diff options
347 files changed, 19008 insertions, 15362 deletions
diff --git a/SConstruct b/SConstruct index 4785e39ec5..e7c10fcf52 100644 --- a/SConstruct +++ b/SConstruct @@ -65,7 +65,7 @@ platform = env['platform'] # derived options x86 = machine == 'x86' gcc = platform in ('linux', 'freebsd', 'darwin') -msvc = platform in ('win32', 'winddk') +msvc = platform in ('windows', 'winddk') Export([ 'debug', @@ -84,8 +84,6 @@ Export([ # TODO: put the compiler specific settings in separate files # TODO: auto-detect as much as possible -common.generate(env) - if platform == 'winddk': env.Tool('winddk', ['.']) @@ -96,37 +94,7 @@ if platform == 'winddk': env['CRT_INC_PATH'], ]) -# Optimization flags -if gcc: - if debug: - env.Append(CFLAGS = '-O0 -g3') - env.Append(CXXFLAGS = '-O0 -g3') - else: - env.Append(CFLAGS = '-O3 -g3') - env.Append(CXXFLAGS = '-O3 -g3') - - env.Append(CFLAGS = '-Wall -Wmissing-prototypes -Wno-long-long -ffast-math -pedantic') - env.Append(CXXFLAGS = '-Wall -pedantic') - - # Be nice to Eclipse - env.Append(CFLAGS = '-fmessage-length=0') - env.Append(CXXFLAGS = '-fmessage-length=0') - -if msvc: - cflags = [ - #'/Wp64', # enable 64 bit porting warnings - ] - env.Append(CFLAGS = cflags) - env.Append(CXXFLAGS = cflags) - # Put debugging information in a separate .pdb file for each object file as - # descrived in the scons manpage - env['CCPDBFLAGS'] = '/Zi /Fd${TARGET}.pdb' - -# Defines -if debug: - env.Append(CPPDEFINES = ['DEBUG']) -else: - env.Append(CPPDEFINES = ['NDEBUG']) +common.generate(env) # Includes @@ -140,12 +108,6 @@ env.Append(CPPPATH = [ # x86 assembly if x86: - env.Append(CPPDEFINES = [ - 'USE_X86_ASM', - 'USE_MMX_ASM', - 'USE_3DNOW_ASM', - 'USE_SSE_ASM', - ]) if gcc: env.Append(CFLAGS = '-m32') env.Append(CXXFLAGS = '-m32') @@ -188,12 +150,11 @@ if llvm: # See also http://www.scons.org/wiki/UsingPkgConfig env.ParseConfig('llvm-config --cflags --ldflags --libs') env.Append(CPPDEFINES = ['MESA_LLVM']) - env.Append(CXXFLAGS = ['-Wno-long-long']) # Force C++ linkage env['LINK'] = env['CXX'] # libGL -if platform not in ('winddk',): +if platform in ('linux', 'freebsd', 'darwin'): env.Append(LIBS = [ 'X11', 'Xext', @@ -34,7 +34,7 @@ default_machine = _machine_map.get(default_machine, 'generic') if default_platform in ('linux', 'freebsd', 'darwin'): default_dri = 'yes' -elif default_platform in ('winddk',): +elif default_platform in ('winddk', 'windows'): default_dri = 'no' else: default_dri = 'no' @@ -44,14 +44,20 @@ else: # Common options def AddOptions(opts): - from SCons.Options.BoolOption import BoolOption - from SCons.Options.EnumOption import EnumOption + try: + from SCons.Options.BoolOption import BoolOption + except ImportError: + from SCons.Variables.BoolVariable import BoolVariable as BoolOption + try: + from SCons.Options.EnumOption import EnumOption + except ImportError: + from SCons.Variables.EnumVariable import EnumVariable as EnumOption opts.Add(BoolOption('debug', 'build debug version', 'no')) #opts.Add(BoolOption('quiet', 'quiet command lines', 'no')) opts.Add(EnumOption('machine', 'use machine-specific assembly code', default_machine, allowed_values=('generic', 'x86', 'x86_64'))) opts.Add(EnumOption('platform', 'target platform', default_platform, - allowed_values=('linux', 'cell', 'winddk'))) + allowed_values=('linux', 'cell', 'windows', 'winddk'))) opts.Add(BoolOption('llvm', 'use LLVM', 'no')) opts.Add(BoolOption('dri', 'build DRI drivers', default_dri)) @@ -133,8 +139,162 @@ def generate(env): # FIXME: this is already too late #if env.get('quiet', False): # quietCommandLines(env) - createConvenienceLibBuilder(env) + + # shortcuts + debug = env['debug'] + machine = env['machine'] + platform = env['platform'] + x86 = env['machine'] == 'x86' + gcc = env['platform'] in ('linux', 'freebsd', 'darwin') + msvc = env['platform'] in ('windows', 'winddk') + + # C preprocessor options + cppdefines = [] + if debug: + cppdefines += ['DEBUG'] + else: + cppdefines += ['NDEBUG'] + if platform == 'windows': + cppdefines += [ + 'WIN32', + '_WINDOWS', + '_UNICODE', + 'UNICODE', + # http://msdn2.microsoft.com/en-us/library/6dwk3a1z.aspx, + 'WIN32_LEAN_AND_MEAN', + 'VC_EXTRALEAN', + '_CRT_SECURE_NO_DEPRECATE', + ] + if debug: + cppdefines += ['_DEBUG'] + if platform == 'winddk': + # Mimic WINDDK's builtin flags. See also: + # - WINDDK's bin/makefile.new i386mk.inc for more info. + # - buildchk_wxp_x86.log files, generated by the WINDDK's build + # - http://alter.org.ua/docs/nt_kernel/vc8_proj/ + cppdefines += [ + ('_X86_', '1'), + ('i386', '1'), + 'STD_CALL', + ('CONDITION_HANDLING', '1'), + ('NT_INST', '0'), + ('WIN32', '100'), + ('_NT1X_', '100'), + ('WINNT', '1'), + ('_WIN32_WINNT', '0x0501'), # minimum required OS version + ('WINVER', '0x0501'), + ('_WIN32_IE', '0x0603'), + ('WIN32_LEAN_AND_MEAN', '1'), + ('DEVL', '1'), + ('__BUILDMACHINE__', 'WinDDK'), + ('FPO', '0'), + ] + if debug: + cppdefines += [('DBG', 1)] + if platform == 'windows': + cppdefines += ['PIPE_SUBSYSTEM_USER'] + if platform == 'winddk': + cppdefines += ['PIPE_SUBSYSTEM_KERNEL'] + env.Append(CPPDEFINES = cppdefines) + + # C compiler options + cflags = [] + if gcc: + if debug: + cflags += ['-O0', '-g3'] + else: + cflags += ['-O3', '-g3'] + cflags += [ + '-Wall', + '-Wmissing-prototypes', + '-Wno-long-long', + '-ffast-math', + '-pedantic', + '-fmessage-length=0', # be nice to Eclipse + ] + if msvc: + # See also: + # - http://msdn2.microsoft.com/en-us/library/y0zzbyt4.aspx + # - cl /? + if debug: + cflags += [ + '/Od', # disable optimizations + '/Oi', # enable intrinsic functions + '/Oy-', # disable frame pointer omission + ] + else: + cflags += [ + '/Ox', # maximum optimizations + '/Oi', # enable intrinsic functions + '/Os', # favor code space + ] + if platform == 'windows': + cflags += [ + # TODO + #'/Wp64', # enable 64 bit porting warnings + ] + if platform == 'winddk': + cflags += [ + '/Zl', # omit default library name in .OBJ + '/Zp8', # 8bytes struct member alignment + '/Gy', # separate functions for linker + '/Gm-', # disable minimal rebuild + '/W3', # warning level + '/WX', # treat warnings as errors + '/Gz', # __stdcall Calling convention + '/GX-', # disable C++ EH + '/GR-', # disable C++ RTTI + '/GF', # enable read-only string pooling + '/GS', # enable security checks + '/G6', # optimize for PPro, P-II, P-III + '/Ze', # enable extensions + #'/Gi-', # ??? + '/QIfdiv-', # disable Pentium FDIV fix + #'/hotpatch', # ??? + #'/Z7', #enable old-style debug info + ] + # Put debugging information in a separate .pdb file for each object file as + # descrived in the scons manpage + env['CCPDBFLAGS'] = '/Zi /Fd${TARGET}.pdb' + env.Append(CFLAGS = cflags) + env.Append(CXXFLAGS = cflags) + # Linker options + if platform == 'winddk': + # See also: + # - http://msdn2.microsoft.com/en-us/library/y0zzbyt4.aspx + env.Append(LINKFLAGS = [ + '/merge:_PAGE=PAGE', + '/merge:_TEXT=.text', + '/section:INIT,d', + '/opt:ref', + '/opt:icf', + '/ignore:4198,4010,4037,4039,4065,4070,4078,4087,4089,4221', + '/incremental:no', + '/fullbuild', + '/release', + '/nodefaultlib', + '/wx', + '/debug', + '/debugtype:cv', + '/version:5.1', + '/osversion:5.1', + '/functionpadmin:5', + '/safeseh', + '/pdbcompress', + '/stack:0x40000,0x1000', + '/driver', + '/align:0x80', + '/subsystem:native,5.01', + '/base:0x10000', + + '/entry:DrvEnableDriver', + ]) + + + createConvenienceLibBuilder(env) + + # for debugging #print env.Dump() diff --git a/configs/default b/configs/default index f0146c50ce..04604f8006 100644 --- a/configs/default +++ b/configs/default @@ -68,7 +68,7 @@ PROGRAM_DIRS = demos redbook samples glsl xdemos # Gallium directories and -GALLIUM_AUXILIARY_DIRS = draw cso_cache pipebuffer tgsi rtasm util sct +GALLIUM_AUXILIARY_DIRS = draw cso_cache pipebuffer tgsi sct translate rtasm util GALLIUM_AUXILIARIES = $(foreach DIR,$(GALLIUM_AUXILIARY_DIRS),$(TOP)/src/gallium/auxiliary/$(DIR)/lib$(DIR).a) GALLIUM_DRIVER_DIRS = softpipe i915simple i965simple failover GALLIUM_DRIVERS = $(foreach DIR,$(GALLIUM_DRIVER_DIRS),$(TOP)/src/gallium/drivers/$(DIR)/lib$(DIR).a) diff --git a/include/GLES/egl.h b/include/GLES/egl.h index f18c86ca62..a211bbc3be 100644 --- a/include/GLES/egl.h +++ b/include/GLES/egl.h @@ -1,64 +1,24 @@ -#ifndef _EGL_H -#define _EGL_H +#ifndef __egl_h_ +#define __egl_h_ /* -** License Applicability. Except to the extent portions of this file are -** made subject to an alternative license as permitted in the SGI Free -** Software License B, Version 1.0 (the "License"), the contents of this -** file are subject only to the provisions of the License. You may not use -** this file except in compliance with the License. You may obtain a copy -** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600 -** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at: +** Copyright 2002-2003 Promoters of the Khronos Group (3Dlabs, ARM Ltd., +** ATI Technologies, Inc., Discreet, Ericsson Mobile, Imagination +** Technologies Group plc, Motorola, Inc., Nokia, Silicon Graphics, Inc., +** SK Telecom, and Sun Microsystems). ** -** http://oss.sgi.com/projects/FreeB -** -** Note that, as provided in the License, the Software is distributed on an -** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS -** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND -** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A -** PARTICULAR PURPOSE, AND NON-INFRINGEMENT. -** -** Original Code. The Original Code is: OpenGL Sample Implementation, -** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics, -** Inc. The Original Code is Copyright (c) 1991-2004 Silicon Graphics, Inc. -** Copyright in any portions created by third parties is as indicated -** elsewhere herein. All Rights Reserved. -** -** Additional Notice Provisions: The application programming interfaces -** established by SGI in conjunction with the Original Code are The -** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released -** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version -** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X -** Window System(R) (Version 1.3), released October 19, 1998. This software -** was created using the OpenGL(R) version 1.2.1 Sample Implementation -** published by SGI, but has not been independently verified as being -** compliant with the OpenGL(R) version 1.2.1 Specification. +** This document is protected by copyright, and contains information +** proprietary to The Khronos Group. Any copying, adaptation, distribution, +** public performance, or public display of this document without the +** express written consent of the copyright holders is strictly prohibited. +** The receipt or possession of this document does not convey any rights to +** reproduce, disclose, or distribute its contents, or to manufacture, use, +** or sell anything that it may describe, in whole or in part. */ -#if 0/*XXX TEMPORARY HACK*/ #include <GLES/gl.h> -#else -#include <GL/gl.h> -#endif #include <GLES/egltypes.h> -/* XXX should go in eglext.h */ -#define GL_OES_VERSION_1_0 1 -#define GL_OES_read_format 1 -#define GL_OES_compressed_paletted_texture 1 -#define GL_IMPLEMENTATION_COLOR_READ_TYPE_OES 0x8B9A -#define GL_IMPLEMENTATION_COLOR_READ_FORMAT_OES 0x8B9B -#define GL_PALETTE4_RGB8_OES 0x8B90 -#define GL_PALETTE4_RGBA8_OES 0x8B91 -#define GL_PALETTE4_R5_G6_B5_OES 0x8B92 -#define GL_PALETTE4_RGBA4_OES 0x8B93 -#define GL_PALETTE4_RGB5_A1_OES 0x8B94 -#define GL_PALETTE8_RGB8_OES 0x8B95 -#define GL_PALETTE8_RGBA8_OES 0x8B96 -#define GL_PALETTE8_R5_G6_B5_OES 0x8B97 -#define GL_PALETTE8_RGBA4_OES 0x8B98 -#define GL_PALETTE8_RGB5_A1_OES 0x8B99 -/* XXX */ /* ** Versioning and extensions @@ -194,7 +154,7 @@ GLAPI EGLDisplay APIENTRY eglGetDisplay (NativeDisplayType display); GLAPI EGLBoolean APIENTRY eglInitialize (EGLDisplay dpy, EGLint *major, EGLint *minor); GLAPI EGLBoolean APIENTRY eglTerminate (EGLDisplay dpy); GLAPI const char * APIENTRY eglQueryString (EGLDisplay dpy, EGLint name); -GLAPI void (* APIENTRY eglGetProcAddress (const char *procname))(void); +GLAPI void (APIENTRY * eglGetProcAddress (const char *procname))(); GLAPI EGLBoolean APIENTRY eglGetConfigs (EGLDisplay dpy, EGLConfig *configs, EGLint config_size, EGLint *num_config); GLAPI EGLBoolean APIENTRY eglChooseConfig (EGLDisplay dpy, const EGLint *attrib_list, EGLConfig *configs, EGLint config_size, EGLint *num_config); @@ -227,98 +187,79 @@ GLAPI EGLBoolean APIENTRY eglWaitNative (EGLint engine); GLAPI EGLBoolean APIENTRY eglSwapBuffers (EGLDisplay dpy, EGLSurface draw); GLAPI EGLBoolean APIENTRY eglCopyBuffers (EGLDisplay dpy, EGLSurface surface, NativePixmapType target); +#ifdef __cplusplus +} +#endif + +#endif /* ___egl_h_ */ -/* EGL_MESA_screen extension >>> PRELIMINARY <<< */ -#ifndef EGL_MESA_screen_surface -#define EGL_MESA_screen_surface 1 -#define EGL_BAD_SCREEN_MESA 0x4000 -#define EGL_BAD_MODE_MESA 0x4001 -#define EGL_SCREEN_COUNT_MESA 0x4002 -#define EGL_SCREEN_POSITION_MESA 0x4003 -#define EGL_SCREEN_POSITION_GRANULARITY_MESA 0x4004 -#define EGL_MODE_ID_MESA 0x4005 -#define EGL_REFRESH_RATE_MESA 0x4006 -#define EGL_OPTIMAL_MESA 0x4007 -#define EGL_INTERLACED_MESA 0x4008 -#define EGL_SCREEN_BIT_MESA 0x08 +/* usage -GLAPI EGLBoolean APIENTRY eglChooseModeMESA(EGLDisplay dpy, EGLScreenMESA screen, const EGLint *attrib_list, EGLModeMESA *modes, EGLint modes_size, EGLint *num_modes); -GLAPI EGLBoolean APIENTRY eglGetModesMESA(EGLDisplay dpy, EGLScreenMESA screen, EGLModeMESA *modes, EGLint modes_size, EGLint *num_modes); -GLAPI EGLBoolean APIENTRY eglGetModeAttribMESA(EGLDisplay dpy, EGLModeMESA mode, EGLint attribute, EGLint *value); -GLAPI EGLBoolean APIENTRY eglGetScreensMESA(EGLDisplay dpy, EGLScreenMESA *screens, EGLint max_screens, EGLint *num_screens); -GLAPI EGLSurface APIENTRY eglCreateScreenSurfaceMESA(EGLDisplay dpy, EGLConfig config, const EGLint *attrib_list); -GLAPI EGLBoolean APIENTRY eglShowScreenSurfaceMESA(EGLDisplay dpy, EGLint screen, EGLSurface surface, EGLModeMESA mode); -GLAPI EGLBoolean APIENTRY eglScreenPositionMESA(EGLDisplay dpy, EGLScreenMESA screen, EGLint x, EGLint y); -GLAPI EGLBoolean APIENTRY eglQueryScreenMESA(EGLDisplay dpy, EGLScreenMESA screen, EGLint attribute, EGLint *value); -GLAPI EGLBoolean APIENTRY eglQueryScreenSurfaceMESA(EGLDisplay dpy, EGLScreenMESA screen, EGLSurface *surface); -GLAPI EGLBoolean APIENTRY eglQueryScreenModeMESA(EGLDisplay dpy, EGLScreenMESA screen, EGLModeMESA *mode); -GLAPI const char * APIENTRY eglQueryModeStringMESA(EGLDisplay dpy, EGLModeMESA mode); + // ---------------------------------------------------------------------- + // Initialization + // ---------------------------------------------------------------------- -#endif /* EGL_MESA_screen_surface */ + HDC hdc = GetDC(0); // the screen or window device context, for example + EGLDisplay display = eglGetDisplay(hdc); + + EGLint major, minor; -#ifndef EGL_MESA_copy_context -#define EGL_MESA_copy_context 1 + if (!eglInitialize(display, &major, &minor)) { + // could not initialize display + } -GLAPI EGLBoolean APIENTRY eglCopyContextMESA(EGLDisplay dpy, EGLContext source, EGLContext dest, EGLint mask); + EGLConfig configs[10]; + EGLint matchingConfigs; + EGLint attribList[] = { 0 }; // extend this -#endif /* EGL_MESA_copy_context */ + if (!eglChooseConfig(display, attribList, &configs, 10, &matchingConfigs)) { + // could not choose config + } + if (matchingConfigs < 1) { + // did not find a suitable configuration + } -/* XXX this is preliminary! */ -#ifndef EGL_VERSION_1_2 -#define EGL_VERSION_1_2 1 + EGLConfig config = configs[0]; // pick any -typedef int EGLenum; -typedef int EGLClientBuffer; + EGLSurface surface = eglCreatePbufferSurface(display, config, attribList); + + // test for error -#define EGL_DISPLAY_SCALING 10000 -#define EGL_UNKNOWN ((EGLint)-1) + EGLContext context = eglCreateContext(display, config, 0, attribList); + + // test for error -#define EGL_OPENGL_ES_BIT 0x1 -#define EGL_OPENVG_BIT 0x2 + // ---------------------------------------------------------------------- + // Rendering Loop + // ---------------------------------------------------------------------- -#define EGL_OPENGL_ES_API 0x30A0 -#define EGL_OPENVG_API 0x30A1 + eglMakeCurrent(display, surface, surface, context); -#define EGL_LUMINANCE_SIZE 0x303D -#define EGL_ALPHA_MASK_SIZE 0x303E -#define EGL_COLOR_BUFFER_TYPE 0x303F -#define EGL_RENDERABLE_TYPE 0x3040 + // actual GL rendering goes here -#define EGL_SINGLE_BUFFER 0x3085 -#define EGL_RENDER_BUFFER 0x3086 -#define EGL_COLORSPACE 0x3087 -#define EGL_ALPHA_FORMAT 0x3088 -#define EGL_COLORSPACE_sRGB 0x3089 -#define EGL_COLORSPACE_LINEAR 0x308A -#define EGL_ALPHA_FORMAT_NONPRE 0x308B -#define EGL_ALPHA_FORMAT_PRE 0x308C -#define EGL_CLIENT_APIS 0x308D -#define EGL_RGB_BUFFER 0x308E -#define EGL_LUMINANCE_BUFFER 0x308F -#define EGL_HORIZONTAL_RESOLUTION 0x3090 -#define EGL_VERTICAL_RESOLUTION 0x3091 -#define EGL_PIXEL_ASPECT_RATIO 0x3092 -#define EGL_SWAP_BEHAVIOR 0x3093 -#define EGL_BUFFER_PRESERVED 0x3094 -#define EGL_BUFFER_DESTROYED 0x3095 -#define EGL_OPENVG_IMAGE 0x3096 -#define EGL_CONTEXT_CLIENT_TYPE 0x3097 + eglWait(); + eglSwapBuffers(display, surface); -GLAPI EGLBoolean APIENTRY eglBindAPI(EGLenum api); -GLAPI EGLenum APIENTRY eglQueryAPI(void); -GLAPI EGLBoolean APIENTRY eglWaitClient(void); -GLAPI EGLBoolean APIENTRY eglReleaseThread(void); -GLAPI EGLSurface APIENTRY eglCreatePbufferFromClientBuffer(EGLDisplay dpy, EGLenum buftype, EGLClientBuffer buffer, EGLConfig config, const EGLint *attrib_list); + // ---------------------------------------------------------------------- + // Cleanup + // ---------------------------------------------------------------------- -#endif /* EGL_VERSION_1_2 */ + if (!eglDestoyContext(display, context)) { + // error deallocating context + } + + if (!eglDestroySurface(display, surface)) { + // error deallocating surface + } -#ifdef __cplusplus -} -#endif + if (!eglTerminate(display)) { + // error while cleaning up display + } -#endif /* _EGL_H */ + +*/ diff --git a/include/GLES/egltypes.h b/include/GLES/egltypes.h index 65cba8d579..ca1f5a9cb3 100644 --- a/include/GLES/egltypes.h +++ b/include/GLES/egltypes.h @@ -1,45 +1,100 @@ +#ifndef __egl_types_h_ +#define __egl_types_h_ + /* -** egltypes.h for Mesa +** egltypes.h is platform dependent. It defines: +** +** - EGL types and resources +** - Native types +** - EGL and native handle values ** -** ONLY egl.h SHOULD INCLUDE THIS FILE! +** EGL types and resources are to be typedef'ed with appropriate platform +** dependent resource handle types. EGLint must be an integer of at least +** 32-bit. ** -** See comments about egltypes.h in the standard egl.h file. +** NativeDisplayType, NativeWindowType and NativePixmapType are to be +** replaced with corresponding types of the native window system in egl.h. +** +** EGL and native handle values must match their types. */ +#if (defined(WIN32) || defined(_WIN32_WCE)) -#include <sys/types.h> +#ifndef WIN32_LEAN_AND_MEAN +#define WIN32_LEAN_AND_MEAN // Exclude rarely-used stuff from Windows headers +#endif +// Windows Header Files: +#include <windows.h> -/* -** These opaque EGL types are implemented as unsigned 32-bit integers: -*/ -typedef u_int32_t EGLDisplay; -typedef u_int32_t EGLConfig; -typedef u_int32_t EGLSurface; -typedef u_int32_t EGLContext; +typedef HDC NativeDisplayType; +typedef HWND NativeWindowType; +typedef HBITMAP NativePixmapType; + +#define EGL_DEFAULT_DISPLAY GetDC(0) + +#elif defined(__SYMBIAN32__) -/* EGL_MESA_screen_surface */ -typedef u_int32_t EGLModeMESA; -typedef u_int32_t EGLScreenMESA; +#include <e32def.h> + +class RWindow; +class CWindowGc; +class CFbsBitmap; + +typedef CWindowGc * NativeDisplayType; +typedef RWindow * NativeWindowType; +typedef CFbsBitmap * NativePixmapType; + +#define EGL_DEFAULT_DISPLAY ((NativeDisplayType) 0) + +#elif defined(__gnu_linux__) + +typedef void * NativeDisplayType; +typedef void * NativeWindowType; +typedef void * NativePixmapType; + +#define EGL_DEFAULT_DISPLAY ((NativeDisplayType) 0) + +#else + +# error "Unsupported Operating System" + +#endif + +#ifdef __cplusplus + +namespace EGL { + class Context; + class Config; + class Surface; +} + +typedef const EGL::Config * EGLConfig; +typedef EGL::Surface * EGLSurface; +typedef EGL::Context * EGLContext; + +#else + +typedef void * EGLConfig; +typedef void * EGLSurface; +typedef void * EGLContext; + +#endif /* -** Other basic EGL types: +** Types and resources */ -typedef u_int8_t EGLBoolean; -typedef int32_t EGLint; - -typedef void * NativeDisplayType; -typedef int NativePixmapType; -typedef int NativeWindowType; +typedef int EGLBoolean; +typedef int EGLint; +typedef void * EGLDisplay; /* -** EGL and native handle null values: +** EGL and native handle values */ -#define EGL_DEFAULT_DISPLAY ((NativeDisplayType) 0) -#define EGL_NO_CONTEXT ((EGLContext) 0) -#define EGL_NO_DISPLAY ((EGLDisplay) 0) -#define EGL_NO_SURFACE ((EGLSurface) 0) +#define EGL_NO_CONTEXT ((EGLContext)0) +#define EGL_NO_DISPLAY ((EGLDisplay)0) +#define EGL_NO_SURFACE ((EGLSurface)0) + -/* EGL_MESA_screen_surface */ -#define EGL_NO_MODE_MESA ((EGLModeMESA) 0) +#endif //ndef __egl_types_h_ diff --git a/include/GLES/gl.h b/include/GLES/gl.h new file mode 100644 index 0000000000..5462c72458 --- /dev/null +++ b/include/GLES/gl.h @@ -0,0 +1,861 @@ +#ifndef __gl_h_ +#define __gl_h_ + +#ifdef __cplusplus +extern "C" { +#endif + +/* +** License Applicability. Except to the extent portions of this file are +** made subject to an alternative license as permitted in the SGI Free +** Software License B, Version 1.0 (the "License"), the contents of this +** file are subject only to the provisions of the License. You may not use +** this file except in compliance with the License. You may obtain a copy +** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600 +** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at: +** +** http://oss.sgi.com/projects/FreeB +** +** Note that, as provided in the License, the Software is distributed on an +** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS +** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND +** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A +** PARTICULAR PURPOSE, AND NON-INFRINGEMENT. +** +** Original Code. The Original Code is: OpenGL Sample Implementation, +** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics, +** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc. +** Copyright in any portions created by third parties is as indicated +** elsewhere herein. All Rights Reserved. +** +** Additional Notice Provisions: The application programming interfaces +** established by SGI in conjunction with the Original Code are The +** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released +** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version +** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X +** Window System(R) (Version 1.3), released October 19, 1998. This software +** was created using the OpenGL(R) version 1.2.1 Sample Implementation +** published by SGI, but has not been independently verified as being +** compliant with the OpenGL(R) version 1.2.1 Specification. +*/ + +#if defined(__SYMBIAN32__) + +# include <e32def.h> + +#elif (defined(WIN32) || defined(_WIN32_WCE)) /*&& !defined(APIENTRY) && !defined(__CYGWIN__)*/ + +# ifndef WIN32_LEAN_AND_MEAN +# define WIN32_LEAN_AND_MEAN // Exclude rarely-used stuff from Windows headers +# endif + +# include <windows.h> + +#elif defined(__gnu_linux__) + +# define GLAPI + +# ifdef APIENTRY +# undef APIENTRY +# endif + +#else + +# error "Unsupported Operating System" + +#endif + + +#ifndef APIENTRY +#define APIENTRY +#endif + +#ifndef GLAPI + #ifdef OGLES_EXPORTS + #define GLAPI __declspec(dllexport) + #else + #define GLAPI __declspec(dllimport) + #endif +#endif + +typedef unsigned int GLenum; +typedef unsigned char GLboolean; +typedef unsigned int GLbitfield; +typedef signed char GLbyte; +typedef short GLshort; +typedef int GLint; +typedef int GLsizei; +typedef unsigned char GLubyte; +typedef unsigned short GLushort; +typedef unsigned int GLuint; +typedef float GLfloat; +typedef float GLclampf; +typedef void GLvoid; +typedef int GLintptr; +typedef int GLsizeiptr; +typedef int GLfixed; +typedef int GLclampx; +/* Internal convenience typedefs */ +typedef void (*_GLfuncptr)(); + +/*************************************************************/ + +/* OpenGL ES core versions */ +#define GL_OES_VERSION_1_0 1 +#define GL_OES_VERSION_1_1 1 + +/* Extensions */ +#define GL_OES_compressed_paletted_texture 1 +/*#define GL_OES_draw_texture 1*/ +#define GL_OES_matrix_get 1 +/*#define GL_OES_matrix_palette 1*/ +#define GL_OES_point_size_array 1 +#define GL_OES_point_sprite 1 +#define GL_OES_read_format 1 + +/* ClearBufferMask */ +#define GL_DEPTH_BUFFER_BIT 0x00000100 +#define GL_STENCIL_BUFFER_BIT 0x00000400 +#define GL_COLOR_BUFFER_BIT 0x00004000 + +/* Boolean */ +#define GL_FALSE 0 +#define GL_TRUE 1 + +/* BeginMode */ +#define GL_POINTS 0x0000 +#define GL_LINES 0x0001 +#define GL_LINE_LOOP 0x0002 +#define GL_LINE_STRIP 0x0003 +#define GL_TRIANGLES 0x0004 +#define GL_TRIANGLE_STRIP 0x0005 +#define GL_TRIANGLE_FAN 0x0006 + +/* AlphaFunction */ +#define GL_NEVER 0x0200 +#define GL_LESS 0x0201 +#define GL_EQUAL 0x0202 +#define GL_LEQUAL 0x0203 +#define GL_GREATER 0x0204 +#define GL_NOTEQUAL 0x0205 +#define GL_GEQUAL 0x0206 +#define GL_ALWAYS 0x0207 + +/* BlendingFactorDest */ +#define GL_ZERO 0 +#define GL_ONE 1 +#define GL_SRC_COLOR 0x0300 +#define GL_ONE_MINUS_SRC_COLOR 0x0301 +#define GL_SRC_ALPHA 0x0302 +#define GL_ONE_MINUS_SRC_ALPHA 0x0303 +#define GL_DST_ALPHA 0x0304 +#define GL_ONE_MINUS_DST_ALPHA 0x0305 + +/* BlendingFactorSrc */ +/* GL_ZERO */ +/* GL_ONE */ +#define GL_DST_COLOR 0x0306 +#define GL_ONE_MINUS_DST_COLOR 0x0307 +#define GL_SRC_ALPHA_SATURATE 0x0308 +/* GL_SRC_ALPHA */ +/* GL_ONE_MINUS_SRC_ALPHA */ +/* GL_DST_ALPHA */ +/* GL_ONE_MINUS_DST_ALPHA */ + +/* ClipPlaneName */ +#define GL_CLIP_PLANE0 0x3000 +#define GL_CLIP_PLANE1 0x3001 +#define GL_CLIP_PLANE2 0x3002 +#define GL_CLIP_PLANE3 0x3003 +#define GL_CLIP_PLANE4 0x3004 +#define GL_CLIP_PLANE5 0x3005 + +/* ColorMaterialFace */ +/* GL_FRONT_AND_BACK */ + +/* ColorMaterialParameter */ +/* GL_AMBIENT_AND_DIFFUSE */ + +/* ColorPointerType */ +/* GL_UNSIGNED_BYTE */ +/* GL_FLOAT */ +/* GL_FIXED */ + +/* CullFaceMode */ +#define GL_FRONT 0x0404 +#define GL_BACK 0x0405 +#define GL_FRONT_AND_BACK 0x0408 + +/* DepthFunction */ +/* GL_NEVER */ +/* GL_LESS */ +/* GL_EQUAL */ +/* GL_LEQUAL */ +/* GL_GREATER */ +/* GL_NOTEQUAL */ +/* GL_GEQUAL */ +/* GL_ALWAYS */ + +/* EnableCap */ +#define GL_FOG 0x0B60 +#define GL_LIGHTING 0x0B50 +#define GL_TEXTURE_2D 0x0DE1 +#define GL_CULL_FACE 0x0B44 +#define GL_ALPHA_TEST 0x0BC0 +#define GL_BLEND 0x0BE2 +#define GL_COLOR_LOGIC_OP 0x0BF2 +#define GL_DITHER 0x0BD0 +#define GL_STENCIL_TEST 0x0B90 +#define GL_DEPTH_TEST 0x0B71 +/* GL_LIGHT0 */ +/* GL_LIGHT1 */ +/* GL_LIGHT2 */ +/* GL_LIGHT3 */ +/* GL_LIGHT4 */ +/* GL_LIGHT5 */ +/* GL_LIGHT6 */ +/* GL_LIGHT7 */ +#define GL_POINT_SMOOTH 0x0B10 +#define GL_LINE_SMOOTH 0x0B20 +#define GL_SCISSOR_TEST 0x0C11 +#define GL_COLOR_MATERIAL 0x0B57 +#define GL_NORMALIZE 0x0BA1 +#define GL_RESCALE_NORMAL 0x803A +#define GL_POLYGON_OFFSET_FILL 0x8037 +#define GL_VERTEX_ARRAY 0x8074 +#define GL_NORMAL_ARRAY 0x8075 +#define GL_COLOR_ARRAY 0x8076 +#define GL_TEXTURE_COORD_ARRAY 0x8078 +#define GL_MULTISAMPLE 0x809D +#define GL_SAMPLE_ALPHA_TO_COVERAGE 0x809E +#define GL_SAMPLE_ALPHA_TO_ONE 0x809F +#define GL_SAMPLE_COVERAGE 0x80A0 + +/* ErrorCode */ +#define GL_NO_ERROR 0 +#define GL_INVALID_ENUM 0x0500 +#define GL_INVALID_VALUE 0x0501 +#define GL_INVALID_OPERATION 0x0502 +#define GL_STACK_OVERFLOW 0x0503 +#define GL_STACK_UNDERFLOW 0x0504 +#define GL_OUT_OF_MEMORY 0x0505 + +/* FogMode */ +/* GL_LINEAR */ +#define GL_EXP 0x0800 +#define GL_EXP2 0x0801 + +/* FogParameter */ +#define GL_FOG_DENSITY 0x0B62 +#define GL_FOG_START 0x0B63 +#define GL_FOG_END 0x0B64 +#define GL_FOG_MODE 0x0B65 +#define GL_FOG_COLOR 0x0B66 + +/* FrontFaceDirection */ +#define GL_CW 0x0900 +#define GL_CCW 0x0901 + +/* GetPName */ +#define GL_CURRENT_COLOR 0x0B00 +#define GL_CURRENT_NORMAL 0x0B02 +#define GL_CURRENT_TEXTURE_COORDS 0x0B03 +#define GL_POINT_SIZE 0x0B11 +#define GL_POINT_SIZE_MIN 0x8126 +#define GL_POINT_SIZE_MAX 0x8127 +#define GL_POINT_FADE_THRESHOLD_SIZE 0x8128 +#define GL_POINT_DISTANCE_ATTENUATION 0x8129 +#define GL_SMOOTH_POINT_SIZE_RANGE 0x0B12 +#define GL_LINE_WIDTH 0x0B21 +#define GL_SMOOTH_LINE_WIDTH_RANGE 0x0B22 +#define GL_ALIASED_POINT_SIZE_RANGE 0x846D +#define GL_ALIASED_LINE_WIDTH_RANGE 0x846E +#define GL_CULL_FACE_MODE 0x0B45 +#define GL_FRONT_FACE 0x0B46 +#define GL_SHADE_MODEL 0x0B54 +#define GL_DEPTH_RANGE 0x0B70 +#define GL_DEPTH_WRITEMASK 0x0B72 +#define GL_DEPTH_CLEAR_VALUE 0x0B73 +#define GL_DEPTH_FUNC 0x0B74 +#define GL_STENCIL_CLEAR_VALUE 0x0B91 +#define GL_STENCIL_FUNC 0x0B92 +#define GL_STENCIL_VALUE_MASK 0x0B93 +#define GL_STENCIL_FAIL 0x0B94 +#define GL_STENCIL_PASS_DEPTH_FAIL 0x0B95 +#define GL_STENCIL_PASS_DEPTH_PASS 0x0B96 +#define GL_STENCIL_REF 0x0B97 +#define GL_STENCIL_WRITEMASK 0x0B98 +#define GL_MATRIX_MODE 0x0BA0 +#define GL_VIEWPORT 0x0BA2 +#define GL_MODELVIEW_STACK_DEPTH 0x0BA3 +#define GL_PROJECTION_STACK_DEPTH 0x0BA4 +#define GL_TEXTURE_STACK_DEPTH 0x0BA5 +#define GL_MODELVIEW_MATRIX 0x0BA6 +#define GL_PROJECTION_MATRIX 0x0BA7 +#define GL_TEXTURE_MATRIX 0x0BA8 +#define GL_ALPHA_TEST_FUNC 0x0BC1 +#define GL_ALPHA_TEST_REF 0x0BC2 +#define GL_BLEND_DST 0x0BE0 +#define GL_BLEND_SRC 0x0BE1 +#define GL_LOGIC_OP_MODE 0x0BF0 +#define GL_SCISSOR_BOX 0x0C10 +#define GL_SCISSOR_TEST 0x0C11 +#define GL_COLOR_CLEAR_VALUE 0x0C22 +#define GL_COLOR_WRITEMASK 0x0C23 +#define GL_UNPACK_ALIGNMENT 0x0CF5 +#define GL_PACK_ALIGNMENT 0x0D05 +#define GL_MAX_LIGHTS 0x0D31 +#define GL_MAX_CLIP_PLANES 0x0D32 +#define GL_MAX_TEXTURE_SIZE 0x0D33 +#define GL_MAX_MODELVIEW_STACK_DEPTH 0x0D36 +#define GL_MAX_PROJECTION_STACK_DEPTH 0x0D38 +#define GL_MAX_TEXTURE_STACK_DEPTH 0x0D39 +#define GL_MAX_VIEWPORT_DIMS 0x0D3A +#define GL_MAX_ELEMENTS_VERTICES 0x80E8 +#define GL_MAX_ELEMENTS_INDICES 0x80E9 +#define GL_MAX_TEXTURE_UNITS 0x84E2 +#define GL_SUBPIXEL_BITS 0x0D50 +#define GL_RED_BITS 0x0D52 +#define GL_GREEN_BITS 0x0D53 +#define GL_BLUE_BITS 0x0D54 +#define GL_ALPHA_BITS 0x0D55 +#define GL_DEPTH_BITS 0x0D56 +#define GL_STENCIL_BITS 0x0D57 +#define GL_POLYGON_OFFSET_UNITS 0x2A00 +#define GL_POLYGON_OFFSET_FILL 0x8037 +#define GL_POLYGON_OFFSET_FACTOR 0x8038 +#define GL_TEXTURE_BINDING_2D 0x8069 +#define GL_VERTEX_ARRAY_SIZE 0x807A +#define GL_VERTEX_ARRAY_TYPE 0x807B +#define GL_VERTEX_ARRAY_STRIDE 0x807C +#define GL_NORMAL_ARRAY_TYPE 0x807E +#define GL_NORMAL_ARRAY_STRIDE 0x807F +#define GL_COLOR_ARRAY_SIZE 0x8081 +#define GL_COLOR_ARRAY_TYPE 0x8082 +#define GL_COLOR_ARRAY_STRIDE 0x8083 +#define GL_TEXTURE_COORD_ARRAY_SIZE 0x8088 +#define GL_TEXTURE_COORD_ARRAY_TYPE 0x8089 +#define GL_TEXTURE_COORD_ARRAY_STRIDE 0x808A +#define GL_VERTEX_ARRAY_POINTER 0x808E +#define GL_NORMAL_ARRAY_POINTER 0x808F +#define GL_COLOR_ARRAY_POINTER 0x8090 +#define GL_TEXTURE_COORD_ARRAY_POINTER 0x8092 +#define GL_SAMPLE_BUFFERS 0x80A8 +#define GL_SAMPLES 0x80A9 +#define GL_SAMPLE_COVERAGE_VALUE 0x80AA +#define GL_SAMPLE_COVERAGE_INVERT 0x80AB + +/* GetTextureParameter */ +/* GL_TEXTURE_MAG_FILTER */ +/* GL_TEXTURE_MIN_FILTER */ +/* GL_TEXTURE_WRAP_S */ +/* GL_TEXTURE_WRAP_T */ + +#define GL_IMPLEMENTATION_COLOR_READ_TYPE_OES 0x8B9A +#define GL_IMPLEMENTATION_COLOR_READ_FORMAT_OES 0x8B9B +#define GL_NUM_COMPRESSED_TEXTURE_FORMATS 0x86A2 +#define GL_COMPRESSED_TEXTURE_FORMATS 0x86A3 + +/* OES_matrix_get */ +#define GL_MODELVIEW_MATRIX_FLOAT_AS_INT_BITS 0x898D +#define GL_PROJECTION_MATRIX_FLOAT_AS_INT_BITS 0x898E +#define GL_TEXTURE_MATRIX_FLOAT_AS_INT_BITS 0x898F + +/* HintMode */ +#define GL_DONT_CARE 0x1100 +#define GL_FASTEST 0x1101 +#define GL_NICEST 0x1102 + +/* HintTarget */ +#define GL_PERSPECTIVE_CORRECTION_HINT 0x0C50 +#define GL_POINT_SMOOTH_HINT 0x0C51 +#define GL_LINE_SMOOTH_HINT 0x0C52 +#define GL_POLYGON_SMOOTH_HINT 0x0C53 +#define GL_FOG_HINT 0x0C54 +#define GL_GENERATE_MIPMAP_HINT 0x8192 + +/* LightModelParameter */ +#define GL_LIGHT_MODEL_AMBIENT 0x0B53 +#define GL_LIGHT_MODEL_TWO_SIDE 0x0B52 + +/* LightParameter */ +#define GL_AMBIENT 0x1200 +#define GL_DIFFUSE 0x1201 +#define GL_SPECULAR 0x1202 +#define GL_POSITION 0x1203 +#define GL_SPOT_DIRECTION 0x1204 +#define GL_SPOT_EXPONENT 0x1205 +#define GL_SPOT_CUTOFF 0x1206 +#define GL_CONSTANT_ATTENUATION 0x1207 +#define GL_LINEAR_ATTENUATION 0x1208 +#define GL_QUADRATIC_ATTENUATION 0x1209 + +/* DataType */ +#define GL_BYTE 0x1400 +#define GL_UNSIGNED_BYTE 0x1401 +#define GL_SHORT 0x1402 +#define GL_UNSIGNED_SHORT 0x1403 +#define GL_FLOAT 0x1406 +#define GL_FIXED 0x140C + +/* LogicOp */ +#define GL_CLEAR 0x1500 +#define GL_AND 0x1501 +#define GL_AND_REVERSE 0x1502 +#define GL_COPY 0x1503 +#define GL_AND_INVERTED 0x1504 +#define GL_NOOP 0x1505 +#define GL_XOR 0x1506 +#define GL_OR 0x1507 +#define GL_NOR 0x1508 +#define GL_EQUIV 0x1509 +#define GL_INVERT 0x150A +#define GL_OR_REVERSE 0x150B +#define GL_COPY_INVERTED 0x150C +#define GL_OR_INVERTED 0x150D +#define GL_NAND 0x150E +#define GL_SET 0x150F + +/* MaterialFace */ +/* GL_FRONT_AND_BACK */ + +/* MaterialParameter */ +#define GL_EMISSION 0x1600 +#define GL_SHININESS 0x1601 +#define GL_AMBIENT_AND_DIFFUSE 0x1602 +/* GL_AMBIENT */ +/* GL_DIFFUSE */ +/* GL_SPECULAR */ + +/* MatrixMode */ +#define GL_MODELVIEW 0x1700 +#define GL_PROJECTION 0x1701 +#define GL_TEXTURE 0x1702 + +/* NormalPointerType */ +/* GL_BYTE */ +/* GL_SHORT */ +/* GL_FLOAT */ +/* GL_FIXED */ + +/* PixelFormat */ +#define GL_ALPHA 0x1906 +#define GL_RGB 0x1907 +#define GL_RGBA 0x1908 +#define GL_LUMINANCE 0x1909 +#define GL_LUMINANCE_ALPHA 0x190A + +/* PixelStoreParameter */ +#define GL_UNPACK_ALIGNMENT 0x0CF5 +#define GL_PACK_ALIGNMENT 0x0D05 + +/* PixelType */ +/* GL_UNSIGNED_BYTE */ +#define GL_UNSIGNED_SHORT_4_4_4_4 0x8033 +#define GL_UNSIGNED_SHORT_5_5_5_1 0x8034 +#define GL_UNSIGNED_SHORT_5_6_5 0x8363 + +/* ShadingModel */ +#define GL_FLAT 0x1D00 +#define GL_SMOOTH 0x1D01 + +/* StencilFunction */ +/* GL_NEVER */ +/* GL_LESS */ +/* GL_EQUAL */ +/* GL_LEQUAL */ +/* GL_GREATER */ +/* GL_NOTEQUAL */ +/* GL_GEQUAL */ +/* GL_ALWAYS */ + +/* StencilOp */ +/* GL_ZERO */ +#define GL_KEEP 0x1E00 +#define GL_REPLACE 0x1E01 +#define GL_INCR 0x1E02 +#define GL_DECR 0x1E03 +/* GL_INVERT */ + +/* StringName */ +#define GL_VENDOR 0x1F00 +#define GL_RENDERER 0x1F01 +#define GL_VERSION 0x1F02 +#define GL_EXTENSIONS 0x1F03 + +/* TexCoordPointerType */ +/* GL_SHORT */ +/* GL_FLOAT */ +/* GL_FIXED */ +/* GL_BYTE */ + +/* TextureEnvMode */ +#define GL_MODULATE 0x2100 +#define GL_DECAL 0x2101 +/* GL_BLEND */ +#define GL_ADD 0x0104 +/* GL_REPLACE */ + +/* TextureEnvParameter */ +#define GL_TEXTURE_ENV_MODE 0x2200 +#define GL_TEXTURE_ENV_COLOR 0x2201 + +/* TextureEnvTarget */ +#define GL_TEXTURE_ENV 0x2300 + +/* TextureMagFilter */ +#define GL_NEAREST 0x2600 +#define GL_LINEAR 0x2601 + +/* TextureMinFilter */ +/* GL_NEAREST */ +/* GL_LINEAR */ +#define GL_NEAREST_MIPMAP_NEAREST 0x2700 +#define GL_LINEAR_MIPMAP_NEAREST 0x2701 +#define GL_NEAREST_MIPMAP_LINEAR 0x2702 +#define GL_LINEAR_MIPMAP_LINEAR 0x2703 + +/* TextureParameterName */ +#define GL_TEXTURE_MAG_FILTER 0x2800 +#define GL_TEXTURE_MIN_FILTER 0x2801 +#define GL_TEXTURE_WRAP_S 0x2802 +#define GL_TEXTURE_WRAP_T 0x2803 +#define GL_GENERATE_MIPMAP 0x8191 + +/* TextureTarget */ +/* GL_TEXTURE_2D */ + +/* TextureUnit */ +#define GL_TEXTURE0 0x84C0 +#define GL_TEXTURE1 0x84C1 +#define GL_TEXTURE2 0x84C2 +#define GL_TEXTURE3 0x84C3 +#define GL_TEXTURE4 0x84C4 +#define GL_TEXTURE5 0x84C5 +#define GL_TEXTURE6 0x84C6 +#define GL_TEXTURE7 0x84C7 +#define GL_TEXTURE8 0x84C8 +#define GL_TEXTURE9 0x84C9 +#define GL_TEXTURE10 0x84CA +#define GL_TEXTURE11 0x84CB +#define GL_TEXTURE12 0x84CC +#define GL_TEXTURE13 0x84CD +#define GL_TEXTURE14 0x84CE +#define GL_TEXTURE15 0x84CF +#define GL_TEXTURE16 0x84D0 +#define GL_TEXTURE17 0x84D1 +#define GL_TEXTURE18 0x84D2 +#define GL_TEXTURE19 0x84D3 +#define GL_TEXTURE20 0x84D4 +#define GL_TEXTURE21 0x84D5 +#define GL_TEXTURE22 0x84D6 +#define GL_TEXTURE23 0x84D7 +#define GL_TEXTURE24 0x84D8 +#define GL_TEXTURE25 0x84D9 +#define GL_TEXTURE26 0x84DA +#define GL_TEXTURE27 0x84DB +#define GL_TEXTURE28 0x84DC +#define GL_TEXTURE29 0x84DD +#define GL_TEXTURE30 0x84DE +#define GL_TEXTURE31 0x84DF +#define GL_ACTIVE_TEXTURE 0x84E0 +#define GL_CLIENT_ACTIVE_TEXTURE 0x84E1 + +/* TextureWrapMode */ +#define GL_REPEAT 0x2901 +#define GL_CLAMP_TO_EDGE 0x812F + +/* PixelInternalFormat */ +#define GL_PALETTE4_RGB8_OES 0x8B90 +#define GL_PALETTE4_RGBA8_OES 0x8B91 +#define GL_PALETTE4_R5_G6_B5_OES 0x8B92 +#define GL_PALETTE4_RGBA4_OES 0x8B93 +#define GL_PALETTE4_RGB5_A1_OES 0x8B94 +#define GL_PALETTE8_RGB8_OES 0x8B95 +#define GL_PALETTE8_RGBA8_OES 0x8B96 +#define GL_PALETTE8_R5_G6_B5_OES 0x8B97 +#define GL_PALETTE8_RGBA4_OES 0x8B98 +#define GL_PALETTE8_RGB5_A1_OES 0x8B99 + +/* VertexPointerType */ +/* GL_SHORT */ +/* GL_FLOAT */ +/* GL_FIXED */ +/* GL_BYTE */ + +/* LightName */ +#define GL_LIGHT0 0x4000 +#define GL_LIGHT1 0x4001 +#define GL_LIGHT2 0x4002 +#define GL_LIGHT3 0x4003 +#define GL_LIGHT4 0x4004 +#define GL_LIGHT5 0x4005 +#define GL_LIGHT6 0x4006 +#define GL_LIGHT7 0x4007 + +/* Buffer Objects */ +#define GL_ARRAY_BUFFER 0x8892 +#define GL_ELEMENT_ARRAY_BUFFER 0x8893 + +#define GL_ARRAY_BUFFER_BINDING 0x8894 +#define GL_ELEMENT_ARRAY_BUFFER_BINDING 0x8895 +#define GL_VERTEX_ARRAY_BUFFER_BINDING 0x8896 +#define GL_NORMAL_ARRAY_BUFFER_BINDING 0x8897 +#define GL_COLOR_ARRAY_BUFFER_BINDING 0x8898 +#define GL_TEXTURE_COORD_ARRAY_BUFFER_BINDING 0x889A + +#define GL_STATIC_DRAW 0x88E4 +#define GL_DYNAMIC_DRAW 0x88E8 + +#define GL_WRITE_ONLY 0x88B9 + +#define GL_BUFFER_SIZE 0x8764 +#define GL_BUFFER_USAGE 0x8765 +#define GL_BUFFER_ACCESS 0x88BB + + +/* Texture combine + dot3 */ +#define GL_SUBTRACT 0x84E7 +#define GL_COMBINE 0x8570 +#define GL_COMBINE_RGB 0x8571 +#define GL_COMBINE_ALPHA 0x8572 +#define GL_RGB_SCALE 0x8573 +#define GL_ADD_SIGNED 0x8574 +#define GL_INTERPOLATE 0x8575 +#define GL_CONSTANT 0x8576 +#define GL_PRIMARY_COLOR 0x8577 +#define GL_PREVIOUS 0x8578 +#define GL_OPERAND0_RGB 0x8590 +#define GL_OPERAND1_RGB 0x8591 +#define GL_OPERAND2_RGB 0x8592 +#define GL_OPERAND0_ALPHA 0x8598 +#define GL_OPERAND1_ALPHA 0x8599 +#define GL_OPERAND2_ALPHA 0x859A + +#define GL_ALPHA_SCALE 0x0D1C + +#define GL_SRC0_RGB 0x8580 +#define GL_SRC1_RGB 0x8581 +#define GL_SRC2_RGB 0x8582 +#define GL_SRC0_ALPHA 0x8588 +#define GL_SRC1_ALPHA 0x8589 +#define GL_SRC2_ALPHA 0x858A + +#define GL_DOT3_RGB 0x86AE +#define GL_DOT3_RGBA 0x86AF + +/*****************************************************************************************/ +/* OES extension functions */ +/*****************************************************************************************/ +/* OES_point_sprite */ +#define GL_POINT_SPRITE_OES 0x8861 +#define GL_COORD_REPLACE_OES 0x8862 + +/* OES_point_size_array */ +#define GL_POINT_SIZE_ARRAY_OES 0x8B9C +#define GL_POINT_SIZE_ARRAY_TYPE_OES 0x898A +#define GL_POINT_SIZE_ARRAY_STRIDE_OES 0x898B +#define GL_POINT_SIZE_ARRAY_POINTER_OES 0x898C +#define GL_POINT_SIZE_ARRAY_BUFFER_BINDING_OES 0x8B9F + +/* OES_matrix_palette */ +#define GL_MAX_VERTEX_UNITS_OES 0x86A4 +#define GL_MAX_PALETTE_MATRICES_OES 0x8842 +#define GL_MATRIX_PALETTE_OES 0x8840 +#define GL_MATRIX_INDEX_ARRAY_OES 0x8844 +#define GL_WEIGHT_ARRAY_OES 0x86AD + +#define GL_MATRIX_INDEX_ARRAY_SIZE_OES 0x8846 +#define GL_MATRIX_INDEX_ARRAY_TYPE_OES 0x8847 +#define GL_MATRIX_INDEX_ARRAY_STRIDE_OES 0x8848 +#define GL_MATRIX_INDEX_ARRAY_POINTER_OES 0x8849 +#define GL_MATRIX_INDEX_ARRAY_BUFFER_BINDING_OES 0x8B9E + +#define GL_WEIGHT_ARRAY_SIZE_OES 0x86AB +#define GL_WEIGHT_ARRAY_TYPE_OES 0x86A9 +#define GL_WEIGHT_ARRAY_STRIDE_OES 0x86AA +#define GL_WEIGHT_ARRAY_POINTER_OES 0x86AC +#define GL_WEIGHT_ARRAY_BUFFER_BINDING_OES 0x889E + +/* OES_draw_texture */ +#define GL_TEXTURE_CROP_RECT_OES 0x8B9D + +/*************************************************************/ + +GLAPI void APIENTRY glActiveTexture (GLenum texture); +GLAPI void APIENTRY glAlphaFunc (GLenum func, GLclampf ref); +GLAPI void APIENTRY glAlphaFuncx (GLenum func, GLclampx ref); +GLAPI void APIENTRY glBindBuffer(GLenum target, GLuint buffer); +GLAPI void APIENTRY glBindTexture (GLenum target, GLuint texture); +GLAPI void APIENTRY glBlendFunc (GLenum sfactor, GLenum dfactor); +GLAPI void APIENTRY glBufferData(GLenum target, GLsizeiptr size, const GLvoid *data, GLenum usage); +GLAPI void APIENTRY glBufferSubData(GLenum target, GLintptr offset, GLsizeiptr size, const GLvoid *data); +GLAPI void APIENTRY glClear (GLbitfield mask); +GLAPI void APIENTRY glClearColor (GLclampf red, GLclampf green, GLclampf blue, GLclampf alpha); +GLAPI void APIENTRY glClearColorx (GLclampx red, GLclampx green, GLclampx blue, GLclampx alpha); +GLAPI void APIENTRY glClearDepthf (GLclampf depth); +GLAPI void APIENTRY glClearDepthx (GLclampx depth); +GLAPI void APIENTRY glClearStencil (GLint s); +GLAPI void APIENTRY glClientActiveTexture (GLenum texture); +GLAPI void APIENTRY glClipPlanex(GLenum plane, const GLfixed *equation); +GLAPI void APIENTRY glClipPlanef(GLenum plane, const GLfloat *equation); +GLAPI void APIENTRY glColor4f (GLfloat red, GLfloat green, GLfloat blue, GLfloat alpha); +GLAPI void APIENTRY glColor4x (GLfixed red, GLfixed green, GLfixed blue, GLfixed alpha); +GLAPI void APIENTRY glColorMask (GLboolean red, GLboolean green, GLboolean blue, GLboolean alpha); +GLAPI void APIENTRY glColorPointer (GLint size, GLenum type, GLsizei stride, const GLvoid *pointer); +GLAPI void APIENTRY glCompressedTexImage2D (GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLint border, GLsizei imageSize, const GLvoid *data); +GLAPI void APIENTRY glCompressedTexSubImage2D (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, GLenum format, GLsizei imageSize, const GLvoid *data); +GLAPI void APIENTRY glCopyTexImage2D (GLenum target, GLint level, GLenum internalformat, GLint x, GLint y, GLsizei width, GLsizei height, GLint border); +GLAPI void APIENTRY glCopyTexSubImage2D (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint x, GLint y, GLsizei width, GLsizei height); +GLAPI void APIENTRY glCullFace (GLenum mode); +GLAPI void APIENTRY glDeleteBuffers(GLsizei n, const GLuint *buffers); +GLAPI void APIENTRY glDeleteTextures (GLsizei n, const GLuint *textures); +GLAPI void APIENTRY glDepthFunc (GLenum func); +GLAPI void APIENTRY glDepthMask (GLboolean flag); +GLAPI void APIENTRY glDepthRangef (GLclampf zNear, GLclampf zFar); +GLAPI void APIENTRY glDepthRangex (GLclampx zNear, GLclampx zFar); +GLAPI void APIENTRY glDisable (GLenum cap); +GLAPI void APIENTRY glDisableClientState (GLenum array); +GLAPI void APIENTRY glDrawArrays (GLenum mode, GLint first, GLsizei count); +GLAPI void APIENTRY glDrawElements (GLenum mode, GLsizei count, GLenum type, const GLvoid *indices); +GLAPI void APIENTRY glEnable (GLenum cap); +GLAPI void APIENTRY glEnableClientState (GLenum array); +GLAPI void APIENTRY glFinish (void); +GLAPI void APIENTRY glFlush (void); +GLAPI void APIENTRY glFogf (GLenum pname, GLfloat param); +GLAPI void APIENTRY glFogfv (GLenum pname, const GLfloat *params); +GLAPI void APIENTRY glFogx (GLenum pname, GLfixed param); +GLAPI void APIENTRY glFogxv (GLenum pname, const GLfixed *params); +GLAPI void APIENTRY glFrontFace (GLenum mode); +GLAPI void APIENTRY glFrustumf (GLfloat left, GLfloat right, GLfloat bottom, GLfloat top, GLfloat zNear, GLfloat zFar); +GLAPI void APIENTRY glFrustumx (GLfixed left, GLfixed right, GLfixed bottom, GLfixed top, GLfixed zNear, GLfixed zFar); +GLAPI void APIENTRY glGetBooleanv(GLenum pname, GLboolean *params); +GLAPI void APIENTRY glGetBufferParameteriv(GLenum target, GLenum pname, GLint *params); +GLAPI void APIENTRY glGetClipPlanef(GLenum pname, GLfloat eqn[4]); +GLAPI void APIENTRY glGetClipPlanex(GLenum pname, GLfixed eqn[4]); +GLAPI void APIENTRY glGenBuffers(GLsizei n, GLuint *buffers); +GLAPI void APIENTRY glGenTextures (GLsizei n, GLuint *textures); +GLAPI GLenum APIENTRY glGetError (void); +GLAPI void APIENTRY glGetFixedv(GLenum pname, GLfixed *params); +GLAPI void APIENTRY glGetFloatv(GLenum pname, GLfloat *params); +GLAPI void APIENTRY glGetIntegerv (GLenum pname, GLint *params); +GLAPI void APIENTRY glGetLightfv(GLenum light, GLenum pname, GLfloat *params); +GLAPI void APIENTRY glGetLightxv(GLenum light, GLenum pname, GLfixed *params); +GLAPI void APIENTRY glGetMaterialfv(GLenum face, GLenum pname, GLfloat *params); +GLAPI void APIENTRY glGetMaterialxv(GLenum face, GLenum pname, GLfixed *params); +GLAPI void APIENTRY glGetPointerv(GLenum pname, void **params); +GLAPI const GLubyte * APIENTRY glGetString (GLenum name); +GLAPI void APIENTRY glGetTexEnviv(GLenum env, GLenum pname, GLint *params); +GLAPI void APIENTRY glGetTexEnvfv(GLenum env, GLenum pname, GLfloat *params); +GLAPI void APIENTRY glGetTexEnvxv(GLenum env, GLenum pname, GLfixed *params); +GLAPI void APIENTRY glGetTexParameteriv(GLenum target, GLenum pname, GLint *params); +GLAPI void APIENTRY glGetTexParameterfv(GLenum target, GLenum pname, GLfloat *params); +GLAPI void APIENTRY glGetTexParameterxv(GLenum target, GLenum pname, GLfixed *params); +GLAPI void APIENTRY glHint (GLenum target, GLenum mode); +GLAPI GLboolean APIENTRY glIsBuffer(GLuint buffer); +GLAPI GLboolean APIENTRY glIsEnabled(GLenum cap); +GLAPI GLboolean APIENTRY glIsTexture(GLuint texture); +GLAPI void APIENTRY glLightModelf (GLenum pname, GLfloat param); +GLAPI void APIENTRY glLightModelfv (GLenum pname, const GLfloat *params); +GLAPI void APIENTRY glLightModelx (GLenum pname, GLfixed param); +GLAPI void APIENTRY glLightModelxv (GLenum pname, const GLfixed *params); +GLAPI void APIENTRY glLightf (GLenum light, GLenum pname, GLfloat param); +GLAPI void APIENTRY glLightfv (GLenum light, GLenum pname, const GLfloat *params); +GLAPI void APIENTRY glLightx (GLenum light, GLenum pname, GLfixed param); +GLAPI void APIENTRY glLightxv (GLenum light, GLenum pname, const GLfixed *params); +GLAPI void APIENTRY glLineWidth (GLfloat width); +GLAPI void APIENTRY glLineWidthx (GLfixed width); +GLAPI void APIENTRY glLoadIdentity (void); +GLAPI void APIENTRY glLoadMatrixf (const GLfloat *m); +GLAPI void APIENTRY glLoadMatrixx (const GLfixed *m); +GLAPI void APIENTRY glLogicOp (GLenum opcode); +GLAPI void APIENTRY glMaterialf (GLenum face, GLenum pname, GLfloat param); +GLAPI void APIENTRY glMaterialfv (GLenum face, GLenum pname, const GLfloat *params); +GLAPI void APIENTRY glMaterialx (GLenum face, GLenum pname, GLfixed param); +GLAPI void APIENTRY glMaterialxv (GLenum face, GLenum pname, const GLfixed *params); +GLAPI void APIENTRY glMatrixMode (GLenum mode); +GLAPI void APIENTRY glMultMatrixf (const GLfloat *m); +GLAPI void APIENTRY glMultMatrixx (const GLfixed *m); +GLAPI void APIENTRY glMultiTexCoord4f (GLenum target, GLfloat s, GLfloat t, GLfloat r, GLfloat q); +GLAPI void APIENTRY glMultiTexCoord4x (GLenum target, GLfixed s, GLfixed t, GLfixed r, GLfixed q); +GLAPI void APIENTRY glNormal3f (GLfloat nx, GLfloat ny, GLfloat nz); +GLAPI void APIENTRY glNormal3x (GLfixed nx, GLfixed ny, GLfixed nz); +GLAPI void APIENTRY glNormalPointer (GLenum type, GLsizei stride, const GLvoid *pointer); +GLAPI void APIENTRY glOrthof (GLfloat left, GLfloat right, GLfloat bottom, GLfloat top, GLfloat zNear, GLfloat zFar); +GLAPI void APIENTRY glOrthox (GLfixed left, GLfixed right, GLfixed bottom, GLfixed top, GLfixed zNear, GLfixed zFar); +GLAPI void APIENTRY glPixelStorei (GLenum pname, GLint param); +GLAPI void APIENTRY glPointParameterf(GLenum pname, GLfloat param); +GLAPI void APIENTRY glPointParameterfv(GLenum pname, const GLfloat *params); +GLAPI void APIENTRY glPointParameterx(GLenum pname, GLfixed param); +GLAPI void APIENTRY glPointParameterxv(GLenum pname, const GLfixed *params); +GLAPI void APIENTRY glPointSize (GLfloat size); +GLAPI void APIENTRY glPointSizex (GLfixed size); +GLAPI void APIENTRY glPolygonOffset (GLfloat factor, GLfloat units); +GLAPI void APIENTRY glPolygonOffsetx (GLfixed factor, GLfixed units); +GLAPI void APIENTRY glPopMatrix (void); +GLAPI void APIENTRY glPushMatrix (void); +GLAPI void APIENTRY glReadPixels (GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, GLenum type, GLvoid *pixels); +GLAPI void APIENTRY glRotatef (GLfloat angle, GLfloat x, GLfloat y, GLfloat z); +GLAPI void APIENTRY glRotatex (GLfixed angle, GLfixed x, GLfixed y, GLfixed z); +GLAPI void APIENTRY glSampleCoverage (GLclampf value, GLboolean invert); +GLAPI void APIENTRY glSampleCoveragex (GLclampx value, GLboolean invert); +GLAPI void APIENTRY glScalef (GLfloat x, GLfloat y, GLfloat z); +GLAPI void APIENTRY glScalex (GLfixed x, GLfixed y, GLfixed z); +GLAPI void APIENTRY glScissor (GLint x, GLint y, GLsizei width, GLsizei height); +GLAPI void APIENTRY glShadeModel (GLenum mode); +GLAPI void APIENTRY glStencilFunc (GLenum func, GLint ref, GLuint mask); +GLAPI void APIENTRY glStencilMask (GLuint mask); +GLAPI void APIENTRY glStencilOp (GLenum fail, GLenum zfail, GLenum zpass); +GLAPI void APIENTRY glTexCoordPointer (GLint size, GLenum type, GLsizei stride, const GLvoid *pointer); +GLAPI void APIENTRY glTexEnvf (GLenum target, GLenum pname, GLfloat param); +GLAPI void APIENTRY glTexEnvfv (GLenum target, GLenum pname, const GLfloat *params); +GLAPI void APIENTRY glTexEnvx (GLenum target, GLenum pname, GLfixed param); +GLAPI void APIENTRY glTexEnvxv (GLenum target, GLenum pname, const GLfixed *params); +GLAPI void APIENTRY glTexImage2D (GLenum target, GLint level, GLint internalformat, GLsizei width, GLsizei height, GLint border, GLenum format, GLenum type, const GLvoid *pixels); +GLAPI void APIENTRY glTexParameteri (GLenum target, GLenum pname, GLint param); +GLAPI void APIENTRY glTexParameterf (GLenum target, GLenum pname, GLfloat param); +GLAPI void APIENTRY glTexParameterx (GLenum target, GLenum pname, GLfixed param); +GLAPI void APIENTRY glTexSubImage2D (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, GLenum format, GLenum type, const GLvoid *pixels); +GLAPI void APIENTRY glTranslatef (GLfloat x, GLfloat y, GLfloat z); +GLAPI void APIENTRY glTranslatex (GLfixed x, GLfixed y, GLfixed z); +GLAPI void APIENTRY glVertexPointer (GLint size, GLenum type, GLsizei stride, const GLvoid *pointer); +GLAPI void APIENTRY glViewport (GLint x, GLint y, GLsizei width, GLsizei height); + +/*****************************************************************************************/ +/* OES extension functions */ +/*****************************************************************************************/ + +#ifdef GL_OES_matrix_palette + +GLAPI void APIENTRY glCurrentPaletteMatrixOES (GLuint matrixpaletteindex); +GLAPI void APIENTRY glLoadPaletteFromModelViewMatrixOES (void); +GLAPI void APIENTRY glMatrixIndexPointerOES (GLint size, GLenum type, GLsizei stride, const GLvoid *pointer); +GLAPI void APIENTRY glWeightPointerOES (GLint size, GLenum type, GLsizei stride, const GLvoid *pointer); + +#endif /* OES_matrix_palette */ + +#ifdef GL_OES_point_size_array + +GLAPI void APIENTRY glPointSizePointerOES(GLenum type, GLsizei stride, const GLvoid *pointer); + +#endif /* OES_point_size_array */ + +#ifdef GL_OES_draw_texture + +GLAPI void APIENTRY glDrawTexsOES (GLshort x, GLshort y, GLshort z, GLshort width, GLshort height); +GLAPI void APIENTRY glDrawTexiOES (GLint x, GLint y, GLint z, GLint width, GLint height); +GLAPI void APIENTRY glDrawTexfOES (GLfloat x, GLfloat y, GLfloat z, GLfloat width, GLfloat height); +GLAPI void APIENTRY glDrawTexxOES (GLfixed x, GLfixed y, GLfixed z, GLfixed width, GLfixed height); + +GLAPI void APIENTRY glDrawTexsvOES (GLshort *coords); +GLAPI void APIENTRY glDrawTexivOES (GLint *coords); +GLAPI void APIENTRY glDrawTexfvOES (GLfloat *coords); +GLAPI void APIENTRY glDrawTexxvOES (GLfixed *coords); + +#endif /* GL_OES_draw_texture */ + +#ifdef __cplusplus +} +#endif + +#endif /* __gl_h_ */ diff --git a/include/GLES/glext.h b/include/GLES/glext.h new file mode 100644 index 0000000000..059c6bf1bb --- /dev/null +++ b/include/GLES/glext.h @@ -0,0 +1,80 @@ +#ifndef __glext_h_ +#define __glext_h_ + +#ifdef __cplusplus +extern "C" { +#endif + +/* +** Copyright 1992-2003, Silicon Graphics, Inc. +** All Rights Reserved. +** +** This is UNPUBLISHED PROPRIETARY SOURCE CODE of Silicon Graphics, Inc.; +** the contents of this file may not be disclosed to third parties, copied or +** duplicated in any form, in whole or in part, without the prior written +** permission of Silicon Graphics, Inc. +** +** RESTRICTED RIGHTS LEGEND: +** Use, duplication or disclosure by the Government is subject to restrictions +** as set forth in subdivision (c)(1)(ii) of the Rights in Technical Data +** and Computer Software clause at DFARS 252.227-7013, and/or in similar or +** successor clauses in the FAR, DOD or NASA FAR Supplement. Unpublished - +** rights reserved under the Copyright Laws of the United States. +** +** -------------------------------------------------------------------------- +** +** Copyright (c) 2004, Hans-Martin Will. All rights reserved. +** +** Redistribution and use in source and binary forms, with or without +** modification, are permitted provided that the following conditions are +** met: +** +** * Redistributions of source code must retain the above copyright +** notice, this list of conditions and the following disclaimer. +** * Redistributions in binary form must reproduce the above copyright +** notice, this list of conditions and the following disclaimer in the +** documentation and/or other materials provided with the distribution. +** +** THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" +** AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE +** IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE +** ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE +** LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, +** OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF +** SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS +** INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN +** CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) +** ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF +** THE POSSIBILITY OF SUCH DAMAGE. +** +*/ + +#ifndef WIN32 +#define APIENTRY +#endif + +#ifndef APIENTRY +#define APIENTRY +#endif + +#ifdef EGL_ON_SYMBIAN +# undef GLAPI +# define GLAPI EXPORT_C +#else +# ifndef GLAPI +# define GLAPI +# endif +#endif + + +#define GL_OES_query_matrix 1 + +/* OES_query_matrix */ +GLAPI GLbitfield APIENTRY glQueryMatrixxOES(GLfixed *mantissa, GLint *exponent); + + +#ifdef __cplusplus +} +#endif + +#endif /* __glext_h_ */ diff --git a/progs/beos/.gitignore b/progs/beos/.gitignore new file mode 100644 index 0000000000..53011ca2e6 --- /dev/null +++ b/progs/beos/.gitignore @@ -0,0 +1,3 @@ +demo +GLInfo +sample diff --git a/progs/demos/drawpix.c b/progs/demos/drawpix.c index d2f5748674..5490bcc635 100644 --- a/progs/demos/drawpix.c +++ b/progs/demos/drawpix.c @@ -28,6 +28,7 @@ static GLfloat Zpos = -1.0; static float Xzoom, Yzoom; static GLboolean DrawFront = GL_FALSE; static GLboolean Dither = GL_TRUE; +static int win = 0; static void Reset( void ) @@ -215,6 +216,7 @@ static void Key( unsigned char key, int x, int y ) printf("glDrawBuffer(%s)\n", DrawFront ? "GL_FRONT" : "GL_BACK"); break; case 27: + glutDestroyWindow(win); exit(0); break; } @@ -343,7 +345,7 @@ int main( int argc, char *argv[] ) else glutInitDisplayMode( GLUT_RGB | GLUT_DOUBLE); - glutCreateWindow(argv[0]); + win = glutCreateWindow(argv[0]); Init(ciMode, filename); Usage(); diff --git a/progs/demos/lodbias.c b/progs/demos/lodbias.c index c5a2a1b457..30b1ed13d5 100644 --- a/progs/demos/lodbias.c +++ b/progs/demos/lodbias.c @@ -42,7 +42,7 @@ static GLfloat Xrot = 0, Yrot = -30, Zrot = 0; static GLboolean Anim = GL_TRUE; static GLint Bias = 0, BiasStepSign = +1; /* ints avoid fp precision problem */ static GLint BiasMin = -400, BiasMax = 400; - +static int win = 0; static void @@ -172,6 +172,7 @@ static void Key( unsigned char key, int x, int y ) Bias = 100.0 * (key - '0'); break; case 27: + glutDestroyWindow(win); exit(0); break; } @@ -281,7 +282,7 @@ int main( int argc, char *argv[] ) glutInitWindowPosition( 0, 0 ); glutInitWindowSize( 350, 350 ); glutInitDisplayMode( GLUT_RGB | GLUT_DOUBLE ); - glutCreateWindow(argv[0]); + win = glutCreateWindow(argv[0]); glutReshapeFunc( Reshape ); glutKeyboardFunc( Key ); glutSpecialFunc( SpecialKey ); diff --git a/progs/demos/shadowtex.c b/progs/demos/shadowtex.c index 59253e8c1e..0574175935 100644 --- a/progs/demos/shadowtex.c +++ b/progs/demos/shadowtex.c @@ -650,6 +650,7 @@ Display(void) glDisable(GL_FRAGMENT_PROGRAM_ARB); } + glDisable(GL_TEXTURE_1D); glDisable(GL_TEXTURE_2D); } diff --git a/progs/demos/streaming_rect.c b/progs/demos/streaming_rect.c index 86e00803c0..4c1d8535fd 100644 --- a/progs/demos/streaming_rect.c +++ b/progs/demos/streaming_rect.c @@ -1,13 +1,9 @@ - /* - * GL_ARB_multitexture demo + * GL_ARB_pixel_buffer_object test * * Command line options: - * -info print GL implementation information - * + * -w WIDTH -h HEIGHT sets window size * - * Brian Paul November 1998 This program is in the public domain. - * Modified on 12 Feb 2002 for > 2 texture units. */ #define GL_GLEXT_PROTOTYPES @@ -25,6 +21,8 @@ #define PBO 11 #define QUIT 100 +static GLuint DrawPBO; + static GLboolean Animate = GL_TRUE; static GLboolean use_pbo = 1; static GLboolean whole_rect = 1; @@ -49,7 +47,7 @@ static void Idle( void ) } } -static int max( int a, int b ) { return a > b ? a : b; } +/*static int max( int a, int b ) { return a > b ? a : b; }*/ static int min( int a, int b ) { return a < b ? a : b; } static void DrawObject() @@ -62,6 +60,7 @@ static void DrawObject() * release the old copy of the texture and allocate a new one * without waiting for outstanding rendering to complete. */ + glBindBufferARB(GL_PIXEL_UNPACK_BUFFER_EXT, DrawPBO); glBufferDataARB(GL_PIXEL_UNPACK_BUFFER_EXT, size, NULL, GL_STREAM_DRAW_ARB); { @@ -69,7 +68,7 @@ static void DrawObject() printf("char %d\n", (unsigned char)(Drift * 255)); - memset(image, size, (unsigned char)(Drift * 255)); + memset(image, (unsigned char)(Drift * 255), size); glUnmapBufferARB(GL_PIXEL_UNPACK_BUFFER_EXT); } @@ -86,7 +85,9 @@ static void DrawObject() if (image == NULL) image = malloc(size); - memset(image, size, (unsigned char)(Drift * 255)); + glBindBufferARB(GL_PIXEL_UNPACK_BUFFER_EXT, 0); + + memset(image, (unsigned char)(Drift * 255), size); /* BGRA should be the fast path for regular uploads as well. */ @@ -227,12 +228,12 @@ static void SpecialKey( int key, int x, int y ) static void Init( int argc, char *argv[] ) { const char *exten = (const char *) glGetString(GL_EXTENSIONS); - GLuint texObj, DrawPBO; + GLuint texObj; GLint size; - if (!strstr(exten, "GL_ARB_multitexture")) { - printf("Sorry, GL_ARB_multitexture not supported by this renderer.\n"); + if (!strstr(exten, "GL_ARB_pixel_buffer_object")) { + printf("Sorry, GL_ARB_pixel_buffer_object not supported by this renderer.\n"); exit(1); } diff --git a/progs/directfb/.gitignore b/progs/directfb/.gitignore new file mode 100644 index 0000000000..55d65fb5b0 --- /dev/null +++ b/progs/directfb/.gitignore @@ -0,0 +1,4 @@ +df_gears +df_morph3d +df_reflect +multi_window diff --git a/progs/egl/.gitignore b/progs/egl/.gitignore new file mode 100644 index 0000000000..1751108235 --- /dev/null +++ b/progs/egl/.gitignore @@ -0,0 +1,5 @@ +demo1 +demo2 +demo3 +eglgears +eglinfo diff --git a/progs/fbdev/.gitignore b/progs/fbdev/.gitignore new file mode 100644 index 0000000000..b9ddf559fd --- /dev/null +++ b/progs/fbdev/.gitignore @@ -0,0 +1 @@ +glfbdevtest diff --git a/progs/glsl/.gitignore b/progs/glsl/.gitignore index 622e0417a8..81ecf5bdd5 100644 --- a/progs/glsl/.gitignore +++ b/progs/glsl/.gitignore @@ -1,5 +1,6 @@ brick bump +convolutions deriv extfuncs.h mandelbrot diff --git a/progs/glsl/Makefile b/progs/glsl/Makefile index 37fa312c30..b9cae66815 100644 --- a/progs/glsl/Makefile +++ b/progs/glsl/Makefile @@ -14,7 +14,8 @@ PROGS = \ mandelbrot \ noise \ toyball \ - texdemo1 + texdemo1 \ + convolutions ##### RULES ##### @@ -62,6 +63,12 @@ texdemo1: texdemo1.o readtex.o texdemo1.o: texdemo1.c readtex.h extfuncs.h $(CC) -c -I$(INCDIR) $(CFLAGS) texdemo1.c +convolutions: convolutions.o readtex.o + $(CC) -I$(INCDIR) $(CFLAGS) convolutions.o readtex.o $(APP_LIB_DEPS) -o $@ + +convolutions.o: convolutions.c readtex.h + $(CC) -c -I$(INCDIR) $(CFLAGS) convolutions.c + clean: -rm -f $(PROGS) diff --git a/progs/glsl/convolution.frag b/progs/glsl/convolution.frag new file mode 100644 index 0000000000..e49b8acf54 --- /dev/null +++ b/progs/glsl/convolution.frag @@ -0,0 +1,21 @@ + +const int KernelSize = 9; + +//texture offsets +uniform vec2 Offset[KernelSize]; +//convolution kernel +uniform vec4 KernelValue[KernelSize]; +uniform sampler2D srcTex; +uniform vec4 ScaleFactor; +uniform vec4 BaseColor; + +void main(void) +{ + int i; + vec4 sum = vec4(0.0); + for (i = 0; i < KernelSize; ++i) { + vec4 tmp = texture2D(srcTex, gl_TexCoord[0].st + Offset[i]); + sum += tmp * KernelValue[i]; + } + gl_FragColor = sum * ScaleFactor + BaseColor; +} diff --git a/progs/glsl/convolution.vert b/progs/glsl/convolution.vert new file mode 100644 index 0000000000..752c54671c --- /dev/null +++ b/progs/glsl/convolution.vert @@ -0,0 +1,5 @@ +void main() { + gl_FrontColor = gl_Color; + gl_TexCoord[0] = gl_MultiTexCoord0; + gl_Position = gl_ModelViewProjectionMatrix * gl_Vertex; +} diff --git a/progs/glsl/convolutions.c b/progs/glsl/convolutions.c new file mode 100644 index 0000000000..be887714c4 --- /dev/null +++ b/progs/glsl/convolutions.c @@ -0,0 +1,441 @@ +#define GL_GLEXT_PROTOTYPES +#include "readtex.h" + +#include <GL/glut.h> +#include <stdio.h> +#include <stdlib.h> +#include <assert.h> +#include <math.h> + +enum Filter { + GAUSSIAN_BLUR, + SHARPEN, + MEAN_REMOVAL, + EMBOSS, + NO_FILTER, + LAST +}; +#define QUIT LAST + +struct BoundingBox { + float minx, miny, minz; + float maxx, maxy, maxz; +}; +struct Texture { + GLuint id; + GLfloat x; + GLfloat y; + GLint width; + GLint height; + GLenum format; +}; + +static const char *textureLocation = "../images/girl2.rgb"; + +static GLfloat viewRotx = 0.0, viewRoty = 0.0, viewRotz = 0.0; +static struct BoundingBox box; +static struct Texture texture; +static GLuint program; +static GLint menuId; +static enum Filter filter = GAUSSIAN_BLUR; + + +static void checkError(int line) +{ + GLenum err = glGetError(); + if (err) { + printf("GL Error %s (0x%x) at line %d\n", + gluErrorString(err), (int) err, line); + } +} + +static void loadAndCompileShader(GLuint shader, const char *text) +{ + GLint stat; + + glShaderSource(shader, 1, (const GLchar **) &text, NULL); + + glCompileShader(shader); + + glGetShaderiv(shader, GL_COMPILE_STATUS, &stat); + if (!stat) { + GLchar log[1000]; + GLsizei len; + glGetShaderInfoLog(shader, 1000, &len, log); + fprintf(stderr, "Problem compiling shader: %s\n", log); + exit(1); + } + else { + printf("Shader compiled OK\n"); + } +} + +static void readShader(GLuint shader, const char *filename) +{ + const int max = 100*1000; + int n; + char *buffer = (char*) malloc(max); + FILE *f = fopen(filename, "r"); + if (!f) { + fprintf(stderr, "Unable to open shader file %s\n", filename); + exit(1); + } + + n = fread(buffer, 1, max, f); + printf("Read %d bytes from shader file %s\n", n, filename); + if (n > 0) { + buffer[n] = 0; + loadAndCompileShader(shader, buffer); + } + + fclose(f); + free(buffer); +} + + +static void +checkLink(GLuint prog) +{ + GLint stat; + glGetProgramiv(prog, GL_LINK_STATUS, &stat); + if (!stat) { + GLchar log[1000]; + GLsizei len; + glGetProgramInfoLog(prog, 1000, &len, log); + fprintf(stderr, "Linker error:\n%s\n", log); + } + else { + fprintf(stderr, "Link success!\n"); + } +} + +static void fillConvolution(GLint *k, + GLfloat *scale, + GLfloat *color) +{ + switch(filter) { + case GAUSSIAN_BLUR: + k[0] = 1; k[1] = 2; k[2] = 1; + k[3] = 2; k[4] = 4; k[5] = 2; + k[6] = 1; k[7] = 2; k[8] = 1; + + *scale = 1./16.; + break; + case SHARPEN: + k[0] = 0; k[1] = -2; k[2] = 0; + k[3] = -2; k[4] = 11; k[5] = -2; + k[6] = 0; k[7] = -2; k[8] = 0; + + *scale = 1./3.; + break; + case MEAN_REMOVAL: + k[0] = -1; k[1] = -1; k[2] = -1; + k[3] = -1; k[4] = 9; k[5] = -1; + k[6] = -1; k[7] = -1; k[8] = -1; + + *scale = 1./1.; + break; + case EMBOSS: + k[0] = -1; k[1] = 0; k[2] = -1; + k[3] = 0; k[4] = 4; k[5] = 0; + k[6] = -1; k[7] = 0; k[8] = -1; + + *scale = 1./1.; + color[0] = 0.5; + color[1] = 0.5; + color[2] = 0.5; + color[3] = 0.5; + break; + case NO_FILTER: + k[0] = 0; k[1] = 0; k[2] = 0; + k[3] = 0; k[4] = 1; k[5] = 0; + k[6] = 0; k[7] = 0; k[8] = 0; + + *scale = 1.; + break; + default: + assert(!"Unhandled switch value"); + } +} + +static void setupConvolution() +{ + GLint *kernel = (GLint*)malloc(sizeof(GLint) * 9); + GLfloat scale; + GLfloat *vecKer = (GLfloat*)malloc(sizeof(GLfloat) * 9 * 4); + GLuint loc; + GLuint i; + GLfloat baseColor[4]; + baseColor[0] = 0; + baseColor[1] = 0; + baseColor[2] = 0; + baseColor[3] = 0; + + fillConvolution(kernel, &scale, baseColor); + /*vector of 4*/ + for (i = 0; i < 9; ++i) { + vecKer[i*4 + 0] = kernel[i]; + vecKer[i*4 + 1] = kernel[i]; + vecKer[i*4 + 2] = kernel[i]; + vecKer[i*4 + 3] = kernel[i]; + } + + loc = glGetUniformLocationARB(program, "KernelValue"); + glUniform4fv(loc, 9, vecKer); + loc = glGetUniformLocationARB(program, "ScaleFactor"); + glUniform4f(loc, scale, scale, scale, scale); + loc = glGetUniformLocationARB(program, "BaseColor"); + glUniform4f(loc, baseColor[0], baseColor[1], + baseColor[2], baseColor[3]); + + free(vecKer); + free(kernel); +} + +static void createProgram(const char *vertProgFile, + const char *fragProgFile) +{ + GLuint fragShader = 0, vertShader = 0; + + program = glCreateProgram(); + if (vertProgFile) { + vertShader = glCreateShader(GL_VERTEX_SHADER); + readShader(vertShader, vertProgFile); + glAttachShader(program, vertShader); + } + + if (fragProgFile) { + fragShader = glCreateShader(GL_FRAGMENT_SHADER); + readShader(fragShader, fragProgFile); + glAttachShader(program, fragShader); + } + + glLinkProgram(program); + checkLink(program); + + glUseProgram(program); + + assert(glIsProgram(program)); + assert(glIsShader(fragShader)); + assert(glIsShader(vertShader)); + + checkError(__LINE__); + {/*texture*/ + GLuint texLoc = glGetUniformLocationARB(program, "srcTex"); + glUniform1iARB(texLoc, 0); + } + {/*setup offsets */ + float offsets[] = { 1.0 / texture.width, 1.0 / texture.height, + 0.0 , 1.0 / texture.height, + -1.0 / texture.width, 1.0 / texture.height, + 1.0 / texture.width, 0.0, + 0.0 , 0.0, + -1.0 / texture.width, 0.0, + 1.0 / texture.width, -1.0 / texture.height, + 0.0 , -1.0 / texture.height, + -1.0 / texture.width, -1.0 / texture.height }; + GLuint offsetLoc = glGetUniformLocationARB(program, "Offset"); + glUniform2fv(offsetLoc, 9, offsets); + } + setupConvolution(); + + checkError(__LINE__); +} + + +static void readTexture(const char *filename) +{ + GLubyte *data; + + texture.x = 0; + texture.y = 0; + + glGenTextures(1, &texture.id); + glBindTexture(GL_TEXTURE_2D, texture.id); + glTexParameteri(GL_TEXTURE_2D, + GL_TEXTURE_MIN_FILTER, GL_NEAREST); + glTexParameteri(GL_TEXTURE_2D, + GL_TEXTURE_MAG_FILTER, GL_NEAREST); + glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP); + glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP); + glTexEnvi(GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, GL_REPLACE); + data = LoadRGBImage(filename, &texture.width, &texture.height, + &texture.format); + if (!data) { + printf("Error: couldn't load texture image '%s'\n", filename); + exit(1); + } + printf("Texture %s (%d x %d)\n", + filename, texture.width, texture.height); + glTexImage2D(GL_TEXTURE_2D, 0, GL_RGB, + texture.width, texture.height, 0, texture.format, + GL_UNSIGNED_BYTE, data); +} + +static void menuSelected(int entry) +{ + switch (entry) { + case QUIT: + exit(0); + break; + default: + filter = (enum Filter)entry; + } + setupConvolution(); + + glutPostRedisplay(); +} + +static void menuInit() +{ + menuId = glutCreateMenu(menuSelected); + + glutAddMenuEntry("Gaussian blur", GAUSSIAN_BLUR); + glutAddMenuEntry("Sharpen", SHARPEN); + glutAddMenuEntry("Mean removal", MEAN_REMOVAL); + glutAddMenuEntry("Emboss", EMBOSS); + glutAddMenuEntry("None", NO_FILTER); + + glutAddMenuEntry("Quit", QUIT); + + glutAttachMenu(GLUT_RIGHT_BUTTON); +} + +static void init() +{ + fprintf(stderr, "GL_RENDERER = %s\n", (char *) glGetString(GL_RENDERER)); + fprintf(stderr, "GL_VERSION = %s\n", (char *) glGetString(GL_VERSION)); + fprintf(stderr, "GL_VENDOR = %s\n", (char *) glGetString(GL_VENDOR)); + + menuInit(); + readTexture(textureLocation); + createProgram("convolution.vert", "convolution.frag"); + + glEnable(GL_TEXTURE_2D); + glClearColor(1.0, 1.0, 1.0, 1.0); + /*glShadeModel(GL_SMOOTH);*/ + glShadeModel(GL_FLAT); +} + +static void reshape(int width, int height) +{ + glViewport(0, 0, width, height); + glMatrixMode(GL_PROJECTION); + glLoadIdentity(); + box.minx = 0; + box.maxx = width; + box.miny = 0; + box.maxy = height; + box.minz = 0; + box.maxz = 1; + glOrtho(box.minx, box.maxx, box.miny, box.maxy, -999999, 999999); + glMatrixMode(GL_MODELVIEW); +} + +static void keyPress(unsigned char key, int x, int y) +{ + switch(key) { + case 27: + exit(0); + default: + return; + } + glutPostRedisplay(); +} + +static void +special(int k, int x, int y) +{ + switch (k) { + case GLUT_KEY_UP: + viewRotx += 2.0; + break; + case GLUT_KEY_DOWN: + viewRotx -= 2.0; + break; + case GLUT_KEY_LEFT: + viewRoty += 2.0; + break; + case GLUT_KEY_RIGHT: + viewRoty -= 2.0; + break; + default: + return; + } + glutPostRedisplay(); +} + + +static void draw() +{ + GLfloat center[2]; + GLfloat anchor[2]; + + glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT); + + glLoadIdentity(); + glPushMatrix(); + + center[0] = box.maxx/2; + center[1] = box.maxy/2; + anchor[0] = center[0] - texture.width/2; + anchor[1] = center[1] - texture.height/2; + + glTranslatef(center[0], center[1], 0); + glRotatef(viewRotx, 1.0, 0.0, 0.0); + glRotatef(viewRoty, 0.0, 1.0, 0.0); + glRotatef(viewRotz, 0.0, 0.0, 1.0); + glTranslatef(-center[0], -center[1], 0); + + glTranslatef(anchor[0], anchor[1], 0); + glBegin(GL_TRIANGLE_STRIP); + { + glColor3f(1., 0., 0.); + glTexCoord2f(0, 0); + glVertex3f(0, 0, 0); + + glColor3f(0., 1., 0.); + glTexCoord2f(0, 1.0); + glVertex3f(0, texture.height, 0); + + glColor3f(1., 0., 0.); + glTexCoord2f(1.0, 0); + glVertex3f(texture.width, 0, 0); + + glColor3f(0., 1., 0.); + glTexCoord2f(1, 1); + glVertex3f(texture.width, texture.height, 0); + } + glEnd(); + + glPopMatrix(); + + glFlush(); + + glutSwapBuffers(); +} + +int main(int argc, char **argv) +{ + glutInit(&argc, argv); + + glutInitWindowPosition(0, 0); + glutInitWindowSize(400, 400); + glutInitDisplayMode(GLUT_RGB | GLUT_ALPHA | GLUT_DOUBLE); + + if (!glutCreateWindow("Image Convolutions")) { + fprintf(stderr, "Couldn't create window!\n"); + exit(1); + } + + init(); + + glutReshapeFunc(reshape); + glutKeyboardFunc(keyPress); + glutSpecialFunc(special); + glutDisplayFunc(draw); + + + glutMainLoop(); + return 0; +} diff --git a/progs/glsl/texdemo1.c b/progs/glsl/texdemo1.c index d29ecf452b..3ceae14b96 100644 --- a/progs/glsl/texdemo1.c +++ b/progs/glsl/texdemo1.c @@ -48,6 +48,7 @@ static GLfloat TexXrot = 0, TexYrot = 0; static GLfloat Xrot = 20.0, Yrot = 20.0, Zrot = 0.0; static GLfloat EyeDist = 10; static GLboolean Anim = GL_TRUE; +static int win = 0; struct uniform_info { @@ -177,6 +178,7 @@ key(unsigned char k, int x, int y) EyeDist = 90; break; case 27: + glutDestroyWindow(win); exit(0); } glutPostRedisplay(); @@ -554,7 +556,7 @@ main(int argc, char *argv[]) glutInit(&argc, argv); glutInitWindowSize(500, 400); glutInitDisplayMode(GLUT_RGB | GLUT_DEPTH | GLUT_DOUBLE); - glutCreateWindow(Demo); + win = glutCreateWindow(Demo); glutReshapeFunc(Reshape); glutKeyboardFunc(key); glutSpecialFunc(specialkey); diff --git a/progs/miniglx/.gitignore b/progs/miniglx/.gitignore new file mode 100644 index 0000000000..f630f591f6 --- /dev/null +++ b/progs/miniglx/.gitignore @@ -0,0 +1,6 @@ +manytex +miniglxsample +miniglxtest +sample_server +sample_server2 +texline diff --git a/progs/osdemos/.gitignore b/progs/osdemos/.gitignore new file mode 100644 index 0000000000..5c78f12481 --- /dev/null +++ b/progs/osdemos/.gitignore @@ -0,0 +1,8 @@ +osdemo +osdemo16 +osdemo32 +ostest1 +readtex.c +readtex.h +showbuffer.c +showbuffer.h diff --git a/progs/samples/.gitignore b/progs/samples/.gitignore index 9635c7f450..f60d6e94ea 100644 --- a/progs/samples/.gitignore +++ b/progs/samples/.gitignore @@ -1,24 +1,41 @@ .cvsignore accum +anywin +bdemo +binfo bitmap1 bitmap2 blendeq blendxor +bugger copy cursor +demo depth eval +ffset fog font +font +incopy line logo +lthreads +lxdemo +lxgears +lxheads +lxinfo +lxpixmap nurb oglinfo olympic overlay +pend point prim quad +readtex.c +readtex.h select shape sphere @@ -28,20 +45,3 @@ stretch texture tri wave -bugger -pend -lthreads -lxdemo -lxgears -lxheads -lxinfo -lxpixmap -anywin -ffset -bdemo -binfo -incopy -demo -font -readtex.c -readtex.h diff --git a/progs/slang/.gitignore b/progs/slang/.gitignore new file mode 100644 index 0000000000..8a42b018e6 --- /dev/null +++ b/progs/slang/.gitignore @@ -0,0 +1,3 @@ +cltest +sotest +vstest diff --git a/progs/tests/.gitignore b/progs/tests/.gitignore index 2c0e16c35e..a72f552f10 100644 --- a/progs/tests/.gitignore +++ b/progs/tests/.gitignore @@ -1,48 +1,75 @@ .cvsignore -getproclist.h +afsmultiarb antialias arbfpspec arbfptest1 arbfptexture arbfptrig +arbnpot +arbnpot-mipmap arbvptest1 arbvptest3 arbvptorus arbvpwarpmesh +arraytexture blendminmax blendsquare bufferobj +bug_3050 bug_3101 bug_3195 +calibrate_rast +copypixrate crossbar cva dinoshade +drawbuffers +extfuncs.h fbotest1 +fbotest2 fbotexture +fillrate floattex fog fogcoord fptest1 fptexture getprocaddress +getproclist.h +interleave invert +jkrahntest manytex +minmag +mipmap_limits +mipmap_view multipal no_s3tc packedpixels pbo +prog_parameter projtex +random +readrate +readtex.c +readtex.h seccolor sharedtex -stencil_wrap stencilwrap +stencil_wrap +subtex +subtexrate tex1d texcmp +texcompress2 +texfilt texgenmix texline texobjshare texrect texwrap +vao-01 +vao-02 vparray vptest1 vptest2 @@ -51,4 +78,6 @@ vptorus vpwarpmesh yuvrect yuvsquare +zcomp +zdrawpix zreaddraw diff --git a/progs/tests/Makefile b/progs/tests/Makefile index 27e2eafdfc..75d1eca08e 100644 --- a/progs/tests/Makefile +++ b/progs/tests/Makefile @@ -84,6 +84,8 @@ SOURCES = \ vpwarpmesh.c \ yuvrect.c \ yuvsquare.c \ + zcomp.c \ + zdrawpix.c \ zreaddraw.c PROGS = $(SOURCES:%.c=%) diff --git a/progs/tests/manytex.c b/progs/tests/manytex.c index 61a1519ae6..900e5834fe 100644 --- a/progs/tests/manytex.c +++ b/progs/tests/manytex.c @@ -30,6 +30,7 @@ static GLboolean LinearFilter = GL_FALSE; static GLboolean RandomSize = GL_FALSE; static GLint Rows, Columns; static GLint LowPriorityCount = 0; +static GLint Win; static void Idle( void ) @@ -128,6 +129,14 @@ static int RandomInt(int min, int max) } +static void DeleteTextures(void) +{ + glDeleteTextures(NumTextures, TextureID); + free(TextureID); + TextureID = NULL; +} + + static void Init( void ) { @@ -306,9 +315,12 @@ static void Key( unsigned char key, int x, int y ) Zrot += step; break; case ' ': + DeleteTextures(); Init(); break; case 27: + DeleteTextures(); + glutDestroyWindow(Win); exit(0); break; } @@ -324,7 +336,7 @@ int main( int argc, char *argv[] ) glutInitWindowPosition( 0, 0 ); glutInitWindowSize( WinWidth, WinHeight ); glutInitDisplayMode( GLUT_RGB | GLUT_DOUBLE ); - glutCreateWindow(argv[0]); + Win = glutCreateWindow(argv[0]); glutReshapeFunc( Reshape ); glutKeyboardFunc( Key ); glutDisplayFunc( Display ); diff --git a/progs/tests/mipmap_limits.c b/progs/tests/mipmap_limits.c index d6d6e467b0..8bdad826f7 100644 --- a/progs/tests/mipmap_limits.c +++ b/progs/tests/mipmap_limits.c @@ -59,6 +59,18 @@ static GLfloat LodBias = 0.0; static GLboolean NearestFilter = GL_TRUE; +static void +InitValues(void) +{ + BaseLevel = 0; + MaxLevel = 8; + MinLod = -1; + MaxLod = 9; + LodBias = 0.0; + NearestFilter = GL_TRUE; +} + + static void MakeImage(int level, int width, int height, const GLubyte color[4]) { const int makeStripes = 0; @@ -108,6 +120,8 @@ static void makeImages(void) static void myinit(void) { + InitValues(); + glEnable(GL_DEPTH_TEST); glDepthFunc(GL_LESS); glShadeModel(GL_FLAT); @@ -125,7 +139,7 @@ static void myinit(void) static void display(void) { GLfloat tcm = 4.0; - printf("BASE_LEVEL = %d MAX_LEVEL = %d MIN_LOD = %f MAX_LOD = %f Bias = %.2g filter = %s\n", + printf("BASE_LEVEL=%d MAX_LEVEL=%d MIN_LOD=%.2g MAX_LOD=%.2g Bias=%.2g Filter=%s\n", BaseLevel, MaxLevel, MinLod, MaxLod, LodBias, NearestFilter ? "NEAREST" : "LINEAR"); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_BASE_LEVEL, BaseLevel); @@ -214,6 +228,9 @@ key(unsigned char k, int x, int y) case 'f': NearestFilter = !NearestFilter; break; + case ' ': + InitValues(); + break; case 27: /* Escape */ exit(0); break; @@ -227,12 +244,13 @@ key(unsigned char k, int x, int y) static void usage(void) { printf("usage:\n"); - printf(" b/B decrease/increase GL_TEXTURE_BASE_LEVEL\n"); - printf(" m/M decrease/increase GL_TEXTURE_MAX_LEVEL\n"); - printf(" n/N decrease/increase GL_TEXTURE_MIN_LOD\n"); - printf(" x/X decrease/increase GL_TEXTURE_MAX_LOD\n"); - printf(" l/L decrease/increase GL_TEXTURE_LOD_BIAS\n"); - printf(" f toggle nearest/linear filtering\n"); + printf(" b/B decrease/increase GL_TEXTURE_BASE_LEVEL\n"); + printf(" m/M decrease/increase GL_TEXTURE_MAX_LEVEL\n"); + printf(" n/N decrease/increase GL_TEXTURE_MIN_LOD\n"); + printf(" x/X decrease/increase GL_TEXTURE_MAX_LOD\n"); + printf(" l/L decrease/increase GL_TEXTURE_LOD_BIAS\n"); + printf(" f toggle nearest/linear filtering\n"); + printf(" SPACE reset values\n"); } diff --git a/progs/tests/zcomp.c b/progs/tests/zcomp.c new file mode 100644 index 0000000000..b53079d07f --- /dev/null +++ b/progs/tests/zcomp.c @@ -0,0 +1,223 @@ +/** + * Test Z compositing with glDrawPixels(GL_DEPTH_COMPONENT) and stencil test. + */ + +#define GL_GLEXT_PROTOTYPES +#include <stdio.h> +#include <stdlib.h> +#include <math.h> +#include <GL/glut.h> +#include "../util/showbuffer.c" + + +static int Win; +static GLfloat Xrot = 0, Yrot = 0, Zpos = 6; +static GLboolean Anim = GL_FALSE; + +static int Width = 400, Height = 200; +static GLfloat *Zimg; +static GLubyte *Cimg; +static GLboolean showZ = 0; + + +static void +Idle(void) +{ + Xrot += 3.0; + Yrot += 4.0; + glutPostRedisplay(); +} + + +/** + * Draw first object, save color+Z images + */ +static void +DrawFirst(void) +{ + static const GLfloat red[4] = { 1.0, 0.0, 0.0, 0.0 }; + + glMaterialfv(GL_FRONT_AND_BACK, GL_AMBIENT_AND_DIFFUSE, red); + + glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT); + + glPushMatrix(); + glTranslatef(-1, 0, 0); + glRotatef(45 + Xrot, 1, 0, 0); + + glutSolidTorus(0.75, 2.0, 10, 20); + + glPopMatrix(); + + glReadPixels(0, 0, Width, Height, GL_DEPTH_COMPONENT, GL_FLOAT, Zimg); + glReadPixels(0, 0, Width, Height, GL_RGBA, GL_UNSIGNED_BYTE, Cimg); +} + + +/** + * Draw second object. + */ +static void +DrawSecond(void) +{ + static const GLfloat blue[4] = { 0.0, 0.0, 1.0, 0.0 }; + + glMaterialfv(GL_FRONT_AND_BACK, GL_AMBIENT_AND_DIFFUSE, blue); + + glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT | GL_STENCIL_BUFFER_BIT); + + glPushMatrix(); + glTranslatef(+1, 0, 0); + glRotatef(-45 + Xrot, 1, 0, 0); + + glutSolidTorus(0.75, 2.0, 10, 20); + + glPopMatrix(); +} + + +/** + * Composite first/saved image over second rendering. + */ +static void +Composite(void) +{ + glWindowPos2i(0, 0); + + /* Draw Z values, set stencil where Z test passes */ + glEnable(GL_STENCIL_TEST); + glStencilFunc(GL_ALWAYS, 1, ~0); + glStencilOp(GL_KEEP, GL_KEEP, GL_REPLACE); + glColorMask(0,0,0,0); + glDrawPixels(Width, Height, GL_DEPTH_COMPONENT, GL_FLOAT, Zimg); + glColorMask(1,1,1,1); + + /* Draw color where stencil==1 */ + glStencilFunc(GL_EQUAL, 1, ~0); + glStencilOp(GL_KEEP, GL_KEEP, GL_KEEP); + glDisable(GL_DEPTH_TEST); + glDrawPixels(Width, Height, GL_RGBA, GL_UNSIGNED_BYTE, Cimg); + glEnable(GL_DEPTH_TEST); + + glDisable(GL_STENCIL_TEST); +} + + +static void +Draw(void) +{ + DrawFirst(); + DrawSecond(); + Composite(); + glutSwapBuffers(); +} + + +static void +Reshape(int width, int height) +{ + GLfloat ar = (float) width / height; + glViewport(0, 0, width, height); + glMatrixMode(GL_PROJECTION); + glLoadIdentity(); + glFrustum(-ar, ar, -1.0, 1.0, 5.0, 30.0); + glMatrixMode(GL_MODELVIEW); + glLoadIdentity(); + glTranslatef(0.0, 0.0, -15.0); + + Width = width; + Height = height; + + if (Zimg) + free(Zimg); + if (Cimg) + free(Cimg); + Zimg = (float *) malloc(width * height * 4); + Cimg = (GLubyte *) malloc(width * height * 4); +} + + +static void +Key(unsigned char key, int x, int y) +{ + const GLfloat step = 1.0; + (void) x; + (void) y; + switch (key) { + case 'a': + Anim = !Anim; + if (Anim) + glutIdleFunc(Idle); + else + glutIdleFunc(NULL); + break; + case 'd': + showZ = !showZ; + break; + case 'z': + Zpos -= step; + break; + case 'Z': + Zpos += step; + break; + case 27: + glutDestroyWindow(Win); + exit(0); + break; + } + glutPostRedisplay(); +} + + +static void +SpecialKey(int key, int x, int y) +{ + const GLfloat step = 3.0; + (void) x; + (void) y; + switch (key) { + case GLUT_KEY_UP: + Xrot -= step; + break; + case GLUT_KEY_DOWN: + Xrot += step; + break; + case GLUT_KEY_LEFT: + Yrot -= step; + break; + case GLUT_KEY_RIGHT: + Yrot += step; + break; + } + glutPostRedisplay(); +} + + +static void +Init(void) +{ + /* setup lighting, etc */ + glEnable(GL_DEPTH_TEST); + glEnable(GL_LIGHTING); + glEnable(GL_LIGHT0); +} + + +int +main(int argc, char *argv[]) +{ + glutInit(&argc, argv); + glutInitWindowPosition(0, 0); + glutInitWindowSize(Width, Height); + glutInitDisplayMode(GLUT_RGB | GLUT_DOUBLE | GLUT_DEPTH | GLUT_STENCIL); + Win = glutCreateWindow(argv[0]); + glutReshapeFunc(Reshape); + glutKeyboardFunc(Key); + glutSpecialFunc(SpecialKey); + glutDisplayFunc(Draw); + if (Anim) + glutIdleFunc(Idle); + Init(); + glutMainLoop(); + return 0; +} diff --git a/progs/tests/zdrawpix.c b/progs/tests/zdrawpix.c new file mode 100644 index 0000000000..dd222e7dd0 --- /dev/null +++ b/progs/tests/zdrawpix.c @@ -0,0 +1,192 @@ +/** + * Test glDrawPixels(GL_DEPTH_COMPONENT) + * + * We load a window-sized buffer of Z values so that Z=1 at the top and + * Z=0 at the bottom (and interpolate between). + * We draw that image into the Z buffer, then draw an ordinary cube. + * The bottom part of the cube should be "clipped" where the cube fails + * the Z test. + * + * Press 'd' to view the Z buffer as a grayscale image. + */ + +#define GL_GLEXT_PROTOTYPES +#include <stdio.h> +#include <stdlib.h> +#include <math.h> +#include <GL/glut.h> +#include "../util/showbuffer.c" + + +static int Win; +static GLfloat Xrot = 50, Yrot = 40, Zpos = 6; +static GLboolean Anim = GL_FALSE; + +static int Width = 200, Height = 200; +static GLfloat *z; +static GLboolean showZ = 0; + + +static void +Idle(void) +{ + Xrot += 3.0; + Yrot += 4.0; + glutPostRedisplay(); +} + + +static void +Draw(void) +{ + glClearColor(0, 0, 0.5, 0); + glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT); + +#if 1 + glColor3f(1, 0, 0); + glWindowPos2i(0,0); + glColorMask(0,0,0,0); + glDrawPixels(Width, Height, GL_DEPTH_COMPONENT, GL_FLOAT, z); +#elif 0 + glPushMatrix(); + glTranslatef(-0.75, 0, Zpos); + glutSolidSphere(1.0, 20, 10); + glPopMatrix(); +#endif + glColorMask(1,1,1,1); + + /* draw cube */ + glPushMatrix(); + glTranslatef(0, 0, Zpos); + glRotatef(Xrot, 1, 0, 0); + glRotatef(Yrot, 0, 1, 0); + glutSolidCube(2.0); + glPopMatrix(); + +#if 0 + /* drawpixels after cube */ + glColor3f(1, 0, 0); + glWindowPos2i(0,0); + //glColorMask(0,0,0,0); + glDrawPixels(Width, Height, GL_DEPTH_COMPONENT, GL_FLOAT, z); +#endif + + if (showZ) { + ShowDepthBuffer(Width, Height, 0.0, 1.0); + } + + glutSwapBuffers(); +} + + +static void +Reshape(int width, int height) +{ + glViewport(0, 0, width, height); + glMatrixMode(GL_PROJECTION); + glLoadIdentity(); + glFrustum(-1.0, 1.0, -1.0, 1.0, 5.0, 30.0); + glMatrixMode(GL_MODELVIEW); + glLoadIdentity(); + glTranslatef(0.0, 0.0, -15.0); + + Width = width; + Height = height; + + z = (float *) malloc(width * height * 4); + { + int i, j, k = 0; + for (i = 0; i < height; i++) { + float zval = (float) i / (height - 1); + for (j = 0; j < width; j++) { + z[k++] = zval; + } + } + } +} + + +static void +Key(unsigned char key, int x, int y) +{ + const GLfloat step = 1.0; + (void) x; + (void) y; + switch (key) { + case 'a': + Anim = !Anim; + if (Anim) + glutIdleFunc(Idle); + else + glutIdleFunc(NULL); + break; + case 'd': + showZ = !showZ; + break; + case 'z': + Zpos -= step; + break; + case 'Z': + Zpos += step; + break; + case 27: + glutDestroyWindow(Win); + exit(0); + break; + } + glutPostRedisplay(); +} + + +static void +SpecialKey(int key, int x, int y) +{ + const GLfloat step = 3.0; + (void) x; + (void) y; + switch (key) { + case GLUT_KEY_UP: + Xrot -= step; + break; + case GLUT_KEY_DOWN: + Xrot += step; + break; + case GLUT_KEY_LEFT: + Yrot -= step; + break; + case GLUT_KEY_RIGHT: + Yrot += step; + break; + } + glutPostRedisplay(); +} + + +static void +Init(void) +{ + /* setup lighting, etc */ + glEnable(GL_DEPTH_TEST); + glEnable(GL_LIGHTING); + glEnable(GL_LIGHT0); +} + + +int +main(int argc, char *argv[]) +{ + glutInit(&argc, argv); + glutInitWindowPosition(0, 0); + glutInitWindowSize(400, 400); + glutInitDisplayMode(GLUT_RGB | GLUT_DOUBLE | GLUT_DEPTH); + Win = glutCreateWindow(argv[0]); + glutReshapeFunc(Reshape); + glutKeyboardFunc(Key); + glutSpecialFunc(SpecialKey); + glutDisplayFunc(Draw); + if (Anim) + glutIdleFunc(Idle); + Init(); + glutMainLoop(); + return 0; +} diff --git a/progs/tools/trace/.gitignore b/progs/tools/trace/.gitignore new file mode 100644 index 0000000000..afe0c5829e --- /dev/null +++ b/progs/tools/trace/.gitignore @@ -0,0 +1 @@ +gltrace.cc diff --git a/progs/trivial/.gitignore b/progs/trivial/.gitignore index aafc4a7291..30f03e9103 100644 --- a/progs/trivial/.gitignore +++ b/progs/trivial/.gitignore @@ -14,6 +14,7 @@ fs-tri line line-clip line-cull +line-smooth line-stipple-wide line-userclip line-userclip-clip @@ -33,6 +34,7 @@ point-clip point-param point-sprite point-wide +point-wide-smooth poly poly-flat poly-unfilled @@ -77,6 +79,7 @@ tri-flat tri-flat-clip tri-fog tri-mask-tri +tri-orig tri-query tri-scissor-tri tri-stencil @@ -84,6 +87,7 @@ tri-tex-3d tri-tri tri-unfilled tri-unfilled-clip +tri-unfilled-edgeflag tri-unfilled-smooth tri-unfilled-userclip tri-unfilled-userclip-stip diff --git a/progs/trivial/Makefile b/progs/trivial/Makefile index 8bc6ad4e38..1dbc4721e8 100644 --- a/progs/trivial/Makefile +++ b/progs/trivial/Makefile @@ -45,6 +45,7 @@ SOURCES = \ point-param.c \ point-sprite.c \ point-wide.c \ + point-wide-smooth.c \ point.c \ poly-flat.c \ poly-unfilled.c \ @@ -69,6 +70,7 @@ SOURCES = \ quadstrip-cont.c \ quadstrip-flat.c \ quadstrip.c \ + tri-orig.c \ tri-alpha.c \ tri-blend-color.c \ tri-blend-max.c \ @@ -92,6 +94,7 @@ SOURCES = \ tri-stencil.c \ tri-tex-3d.c \ tri-tri.c \ + tri-unfilled-edgeflag.c \ tri-unfilled-clip.c \ tri-unfilled-smooth.c \ tri-unfilled-userclip-stip.c \ diff --git a/progs/trivial/point-wide-smooth.c b/progs/trivial/point-wide-smooth.c new file mode 100644 index 0000000000..55cbcaec70 --- /dev/null +++ b/progs/trivial/point-wide-smooth.c @@ -0,0 +1,145 @@ +/* + * Copyright (c) 1991, 1992, 1993 Silicon Graphics, Inc. + * + * Permission to use, copy, modify, distribute, and sell this software and + * its documentation for any purpose is hereby granted without fee, provided + * that (i) the above copyright notices and this permission notice appear in + * all copies of the software and related documentation, and (ii) the name of + * Silicon Graphics may not be used in any advertising or + * publicity relating to the software without the specific, prior written + * permission of Silicon Graphics. + * + * THE SOFTWARE IS PROVIDED "AS-IS" AND WITHOUT WARRANTY OF + * ANY KIND, + * EXPRESS, IMPLIED OR OTHERWISE, INCLUDING WITHOUT LIMITATION, ANY + * WARRANTY OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. + * + * IN NO EVENT SHALL SILICON GRAPHICS BE LIABLE FOR + * ANY SPECIAL, INCIDENTAL, INDIRECT OR CONSEQUENTIAL DAMAGES OF ANY KIND, + * OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, + * WHETHER OR NOT ADVISED OF THE POSSIBILITY OF DAMAGE, AND ON ANY THEORY OF + * LIABILITY, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE + * OF THIS SOFTWARE. + */ + +#include <stdio.h> +#include <string.h> +#include <stdlib.h> +#include <GL/glut.h> + + +#define CI_OFFSET_1 16 +#define CI_OFFSET_2 32 + + +GLenum doubleBuffer; + + +static void Init(void) +{ + fprintf(stderr, "GL_RENDERER = %s\n", (char *) glGetString(GL_RENDERER)); + fprintf(stderr, "GL_VERSION = %s\n", (char *) glGetString(GL_VERSION)); + fprintf(stderr, "GL_VENDOR = %s\n", (char *) glGetString(GL_VENDOR)); + + glClearColor(0.0, 0.0, 1.0, 0.0); +} + +static void Reshape(int width, int height) +{ + + glViewport(0, 0, (GLint)width, (GLint)height); + + glMatrixMode(GL_PROJECTION); + glLoadIdentity(); + glOrtho(-1.0, 1.0, -1.0, 1.0, -0.5, 1000.0); + glMatrixMode(GL_MODELVIEW); +} + +static void Key(unsigned char key, int x, int y) +{ + + switch (key) { + case 27: + exit(1); + default: + return; + } + + glutPostRedisplay(); +} + +static void Draw(void) +{ + glClear(GL_COLOR_BUFFER_BIT); + + glPointSize(8.0); + + glEnable(GL_POINT_SMOOTH); + glEnable(GL_BLEND); + glBlendFunc(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA); + + glBegin(GL_POINTS); + glColor3f(1,0,0); + glVertex3f( 0.9, -0.9, -30.0); + glColor3f(1,1,0); + glVertex3f( 0.9, 0.9, -30.0); + glColor3f(1,0,1); + glVertex3f(-0.9, 0.9, -30.0); + glColor3f(0,1,1); + glVertex3f(-0.9, -0.9, -30.0); + glEnd(); + + glFlush(); + + if (doubleBuffer) { + glutSwapBuffers(); + } +} + +static GLenum Args(int argc, char **argv) +{ + GLint i; + + doubleBuffer = GL_FALSE; + + for (i = 1; i < argc; i++) { + if (strcmp(argv[i], "-sb") == 0) { + doubleBuffer = GL_FALSE; + } else if (strcmp(argv[i], "-db") == 0) { + doubleBuffer = GL_TRUE; + } else { + fprintf(stderr, "%s (Bad option).\n", argv[i]); + return GL_FALSE; + } + } + return GL_TRUE; +} + +int main(int argc, char **argv) +{ + GLenum type; + + glutInit(&argc, argv); + + if (Args(argc, argv) == GL_FALSE) { + exit(1); + } + + glutInitWindowPosition(0, 0); glutInitWindowSize( 250, 250); + + type = GLUT_RGB | GLUT_ALPHA; + type |= (doubleBuffer) ? GLUT_DOUBLE : GLUT_SINGLE; + glutInitDisplayMode(type); + + if (glutCreateWindow("First Tri") == GL_FALSE) { + exit(1); + } + + Init(); + + glutReshapeFunc(Reshape); + glutKeyboardFunc(Key); + glutDisplayFunc(Draw); + glutMainLoop(); + return 0; +} diff --git a/progs/trivial/point-wide.c b/progs/trivial/point-wide.c index d1038ecfa1..265b20097b 100644 --- a/progs/trivial/point-wide.c +++ b/progs/trivial/point-wide.c @@ -33,7 +33,6 @@ GLenum doubleBuffer; -static GLboolean smooth = GL_FALSE; static void Init(void) @@ -60,9 +59,6 @@ static void Key(unsigned char key, int x, int y) { switch (key) { - case 's': - smooth = !smooth; - break; case 27: exit(1); default: @@ -75,19 +71,10 @@ static void Key(unsigned char key, int x, int y) static void Draw(void) { glClear(GL_COLOR_BUFFER_BIT); + glDisable(GL_DEPTH_TEST); glPointSize(8.0); - if (smooth) { - glEnable(GL_POINT_SMOOTH); - glEnable(GL_BLEND); - glBlendFunc(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA); - } - else { - glDisable(GL_POINT_SMOOTH); - glDisable(GL_BLEND); - } - glBegin(GL_POINTS); glColor3f(1,0,0); glVertex3f( 0.9, -0.9, -30.0); diff --git a/progs/trivial/quad-tex-2d.c b/progs/trivial/quad-tex-2d.c index 3a6e5237ed..97c9fc54d8 100644 --- a/progs/trivial/quad-tex-2d.c +++ b/progs/trivial/quad-tex-2d.c @@ -31,6 +31,7 @@ static GLenum Target = GL_TEXTURE_2D; static GLenum Filter = GL_NEAREST; GLenum doubleBuffer; static float Rot = 0; +static int win = 0; static void Init(void) { @@ -88,6 +89,7 @@ static void Reshape(int width, int height) glFrustum(-1, 1, -1, 1, 10, 20); #endif glMatrixMode(GL_MODELVIEW); + glLoadIdentity(); glTranslatef(0, 0, -15); } @@ -101,6 +103,7 @@ static void Key(unsigned char key, int x, int y) Rot -= 10.0; break; case 27: + glutDestroyWindow(win); exit(0); default: return; @@ -170,7 +173,8 @@ int main(int argc, char **argv) type |= (doubleBuffer) ? GLUT_DOUBLE : GLUT_SINGLE; glutInitDisplayMode(type); - if (glutCreateWindow("First Tri") == GL_FALSE) { + win = glutCreateWindow("First Tri"); + if (!win) { exit(1); } diff --git a/progs/trivial/tri-orig.c b/progs/trivial/tri-orig.c new file mode 100644 index 0000000000..d5e6742ce1 --- /dev/null +++ b/progs/trivial/tri-orig.c @@ -0,0 +1,136 @@ +/* + * Copyright (c) 1991, 1992, 1993 Silicon Graphics, Inc. + * + * Permission to use, copy, modify, distribute, and sell this software and + * its documentation for any purpose is hereby granted without fee, provided + * that (i) the above copyright notices and this permission notice appear in + * all copies of the software and related documentation, and (ii) the name of + * Silicon Graphics may not be used in any advertising or + * publicity relating to the software without the specific, prior written + * permission of Silicon Graphics. + * + * THE SOFTWARE IS PROVIDED "AS-IS" AND WITHOUT WARRANTY OF + * ANY KIND, + * EXPRESS, IMPLIED OR OTHERWISE, INCLUDING WITHOUT LIMITATION, ANY + * WARRANTY OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. + * + * IN NO EVENT SHALL SILICON GRAPHICS BE LIABLE FOR + * ANY SPECIAL, INCIDENTAL, INDIRECT OR CONSEQUENTIAL DAMAGES OF ANY KIND, + * OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, + * WHETHER OR NOT ADVISED OF THE POSSIBILITY OF DAMAGE, AND ON ANY THEORY OF + * LIABILITY, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE + * OF THIS SOFTWARE. + */ + +#include <stdio.h> +#include <string.h> +#include <stdlib.h> +#include <GL/glut.h> + + +#define CI_OFFSET_1 16 +#define CI_OFFSET_2 32 + + +GLenum doubleBuffer; + +static void Init(void) +{ + fprintf(stderr, "GL_RENDERER = %s\n", (char *) glGetString(GL_RENDERER)); + fprintf(stderr, "GL_VERSION = %s\n", (char *) glGetString(GL_VERSION)); + fprintf(stderr, "GL_VENDOR = %s\n", (char *) glGetString(GL_VENDOR)); + + glClearColor(0.0, 0.0, 1.0, 0.0); +} + +static void Reshape(int width, int height) +{ + + glViewport(0, 0, (GLint)width, (GLint)height); + + glMatrixMode(GL_PROJECTION); + glLoadIdentity(); +/* glOrtho(-1.0, 1.0, -1.0, 1.0, -0.5, 1000.0); */ + glMatrixMode(GL_MODELVIEW); +} + +static void Key(unsigned char key, int x, int y) +{ + + switch (key) { + case 27: + exit(1); + default: + return; + } + + glutPostRedisplay(); +} + +static void Draw(void) +{ + glClear(GL_COLOR_BUFFER_BIT); + + glBegin(GL_TRIANGLES); + glColor3f(0,0,.7); + glVertex3f( 0.9, -0.9, -0.0); + glColor3f(.8,0,0); + glVertex3f( 0.9, 0.9, -0.0); + glColor3f(0,.9,0); + glVertex3f(-0.9, 0.0, -0.0); + glEnd(); + + glFlush(); + + if (doubleBuffer) { + glutSwapBuffers(); + } +} + +static GLenum Args(int argc, char **argv) +{ + GLint i; + + doubleBuffer = GL_FALSE; + + for (i = 1; i < argc; i++) { + if (strcmp(argv[i], "-sb") == 0) { + doubleBuffer = GL_FALSE; + } else if (strcmp(argv[i], "-db") == 0) { + doubleBuffer = GL_TRUE; + } else { + fprintf(stderr, "%s (Bad option).\n", argv[i]); + return GL_FALSE; + } + } + return GL_TRUE; +} + +int main(int argc, char **argv) +{ + GLenum type; + + glutInit(&argc, argv); + + if (Args(argc, argv) == GL_FALSE) { + exit(1); + } + + glutInitWindowPosition(0, 0); glutInitWindowSize( 250, 250); + + type = GLUT_RGB | GLUT_ALPHA; + type |= (doubleBuffer) ? GLUT_DOUBLE : GLUT_SINGLE; + glutInitDisplayMode(type); + + if (glutCreateWindow("First Tri") == GL_FALSE) { + exit(1); + } + + Init(); + + glutReshapeFunc(Reshape); + glutKeyboardFunc(Key); + glutDisplayFunc(Draw); + glutMainLoop(); + return 0; +} diff --git a/progs/trivial/tri-unfilled-edgeflag.c b/progs/trivial/tri-unfilled-edgeflag.c new file mode 100644 index 0000000000..31f41f6220 --- /dev/null +++ b/progs/trivial/tri-unfilled-edgeflag.c @@ -0,0 +1,141 @@ +/* + * Copyright (c) 1991, 1992, 1993 Silicon Graphics, Inc. + * + * Permission to use, copy, modify, distribute, and sell this software and + * its documentation for any purpose is hereby granted without fee, provided + * that (i) the above copyright notices and this permission notice appear in + * all copies of the software and related documentation, and (ii) the name of + * Silicon Graphics may not be used in any advertising or + * publicity relating to the software without the specific, prior written + * permission of Silicon Graphics. + * + * THE SOFTWARE IS PROVIDED "AS-IS" AND WITHOUT WARRANTY OF + * ANY KIND, + * EXPRESS, IMPLIED OR OTHERWISE, INCLUDING WITHOUT LIMITATION, ANY + * WARRANTY OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. + * + * IN NO EVENT SHALL SILICON GRAPHICS BE LIABLE FOR + * ANY SPECIAL, INCIDENTAL, INDIRECT OR CONSEQUENTIAL DAMAGES OF ANY KIND, + * OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, + * WHETHER OR NOT ADVISED OF THE POSSIBILITY OF DAMAGE, AND ON ANY THEORY OF + * LIABILITY, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE + * OF THIS SOFTWARE. + */ + +#include <stdio.h> +#include <string.h> +#include <stdlib.h> +#include <GL/glut.h> + + +#define CI_OFFSET_1 16 +#define CI_OFFSET_2 32 + + +GLenum doubleBuffer; + +static void Init(void) +{ + fprintf(stderr, "GL_RENDERER = %s\n", (char *) glGetString(GL_RENDERER)); + fprintf(stderr, "GL_VERSION = %s\n", (char *) glGetString(GL_VERSION)); + fprintf(stderr, "GL_VENDOR = %s\n", (char *) glGetString(GL_VENDOR)); + + glClearColor(0.0, 0.0, 1.0, 0.0); +} + +static void Reshape(int width, int height) +{ + + glViewport(0, 0, (GLint)width, (GLint)height); + + glMatrixMode(GL_PROJECTION); + glLoadIdentity(); +/* glOrtho(-1.0, 1.0, -1.0, 1.0, -0.5, 1000.0); */ + glMatrixMode(GL_MODELVIEW); +} + +static void Key(unsigned char key, int x, int y) +{ + + switch (key) { + case 27: + exit(1); + default: + return; + } + + glutPostRedisplay(); +} + +static void Draw(void) +{ + glClear(GL_COLOR_BUFFER_BIT); + glPolygonMode(GL_FRONT, GL_LINE); + glPolygonMode(GL_BACK, GL_LINE); + + glBegin(GL_TRIANGLES); + glEdgeFlag(1); + glColor3f(0,0,.7); + glVertex3f( 0.9, -0.9, -0.0); + glEdgeFlag(0); + glColor3f(.8,0,0); + glVertex3f( 0.9, 0.9, -0.0); + glEdgeFlag(1); + glColor3f(0,.9,0); + glVertex3f(-0.9, 0.0, -0.0); + glEnd(); + + glFlush(); + + if (doubleBuffer) { + glutSwapBuffers(); + } +} + +static GLenum Args(int argc, char **argv) +{ + GLint i; + + doubleBuffer = GL_FALSE; + + for (i = 1; i < argc; i++) { + if (strcmp(argv[i], "-sb") == 0) { + doubleBuffer = GL_FALSE; + } else if (strcmp(argv[i], "-db") == 0) { + doubleBuffer = GL_TRUE; + } else { + fprintf(stderr, "%s (Bad option).\n", argv[i]); + return GL_FALSE; + } + } + return GL_TRUE; +} + +int main(int argc, char **argv) +{ + GLenum type; + + glutInit(&argc, argv); + + if (Args(argc, argv) == GL_FALSE) { + exit(1); + } + + glutInitWindowPosition(0, 0); glutInitWindowSize( 250, 250); + + type = GLUT_RGB | GLUT_ALPHA; + type |= (doubleBuffer) ? GLUT_DOUBLE : GLUT_SINGLE; + glutInitDisplayMode(type); + + if (glutCreateWindow("First Tri") == GL_FALSE) { + exit(1); + } + + Init(); + + glutReshapeFunc(Reshape); + glutKeyboardFunc(Key); + glutDisplayFunc(Draw); + glutMainLoop(); + return 0; +} diff --git a/progs/util/showbuffer.c b/progs/util/showbuffer.c index 17f84dc62b..b0cedea217 100644 --- a/progs/util/showbuffer.c +++ b/progs/util/showbuffer.c @@ -71,6 +71,8 @@ ShowDepthBuffer( GLsizei winWidth, GLsizei winHeight, glPushMatrix(); glLoadIdentity(); + glViewport(0, 0, winWidth, winHeight); + glDisable(GL_STENCIL_TEST); glDisable(GL_DEPTH_TEST); glRasterPos2f(0, 0); @@ -120,6 +122,8 @@ ShowAlphaBuffer( GLsizei winWidth, GLsizei winHeight ) glPushMatrix(); glLoadIdentity(); + glViewport(0, 0, winWidth, winHeight); + glDisable(GL_STENCIL_TEST); glDisable(GL_DEPTH_TEST); glRasterPos2f(0, 0); @@ -170,6 +174,8 @@ ShowStencilBuffer( GLsizei winWidth, GLsizei winHeight, glPushMatrix(); glLoadIdentity(); + glViewport(0, 0, winWidth, winHeight); + glDisable(GL_STENCIL_TEST); glDisable(GL_DEPTH_TEST); glRasterPos2f(0, 0); diff --git a/progs/vp/.gitignore b/progs/vp/.gitignore new file mode 100644 index 0000000000..a5ff993525 --- /dev/null +++ b/progs/vp/.gitignore @@ -0,0 +1 @@ +vp-tris diff --git a/progs/xdemos/.gitignore b/progs/xdemos/.gitignore index 250bc17d98..34551101cb 100644 --- a/progs/xdemos/.gitignore +++ b/progs/xdemos/.gitignore @@ -13,6 +13,7 @@ offset overlay pbdemo pbinfo +sharedtex texture_from_pixmap wincopy xdemo diff --git a/progs/xdemos/Makefile b/progs/xdemos/Makefile index d1d7fecea1..896b160878 100644 --- a/progs/xdemos/Makefile +++ b/progs/xdemos/Makefile @@ -23,6 +23,7 @@ PROGS = glthreads \ overlay \ pbinfo \ pbdemo \ + sharedtex \ texture_from_pixmap \ wincopy \ xfont \ diff --git a/progs/xdemos/sharedtex.c b/progs/xdemos/sharedtex.c new file mode 100644 index 0000000000..81703fe0e7 --- /dev/null +++ b/progs/xdemos/sharedtex.c @@ -0,0 +1,324 @@ +/* + * Test sharing of texture objects by two rendering contexts. + * In particular, test that changing a texture object in one context + * effects the texture in the second context. + * + * Brian Paul + * 30 Apr 2008 + * + * Copyright (C) 2008 Brian Paul All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN + * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + */ + + +#include <GL/gl.h> +#include <GL/glx.h> +#include <assert.h> +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <unistd.h> +#include <X11/keysym.h> + + +#define MAX_CONTEXTS 2 + +#define TEX_SIZE 32 + +static const char *DisplayName = NULL; +static Display *Dpy; +static XVisualInfo *VisInfo; +static Window Win; +static GLXContext Contexts[MAX_CONTEXTS]; +static int WinWidth = 300, WinHeight = 300; + +static int DrawContext = 0, TexContext = 1; + +static GLuint TexObj = 0; +static GLboolean NewTexture = GL_FALSE; + + +static void +Error(const char *msg) +{ + fprintf(stderr, "sharedtex error: %s\n", msg); + exit(1); +} + + +static void +CreateWindow(const char *name) +{ + int attrib[] = { GLX_RGBA, + GLX_RED_SIZE, 1, + GLX_GREEN_SIZE, 1, + GLX_BLUE_SIZE, 1, + GLX_DOUBLEBUFFER, + None }; + int scrnum; + XSetWindowAttributes attr; + unsigned long mask; + Window root; + int xpos = 0, ypos = 0; + static int n = 0; + + scrnum = DefaultScreen(Dpy); + root = RootWindow(Dpy, scrnum); + + VisInfo = glXChooseVisual(Dpy, scrnum, attrib); + if (!VisInfo) { + Error("Unable to find RGB, double-buffered visual"); + } + + /* window attributes */ + xpos = (n % 10) * 100; + ypos = (n / 10) * 100; + n++; + + attr.background_pixel = 0; + attr.border_pixel = 0; + attr.colormap = XCreateColormap(Dpy, root, VisInfo->visual, AllocNone); + attr.event_mask = StructureNotifyMask | ExposureMask | KeyPressMask; + mask = CWBackPixel | CWBorderPixel | CWColormap | CWEventMask; + + Win = XCreateWindow(Dpy, root, xpos, ypos, WinWidth, WinHeight, + 0, VisInfo->depth, InputOutput, + VisInfo->visual, mask, &attr); + if (!Win) { + Error("Couldn't create window"); + } + + { + XSizeHints sizehints; + sizehints.x = xpos; + sizehints.y = ypos; + sizehints.width = WinWidth; + sizehints.height = WinHeight; + sizehints.flags = USSize | USPosition; + XSetNormalHints(Dpy, Win, &sizehints); + XSetStandardProperties(Dpy, Win, name, name, + None, (char **)NULL, 0, &sizehints); + } + + XMapWindow(Dpy, Win); +} + + +/** + * Change texture image, using TexContext + */ +static void +ModifyTexture(void) +{ + GLuint tex[TEX_SIZE][TEX_SIZE]; + GLuint c0, c1; + int i, j; + + if (Win && !glXMakeCurrent(Dpy, Win, Contexts[TexContext])) { + Error("glXMakeCurrent failed"); + } + + /* choose two random colors */ + c0 = rand() & 0xffffffff; + c1 = rand() & 0xffffffff; + + for (i = 0; i < TEX_SIZE; i++) { + for (j = 0; j < TEX_SIZE; j++) { + if (((i / 4) ^ (j / 4)) & 1) { + tex[i][j] = c0; + } + else { + tex[i][j] = c1; + } + } + } + + glBindTexture(GL_TEXTURE_2D, TexObj); + glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, TEX_SIZE, TEX_SIZE, 0, + GL_RGBA, GL_UNSIGNED_BYTE, tex); + + NewTexture = GL_TRUE; +} + + +static void +InitContext(void) +{ + glGenTextures(1, &TexObj); + assert(TexObj); + glBindTexture(GL_TEXTURE_2D, TexObj); + glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR); + glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR); + glEnable(GL_TEXTURE_2D); + + printf("GL_RENDERER = %s\n", (char*) glGetString(GL_RENDERER)); +} + + +static void +Setup(void) +{ + int i; + + Dpy = XOpenDisplay(DisplayName); + if (!Dpy) { + Error("Unable to open display"); + } + + CreateWindow("sharedtex"); + + for (i = 0; i < MAX_CONTEXTS; i++) { + GLXContext share = i > 0 ? Contexts[0] : 0; + + Contexts[i] = glXCreateContext(Dpy, VisInfo, share, True); + if (!Contexts[i]) { + Error("Unable to create GLX context"); + } + + if (!glXMakeCurrent(Dpy, Win, Contexts[i])) { + Error("glXMakeCurrent failed"); + } + + InitContext(); + } + + ModifyTexture(); +} + + +/** + * Redraw window, using DrawContext + */ +static void +Redraw(void) +{ + static float rot = 0.0; + float ar; + + rot += 1.0; + + if (Win && !glXMakeCurrent(Dpy, Win, Contexts[DrawContext])) { + Error("glXMakeCurrent failed"); + } + + glViewport(0, 0, WinWidth, WinHeight); + ar = (float) WinWidth / (float) WinHeight; + glMatrixMode(GL_PROJECTION); + glLoadIdentity(); + glOrtho(-ar, ar, -1.0, 1.0, -1.0, 1.0); + glMatrixMode(GL_MODELVIEW); + + glShadeModel(GL_FLAT); + glClearColor(0.5, 0.5, 0.5, 1.0); + glClear(GL_COLOR_BUFFER_BIT); + + glPushMatrix(); + glRotatef(rot, 0, 0, 1); + glScalef(0.7, 0.7, 0.7); + + if (NewTexture) { + /* rebind to get new contents */ + glBindTexture(GL_TEXTURE_2D, TexObj); + NewTexture = GL_FALSE; + } + + /* draw textured quad */ + glBegin(GL_POLYGON); + glTexCoord2f( 0.0, 0.0 ); glVertex2f( -1.0, -1.0 ); + glTexCoord2f( 1.0, 0.0 ); glVertex2f( 1.0, -1.0 ); + glTexCoord2f( 1.0, 1.0 ); glVertex2f( 1.0, 1.0 ); + glTexCoord2f( 0.0, 1.0 ); glVertex2f( -1.0, 1.0 ); + glEnd(); + + glPopMatrix(); + + if (Win) + glXSwapBuffers(Dpy, Win); +} + + +static void +EventLoop(void) +{ + while (1) { + while (XPending(Dpy) > 0) { + XEvent event; + XNextEvent(Dpy, &event); + + switch (event.type) { + case Expose: + Redraw(); + break; + case ConfigureNotify: + WinWidth = event.xconfigure.width; + WinHeight = event.xconfigure.height; + break; + case KeyPress: + { + char buf[100]; + KeySym keySym; + XComposeStatus stat; + XLookupString(&event.xkey, buf, sizeof(buf), &keySym, &stat); + switch (keySym) { + case XK_Escape: + exit(0); + break; + case XK_t: + case XK_T: + ModifyTexture(); + break; + default: + ; + } + } + Redraw(); + break; + default: + /*no-op*/ ; + } + } + + Redraw(); + usleep(10000); + } +} + + + + +int +main(int argc, char *argv[]) +{ + int i; + + for (i = 1; i < argc; i++) { + if (strcmp(argv[i], "-display") == 0 && i < argc) { + DisplayName = argv[i+1]; + i++; + } + } + + Setup(); + + printf("Press 't' to change texture image/colors\n"); + + EventLoop(); + + return 0; +} diff --git a/src/gallium/README.portability b/src/gallium/README.portability index ab0c197847..18a97f449b 100644 --- a/src/gallium/README.portability +++ b/src/gallium/README.portability @@ -24,6 +24,7 @@ headers in general, should stricly follow these guidelines to ensure * Don't use variable number of macro arguments. Use static inline functions instead. +* Don't use C99 features. = Standard Library = @@ -42,3 +43,67 @@ portable way. * Use the functions/macros in p_debug.h. * Don't include assert.h, call abort, printf, etc. + + += Code Style = + +== Inherantice in C == + +The main thing we do is mimic inheritance by structure containment. + +Here's a silly made-up example: + +/* base class */ +struct buffer +{ + int size; + void (*validate)(struct buffer *buf); +}; + +/* sub-class of bufffer */ +struct texture_buffer +{ + struct buffer base; /* the base class, MUST COME FIRST! */ + int format; + int width, height; +}; + + +Then, we'll typically have cast-wrapper functions to convert base-class +pointers to sub-class pointers where needed: + +static inline struct vertex_buffer *vertex_buffer(struct buffer *buf) +{ + return (struct vertex_buffer *) buf; +} + + +To create/init a sub-classed object: + +struct buffer *create_texture_buffer(int w, int h, int format) +{ + struct texture_buffer *t = malloc(sizeof(*t)); + t->format = format; + t->width = w; + t->height = h; + t->base.size = w * h; + t->base.validate = tex_validate; + return &t->base; +} + +Example sub-class method: + +void tex_validate(struct buffer *buf) +{ + struct texture_buffer *tb = texture_buffer(buf); + assert(tb->format); + assert(tb->width); + assert(tb->height); +} + + +Note that we typically do not use typedefs to make "class names"; we use +'struct whatever' everywhere. + +Gallium's pipe_context and the subclassed psb_context, etc are prime examples +of this. There's also many examples in Mesa and the Mesa state tracker. diff --git a/src/gallium/SConscript b/src/gallium/SConscript index f09778ce99..2653f91bd2 100644 --- a/src/gallium/SConscript +++ b/src/gallium/SConscript @@ -18,6 +18,7 @@ SConscript([ 'auxiliary/rtasm/SConscript', 'auxiliary/tgsi/SConscript', 'auxiliary/cso_cache/SConscript', + 'auxiliary/translate/SConscript', 'auxiliary/draw/SConscript', 'auxiliary/pipebuffer/SConscript', ]) diff --git a/src/gallium/auxiliary/cso_cache/cso_cache.c b/src/gallium/auxiliary/cso_cache/cso_cache.c index 18acab0967..36dc46ff80 100644 --- a/src/gallium/auxiliary/cso_cache/cso_cache.c +++ b/src/gallium/auxiliary/cso_cache/cso_cache.c @@ -43,6 +43,9 @@ struct cso_cache { struct cso_hash *rasterizer_hash; struct cso_hash *sampler_hash; int max_size; + + cso_sanitize_callback sanitize_cb; + void *sanitize_data; }; #if 1 @@ -205,9 +208,20 @@ static INLINE void delete_cso(void *state, enum cso_cache_type type) } } -static INLINE void sanitize_hash(struct cso_hash *hash, enum cso_cache_type type, + +static INLINE void sanitize_hash(struct cso_cache *sc, + struct cso_hash *hash, + enum cso_cache_type type, int max_size) { + if (sc->sanitize_cb) + sc->sanitize_cb(hash, type, max_size, sc->sanitize_data); +} + + +static INLINE void sanitize_cb(struct cso_hash *hash, enum cso_cache_type type, + int max_size, void *user_data) +{ /* if we're approach the maximum size, remove fourth of the entries * otherwise every subsequent call will go through the same */ int hash_size = cso_hash_size(hash); @@ -231,7 +245,7 @@ cso_insert_state(struct cso_cache *sc, void *state) { struct cso_hash *hash = _cso_hash_for_type(sc, type); - sanitize_hash(hash, type, sc->max_size); + sanitize_hash(sc, hash, type, sc->max_size); return cso_hash_insert(hash, hash_key, state); } @@ -255,9 +269,9 @@ void *cso_hash_find_data_from_template( struct cso_hash *hash, while (!cso_hash_iter_is_null(iter)) { void *iter_data = cso_hash_iter_data(iter); if (!memcmp(iter_data, templ, size)) { - /* Return the payload: + /* We found a match */ - return (unsigned char *)iter_data + size; + return iter_data; } iter = cso_hash_iter_next(iter); } @@ -290,6 +304,8 @@ void * cso_take_state(struct cso_cache *sc, struct cso_cache *cso_cache_create(void) { struct cso_cache *sc = MALLOC_STRUCT(cso_cache); + if (sc == NULL) + return NULL; sc->max_size = 4096; sc->blend_hash = cso_hash_create(); @@ -298,6 +314,8 @@ struct cso_cache *cso_cache_create(void) sc->rasterizer_hash = cso_hash_create(); sc->fs_hash = cso_hash_create(); sc->vs_hash = cso_hash_create(); + sc->sanitize_cb = sanitize_cb; + sc->sanitize_data = 0; return sc; } @@ -332,10 +350,10 @@ void cso_for_each_state(struct cso_cache *sc, enum cso_cache_type type, iter = cso_hash_first_node(hash); while (!cso_hash_iter_is_null(iter)) { void *state = cso_hash_iter_data(iter); + iter = cso_hash_iter_next(iter); if (state) { func(state, user_data); } - iter = cso_hash_iter_next(iter); } } @@ -363,13 +381,13 @@ void cso_set_maximum_cache_size(struct cso_cache *sc, int number) { sc->max_size = number; - sanitize_hash(sc->blend_hash, CSO_BLEND, sc->max_size); - sanitize_hash(sc->depth_stencil_hash, CSO_DEPTH_STENCIL_ALPHA, + sanitize_hash(sc, sc->blend_hash, CSO_BLEND, sc->max_size); + sanitize_hash(sc, sc->depth_stencil_hash, CSO_DEPTH_STENCIL_ALPHA, sc->max_size); - sanitize_hash(sc->fs_hash, CSO_FRAGMENT_SHADER, sc->max_size); - sanitize_hash(sc->vs_hash, CSO_VERTEX_SHADER, sc->max_size); - sanitize_hash(sc->rasterizer_hash, CSO_RASTERIZER, sc->max_size); - sanitize_hash(sc->sampler_hash, CSO_SAMPLER, sc->max_size); + sanitize_hash(sc, sc->fs_hash, CSO_FRAGMENT_SHADER, sc->max_size); + sanitize_hash(sc, sc->vs_hash, CSO_VERTEX_SHADER, sc->max_size); + sanitize_hash(sc, sc->rasterizer_hash, CSO_RASTERIZER, sc->max_size); + sanitize_hash(sc, sc->sampler_hash, CSO_SAMPLER, sc->max_size); } int cso_maximum_cache_size(const struct cso_cache *sc) @@ -377,3 +395,11 @@ int cso_maximum_cache_size(const struct cso_cache *sc) return sc->max_size; } +void cso_cache_set_sanitize_callback(struct cso_cache *sc, + cso_sanitize_callback cb, + void *user_data) +{ + sc->sanitize_cb = cb; + sc->sanitize_data = user_data; +} + diff --git a/src/gallium/auxiliary/cso_cache/cso_cache.h b/src/gallium/auxiliary/cso_cache/cso_cache.h index e5edbbb556..6b5c230e8f 100644 --- a/src/gallium/auxiliary/cso_cache/cso_cache.h +++ b/src/gallium/auxiliary/cso_cache/cso_cache.h @@ -84,8 +84,22 @@ extern "C" { #endif +enum cso_cache_type { + CSO_BLEND, + CSO_SAMPLER, + CSO_DEPTH_STENCIL_ALPHA, + CSO_RASTERIZER, + CSO_FRAGMENT_SHADER, + CSO_VERTEX_SHADER +}; + typedef void (*cso_state_callback)(void *ctx, void *obj); +typedef void (*cso_sanitize_callback)(struct cso_hash *hash, + enum cso_cache_type type, + int max_size, + void *user_data); + struct cso_cache; struct cso_blend { @@ -130,21 +144,15 @@ struct cso_sampler { struct pipe_context *context; }; - -enum cso_cache_type { - CSO_BLEND, - CSO_SAMPLER, - CSO_DEPTH_STENCIL_ALPHA, - CSO_RASTERIZER, - CSO_FRAGMENT_SHADER, - CSO_VERTEX_SHADER -}; - unsigned cso_construct_key(void *item, int item_size); struct cso_cache *cso_cache_create(void); void cso_cache_delete(struct cso_cache *sc); +void cso_cache_set_sanitize_callback(struct cso_cache *sc, + cso_sanitize_callback cb, + void *user_data); + struct cso_hash_iter cso_insert_state(struct cso_cache *sc, unsigned hash_key, enum cso_cache_type type, void *state); diff --git a/src/gallium/auxiliary/cso_cache/cso_context.c b/src/gallium/auxiliary/cso_cache/cso_context.c index 4a1a6cb79c..a1a3a9efaf 100644 --- a/src/gallium/auxiliary/cso_cache/cso_context.c +++ b/src/gallium/auxiliary/cso_cache/cso_context.c @@ -25,16 +25,20 @@ * **************************************************************************/ - /* Wrap the cso cache & hash mechanisms in a simplified + /** + * @file + * + * Wrap the cso cache & hash mechanisms in a simplified * pipe-driver-specific interface. * - * Authors: - * Zack Rusin <zack@tungstengraphics.com> - * Keith Whitwell <keith@tungstengraphics.com> + * @author Zack Rusin <zack@tungstengraphics.com> + * @author Keith Whitwell <keith@tungstengraphics.com> */ #include "pipe/p_state.h" #include "pipe/p_util.h" +#include "pipe/p_inlines.h" +#include "tgsi/util/tgsi_parse.h" #include "cso_cache/cso_context.h" #include "cso_cache/cso_cache.h" @@ -76,6 +80,131 @@ struct cso_context { }; +static boolean delete_blend_state(struct cso_context *ctx, void *state) +{ + struct cso_blend *cso = (struct cso_blend *)state; + + if (ctx->blend == cso->data) + return FALSE; + + if (cso->delete_state) + cso->delete_state(cso->context, cso->data); + FREE(state); + return TRUE; +} + +static boolean delete_depth_stencil_state(struct cso_context *ctx, void *state) +{ + struct cso_depth_stencil_alpha *cso = (struct cso_depth_stencil_alpha *)state; + + if (ctx->depth_stencil == cso->data) + return FALSE; + + if (cso->delete_state) + cso->delete_state(cso->context, cso->data); + FREE(state); + + return TRUE; +} + +static boolean delete_sampler_state(struct cso_context *ctx, void *state) +{ + struct cso_sampler *cso = (struct cso_sampler *)state; + if (cso->delete_state) + cso->delete_state(cso->context, cso->data); + FREE(state); + return TRUE; +} + +static boolean delete_rasterizer_state(struct cso_context *ctx, void *state) +{ + struct cso_rasterizer *cso = (struct cso_rasterizer *)state; + + if (ctx->rasterizer == cso->data) + return FALSE; + if (cso->delete_state) + cso->delete_state(cso->context, cso->data); + FREE(state); + return TRUE; +} + +static boolean delete_fs_state(struct cso_context *ctx, void *state) +{ + struct cso_fragment_shader *cso = (struct cso_fragment_shader *)state; + if (ctx->fragment_shader == cso->data) + return FALSE; + if (cso->delete_state) + cso->delete_state(cso->context, cso->data); + FREE(state); + return TRUE; +} + +static boolean delete_vs_state(struct cso_context *ctx, void *state) +{ + struct cso_vertex_shader *cso = (struct cso_vertex_shader *)state; + if (ctx->vertex_shader == cso->data) + return TRUE; + if (cso->delete_state) + cso->delete_state(cso->context, cso->data); + FREE(state); + return FALSE; +} + + +static INLINE boolean delete_cso(struct cso_context *ctx, + void *state, enum cso_cache_type type) +{ + switch (type) { + case CSO_BLEND: + return delete_blend_state(ctx, state); + break; + case CSO_SAMPLER: + return delete_sampler_state(ctx, state); + break; + case CSO_DEPTH_STENCIL_ALPHA: + return delete_depth_stencil_state(ctx, state); + break; + case CSO_RASTERIZER: + return delete_rasterizer_state(ctx, state); + break; + case CSO_FRAGMENT_SHADER: + return delete_fs_state(ctx, state); + break; + case CSO_VERTEX_SHADER: + return delete_vs_state(ctx, state); + break; + default: + assert(0); + FREE(state); + } + return FALSE; +} + +static INLINE void sanitize_hash(struct cso_hash *hash, enum cso_cache_type type, + int max_size, void *user_data) +{ + struct cso_context *ctx = (struct cso_context *)user_data; + /* if we're approach the maximum size, remove fourth of the entries + * otherwise every subsequent call will go through the same */ + int hash_size = cso_hash_size(hash); + int max_entries = (max_size > hash_size) ? max_size : hash_size; + int to_remove = (max_size < max_entries) * max_entries/4; + struct cso_hash_iter iter = cso_hash_first_node(hash); + if (hash_size > max_size) + to_remove += hash_size - max_size; + while (to_remove) { + /*remove elements until we're good */ + /*fixme: currently we pick the nodes to remove at random*/ + void *cso = cso_hash_iter_data(iter); + if (delete_cso(ctx, cso, type)) { + iter = cso_hash_erase(hash, iter); + --to_remove; + } else + iter = cso_hash_iter_next(iter); + } +} + + struct cso_context *cso_create_context( struct pipe_context *pipe ) { struct cso_context *ctx = CALLOC_STRUCT(cso_context); @@ -85,6 +214,9 @@ struct cso_context *cso_create_context( struct pipe_context *pipe ) ctx->cache = cso_cache_create(); if (ctx->cache == NULL) goto out; + cso_cache_set_sanitize_callback(ctx->cache, + sanitize_hash, + ctx); ctx->pipe = pipe; @@ -98,8 +230,14 @@ out: return NULL; } -static void cso_release_all( struct cso_context *ctx ) + +/** + * Prior to context destruction, this function unbinds all state objects. + */ +void cso_release_all( struct cso_context *ctx ) { + unsigned i; + if (ctx->pipe) { ctx->pipe->bind_blend_state( ctx->pipe, NULL ); ctx->pipe->bind_rasterizer_state( ctx->pipe, NULL ); @@ -109,6 +247,11 @@ static void cso_release_all( struct cso_context *ctx ) ctx->pipe->bind_vs_state( ctx->pipe, NULL ); } + for (i = 0; i < PIPE_MAX_SAMPLERS; i++) { + pipe_texture_reference(&ctx->textures[i], NULL); + pipe_texture_reference(&ctx->textures_saved[i], NULL); + } + if (ctx->cache) { cso_cache_delete( ctx->cache ); ctx->cache = NULL; @@ -118,10 +261,10 @@ static void cso_release_all( struct cso_context *ctx ) void cso_destroy_context( struct cso_context *ctx ) { - if (ctx) - cso_release_all( ctx ); - - FREE( ctx ); + if (ctx) { + //cso_release_all( ctx ); + FREE( ctx ); + } } @@ -135,8 +278,8 @@ void cso_destroy_context( struct cso_context *ctx ) * the data member of the cso to be the template itself. */ -void cso_set_blend(struct cso_context *ctx, - const struct pipe_blend_state *templ) +enum pipe_error cso_set_blend(struct cso_context *ctx, + const struct pipe_blend_state *templ) { unsigned hash_key = cso_construct_key((void*)templ, sizeof(struct pipe_blend_state)); struct cso_hash_iter iter = cso_find_state_template(ctx->cache, @@ -146,6 +289,8 @@ void cso_set_blend(struct cso_context *ctx, if (cso_hash_iter_is_null(iter)) { struct cso_blend *cso = MALLOC(sizeof(struct cso_blend)); + if (!cso) + return PIPE_ERROR_OUT_OF_MEMORY; cso->state = *templ; cso->data = ctx->pipe->create_blend_state(ctx->pipe, &cso->state); @@ -153,6 +298,11 @@ void cso_set_blend(struct cso_context *ctx, cso->context = ctx->pipe; iter = cso_insert_state(ctx->cache, hash_key, CSO_BLEND, cso); + if (cso_hash_iter_is_null(iter)) { + FREE(cso); + return PIPE_ERROR_OUT_OF_MEMORY; + } + handle = cso->data; } else { @@ -163,6 +313,7 @@ void cso_set_blend(struct cso_context *ctx, ctx->blend = handle; ctx->pipe->bind_blend_state(ctx->pipe, handle); } + return PIPE_OK; } void cso_save_blend(struct cso_context *ctx) @@ -182,12 +333,12 @@ void cso_restore_blend(struct cso_context *ctx) -void cso_single_sampler(struct cso_context *ctx, - unsigned idx, - const struct pipe_sampler_state *templ) +enum pipe_error cso_single_sampler(struct cso_context *ctx, + unsigned idx, + const struct pipe_sampler_state *templ) { void *handle = NULL; - + if (templ != NULL) { unsigned hash_key = cso_construct_key((void*)templ, sizeof(struct pipe_sampler_state)); struct cso_hash_iter iter = cso_find_state_template(ctx->cache, @@ -196,13 +347,20 @@ void cso_single_sampler(struct cso_context *ctx, if (cso_hash_iter_is_null(iter)) { struct cso_sampler *cso = MALLOC(sizeof(struct cso_sampler)); - + if (!cso) + return PIPE_ERROR_OUT_OF_MEMORY; + cso->state = *templ; cso->data = ctx->pipe->create_sampler_state(ctx->pipe, &cso->state); cso->delete_state = (cso_state_callback)ctx->pipe->delete_sampler_state; cso->context = ctx->pipe; iter = cso_insert_state(ctx->cache, hash_key, CSO_SAMPLER, cso); + if (cso_hash_iter_is_null(iter)) { + FREE(cso); + return PIPE_ERROR_OUT_OF_MEMORY; + } + handle = cso->data; } else { @@ -211,11 +369,12 @@ void cso_single_sampler(struct cso_context *ctx, } ctx->samplers[idx] = handle; + return PIPE_OK; } void cso_single_sampler_done( struct cso_context *ctx ) { - unsigned i; + unsigned i; /* find highest non-null sampler */ for (i = PIPE_MAX_SAMPLERS; i > 0; i--) { @@ -226,8 +385,8 @@ void cso_single_sampler_done( struct cso_context *ctx ) ctx->nr_samplers = i; if (ctx->hw.nr_samplers != ctx->nr_samplers || - memcmp(ctx->hw.samplers, - ctx->samplers, + memcmp(ctx->hw.samplers, + ctx->samplers, ctx->nr_samplers * sizeof(void *)) != 0) { memcpy(ctx->hw.samplers, ctx->samplers, ctx->nr_samplers * sizeof(void *)); @@ -237,22 +396,36 @@ void cso_single_sampler_done( struct cso_context *ctx ) } } -void cso_set_samplers( struct cso_context *ctx, - unsigned nr, - const struct pipe_sampler_state **templates ) +/* + * If the function encouters any errors it will return the + * last one. Done to always try to set as many samplers + * as possible. + */ +enum pipe_error cso_set_samplers( struct cso_context *ctx, + unsigned nr, + const struct pipe_sampler_state **templates ) { unsigned i; - + enum pipe_error temp, error = PIPE_OK; + /* TODO: fastpath */ - for (i = 0; i < nr; i++) - cso_single_sampler( ctx, i, templates[i] ); + for (i = 0; i < nr; i++) { + temp = cso_single_sampler( ctx, i, templates[i] ); + if (temp != PIPE_OK) + error = temp; + } + + for ( ; i < ctx->nr_samplers; i++) { + temp = cso_single_sampler( ctx, i, NULL ); + if (temp != PIPE_OK) + error = temp; + } - for ( ; i < ctx->nr_samplers; i++) - cso_single_sampler( ctx, i, NULL ); - cso_single_sampler_done( ctx ); + + return error; } void cso_save_samplers(struct cso_context *ctx) @@ -263,44 +436,64 @@ void cso_save_samplers(struct cso_context *ctx) void cso_restore_samplers(struct cso_context *ctx) { - cso_set_samplers(ctx, ctx->nr_samplers_saved, - (const struct pipe_sampler_state **) ctx->samplers_saved); + ctx->nr_samplers = ctx->nr_samplers_saved; + memcpy(ctx->samplers, ctx->samplers_saved, sizeof(ctx->samplers)); + cso_single_sampler_done( ctx ); } -void cso_set_sampler_textures( struct cso_context *ctx, - uint count, - struct pipe_texture **textures ) +enum pipe_error cso_set_sampler_textures( struct cso_context *ctx, + uint count, + struct pipe_texture **textures ) { uint i; ctx->nr_textures = count; for (i = 0; i < count; i++) - ctx->textures[i] = textures[i]; + pipe_texture_reference(&ctx->textures[i], textures[i]); for ( ; i < PIPE_MAX_SAMPLERS; i++) - ctx->textures[i] = NULL; + pipe_texture_reference(&ctx->textures[i], NULL); ctx->pipe->set_sampler_textures(ctx->pipe, count, textures); + + return PIPE_OK; } void cso_save_sampler_textures( struct cso_context *ctx ) { + uint i; + ctx->nr_textures_saved = ctx->nr_textures; - memcpy(ctx->textures_saved, ctx->textures, sizeof(ctx->textures)); + for (i = 0; i < ctx->nr_textures; i++) { + assert(!ctx->textures_saved[i]); + pipe_texture_reference(&ctx->textures_saved[i], ctx->textures[i]); + } } void cso_restore_sampler_textures( struct cso_context *ctx ) { - cso_set_sampler_textures(ctx, ctx->nr_textures_saved, ctx->textures_saved); + uint i; + + ctx->nr_textures = ctx->nr_textures_saved; + + for (i = 0; i < ctx->nr_textures; i++) { + pipe_texture_reference(&ctx->textures[i], NULL); + ctx->textures[i] = ctx->textures_saved[i]; + ctx->textures_saved[i] = NULL; + } + for ( ; i < PIPE_MAX_SAMPLERS; i++) + pipe_texture_reference(&ctx->textures[i], NULL); + + ctx->pipe->set_sampler_textures(ctx->pipe, ctx->nr_textures, ctx->textures); + ctx->nr_textures_saved = 0; } - -void cso_set_depth_stencil_alpha(struct cso_context *ctx, - const struct pipe_depth_stencil_alpha_state *templ) +enum pipe_error cso_set_depth_stencil_alpha(struct cso_context *ctx, + const struct pipe_depth_stencil_alpha_state *templ) { unsigned hash_key = cso_construct_key((void*)templ, sizeof(struct pipe_depth_stencil_alpha_state)); @@ -312,13 +505,20 @@ void cso_set_depth_stencil_alpha(struct cso_context *ctx, if (cso_hash_iter_is_null(iter)) { struct cso_depth_stencil_alpha *cso = MALLOC(sizeof(struct cso_depth_stencil_alpha)); + if (!cso) + return PIPE_ERROR_OUT_OF_MEMORY; cso->state = *templ; cso->data = ctx->pipe->create_depth_stencil_alpha_state(ctx->pipe, &cso->state); cso->delete_state = (cso_state_callback)ctx->pipe->delete_depth_stencil_alpha_state; cso->context = ctx->pipe; - cso_insert_state(ctx->cache, hash_key, CSO_DEPTH_STENCIL_ALPHA, cso); + iter = cso_insert_state(ctx->cache, hash_key, CSO_DEPTH_STENCIL_ALPHA, cso); + if (cso_hash_iter_is_null(iter)) { + FREE(cso); + return PIPE_ERROR_OUT_OF_MEMORY; + } + handle = cso->data; } else { @@ -329,6 +529,7 @@ void cso_set_depth_stencil_alpha(struct cso_context *ctx, ctx->depth_stencil = handle; ctx->pipe->bind_depth_stencil_alpha_state(ctx->pipe, handle); } + return PIPE_OK; } void cso_save_depth_stencil_alpha(struct cso_context *ctx) @@ -348,8 +549,8 @@ void cso_restore_depth_stencil_alpha(struct cso_context *ctx) -void cso_set_rasterizer(struct cso_context *ctx, - const struct pipe_rasterizer_state *templ) +enum pipe_error cso_set_rasterizer(struct cso_context *ctx, + const struct pipe_rasterizer_state *templ) { unsigned hash_key = cso_construct_key((void*)templ, sizeof(struct pipe_rasterizer_state)); @@ -360,13 +561,20 @@ void cso_set_rasterizer(struct cso_context *ctx, if (cso_hash_iter_is_null(iter)) { struct cso_rasterizer *cso = MALLOC(sizeof(struct cso_rasterizer)); + if (!cso) + return PIPE_ERROR_OUT_OF_MEMORY; cso->state = *templ; cso->data = ctx->pipe->create_rasterizer_state(ctx->pipe, &cso->state); cso->delete_state = (cso_state_callback)ctx->pipe->delete_rasterizer_state; cso->context = ctx->pipe; - cso_insert_state(ctx->cache, hash_key, CSO_RASTERIZER, cso); + iter = cso_insert_state(ctx->cache, hash_key, CSO_RASTERIZER, cso); + if (cso_hash_iter_is_null(iter)) { + FREE(cso); + return PIPE_ERROR_OUT_OF_MEMORY; + } + handle = cso->data; } else { @@ -377,6 +585,7 @@ void cso_set_rasterizer(struct cso_context *ctx, ctx->rasterizer = handle; ctx->pipe->bind_rasterizer_state(ctx->pipe, handle); } + return PIPE_OK; } void cso_save_rasterizer(struct cso_context *ctx) @@ -395,36 +604,71 @@ void cso_restore_rasterizer(struct cso_context *ctx) } -void cso_set_fragment_shader(struct cso_context *ctx, - const struct pipe_shader_state *templ) + +enum pipe_error cso_set_fragment_shader_handle(struct cso_context *ctx, + void *handle ) { - unsigned hash_key = cso_construct_key((void*)templ, - sizeof(struct pipe_shader_state)); + if (ctx->fragment_shader != handle) { + ctx->fragment_shader = handle; + ctx->pipe->bind_fs_state(ctx->pipe, handle); + } + return PIPE_OK; +} + +void cso_delete_fragment_shader(struct cso_context *ctx, void *handle ) +{ + if (handle == ctx->fragment_shader) { + /* unbind before deleting */ + ctx->pipe->bind_fs_state(ctx->pipe, NULL); + ctx->fragment_shader = NULL; + } + ctx->pipe->delete_fs_state(ctx->pipe, handle); +} + +/* Not really working: + */ +#if 0 +enum pipe_error cso_set_fragment_shader(struct cso_context *ctx, + const struct pipe_shader_state *templ) +{ + const struct tgsi_token *tokens = templ->tokens; + unsigned num_tokens = tgsi_num_tokens(tokens); + size_t tokens_size = num_tokens*sizeof(struct tgsi_token); + unsigned hash_key = cso_construct_key((void*)tokens, tokens_size); struct cso_hash_iter iter = cso_find_state_template(ctx->cache, - hash_key, CSO_FRAGMENT_SHADER, - (void*)templ); + hash_key, + CSO_FRAGMENT_SHADER, + (void*)tokens); void *handle = NULL; if (cso_hash_iter_is_null(iter)) { - struct cso_fragment_shader *cso = MALLOC(sizeof(struct cso_fragment_shader)); + struct cso_fragment_shader *cso = MALLOC(sizeof(struct cso_fragment_shader) + tokens_size); + struct tgsi_token *cso_tokens = (struct tgsi_token *)((char *)cso + sizeof(*cso)); - cso->state = *templ; + if (!cso) + return PIPE_ERROR_OUT_OF_MEMORY; + + memcpy(cso_tokens, tokens, tokens_size); + cso->state.tokens = cso_tokens; cso->data = ctx->pipe->create_fs_state(ctx->pipe, &cso->state); cso->delete_state = (cso_state_callback)ctx->pipe->delete_fs_state; cso->context = ctx->pipe; iter = cso_insert_state(ctx->cache, hash_key, CSO_FRAGMENT_SHADER, cso); + if (cso_hash_iter_is_null(iter)) { + FREE(cso); + return PIPE_ERROR_OUT_OF_MEMORY; + } + handle = cso->data; } else { handle = ((struct cso_fragment_shader *)cso_hash_iter_data(iter))->data; } - if (ctx->fragment_shader != handle) { - ctx->fragment_shader = handle; - ctx->pipe->bind_fs_state(ctx->pipe, handle); - } + return cso_set_fragment_shader_handle( ctx, handle ); } +#endif void cso_save_fragment_shader(struct cso_context *ctx) { @@ -434,7 +678,6 @@ void cso_save_fragment_shader(struct cso_context *ctx) void cso_restore_fragment_shader(struct cso_context *ctx) { - assert(ctx->fragment_shader_saved); if (ctx->fragment_shader_saved != ctx->fragment_shader) { ctx->pipe->bind_fs_state(ctx->pipe, ctx->fragment_shader_saved); ctx->fragment_shader = ctx->fragment_shader_saved; @@ -443,9 +686,32 @@ void cso_restore_fragment_shader(struct cso_context *ctx) } +enum pipe_error cso_set_vertex_shader_handle(struct cso_context *ctx, + void *handle ) +{ + if (ctx->vertex_shader != handle) { + ctx->vertex_shader = handle; + ctx->pipe->bind_vs_state(ctx->pipe, handle); + } + return PIPE_OK; +} + +void cso_delete_vertex_shader(struct cso_context *ctx, void *handle ) +{ + if (handle == ctx->vertex_shader) { + /* unbind before deleting */ + ctx->pipe->bind_vs_state(ctx->pipe, NULL); + ctx->vertex_shader = NULL; + } + ctx->pipe->delete_vs_state(ctx->pipe, handle); +} + -void cso_set_vertex_shader(struct cso_context *ctx, - const struct pipe_shader_state *templ) +/* Not really working: + */ +#if 0 +enum pipe_error cso_set_vertex_shader(struct cso_context *ctx, + const struct pipe_shader_state *templ) { unsigned hash_key = cso_construct_key((void*)templ, sizeof(struct pipe_shader_state)); @@ -457,23 +723,31 @@ void cso_set_vertex_shader(struct cso_context *ctx, if (cso_hash_iter_is_null(iter)) { struct cso_vertex_shader *cso = MALLOC(sizeof(struct cso_vertex_shader)); + if (!cso) + return PIPE_ERROR_OUT_OF_MEMORY; + cso->state = *templ; cso->data = ctx->pipe->create_vs_state(ctx->pipe, &cso->state); cso->delete_state = (cso_state_callback)ctx->pipe->delete_vs_state; cso->context = ctx->pipe; iter = cso_insert_state(ctx->cache, hash_key, CSO_VERTEX_SHADER, cso); + if (cso_hash_iter_is_null(iter)) { + FREE(cso); + return PIPE_ERROR_OUT_OF_MEMORY; + } + handle = cso->data; } else { handle = ((struct cso_vertex_shader *)cso_hash_iter_data(iter))->data; } - if (ctx->vertex_shader != handle) { - ctx->vertex_shader = handle; - ctx->pipe->bind_vs_state(ctx->pipe, handle); - } + return cso_set_vertex_shader_handle( ctx, handle ); } +#endif + + void cso_save_vertex_shader(struct cso_context *ctx) { @@ -483,9 +757,8 @@ void cso_save_vertex_shader(struct cso_context *ctx) void cso_restore_vertex_shader(struct cso_context *ctx) { - assert(ctx->vertex_shader_saved); if (ctx->vertex_shader_saved != ctx->vertex_shader) { - ctx->pipe->bind_fs_state(ctx->pipe, ctx->vertex_shader_saved); + ctx->pipe->bind_vs_state(ctx->pipe, ctx->vertex_shader_saved); ctx->vertex_shader = ctx->vertex_shader_saved; } ctx->vertex_shader_saved = NULL; @@ -493,14 +766,15 @@ void cso_restore_vertex_shader(struct cso_context *ctx) -void cso_set_framebuffer(struct cso_context *ctx, - const struct pipe_framebuffer_state *fb) +enum pipe_error cso_set_framebuffer(struct cso_context *ctx, + const struct pipe_framebuffer_state *fb) { /* XXX this memcmp() fails to detect buffer size changes */ if (1/*memcmp(&ctx->fb, fb, sizeof(*fb))*/) { ctx->fb = *fb; ctx->pipe->set_framebuffer_state(ctx->pipe, fb); } + return PIPE_OK; } void cso_save_framebuffer(struct cso_context *ctx) @@ -517,13 +791,14 @@ void cso_restore_framebuffer(struct cso_context *ctx) } -void cso_set_viewport(struct cso_context *ctx, - const struct pipe_viewport_state *vp) +enum pipe_error cso_set_viewport(struct cso_context *ctx, + const struct pipe_viewport_state *vp) { if (memcmp(&ctx->vp, vp, sizeof(*vp))) { ctx->vp = *vp; ctx->pipe->set_viewport_state(ctx->pipe, vp); } + return PIPE_OK; } void cso_save_viewport(struct cso_context *ctx) @@ -543,11 +818,12 @@ void cso_restore_viewport(struct cso_context *ctx) -void cso_set_blend_color(struct cso_context *ctx, - const struct pipe_blend_color *bc) +enum pipe_error cso_set_blend_color(struct cso_context *ctx, + const struct pipe_blend_color *bc) { if (memcmp(&ctx->blend_color, bc, sizeof(ctx->blend_color))) { ctx->blend_color = *bc; ctx->pipe->set_blend_color(ctx->pipe, bc); } + return PIPE_OK; } diff --git a/src/gallium/auxiliary/cso_cache/cso_context.h b/src/gallium/auxiliary/cso_cache/cso_context.h index 665e8d9911..b04e98bfa1 100644 --- a/src/gallium/auxiliary/cso_cache/cso_context.h +++ b/src/gallium/auxiliary/cso_cache/cso_context.h @@ -31,6 +31,7 @@ #include "pipe/p_context.h" #include "pipe/p_state.h" +#include "pipe/p_error.h" #ifdef __cplusplus @@ -41,51 +42,53 @@ struct cso_context; struct cso_context *cso_create_context( struct pipe_context *pipe ); +void cso_release_all( struct cso_context *ctx ); + void cso_destroy_context( struct cso_context *cso ); -void cso_set_blend( struct cso_context *cso, - const struct pipe_blend_state *blend ); +enum pipe_error cso_set_blend( struct cso_context *cso, + const struct pipe_blend_state *blend ); void cso_save_blend(struct cso_context *cso); void cso_restore_blend(struct cso_context *cso); -void cso_set_depth_stencil_alpha( struct cso_context *cso, - const struct pipe_depth_stencil_alpha_state *dsa ); +enum pipe_error cso_set_depth_stencil_alpha( struct cso_context *cso, + const struct pipe_depth_stencil_alpha_state *dsa ); void cso_save_depth_stencil_alpha(struct cso_context *cso); void cso_restore_depth_stencil_alpha(struct cso_context *cso); -void cso_set_rasterizer( struct cso_context *cso, - const struct pipe_rasterizer_state *rasterizer ); +enum pipe_error cso_set_rasterizer( struct cso_context *cso, + const struct pipe_rasterizer_state *rasterizer ); void cso_save_rasterizer(struct cso_context *cso); void cso_restore_rasterizer(struct cso_context *cso); -void cso_set_samplers( struct cso_context *cso, - unsigned count, - const struct pipe_sampler_state **states ); +enum pipe_error cso_set_samplers( struct cso_context *cso, + unsigned count, + const struct pipe_sampler_state **states ); void cso_save_samplers(struct cso_context *cso); void cso_restore_samplers(struct cso_context *cso); /* Alternate interface to support state trackers that like to modify * samplers one at a time: */ -void cso_single_sampler( struct cso_context *cso, - unsigned nr, - const struct pipe_sampler_state *states ); +enum pipe_error cso_single_sampler( struct cso_context *cso, + unsigned nr, + const struct pipe_sampler_state *states ); void cso_single_sampler_done( struct cso_context *cso ); -void cso_set_sampler_textures( struct cso_context *cso, - uint count, - struct pipe_texture **textures ); +enum pipe_error cso_set_sampler_textures( struct cso_context *cso, + uint count, + struct pipe_texture **textures ); void cso_save_sampler_textures( struct cso_context *cso ); void cso_restore_sampler_textures( struct cso_context *cso ); @@ -96,34 +99,43 @@ void cso_restore_sampler_textures( struct cso_context *cso ); * (eg mesa's internall-generated texenv programs), it will be up to * the state tracker to implement their own specialized caching. */ -void cso_set_fragment_shader( struct cso_context *cso, - const struct pipe_shader_state *shader ); +enum pipe_error cso_set_fragment_shader_handle(struct cso_context *ctx, + void *handle ); +void cso_delete_fragment_shader(struct cso_context *ctx, void *handle ); +/* +enum pipe_error cso_set_fragment_shader( struct cso_context *cso, + const struct pipe_shader_state *shader ); +*/ void cso_save_fragment_shader(struct cso_context *cso); void cso_restore_fragment_shader(struct cso_context *cso); - -void cso_set_vertex_shader( struct cso_context *cso, - const struct pipe_shader_state *shader ); +enum pipe_error cso_set_vertex_shader_handle(struct cso_context *ctx, + void *handle ); +void cso_delete_vertex_shader(struct cso_context *ctx, void *handle ); +/* +enum pipe_error cso_set_vertex_shader( struct cso_context *cso, + const struct pipe_shader_state *shader ); +*/ void cso_save_vertex_shader(struct cso_context *cso); void cso_restore_vertex_shader(struct cso_context *cso); -void cso_set_framebuffer(struct cso_context *cso, - const struct pipe_framebuffer_state *fb); +enum pipe_error cso_set_framebuffer(struct cso_context *cso, + const struct pipe_framebuffer_state *fb); void cso_save_framebuffer(struct cso_context *cso); void cso_restore_framebuffer(struct cso_context *cso); -void cso_set_viewport(struct cso_context *cso, - const struct pipe_viewport_state *vp); +enum pipe_error cso_set_viewport(struct cso_context *cso, + const struct pipe_viewport_state *vp); void cso_save_viewport(struct cso_context *cso); void cso_restore_viewport(struct cso_context *cso); -void cso_set_blend_color(struct cso_context *cso, - const struct pipe_blend_color *bc); +enum pipe_error cso_set_blend_color(struct cso_context *cso, + const struct pipe_blend_color *bc); #ifdef __cplusplus diff --git a/src/gallium/auxiliary/cso_cache/cso_hash.c b/src/gallium/auxiliary/cso_cache/cso_hash.c index ddce3822f7..0646efd952 100644 --- a/src/gallium/auxiliary/cso_cache/cso_hash.c +++ b/src/gallium/auxiliary/cso_cache/cso_hash.c @@ -110,6 +110,10 @@ cso_hash_create_node(struct cso_hash *hash, struct cso_node **anextNode) { struct cso_node *node = cso_data_allocate_node(hash->data.d); + + if (!node) + return NULL; + node->key = akey; node->value = avalue; @@ -219,15 +223,30 @@ struct cso_hash_iter cso_hash_insert(struct cso_hash *hash, { struct cso_node **nextNode = cso_hash_find_node(hash, key); struct cso_node *node = cso_hash_create_node(hash, key, data, nextNode); - struct cso_hash_iter iter = {hash, node}; - return iter; + if (!node) { + struct cso_hash_iter null_iter = {hash, 0}; + return null_iter; + } + + { + struct cso_hash_iter iter = {hash, node}; + return iter; + } } } struct cso_hash * cso_hash_create(void) { struct cso_hash *hash = MALLOC_STRUCT(cso_hash); + if (!hash) + return NULL; + hash->data.d = MALLOC_STRUCT(cso_hash_data); + if (!hash->data.d) { + FREE(hash); + return NULL; + } + hash->data.d->fakeNext = 0; hash->data.d->buckets = 0; hash->data.d->size = 0; diff --git a/src/gallium/auxiliary/cso_cache/cso_hash.h b/src/gallium/auxiliary/cso_cache/cso_hash.h index 73c4742006..85f3e276c6 100644 --- a/src/gallium/auxiliary/cso_cache/cso_hash.h +++ b/src/gallium/auxiliary/cso_cache/cso_hash.h @@ -106,12 +106,12 @@ struct cso_hash_iter cso_hash_iter_prev(struct cso_hash_iter iter); /** - * Convenience routine to iterate over the collision list while doing a memory + * Convenience routine to iterate over the collision list while doing a memory * comparison to see which entry in the list is a direct copy of our template * and returns that entry. */ void *cso_hash_find_data_from_template( struct cso_hash *hash, - unsigned hash_key, + unsigned hash_key, void *templ, int size ); diff --git a/src/gallium/auxiliary/draw/Makefile b/src/gallium/auxiliary/draw/Makefile index 28262a92c6..da7eded21f 100644 --- a/src/gallium/auxiliary/draw/Makefile +++ b/src/gallium/auxiliary/draw/Makefile @@ -4,39 +4,37 @@ include $(TOP)/configs/current LIBNAME = draw C_SOURCES = \ - draw_aaline.c \ - draw_aapoint.c \ - draw_clip.c \ - draw_vs_exec.c \ - draw_vs_sse.c \ - draw_vs_llvm.c \ - draw_context.c\ - draw_cull.c \ - draw_debug.c \ - draw_flatshade.c \ - draw_offset.c \ + draw_context.c \ + draw_pipe.c \ + draw_pipe_aaline.c \ + draw_pipe_aapoint.c \ + draw_pipe_clip.c \ + draw_pipe_cull.c \ + draw_pipe_flatshade.c \ + draw_pipe_offset.c \ + draw_pipe_pstipple.c \ + draw_pipe_stipple.c \ + draw_pipe_twoside.c \ + draw_pipe_unfilled.c \ + draw_pipe_util.c \ + draw_pipe_validate.c \ + draw_pipe_vbuf.c \ + draw_pipe_wide_line.c \ + draw_pipe_wide_point.c \ draw_pt.c \ - draw_pt_vcache.c \ - draw_pt_fetch_emit.c \ - draw_pt_fetch_pipeline.c \ - draw_pt_pipeline.c \ draw_pt_elts.c \ - draw_prim.c \ - draw_pstipple.c \ - draw_stipple.c \ - draw_twoside.c \ - draw_unfilled.c \ - draw_validate.c \ - draw_vbuf.c \ + draw_pt_emit.c \ + draw_pt_fetch.c \ + draw_pt_fetch_emit.c \ + draw_pt_fetch_shade_pipeline.c \ + draw_pt_post_vs.c \ + draw_pt_varray.c \ + draw_pt_vcache.c \ draw_vertex.c \ - draw_vertex_cache.c \ - draw_vertex_fetch.c \ - draw_vertex_shader.c \ - draw_vf.c \ - draw_vf_generic.c \ - draw_vf_sse.c \ - draw_wide_line.c \ - draw_wide_point.c + draw_vs.c \ + draw_vs_exec.c \ + draw_vs_llvm.c \ + draw_vs_sse.c include ../../Makefile.template diff --git a/src/gallium/auxiliary/draw/SConscript b/src/gallium/auxiliary/draw/SConscript index 52107912f5..3b5d5ed492 100644 --- a/src/gallium/auxiliary/draw/SConscript +++ b/src/gallium/auxiliary/draw/SConscript @@ -3,39 +3,37 @@ Import('*') draw = env.ConvenienceLibrary( target = 'draw', source = [ - 'draw_aaline.c', - 'draw_aapoint.c', - 'draw_clip.c', - 'draw_vs_exec.c', - 'draw_vs_sse.c', - 'draw_vs_llvm.c', 'draw_context.c', - 'draw_cull.c', - 'draw_debug.c', - 'draw_flatshade.c', - 'draw_offset.c', + 'draw_pipe.c', + 'draw_pipe_aaline.c', + 'draw_pipe_aapoint.c', + 'draw_pipe_clip.c', + 'draw_pipe_cull.c', + 'draw_pipe_flatshade.c', + 'draw_pipe_offset.c', + 'draw_pipe_pstipple.c', + 'draw_pipe_stipple.c', + 'draw_pipe_twoside.c', + 'draw_pipe_unfilled.c', + 'draw_pipe_util.c', + 'draw_pipe_validate.c', + 'draw_pipe_vbuf.c', + 'draw_pipe_wide_line.c', + 'draw_pipe_wide_point.c', 'draw_pt.c', - 'draw_pt_vcache.c', - 'draw_pt_fetch_emit.c', - 'draw_pt_fetch_pipeline.c', - 'draw_pt_pipeline.c', 'draw_pt_elts.c', - 'draw_prim.c', - 'draw_pstipple.c', - 'draw_stipple.c', - 'draw_twoside.c', - 'draw_unfilled.c', - 'draw_validate.c', - 'draw_vbuf.c', + 'draw_pt_emit.c', + 'draw_pt_fetch.c', + 'draw_pt_fetch_emit.c', + 'draw_pt_fetch_shade_pipeline.c', + 'draw_pt_post_vs.c', + 'draw_pt_varray.c', + 'draw_pt_vcache.c', 'draw_vertex.c', - 'draw_vertex_cache.c', - 'draw_vertex_fetch.c', - 'draw_vertex_shader.c', - 'draw_vf.c', - 'draw_vf_generic.c', - 'draw_vf_sse.c', - 'draw_wide_point.c', - 'draw_wide_line.c' + 'draw_vs.c', + 'draw_vs_exec.c', + 'draw_vs_llvm.c', + 'draw_vs_sse.c', ]) auxiliaries.insert(0, draw) diff --git a/src/gallium/auxiliary/draw/draw_context.c b/src/gallium/auxiliary/draw/draw_context.c index b3c65c90d6..98e23fa830 100644 --- a/src/gallium/auxiliary/draw/draw_context.c +++ b/src/gallium/auxiliary/draw/draw_context.c @@ -33,8 +33,10 @@ #include "pipe/p_util.h" #include "draw_context.h" -#include "draw_private.h" #include "draw_vbuf.h" +#include "draw_vs.h" +#include "draw_pt.h" +#include "draw_pipe.h" struct draw_context *draw_create( void ) @@ -43,38 +45,6 @@ struct draw_context *draw_create( void ) if (draw == NULL) goto fail; -#if defined(__i386__) || defined(__386__) - draw->use_sse = GETENV( "GALLIUM_NOSSE" ) == NULL; -#else - draw->use_sse = FALSE; -#endif - - /* create pipeline stages */ - draw->pipeline.wide_line = draw_wide_line_stage( draw ); - draw->pipeline.wide_point = draw_wide_point_stage( draw ); - draw->pipeline.stipple = draw_stipple_stage( draw ); - draw->pipeline.unfilled = draw_unfilled_stage( draw ); - draw->pipeline.twoside = draw_twoside_stage( draw ); - draw->pipeline.offset = draw_offset_stage( draw ); - draw->pipeline.clip = draw_clip_stage( draw ); - draw->pipeline.flatshade = draw_flatshade_stage( draw ); - draw->pipeline.cull = draw_cull_stage( draw ); - draw->pipeline.validate = draw_validate_stage( draw ); - draw->pipeline.first = draw->pipeline.validate; - - if (!draw->pipeline.wide_line || - !draw->pipeline.wide_point || - !draw->pipeline.stipple || - !draw->pipeline.unfilled || - !draw->pipeline.twoside || - !draw->pipeline.offset || - !draw->pipeline.clip || - !draw->pipeline.flatshade || - !draw->pipeline.cull || - !draw->pipeline.validate) - goto fail; - - ASSIGN_4V( draw->plane[0], -1, 0, 0, 1 ); ASSIGN_4V( draw->plane[1], 1, 0, 0, 1 ); ASSIGN_4V( draw->plane[2], 0, -1, 0, 1 ); @@ -83,31 +53,18 @@ struct draw_context *draw_create( void ) ASSIGN_4V( draw->plane[5], 0, 0, -1, 1 ); /* mesa's a bit wonky */ draw->nr_planes = 6; - /* Statically allocate maximum sized vertices for the cache - could be cleverer... - */ - { - uint i; - const unsigned size = (MAX_VERTEX_SIZE + 0x0f) & ~0x0f; - char *tmp = align_malloc(Elements(draw->vs.queue) * size, 16); - if (!tmp) - goto fail; - - for (i = 0; i < Elements(draw->vs.queue); i++) - draw->vs.queue[i].vertex = (struct vertex_header *)(tmp + i * size); - } - draw->shader_queue_flush = draw_vertex_shader_queue_flush; + draw->reduced_prim = ~0; /* != any of PIPE_PRIM_x */ - /* these defaults are oriented toward the needs of softpipe */ - draw->wide_point_threshold = 1000000.0; /* infinity */ - draw->wide_line_threshold = 1.0; - draw->line_stipple = TRUE; - draw->point_sprite = TRUE; + tgsi_exec_machine_init(&draw->machine); - draw->reduced_prim = ~0; /* != any of PIPE_PRIM_x */ + /* FIXME: give this machine thing a proper constructor: + */ + draw->machine.Inputs = align_malloc(PIPE_MAX_ATTRIBS * sizeof(struct tgsi_exec_vector), 16); + draw->machine.Outputs = align_malloc(PIPE_MAX_ATTRIBS * sizeof(struct tgsi_exec_vector), 16); - draw_vertex_cache_invalidate( draw ); - draw_set_mapped_element_buffer( draw, 0, NULL ); + if (!draw_pipeline_init( draw )) + goto fail; if (!draw_pt_init( draw )) goto fail; @@ -125,39 +82,14 @@ void draw_destroy( struct draw_context *draw ) if (!draw) return; - if (draw->pipeline.wide_line) - draw->pipeline.wide_line->destroy( draw->pipeline.wide_line ); - if (draw->pipeline.wide_point) - draw->pipeline.wide_point->destroy( draw->pipeline.wide_point ); - if (draw->pipeline.stipple) - draw->pipeline.stipple->destroy( draw->pipeline.stipple ); - if (draw->pipeline.unfilled) - draw->pipeline.unfilled->destroy( draw->pipeline.unfilled ); - if (draw->pipeline.twoside) - draw->pipeline.twoside->destroy( draw->pipeline.twoside ); - if (draw->pipeline.offset) - draw->pipeline.offset->destroy( draw->pipeline.offset ); - if (draw->pipeline.clip) - draw->pipeline.clip->destroy( draw->pipeline.clip ); - if (draw->pipeline.flatshade) - draw->pipeline.flatshade->destroy( draw->pipeline.flatshade ); - if (draw->pipeline.cull) - draw->pipeline.cull->destroy( draw->pipeline.cull ); - if (draw->pipeline.validate) - draw->pipeline.validate->destroy( draw->pipeline.validate ); - if (draw->pipeline.aaline) - draw->pipeline.aaline->destroy( draw->pipeline.aaline ); - if (draw->pipeline.aapoint) - draw->pipeline.aapoint->destroy( draw->pipeline.aapoint ); - if (draw->pipeline.pstipple) - draw->pipeline.pstipple->destroy( draw->pipeline.pstipple ); - if (draw->pipeline.rasterize) - draw->pipeline.rasterize->destroy( draw->pipeline.rasterize ); + + if (draw->machine.Inputs) + align_free(draw->machine.Inputs); + + if (draw->machine.Outputs) + align_free(draw->machine.Outputs); tgsi_exec_machine_free_data(&draw->machine); - - if (draw->vs.queue[0].vertex) - align_free( draw->vs.queue[0].vertex ); /* Frees all the vertices. */ /* Not so fast -- we're just borrowing this at the moment. * @@ -165,6 +97,7 @@ void draw_destroy( struct draw_context *draw ) draw->render->destroy( draw->render ); */ + draw_pipeline_destroy( draw ); draw_pt_destroy( draw ); FREE( draw ); @@ -189,6 +122,20 @@ void draw_set_rasterizer_state( struct draw_context *draw, draw_do_flush( draw, DRAW_FLUSH_STATE_CHANGE ); draw->rasterizer = raster; + draw->bypass_clipping = + ((draw->rasterizer && draw->rasterizer->bypass_clipping) || + draw->driver.bypass_clipping); +} + + +void draw_set_driver_clipping( struct draw_context *draw, + boolean bypass_clipping ) +{ + draw_do_flush( draw, DRAW_FLUSH_STATE_CHANGE ); + + draw->driver.bypass_clipping = bypass_clipping; + draw->bypass_clipping = (draw->rasterizer->bypass_clipping || + draw->driver.bypass_clipping); } @@ -247,9 +194,8 @@ draw_set_vertex_buffers(struct draw_context *draw, { assert(count <= PIPE_MAX_ATTRIBS); - draw_do_flush( draw, DRAW_FLUSH_VERTEX_CACHE/*STATE_CHANGE*/ ); - - memcpy(draw->vertex_buffer, buffers, count * sizeof(buffers[0])); + memcpy(draw->pt.vertex_buffer, buffers, count * sizeof(buffers[0])); + draw->pt.nr_vertex_buffers = count; } @@ -260,9 +206,8 @@ draw_set_vertex_elements(struct draw_context *draw, { assert(count <= PIPE_MAX_ATTRIBS); - draw_do_flush( draw, DRAW_FLUSH_VERTEX_CACHE/*STATE_CHANGE*/ ); - - memcpy(draw->vertex_element, elements, count * sizeof(elements[0])); + memcpy(draw->pt.vertex_element, elements, count * sizeof(elements[0])); + draw->pt.nr_vertex_elements = count; } @@ -273,8 +218,7 @@ void draw_set_mapped_vertex_buffer(struct draw_context *draw, unsigned attr, const void *buffer) { - draw_do_flush( draw, DRAW_FLUSH_VERTEX_CACHE/*STATE_CHANGE*/ ); - draw->user.vbuffer[attr] = buffer; + draw->pt.user.vbuffer[attr] = buffer; } @@ -282,8 +226,7 @@ void draw_set_mapped_constant_buffer(struct draw_context *draw, const void *buffer) { - draw_do_flush( draw, DRAW_FLUSH_VERTEX_CACHE/*STATE_CHANGE*/ ); - draw->user.constants = buffer; + draw->pt.user.constants = buffer; } @@ -295,7 +238,7 @@ void draw_wide_point_threshold(struct draw_context *draw, float threshold) { draw_do_flush( draw, DRAW_FLUSH_STATE_CHANGE ); - draw->wide_point_threshold = threshold; + draw->pipeline.wide_point_threshold = threshold; } @@ -307,7 +250,7 @@ void draw_wide_line_threshold(struct draw_context *draw, float threshold) { draw_do_flush( draw, DRAW_FLUSH_STATE_CHANGE ); - draw->wide_line_threshold = threshold; + draw->pipeline.wide_line_threshold = threshold; } @@ -318,7 +261,7 @@ void draw_enable_line_stipple(struct draw_context *draw, boolean enable) { draw_do_flush( draw, DRAW_FLUSH_STATE_CHANGE ); - draw->line_stipple = enable; + draw->pipeline.line_stipple = enable; } @@ -329,7 +272,7 @@ void draw_enable_point_sprites(struct draw_context *draw, boolean enable) { draw_do_flush( draw, DRAW_FLUSH_STATE_CHANGE ); - draw->point_sprite = enable; + draw->pipeline.point_sprite = enable; } @@ -384,79 +327,56 @@ draw_num_vs_outputs(struct draw_context *draw) } -/** - * Allocate space for temporary post-transform vertices, such as for clipping. - */ -void draw_alloc_temp_verts( struct draw_stage *stage, unsigned nr ) -{ - assert(!stage->tmp); - - stage->nr_tmps = nr; - - if (nr) { - ubyte *store = (ubyte *) MALLOC( MAX_VERTEX_SIZE * nr ); - unsigned i; - - stage->tmp = (struct vertex_header **) MALLOC( sizeof(struct vertex_header *) * nr ); - - for (i = 0; i < nr; i++) - stage->tmp[i] = (struct vertex_header *)(store + i * MAX_VERTEX_SIZE); - } -} - -void draw_free_temp_verts( struct draw_stage *stage ) +void draw_set_render( struct draw_context *draw, + struct vbuf_render *render ) { - if (stage->tmp) { - FREE( stage->tmp[0] ); - FREE( stage->tmp ); - stage->tmp = NULL; - } + draw->render = render; } - -boolean draw_use_sse(struct draw_context *draw) +void draw_set_edgeflags( struct draw_context *draw, + const unsigned *edgeflag ) { - return (boolean) draw->use_sse; + draw->pt.user.edgeflag = edgeflag; } -void draw_reset_vertex_ids(struct draw_context *draw) -{ - struct draw_stage *stage = draw->pipeline.first; - - while (stage) { - unsigned i; - - for (i = 0; i < stage->nr_tmps; i++) - stage->tmp[i]->vertex_id = UNDEFINED_VERTEX_ID; - stage = stage->next; - } - draw_vertex_cache_reset_vertex_ids(draw); /* going away soon */ - draw_pt_reset_vertex_ids(draw); +/** + * Tell the drawing context about the index/element buffer to use + * (ala glDrawElements) + * If no element buffer is to be used (i.e. glDrawArrays) then this + * should be called with eltSize=0 and elements=NULL. + * + * \param draw the drawing context + * \param eltSize size of each element (1, 2 or 4 bytes) + * \param elements the element buffer ptr + */ +void +draw_set_mapped_element_buffer( struct draw_context *draw, + unsigned eltSize, void *elements ) +{ + draw->pt.user.elts = elements; + draw->pt.user.eltSize = eltSize; } -void draw_set_render( struct draw_context *draw, - struct vbuf_render *render ) + +/* Revamp me please: + */ +void draw_do_flush( struct draw_context *draw, unsigned flags ) { - draw->render = render; -} + if (!draw->suspend_flushing) + { + assert(!draw->flushing); /* catch inadvertant recursion */ -void draw_set_edgeflags( struct draw_context *draw, - const unsigned *edgeflag ) -{ - draw->user.edgeflag = edgeflag; -} + draw->flushing = TRUE; + draw_pipeline_flush( draw, flags ); -boolean draw_get_edgeflag( struct draw_context *draw, - unsigned idx ) -{ - if (draw->user.edgeflag) - return (draw->user.edgeflag[idx/32] & (1 << (idx%32))) != 0; - else - return 1; + draw->reduced_prim = ~0; /* is reduced_prim needed any more? */ + + draw->flushing = FALSE; + } } diff --git a/src/gallium/auxiliary/draw/draw_context.h b/src/gallium/auxiliary/draw/draw_context.h index c7ac32b452..c5c3d3b09e 100644 --- a/src/gallium/auxiliary/draw/draw_context.h +++ b/src/gallium/auxiliary/draw/draw_context.h @@ -42,37 +42,11 @@ struct pipe_context; -struct vertex_buffer; -struct vertex_info; struct draw_context; struct draw_stage; struct draw_vertex_shader; -/** - * Clipmask flags - */ -/*@{*/ -#define CLIP_RIGHT_BIT 0x01 -#define CLIP_LEFT_BIT 0x02 -#define CLIP_TOP_BIT 0x04 -#define CLIP_BOTTOM_BIT 0x08 -#define CLIP_NEAR_BIT 0x10 -#define CLIP_FAR_BIT 0x20 -/*@}*/ - -/** - * Bitshift for each clip flag - */ -/*@{*/ -#define CLIP_RIGHT_SHIFT 0 -#define CLIP_LEFT_SHIFT 1 -#define CLIP_TOP_SHIFT 2 -#define CLIP_BOTTOM_SHIFT 3 -#define CLIP_NEAR_SHIFT 4 -#define CLIP_FAR_SHIFT 5 -/*@}*/ - struct draw_context *draw_create( void ); @@ -99,15 +73,13 @@ void draw_enable_line_stipple(struct draw_context *draw, boolean enable); void draw_enable_point_sprites(struct draw_context *draw, boolean enable); -boolean draw_use_sse(struct draw_context *draw); - -void +boolean draw_install_aaline_stage(struct draw_context *draw, struct pipe_context *pipe); -void +boolean draw_install_aapoint_stage(struct draw_context *draw, struct pipe_context *pipe); -void +boolean draw_install_pstipple_stage(struct draw_context *draw, struct pipe_context *pipe); @@ -168,17 +140,24 @@ void draw_arrays(struct draw_context *draw, unsigned prim, void draw_flush(struct draw_context *draw); -/*********************************************************************** - * draw_debug.c + +/******************************************************************************* + * Driver backend interface */ -boolean draw_validate_prim( unsigned prim, unsigned length ); -unsigned draw_trim_prim( unsigned mode, unsigned count ); +struct vbuf_render; +void draw_set_render( struct draw_context *draw, + struct vbuf_render *render ); +void draw_set_driver_clipping( struct draw_context *draw, + boolean bypass_clipping ); +/******************************************************************************* + * Draw pipeline + */ +boolean draw_need_pipeline(const struct draw_context *draw, + const struct pipe_rasterizer_state *rasterizer, + unsigned prim ); -struct vbuf_render; -void draw_set_render( struct draw_context *draw, - struct vbuf_render *render ); #endif /* DRAW_CONTEXT_H */ diff --git a/src/gallium/auxiliary/draw/draw_debug.c b/src/gallium/auxiliary/draw/draw_debug.c deleted file mode 100644 index d6220b5f62..0000000000 --- a/src/gallium/auxiliary/draw/draw_debug.c +++ /dev/null @@ -1,113 +0,0 @@ -/************************************************************************** - * - * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - - /* - * Authors: - * Keith Whitwell <keith@tungstengraphics.com> - */ - -#include "draw_private.h" -#include "draw_context.h" - - - -static void -draw_prim_info(unsigned prim, unsigned *first, unsigned *incr) -{ - assert(prim >= PIPE_PRIM_POINTS); - assert(prim <= PIPE_PRIM_POLYGON); - - switch (prim) { - case PIPE_PRIM_POINTS: - *first = 1; - *incr = 1; - break; - case PIPE_PRIM_LINES: - *first = 2; - *incr = 2; - break; - case PIPE_PRIM_LINE_STRIP: - *first = 2; - *incr = 1; - break; - case PIPE_PRIM_LINE_LOOP: - *first = 2; - *incr = 1; - break; - case PIPE_PRIM_TRIANGLES: - *first = 3; - *incr = 3; - break; - case PIPE_PRIM_TRIANGLE_STRIP: - *first = 3; - *incr = 1; - break; - case PIPE_PRIM_TRIANGLE_FAN: - case PIPE_PRIM_POLYGON: - *first = 3; - *incr = 1; - break; - case PIPE_PRIM_QUADS: - *first = 4; - *incr = 4; - break; - case PIPE_PRIM_QUAD_STRIP: - *first = 4; - *incr = 2; - break; - default: - assert(0); - *first = 1; - *incr = 1; - break; - } -} - - -unsigned -draw_trim_prim( unsigned mode, unsigned count ) -{ - unsigned length, first, incr; - - draw_prim_info( mode, &first, &incr ); - - if (count < first) - length = 0; - else - length = count - (count - first) % incr; - - return length; -} - - -boolean -draw_validate_prim( unsigned mode, unsigned count ) -{ - return (count > 0 && - count == draw_trim_prim( mode, count )); -} - diff --git a/src/gallium/auxiliary/draw/draw_pipe.c b/src/gallium/auxiliary/draw/draw_pipe.c new file mode 100644 index 0000000000..46afb0f41f --- /dev/null +++ b/src/gallium/auxiliary/draw/draw_pipe.c @@ -0,0 +1,222 @@ +/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + + /* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#include "pipe/p_util.h" +#include "draw/draw_private.h" +#include "draw/draw_pipe.h" + + + +boolean draw_pipeline_init( struct draw_context *draw ) +{ + /* create pipeline stages */ + draw->pipeline.wide_line = draw_wide_line_stage( draw ); + draw->pipeline.wide_point = draw_wide_point_stage( draw ); + draw->pipeline.stipple = draw_stipple_stage( draw ); + draw->pipeline.unfilled = draw_unfilled_stage( draw ); + draw->pipeline.twoside = draw_twoside_stage( draw ); + draw->pipeline.offset = draw_offset_stage( draw ); + draw->pipeline.clip = draw_clip_stage( draw ); + draw->pipeline.flatshade = draw_flatshade_stage( draw ); + draw->pipeline.cull = draw_cull_stage( draw ); + draw->pipeline.validate = draw_validate_stage( draw ); + draw->pipeline.first = draw->pipeline.validate; + + if (!draw->pipeline.wide_line || + !draw->pipeline.wide_point || + !draw->pipeline.stipple || + !draw->pipeline.unfilled || + !draw->pipeline.twoside || + !draw->pipeline.offset || + !draw->pipeline.clip || + !draw->pipeline.flatshade || + !draw->pipeline.cull || + !draw->pipeline.validate) + return FALSE; + + /* these defaults are oriented toward the needs of softpipe */ + draw->pipeline.wide_point_threshold = 1000000.0; /* infinity */ + draw->pipeline.wide_line_threshold = 1.0; + draw->pipeline.line_stipple = TRUE; + draw->pipeline.point_sprite = TRUE; + + return TRUE; +} + + +void draw_pipeline_destroy( struct draw_context *draw ) +{ + if (draw->pipeline.wide_line) + draw->pipeline.wide_line->destroy( draw->pipeline.wide_line ); + if (draw->pipeline.wide_point) + draw->pipeline.wide_point->destroy( draw->pipeline.wide_point ); + if (draw->pipeline.stipple) + draw->pipeline.stipple->destroy( draw->pipeline.stipple ); + if (draw->pipeline.unfilled) + draw->pipeline.unfilled->destroy( draw->pipeline.unfilled ); + if (draw->pipeline.twoside) + draw->pipeline.twoside->destroy( draw->pipeline.twoside ); + if (draw->pipeline.offset) + draw->pipeline.offset->destroy( draw->pipeline.offset ); + if (draw->pipeline.clip) + draw->pipeline.clip->destroy( draw->pipeline.clip ); + if (draw->pipeline.flatshade) + draw->pipeline.flatshade->destroy( draw->pipeline.flatshade ); + if (draw->pipeline.cull) + draw->pipeline.cull->destroy( draw->pipeline.cull ); + if (draw->pipeline.validate) + draw->pipeline.validate->destroy( draw->pipeline.validate ); + if (draw->pipeline.aaline) + draw->pipeline.aaline->destroy( draw->pipeline.aaline ); + if (draw->pipeline.aapoint) + draw->pipeline.aapoint->destroy( draw->pipeline.aapoint ); + if (draw->pipeline.pstipple) + draw->pipeline.pstipple->destroy( draw->pipeline.pstipple ); + if (draw->pipeline.rasterize) + draw->pipeline.rasterize->destroy( draw->pipeline.rasterize ); +} + + + + + + + +static void do_point( struct draw_context *draw, + const char *v0 ) +{ + struct prim_header prim; + + prim.flags = 0; + prim.pad = 0; + prim.v[0] = (struct vertex_header *)v0; + + draw->pipeline.first->point( draw->pipeline.first, &prim ); +} + + +static void do_line( struct draw_context *draw, + ushort flags, + const char *v0, + const char *v1 ) +{ + struct prim_header prim; + + prim.flags = flags; + prim.pad = 0; + prim.v[0] = (struct vertex_header *)v0; + prim.v[1] = (struct vertex_header *)v1; + + draw->pipeline.first->line( draw->pipeline.first, &prim ); +} + + +static void do_triangle( struct draw_context *draw, + ushort flags, + char *v0, + char *v1, + char *v2 ) +{ + struct prim_header prim; + + prim.v[0] = (struct vertex_header *)v0; + prim.v[1] = (struct vertex_header *)v1; + prim.v[2] = (struct vertex_header *)v2; + prim.flags = flags; + prim.pad = 0; + + draw->pipeline.first->tri( draw->pipeline.first, &prim ); +} + + + + +/* Code to run the pipeline on a fairly arbitary collection of vertices. + * + * Vertex headers must be pre-initialized with the + * UNDEFINED_VERTEX_ID, this code will cause that id to become + * overwritten, so it may have to be reset if there is the intention + * to reuse the vertices. + * + * This code provides a callback to reset the vertex id's which the + * draw_vbuf.c code uses when it has to perform a flush. + */ +void draw_pipeline_run( struct draw_context *draw, + unsigned prim, + struct vertex_header *vertices, + unsigned vertex_count, + unsigned stride, + const ushort *elts, + unsigned count ) +{ + char *verts = (char *)vertices; + unsigned i; + + draw->pipeline.verts = verts; + draw->pipeline.vertex_stride = stride; + draw->pipeline.vertex_count = vertex_count; + + switch (prim) { + case PIPE_PRIM_POINTS: + for (i = 0; i < count; i++) + do_point( draw, + verts + stride * elts[i] ); + break; + case PIPE_PRIM_LINES: + for (i = 0; i+1 < count; i += 2) + do_line( draw, + elts[i+0], /* flags */ + verts + stride * (elts[i+0] & ~DRAW_PIPE_FLAG_MASK), + verts + stride * elts[i+1]); + break; + case PIPE_PRIM_TRIANGLES: + for (i = 0; i+2 < count; i += 3) + do_triangle( draw, + elts[i+0], /* flags */ + verts + stride * (elts[i+0] & ~DRAW_PIPE_FLAG_MASK), + verts + stride * elts[i+1], + verts + stride * elts[i+2]); + break; + } + + draw->pipeline.verts = NULL; + draw->pipeline.vertex_count = 0; +} + + + +void draw_pipeline_flush( struct draw_context *draw, + unsigned flags ) +{ + draw->pipeline.first->flush( draw->pipeline.first, flags ); + draw->pipeline.first = draw->pipeline.validate; +} diff --git a/src/gallium/auxiliary/draw/draw_pipe.h b/src/gallium/auxiliary/draw/draw_pipe.h new file mode 100644 index 0000000000..f1cb0891ca --- /dev/null +++ b/src/gallium/auxiliary/draw/draw_pipe.h @@ -0,0 +1,125 @@ +/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + + /* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#ifndef DRAW_PIPE_H +#define DRAW_PIPE_H + +#include "pipe/p_compiler.h" +#include "draw_private.h" /* for sizeof(vertex_header) */ + + +/** + * Basic info for a point/line/triangle primitive. + */ +struct prim_header { + float det; /**< front/back face determinant */ + ushort flags; + ushort pad; + struct vertex_header *v[3]; /**< 1 to 3 vertex pointers */ +}; + + + +/** + * Base class for all primitive drawing stages. + */ +struct draw_stage +{ + struct draw_context *draw; /**< parent context */ + + struct draw_stage *next; /**< next stage in pipeline */ + + struct vertex_header **tmp; /**< temp vert storage, such as for clipping */ + unsigned nr_tmps; + + void (*point)( struct draw_stage *, + struct prim_header * ); + + void (*line)( struct draw_stage *, + struct prim_header * ); + + void (*tri)( struct draw_stage *, + struct prim_header * ); + + void (*flush)( struct draw_stage *, + unsigned flags ); + + void (*reset_stipple_counter)( struct draw_stage * ); + + void (*destroy)( struct draw_stage * ); +}; + + +extern struct draw_stage *draw_unfilled_stage( struct draw_context *context ); +extern struct draw_stage *draw_twoside_stage( struct draw_context *context ); +extern struct draw_stage *draw_offset_stage( struct draw_context *context ); +extern struct draw_stage *draw_clip_stage( struct draw_context *context ); +extern struct draw_stage *draw_flatshade_stage( struct draw_context *context ); +extern struct draw_stage *draw_cull_stage( struct draw_context *context ); +extern struct draw_stage *draw_stipple_stage( struct draw_context *context ); +extern struct draw_stage *draw_wide_line_stage( struct draw_context *context ); +extern struct draw_stage *draw_wide_point_stage( struct draw_context *context ); +extern struct draw_stage *draw_validate_stage( struct draw_context *context ); + + +extern void draw_free_temp_verts( struct draw_stage *stage ); +extern boolean draw_alloc_temp_verts( struct draw_stage *stage, unsigned nr ); + +extern void draw_reset_vertex_ids( struct draw_context *draw ); + +void draw_pipe_passthrough_tri(struct draw_stage *stage, struct prim_header *header); +void draw_pipe_passthrough_line(struct draw_stage *stage, struct prim_header *header); +void draw_pipe_passthrough_point(struct draw_stage *stage, struct prim_header *header); + + + +/** + * Get a writeable copy of a vertex. + * \param stage drawing stage info + * \param vert the vertex to copy (source) + * \param idx index into stage's tmp[] array to put the copy (dest) + * \return pointer to the copied vertex + */ +static INLINE struct vertex_header * +dup_vert( struct draw_stage *stage, + const struct vertex_header *vert, + unsigned idx ) +{ + struct vertex_header *tmp = stage->tmp[idx]; + const uint vsize = sizeof(struct vertex_header) + + stage->draw->num_vs_outputs * 4 * sizeof(float); + memcpy(tmp, vert, vsize); + tmp->vertex_id = UNDEFINED_VERTEX_ID; + return tmp; +} + +#endif diff --git a/src/gallium/auxiliary/draw/draw_aaline.c b/src/gallium/auxiliary/draw/draw_pipe_aaline.c index e8d2a45102..346a96e37f 100644 --- a/src/gallium/auxiliary/draw/draw_aaline.c +++ b/src/gallium/auxiliary/draw/draw_pipe_aaline.c @@ -43,6 +43,7 @@ #include "draw_context.h" #include "draw_private.h" +#include "draw_pipe.h" /** @@ -333,11 +334,10 @@ aa_transform_inst(struct tgsi_transform_context *ctx, * Generate the frag shader we'll use for drawing AA lines. * This will be the user's shader plus some texture/modulate instructions. */ -static void +static boolean generate_aaline_fs(struct aaline_stage *aaline) { const struct pipe_shader_state *orig_fs = &aaline->fs->state; - //struct draw_context *draw = aaline->stage.draw; struct pipe_shader_state aaline_fs; struct aa_transform_context transform; @@ -345,6 +345,8 @@ generate_aaline_fs(struct aaline_stage *aaline) aaline_fs = *orig_fs; /* copy to init */ aaline_fs.tokens = MALLOC(sizeof(struct tgsi_token) * MAX); + if (aaline_fs.tokens == NULL) + return FALSE; memset(&transform, 0, sizeof(transform)); transform.colorOutput = -1; @@ -369,15 +371,18 @@ generate_aaline_fs(struct aaline_stage *aaline) aaline->fs->aaline_fs = aaline->driver_create_fs_state(aaline->pipe, &aaline_fs); + if (aaline->fs->aaline_fs == NULL) + return FALSE; aaline->fs->generic_attrib = transform.maxGeneric + 1; + return TRUE; } /** * Create the texture map we'll use for antialiasing the lines. */ -static void +static boolean aaline_create_texture(struct aaline_stage *aaline) { struct pipe_context *pipe = aaline->pipe; @@ -387,7 +392,7 @@ aaline_create_texture(struct aaline_stage *aaline) memset(&texTemp, 0, sizeof(texTemp)); texTemp.target = PIPE_TEXTURE_2D; - texTemp.format = PIPE_FORMAT_U_A8; /* XXX verify supported by driver! */ + texTemp.format = PIPE_FORMAT_A8_UNORM; /* XXX verify supported by driver! */ texTemp.last_level = MAX_TEXTURE_LEVEL; texTemp.width[0] = 1 << MAX_TEXTURE_LEVEL; texTemp.height[0] = 1 << MAX_TEXTURE_LEVEL; @@ -395,6 +400,8 @@ aaline_create_texture(struct aaline_stage *aaline) texTemp.cpp = 1; aaline->texture = screen->texture_create(screen, &texTemp); + if (!aaline->texture) + return FALSE; /* Fill in mipmap images. * Basically each level is solid opaque, except for the outermost @@ -410,6 +417,8 @@ aaline_create_texture(struct aaline_stage *aaline) surface = screen->get_tex_surface(screen, aaline->texture, 0, level, 0); data = pipe_surface_map(surface); + if (data == NULL) + return FALSE; for (i = 0; i < size; i++) { for (j = 0; j < size; j++) { @@ -435,6 +444,7 @@ aaline_create_texture(struct aaline_stage *aaline) pipe_surface_reference(&surface, NULL); pipe->texture_update(pipe, aaline->texture, 0, (1 << level)); } + return TRUE; } @@ -443,7 +453,7 @@ aaline_create_texture(struct aaline_stage *aaline) * By using a mipmapped texture, we don't have to generate a different * texture image for each line size. */ -static void +static boolean aaline_create_sampler(struct aaline_stage *aaline) { struct pipe_sampler_state sampler; @@ -461,6 +471,10 @@ aaline_create_sampler(struct aaline_stage *aaline) sampler.max_lod = MAX_TEXTURE_LEVEL; aaline->sampler_cso = pipe->create_sampler_state(pipe, &sampler); + if (aaline->sampler_cso == NULL) + return FALSE; + + return TRUE; } @@ -468,13 +482,20 @@ aaline_create_sampler(struct aaline_stage *aaline) * When we're about to draw our first AA line in a batch, this function is * called to tell the driver to bind our modified fragment shader. */ -static void +static boolean bind_aaline_fragment_shader(struct aaline_stage *aaline) { - if (!aaline->fs->aaline_fs) { - generate_aaline_fs(aaline); - } + struct draw_context *draw = aaline->stage.draw; + + if (!aaline->fs->aaline_fs && + !generate_aaline_fs(aaline)) + return FALSE; + + draw->suspend_flushing = TRUE; aaline->driver_bind_fs_state(aaline->pipe, aaline->fs->aaline_fs); + draw->suspend_flushing = FALSE; + + return TRUE; } @@ -486,20 +507,6 @@ aaline_stage( struct draw_stage *stage ) } -static void -passthrough_point(struct draw_stage *stage, struct prim_header *header) -{ - stage->next->point(stage->next, header); -} - - -static void -passthrough_tri(struct draw_stage *stage, struct prim_header *header) -{ - stage->next->tri(stage->next, header); -} - - /** * Draw a wide line by drawing a quad, using geometry which will * fullfill GL's antialiased line requirements. @@ -637,7 +644,11 @@ aaline_first_line(struct draw_stage *stage, struct prim_header *header) /* * Bind (generate) our fragprog, sampler and texture */ - bind_aaline_fragment_shader(aaline); + if (!bind_aaline_fragment_shader(aaline)) { + stage->line = draw_pipe_passthrough_line; + stage->line(stage, header); + return; + } /* update vertex attrib info */ aaline->tex_slot = draw->num_vs_outputs; @@ -657,8 +668,10 @@ aaline_first_line(struct draw_stage *stage, struct prim_header *header) pipe_texture_reference(&aaline->state.texture[aaline->fs->sampler_unit], aaline->texture); + draw->suspend_flushing = TRUE; aaline->driver_bind_sampler_states(pipe, num_samplers, aaline->state.sampler); aaline->driver_set_sampler_textures(pipe, num_samplers, aaline->state.texture); + draw->suspend_flushing = FALSE; /* now really draw first line */ stage->line = aaline_line; @@ -676,14 +689,14 @@ aaline_flush(struct draw_stage *stage, unsigned flags) stage->line = aaline_first_line; stage->next->flush( stage->next, flags ); - /* restore original frag shader */ + /* restore original frag shader, texture, sampler state */ + draw->suspend_flushing = TRUE; aaline->driver_bind_fs_state(pipe, aaline->fs->driver_fs); - - /* XXX restore original texture, sampler state */ aaline->driver_bind_sampler_states(pipe, aaline->num_samplers, aaline->state.sampler); aaline->driver_set_sampler_textures(pipe, aaline->num_textures, aaline->state.texture); + draw->suspend_flushing = FALSE; draw->extra_vp_outputs.slot = 0; } @@ -700,10 +713,17 @@ static void aaline_destroy(struct draw_stage *stage) { struct aaline_stage *aaline = aaline_stage(stage); + uint i; + + for (i = 0; i < PIPE_MAX_SAMPLERS; i++) { + pipe_texture_reference(&aaline->state.texture[i], NULL); + } - aaline->pipe->delete_sampler_state(aaline->pipe, aaline->sampler_cso); + if (aaline->sampler_cso) + aaline->pipe->delete_sampler_state(aaline->pipe, aaline->sampler_cso); - pipe_texture_release(&aaline->texture); + if (aaline->texture) + pipe_texture_release(&aaline->texture); draw_free_temp_verts( stage ); @@ -715,19 +735,28 @@ static struct aaline_stage * draw_aaline_stage(struct draw_context *draw) { struct aaline_stage *aaline = CALLOC_STRUCT(aaline_stage); + if (aaline == NULL) + return NULL; - draw_alloc_temp_verts( &aaline->stage, 8 ); + if (!draw_alloc_temp_verts( &aaline->stage, 8 )) + goto fail; aaline->stage.draw = draw; aaline->stage.next = NULL; - aaline->stage.point = passthrough_point; + aaline->stage.point = draw_pipe_passthrough_point; aaline->stage.line = aaline_first_line; - aaline->stage.tri = passthrough_tri; + aaline->stage.tri = draw_pipe_passthrough_tri; aaline->stage.flush = aaline_flush; aaline->stage.reset_stipple_counter = aaline_reset_stipple_counter; aaline->stage.destroy = aaline_destroy; return aaline; + + fail: + if (aaline) + aaline_destroy(&aaline->stage); + + return NULL; } @@ -749,13 +778,13 @@ aaline_create_fs_state(struct pipe_context *pipe, { struct aaline_stage *aaline = aaline_stage_from_pipe(pipe); struct aaline_fragment_shader *aafs = CALLOC_STRUCT(aaline_fragment_shader); + if (aafs == NULL) + return NULL; - if (aafs) { - aafs->state = *fs; + aafs->state = *fs; - /* pass-through */ - aafs->driver_fs = aaline->driver_create_fs_state(aaline->pipe, fs); - } + /* pass-through */ + aafs->driver_fs = aaline->driver_create_fs_state(aaline->pipe, fs); return aafs; } @@ -766,6 +795,7 @@ aaline_bind_fs_state(struct pipe_context *pipe, void *fs) { struct aaline_stage *aaline = aaline_stage_from_pipe(pipe); struct aaline_fragment_shader *aafs = (struct aaline_fragment_shader *) fs; + /* save current */ aaline->fs = aafs; /* pass-through */ @@ -790,9 +820,11 @@ aaline_bind_sampler_states(struct pipe_context *pipe, unsigned num, void **sampler) { struct aaline_stage *aaline = aaline_stage_from_pipe(pipe); + /* save current */ memcpy(aaline->state.sampler, sampler, num * sizeof(void *)); aaline->num_samplers = num; + /* pass-through */ aaline->driver_bind_sampler_states(aaline->pipe, num, sampler); } @@ -809,6 +841,9 @@ aaline_set_sampler_textures(struct pipe_context *pipe, for (i = 0; i < num; i++) { pipe_texture_reference(&aaline->state.texture[i], texture[i]); } + for ( ; i < PIPE_MAX_SAMPLERS; i++) { + pipe_texture_reference(&aaline->state.texture[i], NULL); + } aaline->num_textures = num; /* pass-through */ @@ -821,7 +856,7 @@ aaline_set_sampler_textures(struct pipe_context *pipe, * into the draw module's pipeline. This will not be used if the * hardware has native support for AA lines. */ -void +boolean draw_install_aaline_stage(struct draw_context *draw, struct pipe_context *pipe) { struct aaline_stage *aaline; @@ -832,14 +867,17 @@ draw_install_aaline_stage(struct draw_context *draw, struct pipe_context *pipe) * Create / install AA line drawing / prim stage */ aaline = draw_aaline_stage( draw ); - assert(aaline); - draw->pipeline.aaline = &aaline->stage; + if (!aaline) + goto fail; aaline->pipe = pipe; /* create special texture, sampler state */ - aaline_create_texture(aaline); - aaline_create_sampler(aaline); + if (!aaline_create_texture(aaline)) + goto fail; + + if (!aaline_create_sampler(aaline)) + goto fail; /* save original driver functions */ aaline->driver_create_fs_state = pipe->create_fs_state; @@ -856,4 +894,16 @@ draw_install_aaline_stage(struct draw_context *draw, struct pipe_context *pipe) pipe->bind_sampler_states = aaline_bind_sampler_states; pipe->set_sampler_textures = aaline_set_sampler_textures; + + /* Install once everything is known to be OK: + */ + draw->pipeline.aaline = &aaline->stage; + + return TRUE; + + fail: + if (aaline) + aaline->stage.destroy( &aaline->stage ); + + return FALSE; } diff --git a/src/gallium/auxiliary/draw/draw_aapoint.c b/src/gallium/auxiliary/draw/draw_pipe_aapoint.c index fcebe3e7a0..122a48660a 100644 --- a/src/gallium/auxiliary/draw/draw_aapoint.c +++ b/src/gallium/auxiliary/draw/draw_pipe_aapoint.c @@ -48,7 +48,8 @@ #include "tgsi/util/tgsi_dump.h" #include "draw_context.h" -#include "draw_private.h" +#include "draw_vs.h" +#include "draw_pipe.h" /* @@ -480,10 +481,10 @@ aa_transform_inst(struct tgsi_transform_context *ctx, /** - * Generate the frag shader we'll use for drawing AA lines. + * Generate the frag shader we'll use for drawing AA points. * This will be the user's shader plus some texture/modulate instructions. */ -static void +static boolean generate_aapoint_fs(struct aapoint_stage *aapoint) { const struct pipe_shader_state *orig_fs = &aapoint->fs->state; @@ -494,6 +495,8 @@ generate_aapoint_fs(struct aapoint_stage *aapoint) aapoint_fs = *orig_fs; /* copy to init */ aapoint_fs.tokens = MALLOC(sizeof(struct tgsi_token) * MAX); + if (aapoint_fs.tokens == NULL) + return FALSE; memset(&transform, 0, sizeof(transform)); transform.colorOutput = -1; @@ -518,22 +521,33 @@ generate_aapoint_fs(struct aapoint_stage *aapoint) aapoint->fs->aapoint_fs = aapoint->driver_create_fs_state(aapoint->pipe, &aapoint_fs); + if (aapoint->fs->aapoint_fs == NULL) + return FALSE; aapoint->fs->generic_attrib = transform.maxGeneric + 1; + + return TRUE; } /** - * When we're about to draw our first AA line in a batch, this function is + * When we're about to draw our first AA point in a batch, this function is * called to tell the driver to bind our modified fragment shader. */ -static void +static boolean bind_aapoint_fragment_shader(struct aapoint_stage *aapoint) { - if (!aapoint->fs->aapoint_fs) { - generate_aapoint_fs(aapoint); - } + struct draw_context *draw = aapoint->stage.draw; + + if (!aapoint->fs->aapoint_fs && + !generate_aapoint_fs(aapoint)) + return FALSE; + + draw->suspend_flushing = TRUE; aapoint->driver_bind_fs_state(aapoint->pipe, aapoint->fs->aapoint_fs); + draw->suspend_flushing = FALSE; + + return TRUE; } @@ -545,18 +559,6 @@ aapoint_stage( struct draw_stage *stage ) } -static void -passthrough_line(struct draw_stage *stage, struct prim_header *header) -{ - stage->next->line(stage->next, header); -} - - -static void -passthrough_tri(struct draw_stage *stage, struct prim_header *header) -{ - stage->next->tri(stage->next, header); -} /** @@ -700,7 +702,7 @@ aapoint_first_point(struct draw_stage *stage, struct prim_header *header) } } - /* now really draw first line */ + /* now really draw first point */ stage->point = aapoint_point; stage->point(stage, header); } @@ -717,7 +719,9 @@ aapoint_flush(struct draw_stage *stage, unsigned flags) stage->next->flush( stage->next, flags ); /* restore original frag shader */ + draw->suspend_flushing = TRUE; aapoint->driver_bind_fs_state(pipe, aapoint->fs->driver_fs); + draw->suspend_flushing = FALSE; draw->extra_vp_outputs.slot = 0; } @@ -742,19 +746,29 @@ static struct aapoint_stage * draw_aapoint_stage(struct draw_context *draw) { struct aapoint_stage *aapoint = CALLOC_STRUCT(aapoint_stage); + if (aapoint == NULL) + goto fail; - draw_alloc_temp_verts( &aapoint->stage, 4 ); + if (!draw_alloc_temp_verts( &aapoint->stage, 4 )) + goto fail; aapoint->stage.draw = draw; aapoint->stage.next = NULL; aapoint->stage.point = aapoint_first_point; - aapoint->stage.line = passthrough_line; - aapoint->stage.tri = passthrough_tri; + aapoint->stage.line = draw_pipe_passthrough_line; + aapoint->stage.tri = draw_pipe_passthrough_tri; aapoint->stage.flush = aapoint_flush; aapoint->stage.reset_stipple_counter = aapoint_reset_stipple_counter; aapoint->stage.destroy = aapoint_destroy; return aapoint; + + fail: + if (aapoint) + aapoint_destroy(&aapoint->stage); + + return NULL; + } @@ -776,13 +790,13 @@ aapoint_create_fs_state(struct pipe_context *pipe, { struct aapoint_stage *aapoint = aapoint_stage_from_pipe(pipe); struct aapoint_fragment_shader *aafs = CALLOC_STRUCT(aapoint_fragment_shader); + if (aafs == NULL) + return NULL; - if (aafs) { - aafs->state = *fs; + aafs->state = *fs; - /* pass-through */ - aafs->driver_fs = aapoint->driver_create_fs_state(aapoint->pipe, fs); - } + /* pass-through */ + aafs->driver_fs = aapoint->driver_create_fs_state(aapoint->pipe, fs); return aafs; } @@ -817,7 +831,7 @@ aapoint_delete_fs_state(struct pipe_context *pipe, void *fs) * into the draw module's pipeline. This will not be used if the * hardware has native support for AA points. */ -void +boolean draw_install_aapoint_stage(struct draw_context *draw, struct pipe_context *pipe) { @@ -829,8 +843,8 @@ draw_install_aapoint_stage(struct draw_context *draw, * Create / install AA point drawing / prim stage */ aapoint = draw_aapoint_stage( draw ); - assert(aapoint); - draw->pipeline.aapoint = &aapoint->stage; + if (aapoint == NULL) + goto fail; aapoint->pipe = pipe; @@ -843,4 +857,14 @@ draw_install_aapoint_stage(struct draw_context *draw, pipe->create_fs_state = aapoint_create_fs_state; pipe->bind_fs_state = aapoint_bind_fs_state; pipe->delete_fs_state = aapoint_delete_fs_state; + + draw->pipeline.aapoint = &aapoint->stage; + + return TRUE; + + fail: + if (aapoint) + aapoint->stage.destroy( &aapoint->stage ); + + return FALSE; } diff --git a/src/gallium/auxiliary/draw/draw_clip.c b/src/gallium/auxiliary/draw/draw_pipe_clip.c index e24c5d8032..ce80c94163 100644 --- a/src/gallium/auxiliary/draw/draw_clip.c +++ b/src/gallium/auxiliary/draw/draw_pipe_clip.c @@ -35,8 +35,8 @@ #include "pipe/p_util.h" #include "pipe/p_shader_tokens.h" -#include "draw_context.h" -#include "draw_private.h" +#include "draw_vs.h" +#include "draw_pipe.h" #ifndef IS_NEGATIVE @@ -119,7 +119,7 @@ static void interp( const struct clipper *clip, */ { dst->clipmask = 0; - dst->edgeflag = 0; + dst->edgeflag = 0; /* will get overwritten later */ dst->pad = 0; dst->vertex_id = UNDEFINED_VERTEX_ID; } @@ -162,49 +162,50 @@ static void emit_poly( struct draw_stage *stage, struct prim_header header; unsigned i; + const ushort edge_first = DRAW_PIPE_EDGE_FLAG_2; + const ushort edge_middle = DRAW_PIPE_EDGE_FLAG_0; + const ushort edge_last = DRAW_PIPE_EDGE_FLAG_1; + /* later stages may need the determinant, but only the sign matters */ header.det = origPrim->det; + header.flags = DRAW_PIPE_RESET_STIPPLE | edge_first | edge_middle; + header.pad = 0; - for (i = 2; i < n; i++) { + for (i = 2; i < n; i++, header.flags = 0) { header.v[0] = inlist[i-1]; header.v[1] = inlist[i]; header.v[2] = inlist[0]; /* keep in v[2] for flatshading */ - - { - unsigned tmp1 = header.v[1]->edgeflag; - unsigned tmp2 = header.v[2]->edgeflag; - - if (i != n-1) header.v[1]->edgeflag = 0; - if (i != 2) header.v[2]->edgeflag = 0; - - header.edgeflags = ((header.v[0]->edgeflag << 0) | - (header.v[1]->edgeflag << 1) | - (header.v[2]->edgeflag << 2)); - - if (0) { - const struct draw_vertex_shader *vs = stage->draw->vertex_shader; - uint j, k; - debug_printf("Clipped tri:\n"); - for (j = 0; j < 3; j++) { - for (k = 0; k < vs->info.num_outputs; k++) { - debug_printf(" Vert %d: Attr %d: %f %f %f %f\n", j, k, - header.v[j]->data[k][0], - header.v[j]->data[k][1], - header.v[j]->data[k][2], - header.v[j]->data[k][3]); - } + + if (i == n-1) + header.flags |= edge_last; + + if (0) { + const struct draw_vertex_shader *vs = stage->draw->vertex_shader; + uint j, k; + debug_printf("Clipped tri:\n"); + for (j = 0; j < 3; j++) { + for (k = 0; k < vs->info.num_outputs; k++) { + debug_printf(" Vert %d: Attr %d: %f %f %f %f\n", j, k, + header.v[j]->data[k][0], + header.v[j]->data[k][1], + header.v[j]->data[k][2], + header.v[j]->data[k][3]); } } - - stage->next->tri( stage->next, &header ); - - header.v[1]->edgeflag = tmp1; - header.v[2]->edgeflag = tmp2; } + + stage->next->tri( stage->next, &header ); } } - +static INLINE float +dot4(const float *a, const float *b) +{ + return (a[0]*b[0] + + a[1]*b[1] + + a[2]*b[2] + + a[3]*b[3]); +} /* Clip a triangle against the viewport and user clip planes. @@ -486,8 +487,11 @@ static void clip_destroy( struct draw_stage *stage ) struct draw_stage *draw_clip_stage( struct draw_context *draw ) { struct clipper *clipper = CALLOC_STRUCT(clipper); + if (clipper == NULL) + goto fail; - draw_alloc_temp_verts( &clipper->stage, MAX_CLIPPED_VERTICES+1 ); + if (!draw_alloc_temp_verts( &clipper->stage, MAX_CLIPPED_VERTICES+1 )) + goto fail; clipper->stage.draw = draw; clipper->stage.point = clip_point; @@ -500,4 +504,10 @@ struct draw_stage *draw_clip_stage( struct draw_context *draw ) clipper->plane = draw->plane; return &clipper->stage; + + fail: + if (clipper) + clipper->stage.destroy( &clipper->stage ); + + return NULL; } diff --git a/src/gallium/auxiliary/draw/draw_cull.c b/src/gallium/auxiliary/draw/draw_pipe_cull.c index 8177b0ac86..87aaf1f85b 100644 --- a/src/gallium/auxiliary/draw/draw_cull.c +++ b/src/gallium/auxiliary/draw/draw_pipe_cull.c @@ -35,7 +35,7 @@ #include "pipe/p_util.h" #include "pipe/p_defines.h" -#include "draw_private.h" +#include "draw_pipe.h" struct cull_stage { @@ -95,20 +95,6 @@ static void cull_first_tri( struct draw_stage *stage, -static void cull_line( struct draw_stage *stage, - struct prim_header *header ) -{ - stage->next->line( stage->next, header ); -} - - -static void cull_point( struct draw_stage *stage, - struct prim_header *header ) -{ - stage->next->point( stage->next, header ); -} - - static void cull_flush( struct draw_stage *stage, unsigned flags ) { stage->tri = cull_first_tri; @@ -134,17 +120,26 @@ static void cull_destroy( struct draw_stage *stage ) struct draw_stage *draw_cull_stage( struct draw_context *draw ) { struct cull_stage *cull = CALLOC_STRUCT(cull_stage); + if (cull == NULL) + goto fail; - draw_alloc_temp_verts( &cull->stage, 0 ); + if (!draw_alloc_temp_verts( &cull->stage, 0 )) + goto fail; cull->stage.draw = draw; cull->stage.next = NULL; - cull->stage.point = cull_point; - cull->stage.line = cull_line; + cull->stage.point = draw_pipe_passthrough_point; + cull->stage.line = draw_pipe_passthrough_line; cull->stage.tri = cull_first_tri; cull->stage.flush = cull_flush; cull->stage.reset_stipple_counter = cull_reset_stipple_counter; cull->stage.destroy = cull_destroy; return &cull->stage; + + fail: + if (cull) + cull->stage.destroy( &cull->stage ); + + return NULL; } diff --git a/src/gallium/auxiliary/draw/draw_flatshade.c b/src/gallium/auxiliary/draw/draw_pipe_flatshade.c index af2cb05c98..09b68c4559 100644 --- a/src/gallium/auxiliary/draw/draw_flatshade.c +++ b/src/gallium/auxiliary/draw/draw_pipe_flatshade.c @@ -30,7 +30,8 @@ #include "pipe/p_util.h" #include "pipe/p_shader_tokens.h" -#include "draw_private.h" +#include "draw_vs.h" +#include "draw_pipe.h" /** subclass of draw_stage */ @@ -90,7 +91,8 @@ static void flatshade_tri_0( struct draw_stage *stage, struct prim_header tmp; tmp.det = header->det; - tmp.edgeflags = header->edgeflags; + tmp.flags = header->flags; + tmp.pad = header->pad; tmp.v[0] = header->v[0]; tmp.v[1] = dup_vert(stage, header->v[1], 0); tmp.v[2] = dup_vert(stage, header->v[2], 1); @@ -107,7 +109,8 @@ static void flatshade_tri_2( struct draw_stage *stage, struct prim_header tmp; tmp.det = header->det; - tmp.edgeflags = header->edgeflags; + tmp.flags = header->flags; + tmp.pad = header->pad; tmp.v[0] = dup_vert(stage, header->v[0], 0); tmp.v[1] = dup_vert(stage, header->v[1], 1); tmp.v[2] = header->v[2]; @@ -151,13 +154,6 @@ static void flatshade_line_1( struct draw_stage *stage, } -/* Flatshade point -- passthrough. - */ -static void flatshade_point( struct draw_stage *stage, - struct prim_header *header ) -{ - stage->next->point( stage->next, header ); -} static void flatshade_init_state( struct draw_stage *stage ) @@ -230,12 +226,15 @@ static void flatshade_destroy( struct draw_stage *stage ) struct draw_stage *draw_flatshade_stage( struct draw_context *draw ) { struct flat_stage *flatshade = CALLOC_STRUCT(flat_stage); + if (flatshade == NULL) + goto fail; - draw_alloc_temp_verts( &flatshade->stage, 2 ); + if (!draw_alloc_temp_verts( &flatshade->stage, 2 )) + goto fail; flatshade->stage.draw = draw; flatshade->stage.next = NULL; - flatshade->stage.point = flatshade_point; + flatshade->stage.point = draw_pipe_passthrough_point; flatshade->stage.line = flatshade_first_line; flatshade->stage.tri = flatshade_first_tri; flatshade->stage.flush = flatshade_flush; @@ -243,6 +242,12 @@ struct draw_stage *draw_flatshade_stage( struct draw_context *draw ) flatshade->stage.destroy = flatshade_destroy; return &flatshade->stage; + + fail: + if (flatshade) + flatshade->stage.destroy( &flatshade->stage ); + + return NULL; } diff --git a/src/gallium/auxiliary/draw/draw_offset.c b/src/gallium/auxiliary/draw/draw_pipe_offset.c index dbc676deae..ea6de8c571 100644 --- a/src/gallium/auxiliary/draw/draw_offset.c +++ b/src/gallium/auxiliary/draw/draw_pipe_offset.c @@ -33,7 +33,7 @@ */ #include "pipe/p_util.h" -#include "draw_private.h" +#include "draw_pipe.h" @@ -106,7 +106,8 @@ static void offset_tri( struct draw_stage *stage, struct prim_header tmp; tmp.det = header->det; - tmp.edgeflags = header->edgeflags; + tmp.flags = header->flags; + tmp.pad = header->pad; tmp.v[0] = dup_vert(stage, header->v[0], 0); tmp.v[1] = dup_vert(stage, header->v[1], 1); tmp.v[2] = dup_vert(stage, header->v[2], 2); @@ -129,18 +130,6 @@ static void offset_first_tri( struct draw_stage *stage, } -static void offset_line( struct draw_stage *stage, - struct prim_header *header ) -{ - stage->next->line( stage->next, header ); -} - - -static void offset_point( struct draw_stage *stage, - struct prim_header *header ) -{ - stage->next->point( stage->next, header ); -} static void offset_flush( struct draw_stage *stage, @@ -170,17 +159,25 @@ static void offset_destroy( struct draw_stage *stage ) struct draw_stage *draw_offset_stage( struct draw_context *draw ) { struct offset_stage *offset = CALLOC_STRUCT(offset_stage); + if (offset == NULL) + goto fail; draw_alloc_temp_verts( &offset->stage, 3 ); offset->stage.draw = draw; offset->stage.next = NULL; - offset->stage.point = offset_point; - offset->stage.line = offset_line; + offset->stage.point = draw_pipe_passthrough_point; + offset->stage.line = draw_pipe_passthrough_line; offset->stage.tri = offset_first_tri; offset->stage.flush = offset_flush; offset->stage.reset_stipple_counter = offset_reset_stipple_counter; offset->stage.destroy = offset_destroy; return &offset->stage; + + fail: + if (offset) + offset->stage.destroy( &offset->stage ); + + return NULL; } diff --git a/src/gallium/auxiliary/draw/draw_pstipple.c b/src/gallium/auxiliary/draw/draw_pipe_pstipple.c index 4dddb72906..2dfd1800d5 100644 --- a/src/gallium/auxiliary/draw/draw_pstipple.c +++ b/src/gallium/auxiliary/draw/draw_pipe_pstipple.c @@ -44,7 +44,7 @@ #include "tgsi/util/tgsi_dump.h" #include "draw_context.h" -#include "draw_private.h" +#include "draw_pipe.h" @@ -256,7 +256,7 @@ pstip_transform_inst(struct tgsi_transform_context *ctx, uint size = 4; immed = tgsi_default_full_immediate(); immed.Immediate.Size = 1 + size; /* one for the token itself */ - immed.u.ImmediateFloat32 = (struct tgsi_immediate_float32 *) value; + immed.u.Pointer = (void *) value; ctx->emit_immediate(ctx, &immed); } @@ -320,7 +320,7 @@ pstip_transform_inst(struct tgsi_transform_context *ctx, * Generate the frag shader we'll use for doing polygon stipple. * This will be the user's shader prefixed with a TEX and KIL instruction. */ -static void +static boolean generate_pstip_fs(struct pstip_stage *pstip) { const struct pipe_shader_state *orig_fs = &pstip->fs->state; @@ -332,6 +332,8 @@ generate_pstip_fs(struct pstip_stage *pstip) pstip_fs = *orig_fs; /* copy to init */ pstip_fs.tokens = MALLOC(sizeof(struct tgsi_token) * MAX); + if (pstip_fs.tokens == NULL) + return FALSE; memset(&transform, 0, sizeof(transform)); transform.wincoordInput = -1; @@ -355,6 +357,8 @@ generate_pstip_fs(struct pstip_stage *pstip) assert(pstip->fs->sampler_unit < PIPE_MAX_SAMPLERS); pstip->fs->pstip_fs = pstip->driver_create_fs_state(pstip->pipe, &pstip_fs); + + return TRUE; } @@ -404,7 +408,7 @@ pstip_update_texture(struct pstip_stage *pstip) /** * Create the texture map we'll use for stippling. */ -static void +static boolean pstip_create_texture(struct pstip_stage *pstip) { struct pipe_context *pipe = pstip->pipe; @@ -413,7 +417,7 @@ pstip_create_texture(struct pstip_stage *pstip) memset(&texTemp, 0, sizeof(texTemp)); texTemp.target = PIPE_TEXTURE_2D; - texTemp.format = PIPE_FORMAT_U_A8; /* XXX verify supported by driver! */ + texTemp.format = PIPE_FORMAT_A8_UNORM; /* XXX verify supported by driver! */ texTemp.last_level = 0; texTemp.width[0] = 32; texTemp.height[0] = 32; @@ -421,16 +425,17 @@ pstip_create_texture(struct pstip_stage *pstip) texTemp.cpp = 1; pstip->texture = screen->texture_create(screen, &texTemp); - assert(pstip->texture->refcount == 1); + if (pstip->texture == NULL) + return FALSE; + + return TRUE; } /** - * Create the sampler CSO that'll be used for antialiasing. - * By using a mipmapped texture, we don't have to generate a different - * texture image for each line size. + * Create the sampler CSO that'll be used for stippling. */ -static void +static boolean pstip_create_sampler(struct pstip_stage *pstip) { struct pipe_sampler_state sampler; @@ -448,24 +453,32 @@ pstip_create_sampler(struct pstip_stage *pstip) sampler.max_lod = 0.0f; pstip->sampler_cso = pipe->create_sampler_state(pipe, &sampler); + if (pstip->sampler_cso == NULL) + return FALSE; + + return TRUE; } /** - * When we're about to draw our first AA line in a batch, this function is - * called to tell the driver to bind our modified fragment shader. + * When we're about to draw our first stipple polygon in a batch, this function + * is called to tell the driver to bind our modified fragment shader. */ -static void +static boolean bind_pstip_fragment_shader(struct pstip_stage *pstip) { - if (!pstip->fs->pstip_fs) { - generate_pstip_fs(pstip); - } + struct draw_context *draw = pstip->stage.draw; + if (!pstip->fs->pstip_fs && + !generate_pstip_fs(pstip)) + return FALSE; + + draw->suspend_flushing = TRUE; pstip->driver_bind_fs_state(pstip->pipe, pstip->fs->pstip_fs); + draw->suspend_flushing = FALSE; + return TRUE; } - static INLINE struct pstip_stage * pstip_stage( struct draw_stage *stage ) { @@ -474,38 +487,22 @@ pstip_stage( struct draw_stage *stage ) static void -passthrough_point(struct draw_stage *stage, struct prim_header *header) -{ - stage->next->point(stage->next, header); -} - - -static void -passthrough_line(struct draw_stage *stage, struct prim_header *header) -{ - stage->next->line(stage->next, header); -} - - -static void -passthrough_tri(struct draw_stage *stage, struct prim_header *header) -{ - stage->next->tri(stage->next, header); -} - - - -static void pstip_first_tri(struct draw_stage *stage, struct prim_header *header) { struct pstip_stage *pstip = pstip_stage(stage); struct pipe_context *pipe = pstip->pipe; + struct draw_context *draw = stage->draw; uint num_samplers; assert(stage->draw->rasterizer->poly_stipple_enable); /* bind our fragprog */ - bind_pstip_fragment_shader(pstip); + if (!bind_pstip_fragment_shader(pstip)) { + stage->tri = draw_pipe_passthrough_tri; + stage->tri(stage, header); + return; + } + /* how many samplers? */ /* we'll use sampler/texture[pstip->sampler_unit] for the stipple */ @@ -519,11 +516,13 @@ pstip_first_tri(struct draw_stage *stage, struct prim_header *header) assert(num_samplers <= PIPE_MAX_SAMPLERS); + draw->suspend_flushing = TRUE; pstip->driver_bind_sampler_states(pipe, num_samplers, pstip->state.samplers); pstip->driver_set_sampler_textures(pipe, num_samplers, pstip->state.textures); + draw->suspend_flushing = FALSE; - /* now really draw first line */ - stage->tri = passthrough_tri; + /* now really draw first triangle */ + stage->tri = draw_pipe_passthrough_tri; stage->tri(stage, header); } @@ -531,21 +530,21 @@ pstip_first_tri(struct draw_stage *stage, struct prim_header *header) static void pstip_flush(struct draw_stage *stage, unsigned flags) { - /*struct draw_context *draw = stage->draw;*/ + struct draw_context *draw = stage->draw; struct pstip_stage *pstip = pstip_stage(stage); struct pipe_context *pipe = pstip->pipe; stage->tri = pstip_first_tri; stage->next->flush( stage->next, flags ); - /* restore original frag shader */ + /* restore original frag shader, texture, sampler state */ + draw->suspend_flushing = TRUE; pstip->driver_bind_fs_state(pipe, pstip->fs->driver_fs); - - /* XXX restore original texture, sampler state */ pstip->driver_bind_sampler_states(pipe, pstip->num_samplers, pstip->state.samplers); pstip->driver_set_sampler_textures(pipe, pstip->num_textures, pstip->state.textures); + draw->suspend_flushing = FALSE; } @@ -560,6 +559,11 @@ static void pstip_destroy(struct draw_stage *stage) { struct pstip_stage *pstip = pstip_stage(stage); + uint i; + + for (i = 0; i < PIPE_MAX_SAMPLERS; i++) { + pipe_texture_reference(&pstip->state.textures[i], NULL); + } pstip->pipe->delete_sampler_state(pstip->pipe, pstip->sampler_cso); @@ -579,8 +583,8 @@ draw_pstip_stage(struct draw_context *draw) pstip->stage.draw = draw; pstip->stage.next = NULL; - pstip->stage.point = passthrough_point; - pstip->stage.line = passthrough_line; + pstip->stage.point = draw_pipe_passthrough_point; + pstip->stage.line = draw_pipe_passthrough_line; pstip->stage.tri = pstip_first_tri; pstip->stage.flush = pstip_flush; pstip->stage.reset_stipple_counter = pstip_reset_stipple_counter; @@ -690,8 +694,10 @@ pstip_set_polygon_stipple(struct pipe_context *pipe, const struct pipe_poly_stipple *stipple) { struct pstip_stage *pstip = pstip_stage_from_pipe(pipe); + /* save current */ pstip->state.stipple = stipple; + /* pass-through */ pstip->driver_set_polygon_stipple(pstip->pipe, stipple); @@ -699,13 +705,12 @@ pstip_set_polygon_stipple(struct pipe_context *pipe, } - /** - * Called by drivers that want to install this AA line prim stage + * Called by drivers that want to install this polygon stipple stage * into the draw module's pipeline. This will not be used if the - * hardware has native support for AA lines. + * hardware has native support for polygon stipple. */ -void +boolean draw_install_pstipple_stage(struct draw_context *draw, struct pipe_context *pipe) { @@ -714,17 +719,22 @@ draw_install_pstipple_stage(struct draw_context *draw, pipe->draw = (void *) draw; /* - * Create / install AA line drawing / prim stage + * Create / install pgon stipple drawing / prim stage */ pstip = draw_pstip_stage( draw ); - assert(pstip); + if (pstip == NULL) + goto fail; + draw->pipeline.pstipple = &pstip->stage; pstip->pipe = pipe; /* create special texture, sampler state */ - pstip_create_texture(pstip); - pstip_create_sampler(pstip); + if (!pstip_create_texture(pstip)) + goto fail; + + if (!pstip_create_sampler(pstip)) + goto fail; /* save original driver functions */ pstip->driver_create_fs_state = pipe->create_fs_state; @@ -743,4 +753,12 @@ draw_install_pstipple_stage(struct draw_context *draw, pipe->bind_sampler_states = pstip_bind_sampler_states; pipe->set_sampler_textures = pstip_set_sampler_textures; pipe->set_polygon_stipple = pstip_set_polygon_stipple; + + return TRUE; + + fail: + if (pstip) + pstip->stage.destroy( &pstip->stage ); + + return FALSE; } diff --git a/src/gallium/auxiliary/draw/draw_stipple.c b/src/gallium/auxiliary/draw/draw_pipe_stipple.c index 506f33512c..3cbced362e 100644 --- a/src/gallium/auxiliary/draw/draw_stipple.c +++ b/src/gallium/auxiliary/draw/draw_pipe_stipple.c @@ -39,7 +39,7 @@ #include "pipe/p_util.h" #include "pipe/p_defines.h" #include "pipe/p_shader_tokens.h" -#include "draw_private.h" +#include "draw_pipe.h" /** Subclass of draw_stage */ @@ -135,6 +135,10 @@ stipple_line(struct draw_stage *stage, struct prim_header *header) float length = MAX2(dx, dy); int i; + if (header->flags & DRAW_PIPE_RESET_STIPPLE) + stipple->counter = 0; + + /* XXX ToDo: intead of iterating pixel-by-pixel, use a look-up table. */ for (i = 0; i < length; i++) { @@ -195,18 +199,6 @@ stipple_flush(struct draw_stage *stage, unsigned flags) } -static void -passthrough_point(struct draw_stage *stage, struct prim_header *header) -{ - stage->next->point( stage->next, header ); -} - - -static void -passthrough_tri(struct draw_stage *stage, struct prim_header *header) -{ - stage->next->tri(stage->next, header); -} static void @@ -228,9 +220,9 @@ struct draw_stage *draw_stipple_stage( struct draw_context *draw ) stipple->stage.draw = draw; stipple->stage.next = NULL; - stipple->stage.point = passthrough_point; + stipple->stage.point = draw_pipe_passthrough_point; stipple->stage.line = stipple_first_line; - stipple->stage.tri = passthrough_tri; + stipple->stage.tri = draw_pipe_passthrough_tri; stipple->stage.reset_stipple_counter = reset_stipple_counter; stipple->stage.flush = stipple_flush; stipple->stage.destroy = stipple_destroy; diff --git a/src/gallium/auxiliary/draw/draw_twoside.c b/src/gallium/auxiliary/draw/draw_pipe_twoside.c index 3debaac282..50872fdbe9 100644 --- a/src/gallium/auxiliary/draw/draw_twoside.c +++ b/src/gallium/auxiliary/draw/draw_pipe_twoside.c @@ -31,8 +31,8 @@ #include "pipe/p_util.h" #include "pipe/p_defines.h" #include "pipe/p_shader_tokens.h" -#include "draw_private.h" - +#include "draw_vs.h" +#include "draw_pipe.h" struct twoside_stage { struct draw_stage stage; @@ -85,7 +85,8 @@ static void twoside_tri( struct draw_stage *stage, struct prim_header tmp; tmp.det = header->det; - tmp.edgeflags = header->edgeflags; + tmp.flags = header->flags; + tmp.pad = header->pad; /* copy back attribs to front attribs */ tmp.v[0] = copy_bfc(twoside, header->v[0], 0); tmp.v[1] = copy_bfc(twoside, header->v[1], 1); @@ -99,21 +100,6 @@ static void twoside_tri( struct draw_stage *stage, } -static void twoside_line( struct draw_stage *stage, - struct prim_header *header ) -{ - /* pass-through */ - stage->next->line( stage->next, header ); -} - - -static void twoside_point( struct draw_stage *stage, - struct prim_header *header ) -{ - /* pass-through */ - stage->next->point( stage->next, header ); -} - static void twoside_first_tri( struct draw_stage *stage, struct prim_header *header ) @@ -187,17 +173,26 @@ static void twoside_destroy( struct draw_stage *stage ) struct draw_stage *draw_twoside_stage( struct draw_context *draw ) { struct twoside_stage *twoside = CALLOC_STRUCT(twoside_stage); + if (twoside == NULL) + goto fail; - draw_alloc_temp_verts( &twoside->stage, 3 ); + if (!draw_alloc_temp_verts( &twoside->stage, 3 )) + goto fail; twoside->stage.draw = draw; twoside->stage.next = NULL; - twoside->stage.point = twoside_point; - twoside->stage.line = twoside_line; + twoside->stage.point = draw_pipe_passthrough_point; + twoside->stage.line = draw_pipe_passthrough_line; twoside->stage.tri = twoside_first_tri; twoside->stage.flush = twoside_flush; twoside->stage.reset_stipple_counter = twoside_reset_stipple_counter; twoside->stage.destroy = twoside_destroy; return &twoside->stage; + + fail: + if (twoside) + twoside->stage.destroy( &twoside->stage ); + + return NULL; } diff --git a/src/gallium/auxiliary/draw/draw_unfilled.c b/src/gallium/auxiliary/draw/draw_pipe_unfilled.c index b07860cd9e..8f97fdedaa 100644 --- a/src/gallium/auxiliary/draw/draw_unfilled.c +++ b/src/gallium/auxiliary/draw/draw_pipe_unfilled.c @@ -36,6 +36,7 @@ #include "pipe/p_util.h" #include "pipe/p_defines.h" #include "draw_private.h" +#include "draw_pipe.h" struct unfilled_stage { @@ -82,9 +83,9 @@ static void points( struct draw_stage *stage, struct vertex_header *v1 = header->v[1]; struct vertex_header *v2 = header->v[2]; - if (header->edgeflags & 0x1) point( stage, v0 ); - if (header->edgeflags & 0x2) point( stage, v1 ); - if (header->edgeflags & 0x4) point( stage, v2 ); + if ((header->flags & DRAW_PIPE_EDGE_FLAG_0) && v0->edgeflag) point( stage, v0 ); + if ((header->flags & DRAW_PIPE_EDGE_FLAG_1) && v1->edgeflag) point( stage, v1 ); + if ((header->flags & DRAW_PIPE_EDGE_FLAG_2) && v2->edgeflag) point( stage, v2 ); } @@ -95,22 +96,22 @@ static void lines( struct draw_stage *stage, struct vertex_header *v1 = header->v[1]; struct vertex_header *v2 = header->v[2]; -#if 0 - assert(((header->edgeflags & 0x1) >> 0) == header->v[0]->edgeflag); - assert(((header->edgeflags & 0x2) >> 1) == header->v[1]->edgeflag); - assert(((header->edgeflags & 0x4) >> 2) == header->v[2]->edgeflag); -#endif + if (header->flags & DRAW_PIPE_RESET_STIPPLE) + stage->next->reset_stipple_counter( stage->next ); - if (header->edgeflags & 0x4) line( stage, v2, v0 ); - if (header->edgeflags & 0x1) line( stage, v0, v1 ); - if (header->edgeflags & 0x2) line( stage, v1, v2 ); + if ((header->flags & DRAW_PIPE_EDGE_FLAG_2) && v2->edgeflag) line( stage, v2, v0 ); + if ((header->flags & DRAW_PIPE_EDGE_FLAG_0) && v0->edgeflag) line( stage, v0, v1 ); + if ((header->flags & DRAW_PIPE_EDGE_FLAG_1) && v1->edgeflag) line( stage, v1, v2 ); } /* Unfilled tri: * * Note edgeflags in the vertex struct is not sufficient as we will - * need to manipulate them when decomposing primitives??? + * need to manipulate them when decomposing primitives. + * + * We currently keep the vertex edgeflag and primitive edgeflag mask + * separate until the last possible moment. */ static void unfilled_tri( struct draw_stage *stage, struct prim_header *header ) @@ -147,19 +148,6 @@ static void unfilled_first_tri( struct draw_stage *stage, } -static void unfilled_line( struct draw_stage *stage, - struct prim_header *header ) -{ - stage->next->line( stage->next, header ); -} - - -static void unfilled_point( struct draw_stage *stage, - struct prim_header *header ) -{ - stage->next->point( stage->next, header ); -} - static void unfilled_flush( struct draw_stage *stage, unsigned flags ) @@ -189,18 +177,27 @@ static void unfilled_destroy( struct draw_stage *stage ) struct draw_stage *draw_unfilled_stage( struct draw_context *draw ) { struct unfilled_stage *unfilled = CALLOC_STRUCT(unfilled_stage); + if (unfilled == NULL) + goto fail; - draw_alloc_temp_verts( &unfilled->stage, 0 ); + if (!draw_alloc_temp_verts( &unfilled->stage, 0 )) + goto fail; unfilled->stage.draw = draw; unfilled->stage.next = NULL; unfilled->stage.tmp = NULL; - unfilled->stage.point = unfilled_point; - unfilled->stage.line = unfilled_line; + unfilled->stage.point = draw_pipe_passthrough_point; + unfilled->stage.line = draw_pipe_passthrough_line; unfilled->stage.tri = unfilled_first_tri; unfilled->stage.flush = unfilled_flush; unfilled->stage.reset_stipple_counter = unfilled_reset_stipple_counter; unfilled->stage.destroy = unfilled_destroy; return &unfilled->stage; + + fail: + if (unfilled) + unfilled->stage.destroy( &unfilled->stage ); + + return NULL; } diff --git a/src/gallium/auxiliary/draw/draw_pipe_util.c b/src/gallium/auxiliary/draw/draw_pipe_util.c new file mode 100644 index 0000000000..04438f4dd0 --- /dev/null +++ b/src/gallium/auxiliary/draw/draw_pipe_util.c @@ -0,0 +1,137 @@ +/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + + /* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#include "pipe/p_util.h" +#include "draw/draw_private.h" +#include "draw/draw_pipe.h" + + + +void +draw_pipe_passthrough_point(struct draw_stage *stage, struct prim_header *header) +{ + stage->next->point(stage->next, header); +} + +void +draw_pipe_passthrough_line(struct draw_stage *stage, struct prim_header *header) +{ + stage->next->line(stage->next, header); +} + +void +draw_pipe_passthrough_tri(struct draw_stage *stage, struct prim_header *header) +{ + stage->next->tri(stage->next, header); +} + + + + + +/* This is only used for temporary verts. + */ +#define MAX_VERTEX_SIZE ((2 + PIPE_MAX_SHADER_OUTPUTS) * 4 * sizeof(float)) + + +/** + * Allocate space for temporary post-transform vertices, such as for clipping. + */ +boolean draw_alloc_temp_verts( struct draw_stage *stage, unsigned nr ) +{ + assert(!stage->tmp); + + stage->tmp = NULL; + stage->nr_tmps = nr; + + if (nr != 0) + { + unsigned i; + ubyte *store = (ubyte *) MALLOC( MAX_VERTEX_SIZE * nr ); + + if (store == NULL) + return FALSE; + + stage->tmp = (struct vertex_header **) MALLOC( sizeof(struct vertex_header *) * nr ); + if (stage->tmp == NULL) { + FREE(store); + return FALSE; + } + + for (i = 0; i < nr; i++) + stage->tmp[i] = (struct vertex_header *)(store + i * MAX_VERTEX_SIZE); + } + + return TRUE; +} + + +void draw_free_temp_verts( struct draw_stage *stage ) +{ + if (stage->tmp) { + FREE( stage->tmp[0] ); + FREE( stage->tmp ); + stage->tmp = NULL; + } +} + + +/* Reset vertex ids. This is basically a type of flush. + * + * Called only from draw_pipe_vbuf.c + */ +void draw_reset_vertex_ids(struct draw_context *draw) +{ + struct draw_stage *stage = draw->pipeline.first; + + while (stage) { + unsigned i; + + for (i = 0; i < stage->nr_tmps; i++) + stage->tmp[i]->vertex_id = UNDEFINED_VERTEX_ID; + + stage = stage->next; + } + + if (draw->pipeline.verts) + { + unsigned i; + char *verts = draw->pipeline.verts; + unsigned stride = draw->pipeline.vertex_stride; + + for (i = 0; i < draw->pipeline.vertex_count; i++) { + ((struct vertex_header *)verts)->vertex_id = UNDEFINED_VERTEX_ID; + verts += stride; + } + } +} + diff --git a/src/gallium/auxiliary/draw/draw_validate.c b/src/gallium/auxiliary/draw/draw_pipe_validate.c index e163e078f0..6be1d369c3 100644 --- a/src/gallium/auxiliary/draw/draw_validate.c +++ b/src/gallium/auxiliary/draw/draw_pipe_validate.c @@ -31,6 +31,8 @@ #include "pipe/p_util.h" #include "pipe/p_defines.h" #include "draw_private.h" +#include "draw_pipe.h" +#include "draw_context.h" static boolean points( unsigned prim ) { @@ -56,7 +58,8 @@ static boolean triangles( unsigned prim ) * pipeline stages. */ boolean -draw_need_pipeline(const struct draw_context *draw, +draw_need_pipeline(const struct draw_context *draw, + const struct pipe_rasterizer_state *rasterizer, unsigned int prim ) { /* Don't have to worry about triangles turning into lines/points @@ -66,30 +69,30 @@ draw_need_pipeline(const struct draw_context *draw, if (lines(prim)) { /* line stipple */ - if (draw->rasterizer->line_stipple_enable && draw->line_stipple) + if (rasterizer->line_stipple_enable && draw->pipeline.line_stipple) return TRUE; /* wide lines */ - if (draw->rasterizer->line_width > draw->wide_line_threshold) + if (rasterizer->line_width > draw->pipeline.wide_line_threshold) return TRUE; /* AA lines */ - if (draw->rasterizer->line_smooth && draw->pipeline.aaline) + if (rasterizer->line_smooth && draw->pipeline.aaline) return TRUE; } if (points(prim)) { /* large points */ - if (draw->rasterizer->point_size > draw->wide_point_threshold) + if (rasterizer->point_size > draw->pipeline.wide_point_threshold) return TRUE; /* AA points */ - if (draw->rasterizer->point_smooth && draw->pipeline.aapoint) + if (rasterizer->point_smooth && draw->pipeline.aapoint) return TRUE; /* point sprites */ - if (draw->rasterizer->point_sprite && draw->point_sprite) + if (rasterizer->point_sprite && draw->pipeline.point_sprite) return TRUE; } @@ -97,20 +100,20 @@ draw_need_pipeline(const struct draw_context *draw, if (triangles(prim)) { /* polygon stipple */ - if (draw->rasterizer->poly_stipple_enable && draw->pipeline.pstipple) + if (rasterizer->poly_stipple_enable && draw->pipeline.pstipple) return TRUE; /* unfilled polygons */ - if (draw->rasterizer->fill_cw != PIPE_POLYGON_MODE_FILL || - draw->rasterizer->fill_ccw != PIPE_POLYGON_MODE_FILL) + if (rasterizer->fill_cw != PIPE_POLYGON_MODE_FILL || + rasterizer->fill_ccw != PIPE_POLYGON_MODE_FILL) return TRUE; /* polygon offset */ - if (draw->rasterizer->offset_cw || draw->rasterizer->offset_ccw) + if (rasterizer->offset_cw || rasterizer->offset_ccw) return TRUE; /* two-side lighting */ - if (draw->rasterizer->light_twoside) + if (rasterizer->light_twoside) return TRUE; } @@ -119,7 +122,7 @@ draw_need_pipeline(const struct draw_context *draw, * * Generally this isn't a reason to require the pipeline, though. * - if (draw->rasterizer->cull_mode) + if (rasterizer->cull_mode) return TRUE; */ @@ -145,15 +148,15 @@ static struct draw_stage *validate_pipeline( struct draw_stage *stage ) stage->next = next; /* drawing wide lines? */ - wide_lines = (draw->rasterizer->line_width > draw->wide_line_threshold + wide_lines = (draw->rasterizer->line_width > draw->pipeline.wide_line_threshold && !draw->rasterizer->line_smooth); /* drawing large points? */ - if (draw->rasterizer->point_sprite && draw->point_sprite) + if (draw->rasterizer->point_sprite && draw->pipeline.point_sprite) wide_points = TRUE; else if (draw->rasterizer->point_smooth && draw->pipeline.aapoint) wide_points = FALSE; - else if (draw->rasterizer->point_size > draw->wide_point_threshold) + else if (draw->rasterizer->point_size > draw->pipeline.wide_point_threshold) wide_points = TRUE; else wide_points = FALSE; @@ -186,7 +189,7 @@ static struct draw_stage *validate_pipeline( struct draw_stage *stage ) next = draw->pipeline.wide_point; } - if (draw->rasterizer->line_stipple_enable && draw->line_stipple) { + if (draw->rasterizer->line_stipple_enable && draw->pipeline.line_stipple) { draw->pipeline.stipple->next = next; next = draw->pipeline.stipple; precalc_flat = 1; /* only needed for lines really */ @@ -238,7 +241,7 @@ static struct draw_stage *validate_pipeline( struct draw_stage *stage ) /* Clip stage */ - if (!draw->rasterizer->bypass_clipping) + if (!draw->bypass_clipping) { draw->pipeline.clip->next = next; next = draw->pipeline.clip; @@ -298,6 +301,8 @@ static void validate_destroy( struct draw_stage *stage ) struct draw_stage *draw_validate_stage( struct draw_context *draw ) { struct draw_stage *stage = CALLOC_STRUCT(draw_stage); + if (stage == NULL) + return NULL; stage->draw = draw; stage->next = NULL; diff --git a/src/gallium/auxiliary/draw/draw_vbuf.c b/src/gallium/auxiliary/draw/draw_pipe_vbuf.c index f83b441e93..86a7d1c730 100644 --- a/src/gallium/auxiliary/draw/draw_vbuf.c +++ b/src/gallium/auxiliary/draw/draw_pipe_vbuf.c @@ -40,7 +40,9 @@ #include "draw_vbuf.h" #include "draw_private.h" #include "draw_vertex.h" -#include "draw_vf.h" +#include "draw_pipe.h" +#include "translate/translate.h" +#include "translate/translate_cache.h" /** @@ -56,7 +58,7 @@ struct vbuf_stage { /** Vertex size in bytes */ unsigned vertex_size; - struct draw_vertex_fetch *vf; + struct translate *translate; /* FIXME: we have no guarantee that 'unsigned' is 32bit */ @@ -71,8 +73,11 @@ struct vbuf_stage { unsigned max_indices; unsigned nr_indices; - /** Pipe primitive */ - unsigned prim; + /* Cache point size somewhere it's address won't change: + */ + float point_size; + + struct translate_cache *cache; }; @@ -113,68 +118,6 @@ check_space( struct vbuf_stage *vbuf, unsigned nr ) } -#if 0 -static INLINE void -dump_emitted_vertex(const struct vertex_info *vinfo, const uint8_t *data) -{ - assert(vinfo == vbuf->render->get_vertex_info(vbuf->render)); - unsigned i, j, k; - - for (i = 0; i < vinfo->num_attribs; i++) { - j = vinfo->src_index[i]; - switch (vinfo->emit[i]) { - case EMIT_OMIT: - debug_printf("EMIT_OMIT:"); - break; - case EMIT_ALL: - assert(i == 0); - assert(j == 0); - debug_printf("EMIT_ALL:\t"); - for(k = 0; k < vinfo->size*4; ++k) - debug_printf("%02x ", *data++); - break; - case EMIT_1F: - debug_printf("EMIT_1F:\t"); - debug_printf("%f ", *(float *)data); data += sizeof(float); - break; - case EMIT_1F_PSIZE: - debug_printf("EMIT_1F_PSIZE:\t"); - debug_printf("%f ", *(float *)data); data += sizeof(float); - break; - case EMIT_2F: - debug_printf("EMIT_2F:\t"); - debug_printf("%f ", *(float *)data); data += sizeof(float); - debug_printf("%f ", *(float *)data); data += sizeof(float); - break; - case EMIT_3F: - debug_printf("EMIT_3F:\t"); - debug_printf("%f ", *(float *)data); data += sizeof(float); - debug_printf("%f ", *(float *)data); data += sizeof(float); - debug_printf("%f ", *(float *)data); data += sizeof(float); - data += sizeof(float); - break; - case EMIT_4F: - debug_printf("EMIT_4F:\t"); - debug_printf("%f ", *(float *)data); data += sizeof(float); - debug_printf("%f ", *(float *)data); data += sizeof(float); - debug_printf("%f ", *(float *)data); data += sizeof(float); - debug_printf("%f ", *(float *)data); data += sizeof(float); - break; - case EMIT_4UB: - debug_printf("EMIT_4UB:\t"); - debug_printf("%u ", *data++); - debug_printf("%u ", *data++); - debug_printf("%u ", *data++); - debug_printf("%u ", *data++); - break; - default: - assert(0); - } - debug_printf("\n"); - } - debug_printf("\n"); -} -#endif /** @@ -184,109 +127,25 @@ dump_emitted_vertex(const struct vertex_info *vinfo, const uint8_t *data) * have a couple of slots at the beginning (1-dword header, 4-dword * clip pos) that we ignore here. We only use the vertex->data[] fields. */ -static INLINE void +static INLINE ushort emit_vertex( struct vbuf_stage *vbuf, struct vertex_header *vertex ) { -#if 0 - debug_printf("emit vertex %d to %p\n", - vbuf->nr_vertices, vbuf->vertex_ptr); -#endif - - if(vertex->vertex_id != UNDEFINED_VERTEX_ID) { - if(vertex->vertex_id < vbuf->nr_vertices) - return; - else - debug_printf("Bad vertex id 0x%04x (>= 0x%04x)\n", - vertex->vertex_id, vbuf->nr_vertices); - return; - } - - vertex->vertex_id = vbuf->nr_vertices++; - - if(!vbuf->vf) { - const struct vertex_info *vinfo = vbuf->vinfo; - uint i; - uint count = 0; /* for debug/sanity */ + if(vertex->vertex_id == UNDEFINED_VERTEX_ID) { + /* Hmm - vertices are emitted one at a time - better make sure + * set_buffer is efficient. Consider a special one-shot mode for + * translate. + */ + vbuf->translate->set_buffer(vbuf->translate, 0, vertex->data[0], 0); + vbuf->translate->run(vbuf->translate, 0, 1, vbuf->vertex_ptr); + + if (0) draw_dump_emitted_vertex(vbuf->vinfo, (uint8_t *)vbuf->vertex_ptr); - assert(vinfo == vbuf->render->get_vertex_info(vbuf->render)); - - for (i = 0; i < vinfo->num_attribs; i++) { - uint j = vinfo->src_index[i]; - switch (vinfo->emit[i]) { - case EMIT_OMIT: - /* no-op */ - break; - case EMIT_ALL: - /* just copy the whole vertex as-is to the vbuf */ - assert(i == 0); - assert(j == 0); - memcpy(vbuf->vertex_ptr, vertex, vinfo->size * 4); - vbuf->vertex_ptr += vinfo->size; - count += vinfo->size; - break; - case EMIT_HEADER: - memcpy(vbuf->vertex_ptr, vertex, sizeof(*vertex)); - *vbuf->vertex_ptr += sizeof(*vertex) / 4; - count += sizeof(*vertex) / 4; - break; - case EMIT_1F: - *vbuf->vertex_ptr++ = fui(vertex->data[j][0]); - count++; - break; - case EMIT_1F_PSIZE: - *vbuf->vertex_ptr++ = fui(vbuf->stage.draw->rasterizer->point_size); - count++; - break; - case EMIT_2F: - *vbuf->vertex_ptr++ = fui(vertex->data[j][0]); - *vbuf->vertex_ptr++ = fui(vertex->data[j][1]); - count += 2; - break; - case EMIT_3F: - *vbuf->vertex_ptr++ = fui(vertex->data[j][0]); - *vbuf->vertex_ptr++ = fui(vertex->data[j][1]); - *vbuf->vertex_ptr++ = fui(vertex->data[j][2]); - count += 3; - break; - case EMIT_4F: - *vbuf->vertex_ptr++ = fui(vertex->data[j][0]); - *vbuf->vertex_ptr++ = fui(vertex->data[j][1]); - *vbuf->vertex_ptr++ = fui(vertex->data[j][2]); - *vbuf->vertex_ptr++ = fui(vertex->data[j][3]); - count += 4; - break; - case EMIT_4UB: - *vbuf->vertex_ptr++ = pack_ub4(float_to_ubyte( vertex->data[j][2] ), - float_to_ubyte( vertex->data[j][1] ), - float_to_ubyte( vertex->data[j][0] ), - float_to_ubyte( vertex->data[j][3] )); - count += 1; - break; - default: - assert(0); - } - } - assert(count == vinfo->size); -#if 0 - { - static float data[256]; - draw_vf_emit_vertex(vbuf->vf, vertex, data); - if(memcmp((uint8_t *)vbuf->vertex_ptr - vbuf->vertex_size, data, vbuf->vertex_size)) { - debug_printf("With VF:\n"); - dump_emitted_vertex(vbuf->vinfo, (uint8_t *)data); - debug_printf("Without VF:\n"); - dump_emitted_vertex(vbuf->vinfo, (uint8_t *)vbuf->vertex_ptr - vbuf->vertex_size); - assert(0); - } - } -#endif - } - else { - draw_vf_emit_vertex(vbuf->vf, vertex, vbuf->vertex_ptr); - vbuf->vertex_ptr += vbuf->vertex_size/4; + vertex->vertex_id = vbuf->nr_vertices++; } + + return vertex->vertex_id; } @@ -300,9 +159,7 @@ vbuf_tri( struct draw_stage *stage, check_space( vbuf, 3 ); for (i = 0; i < 3; i++) { - emit_vertex( vbuf, prim->v[i] ); - - vbuf->indices[vbuf->nr_indices++] = (ushort) prim->v[i]->vertex_id; + vbuf->indices[vbuf->nr_indices++] = emit_vertex( vbuf, prim->v[i] ); } } @@ -317,9 +174,7 @@ vbuf_line( struct draw_stage *stage, check_space( vbuf, 2 ); for (i = 0; i < 2; i++) { - emit_vertex( vbuf, prim->v[i] ); - - vbuf->indices[vbuf->nr_indices++] = (ushort) prim->v[i]->vertex_id; + vbuf->indices[vbuf->nr_indices++] = emit_vertex( vbuf, prim->v[i] ); } } @@ -332,43 +187,110 @@ vbuf_point( struct draw_stage *stage, check_space( vbuf, 1 ); - emit_vertex( vbuf, prim->v[0] ); - - vbuf->indices[vbuf->nr_indices++] = (ushort) prim->v[0]->vertex_id; + vbuf->indices[vbuf->nr_indices++] = emit_vertex( vbuf, prim->v[0] ); } + + /** * Set the prim type for subsequent vertices. * This may result in a new vertex size. The existing vbuffer (if any) * will be flushed if needed and a new one allocated. */ static void -vbuf_set_prim( struct vbuf_stage *vbuf, uint newprim ) +vbuf_set_prim( struct vbuf_stage *vbuf, uint prim ) { - const struct vertex_info *vinfo; - unsigned vertex_size; - - assert(newprim == PIPE_PRIM_POINTS || - newprim == PIPE_PRIM_LINES || - newprim == PIPE_PRIM_TRIANGLES); + struct translate_key hw_key; + unsigned dst_offset; + unsigned i; - vbuf->prim = newprim; - vbuf->render->set_primitive(vbuf->render, newprim); + vbuf->render->set_primitive(vbuf->render, prim); - vinfo = vbuf->render->get_vertex_info(vbuf->render); - vertex_size = vinfo->size * sizeof(float); + /* Must do this after set_primitive() above: + * + * XXX: need some state managment to track when this needs to be + * recalculated. The driver should tell us whether there was a + * state change. + */ + vbuf->vinfo = vbuf->render->get_vertex_info(vbuf->render); - if (vertex_size != vbuf->vertex_size) + if (vbuf->vertex_size != vbuf->vinfo->size * sizeof(float)) { vbuf_flush_vertices(vbuf); + vbuf->vertex_size = vbuf->vinfo->size * sizeof(float); + } - vbuf->vinfo = vinfo; - vbuf->vertex_size = vertex_size; - if(vbuf->vf) - draw_vf_set_vertex_info(vbuf->vf, - vbuf->vinfo, - vbuf->stage.draw->rasterizer->point_size); - + /* Translate from pipeline vertices to hw vertices. + */ + dst_offset = 0; + + for (i = 0; i < vbuf->vinfo->num_attribs; i++) { + unsigned emit_sz = 0; + unsigned src_buffer = 0; + unsigned output_format; + unsigned src_offset = (vbuf->vinfo->src_index[i] * 4 * sizeof(float) ); + + switch (vbuf->vinfo->emit[i]) { + case EMIT_4F: + output_format = PIPE_FORMAT_R32G32B32A32_FLOAT; + emit_sz = 4 * sizeof(float); + break; + case EMIT_3F: + output_format = PIPE_FORMAT_R32G32B32_FLOAT; + emit_sz = 3 * sizeof(float); + break; + case EMIT_2F: + output_format = PIPE_FORMAT_R32G32_FLOAT; + emit_sz = 2 * sizeof(float); + break; + case EMIT_1F: + output_format = PIPE_FORMAT_R32_FLOAT; + emit_sz = 1 * sizeof(float); + break; + case EMIT_1F_PSIZE: + output_format = PIPE_FORMAT_R32_FLOAT; + emit_sz = 1 * sizeof(float); + src_buffer = 1; + src_offset = 0; + break; + case EMIT_4UB: + output_format = PIPE_FORMAT_B8G8R8A8_UNORM; + emit_sz = 4 * sizeof(ubyte); + break; + default: + assert(0); + output_format = PIPE_FORMAT_NONE; + emit_sz = 0; + break; + } + + hw_key.element[i].input_format = PIPE_FORMAT_R32G32B32A32_FLOAT; + hw_key.element[i].input_buffer = src_buffer; + hw_key.element[i].input_offset = src_offset; + hw_key.element[i].output_format = output_format; + hw_key.element[i].output_offset = dst_offset; + + dst_offset += emit_sz; + } + + hw_key.nr_elements = vbuf->vinfo->num_attribs; + hw_key.output_stride = vbuf->vinfo->size * 4; + + /* Don't bother with caching at this stage: + */ + if (!vbuf->translate || + translate_key_compare(&vbuf->translate->key, &hw_key) != 0) + { + translate_key_sanitize(&hw_key); + vbuf->translate = translate_cache_find(vbuf->cache, &hw_key); + + vbuf->translate->set_buffer(vbuf->translate, 1, &vbuf->point_size, 0); + } + + vbuf->point_size = vbuf->stage.draw->rasterizer->point_size; + + /* Allocate new buffer? + */ if (!vbuf->vertices) vbuf_alloc_vertices(vbuf); } @@ -422,29 +344,9 @@ vbuf_flush_indices( struct vbuf_stage *vbuf ) assert((uint) (vbuf->vertex_ptr - vbuf->vertices) == vbuf->nr_vertices * vbuf->vertex_size / sizeof(unsigned)); - switch(vbuf->prim) { - case PIPE_PRIM_POINTS: - break; - case PIPE_PRIM_LINES: - assert(vbuf->nr_indices % 2 == 0); - break; - case PIPE_PRIM_TRIANGLES: - assert(vbuf->nr_indices % 3 == 0); - break; - default: - assert(0); - } - vbuf->render->draw(vbuf->render, vbuf->indices, vbuf->nr_indices); vbuf->nr_indices = 0; - - /* don't need to reset point/line/tri functions */ -#if 0 - stage->point = vbuf_first_point; - stage->line = vbuf_first_line; - stage->tri = vbuf_first_tri; -#endif } @@ -486,8 +388,8 @@ vbuf_alloc_vertices( struct vbuf_stage *vbuf ) /* Allocate a new vertex buffer */ vbuf->max_vertices = vbuf->render->max_vertex_buffer_bytes / vbuf->vertex_size; vbuf->vertices = (uint *) vbuf->render->allocate_vertices(vbuf->render, - (ushort) vbuf->vertex_size, - (ushort) vbuf->max_vertices); + (ushort) vbuf->vertex_size, + (ushort) vbuf->max_vertices); vbuf->vertex_ptr = vbuf->vertices; } @@ -525,12 +427,15 @@ static void vbuf_destroy( struct draw_stage *stage ) if(vbuf->indices) align_free( vbuf->indices ); - if(vbuf->vf) - draw_vf_destroy( vbuf->vf ); + if(vbuf->translate) + vbuf->translate->release( vbuf->translate ); if (vbuf->render) vbuf->render->destroy( vbuf->render ); + if (vbuf->cache) + translate_cache_destroy(vbuf->cache); + FREE( stage ); } @@ -542,9 +447,8 @@ struct draw_stage *draw_vbuf_stage( struct draw_context *draw, struct vbuf_render *render ) { struct vbuf_stage *vbuf = CALLOC_STRUCT(vbuf_stage); - - if(!vbuf) - return NULL; + if (vbuf == NULL) + goto fail; vbuf->stage.draw = draw; vbuf->stage.point = vbuf_first_point; @@ -555,21 +459,27 @@ struct draw_stage *draw_vbuf_stage( struct draw_context *draw, vbuf->stage.destroy = vbuf_destroy; vbuf->render = render; + vbuf->max_indices = MAX2(render->max_indices, UNDEFINED_VERTEX_ID-1); - assert(render->max_indices < UNDEFINED_VERTEX_ID); - vbuf->max_indices = render->max_indices; - vbuf->indices = (ushort *) - align_malloc( vbuf->max_indices * sizeof(vbuf->indices[0]), 16 ); - if(!vbuf->indices) - vbuf_destroy(&vbuf->stage); + vbuf->indices = (ushort *) align_malloc( vbuf->max_indices * + sizeof(vbuf->indices[0]), + 16 ); + if (!vbuf->indices) + goto fail; + + vbuf->cache = translate_cache_create(); + if (!vbuf->cache) + goto fail; + vbuf->vertices = NULL; vbuf->vertex_ptr = vbuf->vertices; - - vbuf->prim = ~0; - - if(!GETENV("GALLIUM_NOVF")) - vbuf->vf = draw_vf_create(); return &vbuf->stage; + + fail: + if (vbuf) + vbuf_destroy(&vbuf->stage); + + return NULL; } diff --git a/src/gallium/auxiliary/draw/draw_wide_line.c b/src/gallium/auxiliary/draw/draw_pipe_wide_line.c index 9a168ce8bd..878c9c7169 100644 --- a/src/gallium/auxiliary/draw/draw_wide_line.c +++ b/src/gallium/auxiliary/draw/draw_pipe_wide_line.c @@ -32,6 +32,7 @@ #include "pipe/p_defines.h" #include "pipe/p_shader_tokens.h" #include "draw_private.h" +#include "draw_pipe.h" struct wideline_stage { @@ -48,19 +49,6 @@ static INLINE struct wideline_stage *wideline_stage( struct draw_stage *stage ) } -static void wideline_point( struct draw_stage *stage, - struct prim_header *header ) -{ - stage->next->point( stage->next, header ); -} - - -static void wideline_tri( struct draw_stage *stage, - struct prim_header *header ) -{ - stage->next->tri(stage->next, header); -} - /** * Draw a wide line by drawing a quad (two triangles). @@ -179,9 +167,9 @@ struct draw_stage *draw_wide_line_stage( struct draw_context *draw ) wide->stage.draw = draw; wide->stage.next = NULL; - wide->stage.point = wideline_point; + wide->stage.point = draw_pipe_passthrough_point; wide->stage.line = wideline_line; - wide->stage.tri = wideline_tri; + wide->stage.tri = draw_pipe_passthrough_tri; wide->stage.flush = wideline_flush; wide->stage.reset_stipple_counter = wideline_reset_stipple_counter; wide->stage.destroy = wideline_destroy; diff --git a/src/gallium/auxiliary/draw/draw_wide_point.c b/src/gallium/auxiliary/draw/draw_pipe_wide_point.c index 6fc7c9fcd7..ed08573382 100644 --- a/src/gallium/auxiliary/draw/draw_wide_point.c +++ b/src/gallium/auxiliary/draw/draw_pipe_wide_point.c @@ -31,7 +31,8 @@ #include "pipe/p_util.h" #include "pipe/p_defines.h" #include "pipe/p_shader_tokens.h" -#include "draw_private.h" +#include "draw_vs.h" +#include "draw_pipe.h" struct widepoint_stage { @@ -60,23 +61,6 @@ widepoint_stage( struct draw_stage *stage ) } -static void passthrough_point( struct draw_stage *stage, - struct prim_header *header ) -{ - stage->next->point( stage->next, header ); -} - -static void widepoint_line( struct draw_stage *stage, - struct prim_header *header ) -{ - stage->next->line(stage->next, header); -} - -static void widepoint_tri( struct draw_stage *stage, - struct prim_header *header ) -{ - stage->next->tri(stage->next, header); -} /** @@ -199,16 +183,16 @@ static void widepoint_first_point( struct draw_stage *stage, wide->ybias = 0.0; if (draw->rasterizer->gl_rasterization_rules) { - wide->ybias = -0.125; + wide->xbias = 0.125; } /* XXX we won't know the real size if it's computed by the vertex shader! */ - if ((draw->rasterizer->point_size > draw->wide_point_threshold) || - (draw->rasterizer->point_sprite && draw->point_sprite)) { + if ((draw->rasterizer->point_size > draw->pipeline.wide_point_threshold) || + (draw->rasterizer->point_sprite && draw->pipeline.point_sprite)) { stage->point = widepoint_point; } else { - stage->point = passthrough_point; + stage->point = draw_pipe_passthrough_point; } if (draw->rasterizer->point_sprite) { @@ -265,17 +249,26 @@ static void widepoint_destroy( struct draw_stage *stage ) struct draw_stage *draw_wide_point_stage( struct draw_context *draw ) { struct widepoint_stage *wide = CALLOC_STRUCT(widepoint_stage); + if (wide == NULL) + goto fail; - draw_alloc_temp_verts( &wide->stage, 4 ); + if (!draw_alloc_temp_verts( &wide->stage, 4 )) + goto fail; wide->stage.draw = draw; wide->stage.next = NULL; wide->stage.point = widepoint_first_point; - wide->stage.line = widepoint_line; - wide->stage.tri = widepoint_tri; + wide->stage.line = draw_pipe_passthrough_line; + wide->stage.tri = draw_pipe_passthrough_tri; wide->stage.flush = widepoint_flush; wide->stage.reset_stipple_counter = widepoint_reset_stipple_counter; wide->stage.destroy = widepoint_destroy; return &wide->stage; + + fail: + if (wide) + wide->stage.destroy( &wide->stage ); + + return NULL; } diff --git a/src/gallium/auxiliary/draw/draw_prim.c b/src/gallium/auxiliary/draw/draw_prim.c deleted file mode 100644 index 51b6950334..0000000000 --- a/src/gallium/auxiliary/draw/draw_prim.c +++ /dev/null @@ -1,523 +0,0 @@ -/************************************************************************** - * - * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - - /* - * Authors: - * Keith Whitwell <keith@tungstengraphics.com> - */ - -#include "pipe/p_debug.h" -#include "pipe/p_util.h" - -#include "draw_private.h" -#include "draw_context.h" - - - -#define RP_NONE 0 -#define RP_POINT 1 -#define RP_LINE 2 -#define RP_TRI 3 - - -static unsigned reduced_prim[PIPE_PRIM_POLYGON + 1] = { - RP_POINT, - RP_LINE, - RP_LINE, - RP_LINE, - RP_TRI, - RP_TRI, - RP_TRI, - RP_TRI, - RP_TRI, - RP_TRI -}; - - -static void draw_prim_queue_flush( struct draw_context *draw ) -{ - unsigned i; - - if (0) - debug_printf("Flushing with %d prims, %d verts\n", - draw->pq.queue_nr, draw->vs.queue_nr); - - assert (draw->pq.queue_nr != 0); - - /* NOTE: we cannot save draw->pipeline->first in a local var because - * draw->pipeline->first is often changed by the first call to tri(), - * line(), etc. - */ - if (draw->rasterizer->line_stipple_enable) { - switch (draw->reduced_prim) { - case RP_TRI: - for (i = 0; i < draw->pq.queue_nr; i++) { - if (draw->pq.queue[i].reset_line_stipple) - draw->pipeline.first->reset_stipple_counter( draw->pipeline.first ); - - draw->pipeline.first->tri( draw->pipeline.first, &draw->pq.queue[i] ); - } - break; - case RP_LINE: - for (i = 0; i < draw->pq.queue_nr; i++) { - if (draw->pq.queue[i].reset_line_stipple) - draw->pipeline.first->reset_stipple_counter( draw->pipeline.first ); - - draw->pipeline.first->line( draw->pipeline.first, &draw->pq.queue[i] ); - } - break; - case RP_POINT: - draw->pipeline.first->reset_stipple_counter( draw->pipeline.first ); - for (i = 0; i < draw->pq.queue_nr; i++) - draw->pipeline.first->point( draw->pipeline.first, &draw->pq.queue[i] ); - break; - } - } - else { - switch (draw->reduced_prim) { - case RP_TRI: - for (i = 0; i < draw->pq.queue_nr; i++) - draw->pipeline.first->tri( draw->pipeline.first, &draw->pq.queue[i] ); - break; - case RP_LINE: - for (i = 0; i < draw->pq.queue_nr; i++) - draw->pipeline.first->line( draw->pipeline.first, &draw->pq.queue[i] ); - break; - case RP_POINT: - for (i = 0; i < draw->pq.queue_nr; i++) - draw->pipeline.first->point( draw->pipeline.first, &draw->pq.queue[i] ); - break; - } - } - - draw->pq.queue_nr = 0; - draw->vs.post_nr = 0; - draw_vertex_cache_unreference( draw ); -} - -void draw_do_flush( struct draw_context *draw, unsigned flags ) -{ - if (0) - debug_printf("Flushing with %d verts, %d prims\n", - draw->vs.queue_nr, - draw->pq.queue_nr ); - - if (draw->flushing) - return; - - draw->flushing = TRUE; - - if (flags >= DRAW_FLUSH_SHADER_QUEUE) { - if (draw->vs.queue_nr) { - (*draw->shader_queue_flush)(draw); - } - - if (flags >= DRAW_FLUSH_PRIM_QUEUE) { - if (draw->pq.queue_nr) - draw_prim_queue_flush(draw); - - if (flags >= DRAW_FLUSH_VERTEX_CACHE) { - draw_vertex_cache_invalidate(draw); - - if (flags >= DRAW_FLUSH_STATE_CHANGE) { - draw->pipeline.first->flush( draw->pipeline.first, flags ); - draw->pipeline.first = draw->pipeline.validate; - draw->reduced_prim = ~0; - } - } - } - } - - draw->flushing = FALSE; -} - - - -/* Return a pointer to a freshly queued primitive header. Ensure that - * there is room in the vertex cache for a maximum of "nr_verts" new - * vertices. Flush primitive and/or vertex queues if necessary to - * make space. - */ -static struct prim_header *get_queued_prim( struct draw_context *draw, - unsigned nr_verts ) -{ - if (!draw_vertex_cache_check_space( draw, nr_verts )) { -// debug_printf("v"); - draw_do_flush( draw, DRAW_FLUSH_VERTEX_CACHE ); - } - else if (draw->pq.queue_nr == PRIM_QUEUE_LENGTH) { -// debug_printf("p"); - draw_do_flush( draw, DRAW_FLUSH_PRIM_QUEUE ); - } - - assert(draw->pq.queue_nr < PRIM_QUEUE_LENGTH); - - return &draw->pq.queue[draw->pq.queue_nr++]; -} - - - -/** - * Add a point to the primitive queue. - * \param i0 index into user's vertex arrays - */ -static void do_point( struct draw_context *draw, - unsigned i0 ) -{ - struct prim_header *prim = get_queued_prim( draw, 1 ); - - prim->reset_line_stipple = 0; - prim->edgeflags = 1; - prim->pad = 0; - prim->v[0] = draw->vcache.get_vertex( draw, i0 ); -} - - -/** - * Add a line to the primitive queue. - * \param i0 index into user's vertex arrays - * \param i1 index into user's vertex arrays - */ -static void do_line( struct draw_context *draw, - boolean reset_stipple, - unsigned i0, - unsigned i1 ) -{ - struct prim_header *prim = get_queued_prim( draw, 2 ); - - prim->reset_line_stipple = reset_stipple; - prim->edgeflags = 1; - prim->pad = 0; - prim->v[0] = draw->vcache.get_vertex( draw, i0 ); - prim->v[1] = draw->vcache.get_vertex( draw, i1 ); -} - -/** - * Add a triangle to the primitive queue. - */ -static void do_triangle( struct draw_context *draw, - unsigned i0, - unsigned i1, - unsigned i2 ) -{ - struct prim_header *prim = get_queued_prim( draw, 3 ); - -// _mesa_printf("tri %d %d %d\n", i0, i1, i2); - prim->reset_line_stipple = 1; - prim->edgeflags = ~0; - prim->pad = 0; - prim->v[0] = draw->vcache.get_vertex( draw, i0 ); - prim->v[1] = draw->vcache.get_vertex( draw, i1 ); - prim->v[2] = draw->vcache.get_vertex( draw, i2 ); -} - -static void do_ef_triangle( struct draw_context *draw, - boolean reset_stipple, - unsigned ef_mask, - unsigned i0, - unsigned i1, - unsigned i2 ) -{ - struct prim_header *prim = get_queued_prim( draw, 3 ); - struct vertex_header *v0 = draw->vcache.get_vertex( draw, i0 ); - struct vertex_header *v1 = draw->vcache.get_vertex( draw, i1 ); - struct vertex_header *v2 = draw->vcache.get_vertex( draw, i2 ); - - prim->reset_line_stipple = reset_stipple; - - prim->edgeflags = ef_mask & ((v0->edgeflag << 0) | - (v1->edgeflag << 1) | - (v2->edgeflag << 2)); - prim->pad = 0; - prim->v[0] = v0; - prim->v[1] = v1; - prim->v[2] = v2; -} - - -static void do_ef_quad( struct draw_context *draw, - unsigned v0, - unsigned v1, - unsigned v2, - unsigned v3 ) -{ - const unsigned omitEdge2 = ~(1 << 1); - const unsigned omitEdge3 = ~(1 << 2); - do_ef_triangle( draw, 1, omitEdge2, v0, v1, v3 ); - do_ef_triangle( draw, 0, omitEdge3, v1, v2, v3 ); -} - -static void do_quad( struct draw_context *draw, - unsigned v0, - unsigned v1, - unsigned v2, - unsigned v3 ) -{ - do_triangle( draw, v0, v1, v3 ); - do_triangle( draw, v1, v2, v3 ); -} - - -/** - * Main entrypoint to draw some number of points/lines/triangles - */ -static void -draw_prim( struct draw_context *draw, - unsigned prim, unsigned start, unsigned count ) -{ - unsigned i; - boolean unfilled = (draw->rasterizer->fill_cw != PIPE_POLYGON_MODE_FILL || - draw->rasterizer->fill_ccw != PIPE_POLYGON_MODE_FILL); - boolean flatfirst = - (draw->rasterizer->flatshade & draw->rasterizer->flatshade_first) ? TRUE : FALSE; - -// debug_printf("%s (%d) %d/%d\n", __FUNCTION__, draw->prim, start, count ); - - switch (prim) { - case PIPE_PRIM_POINTS: - for (i = 0; i < count; i ++) { - do_point( draw, - start + i ); - } - break; - - case PIPE_PRIM_LINES: - for (i = 0; i+1 < count; i += 2) { - do_line( draw, - TRUE, - start + i + 0, - start + i + 1); - } - break; - - case PIPE_PRIM_LINE_LOOP: - if (count >= 2) { - for (i = 1; i < count; i++) { - do_line( draw, - i == 1, /* XXX: only if vb not split */ - start + i - 1, - start + i ); - } - - do_line( draw, - 0, - start + count - 1, - start + 0 ); - } - break; - - case PIPE_PRIM_LINE_STRIP: - for (i = 1; i < count; i++) { - do_line( draw, - i == 1, - start + i - 1, - start + i ); - } - break; - - case PIPE_PRIM_TRIANGLES: - if (unfilled) { - for (i = 0; i+2 < count; i += 3) { - do_ef_triangle( draw, - 1, - ~0, - start + i + 0, - start + i + 1, - start + i + 2 ); - } - } - else { - for (i = 0; i+2 < count; i += 3) { - do_triangle( draw, - start + i + 0, - start + i + 1, - start + i + 2 ); - } - } - break; - - case PIPE_PRIM_TRIANGLE_STRIP: - if (flatfirst) { - for (i = 0; i+2 < count; i++) { - if (i & 1) { - do_triangle( draw, - start + i + 0, - start + i + 2, - start + i + 1 ); - } - else { - do_triangle( draw, - start + i + 0, - start + i + 1, - start + i + 2 ); - } - } - } - else { - for (i = 0; i+2 < count; i++) { - if (i & 1) { - do_triangle( draw, - start + i + 1, - start + i + 0, - start + i + 2 ); - } - else { - do_triangle( draw, - start + i + 0, - start + i + 1, - start + i + 2 ); - } - } - } - break; - - case PIPE_PRIM_TRIANGLE_FAN: - if (count >= 3) { - if (flatfirst) { - for (i = 0; i+2 < count; i++) { - do_triangle( draw, - start + i + 1, - start + i + 2, - start + 0 ); - } - } - else { - for (i = 0; i+2 < count; i++) { - do_triangle( draw, - start + 0, - start + i + 1, - start + i + 2 ); - } - } - } - break; - - - case PIPE_PRIM_QUADS: - if (unfilled) { - for (i = 0; i+3 < count; i += 4) { - do_ef_quad( draw, - start + i + 0, - start + i + 1, - start + i + 2, - start + i + 3); - } - } - else { - for (i = 0; i+3 < count; i += 4) { - do_quad( draw, - start + i + 0, - start + i + 1, - start + i + 2, - start + i + 3); - } - } - break; - - case PIPE_PRIM_QUAD_STRIP: - if (unfilled) { - for (i = 0; i+3 < count; i += 2) { - do_ef_quad( draw, - start + i + 2, - start + i + 0, - start + i + 1, - start + i + 3); - } - } - else { - for (i = 0; i+3 < count; i += 2) { - do_quad( draw, - start + i + 2, - start + i + 0, - start + i + 1, - start + i + 3); - } - } - break; - - case PIPE_PRIM_POLYGON: - if (unfilled) { - unsigned ef_mask = (1<<2) | (1<<0); - - for (i = 0; i+2 < count; i++) { - - if (i + 3 >= count) - ef_mask |= (1<<1); - - do_ef_triangle( draw, - i == 0, - ef_mask, - start + i + 1, - start + i + 2, - start + 0); - - ef_mask &= ~(1<<2); - } - } - else { - for (i = 0; i+2 < count; i++) { - do_triangle( draw, - start + i + 1, - start + i + 2, - start + 0); - } - } - break; - - default: - assert(0); - break; - } -} - - - - -/** - * Draw vertex arrays - * This is the main entrypoint into the drawing module. - * \param prim one of PIPE_PRIM_x - * \param start index of first vertex to draw - * \param count number of vertices to draw - */ -void -draw_arrays(struct draw_context *draw, unsigned prim, - unsigned start, unsigned count) -{ - if (reduced_prim[prim] != draw->reduced_prim) { - draw_do_flush( draw, DRAW_FLUSH_STATE_CHANGE ); - draw->reduced_prim = reduced_prim[prim]; - } - - /* drawing done here: */ - if (!draw_pt_arrays(draw, prim, start, count)) { - /* we have to run the whole pipeline */ - draw_prim(draw, prim, start, count); - } -} - - diff --git a/src/gallium/auxiliary/draw/draw_private.h b/src/gallium/auxiliary/draw/draw_private.h index 4d056f6dba..cee58bbf73 100644 --- a/src/gallium/auxiliary/draw/draw_private.h +++ b/src/gallium/auxiliary/draw/draw_private.h @@ -44,7 +44,6 @@ #include "pipe/p_state.h" #include "pipe/p_defines.h" -#include "rtasm/rtasm_x86sse.h" #include "tgsi/exec/tgsi_exec.h" #include "tgsi/util/tgsi_scan.h" @@ -52,9 +51,11 @@ struct pipe_context; struct gallivm_prog; struct gallivm_cpu_engine; +struct draw_vertex_shader; +struct draw_context; +struct draw_stage; +struct vbuf_render; -struct draw_pt_middle_end; -struct draw_pt_front_end; /** * Basic vertex info. @@ -68,115 +69,14 @@ struct vertex_header { float clip[4]; - float data[][4]; /* Note variable size */ + /* This will probably become float (*data)[4] soon: + */ + float data[][4]; }; /* NOTE: It should match vertex_id size above */ #define UNDEFINED_VERTEX_ID 0xffff -/* XXX This is too large */ -#define MAX_VERTEX_SIZE ((2 + PIPE_MAX_SHADER_OUTPUTS) * 4 * sizeof(float)) - - - -/** - * Basic info for a point/line/triangle primitive. - */ -struct prim_header { - float det; /**< front/back face determinant */ - unsigned reset_line_stipple:1; - unsigned edgeflags:3; - unsigned pad:28; - struct vertex_header *v[3]; /**< 1 to 3 vertex pointers */ -}; - - - -struct draw_context; - -/** - * Base class for all primitive drawing stages. - */ -struct draw_stage -{ - struct draw_context *draw; /**< parent context */ - - struct draw_stage *next; /**< next stage in pipeline */ - - struct vertex_header **tmp; /**< temp vert storage, such as for clipping */ - unsigned nr_tmps; - - void (*point)( struct draw_stage *, - struct prim_header * ); - - void (*line)( struct draw_stage *, - struct prim_header * ); - - void (*tri)( struct draw_stage *, - struct prim_header * ); - - void (*flush)( struct draw_stage *, - unsigned flags ); - - void (*reset_stipple_counter)( struct draw_stage * ); - - void (*destroy)( struct draw_stage * ); -}; - - -#define PRIM_QUEUE_LENGTH 32 -#define VCACHE_SIZE 32 -#define VCACHE_OVERFLOW 4 -#define VS_QUEUE_LENGTH (VCACHE_SIZE + VCACHE_OVERFLOW + 1) /* can never fill up */ - -/** - * Private version of the compiled vertex_shader - */ -struct draw_vertex_shader { - - /* This member will disappear shortly: - */ - struct pipe_shader_state state; - - struct tgsi_shader_info info; - - void (*prepare)( struct draw_vertex_shader *shader, - struct draw_context *draw ); - - /* Run the shader - this interface will get cleaned up in the - * future: - */ - void (*run)( struct draw_vertex_shader *shader, - struct draw_context *draw, - const unsigned *elts, - unsigned count, - struct vertex_header *vOut[] ); - - - void (*delete)( struct draw_vertex_shader * ); -}; - - -/* Internal function for vertex fetch. - */ -typedef void (*fetch_func)(const void *ptr, float *attrib); - -fetch_func draw_get_fetch_func( enum pipe_format format ); - - - -typedef void (*full_fetch_func)( struct draw_context *draw, - struct tgsi_exec_machine *machine, - const unsigned *elts, - unsigned count ); - -typedef void (*pt_fetch_func)( struct draw_context *draw, - float *out, - unsigned start, - unsigned count ); - - -struct vbuf_render; /** * Private context for the drawing module. @@ -203,6 +103,17 @@ struct draw_context struct draw_stage *wide_line; struct draw_stage *wide_point; struct draw_stage *rasterize; + + float wide_point_threshold; /**< convert pnts to tris if larger than this */ + float wide_line_threshold; /**< convert lines to tris if wider than this */ + boolean line_stipple; /**< do line stipple? */ + boolean point_sprite; /**< convert points to quads for sprites? */ + + /* Temporary storage while the pipeline is being run: + */ + char *verts; + unsigned vertex_stride; + unsigned vertex_count; } pipeline; @@ -211,70 +122,64 @@ struct draw_context /* Support prototype passthrough path: */ struct { - unsigned prim; /* XXX: to be removed */ - unsigned hw_vertex_size; /* XXX: to be removed */ - struct { struct draw_pt_middle_end *fetch_emit; - struct draw_pt_middle_end *fetch_pipeline; - struct draw_pt_middle_end *fetch_shade_emit; - struct draw_pt_middle_end *fetch_shade_cliptest_pipeline_or_emit; + struct draw_pt_middle_end *general; } middle; struct { - struct draw_pt_front_end *noop; - struct draw_pt_front_end *split_arrays; struct draw_pt_front_end *vcache; + struct draw_pt_front_end *varray; } front; + struct pipe_vertex_buffer vertex_buffer[PIPE_MAX_ATTRIBS]; + unsigned nr_vertex_buffers; + + struct pipe_vertex_element vertex_element[PIPE_MAX_ATTRIBS]; + unsigned nr_vertex_elements; + + /* user-space vertex data, buffers */ struct { - char *verts; - unsigned vertex_stride; - unsigned vertex_count; - } pipeline; + const unsigned *edgeflag; + + /** vertex element/index buffer (ex: glDrawElements) */ + const void *elts; + /** bytes per index (0, 1, 2 or 4) */ + unsigned eltSize; + + /** vertex arrays */ + const void *vbuffer[PIPE_MAX_ATTRIBS]; + + /** constant buffer (for vertex shader) */ + const void *constants; + } user; } pt; - boolean flushing; + struct { + boolean bypass_clipping; + } driver; + + boolean flushing; /**< debugging/sanity */ + boolean suspend_flushing; /**< internally set */ + boolean bypass_clipping; /**< set if either api or driver bypass_clipping true */ /* pipe state that we need: */ const struct pipe_rasterizer_state *rasterizer; struct pipe_viewport_state viewport; - struct pipe_vertex_buffer vertex_buffer[PIPE_MAX_ATTRIBS]; - struct pipe_vertex_element vertex_element[PIPE_MAX_ATTRIBS]; + struct draw_vertex_shader *vertex_shader; boolean identity_viewport; uint num_vs_outputs; /**< convenience, from vertex_shader */ - /* user-space vertex data, buffers */ - struct { - const unsigned *edgeflag; - - /** vertex element/index buffer (ex: glDrawElements) */ - const void *elts; - /** bytes per index (0, 1, 2 or 4) */ - unsigned eltSize; - - /** vertex arrays */ - const void *vbuffer[PIPE_MAX_ATTRIBS]; - - /** constant buffer (for vertex shader) */ - const void *constants; - } user; /* Clip derived state: */ float plane[12][4]; unsigned nr_planes; - float wide_point_threshold; /**< convert pnts to tris if larger than this */ - float wide_line_threshold; /**< convert lines to tris if wider than this */ - boolean line_stipple; /**< do line stipple? */ - boolean point_sprite; /**< convert points to quads for sprites? */ - boolean use_sse; - /* If a prim stage introduces new vertex attributes, they'll be stored here */ struct { @@ -288,61 +193,6 @@ struct draw_context /** TGSI program interpreter runtime state */ struct tgsi_exec_machine machine; - /* Vertex fetch internal state - */ - struct { - const ubyte *src_ptr[PIPE_MAX_ATTRIBS]; - unsigned pitch[PIPE_MAX_ATTRIBS]; - fetch_func fetch[PIPE_MAX_ATTRIBS]; - unsigned nr_attrs; - full_fetch_func fetch_func; - pt_fetch_func pt_fetch; - } vertex_fetch; - - /* Post-tnl vertex cache: - */ - struct { - unsigned referenced; /**< bitfield */ - - struct { - unsigned in; /* client array element */ - unsigned out; /* index in vs queue/array */ - } idx[VCACHE_SIZE + VCACHE_OVERFLOW]; - - unsigned overflow; - - /** To find space in the vertex cache: */ - struct vertex_header *(*get_vertex)( struct draw_context *draw, - unsigned i ); - } vcache; - - /* Vertex shader queue: - */ - struct { - struct { - unsigned elt; /**< index into the user's vertex arrays */ - struct vertex_header *vertex; - } queue[VS_QUEUE_LENGTH]; - unsigned queue_nr; - unsigned post_nr; - } vs; - - /** - * Run the vertex shader on all vertices in the vertex queue. - */ - void (*shader_queue_flush)(struct draw_context *draw); - - /* Prim pipeline queue: - */ - struct { - /* Need to queue up primitives until their vertices have been - * transformed by a vs queue flush. - */ - struct prim_header queue[PRIM_QUEUE_LENGTH]; - unsigned queue_nr; - } pq; - - /* This (and the tgsi_exec_machine struct) probably need to be moved somewhere private. */ struct gallivm_cpu_engine *engine; @@ -351,101 +201,70 @@ struct draw_context -extern struct draw_stage *draw_unfilled_stage( struct draw_context *context ); -extern struct draw_stage *draw_twoside_stage( struct draw_context *context ); -extern struct draw_stage *draw_offset_stage( struct draw_context *context ); -extern struct draw_stage *draw_clip_stage( struct draw_context *context ); -extern struct draw_stage *draw_flatshade_stage( struct draw_context *context ); -extern struct draw_stage *draw_cull_stage( struct draw_context *context ); -extern struct draw_stage *draw_stipple_stage( struct draw_context *context ); -extern struct draw_stage *draw_wide_line_stage( struct draw_context *context ); -extern struct draw_stage *draw_wide_point_stage( struct draw_context *context ); -extern struct draw_stage *draw_validate_stage( struct draw_context *context ); -extern void draw_free_temp_verts( struct draw_stage *stage ); -extern void draw_alloc_temp_verts( struct draw_stage *stage, unsigned nr ); -extern void draw_reset_vertex_ids( struct draw_context *draw ); +/******************************************************************************* + * Vertex processing (was passthrough) code: + */ +boolean draw_pt_init( struct draw_context *draw ); +void draw_pt_destroy( struct draw_context *draw ); +void draw_pt_reset_vertex_ids( struct draw_context *draw ); -extern int draw_vertex_cache_check_space( struct draw_context *draw, - unsigned nr_verts ); +/******************************************************************************* + * Primitive processing (pipeline) code: + */ -extern void draw_vertex_cache_invalidate( struct draw_context *draw ); -extern void draw_vertex_cache_unreference( struct draw_context *draw ); -extern void draw_vertex_cache_reset_vertex_ids( struct draw_context *draw ); +boolean draw_pipeline_init( struct draw_context *draw ); +void draw_pipeline_destroy( struct draw_context *draw ); -extern void draw_vertex_shader_queue_flush( struct draw_context *draw ); -extern void draw_update_vertex_fetch( struct draw_context *draw ); -extern boolean draw_need_pipeline(const struct draw_context *draw, - unsigned prim ); -/* Passthrough mode (second attempt): +/* We use the top few bits in the elts[] parameter to convey a little + * API information. This limits the number of vertices we can address + * to only 4096 -- if that becomes a problem, we can switch to 32-bit + * draw indices. + * + * These flags expected at first vertex of lines & triangles when + * unfilled and/or line stipple modes are operational. */ -boolean draw_pt_init( struct draw_context *draw ); -void draw_pt_destroy( struct draw_context *draw ); -boolean draw_pt_arrays( struct draw_context *draw, +#define DRAW_PIPE_EDGE_FLAG_0 (0x1<<12) +#define DRAW_PIPE_EDGE_FLAG_1 (0x2<<12) +#define DRAW_PIPE_EDGE_FLAG_2 (0x4<<12) +#define DRAW_PIPE_EDGE_FLAG_ALL (0x7<<12) +#define DRAW_PIPE_RESET_STIPPLE (0x8<<12) +#define DRAW_PIPE_FLAG_MASK (0xf<<12) + +void draw_pipeline_run( struct draw_context *draw, unsigned prim, - unsigned start, + struct vertex_header *vertices, + unsigned vertex_count, + unsigned stride, + const ushort *elts, unsigned count ); -void draw_pt_reset_vertex_ids( struct draw_context *draw ); -void draw_pt_run_pipeline( struct draw_context *draw, - unsigned prim, - char *verts, - unsigned vertex_stride, - unsigned vertex_count, - const ushort *elts, - unsigned count ); - - -#define DRAW_FLUSH_SHADER_QUEUE 0x1 /* sized not to overflow, never raised */ -#define DRAW_FLUSH_PRIM_QUEUE 0x2 -#define DRAW_FLUSH_VERTEX_CACHE 0x4 + + +void draw_pipeline_flush( struct draw_context *draw, + unsigned flags ); + + + +/******************************************************************************* + * Flushing + */ + #define DRAW_FLUSH_STATE_CHANGE 0x8 #define DRAW_FLUSH_BACKEND 0x10 void draw_do_flush( struct draw_context *draw, unsigned flags ); -boolean draw_get_edgeflag( struct draw_context *draw, - unsigned idx ); - -/** - * Get a writeable copy of a vertex. - * \param stage drawing stage info - * \param vert the vertex to copy (source) - * \param idx index into stage's tmp[] array to put the copy (dest) - * \return pointer to the copied vertex - */ -static INLINE struct vertex_header * -dup_vert( struct draw_stage *stage, - const struct vertex_header *vert, - unsigned idx ) -{ - struct vertex_header *tmp = stage->tmp[idx]; - const uint vsize = sizeof(struct vertex_header) - + stage->draw->num_vs_outputs * 4 * sizeof(float); - memcpy(tmp, vert, vsize); - tmp->vertex_id = UNDEFINED_VERTEX_ID; - return tmp; -} - -static INLINE float -dot4(const float *a, const float *b) -{ - float result = (a[0]*b[0] + - a[1]*b[1] + - a[2]*b[2] + - a[3]*b[3]); - return result; -} #endif /* DRAW_PRIVATE_H */ diff --git a/src/gallium/auxiliary/draw/draw_pt.c b/src/gallium/auxiliary/draw/draw_pt.c index f59fb86f78..c9c5d18313 100644 --- a/src/gallium/auxiliary/draw/draw_pt.c +++ b/src/gallium/auxiliary/draw/draw_pt.c @@ -36,153 +36,63 @@ #include "draw/draw_pt.h" -/* XXX: Shouldn't those two functions below use the '>' operator??? - */ - -static boolean too_many_verts( struct draw_context *draw, - unsigned verts ) -{ - return verts < 1024; -} - -static boolean too_many_elts( struct draw_context *draw, - unsigned elts ) -{ - return elts < (16 * 1024); -} -boolean +/* Overall we split things into: + * - frontend -- prepare fetch_elts, draw_elts - eg vcache + * - middle -- fetch, shade, cliptest, viewport + * - pipeline -- the prim pipeline: clipping, wide lines, etc + * - backend -- the vbuf_render provided by the driver. + */ +static boolean draw_pt_arrays(struct draw_context *draw, unsigned prim, unsigned start, unsigned count) { - const boolean pipeline = draw_need_pipeline(draw, prim); - const boolean cliptest = !draw->rasterizer->bypass_clipping; - const boolean shading = !draw->rasterizer->bypass_vs; struct draw_pt_front_end *frontend = NULL; struct draw_pt_middle_end *middle = NULL; + unsigned opt = 0; - - /* Overall we do: - * - frontend -- prepare fetch_elts, draw_elts - eg vcache - * - middle -- fetch, shade, cliptest, viewport - * - pipeline -- the prim pipeline: clipping, wide lines, etc - * - backend -- the vbuf_render provided by the driver. - */ - - - if (!cliptest && !pipeline && !shading) { - /* This is the 'passthrough' path: - */ - /* Fetch user verts, emit hw verts: - */ - middle = draw->pt.middle.fetch_emit; - } - else if (!cliptest && !shading) { - /* This is the 'passthrough' path targetting the pipeline backend. - */ - /* Fetch user verts, emit pipeline verts, run pipeline: - */ - middle = draw->pt.middle.fetch_pipeline; + if (!draw->render) { + opt |= PT_PIPELINE; } -#if 0 - else if (!cliptest && !pipeline) { - /* Fetch user verts, run vertex shader, emit hw verts: - */ - middle = draw->pt.middle.fetch_shade_emit; - } - else if (!pipeline) { - /* Even though !pipeline, we have to run it to get clipping. We - * do know that the pipeline is just the clipping operation, but - * that probably doesn't help much. - * - * This is going to be the most important path for a lot of - * swtnl cards. - */ - /* Fetch user verts, - * run vertex shader, - * cliptest and viewport trasform - * if no clipped vertices, - * emit hw verts - * else - * run pipline - */ - middle = draw->pt.middle.fetch_shade_cliptest_pipeline_or_emit; - } - else if (!cliptest) { - /* Fetch user verts, run vertex shader, run pipeline: - */ - middle = draw->pt.middle.fetch_shade_pipeline; + + if (draw_need_pipeline(draw, + draw->rasterizer, + prim)) { + opt |= PT_PIPELINE; } - else { - /* This is what we're currently always doing: - */ - /* Fetch user verts, run vertex shader, cliptest, run pipeline: - */ - middle = draw->pt.middle.fetch_shade_cliptest_pipeline; + + if (!draw->bypass_clipping) { + opt |= PT_CLIPTEST; } -#else - else { - return FALSE; + + if (!draw->rasterizer->bypass_vs) { + opt |= PT_SHADE; } -#endif + if (opt) + middle = draw->pt.middle.general; + else + middle = draw->pt.middle.fetch_emit; - /* If !pipeline, need to make sure we respect the driver's limited - * capabilites to receive blocks of vertex data and elements. - */ -#if 0 - if (!pipeline) { - unsigned vertex_mode = passthrough; - unsigned nr_verts = count_vertices( draw, start, count ); - unsigned hw_prim = prim; - - if (is_elts(draw)) { - frontend = draw->pt.front.vcache; - hw_prim = reduced_prim(prim); - } - - if (too_many_verts(nr_verts)) { - /* if (is_verts(draw) && can_split(prim)) { - draw = draw_arrays_split; - } - else */ { - frontend = draw->pt.front.vcache; - hw_prim = reduced_prim(prim); - } - } - - if (too_many_elts(count)) { - - /* if (is_elts(draw) && can_split(prim)) { - draw = draw_elts_split; - } - else */ { - frontend = draw->pt.front.vcache; - hw_prim = reduced_prim(prim); - } - } - - if (!good_prim(hw_prim)) { - draw = draw->pt.front.vcache; - } - } -#else - frontend = draw->pt.front.vcache; -#endif - /* XXX: need to flush to get prim_vbuf.c to release its allocation?? + /* Pick the right frontend */ - draw_do_flush( draw, DRAW_FLUSH_BACKEND ); + if (draw->pt.user.elts || + count >= 256) { + frontend = draw->pt.front.vcache; + } else { + frontend = draw->pt.front.varray; + } - frontend->prepare( frontend, prim, middle ); + frontend->prepare( frontend, prim, middle, opt ); - frontend->run( frontend, - draw_pt_elt_func( draw ), - draw_pt_elt_ptr( draw, start ), - count ); + frontend->run(frontend, + draw_pt_elt_func(draw), + draw_pt_elt_ptr(draw, start), + count); frontend->finish( frontend ); @@ -192,16 +102,20 @@ draw_pt_arrays(struct draw_context *draw, boolean draw_pt_init( struct draw_context *draw ) { - draw->pt.middle.fetch_emit = draw_pt_fetch_emit( draw ); - if (!draw->pt.middle.fetch_emit) + draw->pt.front.vcache = draw_pt_vcache( draw ); + if (!draw->pt.front.vcache) return FALSE; - draw->pt.middle.fetch_pipeline = draw_pt_fetch_pipeline( draw ); - if (!draw->pt.middle.fetch_pipeline) + draw->pt.front.varray = draw_pt_varray(draw); + if (!draw->pt.front.varray) return FALSE; - draw->pt.front.vcache = draw_pt_vcache( draw ); - if (!draw->pt.front.vcache) + draw->pt.middle.fetch_emit = draw_pt_fetch_emit( draw ); + if (!draw->pt.middle.fetch_emit) + return FALSE; + + draw->pt.middle.general = draw_pt_fetch_pipeline_or_emit( draw ); + if (!draw->pt.middle.general) return FALSE; return TRUE; @@ -210,18 +124,68 @@ boolean draw_pt_init( struct draw_context *draw ) void draw_pt_destroy( struct draw_context *draw ) { + if (draw->pt.middle.general) { + draw->pt.middle.general->destroy( draw->pt.middle.general ); + draw->pt.middle.general = NULL; + } + if (draw->pt.middle.fetch_emit) { draw->pt.middle.fetch_emit->destroy( draw->pt.middle.fetch_emit ); draw->pt.middle.fetch_emit = NULL; } - if (draw->pt.middle.fetch_pipeline) { - draw->pt.middle.fetch_pipeline->destroy( draw->pt.middle.fetch_pipeline ); - draw->pt.middle.fetch_pipeline = NULL; - } - if (draw->pt.front.vcache) { draw->pt.front.vcache->destroy( draw->pt.front.vcache ); draw->pt.front.vcache = NULL; } + + if (draw->pt.front.varray) { + draw->pt.front.varray->destroy( draw->pt.front.varray ); + draw->pt.front.varray = NULL; + } +} + + + +static unsigned reduced_prim[PIPE_PRIM_POLYGON + 1] = { + PIPE_PRIM_POINTS, + PIPE_PRIM_LINES, + PIPE_PRIM_LINES, + PIPE_PRIM_LINES, + PIPE_PRIM_TRIANGLES, + PIPE_PRIM_TRIANGLES, + PIPE_PRIM_TRIANGLES, + PIPE_PRIM_TRIANGLES, + PIPE_PRIM_TRIANGLES, + PIPE_PRIM_TRIANGLES +}; + + +/** + * Draw vertex arrays + * This is the main entrypoint into the drawing module. + * \param prim one of PIPE_PRIM_x + * \param start index of first vertex to draw + * \param count number of vertices to draw + */ +void +draw_arrays(struct draw_context *draw, unsigned prim, + unsigned start, unsigned count) +{ + if (reduced_prim[prim] != draw->reduced_prim) { + draw_do_flush( draw, DRAW_FLUSH_STATE_CHANGE ); + draw->reduced_prim = reduced_prim[prim]; + } + + /* drawing done here: */ + draw_pt_arrays(draw, prim, start, count); +} + +boolean draw_pt_get_edgeflag( struct draw_context *draw, + unsigned idx ) +{ + if (draw->pt.user.edgeflag) + return (draw->pt.user.edgeflag[idx/32] & (1 << (idx%32))) != 0; + else + return 1; } diff --git a/src/gallium/auxiliary/draw/draw_pt.h b/src/gallium/auxiliary/draw/draw_pt.h index 590823fd33..2dec376cee 100644 --- a/src/gallium/auxiliary/draw/draw_pt.h +++ b/src/gallium/auxiliary/draw/draw_pt.h @@ -40,14 +40,11 @@ typedef unsigned (*pt_elt_func)( const void *elts, unsigned idx ); struct draw_pt_middle_end; struct draw_context; -/* We use the top couple of bits in the vertex fetch index to convey a - * little API information. This limits the number of vertices we can - * address to only 1 billion -- if that becomes a problem, these could - * be moved out & passed separately. - */ -#define DRAW_PT_EDGEFLAG (1<<30) -#define DRAW_PT_RESET_STIPPLE (1<<31) -#define DRAW_PT_FLAG_MASK (3<<30) + +#define PT_SHADE 0x1 +#define PT_CLIPTEST 0x2 +#define PT_PIPELINE 0x4 +#define PT_MAX_MIDDLE 0x8 /* The "front end" - prepare sets of fetch, draw elements for the @@ -64,7 +61,8 @@ struct draw_context; struct draw_pt_front_end { void (*prepare)( struct draw_pt_front_end *, unsigned prim, - struct draw_pt_middle_end * ); + struct draw_pt_middle_end *, + unsigned opt ); void (*run)( struct draw_pt_front_end *, pt_elt_func elt_func, @@ -82,15 +80,11 @@ struct draw_pt_front_end { * Currently two versions of this: * - fetch, vertex shade, cliptest, prim-pipeline * - fetch, emit (ie passthrough) - * Later: - * - fetch, vertex shade, cliptest, maybe-pipeline, maybe-emit - * - fetch, vertex shade, emit - * - * Currenly only using the passthrough version. */ struct draw_pt_middle_end { void (*prepare)( struct draw_pt_middle_end *, - unsigned prim ); + unsigned prim, + unsigned opt ); void (*run)( struct draw_pt_middle_end *, const unsigned *fetch_elts, @@ -104,12 +98,9 @@ struct draw_pt_middle_end { /* The "back end" - supplied by the driver, defined in draw_vbuf.h. - * - * Not sure whether to wrap the prim pipeline up as an alternate - * backend. Would be a win for everything except pure passthrough - * mode... */ struct vbuf_render; +struct vertex_header; /* Helper functions. @@ -118,12 +109,89 @@ pt_elt_func draw_pt_elt_func( struct draw_context *draw ); const void *draw_pt_elt_ptr( struct draw_context *draw, unsigned start ); -/* Implementations: +/* Frontends: + * + * Currently only the general-purpose vcache implementation, could add + * a special case for tiny vertex buffers. */ struct draw_pt_front_end *draw_pt_vcache( struct draw_context *draw ); +struct draw_pt_front_end *draw_pt_varray(struct draw_context *draw); + +/* Middle-ends: + * + * Currently one general-purpose case which can do all possibilities, + * at the slight expense of creating a vertex_header in some cases + * unecessarily. + * + * The special case fetch_emit code avoids pipeline vertices + * altogether and builds hardware vertices directly from API + * vertex_elements. + */ struct draw_pt_middle_end *draw_pt_fetch_emit( struct draw_context *draw ); -struct draw_pt_middle_end *draw_pt_fetch_pipeline( struct draw_context *draw ); +struct draw_pt_middle_end *draw_pt_fetch_pipeline_or_emit(struct draw_context *draw); + + +/* More helpers: + */ +boolean draw_pt_get_edgeflag( struct draw_context *draw, + unsigned idx ); + + +/******************************************************************************* + * HW vertex emit: + */ +struct pt_emit; + +void draw_pt_emit_prepare( struct pt_emit *emit, + unsigned prim ); + +void draw_pt_emit( struct pt_emit *emit, + const float (*vertex_data)[4], + unsigned vertex_count, + unsigned stride, + const ushort *elts, + unsigned count ); + +void draw_pt_emit_destroy( struct pt_emit *emit ); + +struct pt_emit *draw_pt_emit_create( struct draw_context *draw ); + + +/******************************************************************************* + * API vertex fetch: + */ + +struct pt_fetch; +void draw_pt_fetch_prepare( struct pt_fetch *fetch, + unsigned vertex_size ); + +void draw_pt_fetch_run( struct pt_fetch *fetch, + const unsigned *elts, + unsigned count, + char *verts ); + +void draw_pt_fetch_destroy( struct pt_fetch *fetch ); + +struct pt_fetch *draw_pt_fetch_create( struct draw_context *draw ); + +/******************************************************************************* + * Post-VS: cliptest, rhw, viewport + */ +struct pt_post_vs; + +boolean draw_pt_post_vs_run( struct pt_post_vs *pvs, + struct vertex_header *pipeline_verts, + unsigned stride, + unsigned count ); + +void draw_pt_post_vs_prepare( struct pt_post_vs *pvs, + boolean bypass_clipping, + boolean identity_viewport, + boolean opengl ); + +struct pt_post_vs *draw_pt_post_vs_create( struct draw_context *draw ); +void draw_pt_post_vs_destroy( struct pt_post_vs *pvs ); #endif diff --git a/src/gallium/auxiliary/draw/draw_pt_elts.c b/src/gallium/auxiliary/draw/draw_pt_elts.c index d49770e7b2..2094c081ed 100644 --- a/src/gallium/auxiliary/draw/draw_pt_elts.c +++ b/src/gallium/auxiliary/draw/draw_pt_elts.c @@ -59,7 +59,7 @@ static unsigned elt_vert( const void *elts, unsigned idx ) pt_elt_func draw_pt_elt_func( struct draw_context *draw ) { - switch (draw->user.eltSize) { + switch (draw->pt.user.eltSize) { case 0: return elt_vert; case 1: return elt_ubyte; case 2: return elt_ushort; @@ -71,9 +71,9 @@ pt_elt_func draw_pt_elt_func( struct draw_context *draw ) const void *draw_pt_elt_ptr( struct draw_context *draw, unsigned start ) { - const char *elts = draw->user.elts; + const char *elts = draw->pt.user.elts; - switch (draw->user.eltSize) { + switch (draw->pt.user.eltSize) { case 0: return (const void *)(((const ubyte *)NULL) + start); case 1: diff --git a/src/gallium/auxiliary/draw/draw_pt_emit.c b/src/gallium/auxiliary/draw/draw_pt_emit.c new file mode 100644 index 0000000000..ce3a153f64 --- /dev/null +++ b/src/gallium/auxiliary/draw/draw_pt_emit.c @@ -0,0 +1,203 @@ +/************************************************************************** + * + * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include "pipe/p_util.h" +#include "draw/draw_context.h" +#include "draw/draw_private.h" +#include "draw/draw_vbuf.h" +#include "draw/draw_vertex.h" +#include "draw/draw_pt.h" +#include "translate/translate.h" +#include "translate/translate_cache.h" + +struct pt_emit { + struct draw_context *draw; + + struct translate *translate; + + struct translate_cache *cache; +}; + +void draw_pt_emit_prepare( struct pt_emit *emit, + unsigned prim ) +{ + struct draw_context *draw = emit->draw; + const struct vertex_info *vinfo; + unsigned dst_offset; + struct translate_key hw_key; + unsigned i; + boolean ok; + + ok = draw->render->set_primitive(draw->render, prim); + if (!ok) { + assert(0); + return; + } + + /* Must do this after set_primitive() above: + */ + vinfo = draw->render->get_vertex_info(draw->render); + + + /* Translate from pipeline vertices to hw vertices. + */ + dst_offset = 0; + for (i = 0; i < vinfo->num_attribs; i++) { + unsigned emit_sz = 0; + unsigned src_buffer = 0; + unsigned output_format; + unsigned src_offset = (vinfo->src_index[i] * 4 * sizeof(float) ); + + + + switch (vinfo->emit[i]) { + case EMIT_4F: + output_format = PIPE_FORMAT_R32G32B32A32_FLOAT; + emit_sz = 4 * sizeof(float); + break; + case EMIT_3F: + output_format = PIPE_FORMAT_R32G32B32_FLOAT; + emit_sz = 3 * sizeof(float); + break; + case EMIT_2F: + output_format = PIPE_FORMAT_R32G32_FLOAT; + emit_sz = 2 * sizeof(float); + break; + case EMIT_1F: + output_format = PIPE_FORMAT_R32_FLOAT; + emit_sz = 1 * sizeof(float); + break; + case EMIT_1F_PSIZE: + output_format = PIPE_FORMAT_R32_FLOAT; + emit_sz = 1 * sizeof(float); + src_buffer = 1; + src_offset = 0; + break; + case EMIT_4UB: + output_format = PIPE_FORMAT_B8G8R8A8_UNORM; + emit_sz = 4 * sizeof(ubyte); + default: + assert(0); + output_format = PIPE_FORMAT_NONE; + emit_sz = 0; + break; + } + + hw_key.element[i].input_format = PIPE_FORMAT_R32G32B32A32_FLOAT; + hw_key.element[i].input_buffer = src_buffer; + hw_key.element[i].input_offset = src_offset; + hw_key.element[i].output_format = output_format; + hw_key.element[i].output_offset = dst_offset; + + dst_offset += emit_sz; + } + + hw_key.nr_elements = vinfo->num_attribs; + hw_key.output_stride = vinfo->size * 4; + + if (!emit->translate || + translate_key_compare(&emit->translate->key, &hw_key) != 0) + { + translate_key_sanitize(&hw_key); + emit->translate = translate_cache_find(emit->cache, &hw_key); + } +} + + +void draw_pt_emit( struct pt_emit *emit, + const float (*vertex_data)[4], + unsigned vertex_count, + unsigned stride, + const ushort *elts, + unsigned count ) +{ + struct draw_context *draw = emit->draw; + struct translate *translate = emit->translate; + struct vbuf_render *render = draw->render; + void *hw_verts; + + /* XXX: need to flush to get prim_vbuf.c to release its allocation?? + */ + draw_do_flush( draw, DRAW_FLUSH_BACKEND ); + + hw_verts = render->allocate_vertices(render, + (ushort)translate->key.output_stride, + (ushort)vertex_count); + if (!hw_verts) { + assert(0); + return; + } + + translate->set_buffer(translate, + 0, + vertex_data, + stride ); + + translate->set_buffer(translate, + 1, + &draw->rasterizer->point_size, + 0); + + translate->run( translate, + 0, + vertex_count, + hw_verts ); + + render->draw(render, + elts, + count); + + render->release_vertices(render, + hw_verts, + translate->key.output_stride, + vertex_count); +} + + +struct pt_emit *draw_pt_emit_create( struct draw_context *draw ) +{ + struct pt_emit *emit = CALLOC_STRUCT(pt_emit); + if (!emit) + return NULL; + + emit->draw = draw; + emit->cache = translate_cache_create(); + if (!emit->cache) { + FREE(emit); + return NULL; + } + + return emit; +} + +void draw_pt_emit_destroy( struct pt_emit *emit ) +{ + if (emit->cache) + translate_cache_destroy(emit->cache); + + FREE(emit); +} diff --git a/src/gallium/auxiliary/draw/draw_pt_fetch.c b/src/gallium/auxiliary/draw/draw_pt_fetch.c new file mode 100644 index 0000000000..100117a9ae --- /dev/null +++ b/src/gallium/auxiliary/draw/draw_pt_fetch.c @@ -0,0 +1,191 @@ +/************************************************************************** + * + * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include "pipe/p_util.h" +#include "draw/draw_context.h" +#include "draw/draw_private.h" +#include "draw/draw_vbuf.h" +#include "draw/draw_vertex.h" +#include "draw/draw_pt.h" +#include "translate/translate.h" +#include "translate/translate_cache.h" + + +struct pt_fetch { + struct draw_context *draw; + + struct translate *translate; + + unsigned vertex_size; + boolean need_edgeflags; + + struct translate_cache *cache; +}; + +/* Perform the fetch from API vertex elements & vertex buffers, to a + * contiguous set of float[4] attributes as required for the + * vertex_shader->run_linear() method. + * + * This is used in all cases except pure passthrough + * (draw_pt_fetch_emit.c) which has its own version to translate + * directly to hw vertices. + * + */ +void draw_pt_fetch_prepare( struct pt_fetch *fetch, + unsigned vertex_size ) +{ + struct draw_context *draw = fetch->draw; + unsigned i, nr = 0; + unsigned dst_offset = 0; + struct translate_key key; + + fetch->vertex_size = vertex_size; + + /* Always emit/leave space for a vertex header. + * + * It's worth considering whether the vertex headers should contain + * a pointer to the 'data', rather than having it inline. + * Something to look at after we've fully switched over to the pt + * paths. + */ + { + /* Need to set header->vertex_id = 0xffff somehow. + */ + key.element[nr].input_format = PIPE_FORMAT_R32_FLOAT; + key.element[nr].input_buffer = draw->pt.nr_vertex_buffers; + key.element[nr].input_offset = 0; + key.element[nr].output_format = PIPE_FORMAT_R32_FLOAT; + key.element[nr].output_offset = dst_offset; + dst_offset += 1 * sizeof(float); + nr++; + + + /* Just leave the clip[] array untouched. + */ + dst_offset += 4 * sizeof(float); + } + + + for (i = 0; i < draw->pt.nr_vertex_elements; i++) { + key.element[nr].input_format = draw->pt.vertex_element[i].src_format; + key.element[nr].input_buffer = draw->pt.vertex_element[i].vertex_buffer_index; + key.element[nr].input_offset = draw->pt.vertex_element[i].src_offset; + key.element[nr].output_format = PIPE_FORMAT_R32G32B32A32_FLOAT; + key.element[nr].output_offset = dst_offset; + + dst_offset += 4 * sizeof(float); + nr++; + } + + assert(dst_offset <= vertex_size); + + key.nr_elements = nr; + key.output_stride = vertex_size; + + + if (!fetch->translate || + translate_key_compare(&fetch->translate->key, &key) != 0) + { + translate_key_sanitize(&key); + fetch->translate = translate_cache_find(fetch->cache, &key); + + { + static struct vertex_header vh = { 0, 1, 0, 0xffff }; + fetch->translate->set_buffer(fetch->translate, + draw->pt.nr_vertex_buffers, + &vh, + 0); + } + } + + fetch->need_edgeflags = ((draw->rasterizer->fill_cw != PIPE_POLYGON_MODE_FILL || + draw->rasterizer->fill_ccw != PIPE_POLYGON_MODE_FILL) && + draw->pt.user.edgeflag); +} + + + + +void draw_pt_fetch_run( struct pt_fetch *fetch, + const unsigned *elts, + unsigned count, + char *verts ) +{ + struct draw_context *draw = fetch->draw; + struct translate *translate = fetch->translate; + unsigned i; + + for (i = 0; i < draw->pt.nr_vertex_buffers; i++) { + translate->set_buffer(translate, + i, + ((char *)draw->pt.user.vbuffer[i] + + draw->pt.vertex_buffer[i].buffer_offset), + draw->pt.vertex_buffer[i].pitch ); + } + + translate->run_elts( translate, + elts, + count, + verts ); + + /* Edgeflags are hard to fit into a translate program, populate + * them separately if required. In the setup above they are + * defaulted to one, so only need this if there is reason to change + * that default: + */ + if (fetch->need_edgeflags) { + for (i = 0; i < count; i++) { + struct vertex_header *vh = (struct vertex_header *)(verts + i * fetch->vertex_size); + vh->edgeflag = draw_pt_get_edgeflag( draw, elts[i] ); + } + } +} + + +struct pt_fetch *draw_pt_fetch_create( struct draw_context *draw ) +{ + struct pt_fetch *fetch = CALLOC_STRUCT(pt_fetch); + if (!fetch) + return NULL; + + fetch->draw = draw; + fetch->cache = translate_cache_create(); + if (!fetch->cache) { + FREE(fetch); + return NULL; + } + + return fetch; +} + +void draw_pt_fetch_destroy( struct pt_fetch *fetch ) +{ + translate_cache_destroy(fetch->cache); + + FREE(fetch); +} + diff --git a/src/gallium/auxiliary/draw/draw_pt_fetch_emit.c b/src/gallium/auxiliary/draw/draw_pt_fetch_emit.c index 0806076956..b7b970a297 100644 --- a/src/gallium/auxiliary/draw/draw_pt_fetch_emit.c +++ b/src/gallium/auxiliary/draw/draw_pt_fetch_emit.c @@ -36,6 +36,8 @@ #include "draw/draw_vbuf.h" #include "draw/draw_vertex.h" #include "draw/draw_pt.h" +#include "translate/translate.h" +#include "translate/translate_cache.h" /* The simplest 'middle end' in the new vertex code. * @@ -72,107 +74,30 @@ struct fetch_emit_middle_end { struct draw_pt_middle_end base; struct draw_context *draw; - - struct { - const ubyte *ptr; - unsigned pitch; - void (*fetch)( const void *from, float *attrib); - void (*emit)( const float *attrib, float **out ); - } fetch[PIPE_MAX_ATTRIBS]; - unsigned nr_fetch; - unsigned hw_vertex_size; -}; - - - -static void fetch_R32_FLOAT( const void *from, - float *attrib ) -{ - float *f = (float *) from; - attrib[0] = f[0]; - attrib[1] = 0.0; - attrib[2] = 0.0; - attrib[3] = 1.0; -} - - -static void emit_R32_FLOAT( const float *attrib, - float **out ) -{ - (*out)[0] = attrib[0]; - (*out) += 1; -} - -static void emit_R32G32_FLOAT( const float *attrib, - float **out ) -{ - (*out)[0] = attrib[0]; - (*out)[1] = attrib[1]; - (*out) += 2; -} + struct translate *translate; + const struct vertex_info *vinfo; -static void emit_R32G32B32_FLOAT( const float *attrib, - float **out ) -{ - (*out)[0] = attrib[0]; - (*out)[1] = attrib[1]; - (*out)[2] = attrib[2]; - (*out) += 3; -} + /* Cache point size somewhere it's address won't change: + */ + float point_size; -static void emit_R32G32B32A32_FLOAT( const float *attrib, - float **out ) -{ - (*out)[0] = attrib[0]; - (*out)[1] = attrib[1]; - (*out)[2] = attrib[2]; - (*out)[3] = attrib[3]; - (*out) += 4; -} + struct translate_cache *cache; +}; -/** - * General-purpose fetch from user's vertex arrays, emit to driver's - * vertex buffer. - * - * XXX this is totally temporary. - */ -static void -fetch_store_general( struct fetch_emit_middle_end *feme, - void *out_ptr, - const unsigned *fetch_elts, - unsigned count ) -{ - float *out = (float *)out_ptr; - struct vbuf_render *render = feme->draw->render; - uint i, j; - - for (i = 0; i < count; i++) { - unsigned elt = fetch_elts[i] & ~DRAW_PT_FLAG_MASK; - - for (j = 0; j < feme->nr_fetch; j++) { - float attrib[4]; - const ubyte *from = (feme->fetch[j].ptr + - feme->fetch[j].pitch * elt); - - feme->fetch[j].fetch( from, attrib ); - feme->fetch[j].emit( attrib, &out ); - } - } -} - static void fetch_emit_prepare( struct draw_pt_middle_end *middle, - unsigned prim ) + unsigned prim, + unsigned opt ) { - static const float zero = 0; struct fetch_emit_middle_end *feme = (struct fetch_emit_middle_end *)middle; struct draw_context *draw = feme->draw; const struct vertex_info *vinfo; - unsigned i; + unsigned i, dst_offset; boolean ok; + struct translate_key key; ok = draw->render->set_primitive( draw->render, @@ -184,55 +109,93 @@ static void fetch_emit_prepare( struct draw_pt_middle_end *middle, /* Must do this after set_primitive() above: */ - vinfo = draw->render->get_vertex_info(draw->render); + vinfo = feme->vinfo = draw->render->get_vertex_info(draw->render); + + - for (i = 0; i < vinfo->num_attribs; i++) { - unsigned src_element = vinfo->src_index[i]; - unsigned src_buffer = draw->vertex_element[src_element].vertex_buffer_index; - - feme->fetch[i].ptr = ((const ubyte *)draw->user.vbuffer[src_buffer] + - draw->vertex_buffer[src_buffer].buffer_offset + - draw->vertex_element[src_element].src_offset); + /* Transform from API vertices to HW vertices, skipping the + * pipeline_vertex intermediate step. + */ + dst_offset = 0; + memset(&key, 0, sizeof(key)); - feme->fetch[i].pitch = draw->vertex_buffer[src_buffer].pitch; - - feme->fetch[i].fetch = draw_get_fetch_func(draw->vertex_element[src_element].src_format); + for (i = 0; i < vinfo->num_attribs; i++) { + const struct pipe_vertex_element *src = &draw->pt.vertex_element[vinfo->src_index[i]]; + unsigned emit_sz = 0; + unsigned input_format = src->src_format; + unsigned input_buffer = src->vertex_buffer_index; + unsigned input_offset = src->src_offset; + unsigned output_format; switch (vinfo->emit[i]) { case EMIT_4F: - feme->fetch[i].emit = emit_R32G32B32A32_FLOAT; + output_format = PIPE_FORMAT_R32G32B32A32_FLOAT; + emit_sz = 4 * sizeof(float); break; case EMIT_3F: - feme->fetch[i].emit = emit_R32G32B32_FLOAT; + output_format = PIPE_FORMAT_R32G32B32_FLOAT; + emit_sz = 3 * sizeof(float); break; case EMIT_2F: - feme->fetch[i].emit = emit_R32G32_FLOAT; + output_format = PIPE_FORMAT_R32G32_FLOAT; + emit_sz = 2 * sizeof(float); break; case EMIT_1F: - feme->fetch[i].emit = emit_R32_FLOAT; - break; - case EMIT_HEADER: - feme->fetch[i].ptr = (const ubyte *)&zero; - feme->fetch[i].pitch = 0; - feme->fetch[i].fetch = fetch_R32_FLOAT; - feme->fetch[i].emit = emit_R32_FLOAT; + output_format = PIPE_FORMAT_R32_FLOAT; + emit_sz = 1 * sizeof(float); break; case EMIT_1F_PSIZE: - feme->fetch[i].ptr = (const ubyte *)&feme->draw->rasterizer->point_size; - feme->fetch[i].pitch = 0; - feme->fetch[i].fetch = fetch_R32_FLOAT; - feme->fetch[i].emit = emit_R32_FLOAT; + input_format = PIPE_FORMAT_R32_FLOAT; + input_buffer = draw->pt.nr_vertex_buffers; + input_offset = 0; + output_format = PIPE_FORMAT_R32_FLOAT; + emit_sz = 1 * sizeof(float); break; default: assert(0); - feme->fetch[i].emit = NULL; - break; + output_format = PIPE_FORMAT_NONE; + emit_sz = 0; + continue; } + + key.element[i].input_format = input_format; + key.element[i].input_buffer = input_buffer; + key.element[i].input_offset = input_offset; + key.element[i].output_format = output_format; + key.element[i].output_offset = dst_offset; + + dst_offset += emit_sz; } - feme->nr_fetch = vinfo->num_attribs; - feme->hw_vertex_size = vinfo->size * 4; + key.nr_elements = vinfo->num_attribs; + key.output_stride = vinfo->size * 4; + + /* Don't bother with caching at this stage: + */ + if (!feme->translate || + translate_key_compare(&feme->translate->key, &key) != 0) + { + translate_key_sanitize(&key); + feme->translate = translate_cache_find(feme->cache, + &key); + + + feme->translate->set_buffer(feme->translate, + draw->pt.nr_vertex_buffers, + &feme->point_size, + 0); + } + + feme->point_size = draw->rasterizer->point_size; + + for (i = 0; i < draw->pt.nr_vertex_buffers; i++) { + feme->translate->set_buffer(feme->translate, + i, + ((char *)draw->pt.user.vbuffer[i] + + draw->pt.vertex_buffer[i].buffer_offset), + draw->pt.vertex_buffer[i].pitch ); + } } @@ -249,8 +212,12 @@ static void fetch_emit_run( struct draw_pt_middle_end *middle, struct draw_context *draw = feme->draw; void *hw_verts; + /* XXX: need to flush to get prim_vbuf.c to release its allocation?? + */ + draw_do_flush( draw, DRAW_FLUSH_BACKEND ); + hw_verts = draw->render->allocate_vertices( draw->render, - (ushort)feme->hw_vertex_size, + (ushort)feme->translate->key.output_stride, (ushort)fetch_count ); if (!hw_verts) { assert(0); @@ -260,10 +227,19 @@ static void fetch_emit_run( struct draw_pt_middle_end *middle, /* Single routine to fetch vertices and emit HW verts. */ - fetch_store_general( feme, - hw_verts, - fetch_elts, - fetch_count ); + feme->translate->run_elts( feme->translate, + fetch_elts, + fetch_count, + hw_verts ); + + if (0) { + unsigned i; + for (i = 0; i < fetch_count; i++) { + debug_printf("\n\nvertex %d:\n", i); + draw_dump_emitted_vertex( feme->vinfo, + (const uint8_t *)hw_verts + feme->vinfo->size * 4 * i ); + } + } /* XXX: Draw arrays path to avoid re-emitting index list again and * again. @@ -276,7 +252,7 @@ static void fetch_emit_run( struct draw_pt_middle_end *middle, */ draw->render->release_vertices( draw->render, hw_verts, - feme->hw_vertex_size, + feme->translate->key.output_stride, fetch_count ); } @@ -290,6 +266,10 @@ static void fetch_emit_finish( struct draw_pt_middle_end *middle ) static void fetch_emit_destroy( struct draw_pt_middle_end *middle ) { + struct fetch_emit_middle_end *feme = (struct fetch_emit_middle_end *)middle; + + translate_cache_destroy(feme->cache); + FREE(middle); } @@ -297,7 +277,15 @@ static void fetch_emit_destroy( struct draw_pt_middle_end *middle ) struct draw_pt_middle_end *draw_pt_fetch_emit( struct draw_context *draw ) { struct fetch_emit_middle_end *fetch_emit = CALLOC_STRUCT( fetch_emit_middle_end ); - + if (fetch_emit == NULL) + return NULL; + + fetch_emit->cache = translate_cache_create(); + if (!fetch_emit->cache) { + FREE(fetch_emit); + return NULL; + } + fetch_emit->base.prepare = fetch_emit_prepare; fetch_emit->base.run = fetch_emit_run; fetch_emit->base.finish = fetch_emit_finish; diff --git a/src/gallium/auxiliary/draw/draw_pt_fetch_pipeline.c b/src/gallium/auxiliary/draw/draw_pt_fetch_pipeline.c deleted file mode 100644 index 4c2a281b29..0000000000 --- a/src/gallium/auxiliary/draw/draw_pt_fetch_pipeline.c +++ /dev/null @@ -1,328 +0,0 @@ -/************************************************************************** - * - * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - - /* - * Authors: - * Keith Whitwell <keith@tungstengraphics.com> - */ - -#include "pipe/p_util.h" -#include "draw/draw_context.h" -#include "draw/draw_private.h" -#include "draw/draw_vertex.h" -#include "draw/draw_pt.h" - -/* The simplest 'middle end' in the new vertex code. - * - * The responsibilities of a middle end are to: - * - perform vertex fetch using - * - draw vertex element/buffer state - * - a list of fetch indices we received as an input - * - run the vertex shader - * - cliptest, - * - clip coord calculation - * - viewport transformation - * - if necessary, run the primitive pipeline, passing it: - * - a linear array of vertex_header vertices constructed here - * - a set of draw indices we received as an input - * - otherwise, drive the hw backend, - * - allocate space for hardware format vertices - * - translate the vertex-shader output vertices to hw format - * - calling the backend draw functions. - * - * For convenience, we provide a helper function to drive the hardware - * backend given similar inputs to those required to run the pipeline. - * - * In the case of passthrough mode, many of these actions are disabled - * or noops, so we end up doing: - * - * - perform vertex fetch - * - drive the hw backend - * - * IE, basically just vertex fetch to post-vs-format vertices, - * followed by a call to the backend helper function. - */ - - -struct fetch_pipeline_middle_end { - struct draw_pt_middle_end base; - struct draw_context *draw; - - void (*header)( unsigned idx, float **out); - - struct { - const ubyte *ptr; - unsigned pitch; - void (*fetch)( const void *from, float *attrib); - void (*emit)( const float *attrib, float **out ); - } fetch[PIPE_MAX_ATTRIBS]; - - unsigned nr_fetch; - unsigned pipeline_vertex_size; - unsigned prim; -}; - - - -static void emit_R32_FLOAT( const float *attrib, - float **out ) -{ - (*out)[0] = attrib[0]; - (*out) += 1; -} - -static void emit_R32G32_FLOAT( const float *attrib, - float **out ) -{ - (*out)[0] = attrib[0]; - (*out)[1] = attrib[1]; - (*out) += 2; -} - -static void emit_R32G32B32_FLOAT( const float *attrib, - float **out ) -{ - (*out)[0] = attrib[0]; - (*out)[1] = attrib[1]; - (*out)[2] = attrib[2]; - (*out) += 3; -} - -static void emit_R32G32B32A32_FLOAT( const float *attrib, - float **out ) -{ - (*out)[0] = attrib[0]; - (*out)[1] = attrib[1]; - (*out)[2] = attrib[2]; - (*out)[3] = attrib[3]; - (*out) += 4; -} - -static void header( unsigned idx, - float **out ) -{ - struct vertex_header *header = (struct vertex_header *) (*out); - - header->clipmask = 0; - header->edgeflag = 1; - header->pad = 0; - header->vertex_id = UNDEFINED_VERTEX_ID; - - (*out)[1] = 0; - (*out)[2] = 0; - (*out)[3] = 0; - (*out)[3] = 1; - (*out) += 5; -} - - -static void header_ef( unsigned idx, - float **out ) -{ - struct vertex_header *header = (struct vertex_header *) (*out); - - /* XXX: need a reset_stipple flag in the vertex header too? - */ - header->clipmask = 0; - header->edgeflag = (idx & DRAW_PT_EDGEFLAG) != 0; - header->pad = 0; - header->vertex_id = UNDEFINED_VERTEX_ID; - - (*out)[1] = 0; - (*out)[2] = 0; - (*out)[3] = 0; - (*out)[3] = 1; - (*out) += 5; -} - - -/** - * General-purpose fetch from user's vertex arrays, emit to driver's - * vertex buffer. - * - * XXX this is totally temporary. - */ -static void -fetch_store_general( struct fetch_pipeline_middle_end *fpme, - void *out_ptr, - const unsigned *fetch_elts, - unsigned count ) -{ - float *out = (float *)out_ptr; - uint i, j; - - for (i = 0; i < count; i++) { - unsigned elt = fetch_elts[i]; - - fpme->header( elt, &out ); - elt &= ~DRAW_PT_FLAG_MASK; - - for (j = 0; j < fpme->nr_fetch; j++) { - float attrib[4]; - const ubyte *from = (fpme->fetch[j].ptr + - fpme->fetch[j].pitch * elt); - - fpme->fetch[j].fetch( from, attrib ); - fpme->fetch[j].emit( attrib, &out ); - } - } -} - - -/* We aren't running a vertex shader, but are running the pipeline. - * That means the vertices we need to build look like: - * - * dw0: vertex header (zero?) - * dw1: clip coord 0 - * dw2: clip coord 1 - * dw3: clip coord 2 - * dw4: clip coord 4 - * dw5: screen coord 0 - * dw6: screen coord 0 - * dw7: screen coord 0 - * dw8: screen coord 0 - * dw9: other attribs... - * - */ -static void fetch_pipeline_prepare( struct draw_pt_middle_end *middle, - unsigned prim ) -{ - static const float zero = 0; - struct fetch_pipeline_middle_end *fpme = (struct fetch_pipeline_middle_end *)middle; - struct draw_context *draw = fpme->draw; - unsigned i, nr = 0; - - fpme->prim = prim; - - /* Emit the vertex header and empty clipspace coord field: - */ - if (draw->user.edgeflag) { - fpme->header = header_ef; - } - else { - fpme->header = header; - } - - - /* Need to look at vertex shader inputs (we know it is a - * passthrough shader, so these define the outputs too). If we - * were running a shader, we'd still be looking at the inputs at - * this point. - */ - for (i = 0; i < draw->vertex_shader->info.num_inputs; i++) { - unsigned buf = draw->vertex_element[i].vertex_buffer_index; - enum pipe_format format = draw->vertex_element[i].src_format; - - fpme->fetch[nr].ptr = ((const ubyte *) draw->user.vbuffer[buf] + - draw->vertex_buffer[buf].buffer_offset + - draw->vertex_element[i].src_offset); - - fpme->fetch[nr].pitch = draw->vertex_buffer[buf].pitch; - fpme->fetch[nr].fetch = draw_get_fetch_func( format ); - - /* Always do this -- somewhat redundant... - */ - fpme->fetch[nr].emit = emit_R32G32B32A32_FLOAT; - nr++; - } - - fpme->nr_fetch = nr; - fpme->pipeline_vertex_size = sizeof(struct vertex_header) + nr * 4 * sizeof(float); -} - - - - -static void fetch_pipeline_run( struct draw_pt_middle_end *middle, - const unsigned *fetch_elts, - unsigned fetch_count, - const ushort *draw_elts, - unsigned draw_count ) -{ - struct fetch_pipeline_middle_end *fpme = (struct fetch_pipeline_middle_end *)middle; - struct draw_context *draw = fpme->draw; - char *pipeline_verts; - - pipeline_verts = MALLOC( fpme->pipeline_vertex_size * - fetch_count ); - if (!pipeline_verts) { - assert(0); - return; - } - - - /* Single routine to fetch vertices and emit pipeline verts. - */ - fetch_store_general( fpme, - pipeline_verts, - fetch_elts, - fetch_count ); - - - /* Run the pipeline - */ - draw_pt_run_pipeline( fpme->draw, - fpme->prim, - pipeline_verts, - fpme->pipeline_vertex_size, - fetch_count, - draw_elts, - draw_count ); - - - /* Done -- that was easy, wasn't it: - */ - FREE( pipeline_verts ); -} - - - -static void fetch_pipeline_finish( struct draw_pt_middle_end *middle ) -{ - /* nothing to do */ -} - -static void fetch_pipeline_destroy( struct draw_pt_middle_end *middle ) -{ - FREE(middle); -} - - -struct draw_pt_middle_end *draw_pt_fetch_pipeline( struct draw_context *draw ) -{ - struct fetch_pipeline_middle_end *fetch_pipeline = CALLOC_STRUCT( fetch_pipeline_middle_end ); - - fetch_pipeline->base.prepare = fetch_pipeline_prepare; - fetch_pipeline->base.run = fetch_pipeline_run; - fetch_pipeline->base.finish = fetch_pipeline_finish; - fetch_pipeline->base.destroy = fetch_pipeline_destroy; - - fetch_pipeline->draw = draw; - - return &fetch_pipeline->base; -} - diff --git a/src/gallium/auxiliary/draw/draw_pt_fetch_shade_pipeline.c b/src/gallium/auxiliary/draw/draw_pt_fetch_shade_pipeline.c new file mode 100644 index 0000000000..4ec20493c4 --- /dev/null +++ b/src/gallium/auxiliary/draw/draw_pt_fetch_shade_pipeline.c @@ -0,0 +1,235 @@ +/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include "pipe/p_util.h" +#include "draw/draw_context.h" +#include "draw/draw_vbuf.h" +#include "draw/draw_vertex.h" +#include "draw/draw_pt.h" +#include "draw/draw_vs.h" +#include "translate/translate.h" + + +struct fetch_pipeline_middle_end { + struct draw_pt_middle_end base; + struct draw_context *draw; + + struct pt_emit *emit; + struct pt_fetch *fetch; + struct pt_post_vs *post_vs; + + unsigned vertex_data_offset; + unsigned vertex_size; + unsigned prim; + unsigned opt; +}; + + +static void fetch_pipeline_prepare( struct draw_pt_middle_end *middle, + unsigned prim, + unsigned opt ) +{ + struct fetch_pipeline_middle_end *fpme = (struct fetch_pipeline_middle_end *)middle; + struct draw_context *draw = fpme->draw; + struct draw_vertex_shader *vs = draw->vertex_shader; + + /* Add one to num_outputs because the pipeline occasionally tags on + * an additional texcoord, eg for AA lines. + */ + unsigned nr = MAX2( vs->info.num_inputs, + vs->info.num_outputs + 1 ); + + fpme->prim = prim; + fpme->opt = opt; + + /* Always leave room for the vertex header whether we need it or + * not. It's hard to get rid of it in particular because of the + * viewport code in draw_pt_post_vs.c. + */ + fpme->vertex_size = sizeof(struct vertex_header) + nr * 4 * sizeof(float); + + + + draw_pt_fetch_prepare( fpme->fetch, + fpme->vertex_size ); + + /* XXX: it's not really gl rasterization rules we care about here, + * but gl vs dx9 clip spaces. + */ + draw_pt_post_vs_prepare( fpme->post_vs, + draw->bypass_clipping, + draw->identity_viewport, + draw->rasterizer->gl_rasterization_rules ); + + + if (!(opt & PT_PIPELINE)) + draw_pt_emit_prepare( fpme->emit, + prim ); + + /* No need to prepare the shader. + */ + vs->prepare(vs, draw); + +} + + + + +static void fetch_pipeline_run( struct draw_pt_middle_end *middle, + const unsigned *fetch_elts, + unsigned fetch_count, + const ushort *draw_elts, + unsigned draw_count ) +{ + struct fetch_pipeline_middle_end *fpme = (struct fetch_pipeline_middle_end *)middle; + struct draw_context *draw = fpme->draw; + struct draw_vertex_shader *shader = draw->vertex_shader; + unsigned opt = fpme->opt; + unsigned alloc_count = align_int( fetch_count, 4 ); + + struct vertex_header *pipeline_verts = + (struct vertex_header *)MALLOC(fpme->vertex_size * alloc_count); + + if (!pipeline_verts) { + /* Not much we can do here - just skip the rendering. + */ + assert(0); + return; + } + + /* Fetch into our vertex buffer + */ + draw_pt_fetch_run( fpme->fetch, + fetch_elts, + fetch_count, + (char *)pipeline_verts ); + + /* Run the shader, note that this overwrites the data[] parts of + * the pipeline verts. If there is no shader, ie a bypass shader, + * then the inputs == outputs, and are already in the correct + * place. + */ + if (opt & PT_SHADE) + { + shader->run_linear(shader, + (const float (*)[4])pipeline_verts->data, + ( float (*)[4])pipeline_verts->data, + (const float (*)[4])draw->pt.user.constants, + fetch_count, + fpme->vertex_size, + fpme->vertex_size); + } + + if (draw_pt_post_vs_run( fpme->post_vs, + pipeline_verts, + fetch_count, + fpme->vertex_size )) + { + opt |= PT_PIPELINE; + } + + /* Do we need to run the pipeline? + */ + if (opt & PT_PIPELINE) { + draw_pipeline_run( fpme->draw, + fpme->prim, + pipeline_verts, + fetch_count, + fpme->vertex_size, + draw_elts, + draw_count ); + } + else { + draw_pt_emit( fpme->emit, + (const float (*)[4])pipeline_verts->data, + fetch_count, + fpme->vertex_size, + draw_elts, + draw_count ); + } + + + FREE(pipeline_verts); +} + + + +static void fetch_pipeline_finish( struct draw_pt_middle_end *middle ) +{ + /* nothing to do */ +} + +static void fetch_pipeline_destroy( struct draw_pt_middle_end *middle ) +{ + struct fetch_pipeline_middle_end *fpme = (struct fetch_pipeline_middle_end *)middle; + + if (fpme->fetch) + draw_pt_fetch_destroy( fpme->fetch ); + + if (fpme->emit) + draw_pt_emit_destroy( fpme->emit ); + + if (fpme->post_vs) + draw_pt_post_vs_destroy( fpme->post_vs ); + + FREE(middle); +} + + +struct draw_pt_middle_end *draw_pt_fetch_pipeline_or_emit( struct draw_context *draw ) +{ + struct fetch_pipeline_middle_end *fpme = CALLOC_STRUCT( fetch_pipeline_middle_end ); + if (!fpme) + goto fail; + + fpme->base.prepare = fetch_pipeline_prepare; + fpme->base.run = fetch_pipeline_run; + fpme->base.finish = fetch_pipeline_finish; + fpme->base.destroy = fetch_pipeline_destroy; + + fpme->draw = draw; + + fpme->fetch = draw_pt_fetch_create( draw ); + if (!fpme->fetch) + goto fail; + + fpme->post_vs = draw_pt_post_vs_create( draw ); + if (!fpme->post_vs) + goto fail; + + fpme->emit = draw_pt_emit_create( draw ); + if (!fpme->emit) + goto fail; + + return &fpme->base; + + fail: + if (fpme) + fetch_pipeline_destroy( &fpme->base ); + + return NULL; +} diff --git a/src/gallium/auxiliary/draw/draw_pt_pipeline.c b/src/gallium/auxiliary/draw/draw_pt_pipeline.c deleted file mode 100644 index e70e63d08f..0000000000 --- a/src/gallium/auxiliary/draw/draw_pt_pipeline.c +++ /dev/null @@ -1,168 +0,0 @@ -/************************************************************************** - * - * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - - /* - * Authors: - * Keith Whitwell <keith@tungstengraphics.com> - */ - -#include "pipe/p_util.h" -#include "draw/draw_context.h" -#include "draw/draw_private.h" -#include "draw/draw_vertex.h" -#include "draw/draw_pt.h" - - -/** - * Add a point to the primitive queue. - * \param i0 index into user's vertex arrays - */ -static void do_point( struct draw_context *draw, - const char *v0 ) -{ - struct prim_header prim; - - prim.reset_line_stipple = 0; - prim.edgeflags = 1; - prim.pad = 0; - prim.v[0] = (struct vertex_header *)v0; - - draw->pipeline.first->point( draw->pipeline.first, &prim ); -} - - -/** - * Add a line to the primitive queue. - * \param i0 index into user's vertex arrays - * \param i1 index into user's vertex arrays - */ -static void do_line( struct draw_context *draw, - const char *v0, - const char *v1 ) -{ - struct prim_header prim; - - prim.reset_line_stipple = 1; /* fixme */ - prim.edgeflags = 1; - prim.pad = 0; - prim.v[0] = (struct vertex_header *)v0; - prim.v[1] = (struct vertex_header *)v1; - - draw->pipeline.first->line( draw->pipeline.first, &prim ); -} - -/** - * Add a triangle to the primitive queue. - */ -static void do_triangle( struct draw_context *draw, - char *v0, - char *v1, - char *v2 ) -{ - struct prim_header prim; - - prim.v[0] = (struct vertex_header *)v0; - prim.v[1] = (struct vertex_header *)v1; - prim.v[2] = (struct vertex_header *)v2; - prim.reset_line_stipple = 1; - prim.edgeflags = ((prim.v[0]->edgeflag) | - (prim.v[1]->edgeflag << 1) | - (prim.v[2]->edgeflag << 2)); - prim.pad = 0; - - if (0) debug_printf("tri ef: %d %d %d\n", - prim.v[0]->edgeflag, - prim.v[1]->edgeflag, - prim.v[2]->edgeflag); - - draw->pipeline.first->tri( draw->pipeline.first, &prim ); -} - - - -void draw_pt_reset_vertex_ids( struct draw_context *draw ) -{ - unsigned i; - char *verts = draw->pt.pipeline.verts; - unsigned stride = draw->pt.pipeline.vertex_stride; - - for (i = 0; i < draw->pt.pipeline.vertex_count; i++) { - ((struct vertex_header *)verts)->vertex_id = UNDEFINED_VERTEX_ID; - verts += stride; - } -} - - -/* Code to run the pipeline on a fairly arbitary collection of vertices. - * - * Vertex headers must be pre-initialized with the - * UNDEFINED_VERTEX_ID, this code will cause that id to become - * overwritten, so it may have to be reset if there is the intention - * to reuse the vertices. - * - * This code provides a callback to reset the vertex id's which the - * draw_vbuf.c code uses when it has to perform a flush. - */ -void draw_pt_run_pipeline( struct draw_context *draw, - unsigned prim, - char *verts, - unsigned stride, - unsigned vertex_count, - const ushort *elts, - unsigned count ) -{ - unsigned i; - - draw->pt.pipeline.verts = verts; - draw->pt.pipeline.vertex_stride = stride; - draw->pt.pipeline.vertex_count = vertex_count; - - switch (prim) { - case PIPE_PRIM_POINTS: - for (i = 0; i < count; i++) - do_point( draw, - verts + stride * elts[i] ); - break; - case PIPE_PRIM_LINES: - for (i = 0; i+1 < count; i += 2) - do_line( draw, - verts + stride * elts[i+0], - verts + stride * elts[i+1]); - break; - case PIPE_PRIM_TRIANGLES: - for (i = 0; i+2 < count; i += 3) - do_triangle( draw, - verts + stride * elts[i+0], - verts + stride * elts[i+1], - verts + stride * elts[i+2]); - break; - } - - draw->pt.pipeline.verts = NULL; - draw->pt.pipeline.vertex_count = 0; -} - diff --git a/src/gallium/auxiliary/draw/draw_pt_post_vs.c b/src/gallium/auxiliary/draw/draw_pt_post_vs.c new file mode 100644 index 0000000000..c4a67c8289 --- /dev/null +++ b/src/gallium/auxiliary/draw/draw_pt_post_vs.c @@ -0,0 +1,227 @@ +/************************************************************************** + * + * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include "pipe/p_util.h" +#include "pipe/p_context.h" +#include "draw/draw_context.h" +#include "draw/draw_private.h" +#include "draw/draw_vbuf.h" +#include "draw/draw_vertex.h" +#include "draw/draw_pt.h" + +struct pt_post_vs { + struct draw_context *draw; + + boolean (*run)( struct pt_post_vs *pvs, + struct vertex_header *vertices, + unsigned count, + unsigned stride ); +}; + + + +static INLINE float +dot4(const float *a, const float *b) +{ + return (a[0]*b[0] + + a[1]*b[1] + + a[2]*b[2] + + a[3]*b[3]); +} + + + +static INLINE unsigned +compute_clipmask_gl(const float *clip, /*const*/ float plane[][4], unsigned nr) +{ + unsigned mask = 0x0; + unsigned i; + +#if 0 + debug_printf("compute clipmask %f %f %f %f\n", + clip[0], clip[1], clip[2], clip[3]); + assert(clip[3] != 0.0); +#endif + + /* Do the hardwired planes first: + */ + if (-clip[0] + clip[3] < 0) mask |= (1<<0); + if ( clip[0] + clip[3] < 0) mask |= (1<<1); + if (-clip[1] + clip[3] < 0) mask |= (1<<2); + if ( clip[1] + clip[3] < 0) mask |= (1<<3); + if ( clip[2] + clip[3] < 0) mask |= (1<<4); /* match mesa clipplane numbering - for now */ + if (-clip[2] + clip[3] < 0) mask |= (1<<5); /* match mesa clipplane numbering - for now */ + + /* Followed by any remaining ones: + */ + for (i = 6; i < nr; i++) { + if (dot4(clip, plane[i]) < 0) + mask |= (1<<i); + } + + return mask; +} + + +/* The normal case - cliptest, rhw divide, viewport transform. + * + * Also handle identity viewport here at the expense of a few wasted + * instructions + */ +static boolean post_vs_cliptest_viewport_gl( struct pt_post_vs *pvs, + struct vertex_header *vertices, + unsigned count, + unsigned stride ) +{ + struct vertex_header *out = vertices; + const float *scale = pvs->draw->viewport.scale; + const float *trans = pvs->draw->viewport.translate; + unsigned clipped = 0; + unsigned j; + + if (0) debug_printf("%s\n"); + + for (j = 0; j < count; j++) { + out->clip[0] = out->data[0][0]; + out->clip[1] = out->data[0][1]; + out->clip[2] = out->data[0][2]; + out->clip[3] = out->data[0][3]; + + out->vertex_id = 0xffff; + out->clipmask = compute_clipmask_gl(out->clip, + pvs->draw->plane, + pvs->draw->nr_planes); + clipped += out->clipmask; + + if (out->clipmask == 0) + { + /* divide by w */ + float w = 1.0f / out->data[0][3]; + + /* Viewport mapping */ + out->data[0][0] = out->data[0][0] * w * scale[0] + trans[0]; + out->data[0][1] = out->data[0][1] * w * scale[1] + trans[1]; + out->data[0][2] = out->data[0][2] * w * scale[2] + trans[2]; + out->data[0][3] = w; +#if 0 + debug_printf("post viewport: %f %f %f %f\n", + out->data[0][0], + out->data[0][1], + out->data[0][2], + out->data[0][3]); +#endif + } + + out = (struct vertex_header *)( (char *)out + stride ); + } + + return clipped != 0; +} + + + +/* If bypass_clipping is set, skip cliptest and rhw divide. + */ +static boolean post_vs_viewport( struct pt_post_vs *pvs, + struct vertex_header *vertices, + unsigned count, + unsigned stride ) +{ + struct vertex_header *out = vertices; + const float *scale = pvs->draw->viewport.scale; + const float *trans = pvs->draw->viewport.translate; + unsigned j; + + if (0) debug_printf("%s\n", __FUNCTION__); + for (j = 0; j < count; j++) { + /* Viewport mapping only, no cliptest/rhw divide + */ + out->data[0][0] = out->data[0][0] * scale[0] + trans[0]; + out->data[0][1] = out->data[0][1] * scale[1] + trans[1]; + out->data[0][2] = out->data[0][2] * scale[2] + trans[2]; + + out = (struct vertex_header *)((char *)out + stride); + } + + return FALSE; +} + + +/* If bypass_clipping is set and we have an identity viewport, nothing + * to do. + */ +static boolean post_vs_none( struct pt_post_vs *pvs, + struct vertex_header *vertices, + unsigned count, + unsigned stride ) +{ + if (0) debug_printf("%s\n", __FUNCTION__); + return FALSE; +} + +boolean draw_pt_post_vs_run( struct pt_post_vs *pvs, + struct vertex_header *pipeline_verts, + unsigned count, + unsigned stride ) +{ + return pvs->run( pvs, pipeline_verts, count, stride ); +} + + +void draw_pt_post_vs_prepare( struct pt_post_vs *pvs, + boolean bypass_clipping, + boolean identity_viewport, + boolean opengl ) +{ + if (bypass_clipping) { + if (identity_viewport) + pvs->run = post_vs_none; + else + pvs->run = post_vs_viewport; + } + else { + //if (opengl) + pvs->run = post_vs_cliptest_viewport_gl; + } +} + + +struct pt_post_vs *draw_pt_post_vs_create( struct draw_context *draw ) +{ + struct pt_post_vs *pvs = CALLOC_STRUCT( pt_post_vs ); + if (!pvs) + return NULL; + + pvs->draw = draw; + + return pvs; +} + +void draw_pt_post_vs_destroy( struct pt_post_vs *pvs ) +{ + FREE(pvs); +} diff --git a/src/gallium/auxiliary/draw/draw_pt_varray.c b/src/gallium/auxiliary/draw/draw_pt_varray.c new file mode 100644 index 0000000000..355093f945 --- /dev/null +++ b/src/gallium/auxiliary/draw/draw_pt_varray.c @@ -0,0 +1,249 @@ +/************************************************************************** + * + * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include "pipe/p_util.h" +#include "draw/draw_context.h" +#include "draw/draw_private.h" +#include "draw/draw_pt.h" + +#define FETCH_MAX 256 +#define DRAW_MAX (16*FETCH_MAX) + +struct varray_frontend { + struct draw_pt_front_end base; + struct draw_context *draw; + + ushort draw_elts[DRAW_MAX]; + unsigned fetch_elts[FETCH_MAX]; + + unsigned draw_count; + unsigned fetch_count; + + struct draw_pt_middle_end *middle; + + unsigned input_prim; + unsigned output_prim; +}; + +static void varray_flush(struct varray_frontend *varray) +{ + if (varray->draw_count) { +#if 0 + debug_printf("FLUSH fc = %d, dc = %d\n", + varray->fetch_count, + varray->draw_count); +#endif + varray->middle->run(varray->middle, + varray->fetch_elts, + varray->fetch_count, + varray->draw_elts, + varray->draw_count); + } + + varray->fetch_count = 0; + varray->draw_count = 0; +} + +#if 0 +static void varray_check_flush(struct varray_frontend *varray) +{ + if (varray->draw_count + 6 >= DRAW_MAX/* || + varray->fetch_count + 4 >= FETCH_MAX*/) { + varray_flush(varray); + } +} +#endif + +static INLINE void add_draw_el(struct varray_frontend *varray, + int idx, ushort flags) +{ + varray->draw_elts[varray->draw_count++] = idx | flags; +} + + +static INLINE void varray_triangle( struct varray_frontend *varray, + unsigned i0, + unsigned i1, + unsigned i2 ) +{ + add_draw_el(varray, i0, 0); + add_draw_el(varray, i1, 0); + add_draw_el(varray, i2, 0); +} + +static INLINE void varray_triangle_flags( struct varray_frontend *varray, + ushort flags, + unsigned i0, + unsigned i1, + unsigned i2 ) +{ + add_draw_el(varray, i0, flags); + add_draw_el(varray, i1, 0); + add_draw_el(varray, i2, 0); +} + +static INLINE void varray_line( struct varray_frontend *varray, + unsigned i0, + unsigned i1 ) +{ + add_draw_el(varray, i0, 0); + add_draw_el(varray, i1, 0); +} + + +static INLINE void varray_line_flags( struct varray_frontend *varray, + ushort flags, + unsigned i0, + unsigned i1 ) +{ + add_draw_el(varray, i0, flags); + add_draw_el(varray, i1, 0); +} + + +static INLINE void varray_point( struct varray_frontend *varray, + unsigned i0 ) +{ + add_draw_el(varray, i0, 0); +} + +static INLINE void varray_quad( struct varray_frontend *varray, + unsigned i0, + unsigned i1, + unsigned i2, + unsigned i3 ) +{ + varray_triangle( varray, i0, i1, i3 ); + varray_triangle( varray, i1, i2, i3 ); +} + +static INLINE void varray_ef_quad( struct varray_frontend *varray, + unsigned i0, + unsigned i1, + unsigned i2, + unsigned i3 ) +{ + const unsigned omitEdge1 = DRAW_PIPE_EDGE_FLAG_0 | DRAW_PIPE_EDGE_FLAG_2; + const unsigned omitEdge2 = DRAW_PIPE_EDGE_FLAG_0 | DRAW_PIPE_EDGE_FLAG_1; + + varray_triangle_flags( varray, + DRAW_PIPE_RESET_STIPPLE | omitEdge1, + i0, i1, i3 ); + + varray_triangle_flags( varray, + omitEdge2, + i1, i2, i3 ); +} + +/* At least for now, we're back to using a template include file for + * this. The two paths aren't too different though - it may be + * possible to reunify them. + */ +#define TRIANGLE(vc,flags,i0,i1,i2) varray_triangle_flags(vc,flags,i0,i1,i2) +#define QUAD(vc,i0,i1,i2,i3) varray_ef_quad(vc,i0,i1,i2,i3) +#define LINE(vc,flags,i0,i1) varray_line_flags(vc,flags,i0,i1) +#define POINT(vc,i0) varray_point(vc,i0) +#define FUNC varray_run_extras +#include "draw_pt_varray_tmp.h" + +#define TRIANGLE(vc,flags,i0,i1,i2) varray_triangle(vc,i0,i1,i2) +#define QUAD(vc,i0,i1,i2,i3) varray_quad(vc,i0,i1,i2,i3) +#define LINE(vc,flags,i0,i1) varray_line(vc,i0,i1) +#define POINT(vc,i0) varray_point(vc,i0) +#define FUNC varray_run +#include "draw_pt_varray_tmp.h" + + + +static unsigned reduced_prim[PIPE_PRIM_POLYGON + 1] = { + PIPE_PRIM_POINTS, + PIPE_PRIM_LINES, + PIPE_PRIM_LINES, + PIPE_PRIM_LINES, + PIPE_PRIM_TRIANGLES, + PIPE_PRIM_TRIANGLES, + PIPE_PRIM_TRIANGLES, + PIPE_PRIM_TRIANGLES, + PIPE_PRIM_TRIANGLES, + PIPE_PRIM_TRIANGLES +}; + + + +static void varray_prepare(struct draw_pt_front_end *frontend, + unsigned prim, + struct draw_pt_middle_end *middle, + unsigned opt) +{ + struct varray_frontend *varray = (struct varray_frontend *)frontend; + + if (opt & PT_PIPELINE) + { + varray->base.run = varray_run_extras; + } + else + { + varray->base.run = varray_run; + } + + varray->input_prim = prim; + varray->output_prim = reduced_prim[prim]; + + varray->middle = middle; + middle->prepare(middle, varray->output_prim, opt); +} + + + + +static void varray_finish(struct draw_pt_front_end *frontend) +{ + struct varray_frontend *varray = (struct varray_frontend *)frontend; + varray->middle->finish(varray->middle); + varray->middle = NULL; +} + +static void varray_destroy(struct draw_pt_front_end *frontend) +{ + FREE(frontend); +} + + +struct draw_pt_front_end *draw_pt_varray(struct draw_context *draw) +{ + struct varray_frontend *varray = CALLOC_STRUCT(varray_frontend); + if (varray == NULL) + return NULL; + + varray->base.prepare = varray_prepare; + varray->base.run = NULL; + varray->base.finish = varray_finish; + varray->base.destroy = varray_destroy; + varray->draw = draw; + + return &varray->base; +} diff --git a/src/gallium/auxiliary/draw/draw_pt_varray_tmp.h b/src/gallium/auxiliary/draw/draw_pt_varray_tmp.h new file mode 100644 index 0000000000..b9a319b253 --- /dev/null +++ b/src/gallium/auxiliary/draw/draw_pt_varray_tmp.h @@ -0,0 +1,144 @@ + +static void FUNC(struct draw_pt_front_end *frontend, + pt_elt_func get_elt, + const void *elts, + unsigned count) +{ + struct varray_frontend *varray = (struct varray_frontend *)frontend; + struct draw_context *draw = varray->draw; + unsigned start = (unsigned)elts; + + boolean flatfirst = (draw->rasterizer->flatshade && + draw->rasterizer->flatshade_first); + unsigned i, flags; + +#if 0 + debug_printf("%s (%d) %d/%d\n", __FUNCTION__, draw->prim, start, count); +#endif +#if 0 + debug_printf("INPUT PRIM = %d (start = %d, count = %d)\n", varray->input_prim, + start, count); +#endif + + for (i = 0; i < count; ++i) { + varray->fetch_elts[i] = start + i; + } + varray->fetch_count = count; + + switch (varray->input_prim) { + case PIPE_PRIM_POINTS: + for (i = 0; i < count; i ++) { + POINT(varray, i + 0); + } + break; + + case PIPE_PRIM_LINES: + for (i = 0; i+1 < count; i += 2) { + LINE(varray, DRAW_PIPE_RESET_STIPPLE, + i + 0, i + 1); + } + break; + + case PIPE_PRIM_LINE_LOOP: + if (count >= 2) { + flags = DRAW_PIPE_RESET_STIPPLE; + + for (i = 1; i < count; i++, flags = 0) { + LINE(varray, flags, i - 1, i); + } + LINE(varray, flags, i - 1, 0); + } + break; + + case PIPE_PRIM_LINE_STRIP: + flags = DRAW_PIPE_RESET_STIPPLE; + for (i = 1; i < count; i++, flags = 0) { + LINE(varray, flags, i - 1, i); + } + break; + + case PIPE_PRIM_TRIANGLES: + for (i = 0; i+2 < count; i += 3) { + TRIANGLE(varray, DRAW_PIPE_RESET_STIPPLE | DRAW_PIPE_EDGE_FLAG_ALL, + i + 0, i + 1, i + 2); + } + break; + + case PIPE_PRIM_TRIANGLE_STRIP: + if (flatfirst) { + for (i = 0; i+2 < count; i++) { + TRIANGLE(varray, DRAW_PIPE_RESET_STIPPLE | DRAW_PIPE_EDGE_FLAG_ALL, + i + 0, i + 1 + (i&1), i + 2 - (i&1)); + } + } + else { + for (i = 0; i+2 < count; i++) { + TRIANGLE(varray, DRAW_PIPE_RESET_STIPPLE | DRAW_PIPE_EDGE_FLAG_ALL, + i + 0 + (i&1), i + 1 - (i&1), i + 2); + } + } + break; + + case PIPE_PRIM_TRIANGLE_FAN: + if (count >= 3) { + if (flatfirst) { + flags = DRAW_PIPE_RESET_STIPPLE | DRAW_PIPE_EDGE_FLAG_ALL; + for (i = 0; i+2 < count; i++) { + TRIANGLE(varray, flags, i + 1, i + 2, 0); + } + } + else { + flags = DRAW_PIPE_RESET_STIPPLE | DRAW_PIPE_EDGE_FLAG_ALL; + for (i = 0; i+2 < count; i++) { + TRIANGLE(varray, flags, 0, i + 1, i + 2); + } + } + } + break; + + case PIPE_PRIM_QUADS: + for (i = 0; i+3 < count; i += 4) { + QUAD(varray, i + 0, i + 1, i + 2, i + 3); + } + break; + + case PIPE_PRIM_QUAD_STRIP: + for (i = 0; i+3 < count; i += 2) { + QUAD(varray, i + 2, i + 0, i + 1, i + 3); + } + break; + + case PIPE_PRIM_POLYGON: + { + /* These bitflags look a little odd because we submit the + * vertices as (1,2,0) to satisfy flatshade requirements. + */ + const unsigned edge_first = DRAW_PIPE_EDGE_FLAG_2; + const unsigned edge_middle = DRAW_PIPE_EDGE_FLAG_0; + const unsigned edge_last = DRAW_PIPE_EDGE_FLAG_1; + + flags = DRAW_PIPE_RESET_STIPPLE | edge_first | edge_middle; + + for (i = 0; i+2 < count; i++, flags = edge_middle) { + + if (i + 3 == count) + flags |= edge_last; + + TRIANGLE(varray, flags, i + 1, i + 2, 0); + } + } + break; + + default: + assert(0); + break; + } + + varray_flush(varray); +} + +#undef TRIANGLE +#undef QUAD +#undef POINT +#undef LINE +#undef FUNC diff --git a/src/gallium/auxiliary/draw/draw_pt_vcache.c b/src/gallium/auxiliary/draw/draw_pt_vcache.c index 107dcfc269..6b3fb1406b 100644 --- a/src/gallium/auxiliary/draw/draw_pt_vcache.c +++ b/src/gallium/auxiliary/draw/draw_pt_vcache.c @@ -33,8 +33,6 @@ #include "pipe/p_util.h" #include "draw/draw_context.h" #include "draw/draw_private.h" -//#include "draw/draw_vbuf.h" -//#include "draw/draw_vertex.h" #include "draw/draw_pt.h" @@ -86,8 +84,9 @@ static void vcache_check_flush( struct vcache_frontend *vcache ) } -static void vcache_elt( struct vcache_frontend *vcache, - unsigned felt ) +static INLINE void vcache_elt( struct vcache_frontend *vcache, + unsigned felt, + ushort flags ) { unsigned idx = felt % CACHE_MAX; @@ -99,28 +98,9 @@ static void vcache_elt( struct vcache_frontend *vcache, vcache->fetch_elts[vcache->fetch_count++] = felt; } - vcache->draw_elts[vcache->draw_count++] = vcache->out[idx]; + vcache->draw_elts[vcache->draw_count++] = vcache->out[idx] | flags; } -static unsigned add_edgeflag( struct vcache_frontend *vcache, - unsigned idx, - unsigned mask ) -{ - if (mask && draw_get_edgeflag(vcache->draw, idx)) - return idx | DRAW_PT_EDGEFLAG; - else - return idx; -} - - -static unsigned add_reset_stipple( unsigned idx, - unsigned reset ) -{ - if (reset) - return idx | DRAW_PT_RESET_STIPPLE; - else - return idx; -} static void vcache_triangle( struct vcache_frontend *vcache, @@ -128,48 +108,42 @@ static void vcache_triangle( struct vcache_frontend *vcache, unsigned i1, unsigned i2 ) { - vcache_elt(vcache, i0 | DRAW_PT_EDGEFLAG | DRAW_PT_RESET_STIPPLE); - vcache_elt(vcache, i1 | DRAW_PT_EDGEFLAG); - vcache_elt(vcache, i2 | DRAW_PT_EDGEFLAG); + vcache_elt(vcache, i0, 0); + vcache_elt(vcache, i1, 0); + vcache_elt(vcache, i2, 0); vcache_check_flush(vcache); } -static void vcache_ef_triangle( struct vcache_frontend *vcache, - boolean reset_stipple, - unsigned ef_mask, - unsigned i0, - unsigned i1, - unsigned i2 ) +static void vcache_triangle_flags( struct vcache_frontend *vcache, + ushort flags, + unsigned i0, + unsigned i1, + unsigned i2 ) { - i0 = add_edgeflag( vcache, i0, (ef_mask >> 0) & 1 ); - i1 = add_edgeflag( vcache, i1, (ef_mask >> 1) & 1 ); - i2 = add_edgeflag( vcache, i2, (ef_mask >> 2) & 1 ); - - i0 = add_reset_stipple( i0, reset_stipple ); - - vcache_elt(vcache, i0); - vcache_elt(vcache, i1); - vcache_elt(vcache, i2); + vcache_elt(vcache, i0, flags); + vcache_elt(vcache, i1, 0); + vcache_elt(vcache, i2, 0); vcache_check_flush(vcache); - - if (0) debug_printf("emit tri ef: %d %d %d\n", - !!(i0 & DRAW_PT_EDGEFLAG), - !!(i1 & DRAW_PT_EDGEFLAG), - !!(i2 & DRAW_PT_EDGEFLAG)); - } - static void vcache_line( struct vcache_frontend *vcache, - boolean reset_stipple, unsigned i0, unsigned i1 ) { - i0 = add_reset_stipple( i0, reset_stipple ); + vcache_elt(vcache, i0, 0); + vcache_elt(vcache, i1, 0); + vcache_check_flush(vcache); +} + - vcache_elt(vcache, i0); - vcache_elt(vcache, i1); +static void vcache_line_flags( struct vcache_frontend *vcache, + ushort flags, + unsigned i0, + unsigned i1 ) +{ + vcache_elt(vcache, i0, flags); + vcache_elt(vcache, i1, 0); vcache_check_flush(vcache); } @@ -177,7 +151,7 @@ static void vcache_line( struct vcache_frontend *vcache, static void vcache_point( struct vcache_frontend *vcache, unsigned i0 ) { - vcache_elt(vcache, i0); + vcache_elt(vcache, i0, 0); vcache_check_flush(vcache); } @@ -197,237 +171,36 @@ static void vcache_ef_quad( struct vcache_frontend *vcache, unsigned i2, unsigned i3 ) { - const unsigned omitEdge2 = ~(1 << 1); - const unsigned omitEdge3 = ~(1 << 2); - vcache_ef_triangle( vcache, 1, omitEdge2, i0, i1, i3 ); - vcache_ef_triangle( vcache, 0, omitEdge3, i1, i2, i3 ); -} + const unsigned omitEdge1 = DRAW_PIPE_EDGE_FLAG_0 | DRAW_PIPE_EDGE_FLAG_2; + const unsigned omitEdge2 = DRAW_PIPE_EDGE_FLAG_0 | DRAW_PIPE_EDGE_FLAG_1; + vcache_triangle_flags( vcache, + DRAW_PIPE_RESET_STIPPLE | omitEdge1, + i0, i1, i3 ); + vcache_triangle_flags( vcache, + omitEdge2, + i1, i2, i3 ); +} +/* At least for now, we're back to using a template include file for + * this. The two paths aren't too different though - it may be + * possible to reunify them. + */ +#define TRIANGLE(vc,flags,i0,i1,i2) vcache_triangle_flags(vc,flags,i0,i1,i2) +#define QUAD(vc,i0,i1,i2,i3) vcache_ef_quad(vc,i0,i1,i2,i3) +#define LINE(vc,flags,i0,i1) vcache_line_flags(vc,flags,i0,i1) +#define POINT(vc,i0) vcache_point(vc,i0) +#define FUNC vcache_run_extras +#include "draw_pt_vcache_tmp.h" + +#define TRIANGLE(vc,flags,i0,i1,i2) vcache_triangle(vc,i0,i1,i2) +#define QUAD(vc,i0,i1,i2,i3) vcache_quad(vc,i0,i1,i2,i3) +#define LINE(vc,flags,i0,i1) vcache_line(vc,i0,i1) +#define POINT(vc,i0) vcache_point(vc,i0) +#define FUNC vcache_run +#include "draw_pt_vcache_tmp.h" -static void vcache_run( struct draw_pt_front_end *frontend, - pt_elt_func get_elt, - const void *elts, - unsigned count ) -{ - struct vcache_frontend *vcache = (struct vcache_frontend *)frontend; - struct draw_context *draw = vcache->draw; - - boolean unfilled = (draw->rasterizer->fill_cw != PIPE_POLYGON_MODE_FILL || - draw->rasterizer->fill_ccw != PIPE_POLYGON_MODE_FILL); - - boolean flatfirst = (draw->rasterizer->flatshade && - draw->rasterizer->flatshade_first); - unsigned i; - -// debug_printf("%s (%d) %d/%d\n", __FUNCTION__, draw->prim, start, count ); - - switch (vcache->input_prim) { - case PIPE_PRIM_POINTS: - for (i = 0; i < count; i ++) { - vcache_point( vcache, - get_elt(elts, i + 0) ); - } - break; - - case PIPE_PRIM_LINES: - for (i = 0; i+1 < count; i += 2) { - vcache_line( vcache, - TRUE, - get_elt(elts, i + 0), - get_elt(elts, i + 1)); - } - break; - - case PIPE_PRIM_LINE_LOOP: - if (count >= 2) { - for (i = 1; i < count; i++) { - vcache_line( vcache, - i == 1, /* XXX: only if vb not split */ - get_elt(elts, i - 1), - get_elt(elts, i )); - } - - vcache_line( vcache, - 0, - get_elt(elts, count - 1), - get_elt(elts, 0 )); - } - break; - - case PIPE_PRIM_LINE_STRIP: - for (i = 1; i < count; i++) { - vcache_line( vcache, - i == 1, - get_elt(elts, i - 1), - get_elt(elts, i )); - } - break; - - case PIPE_PRIM_TRIANGLES: - if (unfilled) { - for (i = 0; i+2 < count; i += 3) { - vcache_ef_triangle( vcache, - 1, - ~0, - get_elt(elts, i + 0), - get_elt(elts, i + 1), - get_elt(elts, i + 2 )); - } - } - else { - for (i = 0; i+2 < count; i += 3) { - vcache_triangle( vcache, - get_elt(elts, i + 0), - get_elt(elts, i + 1), - get_elt(elts, i + 2 )); - } - } - break; - - case PIPE_PRIM_TRIANGLE_STRIP: - if (flatfirst) { - for (i = 0; i+2 < count; i++) { - if (i & 1) { - vcache_triangle( vcache, - get_elt(elts, i + 0), - get_elt(elts, i + 2), - get_elt(elts, i + 1 )); - } - else { - vcache_triangle( vcache, - get_elt(elts, i + 0), - get_elt(elts, i + 1), - get_elt(elts, i + 2 )); - } - } - } - else { - for (i = 0; i+2 < count; i++) { - if (i & 1) { - vcache_triangle( vcache, - get_elt(elts, i + 1), - get_elt(elts, i + 0), - get_elt(elts, i + 2 )); - } - else { - vcache_triangle( vcache, - get_elt(elts, i + 0), - get_elt(elts, i + 1), - get_elt(elts, i + 2 )); - } - } - } - break; - - case PIPE_PRIM_TRIANGLE_FAN: - if (count >= 3) { - if (flatfirst) { - for (i = 0; i+2 < count; i++) { - vcache_triangle( vcache, - get_elt(elts, i + 1), - get_elt(elts, i + 2), - get_elt(elts, 0 )); - } - } - else { - for (i = 0; i+2 < count; i++) { - vcache_triangle( vcache, - get_elt(elts, 0), - get_elt(elts, i + 1), - get_elt(elts, i + 2 )); - } - } - } - break; - - - case PIPE_PRIM_QUADS: - if (unfilled) { - for (i = 0; i+3 < count; i += 4) { - vcache_ef_quad( vcache, - get_elt(elts, i + 0), - get_elt(elts, i + 1), - get_elt(elts, i + 2), - get_elt(elts, i + 3)); - } - } - else { - for (i = 0; i+3 < count; i += 4) { - vcache_quad( vcache, - get_elt(elts, i + 0), - get_elt(elts, i + 1), - get_elt(elts, i + 2), - get_elt(elts, i + 3)); - } - } - break; - - case PIPE_PRIM_QUAD_STRIP: - if (unfilled) { - for (i = 0; i+3 < count; i += 2) { - vcache_ef_quad( vcache, - get_elt(elts, i + 2), - get_elt(elts, i + 0), - get_elt(elts, i + 1), - get_elt(elts, i + 3)); - } - } - else { - for (i = 0; i+3 < count; i += 2) { - vcache_quad( vcache, - get_elt(elts, i + 2), - get_elt(elts, i + 0), - get_elt(elts, i + 1), - get_elt(elts, i + 3)); - } - } - break; - - case PIPE_PRIM_POLYGON: - if (unfilled) { - /* These bitflags look a little odd because we submit the - * vertices as (1,2,0) to satisfy flatshade requirements. - */ - const unsigned edge_first = (1<<2); - const unsigned edge_middle = (1<<0); - const unsigned edge_last = (1<<1); - - for (i = 0; i+2 < count; i++) { - unsigned ef_mask = edge_middle; - - if (i == 0) - ef_mask |= edge_first; - - if (i + 3 == count) - ef_mask |= edge_last; - - vcache_ef_triangle( vcache, - i == 0, - ef_mask, - get_elt(elts, i + 1), - get_elt(elts, i + 2), - get_elt(elts, 0)); - } - } - else { - for (i = 0; i+2 < count; i++) { - vcache_triangle( vcache, - get_elt(elts, i + 1), - get_elt(elts, i + 2), - get_elt(elts, 0)); - } - } - break; - - default: - assert(0); - break; - } - - vcache_flush( vcache ); -} @@ -448,23 +221,25 @@ static unsigned reduced_prim[PIPE_PRIM_POLYGON + 1] = { static void vcache_prepare( struct draw_pt_front_end *frontend, unsigned prim, - struct draw_pt_middle_end *middle ) + struct draw_pt_middle_end *middle, + unsigned opt ) { struct vcache_frontend *vcache = (struct vcache_frontend *)frontend; -/* - if (vcache->draw->rasterizer->flatshade_first) - vcache->base.run = vcache_run_pv0; - else - vcache->base.run = vcache_run_pv2; -*/ + if (opt & PT_PIPELINE) + { + vcache->base.run = vcache_run_extras; + } + else + { + vcache->base.run = vcache_run; + } - vcache->base.run = vcache_run; vcache->input_prim = prim; vcache->output_prim = reduced_prim[prim]; vcache->middle = middle; - middle->prepare( middle, vcache->output_prim ); + middle->prepare( middle, vcache->output_prim, opt ); } @@ -486,6 +261,8 @@ static void vcache_destroy( struct draw_pt_front_end *frontend ) struct draw_pt_front_end *draw_pt_vcache( struct draw_context *draw ) { struct vcache_frontend *vcache = CALLOC_STRUCT( vcache_frontend ); + if (vcache == NULL) + return NULL; vcache->base.prepare = vcache_prepare; vcache->base.run = NULL; diff --git a/src/gallium/auxiliary/draw/draw_pt_vcache_tmp.h b/src/gallium/auxiliary/draw/draw_pt_vcache_tmp.h new file mode 100644 index 0000000000..cf9f394aa3 --- /dev/null +++ b/src/gallium/auxiliary/draw/draw_pt_vcache_tmp.h @@ -0,0 +1,174 @@ + + +static void FUNC( struct draw_pt_front_end *frontend, + pt_elt_func get_elt, + const void *elts, + unsigned count ) +{ + struct vcache_frontend *vcache = (struct vcache_frontend *)frontend; + struct draw_context *draw = vcache->draw; + + boolean flatfirst = (draw->rasterizer->flatshade && + draw->rasterizer->flatshade_first); + unsigned i, flags; + + + switch (vcache->input_prim) { + case PIPE_PRIM_POINTS: + for (i = 0; i < count; i ++) { + POINT( vcache, + get_elt(elts, i + 0) ); + } + break; + + case PIPE_PRIM_LINES: + for (i = 0; i+1 < count; i += 2) { + LINE( vcache, + DRAW_PIPE_RESET_STIPPLE, + get_elt(elts, i + 0), + get_elt(elts, i + 1)); + } + break; + + case PIPE_PRIM_LINE_LOOP: + if (count >= 2) { + flags = DRAW_PIPE_RESET_STIPPLE; + + for (i = 1; i < count; i++, flags = 0) { + LINE( vcache, + flags, + get_elt(elts, i - 1), + get_elt(elts, i )); + } + + LINE( vcache, + flags, + get_elt(elts, i - 1), + get_elt(elts, 0 )); + } + break; + + case PIPE_PRIM_LINE_STRIP: + flags = DRAW_PIPE_RESET_STIPPLE; + for (i = 1; i < count; i++, flags = 0) { + LINE( vcache, + flags, + get_elt(elts, i - 1), + get_elt(elts, i )); + } + break; + + case PIPE_PRIM_TRIANGLES: + for (i = 0; i+2 < count; i += 3) { + TRIANGLE( vcache, + DRAW_PIPE_RESET_STIPPLE | DRAW_PIPE_EDGE_FLAG_ALL, + get_elt(elts, i + 0), + get_elt(elts, i + 1), + get_elt(elts, i + 2 )); + } + break; + + case PIPE_PRIM_TRIANGLE_STRIP: + if (flatfirst) { + for (i = 0; i+2 < count; i++) { + TRIANGLE( vcache, + DRAW_PIPE_RESET_STIPPLE | DRAW_PIPE_EDGE_FLAG_ALL, + get_elt(elts, i + 0), + get_elt(elts, i + 1 + (i&1)), + get_elt(elts, i + 2 - (i&1))); + } + } + else { + for (i = 0; i+2 < count; i++) { + TRIANGLE( vcache, + DRAW_PIPE_RESET_STIPPLE | DRAW_PIPE_EDGE_FLAG_ALL, + get_elt(elts, i + 0 + (i&1)), + get_elt(elts, i + 1 - (i&1)), + get_elt(elts, i + 2 )); + } + } + break; + + case PIPE_PRIM_TRIANGLE_FAN: + if (count >= 3) { + if (flatfirst) { + for (i = 0; i+2 < count; i++) { + TRIANGLE( vcache, + DRAW_PIPE_RESET_STIPPLE | DRAW_PIPE_EDGE_FLAG_ALL, + get_elt(elts, i + 1), + get_elt(elts, i + 2), + get_elt(elts, 0 )); + } + } + else { + for (i = 0; i+2 < count; i++) { + TRIANGLE( vcache, + DRAW_PIPE_RESET_STIPPLE | DRAW_PIPE_EDGE_FLAG_ALL, + get_elt(elts, 0), + get_elt(elts, i + 1), + get_elt(elts, i + 2 )); + } + } + } + break; + + + case PIPE_PRIM_QUADS: + for (i = 0; i+3 < count; i += 4) { + QUAD( vcache, + get_elt(elts, i + 0), + get_elt(elts, i + 1), + get_elt(elts, i + 2), + get_elt(elts, i + 3)); + } + break; + + case PIPE_PRIM_QUAD_STRIP: + for (i = 0; i+3 < count; i += 2) { + QUAD( vcache, + get_elt(elts, i + 2), + get_elt(elts, i + 0), + get_elt(elts, i + 1), + get_elt(elts, i + 3)); + } + break; + + case PIPE_PRIM_POLYGON: + { + /* These bitflags look a little odd because we submit the + * vertices as (1,2,0) to satisfy flatshade requirements. + */ + const unsigned edge_first = DRAW_PIPE_EDGE_FLAG_2; + const unsigned edge_middle = DRAW_PIPE_EDGE_FLAG_0; + const unsigned edge_last = DRAW_PIPE_EDGE_FLAG_1; + + flags = DRAW_PIPE_RESET_STIPPLE | edge_first | edge_middle; + + for (i = 0; i+2 < count; i++, flags = edge_middle) { + + if (i + 3 == count) + flags |= edge_last; + + TRIANGLE( vcache, + flags, + get_elt(elts, i + 1), + get_elt(elts, i + 2), + get_elt(elts, 0)); + } + } + break; + + default: + assert(0); + break; + } + + vcache_flush( vcache ); +} + + +#undef TRIANGLE +#undef QUAD +#undef POINT +#undef LINE +#undef FUNC diff --git a/src/gallium/auxiliary/draw/draw_vertex.c b/src/gallium/auxiliary/draw/draw_vertex.c index 970adc95e7..1446f785c5 100644 --- a/src/gallium/auxiliary/draw/draw_vertex.c +++ b/src/gallium/auxiliary/draw/draw_vertex.c @@ -52,9 +52,6 @@ draw_compute_vertex_size(struct vertex_info *vinfo) switch (vinfo->emit[i]) { case EMIT_OMIT: break; - case EMIT_HEADER: - vinfo->size += sizeof(struct vertex_header) / 4; - break; case EMIT_4UB: /* fall-through */ case EMIT_1F_PSIZE: @@ -71,12 +68,62 @@ draw_compute_vertex_size(struct vertex_info *vinfo) case EMIT_4F: vinfo->size += 4; break; - case EMIT_ALL: - /* fall-through */ default: assert(0); } } +} + + +void +draw_dump_emitted_vertex(const struct vertex_info *vinfo, const uint8_t *data) +{ + unsigned i, j; - assert(vinfo->size * 4 <= MAX_VERTEX_SIZE); + for (i = 0; i < vinfo->num_attribs; i++) { + j = vinfo->src_index[i]; + switch (vinfo->emit[i]) { + case EMIT_OMIT: + debug_printf("EMIT_OMIT:"); + break; + case EMIT_1F: + debug_printf("EMIT_1F:\t"); + debug_printf("%f ", *(float *)data); data += sizeof(float); + break; + case EMIT_1F_PSIZE: + debug_printf("EMIT_1F_PSIZE:\t"); + debug_printf("%f ", *(float *)data); data += sizeof(float); + break; + case EMIT_2F: + debug_printf("EMIT_2F:\t"); + debug_printf("%f ", *(float *)data); data += sizeof(float); + debug_printf("%f ", *(float *)data); data += sizeof(float); + break; + case EMIT_3F: + debug_printf("EMIT_3F:\t"); + debug_printf("%f ", *(float *)data); data += sizeof(float); + debug_printf("%f ", *(float *)data); data += sizeof(float); + debug_printf("%f ", *(float *)data); data += sizeof(float); + data += sizeof(float); + break; + case EMIT_4F: + debug_printf("EMIT_4F:\t"); + debug_printf("%f ", *(float *)data); data += sizeof(float); + debug_printf("%f ", *(float *)data); data += sizeof(float); + debug_printf("%f ", *(float *)data); data += sizeof(float); + debug_printf("%f ", *(float *)data); data += sizeof(float); + break; + case EMIT_4UB: + debug_printf("EMIT_4UB:\t"); + debug_printf("%u ", *data++); + debug_printf("%u ", *data++); + debug_printf("%u ", *data++); + debug_printf("%u ", *data++); + break; + default: + assert(0); + } + debug_printf("\n"); + } + debug_printf("\n"); } diff --git a/src/gallium/auxiliary/draw/draw_vertex.h b/src/gallium/auxiliary/draw/draw_vertex.h index abd2017ed3..6d8bac5138 100644 --- a/src/gallium/auxiliary/draw/draw_vertex.h +++ b/src/gallium/auxiliary/draw/draw_vertex.h @@ -47,8 +47,6 @@ */ enum attrib_emit { EMIT_OMIT, /**< don't emit the attribute */ - EMIT_ALL, /**< emit whole post-xform vertex, w/ header */ - EMIT_HEADER, /**< emit vertex_header struct (XXX temp?) */ EMIT_1F, EMIT_1F_PSIZE, /**< insert constant point size */ EMIT_2F, @@ -108,5 +106,7 @@ draw_emit_vertex_attr(struct vertex_info *vinfo, extern void draw_compute_vertex_size(struct vertex_info *vinfo); +void draw_dump_emitted_vertex(const struct vertex_info *vinfo, + const uint8_t *data); #endif /* DRAW_VERTEX_H */ diff --git a/src/gallium/auxiliary/draw/draw_vertex_cache.c b/src/gallium/auxiliary/draw/draw_vertex_cache.c deleted file mode 100644 index c0248979e2..0000000000 --- a/src/gallium/auxiliary/draw/draw_vertex_cache.c +++ /dev/null @@ -1,208 +0,0 @@ -/************************************************************************** - * - * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - - /* - * Authors: - * Keith Whitwell <keith@tungstengraphics.com> - */ - -#include "pipe/p_util.h" -#include "draw_private.h" -#include "draw_context.h" - - -void draw_vertex_cache_invalidate( struct draw_context *draw ) -{ - assert(draw->pq.queue_nr == 0); - assert(draw->vs.queue_nr == 0); - assert(draw->vcache.referenced == 0); - - /* There's an error somewhere in the vcache code that requires this - * memset. The bug is exposed in q3demo demo001, but probably - * elsewhere as well. Will track it down later. - */ - memset(draw->vcache.idx, ~0, sizeof(draw->vcache.idx)); -} - - -/** - * Check if vertex is in cache, otherwise add it. It won't go through - * VS yet, not until there is a flush operation or the VS queue fills up. - * - * Note that cache entries are basically just two pointers: the first - * an index into the user's vertex arrays, the second a location in - * the vertex shader cache for the post-transformed vertex. - * - * \return pointer to location of (post-transformed) vertex header in the cache - */ -static struct vertex_header *get_vertex( struct draw_context *draw, - unsigned i ) -{ - unsigned slot = (i + (i>>5)) % VCACHE_SIZE; - - assert(slot < 32); /* so we don't exceed the bitfield size below */ - - if (draw->vcache.referenced & (1<<slot)) - { - /* Cache hit? - */ - if (draw->vcache.idx[slot].in == i) { -// _mesa_printf("HIT %d %d\n", slot, i); - assert(draw->vcache.idx[slot].out < draw->vs.queue_nr); - return draw->vs.queue[draw->vcache.idx[slot].out].vertex; - } - - /* Otherwise a collision - */ - slot = VCACHE_SIZE + draw->vcache.overflow++; -// _mesa_printf("XXX %d --> %d\n", i, slot); - } - - /* Deal with the cache miss: - */ - { - unsigned out; - - assert(slot < Elements(draw->vcache.idx)); - -// _mesa_printf("NEW %d %d\n", slot, i); - draw->vcache.idx[slot].in = i; - draw->vcache.idx[slot].out = out = draw->vs.queue_nr++; - draw->vcache.referenced |= (1 << slot); - - - /* Add to vertex shader queue: - */ - assert(draw->vs.queue_nr < VS_QUEUE_LENGTH); - - draw->vs.queue[out].elt = i; - draw->vs.queue[out].vertex->clipmask = 0; - draw->vs.queue[out].vertex->edgeflag = draw_get_edgeflag(draw, i); - draw->vs.queue[out].vertex->pad = 0; - draw->vs.queue[out].vertex->vertex_id = UNDEFINED_VERTEX_ID; - - /* Need to set the vertex's edge flag here. If we're being called - * by do_ef_triangle(), that function needs edge flag info! - */ - - return draw->vs.queue[draw->vcache.idx[slot].out].vertex; - } -} - - -static struct vertex_header *get_uint_elt_vertex( struct draw_context *draw, - unsigned i ) -{ - const unsigned *elts = (const unsigned *) draw->user.elts; - return get_vertex( draw, elts[i] ); -} - - -static struct vertex_header *get_ushort_elt_vertex( struct draw_context *draw, - unsigned i ) -{ - const ushort *elts = (const ushort *) draw->user.elts; - return get_vertex( draw, elts[i] ); -} - - -static struct vertex_header *get_ubyte_elt_vertex( struct draw_context *draw, - unsigned i ) -{ - const ubyte *elts = (const ubyte *) draw->user.elts; - return get_vertex( draw, elts[i] ); -} - - -void draw_vertex_cache_reset_vertex_ids( struct draw_context *draw ) -{ - unsigned i; - - for (i = 0; i < draw->vs.post_nr; i++) - draw->vs.queue[i].vertex->vertex_id = UNDEFINED_VERTEX_ID; -} - - -void draw_vertex_cache_unreference( struct draw_context *draw ) -{ - draw->vcache.referenced = 0; - draw->vcache.overflow = 0; -} - - -int draw_vertex_cache_check_space( struct draw_context *draw, - unsigned nr_verts ) -{ - if (draw->vcache.overflow + nr_verts < VCACHE_OVERFLOW) { - /* The vs queue is sized so that this can never happen: - */ - assert(draw->vs.queue_nr + nr_verts < VS_QUEUE_LENGTH); - return TRUE; - } - else - return FALSE; -} - - - -/** - * Tell the drawing context about the index/element buffer to use - * (ala glDrawElements) - * If no element buffer is to be used (i.e. glDrawArrays) then this - * should be called with eltSize=0 and elements=NULL. - * - * \param draw the drawing context - * \param eltSize size of each element (1, 2 or 4 bytes) - * \param elements the element buffer ptr - */ -void -draw_set_mapped_element_buffer( struct draw_context *draw, - unsigned eltSize, void *elements ) -{ -// draw_statechange( draw ); - - /* choose the get_vertex() function to use */ - switch (eltSize) { - case 0: - draw->vcache.get_vertex = get_vertex; - break; - case 1: - draw->vcache.get_vertex = get_ubyte_elt_vertex; - break; - case 2: - draw->vcache.get_vertex = get_ushort_elt_vertex; - break; - case 4: - draw->vcache.get_vertex = get_uint_elt_vertex; - break; - default: - assert(0); - } - draw->user.elts = elements; - draw->user.eltSize = eltSize; -} - diff --git a/src/gallium/auxiliary/draw/draw_vertex_fetch.c b/src/gallium/auxiliary/draw/draw_vertex_fetch.c deleted file mode 100644 index 9041041006..0000000000 --- a/src/gallium/auxiliary/draw/draw_vertex_fetch.c +++ /dev/null @@ -1,528 +0,0 @@ -/************************************************************************** - * - * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - - /* - * Authors: - * Keith Whitwell <keith@tungstengraphics.com> - */ - -#include "pipe/p_util.h" -#include "pipe/p_shader_tokens.h" -#include "draw_private.h" -#include "draw_context.h" - - -#define DRAW_DBG 0 - - -/** - * Fetch a float[4] vertex attribute from memory, doing format/type - * conversion as needed. - * - * This is probably needed/dupliocated elsewhere, eg format - * conversion, texture sampling etc. - */ -#define FETCH_ATTRIB( NAME, SZ, CVT ) \ -static void \ -fetch_##NAME(const void *ptr, float *attrib) \ -{ \ - static const float defaults[4] = { 0,0,0,1 }; \ - int i; \ - \ - for (i = 0; i < SZ; i++) { \ - attrib[i] = CVT(i); \ - } \ - \ - for (; i < 4; i++) { \ - attrib[i] = defaults[i]; \ - } \ -} - -#define CVT_64_FLOAT(i) (float) ((double *) ptr)[i] -#define CVT_32_FLOAT(i) ((float *) ptr)[i] - -#define CVT_8_USCALED(i) (float) ((unsigned char *) ptr)[i] -#define CVT_16_USCALED(i) (float) ((unsigned short *) ptr)[i] -#define CVT_32_USCALED(i) (float) ((unsigned int *) ptr)[i] - -#define CVT_8_SSCALED(i) (float) ((char *) ptr)[i] -#define CVT_16_SSCALED(i) (float) ((short *) ptr)[i] -#define CVT_32_SSCALED(i) (float) ((int *) ptr)[i] - -#define CVT_8_UNORM(i) (float) ((unsigned char *) ptr)[i] / 255.0f -#define CVT_16_UNORM(i) (float) ((unsigned short *) ptr)[i] / 65535.0f -#define CVT_32_UNORM(i) (float) ((unsigned int *) ptr)[i] / 4294967295.0f - -#define CVT_8_SNORM(i) (float) ((char *) ptr)[i] / 127.0f -#define CVT_16_SNORM(i) (float) ((short *) ptr)[i] / 32767.0f -#define CVT_32_SNORM(i) (float) ((int *) ptr)[i] / 2147483647.0f - -FETCH_ATTRIB( R64G64B64A64_FLOAT, 4, CVT_64_FLOAT ) -FETCH_ATTRIB( R64G64B64_FLOAT, 3, CVT_64_FLOAT ) -FETCH_ATTRIB( R64G64_FLOAT, 2, CVT_64_FLOAT ) -FETCH_ATTRIB( R64_FLOAT, 1, CVT_64_FLOAT ) - -FETCH_ATTRIB( R32G32B32A32_FLOAT, 4, CVT_32_FLOAT ) -FETCH_ATTRIB( R32G32B32_FLOAT, 3, CVT_32_FLOAT ) -FETCH_ATTRIB( R32G32_FLOAT, 2, CVT_32_FLOAT ) -FETCH_ATTRIB( R32_FLOAT, 1, CVT_32_FLOAT ) - -FETCH_ATTRIB( R32G32B32A32_USCALED, 4, CVT_32_USCALED ) -FETCH_ATTRIB( R32G32B32_USCALED, 3, CVT_32_USCALED ) -FETCH_ATTRIB( R32G32_USCALED, 2, CVT_32_USCALED ) -FETCH_ATTRIB( R32_USCALED, 1, CVT_32_USCALED ) - -FETCH_ATTRIB( R32G32B32A32_SSCALED, 4, CVT_32_SSCALED ) -FETCH_ATTRIB( R32G32B32_SSCALED, 3, CVT_32_SSCALED ) -FETCH_ATTRIB( R32G32_SSCALED, 2, CVT_32_SSCALED ) -FETCH_ATTRIB( R32_SSCALED, 1, CVT_32_SSCALED ) - -FETCH_ATTRIB( R32G32B32A32_UNORM, 4, CVT_32_UNORM ) -FETCH_ATTRIB( R32G32B32_UNORM, 3, CVT_32_UNORM ) -FETCH_ATTRIB( R32G32_UNORM, 2, CVT_32_UNORM ) -FETCH_ATTRIB( R32_UNORM, 1, CVT_32_UNORM ) - -FETCH_ATTRIB( R32G32B32A32_SNORM, 4, CVT_32_SNORM ) -FETCH_ATTRIB( R32G32B32_SNORM, 3, CVT_32_SNORM ) -FETCH_ATTRIB( R32G32_SNORM, 2, CVT_32_SNORM ) -FETCH_ATTRIB( R32_SNORM, 1, CVT_32_SNORM ) - -FETCH_ATTRIB( R16G16B16A16_USCALED, 4, CVT_16_USCALED ) -FETCH_ATTRIB( R16G16B16_USCALED, 3, CVT_16_USCALED ) -FETCH_ATTRIB( R16G16_USCALED, 2, CVT_16_USCALED ) -FETCH_ATTRIB( R16_USCALED, 1, CVT_16_USCALED ) - -FETCH_ATTRIB( R16G16B16A16_SSCALED, 4, CVT_16_SSCALED ) -FETCH_ATTRIB( R16G16B16_SSCALED, 3, CVT_16_SSCALED ) -FETCH_ATTRIB( R16G16_SSCALED, 2, CVT_16_SSCALED ) -FETCH_ATTRIB( R16_SSCALED, 1, CVT_16_SSCALED ) - -FETCH_ATTRIB( R16G16B16A16_UNORM, 4, CVT_16_UNORM ) -FETCH_ATTRIB( R16G16B16_UNORM, 3, CVT_16_UNORM ) -FETCH_ATTRIB( R16G16_UNORM, 2, CVT_16_UNORM ) -FETCH_ATTRIB( R16_UNORM, 1, CVT_16_UNORM ) - -FETCH_ATTRIB( R16G16B16A16_SNORM, 4, CVT_16_SNORM ) -FETCH_ATTRIB( R16G16B16_SNORM, 3, CVT_16_SNORM ) -FETCH_ATTRIB( R16G16_SNORM, 2, CVT_16_SNORM ) -FETCH_ATTRIB( R16_SNORM, 1, CVT_16_SNORM ) - -FETCH_ATTRIB( R8G8B8A8_USCALED, 4, CVT_8_USCALED ) -FETCH_ATTRIB( R8G8B8_USCALED, 3, CVT_8_USCALED ) -FETCH_ATTRIB( R8G8_USCALED, 2, CVT_8_USCALED ) -FETCH_ATTRIB( R8_USCALED, 1, CVT_8_USCALED ) - -FETCH_ATTRIB( R8G8B8A8_SSCALED, 4, CVT_8_SSCALED ) -FETCH_ATTRIB( R8G8B8_SSCALED, 3, CVT_8_SSCALED ) -FETCH_ATTRIB( R8G8_SSCALED, 2, CVT_8_SSCALED ) -FETCH_ATTRIB( R8_SSCALED, 1, CVT_8_SSCALED ) - -FETCH_ATTRIB( R8G8B8A8_UNORM, 4, CVT_8_UNORM ) -FETCH_ATTRIB( R8G8B8_UNORM, 3, CVT_8_UNORM ) -FETCH_ATTRIB( R8G8_UNORM, 2, CVT_8_UNORM ) -FETCH_ATTRIB( R8_UNORM, 1, CVT_8_UNORM ) - -FETCH_ATTRIB( R8G8B8A8_SNORM, 4, CVT_8_SNORM ) -FETCH_ATTRIB( R8G8B8_SNORM, 3, CVT_8_SNORM ) -FETCH_ATTRIB( R8G8_SNORM, 2, CVT_8_SNORM ) -FETCH_ATTRIB( R8_SNORM, 1, CVT_8_SNORM ) - -FETCH_ATTRIB( A8R8G8B8_UNORM, 4, CVT_8_UNORM ) -//FETCH_ATTRIB( R8G8B8A8_UNORM, 4, CVT_8_UNORM ) - - - -static void -fetch_B8G8R8A8_UNORM(const void *ptr, float *attrib) -{ - attrib[2] = CVT_8_UNORM(0); - attrib[1] = CVT_8_UNORM(1); - attrib[0] = CVT_8_UNORM(2); - attrib[3] = CVT_8_UNORM(3); -} - - -fetch_func draw_get_fetch_func( enum pipe_format format ) -{ -#if 0 - { - char tmp[80]; - pf_sprint_name(tmp, format); - debug_printf("%s: %s\n", __FUNCTION__, tmp); - } -#endif - - switch (format) { - case PIPE_FORMAT_R64_FLOAT: - return fetch_R64_FLOAT; - case PIPE_FORMAT_R64G64_FLOAT: - return fetch_R64G64_FLOAT; - case PIPE_FORMAT_R64G64B64_FLOAT: - return fetch_R64G64B64_FLOAT; - case PIPE_FORMAT_R64G64B64A64_FLOAT: - return fetch_R64G64B64A64_FLOAT; - - case PIPE_FORMAT_R32_FLOAT: - return fetch_R32_FLOAT; - case PIPE_FORMAT_R32G32_FLOAT: - return fetch_R32G32_FLOAT; - case PIPE_FORMAT_R32G32B32_FLOAT: - return fetch_R32G32B32_FLOAT; - case PIPE_FORMAT_R32G32B32A32_FLOAT: - return fetch_R32G32B32A32_FLOAT; - - case PIPE_FORMAT_R32_UNORM: - return fetch_R32_UNORM; - case PIPE_FORMAT_R32G32_UNORM: - return fetch_R32G32_UNORM; - case PIPE_FORMAT_R32G32B32_UNORM: - return fetch_R32G32B32_UNORM; - case PIPE_FORMAT_R32G32B32A32_UNORM: - return fetch_R32G32B32A32_UNORM; - - case PIPE_FORMAT_R32_USCALED: - return fetch_R32_USCALED; - case PIPE_FORMAT_R32G32_USCALED: - return fetch_R32G32_USCALED; - case PIPE_FORMAT_R32G32B32_USCALED: - return fetch_R32G32B32_USCALED; - case PIPE_FORMAT_R32G32B32A32_USCALED: - return fetch_R32G32B32A32_USCALED; - - case PIPE_FORMAT_R32_SNORM: - return fetch_R32_SNORM; - case PIPE_FORMAT_R32G32_SNORM: - return fetch_R32G32_SNORM; - case PIPE_FORMAT_R32G32B32_SNORM: - return fetch_R32G32B32_SNORM; - case PIPE_FORMAT_R32G32B32A32_SNORM: - return fetch_R32G32B32A32_SNORM; - - case PIPE_FORMAT_R32_SSCALED: - return fetch_R32_SSCALED; - case PIPE_FORMAT_R32G32_SSCALED: - return fetch_R32G32_SSCALED; - case PIPE_FORMAT_R32G32B32_SSCALED: - return fetch_R32G32B32_SSCALED; - case PIPE_FORMAT_R32G32B32A32_SSCALED: - return fetch_R32G32B32A32_SSCALED; - - case PIPE_FORMAT_R16_UNORM: - return fetch_R16_UNORM; - case PIPE_FORMAT_R16G16_UNORM: - return fetch_R16G16_UNORM; - case PIPE_FORMAT_R16G16B16_UNORM: - return fetch_R16G16B16_UNORM; - case PIPE_FORMAT_R16G16B16A16_UNORM: - return fetch_R16G16B16A16_UNORM; - - case PIPE_FORMAT_R16_USCALED: - return fetch_R16_USCALED; - case PIPE_FORMAT_R16G16_USCALED: - return fetch_R16G16_USCALED; - case PIPE_FORMAT_R16G16B16_USCALED: - return fetch_R16G16B16_USCALED; - case PIPE_FORMAT_R16G16B16A16_USCALED: - return fetch_R16G16B16A16_USCALED; - - case PIPE_FORMAT_R16_SNORM: - return fetch_R16_SNORM; - case PIPE_FORMAT_R16G16_SNORM: - return fetch_R16G16_SNORM; - case PIPE_FORMAT_R16G16B16_SNORM: - return fetch_R16G16B16_SNORM; - case PIPE_FORMAT_R16G16B16A16_SNORM: - return fetch_R16G16B16A16_SNORM; - - case PIPE_FORMAT_R16_SSCALED: - return fetch_R16_SSCALED; - case PIPE_FORMAT_R16G16_SSCALED: - return fetch_R16G16_SSCALED; - case PIPE_FORMAT_R16G16B16_SSCALED: - return fetch_R16G16B16_SSCALED; - case PIPE_FORMAT_R16G16B16A16_SSCALED: - return fetch_R16G16B16A16_SSCALED; - - case PIPE_FORMAT_R8_UNORM: - return fetch_R8_UNORM; - case PIPE_FORMAT_R8G8_UNORM: - return fetch_R8G8_UNORM; - case PIPE_FORMAT_R8G8B8_UNORM: - return fetch_R8G8B8_UNORM; - case PIPE_FORMAT_R8G8B8A8_UNORM: - return fetch_R8G8B8A8_UNORM; - - case PIPE_FORMAT_R8_USCALED: - return fetch_R8_USCALED; - case PIPE_FORMAT_R8G8_USCALED: - return fetch_R8G8_USCALED; - case PIPE_FORMAT_R8G8B8_USCALED: - return fetch_R8G8B8_USCALED; - case PIPE_FORMAT_R8G8B8A8_USCALED: - return fetch_R8G8B8A8_USCALED; - - case PIPE_FORMAT_R8_SNORM: - return fetch_R8_SNORM; - case PIPE_FORMAT_R8G8_SNORM: - return fetch_R8G8_SNORM; - case PIPE_FORMAT_R8G8B8_SNORM: - return fetch_R8G8B8_SNORM; - case PIPE_FORMAT_R8G8B8A8_SNORM: - return fetch_R8G8B8A8_SNORM; - - case PIPE_FORMAT_R8_SSCALED: - return fetch_R8_SSCALED; - case PIPE_FORMAT_R8G8_SSCALED: - return fetch_R8G8_SSCALED; - case PIPE_FORMAT_R8G8B8_SSCALED: - return fetch_R8G8B8_SSCALED; - case PIPE_FORMAT_R8G8B8A8_SSCALED: - return fetch_R8G8B8A8_SSCALED; - - case PIPE_FORMAT_A8R8G8B8_UNORM: - return fetch_A8R8G8B8_UNORM; - - - case PIPE_FORMAT_B8G8R8A8_UNORM: - return fetch_B8G8R8A8_UNORM; - - case 0: - return NULL; /* not sure why this is needed */ - - default: - /* This can get hit because draw-state-validation is too eager, - and can jump in here validating stuff before the state tracker has set - up everything. - */ - /* assert(0); */ - return NULL; - } -} - - -static void -transpose_4x4( float *out, const float *in ) -{ - /* This can be achieved in 12 sse instructions, plus the final - * stores I guess. This is probably a bit more than that - maybe - * 32 or so? - */ - out[0] = in[0]; out[1] = in[4]; out[2] = in[8]; out[3] = in[12]; - out[4] = in[1]; out[5] = in[5]; out[6] = in[9]; out[7] = in[13]; - out[8] = in[2]; out[9] = in[6]; out[10] = in[10]; out[11] = in[14]; - out[12] = in[3]; out[13] = in[7]; out[14] = in[11]; out[15] = in[15]; -} - - - -static void fetch_xyz_rgb( struct draw_context *draw, - struct tgsi_exec_machine *machine, - const unsigned *elts, - unsigned count ) -{ - const unsigned *pitch = draw->vertex_fetch.pitch; - const ubyte **src = draw->vertex_fetch.src_ptr; - int i; - - assert(count <= 4); - -// debug_printf("%s\n", __FUNCTION__); - - /* loop over vertex attributes (vertex shader inputs) - */ - - for (i = 0; i < 4; i++) { - { - const float *in = (const float *)(src[0] + elts[i] * pitch[0]); - float *out = &machine->Inputs[0].xyzw[0].f[i]; - out[0] = in[0]; - out[4] = in[1]; - out[8] = in[2]; - out[12] = 1.0f; - } - - { - const float *in = (const float *)(src[1] + elts[i] * pitch[1]); - float *out = &machine->Inputs[1].xyzw[0].f[i]; - out[0] = in[0]; - out[4] = in[1]; - out[8] = in[2]; - out[12] = 1.0f; - } - } -} - - - - -static void fetch_xyz_rgb_st( struct draw_context *draw, - struct tgsi_exec_machine *machine, - const unsigned *elts, - unsigned count ) -{ - const unsigned *pitch = draw->vertex_fetch.pitch; - const ubyte **src = draw->vertex_fetch.src_ptr; - int i; - - assert(count <= 4); - - /* loop over vertex attributes (vertex shader inputs) - */ - - for (i = 0; i < 4; i++) { - { - const float *in = (const float *)(src[0] + elts[i] * pitch[0]); - float *out = &machine->Inputs[0].xyzw[0].f[i]; - out[0] = in[0]; - out[4] = in[1]; - out[8] = in[2]; - out[12] = 1.0f; - } - - { - const float *in = (const float *)(src[1] + elts[i] * pitch[1]); - float *out = &machine->Inputs[1].xyzw[0].f[i]; - out[0] = in[0]; - out[4] = in[1]; - out[8] = in[2]; - out[12] = 1.0f; - } - - { - const float *in = (const float *)(src[2] + elts[i] * pitch[2]); - float *out = &machine->Inputs[2].xyzw[0].f[i]; - out[0] = in[0]; - out[4] = in[1]; - out[8] = 0.0f; - out[12] = 1.0f; - } - } -} - - - - -/** - * Fetch vertex attributes for 'count' vertices. - */ -static void generic_vertex_fetch( struct draw_context *draw, - struct tgsi_exec_machine *machine, - const unsigned *elts, - unsigned count ) -{ - unsigned nr_attrs = draw->vertex_fetch.nr_attrs; - unsigned attr; - - assert(count <= 4); - -// debug_printf("%s %d\n", __FUNCTION__, count); - - /* loop over vertex attributes (vertex shader inputs) - */ - for (attr = 0; attr < nr_attrs; attr++) { - - const unsigned pitch = draw->vertex_fetch.pitch[attr]; - const ubyte *src = draw->vertex_fetch.src_ptr[attr]; - const fetch_func fetch = draw->vertex_fetch.fetch[attr]; - unsigned i; - float p[4][4]; - - - /* Fetch four attributes for four vertices. - * - * Could fetch directly into AOS format, but this is meant to be - * a prototype for an sse implementation, which would have - * difficulties doing that. - */ - for (i = 0; i < count; i++) - fetch( src + elts[i] * pitch, p[i] ); - - /* Be nice and zero out any missing vertices: - */ - for ( ; i < 4; i++) - p[i][0] = p[i][1] = p[i][2] = p[i][3] = 0; - - /* Transpose/swizzle into sse-friendly format. Currently - * assuming that all vertex shader inputs are float[4], but this - * isn't true -- if the vertex shader only wants tex0.xy, we - * could optimize for that. - * - * To do so fully without codegen would probably require an - * excessive number of fetch functions, but we could at least - * minimize the transpose step: - */ - transpose_4x4( (float *)&machine->Inputs[attr].xyzw[0].f[0], (float *)p ); - } -} - - - -void draw_update_vertex_fetch( struct draw_context *draw ) -{ - unsigned nr_attrs, i; - -// debug_printf("%s\n", __FUNCTION__); - - /* this may happend during context init */ - if (!draw->vertex_shader) - return; - - nr_attrs = draw->vertex_shader->info.num_inputs; - - for (i = 0; i < nr_attrs; i++) { - unsigned buf = draw->vertex_element[i].vertex_buffer_index; - enum pipe_format format = draw->vertex_element[i].src_format; - - draw->vertex_fetch.src_ptr[i] = (const ubyte *) draw->user.vbuffer[buf] + - draw->vertex_buffer[buf].buffer_offset + - draw->vertex_element[i].src_offset; - - draw->vertex_fetch.pitch[i] = draw->vertex_buffer[buf].pitch; - draw->vertex_fetch.fetch[i] = draw_get_fetch_func( format ); - } - - draw->vertex_fetch.nr_attrs = nr_attrs; - - draw->vertex_fetch.fetch_func = generic_vertex_fetch; - - switch (nr_attrs) { - case 2: - if (draw->vertex_element[0].src_format == PIPE_FORMAT_R32G32B32_FLOAT && - draw->vertex_element[1].src_format == PIPE_FORMAT_R32G32B32_FLOAT) - draw->vertex_fetch.fetch_func = fetch_xyz_rgb; - break; - case 3: - if (draw->vertex_element[0].src_format == PIPE_FORMAT_R32G32B32_FLOAT && - draw->vertex_element[1].src_format == PIPE_FORMAT_R32G32B32_FLOAT && - draw->vertex_element[2].src_format == PIPE_FORMAT_R32G32_FLOAT) - draw->vertex_fetch.fetch_func = fetch_xyz_rgb_st; - break; - default: - break; - } - -} diff --git a/src/gallium/auxiliary/draw/draw_vf.c b/src/gallium/auxiliary/draw/draw_vf.c deleted file mode 100644 index 7bb34ace7a..0000000000 --- a/src/gallium/auxiliary/draw/draw_vf.c +++ /dev/null @@ -1,432 +0,0 @@ -/* - * Copyright 2003 Tungsten Graphics, inc. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * on the rights to use, copy, modify, merge, publish, distribute, sub - * license, and/or sell copies of the Software, and to permit persons to whom - * the Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice (including the next - * paragraph) shall be included in all copies or substantial portions of the - * Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL - * TUNGSTEN GRAPHICS AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM, - * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR - * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE - * USE OR OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: - * Keith Whitwell <keithw@tungstengraphics.com> - */ - - -#include <stddef.h> - -#include "pipe/p_compiler.h" -#include "pipe/p_util.h" -#include "rtasm/rtasm_execmem.h" - -#include "draw_vf.h" - - -#define DRAW_VF_DBG 0 - - -static boolean match_fastpath( struct draw_vertex_fetch *vf, - const struct draw_vf_fastpath *fp) -{ - unsigned j; - - if (vf->attr_count != fp->attr_count) - return FALSE; - - for (j = 0; j < vf->attr_count; j++) - if (vf->attr[j].format != fp->attr[j].format || - vf->attr[j].inputsize != fp->attr[j].size || - vf->attr[j].vertoffset != fp->attr[j].offset) - return FALSE; - - if (fp->match_strides) { - if (vf->vertex_stride != fp->vertex_stride) - return FALSE; - - for (j = 0; j < vf->attr_count; j++) - if (vf->attr[j].inputstride != fp->attr[j].stride) - return FALSE; - } - - return TRUE; -} - -static boolean search_fastpath_emit( struct draw_vertex_fetch *vf ) -{ - struct draw_vf_fastpath *fp = vf->fastpath; - - for ( ; fp ; fp = fp->next) { - if (match_fastpath(vf, fp)) { - vf->emit = fp->func; - return TRUE; - } - } - - return FALSE; -} - -void draw_vf_register_fastpath( struct draw_vertex_fetch *vf, - boolean match_strides ) -{ - struct draw_vf_fastpath *fastpath = CALLOC_STRUCT(draw_vf_fastpath); - unsigned i; - - fastpath->vertex_stride = vf->vertex_stride; - fastpath->attr_count = vf->attr_count; - fastpath->match_strides = match_strides; - fastpath->func = vf->emit; - fastpath->attr = (struct draw_vf_attr_type *) - MALLOC(vf->attr_count * sizeof(fastpath->attr[0])); - - for (i = 0; i < vf->attr_count; i++) { - fastpath->attr[i].format = vf->attr[i].format; - fastpath->attr[i].stride = vf->attr[i].inputstride; - fastpath->attr[i].size = vf->attr[i].inputsize; - fastpath->attr[i].offset = vf->attr[i].vertoffset; - } - - fastpath->next = vf->fastpath; - vf->fastpath = fastpath; -} - - - - -/*********************************************************************** - * Build codegen functions or return generic ones: - */ -static void choose_emit_func( struct draw_vertex_fetch *vf, - unsigned count, - uint8_t *dest) -{ - vf->emit = NULL; - - /* Does this match an existing (hardwired, codegen or known-bad) - * fastpath? - */ - if (search_fastpath_emit(vf)) { - /* Use this result. If it is null, then it is already known - * that the current state will fail for codegen and there is no - * point trying again. - */ - } - else if (vf->codegen_emit) { - vf->codegen_emit( vf ); - } - - if (!vf->emit) { - draw_vf_generate_hardwired_emit(vf); - } - - /* Otherwise use the generic version: - */ - if (!vf->emit) - vf->emit = draw_vf_generic_emit; - - vf->emit( vf, count, dest ); -} - - - - - -/*********************************************************************** - * Public entrypoints, mostly dispatch to the above: - */ - - - -static unsigned -draw_vf_set_vertex_attributes( struct draw_vertex_fetch *vf, - const struct draw_vf_attr_map *map, - unsigned nr, - unsigned vertex_stride ) -{ - unsigned offset = 0; - unsigned i, j; - - assert(nr < PIPE_MAX_ATTRIBS); - - for (j = 0, i = 0; i < nr; i++) { - const unsigned format = map[i].format; - if (format == DRAW_EMIT_PAD) { -#if (DRAW_VF_DBG) - debug_printf("%d: pad %d, offset %d\n", i, - map[i].offset, offset); -#endif - - offset += map[i].offset; - - } - else { - vf->attr[j].attrib = map[i].attrib; - vf->attr[j].format = format; - vf->attr[j].insert = draw_vf_format_info[format].insert; - vf->attr[j].vertattrsize = draw_vf_format_info[format].attrsize; - vf->attr[j].vertoffset = offset; - vf->attr[j].isconst = draw_vf_format_info[format].isconst; - if(vf->attr[j].isconst) - memcpy(vf->attr[j].data, &map[i].data, vf->attr[j].vertattrsize); - -#if (DRAW_VF_DBG) - debug_printf("%d: %s, offset %d\n", i, - draw_vf_format_info[format].name, - vf->attr[j].vertoffset); -#endif - - offset += draw_vf_format_info[format].attrsize; - j++; - } - } - - vf->attr_count = j; - vf->vertex_stride = vertex_stride ? vertex_stride : offset; - vf->emit = choose_emit_func; - - assert(vf->vertex_stride >= offset); - return vf->vertex_stride; -} - - -void draw_vf_set_vertex_info( struct draw_vertex_fetch *vf, - const struct vertex_info *vinfo, - float point_size ) -{ - unsigned i, j, k; - struct draw_vf_attr *a = vf->attr; - struct draw_vf_attr_map attrs[PIPE_MAX_SHADER_INPUTS]; - unsigned count = 0; /* for debug/sanity */ - unsigned nr_attrs = 0; - - for (i = 0; i < vinfo->num_attribs; i++) { - j = vinfo->src_index[i]; - switch (vinfo->emit[i]) { - case EMIT_OMIT: - /* no-op */ - break; - case EMIT_ALL: { - /* just copy the whole vertex as-is to the vbuf */ - unsigned s = vinfo->size; - assert(i == 0); - assert(j == 0); - /* copy the vertex header */ - /* XXX: we actually don't copy the header, just pad it */ - attrs[nr_attrs].attrib = 0; - attrs[nr_attrs].format = DRAW_EMIT_PAD; - attrs[nr_attrs].offset = offsetof(struct vertex_header, data); - s -= offsetof(struct vertex_header, data)/4; - count += offsetof(struct vertex_header, data)/4; - nr_attrs++; - /* copy the vertex data */ - for(k = 0; k < (s & ~0x3); k += 4) { - attrs[nr_attrs].attrib = k/4; - attrs[nr_attrs].format = DRAW_EMIT_4F; - attrs[nr_attrs].offset = 0; - nr_attrs++; - count += 4; - } - /* tail */ - /* XXX: actually, this shouldn't be needed */ - attrs[nr_attrs].attrib = k/4; - attrs[nr_attrs].offset = 0; - switch(s & 0x3) { - case 0: - break; - case 1: - attrs[nr_attrs].format = DRAW_EMIT_1F; - nr_attrs++; - count += 1; - break; - case 2: - attrs[nr_attrs].format = DRAW_EMIT_2F; - nr_attrs++; - count += 2; - break; - case 3: - attrs[nr_attrs].format = DRAW_EMIT_3F; - nr_attrs++; - count += 3; - break; - } - break; - } - case EMIT_HEADER: - /* XXX emit new DRAW_EMIT_HEADER attribute??? */ - attrs[nr_attrs].attrib = 0; - attrs[nr_attrs].format = DRAW_EMIT_PAD; - attrs[nr_attrs].offset = offsetof(struct vertex_header, data); - count += offsetof(struct vertex_header, data)/4; - nr_attrs++; - break; - case EMIT_1F: - attrs[nr_attrs].attrib = j; - attrs[nr_attrs].format = DRAW_EMIT_1F; - attrs[nr_attrs].offset = 0; - nr_attrs++; - count++; - break; - case EMIT_1F_PSIZE: - attrs[nr_attrs].attrib = j; - attrs[nr_attrs].format = DRAW_EMIT_1F_CONST; - attrs[nr_attrs].offset = 0; - attrs[nr_attrs].data.f[0] = point_size; - nr_attrs++; - count++; - break; - case EMIT_2F: - attrs[nr_attrs].attrib = j; - attrs[nr_attrs].format = DRAW_EMIT_2F; - attrs[nr_attrs].offset = 0; - nr_attrs++; - count += 2; - break; - case EMIT_3F: - attrs[nr_attrs].attrib = j; - attrs[nr_attrs].format = DRAW_EMIT_3F; - attrs[nr_attrs].offset = 0; - nr_attrs++; - count += 3; - break; - case EMIT_4F: - attrs[nr_attrs].attrib = j; - attrs[nr_attrs].format = DRAW_EMIT_4F; - attrs[nr_attrs].offset = 0; - nr_attrs++; - count += 4; - break; - case EMIT_4UB: - attrs[nr_attrs].attrib = j; - attrs[nr_attrs].format = DRAW_EMIT_4UB_4F_BGRA; - attrs[nr_attrs].offset = 0; - nr_attrs++; - count += 1; - break; - default: - assert(0); - } - } - - assert(count == vinfo->size); - - draw_vf_set_vertex_attributes(vf, - attrs, - nr_attrs, - vinfo->size * sizeof(float) ); - - for (j = 0; j < vf->attr_count; j++) { - a[j].inputsize = 4; - a[j].do_insert = a[j].insert[4 - 1]; - if(a[j].isconst) { - a[j].inputptr = a[j].data; - a[j].inputstride = 0; - } - } -} - - -#if 0 -/* Set attribute pointers, adjusted for start position: - */ -void draw_vf_set_sources( struct draw_vertex_fetch *vf, - GLvector4f * const sources[], - unsigned start ) -{ - struct draw_vf_attr *a = vf->attr; - unsigned j; - - for (j = 0; j < vf->attr_count; j++) { - const GLvector4f *vptr = sources[a[j].attrib]; - - if ((a[j].inputstride != vptr->stride) || - (a[j].inputsize != vptr->size)) - vf->emit = choose_emit_func; - - a[j].inputstride = vptr->stride; - a[j].inputsize = vptr->size; - a[j].do_insert = a[j].insert[vptr->size - 1]; - a[j].inputptr = ((uint8_t *)vptr->data) + start * vptr->stride; - } -} -#endif - - -/** - * Emit a vertex to dest. - */ -void draw_vf_emit_vertex( struct draw_vertex_fetch *vf, - struct vertex_header *vertex, - void *dest ) -{ - struct draw_vf_attr *a = vf->attr; - unsigned j; - - for (j = 0; j < vf->attr_count; j++) { - if (!a[j].isconst) { - a[j].inputptr = (uint8_t *)&vertex->data[a[j].attrib][0]; - a[j].inputstride = 0; /* XXX: one-vertex-max ATM */ - } - } - - vf->emit( vf, 1, (uint8_t*) dest ); -} - - - -struct draw_vertex_fetch *draw_vf_create( void ) -{ - struct draw_vertex_fetch *vf = CALLOC_STRUCT(draw_vertex_fetch); - unsigned i; - - for (i = 0; i < PIPE_MAX_ATTRIBS; i++) - vf->attr[i].vf = vf; - - vf->identity[0] = 0.0; - vf->identity[1] = 0.0; - vf->identity[2] = 0.0; - vf->identity[3] = 1.0; - - vf->codegen_emit = NULL; - -#ifdef USE_SSE_ASM - if (!GETENV("GALLIUM_NO_CODEGEN")) - vf->codegen_emit = draw_vf_generate_sse_emit; -#endif - - return vf; -} - - -void draw_vf_destroy( struct draw_vertex_fetch *vf ) -{ - struct draw_vf_fastpath *fp, *tmp; - - for (fp = vf->fastpath ; fp ; fp = tmp) { - tmp = fp->next; - FREE(fp->attr); - - /* KW: At the moment, fp->func is constrained to be allocated by - * rtasm_exec_alloc(), as the hardwired fastpaths in - * t_vertex_generic.c are handled specially. It would be nice - * to unify them, but this probably won't change until this - * module gets another overhaul. - */ - //rtasm_exec_free((void *) fp->func); - FREE(fp); - } - - vf->fastpath = NULL; - FREE(vf); -} diff --git a/src/gallium/auxiliary/draw/draw_vf.h b/src/gallium/auxiliary/draw/draw_vf.h deleted file mode 100644 index 7555d1bd58..0000000000 --- a/src/gallium/auxiliary/draw/draw_vf.h +++ /dev/null @@ -1,236 +0,0 @@ -/* - * Copyright 2008 Tungsten Graphics, inc. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * on the rights to use, copy, modify, merge, publish, distribute, sub - * license, and/or sell copies of the Software, and to permit persons to whom - * the Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice (including the next - * paragraph) shall be included in all copies or substantial portions of the - * Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL - * TUNGSTEN GRAPHICS AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM, - * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR - * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE - * USE OR OTHER DEALINGS IN THE SOFTWARE. - */ - - -/** - * Vertex fetch/store/convert code. This functionality is used in two places: - * 1. Vertex fetch/convert - to grab vertex data from incoming vertex - * arrays and convert to format needed by vertex shaders. - * 2. Vertex store/emit - to convert simple float[][4] vertex attributes - * (which is the organization used throughout the draw/prim pipeline) to - * hardware-specific formats and emit into hardware vertex buffers. - * - * - * Authors: - * Keith Whitwell <keithw@tungstengraphics.com> - */ - -#ifndef DRAW_VF_H -#define DRAW_VF_H - - -#include "pipe/p_compiler.h" -#include "pipe/p_state.h" - -#include "draw_vertex.h" -#include "draw_private.h" /* for vertex_header */ - - -enum draw_vf_attr_format { - DRAW_EMIT_1F, - DRAW_EMIT_2F, - DRAW_EMIT_3F, - DRAW_EMIT_4F, - DRAW_EMIT_3F_XYW, /**< for projective texture */ - DRAW_EMIT_1UB_1F, /**< for fog coordinate */ - DRAW_EMIT_3UB_3F_RGB, /**< for specular color */ - DRAW_EMIT_3UB_3F_BGR, /**< for specular color */ - DRAW_EMIT_4UB_4F_RGBA, /**< for color */ - DRAW_EMIT_4UB_4F_BGRA, /**< for color */ - DRAW_EMIT_4UB_4F_ARGB, /**< for color */ - DRAW_EMIT_4UB_4F_ABGR, /**< for color */ - DRAW_EMIT_1F_CONST, - DRAW_EMIT_2F_CONST, - DRAW_EMIT_3F_CONST, - DRAW_EMIT_4F_CONST, - DRAW_EMIT_PAD, /**< leave a hole of 'offset' bytes */ - DRAW_EMIT_MAX -}; - -struct draw_vf_attr_map -{ - /** Input attribute number */ - unsigned attrib; - - enum draw_vf_attr_format format; - - unsigned offset; - - /** - * Constant data for DRAW_EMIT_*_CONST - */ - union { - uint8_t ub[4]; - float f[4]; - } data; -}; - -struct draw_vertex_fetch; - - - -#if 0 -unsigned -draw_vf_set_vertex_attributes( struct draw_vertex_fetch *vf, - const struct draw_vf_attr_map *map, - unsigned nr, - unsigned vertex_stride ); -#endif - -void draw_vf_set_vertex_info( struct draw_vertex_fetch *vf, - const struct vertex_info *vinfo, - float point_size ); - -#if 0 -void -draw_vf_set_sources( struct draw_vertex_fetch *vf, - GLvector4f * const attrib[], - unsigned start ); -#endif - -void -draw_vf_emit_vertex( struct draw_vertex_fetch *vf, - struct vertex_header *vertex, - void *dest ); - -struct draw_vertex_fetch * -draw_vf_create( void ); - -void -draw_vf_destroy( struct draw_vertex_fetch *vf ); - - - -/*********************************************************************** - * Internal functions and structs: - */ - -struct draw_vf_attr; - -typedef void (*draw_vf_extract_func)( const struct draw_vf_attr *a, - float *out, - const uint8_t *v ); - -typedef void (*draw_vf_insert_func)( const struct draw_vf_attr *a, - uint8_t *v, - const float *in ); - -typedef void (*draw_vf_emit_func)( struct draw_vertex_fetch *vf, - unsigned count, - uint8_t *dest ); - - - -/** - * Describes how to convert/move a vertex attribute from a vertex - * array to a vertex structure. - */ -struct draw_vf_attr -{ - struct draw_vertex_fetch *vf; - - unsigned format; - unsigned inputsize; - unsigned inputstride; - unsigned vertoffset; /**< position of the attrib in the vertex struct */ - - boolean isconst; /**< read from const data below */ - uint8_t data[16]; - - unsigned attrib; /**< which vertex attrib (0=position, etc) */ - unsigned vertattrsize; /**< size of the attribute in bytes */ - - uint8_t *inputptr; - const draw_vf_insert_func *insert; - draw_vf_insert_func do_insert; - draw_vf_extract_func extract; -}; - -struct draw_vertex_fetch -{ - struct draw_vf_attr attr[PIPE_MAX_ATTRIBS]; - unsigned attr_count; - unsigned vertex_stride; - - draw_vf_emit_func emit; - - /* Parameters and constants for codegen: - */ - float identity[4]; - - struct draw_vf_fastpath *fastpath; - - void (*codegen_emit)( struct draw_vertex_fetch *vf ); -}; - - -struct draw_vf_attr_type { - unsigned format; - unsigned size; - unsigned stride; - unsigned offset; -}; - -/** XXX this could be moved into draw_vf.c */ -struct draw_vf_fastpath { - unsigned vertex_stride; - unsigned attr_count; - boolean match_strides; - - struct draw_vf_attr_type *attr; - - draw_vf_emit_func func; - struct draw_vf_fastpath *next; -}; - - -void -draw_vf_register_fastpath( struct draw_vertex_fetch *vtx, - boolean match_strides ); - -void -draw_vf_generic_emit( struct draw_vertex_fetch *vf, - unsigned count, - uint8_t *v ); - -void -draw_vf_generate_hardwired_emit( struct draw_vertex_fetch *vf ); - -void -draw_vf_generate_sse_emit( struct draw_vertex_fetch *vf ); - - -/** XXX this type and function could probably be moved into draw_vf.c */ -struct draw_vf_format_info { - const char *name; - draw_vf_insert_func insert[4]; - const unsigned attrsize; - const boolean isconst; -}; - -extern const struct draw_vf_format_info -draw_vf_format_info[DRAW_EMIT_MAX]; - - -#endif diff --git a/src/gallium/auxiliary/draw/draw_vf_generic.c b/src/gallium/auxiliary/draw/draw_vf_generic.c deleted file mode 100644 index 7a60a9db9c..0000000000 --- a/src/gallium/auxiliary/draw/draw_vf_generic.c +++ /dev/null @@ -1,585 +0,0 @@ - -/* - * Copyright 2003 Tungsten Graphics, inc. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * on the rights to use, copy, modify, merge, publish, distribute, sub - * license, and/or sell copies of the Software, and to permit persons to whom - * the Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice (including the next - * paragraph) shall be included in all copies or substantial portions of the - * Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL - * TUNGSTEN GRAPHICS AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM, - * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR - * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE - * USE OR OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: - * Keith Whitwell <keithw@tungstengraphics.com> - */ - - -#include "pipe/p_compiler.h" -#include "pipe/p_debug.h" -#include "pipe/p_util.h" - -#include "draw_vf.h" - - - -static INLINE void insert_4f_4( const struct draw_vf_attr *a, uint8_t *v, const float *in ) -{ - float *out = (float *)(v); - (void) a; - - out[0] = in[0]; - out[1] = in[1]; - out[2] = in[2]; - out[3] = in[3]; -} - -static INLINE void insert_4f_3( const struct draw_vf_attr *a, uint8_t *v, const float *in ) -{ - float *out = (float *)(v); - (void) a; - - out[0] = in[0]; - out[1] = in[1]; - out[2] = in[2]; - out[3] = 1; -} - -static INLINE void insert_4f_2( const struct draw_vf_attr *a, uint8_t *v, const float *in ) -{ - float *out = (float *)(v); - (void) a; - - out[0] = in[0]; - out[1] = in[1]; - out[2] = 0; - out[3] = 1; -} - -static INLINE void insert_4f_1( const struct draw_vf_attr *a, uint8_t *v, const float *in ) -{ - float *out = (float *)(v); - (void) a; - - out[0] = in[0]; - out[1] = 0; - out[2] = 0; - out[3] = 1; -} - -static INLINE void insert_3f_xyw_4( const struct draw_vf_attr *a, uint8_t *v, const float *in ) -{ - float *out = (float *)(v); - (void) a; - - out[0] = in[0]; - out[1] = in[1]; - out[2] = in[3]; -} - -static INLINE void insert_3f_xyw_err( const struct draw_vf_attr *a, uint8_t *v, const float *in ) -{ - (void) a; (void) v; (void) in; - assert(0); -} - -static INLINE void insert_3f_3( const struct draw_vf_attr *a, uint8_t *v, const float *in ) -{ - float *out = (float *)(v); - (void) a; - - out[0] = in[0]; - out[1] = in[1]; - out[2] = in[2]; -} - -static INLINE void insert_3f_2( const struct draw_vf_attr *a, uint8_t *v, const float *in ) -{ - float *out = (float *)(v); - (void) a; - - out[0] = in[0]; - out[1] = in[1]; - out[2] = 0; -} - -static INLINE void insert_3f_1( const struct draw_vf_attr *a, uint8_t *v, const float *in ) -{ - float *out = (float *)(v); - (void) a; - - out[0] = in[0]; - out[1] = 0; - out[2] = 0; -} - - -static INLINE void insert_2f_2( const struct draw_vf_attr *a, uint8_t *v, const float *in ) -{ - float *out = (float *)(v); - (void) a; - - out[0] = in[0]; - out[1] = in[1]; -} - -static INLINE void insert_2f_1( const struct draw_vf_attr *a, uint8_t *v, const float *in ) -{ - float *out = (float *)(v); - (void) a; - - out[0] = in[0]; - out[1] = 0; -} - -static INLINE void insert_1f_1( const struct draw_vf_attr *a, uint8_t *v, const float *in ) -{ - float *out = (float *)(v); - (void) a; - - out[0] = in[0]; -} - -static INLINE void insert_null( const struct draw_vf_attr *a, uint8_t *v, const float *in ) -{ - (void) a; (void) v; (void) in; -} - -static INLINE void insert_4ub_4f_rgba_4( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - (void) a; - UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[0]); - UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[1]); - UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[2]); - UNCLAMPED_FLOAT_TO_UBYTE(v[3], in[3]); -} - -static INLINE void insert_4ub_4f_rgba_3( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - (void) a; - UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[0]); - UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[1]); - UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[2]); - v[3] = 0xff; -} - -static INLINE void insert_4ub_4f_rgba_2( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - (void) a; - UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[0]); - UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[1]); - v[2] = 0; - v[3] = 0xff; -} - -static INLINE void insert_4ub_4f_rgba_1( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - (void) a; - UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[0]); - v[1] = 0; - v[2] = 0; - v[3] = 0xff; -} - -static INLINE void insert_4ub_4f_bgra_4( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - (void) a; - UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[0]); - UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[1]); - UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[2]); - UNCLAMPED_FLOAT_TO_UBYTE(v[3], in[3]); -} - -static INLINE void insert_4ub_4f_bgra_3( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - (void) a; - UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[0]); - UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[1]); - UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[2]); - v[3] = 0xff; -} - -static INLINE void insert_4ub_4f_bgra_2( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - (void) a; - UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[0]); - UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[1]); - v[0] = 0; - v[3] = 0xff; -} - -static INLINE void insert_4ub_4f_bgra_1( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - (void) a; - UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[0]); - v[1] = 0; - v[0] = 0; - v[3] = 0xff; -} - -static INLINE void insert_4ub_4f_argb_4( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - (void) a; - UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[0]); - UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[1]); - UNCLAMPED_FLOAT_TO_UBYTE(v[3], in[2]); - UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[3]); -} - -static INLINE void insert_4ub_4f_argb_3( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - (void) a; - UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[0]); - UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[1]); - UNCLAMPED_FLOAT_TO_UBYTE(v[3], in[2]); - v[0] = 0xff; -} - -static INLINE void insert_4ub_4f_argb_2( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - (void) a; - UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[0]); - UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[1]); - v[3] = 0x00; - v[0] = 0xff; -} - -static INLINE void insert_4ub_4f_argb_1( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - (void) a; - UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[0]); - v[2] = 0x00; - v[3] = 0x00; - v[0] = 0xff; -} - -static INLINE void insert_4ub_4f_abgr_4( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - (void) a; - UNCLAMPED_FLOAT_TO_UBYTE(v[3], in[0]); - UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[1]); - UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[2]); - UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[3]); -} - -static INLINE void insert_4ub_4f_abgr_3( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - (void) a; - UNCLAMPED_FLOAT_TO_UBYTE(v[3], in[0]); - UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[1]); - UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[2]); - v[0] = 0xff; -} - -static INLINE void insert_4ub_4f_abgr_2( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - (void) a; - UNCLAMPED_FLOAT_TO_UBYTE(v[3], in[0]); - UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[1]); - v[1] = 0x00; - v[0] = 0xff; -} - -static INLINE void insert_4ub_4f_abgr_1( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - (void) a; - UNCLAMPED_FLOAT_TO_UBYTE(v[3], in[0]); - v[2] = 0x00; - v[1] = 0x00; - v[0] = 0xff; -} - -static INLINE void insert_3ub_3f_rgb_3( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - (void) a; - UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[0]); - UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[1]); - UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[2]); -} - -static INLINE void insert_3ub_3f_rgb_2( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - (void) a; - UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[0]); - UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[1]); - v[2] = 0; -} - -static INLINE void insert_3ub_3f_rgb_1( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - (void) a; - UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[0]); - v[1] = 0; - v[2] = 0; -} - -static INLINE void insert_3ub_3f_bgr_3( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - (void) a; - UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[0]); - UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[1]); - UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[2]); -} - -static INLINE void insert_3ub_3f_bgr_2( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - (void) a; - UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[0]); - UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[1]); - v[0] = 0; -} - -static INLINE void insert_3ub_3f_bgr_1( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - (void) a; - UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[0]); - v[1] = 0; - v[0] = 0; -} - - -static INLINE void insert_1ub_1f_1( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - (void) a; - UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[0]); -} - - -const struct draw_vf_format_info draw_vf_format_info[DRAW_EMIT_MAX] = -{ - { "1f", - { insert_1f_1, insert_1f_1, insert_1f_1, insert_1f_1 }, - sizeof(float), FALSE }, - - { "2f", - { insert_2f_1, insert_2f_2, insert_2f_2, insert_2f_2 }, - 2 * sizeof(float), FALSE }, - - { "3f", - { insert_3f_1, insert_3f_2, insert_3f_3, insert_3f_3 }, - 3 * sizeof(float), FALSE }, - - { "4f", - { insert_4f_1, insert_4f_2, insert_4f_3, insert_4f_4 }, - 4 * sizeof(float), FALSE }, - - { "3f_xyw", - { insert_3f_xyw_err, insert_3f_xyw_err, insert_3f_xyw_err, - insert_3f_xyw_4 }, - 3 * sizeof(float), FALSE }, - - { "1ub_1f", - { insert_1ub_1f_1, insert_1ub_1f_1, insert_1ub_1f_1, insert_1ub_1f_1 }, - sizeof(uint8_t), FALSE }, - - { "3ub_3f_rgb", - { insert_3ub_3f_rgb_1, insert_3ub_3f_rgb_2, insert_3ub_3f_rgb_3, - insert_3ub_3f_rgb_3 }, - 3 * sizeof(uint8_t), FALSE }, - - { "3ub_3f_bgr", - { insert_3ub_3f_bgr_1, insert_3ub_3f_bgr_2, insert_3ub_3f_bgr_3, - insert_3ub_3f_bgr_3 }, - 3 * sizeof(uint8_t), FALSE }, - - { "4ub_4f_rgba", - { insert_4ub_4f_rgba_1, insert_4ub_4f_rgba_2, insert_4ub_4f_rgba_3, - insert_4ub_4f_rgba_4 }, - 4 * sizeof(uint8_t), FALSE }, - - { "4ub_4f_bgra", - { insert_4ub_4f_bgra_1, insert_4ub_4f_bgra_2, insert_4ub_4f_bgra_3, - insert_4ub_4f_bgra_4 }, - 4 * sizeof(uint8_t), FALSE }, - - { "4ub_4f_argb", - { insert_4ub_4f_argb_1, insert_4ub_4f_argb_2, insert_4ub_4f_argb_3, - insert_4ub_4f_argb_4 }, - 4 * sizeof(uint8_t), FALSE }, - - { "4ub_4f_abgr", - { insert_4ub_4f_abgr_1, insert_4ub_4f_abgr_2, insert_4ub_4f_abgr_3, - insert_4ub_4f_abgr_4 }, - 4 * sizeof(uint8_t), FALSE }, - - { "1f_const", - { insert_1f_1, insert_1f_1, insert_1f_1, insert_1f_1 }, - sizeof(float), TRUE }, - - { "2f_const", - { insert_2f_1, insert_2f_2, insert_2f_2, insert_2f_2 }, - 2 * sizeof(float), TRUE }, - - { "3f_const", - { insert_3f_1, insert_3f_2, insert_3f_3, insert_3f_3 }, - 3 * sizeof(float), TRUE }, - - { "4f_const", - { insert_4f_1, insert_4f_2, insert_4f_3, insert_4f_4 }, - 4 * sizeof(float), TRUE }, - - { "pad", - { NULL, NULL, NULL, NULL }, - 0, FALSE }, - -}; - - - - -/*********************************************************************** - * Hardwired fastpaths for emitting whole vertices or groups of - * vertices - */ -#define EMIT5(NR, F0, F1, F2, F3, F4, NAME) \ -static void NAME( struct draw_vertex_fetch *vf, \ - unsigned count, \ - uint8_t *v ) \ -{ \ - struct draw_vf_attr *a = vf->attr; \ - unsigned i; \ - \ - for (i = 0 ; i < count ; i++, v += vf->vertex_stride) { \ - if (NR > 0) { \ - F0( &a[0], v + a[0].vertoffset, (float *)a[0].inputptr ); \ - a[0].inputptr += a[0].inputstride; \ - } \ - \ - if (NR > 1) { \ - F1( &a[1], v + a[1].vertoffset, (float *)a[1].inputptr ); \ - a[1].inputptr += a[1].inputstride; \ - } \ - \ - if (NR > 2) { \ - F2( &a[2], v + a[2].vertoffset, (float *)a[2].inputptr ); \ - a[2].inputptr += a[2].inputstride; \ - } \ - \ - if (NR > 3) { \ - F3( &a[3], v + a[3].vertoffset, (float *)a[3].inputptr ); \ - a[3].inputptr += a[3].inputstride; \ - } \ - \ - if (NR > 4) { \ - F4( &a[4], v + a[4].vertoffset, (float *)a[4].inputptr ); \ - a[4].inputptr += a[4].inputstride; \ - } \ - } \ -} - - -#define EMIT2(F0, F1, NAME) EMIT5(2, F0, F1, insert_null, \ - insert_null, insert_null, NAME) - -#define EMIT3(F0, F1, F2, NAME) EMIT5(3, F0, F1, F2, insert_null, \ - insert_null, NAME) - -#define EMIT4(F0, F1, F2, F3, NAME) EMIT5(4, F0, F1, F2, F3, \ - insert_null, NAME) - - -EMIT2(insert_3f_3, insert_4ub_4f_rgba_4, emit_xyz3_rgba4) - -EMIT3(insert_4f_4, insert_4ub_4f_rgba_4, insert_2f_2, emit_xyzw4_rgba4_st2) - -EMIT4(insert_4f_4, insert_4ub_4f_rgba_4, insert_2f_2, insert_2f_2, emit_xyzw4_rgba4_st2_st2) - - -/* Use the codegen paths to select one of a number of hardwired - * fastpaths. - */ -void draw_vf_generate_hardwired_emit( struct draw_vertex_fetch *vf ) -{ - draw_vf_emit_func func = NULL; - - /* Does it fit a hardwired fastpath? Help! this is growing out of - * control! - */ - switch (vf->attr_count) { - case 2: - if (vf->attr[0].do_insert == insert_3f_3 && - vf->attr[1].do_insert == insert_4ub_4f_rgba_4) { - func = emit_xyz3_rgba4; - } - break; - case 3: - if (vf->attr[2].do_insert == insert_2f_2) { - if (vf->attr[1].do_insert == insert_4ub_4f_rgba_4) { - if (vf->attr[0].do_insert == insert_4f_4) - func = emit_xyzw4_rgba4_st2; - } - } - break; - case 4: - if (vf->attr[2].do_insert == insert_2f_2 && - vf->attr[3].do_insert == insert_2f_2) { - if (vf->attr[1].do_insert == insert_4ub_4f_rgba_4) { - if (vf->attr[0].do_insert == insert_4f_4) - func = emit_xyzw4_rgba4_st2_st2; - } - } - break; - } - - vf->emit = func; -} - -/*********************************************************************** - * Generic (non-codegen) functions for whole vertices or groups of - * vertices - */ - -void draw_vf_generic_emit( struct draw_vertex_fetch *vf, - unsigned count, - uint8_t *v ) -{ - struct draw_vf_attr *a = vf->attr; - const unsigned attr_count = vf->attr_count; - const unsigned stride = vf->vertex_stride; - unsigned i, j; - - for (i = 0 ; i < count ; i++, v += stride) { - for (j = 0; j < attr_count; j++) { - float *in = (float *)a[j].inputptr; - a[j].inputptr += a[j].inputstride; - a[j].do_insert( &a[j], v + a[j].vertoffset, in ); - } - } -} - - diff --git a/src/gallium/auxiliary/draw/draw_vf_sse.c b/src/gallium/auxiliary/draw/draw_vf_sse.c deleted file mode 100644 index aff4ffd985..0000000000 --- a/src/gallium/auxiliary/draw/draw_vf_sse.c +++ /dev/null @@ -1,613 +0,0 @@ -/* - * Copyright 2003 Tungsten Graphics, inc. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * on the rights to use, copy, modify, merge, publish, distribute, sub - * license, and/or sell copies of the Software, and to permit persons to whom - * the Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice (including the next - * paragraph) shall be included in all copies or substantial portions of the - * Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL - * TUNGSTEN GRAPHICS AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM, - * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR - * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE - * USE OR OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: - * Keith Whitwell <keithw@tungstengraphics.com> - */ - - -#include "pipe/p_compiler.h" -#include "util/u_simple_list.h" - -#include "draw_vf.h" - - -#if defined(USE_SSE_ASM) - -#include "rtasm/rtasm_cpu.h" -#include "rtasm/rtasm_x86sse.h" - - -#define X 0 -#define Y 1 -#define Z 2 -#define W 3 - - -struct x86_program { - struct x86_function func; - - struct draw_vertex_fetch *vf; - boolean inputs_safe; - boolean outputs_safe; - boolean have_sse2; - - struct x86_reg identity; - struct x86_reg chan0; -}; - - -static struct x86_reg get_identity( struct x86_program *p ) -{ - return p->identity; -} - -static void emit_load4f_4( struct x86_program *p, - struct x86_reg dest, - struct x86_reg arg0 ) -{ - sse_movups(&p->func, dest, arg0); -} - -static void emit_load4f_3( struct x86_program *p, - struct x86_reg dest, - struct x86_reg arg0 ) -{ - /* Have to jump through some hoops: - * - * c 0 0 0 - * c 0 0 1 - * 0 0 c 1 - * a b c 1 - */ - sse_movss(&p->func, dest, x86_make_disp(arg0, 8)); - sse_shufps(&p->func, dest, get_identity(p), SHUF(X,Y,Z,W) ); - sse_shufps(&p->func, dest, dest, SHUF(Y,Z,X,W) ); - sse_movlps(&p->func, dest, arg0); -} - -static void emit_load4f_2( struct x86_program *p, - struct x86_reg dest, - struct x86_reg arg0 ) -{ - /* Initialize from identity, then pull in low two words: - */ - sse_movups(&p->func, dest, get_identity(p)); - sse_movlps(&p->func, dest, arg0); -} - -static void emit_load4f_1( struct x86_program *p, - struct x86_reg dest, - struct x86_reg arg0 ) -{ - /* Pull in low word, then swizzle in identity */ - sse_movss(&p->func, dest, arg0); - sse_shufps(&p->func, dest, get_identity(p), SHUF(X,Y,Z,W) ); -} - - - -static void emit_load3f_3( struct x86_program *p, - struct x86_reg dest, - struct x86_reg arg0 ) -{ - /* Over-reads by 1 dword - potential SEGV if input is a vertex - * array. - */ - if (p->inputs_safe) { - sse_movups(&p->func, dest, arg0); - } - else { - /* c 0 0 0 - * c c c c - * a b c c - */ - sse_movss(&p->func, dest, x86_make_disp(arg0, 8)); - sse_shufps(&p->func, dest, dest, SHUF(X,X,X,X)); - sse_movlps(&p->func, dest, arg0); - } -} - -static void emit_load3f_2( struct x86_program *p, - struct x86_reg dest, - struct x86_reg arg0 ) -{ - emit_load4f_2(p, dest, arg0); -} - -static void emit_load3f_1( struct x86_program *p, - struct x86_reg dest, - struct x86_reg arg0 ) -{ - emit_load4f_1(p, dest, arg0); -} - -static void emit_load2f_2( struct x86_program *p, - struct x86_reg dest, - struct x86_reg arg0 ) -{ - sse_movlps(&p->func, dest, arg0); -} - -static void emit_load2f_1( struct x86_program *p, - struct x86_reg dest, - struct x86_reg arg0 ) -{ - emit_load4f_1(p, dest, arg0); -} - -static void emit_load1f_1( struct x86_program *p, - struct x86_reg dest, - struct x86_reg arg0 ) -{ - sse_movss(&p->func, dest, arg0); -} - -static void (*load[4][4])( struct x86_program *p, - struct x86_reg dest, - struct x86_reg arg0 ) = { - { emit_load1f_1, - emit_load1f_1, - emit_load1f_1, - emit_load1f_1 }, - - { emit_load2f_1, - emit_load2f_2, - emit_load2f_2, - emit_load2f_2 }, - - { emit_load3f_1, - emit_load3f_2, - emit_load3f_3, - emit_load3f_3 }, - - { emit_load4f_1, - emit_load4f_2, - emit_load4f_3, - emit_load4f_4 } -}; - -static void emit_load( struct x86_program *p, - struct x86_reg dest, - unsigned sz, - struct x86_reg src, - unsigned src_sz) -{ - load[sz-1][src_sz-1](p, dest, src); -} - -static void emit_store4f( struct x86_program *p, - struct x86_reg dest, - struct x86_reg arg0 ) -{ - sse_movups(&p->func, dest, arg0); -} - -static void emit_store3f( struct x86_program *p, - struct x86_reg dest, - struct x86_reg arg0 ) -{ - if (p->outputs_safe) { - /* Emit the extra dword anyway. This may hurt writecombining, - * may cause other problems. - */ - sse_movups(&p->func, dest, arg0); - } - else { - /* Alternate strategy - emit two, shuffle, emit one. - */ - sse_movlps(&p->func, dest, arg0); - sse_shufps(&p->func, arg0, arg0, SHUF(Z,Z,Z,Z) ); /* NOTE! destructive */ - sse_movss(&p->func, x86_make_disp(dest,8), arg0); - } -} - -static void emit_store2f( struct x86_program *p, - struct x86_reg dest, - struct x86_reg arg0 ) -{ - sse_movlps(&p->func, dest, arg0); -} - -static void emit_store1f( struct x86_program *p, - struct x86_reg dest, - struct x86_reg arg0 ) -{ - sse_movss(&p->func, dest, arg0); -} - - -static void (*store[4])( struct x86_program *p, - struct x86_reg dest, - struct x86_reg arg0 ) = -{ - emit_store1f, - emit_store2f, - emit_store3f, - emit_store4f -}; - -static void emit_store( struct x86_program *p, - struct x86_reg dest, - unsigned sz, - struct x86_reg temp ) - -{ - store[sz-1](p, dest, temp); -} - -static void emit_pack_store_4ub( struct x86_program *p, - struct x86_reg dest, - struct x86_reg temp ) -{ - /* Scale by 255.0 - */ - sse_mulps(&p->func, temp, p->chan0); - - if (p->have_sse2) { - sse2_cvtps2dq(&p->func, temp, temp); - sse2_packssdw(&p->func, temp, temp); - sse2_packuswb(&p->func, temp, temp); - sse_movss(&p->func, dest, temp); - } - else { - struct x86_reg mmx0 = x86_make_reg(file_MMX, 0); - struct x86_reg mmx1 = x86_make_reg(file_MMX, 1); - sse_cvtps2pi(&p->func, mmx0, temp); - sse_movhlps(&p->func, temp, temp); - sse_cvtps2pi(&p->func, mmx1, temp); - mmx_packssdw(&p->func, mmx0, mmx1); - mmx_packuswb(&p->func, mmx0, mmx0); - mmx_movd(&p->func, dest, mmx0); - } -} - -static int get_offset( const void *a, const void *b ) -{ - return (const char *)b - (const char *)a; -} - -/* Not much happens here. Eventually use this function to try and - * avoid saving/reloading the source pointers each vertex (if some of - * them can fit in registers). - */ -static void get_src_ptr( struct x86_program *p, - struct x86_reg srcREG, - struct x86_reg vfREG, - struct draw_vf_attr *a ) -{ - struct draw_vertex_fetch *vf = p->vf; - struct x86_reg ptr_to_src = x86_make_disp(vfREG, get_offset(vf, &a->inputptr)); - - /* Load current a[j].inputptr - */ - x86_mov(&p->func, srcREG, ptr_to_src); -} - -static void update_src_ptr( struct x86_program *p, - struct x86_reg srcREG, - struct x86_reg vfREG, - struct draw_vf_attr *a ) -{ - if (a->inputstride) { - struct draw_vertex_fetch *vf = p->vf; - struct x86_reg ptr_to_src = x86_make_disp(vfREG, get_offset(vf, &a->inputptr)); - - /* add a[j].inputstride (hardcoded value - could just as easily - * pull the stride value from memory each time). - */ - x86_lea(&p->func, srcREG, x86_make_disp(srcREG, a->inputstride)); - - /* save new value of a[j].inputptr - */ - x86_mov(&p->func, ptr_to_src, srcREG); - } -} - - -/* Lots of hardcoding - * - * EAX -- pointer to current output vertex - * ECX -- pointer to current attribute - * - */ -static boolean build_vertex_emit( struct x86_program *p ) -{ - struct draw_vertex_fetch *vf = p->vf; - unsigned j = 0; - - struct x86_reg vertexEAX = x86_make_reg(file_REG32, reg_AX); - struct x86_reg srcECX = x86_make_reg(file_REG32, reg_CX); - struct x86_reg countEBP = x86_make_reg(file_REG32, reg_BP); - struct x86_reg vfESI = x86_make_reg(file_REG32, reg_SI); - struct x86_reg temp = x86_make_reg(file_XMM, 0); - uint8_t *fixup, *label; - - /* Push a few regs? - */ - x86_push(&p->func, countEBP); - x86_push(&p->func, vfESI); - - - /* Get vertex count, compare to zero - */ - x86_xor(&p->func, srcECX, srcECX); - x86_mov(&p->func, countEBP, x86_fn_arg(&p->func, 2)); - x86_cmp(&p->func, countEBP, srcECX); - fixup = x86_jcc_forward(&p->func, cc_E); - - /* Initialize destination register. - */ - x86_mov(&p->func, vertexEAX, x86_fn_arg(&p->func, 3)); - - /* Move argument 1 (vf) into a reg: - */ - x86_mov(&p->func, vfESI, x86_fn_arg(&p->func, 1)); - - - /* always load, needed or not: - */ - sse_movups(&p->func, p->identity, x86_make_disp(vfESI, get_offset(vf, &vf->identity[0]))); - - /* Note address for loop jump */ - label = x86_get_label(&p->func); - - /* Emit code for each of the attributes. Currently routes - * everything through SSE registers, even when it might be more - * efficient to stick with regular old x86. No optimization or - * other tricks - enough new ground to cover here just getting - * things working. - */ - while (j < vf->attr_count) { - struct draw_vf_attr *a = &vf->attr[j]; - struct x86_reg dest = x86_make_disp(vertexEAX, a->vertoffset); - - /* Now, load an XMM reg from src, perhaps transform, then save. - * Could be shortcircuited in specific cases: - */ - switch (a->format) { - case DRAW_EMIT_1F: - case DRAW_EMIT_1F_CONST: - get_src_ptr(p, srcECX, vfESI, a); - emit_load(p, temp, 1, x86_deref(srcECX), a->inputsize); - emit_store(p, dest, 1, temp); - update_src_ptr(p, srcECX, vfESI, a); - break; - case DRAW_EMIT_2F: - case DRAW_EMIT_2F_CONST: - get_src_ptr(p, srcECX, vfESI, a); - emit_load(p, temp, 2, x86_deref(srcECX), a->inputsize); - emit_store(p, dest, 2, temp); - update_src_ptr(p, srcECX, vfESI, a); - break; - case DRAW_EMIT_3F: - case DRAW_EMIT_3F_CONST: - /* Potentially the worst case - hardcode 2+1 copying: - */ - if (0) { - get_src_ptr(p, srcECX, vfESI, a); - emit_load(p, temp, 3, x86_deref(srcECX), a->inputsize); - emit_store(p, dest, 3, temp); - update_src_ptr(p, srcECX, vfESI, a); - } - else { - get_src_ptr(p, srcECX, vfESI, a); - emit_load(p, temp, 2, x86_deref(srcECX), a->inputsize); - emit_store(p, dest, 2, temp); - if (a->inputsize > 2) { - emit_load(p, temp, 1, x86_make_disp(srcECX, 8), 1); - emit_store(p, x86_make_disp(dest,8), 1, temp); - } - else { - sse_movss(&p->func, x86_make_disp(dest,8), get_identity(p)); - } - update_src_ptr(p, srcECX, vfESI, a); - } - break; - case DRAW_EMIT_4F: - case DRAW_EMIT_4F_CONST: - get_src_ptr(p, srcECX, vfESI, a); - emit_load(p, temp, 4, x86_deref(srcECX), a->inputsize); - emit_store(p, dest, 4, temp); - update_src_ptr(p, srcECX, vfESI, a); - break; - case DRAW_EMIT_3F_XYW: - get_src_ptr(p, srcECX, vfESI, a); - emit_load(p, temp, 4, x86_deref(srcECX), a->inputsize); - sse_shufps(&p->func, temp, temp, SHUF(X,Y,W,Z)); - emit_store(p, dest, 3, temp); - update_src_ptr(p, srcECX, vfESI, a); - break; - - case DRAW_EMIT_1UB_1F: - /* Test for PAD3 + 1UB: - */ - if (j > 0 && - a[-1].vertoffset + a[-1].vertattrsize <= a->vertoffset - 3) - { - get_src_ptr(p, srcECX, vfESI, a); - emit_load(p, temp, 1, x86_deref(srcECX), a->inputsize); - sse_shufps(&p->func, temp, temp, SHUF(X,X,X,X)); - emit_pack_store_4ub(p, x86_make_disp(dest, -3), temp); /* overkill! */ - update_src_ptr(p, srcECX, vfESI, a); - } - else { - debug_printf("Can't emit 1ub %x %x %d\n", - a->vertoffset, a[-1].vertoffset, a[-1].vertattrsize ); - return FALSE; - } - break; - case DRAW_EMIT_3UB_3F_RGB: - case DRAW_EMIT_3UB_3F_BGR: - /* Test for 3UB + PAD1: - */ - if (j == vf->attr_count - 1 || - a[1].vertoffset >= a->vertoffset + 4) { - get_src_ptr(p, srcECX, vfESI, a); - emit_load(p, temp, 3, x86_deref(srcECX), a->inputsize); - if (a->format == DRAW_EMIT_3UB_3F_BGR) - sse_shufps(&p->func, temp, temp, SHUF(Z,Y,X,W)); - emit_pack_store_4ub(p, dest, temp); - update_src_ptr(p, srcECX, vfESI, a); - } - /* Test for 3UB + 1UB: - */ - else if (j < vf->attr_count - 1 && - a[1].format == DRAW_EMIT_1UB_1F && - a[1].vertoffset == a->vertoffset + 3) { - get_src_ptr(p, srcECX, vfESI, a); - emit_load(p, temp, 3, x86_deref(srcECX), a->inputsize); - update_src_ptr(p, srcECX, vfESI, a); - - /* Make room for incoming value: - */ - sse_shufps(&p->func, temp, temp, SHUF(W,X,Y,Z)); - - get_src_ptr(p, srcECX, vfESI, &a[1]); - emit_load(p, temp, 1, x86_deref(srcECX), a[1].inputsize); - update_src_ptr(p, srcECX, vfESI, &a[1]); - - /* Rearrange and possibly do BGR conversion: - */ - if (a->format == DRAW_EMIT_3UB_3F_BGR) - sse_shufps(&p->func, temp, temp, SHUF(W,Z,Y,X)); - else - sse_shufps(&p->func, temp, temp, SHUF(Y,Z,W,X)); - - emit_pack_store_4ub(p, dest, temp); - j++; /* NOTE: two attrs consumed */ - } - else { - debug_printf("Can't emit 3ub\n"); - } - return FALSE; /* add this later */ - break; - - case DRAW_EMIT_4UB_4F_RGBA: - get_src_ptr(p, srcECX, vfESI, a); - emit_load(p, temp, 4, x86_deref(srcECX), a->inputsize); - emit_pack_store_4ub(p, dest, temp); - update_src_ptr(p, srcECX, vfESI, a); - break; - case DRAW_EMIT_4UB_4F_BGRA: - get_src_ptr(p, srcECX, vfESI, a); - emit_load(p, temp, 4, x86_deref(srcECX), a->inputsize); - sse_shufps(&p->func, temp, temp, SHUF(Z,Y,X,W)); - emit_pack_store_4ub(p, dest, temp); - update_src_ptr(p, srcECX, vfESI, a); - break; - case DRAW_EMIT_4UB_4F_ARGB: - get_src_ptr(p, srcECX, vfESI, a); - emit_load(p, temp, 4, x86_deref(srcECX), a->inputsize); - sse_shufps(&p->func, temp, temp, SHUF(W,X,Y,Z)); - emit_pack_store_4ub(p, dest, temp); - update_src_ptr(p, srcECX, vfESI, a); - break; - case DRAW_EMIT_4UB_4F_ABGR: - get_src_ptr(p, srcECX, vfESI, a); - emit_load(p, temp, 4, x86_deref(srcECX), a->inputsize); - sse_shufps(&p->func, temp, temp, SHUF(W,Z,Y,X)); - emit_pack_store_4ub(p, dest, temp); - update_src_ptr(p, srcECX, vfESI, a); - break; - default: - debug_printf("unknown a[%d].format %d\n", j, a->format); - return FALSE; /* catch any new opcodes */ - } - - /* Increment j by at least 1 - may have been incremented above also: - */ - j++; - } - - /* Next vertex: - */ - x86_lea(&p->func, vertexEAX, x86_make_disp(vertexEAX, vf->vertex_stride)); - - /* decr count, loop if not zero - */ - x86_dec(&p->func, countEBP); - x86_test(&p->func, countEBP, countEBP); - x86_jcc(&p->func, cc_NZ, label); - - /* Exit mmx state? - */ - if (p->func.need_emms) - mmx_emms(&p->func); - - /* Land forward jump here: - */ - x86_fixup_fwd_jump(&p->func, fixup); - - /* Pop regs and return - */ - x86_pop(&p->func, x86_get_base_reg(vfESI)); - x86_pop(&p->func, countEBP); - x86_ret(&p->func); - - vf->emit = (draw_vf_emit_func)x86_get_func(&p->func); - return TRUE; -} - - - -void draw_vf_generate_sse_emit( struct draw_vertex_fetch *vf ) -{ - struct x86_program p; - - if (!rtasm_cpu_has_sse()) { - vf->codegen_emit = NULL; - return; - } - - memset(&p, 0, sizeof(p)); - - p.vf = vf; - p.inputs_safe = 0; /* for now */ - p.outputs_safe = 1; /* for now */ - p.have_sse2 = rtasm_cpu_has_sse2(); - p.identity = x86_make_reg(file_XMM, 6); - p.chan0 = x86_make_reg(file_XMM, 7); - - x86_init_func(&p.func); - - if (build_vertex_emit(&p)) { - draw_vf_register_fastpath( vf, TRUE ); - } - else { - /* Note the failure so that we don't keep trying to codegen an - * impossible state: - */ - draw_vf_register_fastpath( vf, FALSE ); - x86_release_func(&p.func); - } -} - -#else - -void draw_vf_generate_sse_emit( struct draw_vertex_fetch *vf ) -{ - /* Dummy version for when USE_SSE_ASM not defined */ -} - -#endif diff --git a/src/gallium/auxiliary/draw/draw_vertex_shader.c b/src/gallium/auxiliary/draw/draw_vs.c index 133418baca..03fe00a951 100644 --- a/src/gallium/auxiliary/draw/draw_vertex_shader.c +++ b/src/gallium/auxiliary/draw/draw_vs.c @@ -39,51 +39,6 @@ -/** - * Run the vertex shader on all vertices in the vertex queue. - * Called by the draw module when the vertx cache needs to be flushed. - */ -void -draw_vertex_shader_queue_flush(struct draw_context *draw) -{ - struct draw_vertex_shader *shader = draw->vertex_shader; - unsigned i; - - assert(draw->vs.queue_nr != 0); - - /* XXX: do this on statechange: - */ - shader->prepare( shader, draw ); - -// fprintf(stderr, "%s %d\n", __FUNCTION__, draw->vs.queue_nr ); - - /* run vertex shader on vertex cache entries, four per invokation */ - for (i = 0; i < draw->vs.queue_nr; i += 4) { - struct vertex_header *dests[4]; - unsigned elts[4]; - int j, n = MIN2(4, draw->vs.queue_nr - i); - - for (j = 0; j < n; j++) { - elts[j] = draw->vs.queue[i + j].elt; - dests[j] = draw->vs.queue[i + j].vertex; - } - - for ( ; j < 4; j++) { - elts[j] = elts[0]; - dests[j] = draw->vs.queue[i + j].vertex; - } - - assert(n > 0); - assert(n <= 4); - - shader->run(shader, draw, elts, n, dests); - } - - draw->vs.post_nr = draw->vs.queue_nr; - draw->vs.queue_nr = 0; -} - - struct draw_vertex_shader * draw_create_vertex_shader(struct draw_context *draw, const struct pipe_shader_state *shader) @@ -97,10 +52,8 @@ draw_create_vertex_shader(struct draw_context *draw, vs = draw_create_vs_exec( draw, shader ); } } - assert(vs); - - tgsi_scan_shader(shader->tokens, &vs->info); + assert(vs); return vs; } @@ -115,9 +68,6 @@ draw_bind_vertex_shader(struct draw_context *draw, { draw->vertex_shader = dvs; draw->num_vs_outputs = dvs->info.num_outputs; - - tgsi_exec_machine_init(&draw->machine); - dvs->prepare( dvs, draw ); } else { diff --git a/src/gallium/auxiliary/draw/draw_vs.h b/src/gallium/auxiliary/draw/draw_vs.h index 4ee7e705e9..f9772b83b8 100644 --- a/src/gallium/auxiliary/draw/draw_vs.h +++ b/src/gallium/auxiliary/draw/draw_vs.h @@ -31,10 +31,43 @@ #ifndef DRAW_VS_H #define DRAW_VS_H -struct draw_vertex_shader; +#include "draw_context.h" +#include "draw_private.h" + + struct draw_context; struct pipe_shader_state; +/** + * Private version of the compiled vertex_shader + */ +struct draw_vertex_shader { + + /* This member will disappear shortly: + */ + struct pipe_shader_state state; + + struct tgsi_shader_info info; + + void (*prepare)( struct draw_vertex_shader *shader, + struct draw_context *draw ); + + /* Run the shader - this interface will get cleaned up in the + * future: + */ + void (*run_linear)( struct draw_vertex_shader *shader, + const float (*input)[4], + float (*output)[4], + const float (*constants)[4], + unsigned count, + unsigned input_stride, + unsigned output_stride ); + + + void (*delete)( struct draw_vertex_shader * ); +}; + + struct draw_vertex_shader * draw_create_vs_exec(struct draw_context *draw, const struct pipe_shader_state *templ); @@ -47,4 +80,8 @@ struct draw_vertex_shader * draw_create_vs_llvm(struct draw_context *draw, const struct pipe_shader_state *templ); + +#define MAX_TGSI_VERTICES 4 + + #endif diff --git a/src/gallium/auxiliary/draw/draw_vs_exec.c b/src/gallium/auxiliary/draw/draw_vs_exec.c index c6e503686a..7a02f6334b 100644 --- a/src/gallium/auxiliary/draw/draw_vs_exec.c +++ b/src/gallium/auxiliary/draw/draw_vs_exec.c @@ -39,163 +39,120 @@ #include "draw_vs.h" #include "tgsi/util/tgsi_parse.h" +#include "tgsi/util/tgsi_scan.h" -static INLINE unsigned -compute_clipmask(const float *clip, /*const*/ float plane[][4], unsigned nr) -{ - unsigned mask = 0; - unsigned i; - - /* Do the hardwired planes first: - */ - if (-clip[0] + clip[3] < 0) mask |= CLIP_RIGHT_BIT; - if ( clip[0] + clip[3] < 0) mask |= CLIP_LEFT_BIT; - if (-clip[1] + clip[3] < 0) mask |= CLIP_TOP_BIT; - if ( clip[1] + clip[3] < 0) mask |= CLIP_BOTTOM_BIT; - if (-clip[2] + clip[3] < 0) mask |= CLIP_FAR_BIT; - if ( clip[2] + clip[3] < 0) mask |= CLIP_NEAR_BIT; - - /* Followed by any remaining ones: - */ - for (i = 6; i < nr; i++) { - if (dot4(clip, plane[i]) < 0) - mask |= (1<<i); - } +struct exec_vertex_shader { + struct draw_vertex_shader base; + struct tgsi_exec_machine *machine; +}; - return mask; +static struct exec_vertex_shader *exec_vertex_shader( struct draw_vertex_shader *vs ) +{ + return (struct exec_vertex_shader *)vs; } +/* Not required for run_linear. + */ static void vs_exec_prepare( struct draw_vertex_shader *shader, struct draw_context *draw ) { + struct exec_vertex_shader *evs = exec_vertex_shader(shader); + /* specify the vertex program to interpret/execute */ - tgsi_exec_machine_bind_shader(&draw->machine, + tgsi_exec_machine_bind_shader(evs->machine, shader->state.tokens, PIPE_MAX_SAMPLERS, NULL /*samplers*/ ); - draw_update_vertex_fetch( draw ); } -/** - * Transform vertices with the current vertex program/shader - * Up to four vertices can be shaded at a time. - * \param vbuffer the input vertex data - * \param elts indexes of four input vertices - * \param count number of vertices to shade [1..4] - * \param vOut array of pointers to four output vertices + + +/* Simplified vertex shader interface for the pt paths. Given the + * complexity of code-generating all the above operations together, + * it's time to try doing all the other stuff separately. */ static void -vs_exec_run( struct draw_vertex_shader *shader, - struct draw_context *draw, - const unsigned *elts, - unsigned count, - struct vertex_header *vOut[] ) +vs_exec_run_linear( struct draw_vertex_shader *shader, + const float (*input)[4], + float (*output)[4], + const float (*constants)[4], + unsigned count, + unsigned input_stride, + unsigned output_stride ) { - struct tgsi_exec_machine *machine = &draw->machine; - unsigned int j; - - ALIGN16_DECL(struct tgsi_exec_vector, inputs, PIPE_MAX_ATTRIBS); - ALIGN16_DECL(struct tgsi_exec_vector, outputs, PIPE_MAX_ATTRIBS); - const float *scale = draw->viewport.scale; - const float *trans = draw->viewport.translate; - - assert(count <= 4); - assert(draw->vertex_shader->info.output_semantic_name[0] - == TGSI_SEMANTIC_POSITION); - - machine->Consts = (float (*)[4]) draw->user.constants; - machine->Inputs = ALIGN16_ASSIGN(inputs); - if (draw->rasterizer->bypass_vs) { - /* outputs are just the inputs */ - machine->Outputs = machine->Inputs; - } - else { - machine->Outputs = ALIGN16_ASSIGN(outputs); - } + struct exec_vertex_shader *evs = exec_vertex_shader(shader); + struct tgsi_exec_machine *machine = evs->machine; + unsigned int i, j; + unsigned slot; + + machine->Consts = constants; + + for (i = 0; i < count; i += MAX_TGSI_VERTICES) { + unsigned int max_vertices = MIN2(MAX_TGSI_VERTICES, count - i); - draw->vertex_fetch.fetch_func( draw, machine, elts, count ); + /* Swizzle inputs. + */ + for (j = 0; j < max_vertices; j++) { +#if 0 + debug_printf("%d) Input vert:\n", i + j); + for (slot = 0; slot < shader->info.num_inputs; slot++) { + debug_printf("\t%d: %f %f %f %f\n", slot, + input[slot][0], + input[slot][1], + input[slot][2], + input[slot][3]); + } +#endif + + for (slot = 0; slot < shader->info.num_inputs; slot++) { + machine->Inputs[slot].xyzw[0].f[j] = input[slot][0]; + machine->Inputs[slot].xyzw[1].f[j] = input[slot][1]; + machine->Inputs[slot].xyzw[2].f[j] = input[slot][2]; + machine->Inputs[slot].xyzw[3].f[j] = input[slot][3]; + } + + input = (const float (*)[4])((const char *)input + input_stride); + } - if (!draw->rasterizer->bypass_vs) { /* run interpreter */ tgsi_exec_machine_run( machine ); - } - - /* store machine results */ - for (j = 0; j < count; j++) { - unsigned slot; - float x, y, z, w; - /* Handle attr[0] (position) specially: - * - * XXX: Computing the clipmask should be done in the vertex - * program as a set of DP4 instructions appended to the - * user-provided code. + /* Unswizzle all output results. */ - x = vOut[j]->clip[0] = machine->Outputs[0].xyzw[0].f[j]; - y = vOut[j]->clip[1] = machine->Outputs[0].xyzw[1].f[j]; - z = vOut[j]->clip[2] = machine->Outputs[0].xyzw[2].f[j]; - w = vOut[j]->clip[3] = machine->Outputs[0].xyzw[3].f[j]; - - if (!draw->rasterizer->bypass_clipping) { - vOut[j]->clipmask = compute_clipmask(vOut[j]->clip, draw->plane, draw->nr_planes); - - /* divide by w */ - w = 1.0f / w; - x *= w; - y *= w; - z *= w; - } - else { - vOut[j]->clipmask = 0; - } - vOut[j]->edgeflag = 1; - - if (!draw->identity_viewport) { - /* Viewport mapping */ - vOut[j]->data[0][0] = x * scale[0] + trans[0]; - vOut[j]->data[0][1] = y * scale[1] + trans[1]; - vOut[j]->data[0][2] = z * scale[2] + trans[2]; - vOut[j]->data[0][3] = w; - } - else { - vOut[j]->data[0][0] = x; - vOut[j]->data[0][1] = y; - vOut[j]->data[0][2] = z; - vOut[j]->data[0][3] = w; - } - - /* Remaining attributes are packed into sequential post-transform - * vertex attrib slots. - */ - for (slot = 1; slot < draw->num_vs_outputs; slot++) { - vOut[j]->data[slot][0] = machine->Outputs[slot].xyzw[0].f[j]; - vOut[j]->data[slot][1] = machine->Outputs[slot].xyzw[1].f[j]; - vOut[j]->data[slot][2] = machine->Outputs[slot].xyzw[2].f[j]; - vOut[j]->data[slot][3] = machine->Outputs[slot].xyzw[3].f[j]; - } - -#if 0 /*DEBUG*/ - printf("Post xform vert:\n"); - for (slot = 0; slot < draw->num_vs_outputs; slot++) { - printf("%d: %f %f %f %f\n", slot, - vOut[j]->data[slot][0], - vOut[j]->data[slot][1], - vOut[j]->data[slot][2], - vOut[j]->data[slot][3]); - } -#endif - - - } /* loop over vertices */ + for (j = 0; j < max_vertices; j++) { + for (slot = 0; slot < shader->info.num_outputs; slot++) { + output[slot][0] = machine->Outputs[slot].xyzw[0].f[j]; + output[slot][1] = machine->Outputs[slot].xyzw[1].f[j]; + output[slot][2] = machine->Outputs[slot].xyzw[2].f[j]; + output[slot][3] = machine->Outputs[slot].xyzw[3].f[j]; + + } + +#if 0 + debug_printf("%d) Post xform vert:\n", i + j); + for (slot = 0; slot < shader->info.num_outputs; slot++) { + debug_printf("\t%d: %f %f %f %f\n", slot, + output[slot][0], + output[slot][1], + output[slot][2], + output[slot][3]); + } +#endif + + output = (float (*)[4])((char *)output + output_stride); + } + + } } + static void vs_exec_delete( struct draw_vertex_shader *dvs ) { @@ -208,17 +165,24 @@ struct draw_vertex_shader * draw_create_vs_exec(struct draw_context *draw, const struct pipe_shader_state *state) { - struct draw_vertex_shader *vs = CALLOC_STRUCT( draw_vertex_shader ); - uint nt = tgsi_num_tokens(state->tokens); + struct exec_vertex_shader *vs = CALLOC_STRUCT( exec_vertex_shader ); if (vs == NULL) return NULL; /* we make a private copy of the tokens */ - vs->state.tokens = mem_dup(state->tokens, nt * sizeof(state->tokens[0])); - vs->prepare = vs_exec_prepare; - vs->run = vs_exec_run; - vs->delete = vs_exec_delete; + vs->base.state.tokens = tgsi_dup_tokens(state->tokens); + if (!vs->base.state.tokens) { + FREE(vs); + return NULL; + } + + tgsi_scan_shader(state->tokens, &vs->base.info); + + vs->base.prepare = vs_exec_prepare; + vs->base.run_linear = vs_exec_run_linear; + vs->base.delete = vs_exec_delete; + vs->machine = &draw->machine; - return vs; + return &vs->base; } diff --git a/src/gallium/auxiliary/draw/draw_vs_llvm.c b/src/gallium/auxiliary/draw/draw_vs_llvm.c index c8268317ef..171da51dd5 100644 --- a/src/gallium/auxiliary/draw/draw_vs_llvm.c +++ b/src/gallium/auxiliary/draw/draw_vs_llvm.c @@ -1,8 +1,8 @@ /************************************************************************** - * + * * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. * All Rights Reserved. - * + * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the * "Software"), to deal in the Software without restriction, including @@ -10,11 +10,11 @@ * distribute, sub license, and/or sell copies of the Software, and to * permit persons to whom the Software is furnished to do so, subject to * the following conditions: - * + * * The above copyright notice and this permission notice (including the * next paragraph) shall be included in all copies or substantial portions * of the Software. - * + * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. @@ -22,7 +22,7 @@ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * + * **************************************************************************/ /* @@ -47,154 +47,42 @@ struct draw_llvm_vertex_shader { struct draw_vertex_shader base; struct gallivm_prog *llvm_prog; + struct tgsi_exec_machine *machine; }; -static INLINE unsigned -compute_clipmask(const float *clip, /*const*/ float plane[][4], unsigned nr) -{ - unsigned mask = 0; - unsigned i; - - /* Do the hardwired planes first: - */ - if (-clip[0] + clip[3] < 0) mask |= CLIP_RIGHT_BIT; - if ( clip[0] + clip[3] < 0) mask |= CLIP_LEFT_BIT; - if (-clip[1] + clip[3] < 0) mask |= CLIP_TOP_BIT; - if ( clip[1] + clip[3] < 0) mask |= CLIP_BOTTOM_BIT; - if (-clip[2] + clip[3] < 0) mask |= CLIP_FAR_BIT; - if ( clip[2] + clip[3] < 0) mask |= CLIP_NEAR_BIT; - - /* Followed by any remaining ones: - */ - for (i = 6; i < nr; i++) { - if (dot4(clip, plane[i]) < 0) - mask |= (1<<i); - } - - return mask; -} - - - static void vs_llvm_prepare( struct draw_vertex_shader *base, struct draw_context *draw ) { - draw_update_vertex_fetch( draw ); } -/** - * Transform vertices with the current vertex program/shader - * Up to four vertices can be shaded at a time. - * \param vbuffer the input vertex data - * \param elts indexes of four input vertices - * \param count number of vertices to shade [1..4] - * \param vOut array of pointers to four output vertices - */ + static void -vs_llvm_run( struct draw_vertex_shader *base, - struct draw_context *draw, - const unsigned *elts, - unsigned count, - struct vertex_header *vOut[] ) +vs_llvm_run_linear( struct draw_vertex_shader *base, + const float (*input)[4], + float (*output)[4], + const float (*constants)[4], + unsigned count, + unsigned input_stride, + unsigned output_stride ) { - struct draw_llvm_vertex_shader *shader = + struct draw_llvm_vertex_shader *shader = (struct draw_llvm_vertex_shader *)base; - struct tgsi_exec_machine *machine = &draw->machine; - unsigned int j; - - ALIGN16_DECL(struct tgsi_exec_vector, inputs, PIPE_MAX_ATTRIBS); - ALIGN16_DECL(struct tgsi_exec_vector, outputs, PIPE_MAX_ATTRIBS); - const float *scale = draw->viewport.scale; - const float *trans = draw->viewport.translate; - - - assert(count <= 4); - assert(draw->vertex_shader->state->output_semantic_name[0] - == TGSI_SEMANTIC_POSITION); - - /* Consts does not require 16 byte alignment. */ - machine->Consts = (float (*)[4]) draw->user.constants; - - machine->Inputs = ALIGN16_ASSIGN(inputs); - if (draw->rasterizer->bypass_vs) { - /* outputs are just the inputs */ - machine->Outputs = machine->Inputs; - } - else { - machine->Outputs = ALIGN16_ASSIGN(outputs); - } - - - draw->vertex_fetch.fetch_func( draw, machine, elts, count ); + gallivm_cpu_vs_exec(shader->llvm_prog, shader->machine, + input, base->info.num_inputs, output, base->info.num_outputs, + constants, count, input_stride, output_stride); +} - if (!draw->rasterizer->bypass_vs) { - /* run shader */ - gallivm_cpu_vs_exec(shader->llvm_prog, - machine->Inputs, - machine->Outputs, - machine->Consts, - machine->Temps); - } - /* store machine results */ - for (j = 0; j < count; j++) { - unsigned slot; - float x, y, z, w; - - x = vOut[j]->clip[0] = machine->Outputs[0].xyzw[0].f[j]; - y = vOut[j]->clip[1] = machine->Outputs[0].xyzw[1].f[j]; - z = vOut[j]->clip[2] = machine->Outputs[0].xyzw[2].f[j]; - w = vOut[j]->clip[3] = machine->Outputs[0].xyzw[3].f[j]; - - if (!draw->rasterizer->bypass_clipping) { - vOut[j]->clipmask = compute_clipmask(vOut[j]->clip, draw->plane, draw->nr_planes); - - /* divide by w */ - w = 1.0f / w; - x *= w; - y *= w; - z *= w; - } - else { - vOut[j]->clipmask = 0; - } - vOut[j]->edgeflag = 1; - - if (!draw->identity_viewport) { - /* Viewport mapping */ - vOut[j]->data[0][0] = x * scale[0] + trans[0]; - vOut[j]->data[0][1] = y * scale[1] + trans[1]; - vOut[j]->data[0][2] = z * scale[2] + trans[2]; - vOut[j]->data[0][3] = w; - } - else { - vOut[j]->data[0][0] = x; - vOut[j]->data[0][1] = y; - vOut[j]->data[0][2] = z; - vOut[j]->data[0][3] = w; - } - - /* Remaining attributes are packed into sequential post-transform - * vertex attrib slots. - */ - for (slot = 1; slot < draw->num_vs_outputs; slot++) { - vOut[j]->data[slot][0] = machine->Outputs[slot].xyzw[0].f[j]; - vOut[j]->data[slot][1] = machine->Outputs[slot].xyzw[1].f[j]; - vOut[j]->data[slot][2] = machine->Outputs[slot].xyzw[2].f[j]; - vOut[j]->data[slot][3] = machine->Outputs[slot].xyzw[3].f[j]; - } - } /* loop over vertices */ -} static void vs_llvm_delete( struct draw_vertex_shader *base ) { - struct draw_llvm_vertex_shader *shader = + struct draw_llvm_vertex_shader *shader = (struct draw_llvm_vertex_shader *)base; /* Do something to free compiled shader: @@ -212,23 +100,30 @@ draw_create_vs_llvm(struct draw_context *draw, const struct pipe_shader_state *templ) { struct draw_llvm_vertex_shader *vs; - uint nt = tgsi_num_tokens(templ->tokens); vs = CALLOC_STRUCT( draw_llvm_vertex_shader ); - if (vs == NULL) + if (vs == NULL) return NULL; /* we make a private copy of the tokens */ - vs->base.state.tokens = mem_dup(templ->tokens, nt * sizeof(templ->tokens[0])); + vs->base.state.tokens = tgsi_dup_tokens(templ->tokens); + if (!vs->base.state.tokens) { + FREE(vs); + return NULL; + } + + tgsi_scan_shader(vs->base.state.tokens, &vs->base.info); + vs->base.prepare = vs_llvm_prepare; - vs->base.run = vs_llvm_run; + vs->base.run_linear = vs_llvm_run_linear; vs->base.delete = vs_llvm_delete; + vs->machine = &draw->machine; { struct gallivm_ir *ir = gallivm_ir_new(GALLIVM_VS); gallivm_ir_set_layout(ir, GALLIVM_SOA); gallivm_ir_set_components(ir, 4); - gallivm_ir_fill_from_tgsi(ir, vs->base.state->tokens); + gallivm_ir_fill_from_tgsi(ir, vs->base.state.tokens); vs->llvm_prog = gallivm_ir_compile(ir); gallivm_ir_delete(ir); } diff --git a/src/gallium/auxiliary/draw/draw_vs_sse.c b/src/gallium/auxiliary/draw/draw_vs_sse.c index f40d65df08..e3f4e67472 100644 --- a/src/gallium/auxiliary/draw/draw_vs_sse.c +++ b/src/gallium/auxiliary/draw/draw_vs_sse.c @@ -41,173 +41,136 @@ #include "draw_private.h" #include "draw_context.h" +#include "rtasm/rtasm_cpu.h" #include "rtasm/rtasm_x86sse.h" #include "tgsi/exec/tgsi_sse2.h" #include "tgsi/util/tgsi_parse.h" +#define SSE_MAX_VERTICES 4 +#define SSE_SWIZZLES 1 +#if SSE_SWIZZLES +typedef void (XSTDCALL *codegen_function) ( + const struct tgsi_exec_vector *input, /* 1 */ + struct tgsi_exec_vector *output, /* 2 */ + float (*constant)[4], /* 3 */ + struct tgsi_exec_vector *temporary, /* 4 */ + float (*immediates)[4], /* 5 */ + const float (*aos_input)[4], /* 6 */ + uint num_inputs, /* 7 */ + uint input_stride, /* 8 */ + float (*aos_output)[4], /* 9 */ + uint num_outputs, /* 10 */ + uint output_stride ); /* 11 */ +#else typedef void (XSTDCALL *codegen_function) ( const struct tgsi_exec_vector *input, struct tgsi_exec_vector *output, float (*constant)[4], - struct tgsi_exec_vector *temporary ); - + struct tgsi_exec_vector *temporary, + float (*immediates)[4] ); +#endif struct draw_sse_vertex_shader { struct draw_vertex_shader base; struct x86_function sse2_program; - codegen_function func; -}; + codegen_function func; + + struct tgsi_exec_machine *machine; -/* Should be part of the generated shader: - */ -static INLINE unsigned -compute_clipmask(const float *clip, /*const*/ float plane[][4], unsigned nr) -{ - unsigned mask = 0; - unsigned i; - - /* Do the hardwired planes first: - */ - if (-clip[0] + clip[3] < 0) mask |= CLIP_RIGHT_BIT; - if ( clip[0] + clip[3] < 0) mask |= CLIP_LEFT_BIT; - if (-clip[1] + clip[3] < 0) mask |= CLIP_TOP_BIT; - if ( clip[1] + clip[3] < 0) mask |= CLIP_BOTTOM_BIT; - if (-clip[2] + clip[3] < 0) mask |= CLIP_FAR_BIT; - if ( clip[2] + clip[3] < 0) mask |= CLIP_NEAR_BIT; - - /* Followed by any remaining ones: - */ - for (i = 6; i < nr; i++) { - if (dot4(clip, plane[i]) < 0) - mask |= (1<<i); - } - - return mask; -} + float immediates[TGSI_EXEC_NUM_IMMEDIATES][4]; +}; static void vs_sse_prepare( struct draw_vertex_shader *base, struct draw_context *draw ) { - draw_update_vertex_fetch( draw ); } -/** - * Transform vertices with the current vertex program/shader - * Up to four vertices can be shaded at a time. - * \param vbuffer the input vertex data - * \param elts indexes of four input vertices - * \param count number of vertices to shade [1..4] - * \param vOut array of pointers to four output vertices + + +/* Simplified vertex shader interface for the pt paths. Given the + * complexity of code-generating all the above operations together, + * it's time to try doing all the other stuff separately. */ static void -vs_sse_run( struct draw_vertex_shader *base, - struct draw_context *draw, - const unsigned *elts, - unsigned count, - struct vertex_header *vOut[] ) +vs_sse_run_linear( struct draw_vertex_shader *base, + const float (*input)[4], + float (*output)[4], + const float (*constants)[4], + unsigned count, + unsigned input_stride, + unsigned output_stride ) { struct draw_sse_vertex_shader *shader = (struct draw_sse_vertex_shader *)base; - struct tgsi_exec_machine *machine = &draw->machine; - unsigned int j; - - ALIGN16_DECL(struct tgsi_exec_vector, inputs, PIPE_MAX_ATTRIBS); - ALIGN16_DECL(struct tgsi_exec_vector, outputs, PIPE_MAX_ATTRIBS); - const float *scale = draw->viewport.scale; - const float *trans = draw->viewport.translate; - - assert(count <= 4); - assert(draw->vertex_shader->info.output_semantic_name[0] - == TGSI_SEMANTIC_POSITION); - - /* Consts does not require 16 byte alignment. */ - machine->Consts = (float (*)[4]) draw->user.constants; - machine->Inputs = ALIGN16_ASSIGN(inputs); - if (draw->rasterizer->bypass_vs) { - /* outputs are just the inputs */ - machine->Outputs = machine->Inputs; - } - else { - machine->Outputs = ALIGN16_ASSIGN(outputs); - } - - - /* Fetch vertices. This may at some point be integrated into the - * compiled shader -- that would require a reorganization where - * multiple versions of the compiled shader might exist, - * specialized for each fetch state. - */ - draw->vertex_fetch.fetch_func( draw, machine, elts, count ); + struct tgsi_exec_machine *machine = shader->machine; + unsigned int i; + for (i = 0; i < count; i += MAX_TGSI_VERTICES) { + unsigned int max_vertices = MIN2(MAX_TGSI_VERTICES, count - i); - if (!draw->rasterizer->bypass_vs) { +#if SSE_SWIZZLES /* run compiled shader - */ + */ shader->func(machine->Inputs, - machine->Outputs, - machine->Consts, - machine->Temps ); - } - - - /* XXX: Computing the clipmask and emitting results should be done - * in the vertex program as a set of instructions appended to - * the user-provided code. - */ - for (j = 0; j < count; j++) { - unsigned slot; - float x, y, z, w; - - x = vOut[j]->clip[0] = machine->Outputs[0].xyzw[0].f[j]; - y = vOut[j]->clip[1] = machine->Outputs[0].xyzw[1].f[j]; - z = vOut[j]->clip[2] = machine->Outputs[0].xyzw[2].f[j]; - w = vOut[j]->clip[3] = machine->Outputs[0].xyzw[3].f[j]; - - if (!draw->rasterizer->bypass_clipping) { - vOut[j]->clipmask = compute_clipmask(vOut[j]->clip, draw->plane, draw->nr_planes); + machine->Outputs, + (float (*)[4])constants, + machine->Temps, + shader->immediates, + input, + base->info.num_inputs, + input_stride, + output, + base->info.num_outputs, + output_stride ); + + input = (const float (*)[4])((const char *)input + input_stride * max_vertices); + output = (float (*)[4])((char *)output + output_stride * max_vertices); +#else + unsigned int j, slot; - /* divide by w */ - w = 1.0f / w; - x *= w; - y *= w; - z *= w; - } - else { - vOut[j]->clipmask = 0; - } - vOut[j]->edgeflag = 1; - - if (!draw->identity_viewport) { - /* Viewport mapping */ - vOut[j]->data[0][0] = x * scale[0] + trans[0]; - vOut[j]->data[0][1] = y * scale[1] + trans[1]; - vOut[j]->data[0][2] = z * scale[2] + trans[2]; - vOut[j]->data[0][3] = w; - } - else { - vOut[j]->data[0][0] = x; - vOut[j]->data[0][1] = y; - vOut[j]->data[0][2] = z; - vOut[j]->data[0][3] = w; + /* Swizzle inputs. + */ + for (j = 0; j < max_vertices; j++) { + for (slot = 0; slot < base->info.num_inputs; slot++) { + machine->Inputs[slot].xyzw[0].f[j] = input[slot][0]; + machine->Inputs[slot].xyzw[1].f[j] = input[slot][1]; + machine->Inputs[slot].xyzw[2].f[j] = input[slot][2]; + machine->Inputs[slot].xyzw[3].f[j] = input[slot][3]; + } + + input = (const float (*)[4])((const char *)input + input_stride); } - /* Remaining attributes are packed into sequential post-transform - * vertex attrib slots. + /* run compiled shader */ - for (slot = 1; slot < draw->num_vs_outputs; slot++) { - vOut[j]->data[slot][0] = machine->Outputs[slot].xyzw[0].f[j]; - vOut[j]->data[slot][1] = machine->Outputs[slot].xyzw[1].f[j]; - vOut[j]->data[slot][2] = machine->Outputs[slot].xyzw[2].f[j]; - vOut[j]->data[slot][3] = machine->Outputs[slot].xyzw[3].f[j]; + shader->func(machine->Inputs, + machine->Outputs, + (float (*)[4])constants, + machine->Temps, + shader->immediates); + + /* Unswizzle all output results. + */ + for (j = 0; j < max_vertices; j++) { + for (slot = 0; slot < base->info.num_outputs; slot++) { + output[slot][0] = machine->Outputs[slot].xyzw[0].f[j]; + output[slot][1] = machine->Outputs[slot].xyzw[1].f[j]; + output[slot][2] = machine->Outputs[slot].xyzw[2].f[j]; + output[slot][3] = machine->Outputs[slot].xyzw[3].f[j]; + } + + output = (float (*)[4])((char *)output + output_stride); } - } +#endif + } } + static void vs_sse_delete( struct draw_vertex_shader *base ) { @@ -225,9 +188,8 @@ draw_create_vs_sse(struct draw_context *draw, const struct pipe_shader_state *templ) { struct draw_sse_vertex_shader *vs; - uint nt = tgsi_num_tokens(templ->tokens); - if (!draw->use_sse) + if (!rtasm_cpu_has_sse2()) return NULL; vs = CALLOC_STRUCT( draw_sse_vertex_shader ); @@ -235,18 +197,27 @@ draw_create_vs_sse(struct draw_context *draw, return NULL; /* we make a private copy of the tokens */ - vs->base.state.tokens = mem_dup(templ->tokens, nt * sizeof(templ->tokens[0])); + vs->base.state.tokens = tgsi_dup_tokens(templ->tokens); + if (!vs->base.state.tokens) + goto fail; + + tgsi_scan_shader(templ->tokens, &vs->base.info); + vs->base.prepare = vs_sse_prepare; - vs->base.run = vs_sse_run; + vs->base.run_linear = vs_sse_run_linear; vs->base.delete = vs_sse_delete; + vs->machine = &draw->machine; x86_init_func( &vs->sse2_program ); if (!tgsi_emit_sse2( (struct tgsi_token *) vs->base.state.tokens, - &vs->sse2_program )) + &vs->sse2_program, vs->immediates, SSE_SWIZZLES )) goto fail; vs->func = (codegen_function) x86_get_func( &vs->sse2_program ); + if (!vs->func) { + goto fail; + } return &vs->base; diff --git a/src/gallium/auxiliary/draw/draw_wide_prims.c b/src/gallium/auxiliary/draw/draw_wide_prims.c deleted file mode 100644 index d6bff110b4..0000000000 --- a/src/gallium/auxiliary/draw/draw_wide_prims.c +++ /dev/null @@ -1,366 +0,0 @@ -/************************************************************************** - * - * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - -/* Authors: Keith Whitwell <keith@tungstengraphics.com> - */ - -#include "pipe/p_util.h" -#include "pipe/p_defines.h" -#include "pipe/p_shader_tokens.h" -#include "draw_private.h" - - -struct wide_stage { - struct draw_stage stage; - - float half_line_width; - float half_point_size; - - uint texcoord_slot[PIPE_MAX_SHADER_OUTPUTS]; - uint texcoord_mode[PIPE_MAX_SHADER_OUTPUTS]; - uint num_texcoords; - - int psize_slot; -}; - - - -static INLINE struct wide_stage *wide_stage( struct draw_stage *stage ) -{ - return (struct wide_stage *)stage; -} - - -static void passthrough_point( struct draw_stage *stage, - struct prim_header *header ) -{ - stage->next->point( stage->next, header ); -} - -static void passthrough_line( struct draw_stage *stage, - struct prim_header *header ) -{ - stage->next->line(stage->next, header); -} - -static void passthrough_tri( struct draw_stage *stage, - struct prim_header *header ) -{ - stage->next->tri(stage->next, header); -} - - -/** - * Draw a wide line by drawing a quad (two triangles). - * XXX need to disable polygon stipple. - */ -static void wide_line( struct draw_stage *stage, - struct prim_header *header ) -{ - const struct wide_stage *wide = wide_stage(stage); - const float half_width = wide->half_line_width; - - struct prim_header tri; - - struct vertex_header *v0 = dup_vert(stage, header->v[0], 0); - struct vertex_header *v1 = dup_vert(stage, header->v[0], 1); - struct vertex_header *v2 = dup_vert(stage, header->v[1], 2); - struct vertex_header *v3 = dup_vert(stage, header->v[1], 3); - - float *pos0 = v0->data[0]; - float *pos1 = v1->data[0]; - float *pos2 = v2->data[0]; - float *pos3 = v3->data[0]; - - const float dx = FABSF(pos0[0] - pos2[0]); - const float dy = FABSF(pos0[1] - pos2[1]); - - /* - * Draw wide line as a quad (two tris) by "stretching" the line along - * X or Y. - * We need to tweak coords in several ways to be conformant here. - */ - - if (dx > dy) { - /* x-major line */ - pos0[1] = pos0[1] - half_width - 0.25f; - pos1[1] = pos1[1] + half_width - 0.25f; - pos2[1] = pos2[1] - half_width - 0.25f; - pos3[1] = pos3[1] + half_width - 0.25f; - if (pos0[0] < pos2[0]) { - /* left to right line */ - pos0[0] -= 0.5f; - pos1[0] -= 0.5f; - pos2[0] -= 0.5f; - pos3[0] -= 0.5f; - } - else { - /* right to left line */ - pos0[0] += 0.5f; - pos1[0] += 0.5f; - pos2[0] += 0.5f; - pos3[0] += 0.5f; - } - } - else { - /* y-major line */ - pos0[0] = pos0[0] - half_width + 0.25f; - pos1[0] = pos1[0] + half_width + 0.25f; - pos2[0] = pos2[0] - half_width + 0.25f; - pos3[0] = pos3[0] + half_width + 0.25f; - if (pos0[1] < pos2[1]) { - /* top to bottom line */ - pos0[1] -= 0.5f; - pos1[1] -= 0.5f; - pos2[1] -= 0.5f; - pos3[1] -= 0.5f; - } - else { - /* bottom to top line */ - pos0[1] += 0.5f; - pos1[1] += 0.5f; - pos2[1] += 0.5f; - pos3[1] += 0.5f; - } - } - - tri.det = header->det; /* only the sign matters */ - tri.v[0] = v0; - tri.v[1] = v2; - tri.v[2] = v3; - stage->next->tri( stage->next, &tri ); - - tri.v[0] = v0; - tri.v[1] = v3; - tri.v[2] = v1; - stage->next->tri( stage->next, &tri ); -} - - -/** - * Set the vertex texcoords for sprite mode. - * Coords may be left untouched or set to a right-side-up or upside-down - * orientation. - */ -static void set_texcoords(const struct wide_stage *wide, - struct vertex_header *v, const float tc[4]) -{ - uint i; - for (i = 0; i < wide->num_texcoords; i++) { - if (wide->texcoord_mode[i] != PIPE_SPRITE_COORD_NONE) { - uint j = wide->texcoord_slot[i]; - v->data[j][0] = tc[0]; - if (wide->texcoord_mode[i] == PIPE_SPRITE_COORD_LOWER_LEFT) - v->data[j][1] = 1.0f - tc[1]; - else - v->data[j][1] = tc[1]; - v->data[j][2] = tc[2]; - v->data[j][3] = tc[3]; - } - } -} - - -/* If there are lots of sprite points (and why wouldn't there be?) it - * would probably be more sensible to change hardware setup to - * optimize this rather than doing the whole thing in software like - * this. - */ -static void wide_point( struct draw_stage *stage, - struct prim_header *header ) -{ - const struct wide_stage *wide = wide_stage(stage); - const boolean sprite = (boolean) stage->draw->rasterizer->point_sprite; - float half_size; - float left_adj, right_adj; - - struct prim_header tri; - - /* four dups of original vertex */ - struct vertex_header *v0 = dup_vert(stage, header->v[0], 0); - struct vertex_header *v1 = dup_vert(stage, header->v[0], 1); - struct vertex_header *v2 = dup_vert(stage, header->v[0], 2); - struct vertex_header *v3 = dup_vert(stage, header->v[0], 3); - - float *pos0 = v0->data[0]; - float *pos1 = v1->data[0]; - float *pos2 = v2->data[0]; - float *pos3 = v3->data[0]; - - /* point size is either per-vertex or fixed size */ - if (wide->psize_slot >= 0) { - half_size = 0.5f * header->v[0]->data[wide->psize_slot][0]; - } - else { - half_size = wide->half_point_size; - } - - left_adj = -half_size; /* + 0.25f;*/ - right_adj = half_size; /* + 0.25f;*/ - - pos0[0] += left_adj; - pos0[1] -= half_size; - - pos1[0] += left_adj; - pos1[1] += half_size; - - pos2[0] += right_adj; - pos2[1] -= half_size; - - pos3[0] += right_adj; - pos3[1] += half_size; - - if (sprite) { - static const float tex00[4] = { 0, 0, 0, 1 }; - static const float tex01[4] = { 0, 1, 0, 1 }; - static const float tex11[4] = { 1, 1, 0, 1 }; - static const float tex10[4] = { 1, 0, 0, 1 }; - set_texcoords( wide, v0, tex00 ); - set_texcoords( wide, v1, tex01 ); - set_texcoords( wide, v2, tex10 ); - set_texcoords( wide, v3, tex11 ); - } - - tri.det = header->det; /* only the sign matters */ - tri.v[0] = v0; - tri.v[1] = v2; - tri.v[2] = v3; - stage->next->tri( stage->next, &tri ); - - tri.v[0] = v0; - tri.v[1] = v3; - tri.v[2] = v1; - stage->next->tri( stage->next, &tri ); -} - - -static void wide_first_point( struct draw_stage *stage, - struct prim_header *header ) -{ - struct wide_stage *wide = wide_stage(stage); - struct draw_context *draw = stage->draw; - - wide->half_point_size = 0.5f * draw->rasterizer->point_size; - - /* XXX we won't know the real size if it's computed by the vertex shader! */ - if (draw->rasterizer->point_size > draw->wide_point_threshold) { - stage->point = wide_point; - } - else { - stage->point = passthrough_point; - } - - if (draw->rasterizer->point_sprite) { - /* find vertex shader texcoord outputs */ - const struct draw_vertex_shader *vs = draw->vertex_shader; - uint i, j = 0; - for (i = 0; i < vs->info.num_outputs; i++) { - if (vs->info.output_semantic_name[i] == TGSI_SEMANTIC_GENERIC) { - wide->texcoord_slot[j] = i; - wide->texcoord_mode[j] = draw->rasterizer->sprite_coord_mode[j]; - j++; - } - } - wide->num_texcoords = j; - } - - wide->psize_slot = -1; - - if (draw->rasterizer->point_size_per_vertex) { - /* find PSIZ vertex output */ - const struct draw_vertex_shader *vs = draw->vertex_shader; - uint i; - for (i = 0; i < vs->info.num_outputs; i++) { - if (vs->info.output_semantic_name[i] == TGSI_SEMANTIC_PSIZE) { - wide->psize_slot = i; - break; - } - } - } - - stage->point( stage, header ); -} - - - -static void wide_first_line( struct draw_stage *stage, - struct prim_header *header ) -{ - struct wide_stage *wide = wide_stage(stage); - struct draw_context *draw = stage->draw; - - wide->half_line_width = 0.5f * draw->rasterizer->line_width; - - if (draw->rasterizer->line_width != 1.0) { - wide->stage.line = wide_line; - } - else { - wide->stage.line = passthrough_line; - } - - stage->line( stage, header ); -} - - -static void wide_flush( struct draw_stage *stage, unsigned flags ) -{ - stage->line = wide_first_line; - stage->point = wide_first_point; - stage->next->flush( stage->next, flags ); -} - - -static void wide_reset_stipple_counter( struct draw_stage *stage ) -{ - stage->next->reset_stipple_counter( stage->next ); -} - - -static void wide_destroy( struct draw_stage *stage ) -{ - draw_free_temp_verts( stage ); - FREE( stage ); -} - - -struct draw_stage *draw_wide_stage( struct draw_context *draw ) -{ - struct wide_stage *wide = CALLOC_STRUCT(wide_stage); - - draw_alloc_temp_verts( &wide->stage, 4 ); - - wide->stage.draw = draw; - wide->stage.next = NULL; - wide->stage.point = wide_first_point; - wide->stage.line = wide_first_line; - wide->stage.tri = passthrough_tri; - wide->stage.flush = wide_flush; - wide->stage.reset_stipple_counter = wide_reset_stipple_counter; - wide->stage.destroy = wide_destroy; - - return &wide->stage; -} diff --git a/src/gallium/auxiliary/gallivm/Makefile b/src/gallium/auxiliary/gallivm/Makefile index c24e19e062..c3f7bfba93 100644 --- a/src/gallium/auxiliary/gallivm/Makefile +++ b/src/gallium/auxiliary/gallivm/Makefile @@ -65,10 +65,14 @@ depend: $(C_SOURCES) $(CPP_SOURCES) $(ASM_SOURCES) $(INC_SOURCES) gallivm_builtins.cpp: llvm_builtins.c - clang --emit-llvm < $< |llvm-as|opt -std-compile-opts|llvm2cpp -gen-contents -o=$@ -f -for=shader -funcname=createGallivmBuiltins + clang --emit-llvm < $< |llvm-as|opt -std-compile-opts > temp1.bin + (echo "static const unsigned char llvm_builtins_data[] = {"; od -txC temp1.bin | sed -e "s/^[0-9]*//" -e s"/ \([0-9a-f][0-9a-f]\)/0x\1,/g" -e"\$$d" | sed -e"\$$s/,$$/};/") >$@ + rm temp1.bin gallivmsoabuiltins.cpp: soabuiltins.c - clang --emit-llvm < $< |llvm-as|opt -std-compile-opts|llvm2cpp -gen-module -o=$@ -f -for=shader -funcname=createSoaBuiltins + clang --emit-llvm < $< |llvm-as|opt -std-compile-opts > temp2.bin + (echo "static const unsigned char soabuiltins_data[] = {"; od -txC temp2.bin | sed -e "s/^[0-9]*//" -e s"/ \([0-9a-f][0-9a-f]\)/0x\1,/g" -e"\$$d" | sed -e"\$$s/,$$/};/") >$@ + rm temp2.bin # Emacs tags tags: diff --git a/src/gallium/auxiliary/gallivm/gallivm.cpp b/src/gallium/auxiliary/gallivm/gallivm.cpp index b6f641a3f8..77900e342b 100644 --- a/src/gallium/auxiliary/gallivm/gallivm.cpp +++ b/src/gallium/auxiliary/gallivm/gallivm.cpp @@ -288,10 +288,7 @@ void gallivm_ir_fill_from_tgsi(struct gallivm_ir *ir, std::cout << "Creating llvm from: " <<std::endl; tgsi_dump(tokens, 0); - llvm::Module *mod = tgsi_to_llvmir(ir, tokens); - - //llvm::Module *mod = tgsi_to_llvm(ir, tokens); ir->module = mod; gallivm_ir_dump(ir, 0); } @@ -306,11 +303,11 @@ struct gallivm_prog * gallivm_ir_compile(struct gallivm_ir *ir) { struct gallivm_prog *prog = (struct gallivm_prog *)calloc(1, sizeof(struct gallivm_prog)); - + std::cout << "Before optimizations:"<<std::endl; ir->module->dump(); std::cout<<"-------------------------------"<<std::endl; - + PassManager veri; veri.add(createVerifierPass()); veri.run(*ir->module); @@ -318,7 +315,7 @@ struct gallivm_prog * gallivm_ir_compile(struct gallivm_ir *ir) prog->num_consts = ir->num_consts; memcpy(prog->interpolators, ir->interpolators, sizeof(prog->interpolators)); prog->num_interp = ir->num_interp; - + /* Run optimization passes over it */ PassManager passes; passes.add(new TargetData(mod)); diff --git a/src/gallium/auxiliary/gallivm/gallivm.h b/src/gallium/auxiliary/gallivm/gallivm.h index b4d6555d2f..36a64a7747 100644 --- a/src/gallium/auxiliary/gallivm/gallivm.h +++ b/src/gallium/auxiliary/gallivm/gallivm.h @@ -90,10 +90,15 @@ void gallivm_prog_dump(struct gallivm_prog *prog, const char *file_prefix); struct gallivm_cpu_engine *gallivm_cpu_engine_create(struct gallivm_prog *prog); struct gallivm_cpu_engine *gallivm_global_cpu_engine(); int gallivm_cpu_vs_exec(struct gallivm_prog *prog, - struct tgsi_exec_vector *inputs, - struct tgsi_exec_vector *dests, - float (*consts)[4], - struct tgsi_exec_vector *temps); + struct tgsi_exec_machine *machine, + const float (*input)[4], + unsigned num_inputs, + float (*output)[4], + unsigned num_outputs, + const float (*constants)[4], + unsigned count, + unsigned input_stride, + unsigned output_stride); int gallivm_cpu_fs_exec(struct gallivm_prog *prog, float x, float y, float (*dests)[PIPE_MAX_SHADER_INPUTS][4], diff --git a/src/gallium/auxiliary/gallivm/gallivm_builtins.cpp b/src/gallium/auxiliary/gallivm/gallivm_builtins.cpp index 1796f0a177..0fc5c4ec5c 100644 --- a/src/gallium/auxiliary/gallivm/gallivm_builtins.cpp +++ b/src/gallium/auxiliary/gallivm/gallivm_builtins.cpp @@ -1,567 +1,140 @@ -// Generated by llvm2cpp - DO NOT MODIFY! - - -Module* createGallivmBuiltins(Module *mod) { - -mod->setModuleIdentifier("shader"); - -// Type Definitions -ArrayType* ArrayTy_0 = ArrayType::get(IntegerType::get(8), 25); - -PointerType* PointerTy_1 = PointerType::get(ArrayTy_0, 0); - -std::vector<const Type*>FuncTy_2_args; -FuncTy_2_args.push_back(Type::FloatTy); -FuncTy_2_args.push_back(Type::FloatTy); -FunctionType* FuncTy_2 = FunctionType::get( - /*Result=*/Type::FloatTy, - /*Params=*/FuncTy_2_args, - /*isVarArg=*/false); - -PointerType* PointerTy_3 = PointerType::get(FuncTy_2, 0); - -VectorType* VectorTy_4 = VectorType::get(Type::FloatTy, 4); - -std::vector<const Type*>FuncTy_5_args; -FuncTy_5_args.push_back(VectorTy_4); -FunctionType* FuncTy_5 = FunctionType::get( - /*Result=*/VectorTy_4, - /*Params=*/FuncTy_5_args, - /*isVarArg=*/false); - -std::vector<const Type*>FuncTy_6_args; -FuncTy_6_args.push_back(VectorTy_4); -FuncTy_6_args.push_back(VectorTy_4); -FuncTy_6_args.push_back(VectorTy_4); -FunctionType* FuncTy_6 = FunctionType::get( - /*Result=*/VectorTy_4, - /*Params=*/FuncTy_6_args, - /*isVarArg=*/false); - -VectorType* VectorTy_7 = VectorType::get(IntegerType::get(32), 4); - -std::vector<const Type*>FuncTy_9_args; -FunctionType* FuncTy_9 = FunctionType::get( - /*Result=*/IntegerType::get(32), - /*Params=*/FuncTy_9_args, - /*isVarArg=*/true); - -PointerType* PointerTy_8 = PointerType::get(FuncTy_9, 0); - -PointerType* PointerTy_10 = PointerType::get(IntegerType::get(8), 0); - -std::vector<const Type*>FuncTy_12_args; -FuncTy_12_args.push_back(Type::FloatTy); -FunctionType* FuncTy_12 = FunctionType::get( - /*Result=*/Type::FloatTy, - /*Params=*/FuncTy_12_args, - /*isVarArg=*/false); - -PointerType* PointerTy_11 = PointerType::get(FuncTy_12, 0); - -std::vector<const Type*>FuncTy_13_args; -FuncTy_13_args.push_back(VectorTy_4); -FunctionType* FuncTy_13 = FunctionType::get( - /*Result=*/IntegerType::get(32), - /*Params=*/FuncTy_13_args, - /*isVarArg=*/false); - - -// Function Declarations - -Function* func_approx = new Function( - /*Type=*/FuncTy_2, - /*Linkage=*/GlobalValue::WeakLinkage, - /*Name=*/"approx", mod); -func_approx->setCallingConv(CallingConv::C); -const ParamAttrsList *func_approx_PAL = 0; -func_approx->setParamAttrs(func_approx_PAL); - -Function* func_powf = new Function( - /*Type=*/FuncTy_2, - /*Linkage=*/GlobalValue::ExternalLinkage, - /*Name=*/"powf", mod); // (external, no body) -func_powf->setCallingConv(CallingConv::C); -const ParamAttrsList *func_powf_PAL = 0; -func_powf->setParamAttrs(func_powf_PAL); - -Function* func_lit = new Function( - /*Type=*/FuncTy_5, - /*Linkage=*/GlobalValue::WeakLinkage, - /*Name=*/"lit", mod); -func_lit->setCallingConv(CallingConv::C); -const ParamAttrsList *func_lit_PAL = 0; -func_lit->setParamAttrs(func_lit_PAL); - -Function* func_cmp = new Function( - /*Type=*/FuncTy_6, - /*Linkage=*/GlobalValue::WeakLinkage, - /*Name=*/"cmp", mod); -func_cmp->setCallingConv(CallingConv::C); -const ParamAttrsList *func_cmp_PAL = 0; -{ - ParamAttrsVector Attrs; - ParamAttrsWithIndex PAWI; - PAWI.index = 0; PAWI.attrs = 0 | ParamAttr::NoUnwind; - Attrs.push_back(PAWI); - func_cmp_PAL = ParamAttrsList::get(Attrs); - -} -func_cmp->setParamAttrs(func_cmp_PAL); - -Function* func_vcos = new Function( - /*Type=*/FuncTy_5, - /*Linkage=*/GlobalValue::WeakLinkage, - /*Name=*/"vcos", mod); -func_vcos->setCallingConv(CallingConv::C); -const ParamAttrsList *func_vcos_PAL = 0; -func_vcos->setParamAttrs(func_vcos_PAL); - -Function* func_printf = new Function( - /*Type=*/FuncTy_9, - /*Linkage=*/GlobalValue::ExternalLinkage, - /*Name=*/"printf", mod); // (external, no body) -func_printf->setCallingConv(CallingConv::C); -const ParamAttrsList *func_printf_PAL = 0; -func_printf->setParamAttrs(func_printf_PAL); - -Function* func_cosf = new Function( - /*Type=*/FuncTy_12, - /*Linkage=*/GlobalValue::ExternalLinkage, - /*Name=*/"cosf", mod); // (external, no body) -func_cosf->setCallingConv(CallingConv::C); -const ParamAttrsList *func_cosf_PAL = 0; -func_cosf->setParamAttrs(func_cosf_PAL); - -Function* func_scs = new Function( - /*Type=*/FuncTy_5, - /*Linkage=*/GlobalValue::WeakLinkage, - /*Name=*/"scs", mod); -func_scs->setCallingConv(CallingConv::C); -const ParamAttrsList *func_scs_PAL = 0; -func_scs->setParamAttrs(func_scs_PAL); - -Function* func_sinf = new Function( - /*Type=*/FuncTy_12, - /*Linkage=*/GlobalValue::ExternalLinkage, - /*Name=*/"sinf", mod); // (external, no body) -func_sinf->setCallingConv(CallingConv::C); -const ParamAttrsList *func_sinf_PAL = 0; -func_sinf->setParamAttrs(func_sinf_PAL); - -Function* func_vsin = new Function( - /*Type=*/FuncTy_5, - /*Linkage=*/GlobalValue::WeakLinkage, - /*Name=*/"vsin", mod); -func_vsin->setCallingConv(CallingConv::C); -const ParamAttrsList *func_vsin_PAL = 0; -func_vsin->setParamAttrs(func_vsin_PAL); - -Function* func_kilp = new Function( - /*Type=*/FuncTy_13, - /*Linkage=*/GlobalValue::WeakLinkage, - /*Name=*/"kilp", mod); -func_kilp->setCallingConv(CallingConv::C); -const ParamAttrsList *func_kilp_PAL = 0; -{ - ParamAttrsVector Attrs; - ParamAttrsWithIndex PAWI; - PAWI.index = 0; PAWI.attrs = 0 | ParamAttr::NoUnwind; - Attrs.push_back(PAWI); - func_kilp_PAL = ParamAttrsList::get(Attrs); - -} -func_kilp->setParamAttrs(func_kilp_PAL); - -// Global Variable Declarations - - -GlobalVariable* gvar_array__str = new GlobalVariable( -/*Type=*/ArrayTy_0, -/*isConstant=*/true, -/*Linkage=*/GlobalValue::InternalLinkage, -/*Initializer=*/0, // has initializer, specified below -/*Name=*/".str", -mod); - -GlobalVariable* gvar_array__str1 = new GlobalVariable( -/*Type=*/ArrayTy_0, -/*isConstant=*/true, -/*Linkage=*/GlobalValue::InternalLinkage, -/*Initializer=*/0, // has initializer, specified below -/*Name=*/".str1", -mod); - -// Constant Definitions -Constant* const_array_14 = ConstantArray::get("VEC IN is %f %f %f %f\x0A", true); -Constant* const_array_15 = ConstantArray::get("VEC OUT is %f %f %f %f\x0A", true); -ConstantFP* const_float_16 = ConstantFP::get(Type::FloatTy, APFloat(-1.280000e+02f)); -ConstantFP* const_float_17 = ConstantFP::get(Type::FloatTy, APFloat(1.280000e+02f)); -Constant* const_float_18 = Constant::getNullValue(Type::FloatTy); -Constant* const_int32_19 = Constant::getNullValue(IntegerType::get(32)); -std::vector<Constant*> const_packed_20_elems; -ConstantFP* const_float_21 = ConstantFP::get(Type::FloatTy, APFloat(1.000000e+00f)); -const_packed_20_elems.push_back(const_float_21); -UndefValue* const_float_22 = UndefValue::get(Type::FloatTy); -const_packed_20_elems.push_back(const_float_22); -const_packed_20_elems.push_back(const_float_22); -const_packed_20_elems.push_back(const_float_21); -Constant* const_packed_20 = ConstantVector::get(VectorTy_4, const_packed_20_elems); -ConstantInt* const_int32_23 = ConstantInt::get(APInt(32, "1", 10)); -ConstantInt* const_int32_24 = ConstantInt::get(APInt(32, "3", 10)); -ConstantInt* const_int32_25 = ConstantInt::get(APInt(32, "2", 10)); -std::vector<Constant*> const_packed_26_elems; -const_packed_26_elems.push_back(const_float_21); -const_packed_26_elems.push_back(const_float_18); -const_packed_26_elems.push_back(const_float_18); -const_packed_26_elems.push_back(const_float_21); -Constant* const_packed_26 = ConstantVector::get(VectorTy_4, const_packed_26_elems); -Constant* const_double_27 = Constant::getNullValue(Type::DoubleTy); -std::vector<Constant*> const_packed_28_elems; -const_packed_28_elems.push_back(const_int32_19); -ConstantInt* const_int32_29 = ConstantInt::get(APInt(32, "5", 10)); -const_packed_28_elems.push_back(const_int32_29); -const_packed_28_elems.push_back(const_int32_25); -const_packed_28_elems.push_back(const_int32_24); -Constant* const_packed_28 = ConstantVector::get(VectorTy_7, const_packed_28_elems); -std::vector<Constant*> const_packed_30_elems; -const_packed_30_elems.push_back(const_int32_19); -const_packed_30_elems.push_back(const_int32_23); -ConstantInt* const_int32_31 = ConstantInt::get(APInt(32, "6", 10)); -const_packed_30_elems.push_back(const_int32_31); -const_packed_30_elems.push_back(const_int32_24); -Constant* const_packed_30 = ConstantVector::get(VectorTy_7, const_packed_30_elems); -std::vector<Constant*> const_packed_32_elems; -const_packed_32_elems.push_back(const_int32_19); -const_packed_32_elems.push_back(const_int32_23); -const_packed_32_elems.push_back(const_int32_25); -ConstantInt* const_int32_33 = ConstantInt::get(APInt(32, "7", 10)); -const_packed_32_elems.push_back(const_int32_33); -Constant* const_packed_32 = ConstantVector::get(VectorTy_7, const_packed_32_elems); -std::vector<Constant*> const_ptr_34_indices; -const_ptr_34_indices.push_back(const_int32_19); -const_ptr_34_indices.push_back(const_int32_19); -Constant* const_ptr_34 = ConstantExpr::getGetElementPtr(gvar_array__str, &const_ptr_34_indices[0], const_ptr_34_indices.size() ); -UndefValue* const_packed_35 = UndefValue::get(VectorTy_4); -std::vector<Constant*> const_ptr_36_indices; -const_ptr_36_indices.push_back(const_int32_19); -const_ptr_36_indices.push_back(const_int32_19); -Constant* const_ptr_36 = ConstantExpr::getGetElementPtr(gvar_array__str1, &const_ptr_36_indices[0], const_ptr_36_indices.size() ); - -// Global Variable Definitions -gvar_array__str->setInitializer(const_array_14); -gvar_array__str1->setInitializer(const_array_15); - -// Function Definitions - -// Function: approx (func_approx) -{ - Function::arg_iterator args = func_approx->arg_begin(); - Value* float_a = args++; - float_a->setName("a"); - Value* float_b = args++; - float_b->setName("b"); - - BasicBlock* label_entry = new BasicBlock("entry",func_approx,0); - - // Block entry (label_entry) - FCmpInst* int1_cmp = new FCmpInst(FCmpInst::FCMP_OLT, float_b, const_float_16, "cmp", label_entry); - SelectInst* float_b_addr_0 = new SelectInst(int1_cmp, const_float_16, float_b, "b.addr.0", label_entry); - FCmpInst* int1_cmp3 = new FCmpInst(FCmpInst::FCMP_OGT, float_b_addr_0, const_float_17, "cmp3", label_entry); - SelectInst* float_b_addr_1 = new SelectInst(int1_cmp3, const_float_17, float_b_addr_0, "b.addr.1", label_entry); - FCmpInst* int1_cmp7 = new FCmpInst(FCmpInst::FCMP_OLT, float_a, const_float_18, "cmp7", label_entry); - SelectInst* float_a_addr_0 = new SelectInst(int1_cmp7, const_float_18, float_a, "a.addr.0", label_entry); - std::vector<Value*> float_call_params; - float_call_params.push_back(float_a_addr_0); - float_call_params.push_back(float_b_addr_1); - CallInst* float_call = new CallInst(func_powf, float_call_params.begin(), float_call_params.end(), "call", label_entry); - float_call->setCallingConv(CallingConv::C); - float_call->setTailCall(true);const ParamAttrsList *float_call_PAL = 0; - float_call->setParamAttrs(float_call_PAL); - - new ReturnInst(float_call, label_entry); - -} - -// Function: lit (func_lit) -{ - Function::arg_iterator args = func_lit->arg_begin(); - Value* packed_tmp = args++; - packed_tmp->setName("tmp"); - - BasicBlock* label_entry_38 = new BasicBlock("entry",func_lit,0); - BasicBlock* label_ifthen = new BasicBlock("ifthen",func_lit,0); - BasicBlock* label_UnifiedReturnBlock = new BasicBlock("UnifiedReturnBlock",func_lit,0); - - // Block entry (label_entry_38) - ExtractElementInst* float_tmp6 = new ExtractElementInst(packed_tmp, const_int32_19, "tmp6", label_entry_38); - FCmpInst* int1_cmp_39 = new FCmpInst(FCmpInst::FCMP_OGT, float_tmp6, const_float_18, "cmp", label_entry_38); - new BranchInst(label_ifthen, label_UnifiedReturnBlock, int1_cmp_39, label_entry_38); - - // Block ifthen (label_ifthen) - InsertElementInst* packed_tmp10 = new InsertElementInst(const_packed_20, float_tmp6, const_int32_23, "tmp10", label_ifthen); - ExtractElementInst* float_tmp12 = new ExtractElementInst(packed_tmp, const_int32_23, "tmp12", label_ifthen); - ExtractElementInst* float_tmp14 = new ExtractElementInst(packed_tmp, const_int32_24, "tmp14", label_ifthen); - std::vector<Value*> float_call_41_params; - float_call_41_params.push_back(float_tmp12); - float_call_41_params.push_back(float_tmp14); - CallInst* float_call_41 = new CallInst(func_approx, float_call_41_params.begin(), float_call_41_params.end(), "call", label_ifthen); - float_call_41->setCallingConv(CallingConv::C); - float_call_41->setTailCall(true);const ParamAttrsList *float_call_41_PAL = 0; - float_call_41->setParamAttrs(float_call_41_PAL); - - InsertElementInst* packed_tmp16 = new InsertElementInst(packed_tmp10, float_call_41, const_int32_25, "tmp16", label_ifthen); - new ReturnInst(packed_tmp16, label_ifthen); - - // Block UnifiedReturnBlock (label_UnifiedReturnBlock) - new ReturnInst(const_packed_26, label_UnifiedReturnBlock); - -} - -// Function: cmp (func_cmp) -{ - Function::arg_iterator args = func_cmp->arg_begin(); - Value* packed_tmp0 = args++; - packed_tmp0->setName("tmp0"); - Value* packed_tmp1 = args++; - packed_tmp1->setName("tmp1"); - Value* packed_tmp2 = args++; - packed_tmp2->setName("tmp2"); - - BasicBlock* label_entry_44 = new BasicBlock("entry",func_cmp,0); - BasicBlock* label_cond__14 = new BasicBlock("cond.?14",func_cmp,0); - BasicBlock* label_cond_cont20 = new BasicBlock("cond.cont20",func_cmp,0); - BasicBlock* label_cond__28 = new BasicBlock("cond.?28",func_cmp,0); - BasicBlock* label_cond_cont34 = new BasicBlock("cond.cont34",func_cmp,0); - BasicBlock* label_cond__42 = new BasicBlock("cond.?42",func_cmp,0); - BasicBlock* label_cond_cont48 = new BasicBlock("cond.cont48",func_cmp,0); - - // Block entry (label_entry_44) - ExtractElementInst* float_tmp3 = new ExtractElementInst(packed_tmp0, const_int32_19, "tmp3", label_entry_44); - CastInst* double_conv = new FPExtInst(float_tmp3, Type::DoubleTy, "conv", label_entry_44); - FCmpInst* int1_cmp_45 = new FCmpInst(FCmpInst::FCMP_OLT, double_conv, const_double_27, "cmp", label_entry_44); - ExtractElementInst* float_tmp11 = new ExtractElementInst(packed_tmp0, const_int32_23, "tmp11", label_entry_44); - CastInst* double_conv12 = new FPExtInst(float_tmp11, Type::DoubleTy, "conv12", label_entry_44); - FCmpInst* int1_cmp13 = new FCmpInst(FCmpInst::FCMP_OLT, double_conv12, const_double_27, "cmp13", label_entry_44); - SelectInst* packed_tmp1_tmp2 = new SelectInst(int1_cmp_45, packed_tmp1, packed_tmp2, "tmp1.tmp2", label_entry_44); - new BranchInst(label_cond__14, label_cond_cont20, int1_cmp13, label_entry_44); - - // Block cond.?14 (label_cond__14) - ShuffleVectorInst* packed_tmp233 = new ShuffleVectorInst(packed_tmp1_tmp2, packed_tmp1, const_packed_28, "tmp233", label_cond__14); - ExtractElementInst* float_tmp254 = new ExtractElementInst(packed_tmp0, const_int32_25, "tmp254", label_cond__14); - CastInst* double_conv265 = new FPExtInst(float_tmp254, Type::DoubleTy, "conv265", label_cond__14); - FCmpInst* int1_cmp276 = new FCmpInst(FCmpInst::FCMP_OLT, double_conv265, const_double_27, "cmp276", label_cond__14); - new BranchInst(label_cond__28, label_cond_cont34, int1_cmp276, label_cond__14); - - // Block cond.cont20 (label_cond_cont20) - ShuffleVectorInst* packed_tmp23 = new ShuffleVectorInst(packed_tmp1_tmp2, packed_tmp2, const_packed_28, "tmp23", label_cond_cont20); - ExtractElementInst* float_tmp25 = new ExtractElementInst(packed_tmp0, const_int32_25, "tmp25", label_cond_cont20); - CastInst* double_conv26 = new FPExtInst(float_tmp25, Type::DoubleTy, "conv26", label_cond_cont20); - FCmpInst* int1_cmp27 = new FCmpInst(FCmpInst::FCMP_OLT, double_conv26, const_double_27, "cmp27", label_cond_cont20); - new BranchInst(label_cond__28, label_cond_cont34, int1_cmp27, label_cond_cont20); - - // Block cond.?28 (label_cond__28) - PHINode* packed_tmp23_reg2mem_0 = new PHINode(VectorTy_4, "tmp23.reg2mem.0", label_cond__28); - packed_tmp23_reg2mem_0->reserveOperandSpace(2); - packed_tmp23_reg2mem_0->addIncoming(packed_tmp233, label_cond__14); - packed_tmp23_reg2mem_0->addIncoming(packed_tmp23, label_cond_cont20); - - ShuffleVectorInst* packed_tmp378 = new ShuffleVectorInst(packed_tmp23_reg2mem_0, packed_tmp1, const_packed_30, "tmp378", label_cond__28); - ExtractElementInst* float_tmp399 = new ExtractElementInst(packed_tmp0, const_int32_24, "tmp399", label_cond__28); - CastInst* double_conv4010 = new FPExtInst(float_tmp399, Type::DoubleTy, "conv4010", label_cond__28); - FCmpInst* int1_cmp4111 = new FCmpInst(FCmpInst::FCMP_OLT, double_conv4010, const_double_27, "cmp4111", label_cond__28); - new BranchInst(label_cond__42, label_cond_cont48, int1_cmp4111, label_cond__28); - - // Block cond.cont34 (label_cond_cont34) - PHINode* packed_tmp23_reg2mem_1 = new PHINode(VectorTy_4, "tmp23.reg2mem.1", label_cond_cont34); - packed_tmp23_reg2mem_1->reserveOperandSpace(2); - packed_tmp23_reg2mem_1->addIncoming(packed_tmp233, label_cond__14); - packed_tmp23_reg2mem_1->addIncoming(packed_tmp23, label_cond_cont20); - - ShuffleVectorInst* packed_tmp37 = new ShuffleVectorInst(packed_tmp23_reg2mem_1, packed_tmp2, const_packed_30, "tmp37", label_cond_cont34); - ExtractElementInst* float_tmp39 = new ExtractElementInst(packed_tmp0, const_int32_24, "tmp39", label_cond_cont34); - CastInst* double_conv40 = new FPExtInst(float_tmp39, Type::DoubleTy, "conv40", label_cond_cont34); - FCmpInst* int1_cmp41 = new FCmpInst(FCmpInst::FCMP_OLT, double_conv40, const_double_27, "cmp41", label_cond_cont34); - new BranchInst(label_cond__42, label_cond_cont48, int1_cmp41, label_cond_cont34); - - // Block cond.?42 (label_cond__42) - PHINode* packed_tmp37_reg2mem_0 = new PHINode(VectorTy_4, "tmp37.reg2mem.0", label_cond__42); - packed_tmp37_reg2mem_0->reserveOperandSpace(2); - packed_tmp37_reg2mem_0->addIncoming(packed_tmp378, label_cond__28); - packed_tmp37_reg2mem_0->addIncoming(packed_tmp37, label_cond_cont34); - - ShuffleVectorInst* packed_tmp5113 = new ShuffleVectorInst(packed_tmp37_reg2mem_0, packed_tmp1, const_packed_32, "tmp5113", label_cond__42); - new ReturnInst(packed_tmp5113, label_cond__42); - - // Block cond.cont48 (label_cond_cont48) - PHINode* packed_tmp37_reg2mem_1 = new PHINode(VectorTy_4, "tmp37.reg2mem.1", label_cond_cont48); - packed_tmp37_reg2mem_1->reserveOperandSpace(2); - packed_tmp37_reg2mem_1->addIncoming(packed_tmp378, label_cond__28); - packed_tmp37_reg2mem_1->addIncoming(packed_tmp37, label_cond_cont34); - - ShuffleVectorInst* packed_tmp51 = new ShuffleVectorInst(packed_tmp37_reg2mem_1, packed_tmp2, const_packed_32, "tmp51", label_cond_cont48); - new ReturnInst(packed_tmp51, label_cond_cont48); - -} - -// Function: vcos (func_vcos) -{ - Function::arg_iterator args = func_vcos->arg_begin(); - Value* packed_val = args++; - packed_val->setName("val"); - - BasicBlock* label_entry_53 = new BasicBlock("entry",func_vcos,0); - - // Block entry (label_entry_53) - ExtractElementInst* float_tmp1 = new ExtractElementInst(packed_val, const_int32_19, "tmp1", label_entry_53); - CastInst* double_conv_54 = new FPExtInst(float_tmp1, Type::DoubleTy, "conv", label_entry_53); - ExtractElementInst* float_tmp3_55 = new ExtractElementInst(packed_val, const_int32_23, "tmp3", label_entry_53); - CastInst* double_conv4 = new FPExtInst(float_tmp3_55, Type::DoubleTy, "conv4", label_entry_53); - ExtractElementInst* float_tmp6_56 = new ExtractElementInst(packed_val, const_int32_25, "tmp6", label_entry_53); - CastInst* double_conv7 = new FPExtInst(float_tmp6_56, Type::DoubleTy, "conv7", label_entry_53); - ExtractElementInst* float_tmp9 = new ExtractElementInst(packed_val, const_int32_24, "tmp9", label_entry_53); - CastInst* double_conv10 = new FPExtInst(float_tmp9, Type::DoubleTy, "conv10", label_entry_53); - std::vector<Value*> int32_call_params; - int32_call_params.push_back(const_ptr_34); - int32_call_params.push_back(double_conv_54); - int32_call_params.push_back(double_conv4); - int32_call_params.push_back(double_conv7); - int32_call_params.push_back(double_conv10); - CallInst* int32_call = new CallInst(func_printf, int32_call_params.begin(), int32_call_params.end(), "call", label_entry_53); - int32_call->setCallingConv(CallingConv::C); - int32_call->setTailCall(true);const ParamAttrsList *int32_call_PAL = 0; - int32_call->setParamAttrs(int32_call_PAL); - - CallInst* float_call13 = new CallInst(func_cosf, float_tmp1, "call13", label_entry_53); - float_call13->setCallingConv(CallingConv::C); - float_call13->setTailCall(true);const ParamAttrsList *float_call13_PAL = 0; - float_call13->setParamAttrs(float_call13_PAL); - - InsertElementInst* packed_tmp15 = new InsertElementInst(const_packed_35, float_call13, const_int32_19, "tmp15", label_entry_53); - CallInst* float_call18 = new CallInst(func_cosf, float_tmp1, "call18", label_entry_53); - float_call18->setCallingConv(CallingConv::C); - float_call18->setTailCall(true);const ParamAttrsList *float_call18_PAL = 0; - float_call18->setParamAttrs(float_call18_PAL); - - InsertElementInst* packed_tmp20 = new InsertElementInst(packed_tmp15, float_call18, const_int32_23, "tmp20", label_entry_53); - CallInst* float_call23 = new CallInst(func_cosf, float_tmp1, "call23", label_entry_53); - float_call23->setCallingConv(CallingConv::C); - float_call23->setTailCall(true);const ParamAttrsList *float_call23_PAL = 0; - float_call23->setParamAttrs(float_call23_PAL); - - InsertElementInst* packed_tmp25 = new InsertElementInst(packed_tmp20, float_call23, const_int32_25, "tmp25", label_entry_53); - CallInst* float_call28 = new CallInst(func_cosf, float_tmp1, "call28", label_entry_53); - float_call28->setCallingConv(CallingConv::C); - float_call28->setTailCall(true);const ParamAttrsList *float_call28_PAL = 0; - float_call28->setParamAttrs(float_call28_PAL); - - InsertElementInst* packed_tmp30 = new InsertElementInst(packed_tmp25, float_call28, const_int32_24, "tmp30", label_entry_53); - CastInst* double_conv33 = new FPExtInst(float_call13, Type::DoubleTy, "conv33", label_entry_53); - CastInst* double_conv36 = new FPExtInst(float_call18, Type::DoubleTy, "conv36", label_entry_53); - CastInst* double_conv39 = new FPExtInst(float_call23, Type::DoubleTy, "conv39", label_entry_53); - CastInst* double_conv42 = new FPExtInst(float_call28, Type::DoubleTy, "conv42", label_entry_53); - std::vector<Value*> int32_call43_params; - int32_call43_params.push_back(const_ptr_36); - int32_call43_params.push_back(double_conv33); - int32_call43_params.push_back(double_conv36); - int32_call43_params.push_back(double_conv39); - int32_call43_params.push_back(double_conv42); - CallInst* int32_call43 = new CallInst(func_printf, int32_call43_params.begin(), int32_call43_params.end(), "call43", label_entry_53); - int32_call43->setCallingConv(CallingConv::C); - int32_call43->setTailCall(true);const ParamAttrsList *int32_call43_PAL = 0; - int32_call43->setParamAttrs(int32_call43_PAL); - - new ReturnInst(packed_tmp30, label_entry_53); - -} - -// Function: scs (func_scs) -{ - Function::arg_iterator args = func_scs->arg_begin(); - Value* packed_val_58 = args++; - packed_val_58->setName("val"); - - BasicBlock* label_entry_59 = new BasicBlock("entry",func_scs,0); - - // Block entry (label_entry_59) - ExtractElementInst* float_tmp2 = new ExtractElementInst(packed_val_58, const_int32_19, "tmp2", label_entry_59); - CallInst* float_call_60 = new CallInst(func_cosf, float_tmp2, "call", label_entry_59); - float_call_60->setCallingConv(CallingConv::C); - float_call_60->setTailCall(true);const ParamAttrsList *float_call_60_PAL = 0; - float_call_60->setParamAttrs(float_call_60_PAL); - - InsertElementInst* packed_tmp5 = new InsertElementInst(const_packed_35, float_call_60, const_int32_19, "tmp5", label_entry_59); - CallInst* float_call7 = new CallInst(func_sinf, float_tmp2, "call7", label_entry_59); - float_call7->setCallingConv(CallingConv::C); - float_call7->setTailCall(true);const ParamAttrsList *float_call7_PAL = 0; - float_call7->setParamAttrs(float_call7_PAL); - - InsertElementInst* packed_tmp9 = new InsertElementInst(packed_tmp5, float_call7, const_int32_23, "tmp9", label_entry_59); - new ReturnInst(packed_tmp9, label_entry_59); - -} - -// Function: vsin (func_vsin) -{ - Function::arg_iterator args = func_vsin->arg_begin(); - Value* packed_val_62 = args++; - packed_val_62->setName("val"); - - BasicBlock* label_entry_63 = new BasicBlock("entry",func_vsin,0); - - // Block entry (label_entry_63) - ExtractElementInst* float_tmp2_64 = new ExtractElementInst(packed_val_62, const_int32_19, "tmp2", label_entry_63); - CallInst* float_call_65 = new CallInst(func_sinf, float_tmp2_64, "call", label_entry_63); - float_call_65->setCallingConv(CallingConv::C); - float_call_65->setTailCall(true);const ParamAttrsList *float_call_65_PAL = 0; - float_call_65->setParamAttrs(float_call_65_PAL); - - InsertElementInst* packed_tmp6 = new InsertElementInst(const_packed_35, float_call_65, const_int32_19, "tmp6", label_entry_63); - InsertElementInst* packed_tmp9_66 = new InsertElementInst(packed_tmp6, float_call_65, const_int32_23, "tmp9", label_entry_63); - InsertElementInst* packed_tmp12 = new InsertElementInst(packed_tmp9_66, float_call_65, const_int32_25, "tmp12", label_entry_63); - InsertElementInst* packed_tmp15_67 = new InsertElementInst(packed_tmp12, float_call_65, const_int32_24, "tmp15", label_entry_63); - new ReturnInst(packed_tmp15_67, label_entry_63); - -} - -// Function: kilp (func_kilp) -{ - Function::arg_iterator args = func_kilp->arg_begin(); - Value* packed_val_69 = args++; - packed_val_69->setName("val"); - - BasicBlock* label_entry_70 = new BasicBlock("entry",func_kilp,0); - BasicBlock* label_lor_rhs = new BasicBlock("lor_rhs",func_kilp,0); - BasicBlock* label_lor_rhs5 = new BasicBlock("lor_rhs5",func_kilp,0); - BasicBlock* label_lor_rhs11 = new BasicBlock("lor_rhs11",func_kilp,0); - BasicBlock* label_UnifiedReturnBlock_71 = new BasicBlock("UnifiedReturnBlock",func_kilp,0); - - // Block entry (label_entry_70) - ExtractElementInst* float_tmp1_72 = new ExtractElementInst(packed_val_69, const_int32_19, "tmp1", label_entry_70); - FCmpInst* int1_cmp_73 = new FCmpInst(FCmpInst::FCMP_OLT, float_tmp1_72, const_float_18, "cmp", label_entry_70); - new BranchInst(label_UnifiedReturnBlock_71, label_lor_rhs, int1_cmp_73, label_entry_70); - - // Block lor_rhs (label_lor_rhs) - ExtractElementInst* float_tmp3_75 = new ExtractElementInst(packed_val_69, const_int32_23, "tmp3", label_lor_rhs); - FCmpInst* int1_cmp4 = new FCmpInst(FCmpInst::FCMP_OLT, float_tmp3_75, const_float_18, "cmp4", label_lor_rhs); - new BranchInst(label_UnifiedReturnBlock_71, label_lor_rhs5, int1_cmp4, label_lor_rhs); - - // Block lor_rhs5 (label_lor_rhs5) - ExtractElementInst* float_tmp7 = new ExtractElementInst(packed_val_69, const_int32_25, "tmp7", label_lor_rhs5); - FCmpInst* int1_cmp8 = new FCmpInst(FCmpInst::FCMP_OLT, float_tmp7, const_float_18, "cmp8", label_lor_rhs5); - new BranchInst(label_UnifiedReturnBlock_71, label_lor_rhs11, int1_cmp8, label_lor_rhs5); - - // Block lor_rhs11 (label_lor_rhs11) - ExtractElementInst* float_tmp13 = new ExtractElementInst(packed_val_69, const_int32_24, "tmp13", label_lor_rhs11); - FCmpInst* int1_cmp14 = new FCmpInst(FCmpInst::FCMP_OLT, float_tmp13, const_float_18, "cmp14", label_lor_rhs11); - CastInst* int32_retval = new ZExtInst(int1_cmp14, IntegerType::get(32), "retval", label_lor_rhs11); - new ReturnInst(int32_retval, label_lor_rhs11); - - // Block UnifiedReturnBlock (label_UnifiedReturnBlock_71) - new ReturnInst(const_int32_23, label_UnifiedReturnBlock_71); - -} - -return mod; - -} +static const unsigned char llvm_builtins_data[] = { +0x42,0x43,0xc0,0xde,0x21,0x0c,0x00,0x00,0x29,0x02,0x00,0x00,0x01,0x10,0x00,0x00, +0x10,0x00,0x00,0x00,0x07,0x81,0x23,0x91,0x41,0xc8,0x04,0x49,0x06,0x10,0x32,0x39, +0x92,0x01,0x84,0x0c,0x25,0x05,0x08,0x19,0x1e,0x04,0x8b,0x62,0x80,0x14,0x45,0x02, +0x42,0x92,0x0b,0x42,0xa4,0x10,0x32,0x14,0x38,0x08,0x18,0x49,0x0a,0x32,0x44,0x24, +0x48,0x0a,0x90,0x21,0x23,0x44,0x72,0x80,0x8c,0x14,0x21,0x86,0x0a,0x8a,0x0a,0x64, +0x0c,0x1f,0x00,0x00,0x49,0x18,0x00,0x00,0x02,0x00,0x00,0x00,0x0b,0x04,0x00,0x0c, +0x00,0x00,0x00,0x00,0x51,0x20,0x00,0x00,0x12,0x00,0x00,0x00,0x32,0x22,0x48,0x09, +0x20,0x65,0x82,0x84,0x00,0x26,0x45,0x48,0x05,0x09,0x26,0x45,0xc6,0x05,0x42,0x52, +0x26,0x08,0xae,0x19,0x80,0x61,0x04,0x02,0x98,0x23,0x00,0x83,0x29,0x80,0x21,0x00, +0xb2,0x73,0x04,0x01,0x51,0x8a,0xf4,0x08,0x92,0xa4,0x39,0x47,0x80,0x50,0x2b,0x03, +0x00,0xa0,0x08,0x21,0x5c,0x46,0x2b,0x44,0x08,0x21,0xd4,0x40,0x14,0x01,0x80,0x11, +0x80,0x22,0x88,0x00,0x13,0xa2,0x74,0xb0,0x03,0x3c,0xb0,0x83,0x36,0x80,0x87,0x71, +0x68,0x03,0x76,0x48,0x07,0x77,0xa8,0x07,0x7c,0x68,0x83,0x73,0x70,0x87,0x7a,0xd8, +0x70,0x0f,0xe5,0xd0,0x06,0xf0,0xa0,0x07,0x73,0x20,0x07,0x7a,0x30,0x07,0x72,0xa0, +0x07,0x73,0x20,0x07,0x6d,0x90,0x0e,0x71,0xa0,0x07,0x78,0xa0,0x07,0x78,0xd0,0x06, +0xe9,0x80,0x07,0x7a,0x80,0x07,0x7a,0x80,0x07,0x6d,0x90,0x0e,0x71,0x60,0x07,0x7a, +0x10,0x07,0x76,0xa0,0x07,0x71,0x60,0x07,0x6d,0x90,0x0e,0x73,0x20,0x07,0x7a,0x30, +0x07,0x72,0xa0,0x07,0x73,0x20,0x07,0x6d,0x90,0x0e,0x76,0x40,0x07,0x7a,0x30,0x07, +0x72,0xa0,0x07,0x76,0x40,0x07,0x6d,0x60,0x0e,0x73,0x20,0x07,0x7a,0x30,0x07,0x72, +0xa0,0x07,0x73,0x20,0x07,0x6d,0x60,0x0e,0x76,0x40,0x07,0x7a,0x30,0x07,0x72,0xa0, +0x07,0x76,0x40,0x07,0x6d,0x60,0x0f,0x76,0x40,0x07,0x7a,0x60,0x07,0x74,0xa0,0x07, +0x76,0x40,0x07,0x6d,0x60,0x0f,0x71,0x20,0x07,0x78,0xa0,0x07,0x71,0x20,0x07,0x78, +0xa0,0x07,0x71,0x20,0x07,0x78,0xd0,0x06,0xe1,0x00,0x07,0x7a,0x00,0x07,0x7a,0x60, +0x07,0x74,0xd0,0x06,0xe6,0x80,0x07,0x70,0xa0,0x07,0x71,0x20,0x07,0x78,0xa0,0x07, +0x71,0x20,0x07,0x78,0xa0,0xf3,0x40,0x88,0x04,0x32,0x32,0x02,0x04,0x20,0x76,0x46, +0xfc,0x6c,0x48,0x92,0x00,0x40,0x00,0x00,0x00,0x00,0x0c,0x49,0x12,0x20,0x00,0x00, +0x00,0x00,0x80,0x21,0x89,0x02,0x00,0x01,0x00,0x00,0x00,0x30,0x24,0x59,0x00,0x20, +0x08,0x00,0x00,0x00,0x86,0x24,0x0a,0x00,0x04,0x00,0x00,0x00,0xc0,0x90,0x84,0x01, +0x02,0x00,0x00,0x00,0x00,0x18,0x92,0x1c,0x40,0x00,0x00,0x00,0x00,0x00,0x43,0x12, +0x05,0x00,0x02,0x00,0x00,0x00,0x60,0x48,0x72,0x00,0x01,0x00,0x00,0x00,0x00,0x0c, +0x49,0x14,0x00,0x08,0x00,0x00,0x00,0x80,0x21,0x49,0x01,0x00,0x41,0x00,0x00,0x00, +0x90,0x05,0x02,0x00,0x10,0x00,0x00,0x00,0x32,0x1e,0x98,0x10,0x19,0x11,0x4c,0x90, +0x8c,0x09,0x26,0x47,0xc6,0x04,0x43,0x8a,0x8a,0x59,0x8b,0x43,0x50,0xd2,0x09,0x02, +0x81,0xd2,0x73,0x50,0xc9,0x0c,0x2a,0x99,0x41,0x25,0x33,0xa8,0x64,0x56,0x28,0x66, +0x2d,0x0e,0x41,0xcf,0x2a,0x15,0x04,0x4a,0xcf,0x41,0x25,0x33,0xa8,0x64,0x06,0x95, +0xcc,0xa0,0x92,0x59,0x01,0x00,0x00,0x00,0x53,0x82,0x26,0x0c,0x04,0x00,0x00,0x00, +0x22,0x00,0x00,0x00,0x13,0x04,0x41,0x2c,0x10,0x00,0x00,0x00,0x05,0x00,0x00,0x00, +0x04,0xc6,0x08,0x40,0x10,0x04,0xe1,0x70,0x18,0x23,0x00,0x41,0x10,0x84,0xc3,0x60, +0x04,0x00,0x00,0x00,0x93,0x0c,0xce,0x43,0x4c,0x31,0x3c,0x8e,0x34,0xc9,0x30,0x41, +0xc2,0x14,0x03,0x34,0x51,0x93,0x0c,0x4d,0x44,0x4c,0x31,0x44,0x8d,0x35,0x56,0x01, +0x04,0xc3,0x55,0x21,0x16,0x0e,0x04,0x00,0x0f,0x00,0x00,0x00,0x46,0x41,0x08,0xcc, +0x73,0x9b,0x05,0x21,0x30,0xcf,0x6e,0x18,0x84,0x00,0x2c,0x8b,0x35,0x04,0x80,0x39, +0x04,0x81,0x5d,0x20,0x80,0x0f,0x0c,0x43,0xe4,0xd3,0x36,0x81,0x04,0x3e,0x30,0x0c, +0x91,0x4f,0x5b,0x05,0x12,0xf8,0xc0,0x30,0x44,0x7e,0x7d,0x00,0x05,0xd1,0x4c,0x11, +0x66,0x12,0x83,0xc0,0x3c,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x61,0x20,0x00,0x00, +0x2a,0x00,0x00,0x00,0x13,0x04,0x43,0x2c,0x10,0x00,0x00,0x00,0x08,0x00,0x00,0x00, +0x24,0x8a,0xa0,0x0c,0x46,0x00,0x4a,0x80,0xc2,0x1c,0x84,0x55,0x55,0xd6,0x1c,0x84, +0x45,0x51,0x16,0x81,0x19,0x80,0x11,0x80,0x31,0x02,0x10,0x04,0x41,0xfc,0x03,0x00, +0x63,0x08,0x0d,0x34,0xc9,0x70,0x55,0xc2,0x2c,0x43,0x20,0x60,0x73,0x0c,0xd3,0x15, +0x8d,0x21,0x34,0xd1,0x18,0x42,0xf3,0x8c,0x55,0x00,0x81,0xa0,0x6d,0x73,0x0c,0x19, +0xe7,0x60,0x87,0x52,0x38,0x10,0x00,0x00,0x13,0x00,0x00,0x00,0x17,0x60,0x20,0xc5, +0x74,0x10,0x8d,0x65,0x14,0x13,0xf3,0xd4,0xb4,0x6d,0x14,0x13,0xf3,0xd4,0xb8,0x69, +0x14,0x13,0xf3,0xd4,0xb6,0x75,0x14,0x13,0xf3,0xd4,0xba,0x35,0x0c,0x13,0xf3,0x9c, +0x80,0xe4,0x36,0x48,0x81,0x10,0xc3,0x4a,0x4c,0x54,0xd4,0x6c,0x8b,0x23,0x28,0x76, +0x41,0x4c,0xcc,0xa3,0x1b,0x07,0x21,0x00,0xcb,0x72,0x00,0x05,0xd1,0x4c,0x11,0x66, +0x18,0x83,0xc0,0x3c,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x61,0x20,0x00,0x00, +0x81,0x00,0x00,0x00,0x13,0x04,0x4d,0x2c,0x10,0x00,0x00,0x00,0x04,0x00,0x00,0x00, +0x24,0xca,0x60,0x04,0xa0,0x04,0x8a,0x80,0xc2,0x0c,0x00,0x91,0x11,0x00,0x00,0x00, +0x63,0x08,0x4d,0x64,0x16,0xc1,0x49,0x86,0xab,0x22,0x66,0x19,0x02,0x01,0x1b,0x43, +0x70,0xa2,0x59,0x82,0x61,0x0c,0xe1,0x89,0x66,0x09,0x86,0x81,0x0a,0x20,0x0b,0x34, +0x61,0x8e,0x81,0xda,0xa2,0x31,0x84,0x46,0xb2,0x8e,0xe0,0x24,0x83,0x57,0x11,0xb3, +0x0c,0x44,0xf1,0x8d,0x21,0x38,0xd2,0x2c,0x81,0x31,0x86,0xf0,0x48,0xb3,0x04,0xc6, +0x40,0x05,0x00,0x06,0x44,0x18,0x14,0x73,0x0c,0x9c,0x18,0x48,0x63,0x08,0xcd,0x64, +0x64,0x40,0x70,0x92,0xa1,0x0c,0x2a,0x62,0x96,0xe1,0x40,0xcc,0x60,0x0c,0xc1,0x99, +0x66,0x09,0x92,0x31,0x84,0x67,0x9a,0x25,0x48,0x06,0x2a,0x80,0x33,0x38,0xd0,0x00, +0x99,0x63,0x18,0x83,0x34,0x98,0xc6,0x10,0x1a,0xc8,0xd6,0x80,0xe0,0x24,0x03,0x1b, +0x54,0xc4,0x2c,0x83,0xb2,0xb4,0xc1,0x18,0x82,0x03,0xcd,0x12,0x30,0x63,0x08,0x0f, +0x34,0x4b,0xc0,0x0c,0x54,0x00,0x6e,0xa0,0xbc,0xc1,0x32,0xc7,0xa0,0x06,0x70,0x00, +0x61,0x1c,0x84,0x03,0x01,0x00,0x00,0x00,0x4e,0x00,0x00,0x00,0x76,0x52,0x4c,0xcc, +0x73,0xd3,0x24,0x05,0x64,0xec,0xcd,0x8d,0xcc,0xe5,0x87,0x46,0xc6,0x50,0x8a,0x89, +0x79,0xee,0xdb,0x54,0x8a,0x89,0x79,0xee,0xdd,0x1a,0x88,0x89,0x79,0x68,0x73,0x20, +0x26,0xe6,0xa9,0xed,0x81,0x98,0x98,0xc7,0x36,0x0b,0x62,0x62,0x9e,0xdb,0x32,0x88, +0x89,0x79,0x72,0xd3,0x20,0x26,0xe6,0xd9,0x8d,0x83,0x98,0x98,0xa7,0xb7,0x95,0x62, +0x62,0x9e,0xbb,0x27,0x2d,0x20,0x63,0x6f,0x6e,0x64,0x2e,0x3a,0x34,0x35,0x56,0x62, +0x08,0x4e,0x53,0xd9,0xba,0xb5,0x14,0x02,0xf3,0xe0,0xf5,0x25,0x2c,0x82,0xd3,0x0c, +0xbe,0xe0,0x34,0xd3,0x8d,0x9b,0x88,0x21,0x38,0xcd,0x60,0xd7,0x24,0x01,0x63,0xec, +0xcd,0x8d,0xcc,0x45,0x87,0x44,0x80,0x8c,0xbd,0xb9,0x91,0xb9,0xfc,0xc4,0xd0,0x90, +0x02,0x8c,0xb1,0x37,0x37,0x32,0x97,0x1f,0x73,0x29,0x26,0xe6,0xc1,0x71,0x7b,0x29, +0x26,0xe6,0xc1,0x77,0xfb,0x28,0x04,0xe6,0xa9,0x6f,0x52,0x01,0x32,0xf6,0xe6,0x46, +0xe6,0xa2,0x13,0x73,0x63,0x18,0x83,0xc0,0x3c,0xb6,0x41,0x08,0x4e,0x33,0x58,0x47, +0x31,0x31,0x4f,0x5d,0x1f,0xc3,0x22,0x38,0xcd,0xe0,0x0b,0x4e,0x33,0xe1,0xbc,0xa5, +0x18,0x82,0xd3,0x0c,0x77,0x6e,0x20,0xc5,0xc4,0x3c,0xb5,0x4e,0x3a,0x40,0xc6,0xde, +0xdc,0xc8,0x5c,0x7e,0x64,0x70,0x2c,0xa4,0x98,0x98,0xa7,0xee,0xed,0x82,0x10,0x9c, +0xa6,0xba,0x81,0x44,0x70,0x9a,0xc1,0x17,0x9c,0x66,0x32,0x93,0x42,0x60,0x1e,0x7b, +0xb7,0x98,0x62,0x62,0x9e,0xbc,0x36,0x16,0x43,0x70,0x9a,0x0a,0xa7,0x6d,0xa4,0x98, +0x98,0xc7,0xbe,0x8d,0xa4,0x98,0x98,0xc7,0xce,0x0d,0xc6,0x10,0x9c,0x66,0xc0,0x7b, +0x12,0x02,0x32,0xf6,0xe6,0x46,0xe6,0xa2,0x33,0x13,0x73,0x06,0x8b,0xe0,0x34,0x83, +0x2f,0x38,0xcd,0x64,0xd3,0xe6,0x61,0x08,0x4e,0x53,0xd5,0xf6,0x01,0x14,0x44,0x33, +0x45,0x18,0x00,0x00,0x00,0x00,0x00,0x00,0x61,0x20,0x00,0x00,0x4a,0x00,0x00,0x00, +0x13,0x04,0x41,0x2c,0x10,0x00,0x00,0x00,0x07,0x00,0x00,0x00,0x24,0xca,0x60,0x04, +0xa0,0x04,0x8a,0x80,0xc2,0x0c,0x00,0xb9,0x61,0x0c,0x04,0x10,0x1e,0xe1,0x19,0xc6, +0x40,0x02,0xe1,0x11,0x1e,0x00,0x00,0x00,0x63,0x08,0xcd,0x63,0x15,0xc1,0x31,0x84, +0x06,0xb2,0x8b,0xe0,0x18,0x42,0x13,0x59,0x46,0x70,0x0c,0xa1,0x71,0x6c,0x23,0x38, +0x16,0x02,0x04,0xc7,0x64,0x61,0x1a,0x37,0x16,0x01,0x04,0x48,0x35,0xc7,0x20,0x79, +0xcf,0x58,0x04,0x10,0x20,0xd5,0x1c,0xc3,0x07,0x06,0xd0,0x58,0x04,0x10,0x20,0xd5, +0x1c,0x43,0x18,0x88,0x41,0x34,0x16,0x01,0x04,0x48,0x35,0xc7,0x30,0x06,0x64,0xe0, +0x98,0x47,0xd0,0xc0,0x80,0xa0,0x89,0x01,0x41,0x23,0x03,0x82,0x63,0x21,0x40,0x70, +0x50,0x66,0x70,0x06,0x68,0x90,0x06,0x58,0x06,0xe1,0x40,0x00,0x25,0x00,0x00,0x00, +0x56,0x52,0x4c,0xcc,0x73,0xd3,0x56,0x41,0x4c,0xcc,0x53,0xdb,0x05,0x31,0x31,0xcf, +0x6d,0x19,0xc4,0xc4,0x3c,0xba,0x6d,0x10,0x13,0xf3,0xf4,0xd6,0x41,0x08,0xc0,0xb2, +0x18,0x46,0x21,0x38,0x4d,0x85,0x9b,0x46,0x21,0x38,0x4d,0xb5,0x9b,0x8a,0x21,0x00, +0xcb,0x82,0xdf,0x66,0x62,0x08,0x4e,0x53,0xdd,0xb7,0x9d,0x18,0x82,0xd3,0x54,0xb7, +0x6e,0x28,0x86,0xe0,0x34,0xd5,0xdd,0xdb,0x47,0x31,0x31,0x4f,0x9d,0x9b,0x87,0x21, +0x00,0xcb,0x52,0xdf,0x06,0x62,0x08,0xc0,0xb2,0xd4,0xbc,0x59,0x10,0x82,0xd3,0x54, +0x96,0x62,0x08,0x4e,0x53,0xe1,0xb6,0x85,0x14,0x13,0xf3,0xd8,0xb4,0x8d,0x14,0x13, +0xf3,0xd8,0xb9,0x89,0x18,0x02,0xb0,0x2c,0xf6,0x6d,0x24,0x86,0x00,0x2c,0x8b,0xcd, +0x1b,0x87,0x21,0x38,0x4d,0x55,0xd3,0xd6,0x30,0x54,0xc0,0x72,0x00,0x05,0xd1,0x4c, +0x11,0x06,0x00,0x00,0x00,0x00,0x00,0x00,0x61,0x20,0x00,0x00,0x19,0x00,0x00,0x00, +0x13,0x04,0x41,0x2c,0x10,0x00,0x00,0x00,0x03,0x00,0x00,0x00,0x24,0x4a,0x60,0x04, +0x80,0xc2,0x0c,0x00,0x00,0x00,0x00,0x00,0x63,0x08,0xcd,0x33,0x16,0x01,0x04,0x48, +0x34,0xc7,0x00,0x49,0xcf,0x58,0x04,0x10,0x28,0xd1,0x1c,0xc3,0x44,0x39,0x58,0x85, +0x03,0x01,0x00,0x00,0x0a,0x00,0x00,0x00,0x16,0x41,0x4c,0xcc,0x63,0xdb,0x04,0x31, +0x31,0x4f,0x6e,0x0d,0x43,0x05,0x2c,0x07,0x50,0x10,0xcd,0x14,0x61,0x56,0x41,0x4c, +0xcc,0xd3,0x1b,0x45,0x21,0x00,0xcb,0xb2,0x9b,0x04,0x21,0x00,0xcb,0x02,0x00,0x00, +0x00,0x00,0x00,0x00,0x61,0x20,0x00,0x00,0x1b,0x00,0x00,0x00,0x13,0x04,0x41,0x2c, +0x10,0x00,0x00,0x00,0x03,0x00,0x00,0x00,0x24,0xca,0x60,0x04,0xa0,0x04,0x8a,0x80, +0xc2,0x0c,0x00,0x00,0x63,0x08,0xcd,0x33,0x16,0x01,0x04,0xca,0x34,0xc7,0x20,0x51, +0xcf,0x1c,0x43,0x45,0x41,0x73,0x0c,0x16,0x15,0xcd,0x31,0x5c,0x94,0x83,0x58,0x38, +0x10,0x00,0x00,0x00,0x0b,0x00,0x00,0x00,0x86,0x51,0x4c,0xcc,0x53,0xe7,0x76,0x51, +0x4c,0xcc,0x53,0xdb,0x36,0x41,0x4c,0xcc,0x63,0x5b,0x05,0x31,0x31,0x8f,0x6e,0x0d, +0x43,0x05,0x2c,0x66,0x41,0x4c,0xcc,0xd3,0x1f,0x40,0x41,0x34,0x53,0x84,0x19,0x05, +0x21,0x00,0xcb,0x02,0x00,0x00,0x00,0x00,0x61,0x20,0x00,0x00,0x2f,0x00,0x00,0x00, +0x13,0x04,0x45,0x2c,0x10,0x00,0x00,0x00,0x03,0x00,0x00,0x00,0x24,0xca,0xa0,0x04, +0x46,0x00,0x8a,0x80,0xc0,0x08,0x00,0x00,0x63,0x08,0x0d,0x34,0xc9,0x30,0x49,0xc4, +0x2c,0x03,0x11,0x50,0x63,0x08,0xcd,0x33,0xc9,0x50,0x49,0xc4,0x2c,0x03,0x21,0x58, +0x63,0x08,0x4d,0x34,0xc9,0x70,0x49,0xc4,0x2c,0x03,0x31,0x60,0x63,0x08,0x8d,0x33, +0xc9,0x90,0x49,0x84,0x69,0x22,0x70,0xc3,0x27,0x1c,0x08,0x00,0x1a,0x00,0x00,0x00, +0x96,0x51,0x4c,0xcc,0x53,0xdf,0x66,0x41,0x08,0xcc,0x83,0xdb,0x04,0x31,0x31,0x4f, +0x6d,0x15,0xc4,0xc4,0x3c,0xb7,0x61,0x10,0x02,0xf3,0xf0,0x47,0x20,0xb9,0x0d,0x52, +0x20,0xc4,0xb0,0x12,0x13,0x15,0x35,0xdb,0xe2,0x08,0x8a,0x5d,0x10,0x13,0xf3,0xec, +0x37,0x90,0x2c,0x4e,0xf4,0x47,0x87,0x54,0xd7,0x17,0x70,0x2c,0x4e,0xf4,0x47,0x87, +0x74,0x02,0xc8,0xe2,0x44,0x7f,0x74,0x48,0xb9,0x69,0x14,0x02,0xf3,0xd4,0xb8,0x6d, +0x18,0x11,0x31,0x55,0xc0,0x62,0x0d,0x43,0x05,0x2c,0x07,0x50,0x10,0xcd,0x14,0x61, +0x46,0x31,0x08,0xcc,0x03,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x71,0x20,0x00,0x00, +0x12,0x00,0x00,0x00,0x66,0x40,0x54,0x82,0x23,0x59,0xc2,0x20,0x09,0x92,0x1d,0x18, +0x4f,0x84,0x34,0x53,0x61,0x03,0xc4,0xe3,0x58,0x85,0x05,0x14,0xbe,0x34,0x45,0xb5, +0x21,0x10,0x82,0x23,0x15,0x46,0x30,0x2c,0xc8,0x64,0x02,0x06,0xf0,0x3c,0x91,0x73, +0x19,0x00,0xe1,0x4b,0x53,0x64,0x0a,0x84,0x84,0x34,0x85,0x31,0x10,0x0a,0xb2,0x3c, +0x56,0x30,0x08,0xcc,0x63,0x0b,0x44,0x25,0x21,0x0d,0x00,0x00,0x00,0x00,0x00,0x00}; diff --git a/src/gallium/auxiliary/gallivm/gallivm_cpu.cpp b/src/gallium/auxiliary/gallivm/gallivm_cpu.cpp index 8f9830d0b1..857c190f7b 100644 --- a/src/gallium/auxiliary/gallivm/gallivm_cpu.cpp +++ b/src/gallium/auxiliary/gallivm/gallivm_cpu.cpp @@ -41,6 +41,7 @@ #include "pipe/p_context.h" #include "pipe/p_shader_tokens.h" +#include "pipe/p_util.h" #include "tgsi/exec/tgsi_exec.h" #include "tgsi/util/tgsi_dump.h" @@ -179,22 +180,63 @@ typedef void (*vertex_shader_runner)(void *ainputs, float (*aconsts)[4], void *temps); - +#define MAX_TGSI_VERTICES 4 /*! This function is used to execute the gallivm_prog in software. Before calling this function the gallivm_prog has to be JIT compiled with the gallivm_cpu_jit_compile function. */ int gallivm_cpu_vs_exec(struct gallivm_prog *prog, - struct tgsi_exec_vector *inputs, - struct tgsi_exec_vector *dests, - float (*consts)[4], - struct tgsi_exec_vector *temps) + struct tgsi_exec_machine *machine, + const float (*input)[4], + unsigned num_inputs, + float (*output)[4], + unsigned num_outputs, + const float (*constants)[4], + unsigned count, + unsigned input_stride, + unsigned output_stride ) { + unsigned int i, j; + unsigned slot; vertex_shader_runner runner = reinterpret_cast<vertex_shader_runner>(prog->function); + assert(runner); - /*FIXME*/ - runner(inputs, dests, consts, temps); + + for (i = 0; i < count; i += MAX_TGSI_VERTICES) { + unsigned int max_vertices = MIN2(MAX_TGSI_VERTICES, count - i); + + /* Swizzle inputs. + */ + for (j = 0; j < max_vertices; j++) { + for (slot = 0; slot < num_inputs; slot++) { + machine->Inputs[slot].xyzw[0].f[j] = input[slot][0]; + machine->Inputs[slot].xyzw[1].f[j] = input[slot][1]; + machine->Inputs[slot].xyzw[2].f[j] = input[slot][2]; + machine->Inputs[slot].xyzw[3].f[j] = input[slot][3]; + } + + input = (const float (*)[4])((const char *)input + input_stride); + } + + /* run shader */ + runner(machine->Inputs, + machine->Outputs, + (float (*)[4]) constants, + machine->Temps); + + /* Unswizzle all output results + */ + for (j = 0; j < max_vertices; j++) { + for (slot = 0; slot < num_outputs; slot++) { + output[slot][0] = machine->Outputs[slot].xyzw[0].f[j]; + output[slot][1] = machine->Outputs[slot].xyzw[1].f[j]; + output[slot][2] = machine->Outputs[slot].xyzw[2].f[j]; + output[slot][3] = machine->Outputs[slot].xyzw[3].f[j]; + } + output = (float (*)[4])((char *)output + output_stride); + } + } return 0; } diff --git a/src/gallium/auxiliary/gallivm/instructions.cpp b/src/gallium/auxiliary/gallivm/instructions.cpp index 8919491792..1a98491b82 100644 --- a/src/gallium/auxiliary/gallivm/instructions.cpp +++ b/src/gallium/auxiliary/gallivm/instructions.cpp @@ -35,6 +35,8 @@ #include "storage.h" +#include "pipe/p_util.h" + #include <llvm/CallingConv.h> #include <llvm/Constants.h> #include <llvm/DerivedTypes.h> @@ -42,7 +44,8 @@ #include <llvm/InstrTypes.h> #include <llvm/Instructions.h> #include <llvm/ParameterAttributes.h> -#include <llvm/ParamAttrsList.h> +#include <llvm/Support/MemoryBuffer.h> +#include <llvm/Bitcode/ReaderWriter.h> #include <sstream> #include <fstream> @@ -53,7 +56,6 @@ using namespace llvm; #include "gallivm_builtins.cpp" #if 0 - llvm::Value *arrayFromChannels(std::vector<llvm::Value*> &vals) { VectorType *vectorType = VectorType::get(Type::FloatTy, 4); @@ -84,7 +86,10 @@ Instructions::Instructions(llvm::Module *mod, llvm::Function *func, llvm::BasicB m_llvmLit = 0; m_fmtPtr = 0; - createGallivmBuiltins(m_mod); + MemoryBuffer *buffer = MemoryBuffer::getMemBuffer( + (const char*)&llvm_builtins_data[0], + (const char*)&llvm_builtins_data[Elements(llvm_builtins_data)-1]); + m_mod = ParseBitcodeFile(buffer); } llvm::Value * Instructions::add(llvm::Value *in1, llvm::Value *in2) @@ -134,12 +139,12 @@ llvm::Value *Instructions::callFSqrt(llvm::Value *val) // predeclare the intrinsic std::vector<const Type*> fsqrtArgs; fsqrtArgs.push_back(Type::FloatTy); - ParamAttrsList *fsqrtPal = 0; + PAListPtr fsqrtPal; FunctionType* fsqrtType = FunctionType::get( /*Result=*/Type::FloatTy, /*Params=*/fsqrtArgs, /*isVarArg=*/false); - m_llvmFSqrt = new Function( + m_llvmFSqrt = Function::Create( /*Type=*/fsqrtType, /*Linkage=*/GlobalValue::ExternalLinkage, /*Name=*/"llvm.sqrt.f32", m_mod); @@ -161,10 +166,9 @@ llvm::Value * Instructions::rsq(llvm::Value *in1) Value *abs = callFAbs(x); Value *sqrt = callFSqrt(abs); - Value *rsqrt = m_builder.CreateFDiv(ConstantFP::get(Type::FloatTy, - APFloat(1.f)), - sqrt, - name("rsqrt")); + Value *rsqrt = m_builder.CreateFDiv(ConstantFP::get(APFloat(1.f)), + sqrt, + name("rsqrt")); return vectorFromVals(rsqrt, rsqrt, rsqrt, rsqrt); } @@ -191,12 +195,12 @@ llvm::Value *Instructions::callFAbs(llvm::Value *val) // predeclare the intrinsic std::vector<const Type*> fabsArgs; fabsArgs.push_back(Type::FloatTy); - ParamAttrsList *fabsPal = 0; + PAListPtr fabsPal; FunctionType* fabsType = FunctionType::get( /*Result=*/Type::FloatTy, /*Params=*/fabsArgs, /*isVarArg=*/false); - m_llvmFAbs = new Function( + m_llvmFAbs = Function::Create( /*Type=*/fabsType, /*Linkage=*/GlobalValue::ExternalLinkage, /*Name=*/"fabs", m_mod); @@ -234,12 +238,12 @@ llvm::Value * Instructions::callPow(llvm::Value *val1, llvm::Value *val2) std::vector<const Type*> powArgs; powArgs.push_back(Type::FloatTy); powArgs.push_back(Type::FloatTy); - ParamAttrsList *powPal = 0; + PAListPtr powPal; FunctionType* powType = FunctionType::get( /*Result=*/Type::FloatTy, /*Params=*/powArgs, /*isVarArg=*/false); - m_llvmPow = new Function( + m_llvmPow = Function::Create( /*Type=*/powType, /*Linkage=*/GlobalValue::ExternalLinkage, /*Name=*/"llvm.pow.f32", m_mod); @@ -273,9 +277,8 @@ llvm::Value * Instructions::rcp(llvm::Value *in1) Value *x1 = m_builder.CreateExtractElement(in1, m_storage->constantInt(0), name("x1")); - Value *res = m_builder.CreateFDiv(ConstantFP::get(Type::FloatTy, - APFloat(1.f)), - x1, name("rcp")); + Value *res = m_builder.CreateFDiv(ConstantFP::get(APFloat(1.f)), + x1, name("rcp")); return vectorFromVals(res, res, res, res); } @@ -314,13 +317,13 @@ llvm::Value * Instructions::dst(llvm::Value *in1, llvm::Value *in2) m_storage->constantInt(3), name("w")); Value *ry = m_builder.CreateMul(y1, y2, name("tyuy")); - return vectorFromVals(ConstantFP::get(Type::FloatTy, APFloat(1.f)), + return vectorFromVals(ConstantFP::get(APFloat(1.f)), ry, z, w); } llvm::Value * Instructions::ex2(llvm::Value *in) { - llvm::Value *val = callPow(ConstantFP::get(Type::FloatTy, APFloat(2.f)), + llvm::Value *val = callPow(ConstantFP::get(APFloat(2.f)), m_builder.CreateExtractElement( in, m_storage->constantInt(0), name("x1"))); @@ -333,12 +336,12 @@ llvm::Value * Instructions::callFloor(llvm::Value *val) // predeclare the intrinsic std::vector<const Type*> floorArgs; floorArgs.push_back(Type::FloatTy); - ParamAttrsList *floorPal = 0; + PAListPtr floorPal; FunctionType* floorType = FunctionType::get( /*Result=*/Type::FloatTy, /*Params=*/floorArgs, /*isVarArg=*/false); - m_llvmFloor = new Function( + m_llvmFloor = Function::Create( /*Type=*/floorType, /*Linkage=*/GlobalValue::ExternalLinkage, /*Name=*/"floorf", m_mod); @@ -376,12 +379,12 @@ llvm::Value * Instructions::callFLog(llvm::Value *val) // predeclare the intrinsic std::vector<const Type*> flogArgs; flogArgs.push_back(Type::FloatTy); - ParamAttrsList *flogPal = 0; + PAListPtr flogPal; FunctionType* flogType = FunctionType::get( /*Result=*/Type::FloatTy, /*Params=*/flogArgs, /*isVarArg=*/false); - m_llvmFlog = new Function( + m_llvmFlog = Function::Create( /*Type=*/flogType, /*Linkage=*/GlobalValue::ExternalLinkage, /*Name=*/"logf", m_mod); @@ -504,12 +507,12 @@ void Instructions::printVector(llvm::Value *val) llvm::Function * Instructions::declarePrintf() { std::vector<const Type*> args; - ParamAttrsList *params = 0; + PAListPtr params; FunctionType* funcTy = FunctionType::get( /*Result=*/IntegerType::get(32), /*Params=*/args, /*isVarArg=*/true); - Function* func_printf = new Function( + Function* func_printf = Function::Create( /*Type=*/funcTy, /*Linkage=*/GlobalValue::ExternalLinkage, /*Name=*/"printf", m_mod); @@ -521,7 +524,7 @@ llvm::Function * Instructions::declarePrintf() llvm::Value * Instructions::sgt(llvm::Value *in1, llvm::Value *in2) { - Constant *const1f = ConstantFP::get(Type::FloatTy, APFloat(1.000000e+00f)); + Constant *const1f = ConstantFP::get(APFloat(1.000000e+00f)); Constant *const0f = Constant::getNullValue(Type::FloatTy); std::vector<llvm::Value*> vec1 = extractVector(in1); @@ -542,7 +545,7 @@ llvm::Value * Instructions::sgt(llvm::Value *in1, llvm::Value *in2) } llvm::Value * Instructions::sge(llvm::Value *in1, llvm::Value *in2) { - Constant *const1f = ConstantFP::get(Type::FloatTy, APFloat(1.000000e+00f)); + Constant *const1f = ConstantFP::get(APFloat(1.000000e+00f)); Constant *const0f = Constant::getNullValue(Type::FloatTy); std::vector<llvm::Value*> vec1 = extractVector(in1); @@ -566,7 +569,7 @@ llvm::Value * Instructions::sge(llvm::Value *in1, llvm::Value *in2) llvm::Value * Instructions::slt(llvm::Value *in1, llvm::Value *in2) { - Constant *const1f = ConstantFP::get(Type::FloatTy, APFloat(1.000000e+00f)); + Constant *const1f = ConstantFP::get(APFloat(1.000000e+00f)); Constant *const0f = Constant::getNullValue(Type::FloatTy); std::vector<llvm::Value*> vec1 = extractVector(in1); @@ -633,8 +636,8 @@ llvm::Value * Instructions::abs(llvm::Value *in) void Instructions::ifop(llvm::Value *in) { - BasicBlock *ifthen = new BasicBlock(name("ifthen"), m_func,0); - BasicBlock *ifend = new BasicBlock(name("ifthenend"), m_func,0); + BasicBlock *ifthen = BasicBlock::Create(name("ifthen"), m_func,0); + BasicBlock *ifend = BasicBlock::Create(name("ifthenend"), m_func,0); //BasicBlock *yblock = new BasicBlock(name("yblock"), m_func,0); //BasicBlock *zblock = new BasicBlock(name("zblock"), m_func,0); @@ -660,7 +663,7 @@ llvm::BasicBlock * Instructions::currentBlock() const void Instructions::elseop() { assert(!m_ifStack.empty()); - BasicBlock *ifend = new BasicBlock(name("ifend"), m_func,0); + BasicBlock *ifend = BasicBlock::Create(name("ifend"), m_func,0); m_builder.CreateBr(ifend); m_builder.SetInsertPoint(m_ifStack.top()); currentBlock()->setName(name("ifelse")); @@ -687,8 +690,8 @@ llvm::Value * Instructions::lerp(llvm::Value *in1, llvm::Value *in2, void Instructions::beginLoop() { - BasicBlock *begin = new BasicBlock(name("loop"), m_func,0); - BasicBlock *end = new BasicBlock(name("endloop"), m_func,0); + BasicBlock *begin = BasicBlock::Create(name("loop"), m_func,0); + BasicBlock *end = BasicBlock::Create(name("endloop"), m_func,0); m_builder.CreateBr(begin); Loop loop; @@ -711,7 +714,7 @@ void Instructions::endLoop() void Instructions::brk() { assert(!m_loopStack.empty()); - BasicBlock *unr = new BasicBlock(name("unreachable"), m_func,0); + BasicBlock *unr = BasicBlock::Create(name("unreachable"), m_func,0); m_builder.CreateBr(m_loopStack.top().end); m_builder.SetInsertPoint(unr); } @@ -760,13 +763,13 @@ llvm::Function * Instructions::declareFunc(int label) args.push_back(vecPtr); args.push_back(vecPtr); args.push_back(vecPtr); - ParamAttrsList *params = 0; + PAListPtr params; FunctionType *funcType = FunctionType::get( /*Result=*/Type::VoidTy, /*Params=*/args, /*isVarArg=*/false); std::string name = createFuncName(label); - Function *func = new Function( + Function *func = Function::Create( /*Type=*/funcType, /*Linkage=*/GlobalValue::ExternalLinkage, /*Name=*/name.c_str(), m_mod); @@ -784,7 +787,7 @@ void Instructions::bgnSub(unsigned label) ptr_INPUT->setName("INPUT"); m_storage->pushArguments(ptr_INPUT); - llvm::BasicBlock *entry = new BasicBlock("entry", func, 0); + llvm::BasicBlock *entry = BasicBlock::Create("entry", func, 0); m_func = func; m_builder.SetInsertPoint(entry); @@ -809,10 +812,10 @@ llvm::Function * Instructions::findFunction(int label) llvm::Value * Instructions::constVector(float x, float y, float z, float w) { std::vector<Constant*> vec(4); - vec[0] = ConstantFP::get(Type::FloatTy, APFloat(x)); - vec[1] = ConstantFP::get(Type::FloatTy, APFloat(y)); - vec[2] = ConstantFP::get(Type::FloatTy, APFloat(z)); - vec[3] = ConstantFP::get(Type::FloatTy, APFloat(w)); + vec[0] = ConstantFP::get(APFloat(x)); + vec[1] = ConstantFP::get(APFloat(y)); + vec[2] = ConstantFP::get(APFloat(z)); + vec[3] = ConstantFP::get(APFloat(w)); return ConstantVector::get(m_floatVecType, vec); } diff --git a/src/gallium/auxiliary/gallivm/instructions.h b/src/gallium/auxiliary/gallivm/instructions.h index 9ebc17dd8e..19ca84ddc6 100644 --- a/src/gallium/auxiliary/gallivm/instructions.h +++ b/src/gallium/auxiliary/gallivm/instructions.h @@ -36,7 +36,7 @@ #include <llvm/BasicBlock.h> #include <llvm/Module.h> #include <llvm/Value.h> -#include <llvm/Support/LLVMBuilder.h> +#include <llvm/Support/IRBuilder.h> #include <map> #include <stack> @@ -125,7 +125,7 @@ private: llvm::Module *m_mod; llvm::Function *m_func; char m_name[32]; - llvm::LLVMFoldingBuilder m_builder; + llvm::IRBuilder m_builder; int m_idx; llvm::VectorType *m_floatVecType; diff --git a/src/gallium/auxiliary/gallivm/instructionssoa.cpp b/src/gallium/auxiliary/gallivm/instructionssoa.cpp index 6f83b56a72..76049ade7c 100644 --- a/src/gallium/auxiliary/gallivm/instructionssoa.cpp +++ b/src/gallium/auxiliary/gallivm/instructionssoa.cpp @@ -1,8 +1,35 @@ +/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ #include "instructionssoa.h" #include "storagesoa.h" #include "pipe/p_shader_tokens.h" +#include "pipe/p_util.h" #include <llvm/CallingConv.h> #include <llvm/Constants.h> @@ -10,7 +37,10 @@ #include <llvm/Function.h> #include <llvm/Instructions.h> #include <llvm/Transforms/Utils/Cloning.h> -#include <llvm/ParamAttrsList.h> +#include <llvm/ParameterAttributes.h> +#include <llvm/Support/MemoryBuffer.h> +#include <llvm/Bitcode/ReaderWriter.h> + #include <iostream> @@ -143,16 +173,56 @@ std::vector<llvm::Value*> InstructionsSoa::extractVector(llvm::Value *vector) void InstructionsSoa::createFunctionMap() { + m_functionsMap[TGSI_OPCODE_ABS] = "abs"; m_functionsMap[TGSI_OPCODE_DP3] = "dp3"; m_functionsMap[TGSI_OPCODE_DP4] = "dp4"; + m_functionsMap[TGSI_OPCODE_MIN] = "min"; + m_functionsMap[TGSI_OPCODE_MAX] = "max"; m_functionsMap[TGSI_OPCODE_POWER] = "pow"; + m_functionsMap[TGSI_OPCODE_LIT] = "lit"; + m_functionsMap[TGSI_OPCODE_RSQ] = "rsq"; } void InstructionsSoa::createDependencies() { - std::vector<std::string> powDeps(1); - powDeps[0] = "powf"; - m_builtinDependencies["pow"] = powDeps; + { + std::vector<std::string> powDeps(2); + powDeps[0] = "powf"; + powDeps[1] = "powvec"; + m_builtinDependencies["pow"] = powDeps; + } + { + std::vector<std::string> absDeps(2); + absDeps[0] = "fabsf"; + absDeps[1] = "absvec"; + m_builtinDependencies["abs"] = absDeps; + } + { + std::vector<std::string> maxDeps(1); + maxDeps[0] = "maxvec"; + m_builtinDependencies["max"] = maxDeps; + } + { + std::vector<std::string> minDeps(1); + minDeps[0] = "minvec"; + m_builtinDependencies["min"] = minDeps; + } + { + std::vector<std::string> litDeps(4); + litDeps[0] = "minvec"; + litDeps[1] = "maxvec"; + litDeps[2] = "powf"; + litDeps[3] = "powvec"; + m_builtinDependencies["lit"] = litDeps; + } + { + std::vector<std::string> rsqDeps(4); + rsqDeps[0] = "sqrtf"; + rsqDeps[1] = "sqrtvec"; + rsqDeps[2] = "fabsf"; + rsqDeps[3] = "absvec"; + m_builtinDependencies["rsq"] = rsqDeps; + } } llvm::Function * InstructionsSoa::function(int op) @@ -162,9 +232,13 @@ llvm::Function * InstructionsSoa::function(int op) std::string name = m_functionsMap[op]; + std::cout <<"For op = "<<op<<", func is '"<<name<<"'"<<std::endl; + std::vector<std::string> deps = m_builtinDependencies[name]; for (unsigned int i = 0; i < deps.size(); ++i) { - injectFunction(m_builtins->getFunction(deps[i])); + llvm::Function *func = m_builtins->getFunction(deps[i]); + std::cout <<"\tinjecting dep = '"<<func->getName()<<"'"<<std::endl; + injectFunction(func); } llvm::Function *originalFunc = m_builtins->getFunction(name); @@ -183,10 +257,22 @@ llvm::Module * InstructionsSoa::currentModule() const void InstructionsSoa::createBuiltins() { - m_builtins = createSoaBuiltins(); + MemoryBuffer *buffer = MemoryBuffer::getMemBuffer( + (const char*)&soabuiltins_data[0], + (const char*)&soabuiltins_data[Elements(soabuiltins_data)]); + m_builtins = ParseBitcodeFile(buffer); + std::cout<<"Builtins created at "<<m_builtins<<std::endl; + assert(m_builtins); createDependencies(); } + +std::vector<llvm::Value*> InstructionsSoa::abs(const std::vector<llvm::Value*> in1) +{ + llvm::Function *func = function(TGSI_OPCODE_ABS); + return callBuiltin(func, in1); +} + std::vector<llvm::Value*> InstructionsSoa::dp3(const std::vector<llvm::Value*> in1, const std::vector<llvm::Value*> in2) { @@ -204,32 +290,32 @@ llvm::Value * InstructionsSoa::allocaTemp() std::vector<Value*> indices; indices.push_back(m_storage->constantInt(0)); indices.push_back(m_storage->constantInt(0)); - GetElementPtrInst *getElem = new GetElementPtrInst(alloca, - indices.begin(), - indices.end(), - name("allocaPtr"), - m_builder.GetInsertBlock()); + GetElementPtrInst *getElem = GetElementPtrInst::Create(alloca, + indices.begin(), + indices.end(), + name("allocaPtr"), + m_builder.GetInsertBlock()); return getElem; } std::vector<llvm::Value*> InstructionsSoa::allocaToResult(llvm::Value *allocaPtr) { - GetElementPtrInst *xElemPtr = new GetElementPtrInst(allocaPtr, - m_storage->constantInt(0), - name("xPtr"), - m_builder.GetInsertBlock()); - GetElementPtrInst *yElemPtr = new GetElementPtrInst(allocaPtr, - m_storage->constantInt(1), - name("yPtr"), - m_builder.GetInsertBlock()); - GetElementPtrInst *zElemPtr = new GetElementPtrInst(allocaPtr, - m_storage->constantInt(2), - name("zPtr"), - m_builder.GetInsertBlock()); - GetElementPtrInst *wElemPtr = new GetElementPtrInst(allocaPtr, - m_storage->constantInt(3), - name("wPtr"), - m_builder.GetInsertBlock()); + GetElementPtrInst *xElemPtr = GetElementPtrInst::Create(allocaPtr, + m_storage->constantInt(0), + name("xPtr"), + m_builder.GetInsertBlock()); + GetElementPtrInst *yElemPtr = GetElementPtrInst::Create(allocaPtr, + m_storage->constantInt(1), + name("yPtr"), + m_builder.GetInsertBlock()); + GetElementPtrInst *zElemPtr = GetElementPtrInst::Create(allocaPtr, + m_storage->constantInt(2), + name("zPtr"), + m_builder.GetInsertBlock()); + GetElementPtrInst *wElemPtr = GetElementPtrInst::Create(allocaPtr, + m_storage->constantInt(3), + name("wPtr"), + m_builder.GetInsertBlock()); std::vector<llvm::Value*> res(4); res[0] = new LoadInst(xElemPtr, name("xRes"), false, m_builder.GetInsertBlock()); @@ -320,6 +406,21 @@ std::vector<llvm::Value*> InstructionsSoa::pow(const std::vector<llvm::Value*> i return callBuiltin(func, in1, in2); } +std::vector<llvm::Value*> InstructionsSoa::min(const std::vector<llvm::Value*> in1, + const std::vector<llvm::Value*> in2) +{ + llvm::Function *func = function(TGSI_OPCODE_MIN); + return callBuiltin(func, in1, in2); +} + + +std::vector<llvm::Value*> InstructionsSoa::max(const std::vector<llvm::Value*> in1, + const std::vector<llvm::Value*> in2) +{ + llvm::Function *func = function(TGSI_OPCODE_MAX); + return callBuiltin(func, in1, in2); +} + void checkFunction(Function *func) { for (Function::const_iterator BI = func->begin(), BE = func->end(); @@ -354,29 +455,56 @@ void InstructionsSoa::injectFunction(llvm::Function *originalFunc, int op) } llvm::Function *func = 0; if (originalFunc->isDeclaration()) { - std::cout << "function decleration" <<std::endl; - func = new Function(originalFunc->getFunctionType(), GlobalValue::ExternalLinkage, - originalFunc->getName(), currentModule()); + func = Function::Create(originalFunc->getFunctionType(), GlobalValue::ExternalLinkage, + originalFunc->getName(), currentModule()); func->setCallingConv(CallingConv::C); - const ParamAttrsList *pal = 0; + const PAListPtr pal; func->setParamAttrs(pal); currentModule()->dump(); } else { DenseMap<const Value*, Value *> val; + val[m_builtins->getFunction("fabsf")] = currentModule()->getFunction("fabsf"); val[m_builtins->getFunction("powf")] = currentModule()->getFunction("powf"); + val[m_builtins->getFunction("sqrtf")] = currentModule()->getFunction("sqrtf"); + func = CloneFunction(originalFunc, val); +#if 0 std::cout <<" replacing "<<m_builtins->getFunction("powf") <<", with " <<currentModule()->getFunction("powf")<<std::endl; - func = CloneFunction(originalFunc, val); std::cout<<"1111-------------------------------"<<std::endl; checkFunction(originalFunc); std::cout<<"2222-------------------------------"<<std::endl; checkFunction(func); std::cout <<"XXXX = " <<val[m_builtins->getFunction("powf")]<<std::endl; +#endif currentModule()->getFunctionList().push_back(func); - std::cout << "Func parent is "<<func->getParent() - <<", cur is "<<currentModule() <<std::endl; } if (op != TGSI_OPCODE_LAST) { m_functions[op] = func; } } + +std::vector<llvm::Value*> InstructionsSoa::sub(const std::vector<llvm::Value*> in1, + const std::vector<llvm::Value*> in2) +{ + std::vector<llvm::Value*> res(4); + + res[0] = m_builder.CreateSub(in1[0], in2[0], name("subx")); + res[1] = m_builder.CreateSub(in1[1], in2[1], name("suby")); + res[2] = m_builder.CreateSub(in1[2], in2[2], name("subz")); + res[3] = m_builder.CreateSub(in1[3], in2[3], name("subw")); + + return res; +} + +std::vector<llvm::Value*> InstructionsSoa::lit(const std::vector<llvm::Value*> in) +{ + llvm::Function *func = function(TGSI_OPCODE_LIT); + return callBuiltin(func, in); +} + +std::vector<llvm::Value*> InstructionsSoa::rsq(const std::vector<llvm::Value*> in) +{ + llvm::Function *func = function(TGSI_OPCODE_RSQ); + return callBuiltin(func, in); +} + diff --git a/src/gallium/auxiliary/gallivm/instructionssoa.h b/src/gallium/auxiliary/gallivm/instructionssoa.h index b9104ea286..3e20b652dd 100644 --- a/src/gallium/auxiliary/gallivm/instructionssoa.h +++ b/src/gallium/auxiliary/gallivm/instructionssoa.h @@ -29,7 +29,7 @@ #define INSTRUCTIONSSOA_H #include <pipe/p_shader_tokens.h> -#include <llvm/Support/LLVMBuilder.h> +#include <llvm/Support/IRBuilder.h> #include <map> #include <vector> @@ -48,6 +48,7 @@ public: InstructionsSoa(llvm::Module *mod, llvm::Function *func, llvm::BasicBlock *block, StorageSoa *storage); + std::vector<llvm::Value*> abs(const std::vector<llvm::Value*> in1); std::vector<llvm::Value*> arl(const std::vector<llvm::Value*> in); std::vector<llvm::Value*> add(const std::vector<llvm::Value*> in1, const std::vector<llvm::Value*> in2); @@ -55,13 +56,21 @@ public: const std::vector<llvm::Value*> in2); std::vector<llvm::Value*> dp4(const std::vector<llvm::Value*> in1, const std::vector<llvm::Value*> in2); + std::vector<llvm::Value*> lit(const std::vector<llvm::Value*> in); std::vector<llvm::Value*> madd(const std::vector<llvm::Value*> in1, const std::vector<llvm::Value*> in2, const std::vector<llvm::Value*> in3); + std::vector<llvm::Value*> max(const std::vector<llvm::Value*> in1, + const std::vector<llvm::Value*> in2); + std::vector<llvm::Value*> min(const std::vector<llvm::Value*> in1, + const std::vector<llvm::Value*> in2); std::vector<llvm::Value*> mul(const std::vector<llvm::Value*> in1, const std::vector<llvm::Value*> in2); std::vector<llvm::Value*> pow(const std::vector<llvm::Value*> in1, const std::vector<llvm::Value*> in2); + std::vector<llvm::Value*> rsq(const std::vector<llvm::Value*> in1); + std::vector<llvm::Value*> sub(const std::vector<llvm::Value*> in1, + const std::vector<llvm::Value*> in2); void end(); std::vector<llvm::Value*> extractVector(llvm::Value *vector); @@ -87,7 +96,7 @@ private: const std::vector<llvm::Value*> in3); void injectFunction(llvm::Function *originalFunc, int op = TGSI_OPCODE_LAST); private: - llvm::LLVMFoldingBuilder m_builder; + llvm::IRBuilder m_builder; StorageSoa *m_storage; std::map<int, std::string> m_functionsMap; diff --git a/src/gallium/auxiliary/gallivm/llvm_builtins.c b/src/gallium/auxiliary/gallivm/llvm_builtins.c index 4f98d754ba..6b9d626ed4 100644 --- a/src/gallium/auxiliary/gallivm/llvm_builtins.c +++ b/src/gallium/auxiliary/gallivm/llvm_builtins.c @@ -1,4 +1,3 @@ -/*clang --emit-llvm llvm_builtins.c |llvm-as|opt -std-compile-opts|llvm2cpp -gen-contents -o=gallivm_builtins.cpp -f -for=shader -funcname=createGallivmBuiltins*/ /************************************************************************** * * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. @@ -30,7 +29,7 @@ * Authors: * Zack Rusin zack@tungstengraphics.com */ -typedef __attribute__(( ocu_vector_type(4) )) float float4; +typedef __attribute__(( ext_vector_type(4) )) float float4; extern float powf(float a, float b); diff --git a/src/gallium/auxiliary/gallivm/soabuiltins.c b/src/gallium/auxiliary/gallivm/soabuiltins.c index 4d658be520..78f84510e2 100644 --- a/src/gallium/auxiliary/gallivm/soabuiltins.c +++ b/src/gallium/auxiliary/gallivm/soabuiltins.c @@ -31,7 +31,30 @@ * Authors: * Zack Rusin zack@tungstengraphics.com */ -typedef __attribute__(( ocu_vector_type(4) )) float float4; +typedef __attribute__(( ext_vector_type(4) )) float float4; + + +extern float fabsf(float val); + +float4 absvec(float4 vec) +{ + float4 res; + res.x = fabsf(vec.x); + res.y = fabsf(vec.y); + res.z = fabsf(vec.z); + res.w = fabsf(vec.w); + + return res; +} + +void abs(float4 *res, + float4 tmp0x, float4 tmp0y, float4 tmp0z, float4 tmp0w) +{ + res[0] = absvec(tmp0x); + res[1] = absvec(tmp0y); + res[2] = absvec(tmp0z); + res[3] = absvec(tmp0w); +} void dp3(float4 *res, float4 tmp0x, float4 tmp0y, float4 tmp0z, float4 tmp0w, @@ -61,30 +84,104 @@ void dp4(float4 *res, } extern float powf(float num, float p); +extern float sqrtf(float x); + +float4 powvec(float4 vec, float4 q) +{ + float4 p; + p.x = powf(vec.x, q.x); + p.y = powf(vec.y, q.y); + p.z = powf(vec.z, q.z); + p.w = powf(vec.w, q.w); + return p; +} void pow(float4 *res, float4 tmp0x, float4 tmp0y, float4 tmp0z, float4 tmp0w, float4 tmp1x, float4 tmp1y, float4 tmp1z, float4 tmp1w) { - float4 p; - p.x = powf(tmp0x.x, tmp1x.x); - p.y = powf(tmp0x.y, tmp1x.y); - p.z = powf(tmp0x.z, tmp1x.z); - p.w = powf(tmp0x.w, tmp1x.w); - - res[0] = p; - res[1] = p; - res[2] = p; - res[3] = p; + res[0] = powvec(tmp0x, tmp1x); + res[1] = res[0]; + res[2] = res[0]; + res[3] = res[0]; +} + +float4 minvec(float4 a, float4 b) +{ + return (float4){(a.x < b.x) ? a.x : b.x, + (a.y < b.y) ? a.y : b.y, + (a.z < b.z) ? a.z : b.z, + (a.w < b.w) ? a.w : b.w}; +} + +void min(float4 *res, + float4 tmp0x, float4 tmp0y, float4 tmp0z, float4 tmp0w, + float4 tmp1x, float4 tmp1y, float4 tmp1z, float4 tmp1w) +{ + res[0] = minvec(tmp0x, tmp1x); + res[1] = minvec(tmp0y, tmp1y); + res[2] = minvec(tmp0z, tmp1z); + res[3] = minvec(tmp0w, tmp1w); +} + + +float4 maxvec(float4 a, float4 b) +{ + return (float4){(a.x > b.x) ? a.x : b.x, + (a.y > b.y) ? a.y : b.y, + (a.z > b.z) ? a.z : b.z, + (a.w > b.w) ? a.w : b.w}; +} + +void max(float4 *res, + float4 tmp0x, float4 tmp0y, float4 tmp0z, float4 tmp0w, + float4 tmp1x, float4 tmp1y, float4 tmp1z, float4 tmp1w) +{ + res[0] = maxvec(tmp0x, tmp1x); + res[1] = maxvec(tmp0y, tmp1y); + res[2] = maxvec(tmp0z, tmp1z); + res[3] = maxvec(tmp0w, tmp1w); +} + + +void lit(float4 *res, + float4 tmp0x, float4 tmp0y, float4 tmp0z, float4 tmp0w) +{ + const float4 zerovec = (float4) {0.0, 0.0, 0.0, 0.0}; + const float4 min128 = (float4) {-128.f, -128.f, -128.f, -128.f}; + const float4 plus128 = (float4) {128.f, 128.f, 128.f, 128.f}; + + res[0] = (float4){1.0, 1.0, 1.0, 1.0}; + if (tmp0x.x > 0) { + float4 tmpy = maxvec(tmp0y, zerovec); + float4 tmpw = minvec(tmp0w, plus128); + tmpw = maxvec(tmpw, min128); + res[1] = tmp0x; + res[2] = powvec(tmpy, tmpw); + } else { + res[1] = zerovec; + res[2] = zerovec; + } + res[3] = (float4){1.0, 1.0, 1.0, 1.0}; } -#if 0 -void yo(float4 *out, float4 *in) + +float4 sqrtvec(float4 vec) { - float4 res[4]; + float4 p; + p.x = sqrtf(vec.x); + p.y = sqrtf(vec.y); + p.z = sqrtf(vec.z); + p.w = sqrtf(vec.w); + return p; +} - dp3(res, in[0], in[1], in[2], in[3], - in[4], in[5], in[6], in[7]); - out[1] = res[1]; +void rsq(float4 *res, + float4 tmp0x, float4 tmp0y, float4 tmp0z, float4 tmp0w) +{ + const float4 onevec = (float4) {1., 1., 1., 1.}; + res[0] = onevec/sqrtvec(absvec(tmp0x)); + res[1] = onevec/sqrtvec(absvec(tmp0y)); + res[2] = onevec/sqrtvec(absvec(tmp0z)); + res[3] = onevec/sqrtvec(absvec(tmp0w)); } -#endif diff --git a/src/gallium/auxiliary/gallivm/storage.cpp b/src/gallium/auxiliary/gallivm/storage.cpp index c4326de8c5..6f373f6dd5 100644 --- a/src/gallium/auxiliary/gallivm/storage.cpp +++ b/src/gallium/auxiliary/gallivm/storage.cpp @@ -69,10 +69,10 @@ llvm::Constant *Storage::shuffleMask(int vec) { if (!m_extSwizzleVec) { std::vector<Constant*> elems; - elems.push_back(ConstantFP::get(Type::FloatTy, APFloat(0.f))); - elems.push_back(ConstantFP::get(Type::FloatTy, APFloat(1.f))); - elems.push_back(ConstantFP::get(Type::FloatTy, APFloat(0.f))); - elems.push_back(ConstantFP::get(Type::FloatTy, APFloat(1.f))); + elems.push_back(ConstantFP::get(APFloat(0.f))); + elems.push_back(ConstantFP::get(APFloat(1.f))); + elems.push_back(ConstantFP::get(APFloat(0.f))); + elems.push_back(ConstantFP::get(APFloat(1.f))); m_extSwizzleVec = ConstantVector::get(m_floatVecType, elems); } @@ -186,26 +186,26 @@ llvm::Value *Storage::maskWrite(llvm::Value *src, int mask, llvm::Value *templ) if ((mask & TGSI_WRITEMASK_X)) { llvm::Value *x = new ExtractElementInst(src, unsigned(0), name("x"), m_block); - dst = new InsertElementInst(dst, x, unsigned(0), - name("dstx"), m_block); + dst = InsertElementInst::Create(dst, x, unsigned(0), + name("dstx"), m_block); } if ((mask & TGSI_WRITEMASK_Y)) { llvm::Value *y = new ExtractElementInst(src, unsigned(1), name("y"), m_block); - dst = new InsertElementInst(dst, y, unsigned(1), - name("dsty"), m_block); + dst = InsertElementInst::Create(dst, y, unsigned(1), + name("dsty"), m_block); } if ((mask & TGSI_WRITEMASK_Z)) { llvm::Value *z = new ExtractElementInst(src, unsigned(2), name("z"), m_block); - dst = new InsertElementInst(dst, z, unsigned(2), - name("dstz"), m_block); + dst = InsertElementInst::Create(dst, z, unsigned(2), + name("dstz"), m_block); } if ((mask & TGSI_WRITEMASK_W)) { llvm::Value *w = new ExtractElementInst(src, unsigned(3), name("w"), m_block); - dst = new InsertElementInst(dst, w, unsigned(3), - name("dstw"), m_block); + dst = InsertElementInst::Create(dst, w, unsigned(3), + name("dstw"), m_block); } return dst; } @@ -295,10 +295,10 @@ llvm::Value * Storage::immediateElement(int idx) void Storage::addImmediate(float *val) { std::vector<Constant*> vec(4); - vec[0] = ConstantFP::get(Type::FloatTy, APFloat(val[0])); - vec[1] = ConstantFP::get(Type::FloatTy, APFloat(val[1])); - vec[2] = ConstantFP::get(Type::FloatTy, APFloat(val[2])); - vec[3] = ConstantFP::get(Type::FloatTy, APFloat(val[3])); + vec[0] = ConstantFP::get(APFloat(val[0])); + vec[1] = ConstantFP::get(APFloat(val[1])); + vec[2] = ConstantFP::get(APFloat(val[2])); + vec[3] = ConstantFP::get(APFloat(val[3])); m_immediates.push_back(ConstantVector::get(m_floatVecType, vec)); } @@ -308,11 +308,11 @@ llvm::Value * Storage::elemPtr(Args arg) std::vector<Value*> indices; indices.push_back(constantInt(0)); indices.push_back(constantInt(static_cast<int>(arg))); - GetElementPtrInst *getElem = new GetElementPtrInst(m_INPUT, - indices.begin(), - indices.end(), - name("input_ptr"), - m_block); + GetElementPtrInst *getElem = GetElementPtrInst::Create(m_INPUT, + indices.begin(), + indices.end(), + name("input_ptr"), + m_block); return new LoadInst(getElem, name("input_field"), false, m_block); } @@ -322,7 +322,7 @@ llvm::Value * Storage::elemIdx(llvm::Value *ptr, int idx, GetElementPtrInst *getElem = 0; if (indIdx) { - getElem = new GetElementPtrInst(ptr, + getElem = GetElementPtrInst::Create(ptr, BinaryOperator::create(Instruction::Add, indIdx, constantInt(idx), @@ -331,7 +331,7 @@ llvm::Value * Storage::elemIdx(llvm::Value *ptr, int idx, name("field"), m_block); } else { - getElem = new GetElementPtrInst(ptr, + getElem = GetElementPtrInst::Create(ptr, constantInt(idx), name("field"), m_block); @@ -350,7 +350,7 @@ void Storage::setKilElement(llvm::Value *val) std::vector<Value*> indices; indices.push_back(constantInt(0)); indices.push_back(constantInt(static_cast<int>(KilArg))); - GetElementPtrInst *elem = new GetElementPtrInst(m_INPUT, + GetElementPtrInst *elem = GetElementPtrInst::Create(m_INPUT, indices.begin(), indices.end(), name("kil_ptr"), diff --git a/src/gallium/auxiliary/gallivm/storagesoa.cpp b/src/gallium/auxiliary/gallivm/storagesoa.cpp index bb6fe3d7e1..78d754371f 100644 --- a/src/gallium/auxiliary/gallivm/storagesoa.cpp +++ b/src/gallium/auxiliary/gallivm/storagesoa.cpp @@ -207,11 +207,11 @@ llvm::Value * StorageSoa::elementPointer(llvm::Value *ptr, llvm::Value *index, indices.push_back(index); indices.push_back(constantInt(channel)); - GetElementPtrInst *getElem = new GetElementPtrInst(ptr, - indices.begin(), - indices.end(), - name("ptr"), - m_block); + GetElementPtrInst *getElem = GetElementPtrInst::Create(ptr, + indices.begin(), + indices.end(), + name("ptr"), + m_block); return getElem; } @@ -264,10 +264,10 @@ llvm::Constant * StorageSoa::createConstGlobalVector(const std::vector<float> &v { VectorType *vectorType = VectorType::get(Type::FloatTy, 4); std::vector<Constant*> immValues; - ConstantFP *constx = ConstantFP::get(Type::FloatTy, APFloat(vec[0])); - ConstantFP *consty = ConstantFP::get(Type::FloatTy, APFloat(vec[1])); - ConstantFP *constz = ConstantFP::get(Type::FloatTy, APFloat(vec[2])); - ConstantFP *constw = ConstantFP::get(Type::FloatTy, APFloat(vec[3])); + ConstantFP *constx = ConstantFP::get(APFloat(vec[0])); + ConstantFP *consty = ConstantFP::get(APFloat(vec[1])); + ConstantFP *constz = ConstantFP::get(APFloat(vec[2])); + ConstantFP *constw = ConstantFP::get(APFloat(vec[3])); immValues.push_back(constx); immValues.push_back(consty); immValues.push_back(constz); diff --git a/src/gallium/auxiliary/gallivm/tgsitollvm.cpp b/src/gallium/auxiliary/gallivm/tgsitollvm.cpp index ab9e7a06fb..9695358ab8 100644 --- a/src/gallium/auxiliary/gallivm/tgsitollvm.cpp +++ b/src/gallium/auxiliary/gallivm/tgsitollvm.cpp @@ -692,12 +692,14 @@ translate_instructionir(llvm::Module *module, } break; case TGSI_OPCODE_LIT: { + out = instr->lit(inputs[0]); } break; case TGSI_OPCODE_RCP: { } break; case TGSI_OPCODE_RSQ: { + out = instr->rsq(inputs[0]); } break; case TGSI_OPCODE_EXP: @@ -724,9 +726,11 @@ translate_instructionir(llvm::Module *module, } break; case TGSI_OPCODE_MIN: { + out = instr->min(inputs[0], inputs[1]); } break; case TGSI_OPCODE_MAX: { + out = instr->max(inputs[0], inputs[1]); } break; case TGSI_OPCODE_SLT: { @@ -740,6 +744,7 @@ translate_instructionir(llvm::Module *module, } break; case TGSI_OPCODE_SUB: { + out = instr->sub(inputs[0], inputs[1]); } break; case TGSI_OPCODE_LERP: { @@ -781,6 +786,7 @@ translate_instructionir(llvm::Module *module, case TGSI_OPCODE_MULTIPLYMATRIX: break; case TGSI_OPCODE_ABS: { + out = instr->abs(inputs[0]); } break; case TGSI_OPCODE_RCC: @@ -1014,7 +1020,7 @@ tgsi_to_llvm(struct gallivm_ir *ir, const struct tgsi_token *tokens) Value *ptr_INPUT = args++; ptr_INPUT->setName("input"); - BasicBlock *label_entry = new BasicBlock("entry", shader, 0); + BasicBlock *label_entry = BasicBlock::Create("entry", shader, 0); tgsi_parse_init(&parse, tokens); @@ -1085,7 +1091,7 @@ llvm::Module * tgsi_to_llvmir(struct gallivm_ir *ir, Value *temps = args++; temps->setName("temps"); - BasicBlock *label_entry = new BasicBlock("entry", shader, 0); + BasicBlock *label_entry = BasicBlock::Create("entry", shader, 0); tgsi_parse_init(&parse, tokens); diff --git a/src/gallium/auxiliary/pipebuffer/Makefile b/src/gallium/auxiliary/pipebuffer/Makefile index a9fa518c67..d654dbcc91 100644 --- a/src/gallium/auxiliary/pipebuffer/Makefile +++ b/src/gallium/auxiliary/pipebuffer/Makefile @@ -6,9 +6,11 @@ LIBNAME = pipebuffer C_SOURCES = \ pb_buffer_fenced.c \ pb_buffer_malloc.c \ + pb_bufmgr_cache.c \ pb_bufmgr_fenced.c \ pb_bufmgr_mm.c \ pb_bufmgr_pool.c \ + pb_bufmgr_slab.c \ pb_winsys.c include ../../Makefile.template diff --git a/src/gallium/auxiliary/pipebuffer/SConscript b/src/gallium/auxiliary/pipebuffer/SConscript index 3d41fd84a6..604a217982 100644 --- a/src/gallium/auxiliary/pipebuffer/SConscript +++ b/src/gallium/auxiliary/pipebuffer/SConscript @@ -5,9 +5,11 @@ pipebuffer = env.ConvenienceLibrary( source = [ 'pb_buffer_fenced.c', 'pb_buffer_malloc.c', + 'pb_bufmgr_cache.c', 'pb_bufmgr_fenced.c', 'pb_bufmgr_mm.c', 'pb_bufmgr_pool.c', + 'pb_bufmgr_slab.c', 'pb_winsys.c', ]) diff --git a/src/gallium/auxiliary/pipebuffer/pb_buffer.h b/src/gallium/auxiliary/pipebuffer/pb_buffer.h index 4b09c80b2a..49705cb862 100644 --- a/src/gallium/auxiliary/pipebuffer/pb_buffer.h +++ b/src/gallium/auxiliary/pipebuffer/pb_buffer.h @@ -193,6 +193,17 @@ pb_reference(struct pb_buffer **dst, /** + * Utility function to check whether a requested alignment is consistent with + * the provided alignment or not. + */ +static INLINE int +pb_check_alignment(size_t requested, size_t provided) +{ + return requested <= provided && (provided % requested) == 0; +} + + +/** * Malloc-based buffer to store data that can't be used by the graphics * hardware. */ diff --git a/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c b/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c index 24ba61a0b7..2fa0842971 100644 --- a/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c +++ b/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c @@ -34,7 +34,14 @@ */ +#include "pipe/p_config.h" + +#if defined(PIPE_OS_LINUX) +#include <unistd.h> +#endif + #include "pipe/p_compiler.h" +#include "pipe/p_error.h" #include "pipe/p_debug.h" #include "pipe/p_winsys.h" #include "pipe/p_thread.h" @@ -44,9 +51,6 @@ #include "pb_buffer.h" #include "pb_buffer_fenced.h" -#ifndef WIN32 -#include <unistd.h> -#endif /** @@ -54,6 +58,13 @@ */ #define SUPER(__derived) (&(__derived)->base) +#define PIPE_BUFFER_USAGE_CPU_READ_WRITE \ + ( PIPE_BUFFER_USAGE_CPU_READ | PIPE_BUFFER_USAGE_CPU_WRITE ) +#define PIPE_BUFFER_USAGE_GPU_READ_WRITE \ + ( PIPE_BUFFER_USAGE_GPU_READ | PIPE_BUFFER_USAGE_GPU_WRITE ) +#define PIPE_BUFFER_USAGE_WRITE \ + ( PIPE_BUFFER_USAGE_CPU_WRITE | PIPE_BUFFER_USAGE_GPU_WRITE ) + struct fenced_buffer_list { @@ -76,6 +87,15 @@ struct fenced_buffer struct pb_buffer *buffer; + /* FIXME: protect access with mutex */ + + /** + * A bitmask of PIPE_BUFFER_USAGE_CPU/GPU_READ/WRITE describing the current + * buffer usage. + */ + unsigned flags; + + unsigned mapcount; struct pipe_fence_handle *fence; struct list_head head; @@ -97,7 +117,10 @@ _fenced_buffer_add(struct fenced_buffer *fenced_buf) { struct fenced_buffer_list *fenced_list = fenced_buf->list; + assert(fenced_buf->base.base.refcount); + assert(fenced_buf->flags & PIPE_BUFFER_USAGE_GPU_READ_WRITE); assert(fenced_buf->fence); + assert(!fenced_buf->head.prev); assert(!fenced_buf->head.next); LIST_ADDTAIL(&fenced_buf->head, &fenced_list->delayed); @@ -111,8 +134,6 @@ _fenced_buffer_add(struct fenced_buffer *fenced_buf) static INLINE void _fenced_buffer_destroy(struct fenced_buffer *fenced_buf) { - struct fenced_buffer_list *fenced_list = fenced_buf->list; - assert(!fenced_buf->base.base.refcount); assert(!fenced_buf->fence); pb_reference(&fenced_buf->buffer, NULL); @@ -128,8 +149,8 @@ _fenced_buffer_remove(struct fenced_buffer *fenced_buf) assert(fenced_buf->fence); - assert(winsys->fence_signalled(winsys, fenced_buf->fence, 0) == 0); winsys->fence_reference(winsys, &fenced_buf->fence, NULL); + fenced_buf->flags &= ~PIPE_BUFFER_USAGE_GPU_READ_WRITE; assert(fenced_buf->head.prev); assert(fenced_buf->head.next); @@ -174,6 +195,9 @@ _fenced_buffer_list_check_free(struct fenced_buffer_list *fenced_list, break; prev_fence = fenced_buf->fence; } + else { + assert(winsys->fence_signalled(winsys, fenced_buf->fence, 0) == 0); + } _fenced_buffer_remove(fenced_buf); @@ -183,6 +207,40 @@ _fenced_buffer_list_check_free(struct fenced_buffer_list *fenced_list, } +/** + * Serialize writes, but allow concurrent reads. + */ +static INLINE enum pipe_error +fenced_buffer_serialize(struct fenced_buffer *fenced_buf, unsigned flags) +{ + struct fenced_buffer_list *fenced_list = fenced_buf->list; + struct pipe_winsys *winsys = fenced_list->winsys; + + /* Allow concurrent reads */ + if(((fenced_buf->flags | flags) & PIPE_BUFFER_USAGE_WRITE) == 0) + return PIPE_OK; + + /* Wait for the CPU to finish */ + if(fenced_buf->mapcount) { + /* FIXME: Use thread conditions variables to signal when mapcount + * reaches zero */ + debug_warning("attemp to write concurrently to buffer"); + /* XXX: we must not fail here in order to support texture mipmap generation + return PIPE_ERROR_RETRY; + */ + } + + /* Wait for the GPU to finish */ + if(fenced_buf->fence) { + if(winsys->fence_finish(winsys, fenced_buf->fence, 0) != 0) + return PIPE_ERROR_RETRY; + _fenced_buffer_remove(fenced_buf); + } + + return PIPE_OK; +} + + static void fenced_buffer_destroy(struct pb_buffer *buf) { @@ -199,6 +257,7 @@ fenced_buffer_destroy(struct pb_buffer *buf) prev = curr->prev; do { fenced_buf = LIST_ENTRY(struct fenced_buffer, curr, head); + assert(winsys->fence_signalled(winsys, fenced_buf->fence, 0) == 0); _fenced_buffer_remove(fenced_buf); curr = prev; prev = curr->prev; @@ -219,16 +278,30 @@ static void * fenced_buffer_map(struct pb_buffer *buf, unsigned flags) { - struct fenced_buffer *fenced_buf = fenced_buffer(buf); - return pb_map(fenced_buf->buffer, flags); + struct fenced_buffer *fenced_buf = fenced_buffer(buf); + void *map; + assert((flags & ~PIPE_BUFFER_USAGE_CPU_READ_WRITE) == 0); + + if(fenced_buffer_serialize(fenced_buf, flags) != PIPE_OK) + return NULL; + + map = pb_map(fenced_buf->buffer, flags); + if(map) + ++fenced_buf->mapcount; + fenced_buf->flags |= flags & PIPE_BUFFER_USAGE_CPU_READ_WRITE; + return map; } static void fenced_buffer_unmap(struct pb_buffer *buf) { - struct fenced_buffer *fenced_buf = fenced_buffer(buf); + struct fenced_buffer *fenced_buf = fenced_buffer(buf); + assert(fenced_buf->mapcount); pb_unmap(fenced_buf->buffer); + --fenced_buf->mapcount; + if(!fenced_buf->mapcount) + fenced_buf->flags &= ~PIPE_BUFFER_USAGE_CPU_READ_WRITE; } @@ -281,16 +354,49 @@ void buffer_fence(struct pb_buffer *buf, struct pipe_fence_handle *fence) { - struct fenced_buffer *fenced_buf = fenced_buffer(buf); - struct fenced_buffer_list *fenced_list = fenced_buf->list; - struct pipe_winsys *winsys = fenced_list->winsys; + struct fenced_buffer *fenced_buf; + struct fenced_buffer_list *fenced_list; + struct pipe_winsys *winsys; + /* FIXME: receive this as a parameter */ + unsigned flags = fence ? PIPE_BUFFER_USAGE_GPU_READ_WRITE : 0; + + /* This is a public function, so be extra cautious with the buffer passed, + * as happens frequently to receive null buffers, or pointer to buffers + * other than fenced buffers. */ + assert(buf); + if(!buf) + return; + assert(buf->vtbl == &fenced_buffer_vtbl); + if(buf->vtbl != &fenced_buffer_vtbl) + return; + + fenced_buf = fenced_buffer(buf); + fenced_list = fenced_buf->list; + winsys = fenced_list->winsys; + + if(fence == fenced_buf->fence) { + /* Handle the same fence case specially, not only because it is a fast + * path, but mostly to avoid serializing two writes with the same fence, + * as that would bring the hardware down to synchronous operation without + * any benefit. + */ + fenced_buf->flags |= flags & PIPE_BUFFER_USAGE_GPU_READ_WRITE; + return; + } + + if(fenced_buffer_serialize(fenced_buf, flags) != PIPE_OK) { + /* FIXME: propagate error */ + (void)0; + } _glthread_LOCK_MUTEX(fenced_list->mutex); if (fenced_buf->fence) _fenced_buffer_remove(fenced_buf); - winsys->fence_reference(winsys, &fenced_buf->fence, fence); - if (fenced_buf->fence) + if (fence) { + winsys->fence_reference(winsys, &fenced_buf->fence, fence); + fenced_buf->flags |= flags & PIPE_BUFFER_USAGE_GPU_READ_WRITE; _fenced_buffer_add(fenced_buf); + } _glthread_UNLOCK_MUTEX(fenced_list->mutex); } @@ -334,7 +440,7 @@ fenced_buffer_list_destroy(struct fenced_buffer_list *fenced_list) /* Wait on outstanding fences */ while (fenced_list->numDelayed) { _glthread_UNLOCK_MUTEX(fenced_list->mutex); -#ifndef WIN32 +#if defined(PIPE_OS_LINUX) sched_yield(); #endif _fenced_buffer_list_check_free(fenced_list, 1); diff --git a/src/gallium/auxiliary/pipebuffer/pb_bufmgr.h b/src/gallium/auxiliary/pipebuffer/pb_bufmgr.h index 0cf8e92e37..8de286e3f9 100644 --- a/src/gallium/auxiliary/pipebuffer/pb_bufmgr.h +++ b/src/gallium/auxiliary/pipebuffer/pb_bufmgr.h @@ -80,7 +80,7 @@ struct pb_manager /** - * Static buffer pool manager. + * Static buffer pool sub-allocator. * * Manages the allocation of equally sized buffers. It does so by allocating * a single big buffer and divide it equally sized buffers. @@ -94,7 +94,7 @@ pool_bufmgr_create(struct pb_manager *provider, /** - * Wraper around the old memory manager. + * Static sub-allocator based the old memory manager. * * It managers buffers of different sizes. It does so by allocating a buffer * with the size of the heap, and then using the old mm memory manager to manage @@ -114,6 +114,40 @@ mm_bufmgr_create_from_buffer(struct pb_buffer *buffer, size_t size, size_t align2); +/** + * Slab sub-allocator. + */ +struct pb_manager * +pb_slab_manager_create(struct pb_manager *provider, + size_t bufSize, + size_t slabSize, + const struct pb_desc *desc); + +/** + * Allow a range of buffer size, by aggregating multiple slabs sub-allocators + * with different bucket sizes. + */ +struct pb_manager * +pb_slab_range_manager_create(struct pb_manager *provider, + size_t minBufSize, + size_t maxBufSize, + size_t slabSize, + const struct pb_desc *desc); + + +/** + * Time-based buffer cache. + * + * This manager keeps a cache of destroyed buffers during a time interval. + */ +struct pb_manager * +pb_cache_manager_create(struct pb_manager *provider, + unsigned usecs); + +void +pb_cache_flush(struct pb_manager *mgr); + + /** * Fenced buffer manager. * diff --git a/src/gallium/auxiliary/pipebuffer/pb_bufmgr_cache.c b/src/gallium/auxiliary/pipebuffer/pb_bufmgr_cache.c new file mode 100644 index 0000000000..4bd3f94a6c --- /dev/null +++ b/src/gallium/auxiliary/pipebuffer/pb_bufmgr_cache.c @@ -0,0 +1,344 @@ +/************************************************************************** + * + * Copyright 2007-2008 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +/** + * \file + * Buffer cache. + * + * \author José Fonseca <jrfonseca-at-tungstengraphics-dot-com> + * \author Thomas Hellström <thomas-at-tungstengraphics-dot-com> + */ + + +#include "pipe/p_compiler.h" +#include "pipe/p_debug.h" +#include "pipe/p_winsys.h" +#include "pipe/p_thread.h" +#include "pipe/p_util.h" +#include "util/u_double_list.h" +#include "util/u_time.h" + +#include "pb_buffer.h" +#include "pb_bufmgr.h" + + +/** + * Convenience macro (type safe). + */ +#define SUPER(__derived) (&(__derived)->base) + + +struct pb_cache_manager; + + +/** + * Wrapper around a pipe buffer which adds delayed destruction. + */ +struct pb_cache_buffer +{ + struct pb_buffer base; + + struct pb_buffer *buffer; + struct pb_cache_manager *mgr; + + /** Caching time interval */ + struct util_time start, end; + + struct list_head head; +}; + + +struct pb_cache_manager +{ + struct pb_manager base; + + struct pb_manager *provider; + unsigned usecs; + + _glthread_Mutex mutex; + + struct list_head delayed; + size_t numDelayed; +}; + + +static INLINE struct pb_cache_buffer * +pb_cache_buffer(struct pb_buffer *buf) +{ + assert(buf); + return (struct pb_cache_buffer *)buf; +} + + +static INLINE struct pb_cache_manager * +pb_cache_manager(struct pb_manager *mgr) +{ + assert(mgr); + return (struct pb_cache_manager *)mgr; +} + + +/** + * Actually destroy the buffer. + */ +static INLINE void +_pb_cache_buffer_destroy(struct pb_cache_buffer *buf) +{ + struct pb_cache_manager *mgr = buf->mgr; + + LIST_DEL(&buf->head); + assert(mgr->numDelayed); + --mgr->numDelayed; + assert(!buf->base.base.refcount); + pb_reference(&buf->buffer, NULL); + FREE(buf); +} + + +/** + * Free as many cache buffers from the list head as possible. + */ +static void +_pb_cache_buffer_list_check_free(struct pb_cache_manager *mgr) +{ + struct list_head *curr, *next; + struct pb_cache_buffer *buf; + struct util_time now; + + util_time_get(&now); + + curr = mgr->delayed.next; + next = curr->next; + while(curr != &mgr->delayed) { + buf = LIST_ENTRY(struct pb_cache_buffer, curr, head); + + if(!util_time_timeout(&buf->start, &buf->end, &now)) + break; + + _pb_cache_buffer_destroy(buf); + + curr = next; + next = curr->next; + } +} + + +static void +pb_cache_buffer_destroy(struct pb_buffer *_buf) +{ + struct pb_cache_buffer *buf = pb_cache_buffer(_buf); + struct pb_cache_manager *mgr = buf->mgr; + + _glthread_LOCK_MUTEX(mgr->mutex); + assert(buf->base.base.refcount == 0); + + _pb_cache_buffer_list_check_free(mgr); + + util_time_get(&buf->start); + util_time_add(&buf->start, mgr->usecs, &buf->end); + LIST_ADDTAIL(&buf->head, &mgr->delayed); + ++mgr->numDelayed; + _glthread_UNLOCK_MUTEX(mgr->mutex); +} + + +static void * +pb_cache_buffer_map(struct pb_buffer *_buf, + unsigned flags) +{ + struct pb_cache_buffer *buf = pb_cache_buffer(_buf); + return pb_map(buf->buffer, flags); +} + + +static void +pb_cache_buffer_unmap(struct pb_buffer *_buf) +{ + struct pb_cache_buffer *buf = pb_cache_buffer(_buf); + pb_unmap(buf->buffer); +} + + +static void +pb_cache_buffer_get_base_buffer(struct pb_buffer *_buf, + struct pb_buffer **base_buf, + unsigned *offset) +{ + struct pb_cache_buffer *buf = pb_cache_buffer(_buf); + pb_get_base_buffer(buf->buffer, base_buf, offset); +} + + +const struct pb_vtbl +pb_cache_buffer_vtbl = { + pb_cache_buffer_destroy, + pb_cache_buffer_map, + pb_cache_buffer_unmap, + pb_cache_buffer_get_base_buffer +}; + + +static INLINE boolean +pb_cache_is_buffer_compat(struct pb_cache_buffer *buf, + size_t size, + const struct pb_desc *desc) +{ + /* TODO: be more lenient with size */ + if(buf->base.base.size != size) + return FALSE; + + if(!pb_check_alignment(desc->alignment, buf->base.base.alignment)) + return FALSE; + + /* XXX: check usage too? */ + + return TRUE; +} + + +static struct pb_buffer * +pb_cache_manager_create_buffer(struct pb_manager *_mgr, + size_t size, + const struct pb_desc *desc) +{ + struct pb_cache_manager *mgr = pb_cache_manager(_mgr); + struct pb_cache_buffer *buf; + struct pb_cache_buffer *curr_buf; + struct list_head *curr, *next; + struct util_time now; + + _glthread_LOCK_MUTEX(mgr->mutex); + + buf = NULL; + curr = mgr->delayed.next; + next = curr->next; + + /* search in the expired buffers, freeing them in the process */ + util_time_get(&now); + while(curr != &mgr->delayed) { + curr_buf = LIST_ENTRY(struct pb_cache_buffer, curr, head); + if(!buf && pb_cache_is_buffer_compat(curr_buf, size, desc)) + buf = curr_buf; + else if(util_time_timeout(&curr_buf->start, &curr_buf->end, &now)) + _pb_cache_buffer_destroy(curr_buf); + curr = next; + next = curr->next; + } + + /* keep searching in the hot buffers */ + while(!buf && curr != &mgr->delayed) { + curr_buf = LIST_ENTRY(struct pb_cache_buffer, curr, head); + if(pb_cache_is_buffer_compat(curr_buf, size, desc)) + buf = curr_buf; + curr = next; + next = curr->next; + } + + if(buf) { + LIST_DEL(&buf->head); + _glthread_UNLOCK_MUTEX(mgr->mutex); + ++buf->base.base.refcount; + return &buf->base; + } + + _glthread_UNLOCK_MUTEX(mgr->mutex); + + buf = CALLOC_STRUCT(pb_cache_buffer); + if(!buf) + return NULL; + + buf->buffer = mgr->provider->create_buffer(mgr->provider, size, desc); + if(!buf->buffer) { + FREE(buf); + return NULL; + } + + assert(buf->buffer->base.refcount >= 1); + assert(pb_check_alignment(desc->alignment, buf->buffer->base.alignment)); + assert((buf->buffer->base.usage & desc->usage) == desc->usage); + assert(buf->buffer->base.size >= size); + + buf->base.base.refcount = 1; + buf->base.base.alignment = buf->buffer->base.alignment; + buf->base.base.usage = buf->buffer->base.usage; + buf->base.base.size = buf->buffer->base.size; + + buf->base.vtbl = &pb_cache_buffer_vtbl; + buf->mgr = mgr; + + return &buf->base; +} + + +void +pb_cache_flush(struct pb_manager *_mgr) +{ + struct pb_cache_manager *mgr = pb_cache_manager(_mgr); + struct list_head *curr, *next; + struct pb_cache_buffer *buf; + + _glthread_LOCK_MUTEX(mgr->mutex); + curr = mgr->delayed.next; + next = curr->next; + while(curr != &mgr->delayed) { + buf = LIST_ENTRY(struct pb_cache_buffer, curr, head); + _pb_cache_buffer_destroy(buf); + curr = next; + next = curr->next; + } + _glthread_UNLOCK_MUTEX(mgr->mutex); +} + + +static void +pb_cache_manager_destroy(struct pb_manager *mgr) +{ + pb_cache_flush(mgr); + FREE(mgr); +} + + +struct pb_manager * +pb_cache_manager_create(struct pb_manager *provider, + unsigned usecs) +{ + struct pb_cache_manager *mgr; + + mgr = (struct pb_cache_manager *)CALLOC(1, sizeof(*mgr)); + if (!mgr) + return NULL; + + mgr->base.destroy = pb_cache_manager_destroy; + mgr->base.create_buffer = pb_cache_manager_create_buffer; + mgr->provider = provider; + mgr->usecs = usecs; + LIST_INITHEAD(&mgr->delayed); + mgr->numDelayed = 0; + _glthread_INIT_MUTEX(mgr->mutex); + + return &mgr->base; +} diff --git a/src/gallium/auxiliary/pipebuffer/pb_bufmgr_fenced.c b/src/gallium/auxiliary/pipebuffer/pb_bufmgr_fenced.c index bffca5b244..9d809e2f9b 100644 --- a/src/gallium/auxiliary/pipebuffer/pb_bufmgr_fenced.c +++ b/src/gallium/auxiliary/pipebuffer/pb_bufmgr_fenced.c @@ -30,7 +30,7 @@ * \file * A buffer manager that wraps buffers in fenced buffers. * - * \author José Fonseca <jrfonseca@tungstengraphics.dot.com> + * \author José Fonseca <jrfonseca@tungstengraphics.dot.com> */ @@ -101,7 +101,8 @@ fenced_bufmgr_destroy(struct pb_manager *mgr) fenced_buffer_list_destroy(fenced_mgr->fenced_list); - fenced_mgr->provider->destroy(fenced_mgr->provider); + if(fenced_mgr->provider) + fenced_mgr->provider->destroy(fenced_mgr->provider); FREE(fenced_mgr); } @@ -113,6 +114,9 @@ fenced_bufmgr_create(struct pb_manager *provider, { struct fenced_pb_manager *fenced_mgr; + if(!provider) + return NULL; + fenced_mgr = (struct fenced_pb_manager *)CALLOC(1, sizeof(*fenced_mgr)); if (!fenced_mgr) return NULL; diff --git a/src/gallium/auxiliary/pipebuffer/pb_bufmgr_slab.c b/src/gallium/auxiliary/pipebuffer/pb_bufmgr_slab.c new file mode 100644 index 0000000000..b931455056 --- /dev/null +++ b/src/gallium/auxiliary/pipebuffer/pb_bufmgr_slab.c @@ -0,0 +1,495 @@ +/************************************************************************** + * + * Copyright 2006-2008 Tungsten Graphics, Inc., Cedar Park, TX., USA + * All Rights Reserved. + * + * Permission is hereby granted, FREE of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * + **************************************************************************/ + +/** + * @file + * S-lab pool implementation. + * + * @author Thomas Hellstrom <thomas-at-tungstengraphics-dot-com> + * @author Jose Fonseca <jrfonseca@tungstengraphics.com> + */ + +#include "pipe/p_compiler.h" +#include "pipe/p_error.h" +#include "pipe/p_debug.h" +#include "pipe/p_thread.h" +#include "pipe/p_defines.h" +#include "pipe/p_util.h" +#include "util/u_double_list.h" +#include "util/u_time.h" + +#include "pb_buffer.h" +#include "pb_bufmgr.h" + + +#define DRI_SLABPOOL_ALLOC_RETRIES 100 + + +struct pb_slab; + +struct pb_slab_buffer +{ + struct pb_buffer base; + + struct pb_slab *slab; + struct list_head head; + unsigned mapCount; + size_t start; + _glthread_Cond event; +}; + +struct pb_slab +{ + struct list_head head; + struct list_head freeBuffers; + size_t numBuffers; + size_t numFree; + struct pb_slab_buffer *buffers; + struct pb_slab_manager *mgr; + + struct pb_buffer *bo; + void *virtual; +}; + +struct pb_slab_manager +{ + struct pb_manager base; + + struct pb_manager *provider; + size_t bufSize; + size_t slabSize; + struct pb_desc desc; + + struct list_head slabs; + struct list_head freeSlabs; + + _glthread_Mutex mutex; +}; + +/** + * The data of this structure remains constant after + * initialization and thus needs no mutex protection. + */ +struct pb_slab_range_manager +{ + struct pb_manager base; + + struct pb_manager *provider; + size_t minBufSize; + size_t maxBufSize; + struct pb_desc desc; + + unsigned numBuckets; + size_t *bucketSizes; + struct pb_manager **buckets; +}; + + +static INLINE struct pb_slab_buffer * +pb_slab_buffer(struct pb_buffer *buf) +{ + assert(buf); + return (struct pb_slab_buffer *)buf; +} + + +static INLINE struct pb_slab_manager * +pb_slab_manager(struct pb_manager *mgr) +{ + assert(mgr); + return (struct pb_slab_manager *)mgr; +} + + +static INLINE struct pb_slab_range_manager * +pb_slab_range_manager(struct pb_manager *mgr) +{ + assert(mgr); + return (struct pb_slab_range_manager *)mgr; +} + + +/** + * Delete a buffer from the slab delayed list and put + * it on the slab FREE list. + */ +static void +pb_slab_buffer_destroy(struct pb_buffer *_buf) +{ + struct pb_slab_buffer *buf = pb_slab_buffer(_buf); + struct pb_slab *slab = buf->slab; + struct pb_slab_manager *mgr = slab->mgr; + struct list_head *list = &buf->head; + + _glthread_LOCK_MUTEX(mgr->mutex); + + assert(buf->base.base.refcount == 0); + + buf->mapCount = 0; + + LIST_DEL(list); + LIST_ADDTAIL(list, &slab->freeBuffers); + slab->numFree++; + + if (slab->head.next == &slab->head) + LIST_ADDTAIL(&slab->head, &mgr->slabs); + + if (slab->numFree == slab->numBuffers) { + list = &slab->head; + LIST_DEL(list); + LIST_ADDTAIL(list, &mgr->freeSlabs); + } + + if (mgr->slabs.next == &mgr->slabs || slab->numFree + != slab->numBuffers) { + + struct list_head *next; + + for (list = mgr->freeSlabs.next, next = list->next; list + != &mgr->freeSlabs; list = next, next = list->next) { + + slab = LIST_ENTRY(struct pb_slab, list, head); + + LIST_DELINIT(list); + pb_reference(&slab->bo, NULL); + FREE(slab->buffers); + FREE(slab); + } + } + + _glthread_UNLOCK_MUTEX(mgr->mutex); +} + + +static void * +pb_slab_buffer_map(struct pb_buffer *_buf, + unsigned flags) +{ + struct pb_slab_buffer *buf = pb_slab_buffer(_buf); + + ++buf->mapCount; + return (void *) ((uint8_t *) buf->slab->virtual + buf->start); +} + + +static void +pb_slab_buffer_unmap(struct pb_buffer *_buf) +{ + struct pb_slab_buffer *buf = pb_slab_buffer(_buf); + + --buf->mapCount; + if (buf->mapCount == 0) + _glthread_COND_BROADCAST(buf->event); +} + + +static void +pb_slab_buffer_get_base_buffer(struct pb_buffer *_buf, + struct pb_buffer **base_buf, + unsigned *offset) +{ + struct pb_slab_buffer *buf = pb_slab_buffer(_buf); + pb_get_base_buffer(buf->slab->bo, base_buf, offset); + *offset += buf->start; +} + + +static const struct pb_vtbl +pb_slab_buffer_vtbl = { + pb_slab_buffer_destroy, + pb_slab_buffer_map, + pb_slab_buffer_unmap, + pb_slab_buffer_get_base_buffer +}; + + +static enum pipe_error +pb_slab_create(struct pb_slab_manager *mgr) +{ + struct pb_slab *slab; + struct pb_slab_buffer *buf; + unsigned numBuffers; + unsigned i; + enum pipe_error ret; + + slab = CALLOC_STRUCT(pb_slab); + if (!slab) + return PIPE_ERROR_OUT_OF_MEMORY; + + /* + * FIXME: We should perhaps allow some variation in slabsize in order + * to efficiently reuse slabs. + */ + + slab->bo = mgr->provider->create_buffer(mgr->provider, mgr->slabSize, &mgr->desc); + if(!slab->bo) { + ret = PIPE_ERROR_OUT_OF_MEMORY; + goto out_err0; + } + + slab->virtual = pb_map(slab->bo, + PIPE_BUFFER_USAGE_CPU_READ | + PIPE_BUFFER_USAGE_CPU_WRITE); + if(!slab->virtual) { + ret = PIPE_ERROR_OUT_OF_MEMORY; + goto out_err1; + } + + pb_unmap(slab->bo); + + numBuffers = slab->bo->base.size / mgr->bufSize; + + slab->buffers = CALLOC(numBuffers, sizeof(*slab->buffers)); + if (!slab->buffers) { + ret = PIPE_ERROR_OUT_OF_MEMORY; + goto out_err1; + } + + LIST_INITHEAD(&slab->head); + LIST_INITHEAD(&slab->freeBuffers); + slab->numBuffers = numBuffers; + slab->numFree = 0; + slab->mgr = mgr; + + buf = slab->buffers; + for (i=0; i < numBuffers; ++i) { + buf->base.base.refcount = 0; + buf->base.base.size = mgr->bufSize; + buf->base.base.alignment = 0; + buf->base.base.usage = 0; + buf->base.vtbl = &pb_slab_buffer_vtbl; + buf->slab = slab; + buf->start = i* mgr->bufSize; + buf->mapCount = 0; + _glthread_INIT_COND(buf->event); + LIST_ADDTAIL(&buf->head, &slab->freeBuffers); + slab->numFree++; + buf++; + } + + LIST_ADDTAIL(&slab->head, &mgr->slabs); + + return PIPE_OK; + +out_err1: + pb_reference(&slab->bo, NULL); +out_err0: + FREE(slab); + return ret; +} + + +static struct pb_buffer * +pb_slab_manager_create_buffer(struct pb_manager *_mgr, + size_t size, + const struct pb_desc *desc) +{ + struct pb_slab_manager *mgr = pb_slab_manager(_mgr); + static struct pb_slab_buffer *buf; + struct pb_slab *slab; + struct list_head *list; + int count = DRI_SLABPOOL_ALLOC_RETRIES; + + /* check size */ + assert(size == mgr->bufSize); + if(size != mgr->bufSize) + return NULL; + + /* check if we can provide the requested alignment */ + assert(pb_check_alignment(desc->alignment, mgr->desc.alignment)); + if(!pb_check_alignment(desc->alignment, mgr->desc.alignment)) + return NULL; + assert(pb_check_alignment(desc->alignment, mgr->bufSize)); + if(!pb_check_alignment(desc->alignment, mgr->bufSize)) + return NULL; + + /* XXX: check for compatible buffer usage too? */ + + _glthread_LOCK_MUTEX(mgr->mutex); + while (mgr->slabs.next == &mgr->slabs && count > 0) { + if (mgr->slabs.next != &mgr->slabs) + break; + + _glthread_UNLOCK_MUTEX(mgr->mutex); + if (count != DRI_SLABPOOL_ALLOC_RETRIES) + util_time_sleep(1); + _glthread_LOCK_MUTEX(mgr->mutex); + (void) pb_slab_create(mgr); + count--; + } + + list = mgr->slabs.next; + if (list == &mgr->slabs) { + _glthread_UNLOCK_MUTEX(mgr->mutex); + return NULL; + } + slab = LIST_ENTRY(struct pb_slab, list, head); + if (--slab->numFree == 0) + LIST_DELINIT(list); + + list = slab->freeBuffers.next; + LIST_DELINIT(list); + + _glthread_UNLOCK_MUTEX(mgr->mutex); + buf = LIST_ENTRY(struct pb_slab_buffer, list, head); + + ++buf->base.base.refcount; + buf->base.base.alignment = desc->alignment; + buf->base.base.usage = desc->usage; + + return &buf->base; +} + + +static void +pb_slab_manager_destroy(struct pb_manager *_mgr) +{ + struct pb_slab_manager *mgr = pb_slab_manager(_mgr); + + /* TODO: cleanup all allocated buffers */ + FREE(mgr); +} + + +struct pb_manager * +pb_slab_manager_create(struct pb_manager *provider, + size_t bufSize, + size_t slabSize, + const struct pb_desc *desc) +{ + struct pb_slab_manager *mgr; + + mgr = CALLOC_STRUCT(pb_slab_manager); + if (!mgr) + return NULL; + + mgr->base.destroy = pb_slab_manager_destroy; + mgr->base.create_buffer = pb_slab_manager_create_buffer; + + mgr->provider = provider; + mgr->bufSize = bufSize; + mgr->slabSize = slabSize; + mgr->desc = *desc; + + LIST_INITHEAD(&mgr->slabs); + LIST_INITHEAD(&mgr->freeSlabs); + + _glthread_INIT_MUTEX(mgr->mutex); + + return &mgr->base; +} + + +static struct pb_buffer * +pb_slab_range_manager_create_buffer(struct pb_manager *_mgr, + size_t size, + const struct pb_desc *desc) +{ + struct pb_slab_range_manager *mgr = pb_slab_range_manager(_mgr); + size_t bufSize; + unsigned i; + + bufSize = mgr->minBufSize; + for (i = 0; i < mgr->numBuckets; ++i) { + if(bufSize >= size) + return mgr->buckets[i]->create_buffer(mgr->buckets[i], size, desc); + bufSize *= 2; + } + + /* Fall back to allocate a buffer object directly from the provider. */ + return mgr->provider->create_buffer(mgr->provider, size, desc); +} + + +static void +pb_slab_range_manager_destroy(struct pb_manager *_mgr) +{ + struct pb_slab_range_manager *mgr = pb_slab_range_manager(_mgr); + unsigned i; + + for (i = 0; i < mgr->numBuckets; ++i) + mgr->buckets[i]->destroy(mgr->buckets[i]); + FREE(mgr->buckets); + FREE(mgr->bucketSizes); + FREE(mgr); +} + + +struct pb_manager * +pb_slab_range_manager_create(struct pb_manager *provider, + size_t minBufSize, + size_t maxBufSize, + size_t slabSize, + const struct pb_desc *desc) +{ + struct pb_slab_range_manager *mgr; + size_t bufSize; + unsigned i; + + mgr = CALLOC_STRUCT(pb_slab_range_manager); + if (!mgr) + goto out_err0; + + mgr->base.destroy = pb_slab_range_manager_destroy; + mgr->base.create_buffer = pb_slab_range_manager_create_buffer; + + mgr->provider = provider; + mgr->minBufSize = minBufSize; + mgr->maxBufSize = maxBufSize; + + mgr->numBuckets = 1; + bufSize = minBufSize; + while(bufSize < maxBufSize) { + bufSize *= 2; + ++mgr->numBuckets; + } + + mgr->buckets = CALLOC(mgr->numBuckets, sizeof(*mgr->buckets)); + if (!mgr->buckets) + goto out_err1; + + bufSize = minBufSize; + for (i = 0; i < mgr->numBuckets; ++i) { + mgr->buckets[i] = pb_slab_manager_create(provider, bufSize, slabSize, desc); + if(!mgr->buckets[i]) + goto out_err2; + bufSize *= 2; + } + + return &mgr->base; + +out_err2: + for (i = 0; i < mgr->numBuckets; ++i) + if(mgr->buckets[i]) + mgr->buckets[i]->destroy(mgr->buckets[i]); + FREE(mgr->buckets); +out_err1: + FREE(mgr); +out_err0: + return NULL; +} diff --git a/src/gallium/auxiliary/rtasm/rtasm_cpu.c b/src/gallium/auxiliary/rtasm/rtasm_cpu.c index eb3359750b..f01e12faa0 100644 --- a/src/gallium/auxiliary/rtasm/rtasm_cpu.c +++ b/src/gallium/auxiliary/rtasm/rtasm_cpu.c @@ -26,14 +26,29 @@ **************************************************************************/ +#include "pipe/p_debug.h" #include "rtasm_cpu.h" +static boolean rtasm_sse_enabled(void) +{ + static boolean firsttime = 1; + static boolean enabled; + + /* This gets called quite often at the moment: + */ + if (firsttime) { + enabled = !debug_get_bool_option("GALLIUM_NOSSE", FALSE); + firsttime = FALSE; + } + return enabled; +} + int rtasm_cpu_has_sse(void) { /* FIXME: actually detect this at run-time */ -#if defined(__i386__) || defined(__386__) - return 1; +#if defined(__i386__) || defined(__386__) || defined(i386) + return rtasm_sse_enabled(); #else return 0; #endif @@ -42,8 +57,8 @@ int rtasm_cpu_has_sse(void) int rtasm_cpu_has_sse2(void) { /* FIXME: actually detect this at run-time */ -#if defined(__i386__) || defined(__386__) - return 1; +#if defined(__i386__) || defined(__386__) || defined(i386) + return rtasm_sse_enabled(); #else return 0; #endif diff --git a/src/gallium/auxiliary/rtasm/rtasm_x86sse.c b/src/gallium/auxiliary/rtasm/rtasm_x86sse.c index 4d33950e99..4e036d9032 100644 --- a/src/gallium/auxiliary/rtasm/rtasm_x86sse.c +++ b/src/gallium/auxiliary/rtasm/rtasm_x86sse.c @@ -33,15 +33,114 @@ #define DISASSEM 0 #define X86_TWOB 0x0f -static unsigned char *cptr( void (*label)() ) -{ - return (unsigned char *) label; -} + +#define DUMP_SSE 0 + +#if DUMP_SSE + +static void +_print_reg( + struct x86_reg reg ) +{ + if (reg.mod != mod_REG) + debug_printf( "[" ); + + switch( reg.file ) { + case file_REG32: + switch( reg.idx ) { + case reg_AX: debug_printf( "EAX" ); break; + case reg_CX: debug_printf( "ECX" ); break; + case reg_DX: debug_printf( "EDX" ); break; + case reg_BX: debug_printf( "EBX" ); break; + case reg_SP: debug_printf( "ESP" ); break; + case reg_BP: debug_printf( "EBP" ); break; + case reg_SI: debug_printf( "ESI" ); break; + case reg_DI: debug_printf( "EDI" ); break; + } + break; + case file_MMX: + debug_printf( "MMX%u", reg.idx ); + break; + case file_XMM: + debug_printf( "XMM%u", reg.idx ); + break; + case file_x87: + debug_printf( "fp%u", reg.idx ); + break; + } + + if (reg.mod == mod_DISP8 || + reg.mod == mod_DISP32) + debug_printf("+%d", reg.disp); + + if (reg.mod != mod_REG) + debug_printf( "]" ); +} + + +#define DUMP_START() debug_printf( "\n" ) +#define DUMP_END() debug_printf( "\n" ) + +#define DUMP() do { \ + const char *foo = __FUNCTION__; \ + while (*foo && *foo != '_') \ + foo++; \ + if (*foo) \ + foo++; \ + debug_printf( "\n% 15s ", foo ); \ +} while (0) + +#define DUMP_I( I ) do { \ + DUMP(); \ + debug_printf( "%u", I ); \ +} while( 0 ) + +#define DUMP_R( R0 ) do { \ + DUMP(); \ + _print_reg( R0 ); \ +} while( 0 ) + +#define DUMP_RR( R0, R1 ) do { \ + DUMP(); \ + _print_reg( R0 ); \ + debug_printf( ", " ); \ + _print_reg( R1 ); \ +} while( 0 ) + +#define DUMP_RI( R0, I ) do { \ + DUMP(); \ + _print_reg( R0 ); \ + debug_printf( ", %u", I ); \ +} while( 0 ) + +#define DUMP_RRI( R0, R1, I ) do { \ + DUMP(); \ + _print_reg( R0 ); \ + debug_printf( ", " ); \ + _print_reg( R1 ); \ + debug_printf( ", %u", I ); \ +} while( 0 ) + +#else + +#define DUMP_START() +#define DUMP_END() +#define DUMP( ) +#define DUMP_I( I ) +#define DUMP_R( R0 ) +#define DUMP_RR( R0, R1 ) +#define DUMP_RI( R0, I ) +#define DUMP_RRI( R0, R1, I ) + +#endif static void do_realloc( struct x86_function *p ) { - if (p->size == 0) { + if (p->store == p->error_overflow) { + p->csr = p->store; + } + else if (p->size == 0) { p->size = 1024; p->store = rtasm_exec_malloc(p->size); p->csr = p->store; @@ -51,10 +150,22 @@ static void do_realloc( struct x86_function *p ) unsigned char *tmp = p->store; p->size *= 2; p->store = rtasm_exec_malloc(p->size); - memcpy(p->store, tmp, used); - p->csr = p->store + used; + + if (p->store) { + memcpy(p->store, tmp, used); + p->csr = p->store + used; + } + else { + p->csr = p->store; + } + rtasm_exec_free(tmp); } + + if (p->store == NULL) { + p->store = p->csr = p->error_overflow; + p->size = sizeof(p->error_overflow); + } } /* Emit bytes to the instruction stream: @@ -236,9 +347,9 @@ struct x86_reg x86_get_base_reg( struct x86_reg reg ) return x86_make_reg( reg.file, reg.idx ); } -unsigned char *x86_get_label( struct x86_function *p ) +int x86_get_label( struct x86_function *p ) { - return p->csr; + return p->csr - p->store; } @@ -250,16 +361,22 @@ unsigned char *x86_get_label( struct x86_function *p ) void x86_jcc( struct x86_function *p, enum x86_cc cc, - unsigned char *label ) + int label ) { - intptr_t offset = pointer_to_intptr( label ) - (pointer_to_intptr( x86_get_label(p) ) + 2); + int offset = label - (x86_get_label(p) + 2); + DUMP_I(cc); + if (offset < 0) { + int amt = p->csr - p->store; + assert(amt > -offset); + } + if (offset <= 127 && offset >= -128) { emit_1ub(p, 0x70 + cc); emit_1b(p, (char) offset); } else { - offset = pointer_to_intptr( label ) - (pointer_to_intptr( x86_get_label(p) ) + 6); + offset = label - (x86_get_label(p) + 6); emit_2ub(p, 0x0f, 0x80 + cc); emit_1i(p, offset); } @@ -267,23 +384,27 @@ void x86_jcc( struct x86_function *p, /* Always use a 32bit offset for forward jumps: */ -unsigned char *x86_jcc_forward( struct x86_function *p, - enum x86_cc cc ) +int x86_jcc_forward( struct x86_function *p, + enum x86_cc cc ) { + DUMP_I(cc); emit_2ub(p, 0x0f, 0x80 + cc); emit_1i(p, 0); return x86_get_label(p); } -unsigned char *x86_jmp_forward( struct x86_function *p) +int x86_jmp_forward( struct x86_function *p) { + DUMP(); emit_1ub(p, 0xe9); emit_1i(p, 0); return x86_get_label(p); } -unsigned char *x86_call_forward( struct x86_function *p) +int x86_call_forward( struct x86_function *p) { + DUMP(); + emit_1ub(p, 0xe8); emit_1i(p, 0); return x86_get_label(p); @@ -292,34 +413,24 @@ unsigned char *x86_call_forward( struct x86_function *p) /* Fixup offset from forward jump: */ void x86_fixup_fwd_jump( struct x86_function *p, - unsigned char *fixup ) + int fixup ) { - *(int *)(fixup - 4) = pointer_to_intptr( x86_get_label(p) ) - pointer_to_intptr( fixup ); + *(int *)(p->store + fixup - 4) = x86_get_label(p) - fixup; } -void x86_jmp( struct x86_function *p, unsigned char *label) +void x86_jmp( struct x86_function *p, int label) { + DUMP_I( label ); emit_1ub(p, 0xe9); - emit_1i(p, pointer_to_intptr( label ) - pointer_to_intptr( x86_get_label(p) ) - 4); + emit_1i(p, label - x86_get_label(p) - 4); } -#if 0 -/* This doesn't work once we start reallocating & copying the - * generated code on buffer fills, because the call is relative to the - * current pc. - */ -void x86_call( struct x86_function *p, void (*label)()) -{ - emit_1ub(p, 0xe8); - emit_1i(p, cptr(label) - x86_get_label(p) - 4); -} -#else void x86_call( struct x86_function *p, struct x86_reg reg) { + DUMP_R( reg ); emit_1ub(p, 0xff); - emit_modrm(p, reg, reg); + emit_modrm_noreg(p, 2, reg); } -#endif /* michal: @@ -328,6 +439,7 @@ void x86_call( struct x86_function *p, struct x86_reg reg) */ void x86_mov_reg_imm( struct x86_function *p, struct x86_reg dst, int imm ) { + DUMP_RI( dst, imm ); assert(dst.mod == mod_REG); emit_1ub(p, 0xb8 + dst.idx); emit_1i(p, imm); @@ -336,14 +448,23 @@ void x86_mov_reg_imm( struct x86_function *p, struct x86_reg dst, int imm ) void x86_push( struct x86_function *p, struct x86_reg reg ) { - assert(reg.mod == mod_REG); - emit_1ub(p, 0x50 + reg.idx); + DUMP_R( reg ); + if (reg.mod == mod_REG) + emit_1ub(p, 0x50 + reg.idx); + else + { + emit_1ub(p, 0xff); + emit_modrm_noreg(p, 6, reg); + } + + p->stack_offset += 4; } void x86_pop( struct x86_function *p, struct x86_reg reg ) { + DUMP_R( reg ); assert(reg.mod == mod_REG); emit_1ub(p, 0x58 + reg.idx); p->stack_offset -= 4; @@ -352,6 +473,7 @@ void x86_pop( struct x86_function *p, void x86_inc( struct x86_function *p, struct x86_reg reg ) { + DUMP_R( reg ); assert(reg.mod == mod_REG); emit_1ub(p, 0x40 + reg.idx); } @@ -359,17 +481,27 @@ void x86_inc( struct x86_function *p, void x86_dec( struct x86_function *p, struct x86_reg reg ) { + DUMP_R( reg ); assert(reg.mod == mod_REG); emit_1ub(p, 0x48 + reg.idx); } void x86_ret( struct x86_function *p ) { + DUMP(); + assert(p->stack_offset == 0); emit_1ub(p, 0xc3); } +void x86_retw( struct x86_function *p, unsigned short imm ) +{ + DUMP(); + emit_3ub(p, 0xc2, imm & 0xff, (imm >> 8) & 0xff); +} + void x86_sahf( struct x86_function *p ) { + DUMP(); emit_1ub(p, 0x9e); } @@ -377,6 +509,7 @@ void x86_mov( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_op_modrm( p, 0x8b, 0x89, dst, src ); } @@ -384,6 +517,7 @@ void x86_xor( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_op_modrm( p, 0x33, 0x31, dst, src ); } @@ -391,6 +525,7 @@ void x86_cmp( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_op_modrm( p, 0x3b, 0x39, dst, src ); } @@ -398,6 +533,7 @@ void x86_lea( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_1ub(p, 0x8d); emit_modrm( p, dst, src ); } @@ -406,6 +542,7 @@ void x86_test( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_1ub(p, 0x85); emit_modrm( p, dst, src ); } @@ -414,20 +551,36 @@ void x86_add( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_op_modrm(p, 0x03, 0x01, dst, src ); } +/* Calculate EAX * src, results in EDX:EAX. + */ void x86_mul( struct x86_function *p, struct x86_reg src ) { - assert (src.file == file_REG32 && src.mod == mod_REG); - emit_op_modrm(p, 0xf7, 0, x86_make_reg (file_REG32, reg_SP), src ); + DUMP_R( src ); + emit_1ub(p, 0xf7); + emit_modrm_noreg(p, 4, src ); +} + + +void x86_imul( struct x86_function *p, + struct x86_reg dst, + struct x86_reg src ) +{ + DUMP_RR( dst, src ); + emit_2ub(p, X86_TWOB, 0xAF); + emit_modrm(p, dst, src); } + void x86_sub( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_op_modrm(p, 0x2b, 0x29, dst, src ); } @@ -435,6 +588,7 @@ void x86_or( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_op_modrm( p, 0x0b, 0x09, dst, src ); } @@ -442,6 +596,7 @@ void x86_and( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_op_modrm( p, 0x23, 0x21, dst, src ); } @@ -456,6 +611,7 @@ void sse_movss( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_2ub(p, 0xF3, X86_TWOB); emit_op_modrm( p, 0x10, 0x11, dst, src ); } @@ -464,6 +620,7 @@ void sse_movaps( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_1ub(p, X86_TWOB); emit_op_modrm( p, 0x28, 0x29, dst, src ); } @@ -472,6 +629,7 @@ void sse_movups( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_1ub(p, X86_TWOB); emit_op_modrm( p, 0x10, 0x11, dst, src ); } @@ -480,6 +638,7 @@ void sse_movhps( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); assert(dst.mod != mod_REG || src.mod != mod_REG); emit_1ub(p, X86_TWOB); emit_op_modrm( p, 0x16, 0x17, dst, src ); /* cf movlhps */ @@ -489,6 +648,7 @@ void sse_movlps( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); assert(dst.mod != mod_REG || src.mod != mod_REG); emit_1ub(p, X86_TWOB); emit_op_modrm( p, 0x12, 0x13, dst, src ); /* cf movhlps */ @@ -498,6 +658,7 @@ void sse_maxps( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_2ub(p, X86_TWOB, 0x5F); emit_modrm( p, dst, src ); } @@ -506,6 +667,7 @@ void sse_maxss( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_3ub(p, 0xF3, X86_TWOB, 0x5F); emit_modrm( p, dst, src ); } @@ -514,6 +676,7 @@ void sse_divss( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_3ub(p, 0xF3, X86_TWOB, 0x5E); emit_modrm( p, dst, src ); } @@ -522,6 +685,7 @@ void sse_minps( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_2ub(p, X86_TWOB, 0x5D); emit_modrm( p, dst, src ); } @@ -530,6 +694,7 @@ void sse_subps( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_2ub(p, X86_TWOB, 0x5C); emit_modrm( p, dst, src ); } @@ -538,6 +703,7 @@ void sse_mulps( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_2ub(p, X86_TWOB, 0x59); emit_modrm( p, dst, src ); } @@ -546,6 +712,7 @@ void sse_mulss( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_3ub(p, 0xF3, X86_TWOB, 0x59); emit_modrm( p, dst, src ); } @@ -554,6 +721,7 @@ void sse_addps( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_2ub(p, X86_TWOB, 0x58); emit_modrm( p, dst, src ); } @@ -562,6 +730,7 @@ void sse_addss( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_3ub(p, 0xF3, X86_TWOB, 0x58); emit_modrm( p, dst, src ); } @@ -570,6 +739,7 @@ void sse_andnps( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_2ub(p, X86_TWOB, 0x55); emit_modrm( p, dst, src ); } @@ -578,6 +748,7 @@ void sse_andps( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_2ub(p, X86_TWOB, 0x54); emit_modrm( p, dst, src ); } @@ -586,6 +757,7 @@ void sse_rsqrtps( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_2ub(p, X86_TWOB, 0x52); emit_modrm( p, dst, src ); } @@ -594,6 +766,7 @@ void sse_rsqrtss( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_3ub(p, 0xF3, X86_TWOB, 0x52); emit_modrm( p, dst, src ); @@ -603,6 +776,7 @@ void sse_movhlps( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); assert(dst.mod == mod_REG && src.mod == mod_REG); emit_2ub(p, X86_TWOB, 0x12); emit_modrm( p, dst, src ); @@ -612,6 +786,7 @@ void sse_movlhps( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); assert(dst.mod == mod_REG && src.mod == mod_REG); emit_2ub(p, X86_TWOB, 0x16); emit_modrm( p, dst, src ); @@ -621,6 +796,7 @@ void sse_orps( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_2ub(p, X86_TWOB, 0x56); emit_modrm( p, dst, src ); } @@ -629,6 +805,7 @@ void sse_xorps( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_2ub(p, X86_TWOB, 0x57); emit_modrm( p, dst, src ); } @@ -637,6 +814,7 @@ void sse_cvtps2pi( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); assert(dst.file == file_MMX && (src.file == file_XMM || src.mod != mod_REG)); @@ -646,36 +824,62 @@ void sse_cvtps2pi( struct x86_function *p, emit_modrm( p, dst, src ); } +void sse2_cvtdq2ps( struct x86_function *p, + struct x86_reg dst, + struct x86_reg src ) +{ + DUMP_RR( dst, src ); + emit_2ub(p, X86_TWOB, 0x5b); + emit_modrm( p, dst, src ); +} + /* Shufps can also be used to implement a reduced swizzle when dest == * arg0. */ void sse_shufps( struct x86_function *p, - struct x86_reg dest, - struct x86_reg arg0, + struct x86_reg dst, + struct x86_reg src, unsigned char shuf) { + DUMP_RRI( dst, src, shuf ); emit_2ub(p, X86_TWOB, 0xC6); - emit_modrm(p, dest, arg0); + emit_modrm(p, dst, src); emit_1ub(p, shuf); } +void sse_unpckhps( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) +{ + DUMP_RR( dst, src ); + emit_2ub( p, X86_TWOB, 0x15 ); + emit_modrm( p, dst, src ); +} + +void sse_unpcklps( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) +{ + DUMP_RR( dst, src ); + emit_2ub( p, X86_TWOB, 0x14 ); + emit_modrm( p, dst, src ); +} + void sse_cmpps( struct x86_function *p, - struct x86_reg dest, - struct x86_reg arg0, + struct x86_reg dst, + struct x86_reg src, unsigned char cc) { + DUMP_RRI( dst, src, cc ); emit_2ub(p, X86_TWOB, 0xC2); - emit_modrm(p, dest, arg0); + emit_modrm(p, dst, src); emit_1ub(p, cc); } void sse_pmovmskb( struct x86_function *p, - struct x86_reg dest, + struct x86_reg dst, struct x86_reg src) { - emit_3ub(p, 0x66, X86_TWOB, 0xD7); - emit_modrm(p, dest, src); + DUMP_RR( dst, src ); + emit_3ub(p, 0x66, X86_TWOB, 0xD7); + emit_modrm(p, dst, src); } /*********************************************************************** @@ -686,12 +890,13 @@ void sse_pmovmskb( struct x86_function *p, * Perform a reduced swizzle: */ void sse2_pshufd( struct x86_function *p, - struct x86_reg dest, - struct x86_reg arg0, + struct x86_reg dst, + struct x86_reg src, unsigned char shuf) { + DUMP_RRI( dst, src, shuf ); emit_3ub(p, 0x66, X86_TWOB, 0x70); - emit_modrm(p, dest, arg0); + emit_modrm(p, dst, src); emit_1ub(p, shuf); } @@ -699,6 +904,7 @@ void sse2_cvttps2dq( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_3ub( p, 0xF3, X86_TWOB, 0x5B ); emit_modrm( p, dst, src ); } @@ -707,6 +913,7 @@ void sse2_cvtps2dq( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_3ub(p, 0x66, X86_TWOB, 0x5B); emit_modrm( p, dst, src ); } @@ -715,6 +922,7 @@ void sse2_packssdw( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_3ub(p, 0x66, X86_TWOB, 0x6B); emit_modrm( p, dst, src ); } @@ -723,6 +931,7 @@ void sse2_packsswb( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_3ub(p, 0x66, X86_TWOB, 0x63); emit_modrm( p, dst, src ); } @@ -731,14 +940,26 @@ void sse2_packuswb( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_3ub(p, 0x66, X86_TWOB, 0x67); emit_modrm( p, dst, src ); } +void sse2_punpcklbw( struct x86_function *p, + struct x86_reg dst, + struct x86_reg src ) +{ + DUMP_RR( dst, src ); + emit_3ub(p, 0x66, X86_TWOB, 0x60); + emit_modrm( p, dst, src ); +} + + void sse2_rcpps( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_2ub(p, X86_TWOB, 0x53); emit_modrm( p, dst, src ); } @@ -747,6 +968,7 @@ void sse2_rcpss( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_3ub(p, 0xF3, X86_TWOB, 0x53); emit_modrm( p, dst, src ); } @@ -755,6 +977,7 @@ void sse2_movd( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); emit_2ub(p, 0x66, X86_TWOB); emit_op_modrm( p, 0x6e, 0x7e, dst, src ); } @@ -767,30 +990,35 @@ void sse2_movd( struct x86_function *p, */ void x87_fist( struct x86_function *p, struct x86_reg dst ) { + DUMP_R( dst ); emit_1ub(p, 0xdb); emit_modrm_noreg(p, 2, dst); } void x87_fistp( struct x86_function *p, struct x86_reg dst ) { + DUMP_R( dst ); emit_1ub(p, 0xdb); emit_modrm_noreg(p, 3, dst); } void x87_fild( struct x86_function *p, struct x86_reg arg ) { + DUMP_R( arg ); emit_1ub(p, 0xdf); emit_modrm_noreg(p, 0, arg); } void x87_fldz( struct x86_function *p ) { + DUMP(); emit_2ub(p, 0xd9, 0xee); } void x87_fldcw( struct x86_function *p, struct x86_reg arg ) { + DUMP_R( arg ); assert(arg.file == file_REG32); assert(arg.mod != mod_REG); emit_1ub(p, 0xd9); @@ -799,26 +1027,31 @@ void x87_fldcw( struct x86_function *p, struct x86_reg arg ) void x87_fld1( struct x86_function *p ) { + DUMP(); emit_2ub(p, 0xd9, 0xe8); } void x87_fldl2e( struct x86_function *p ) { + DUMP(); emit_2ub(p, 0xd9, 0xea); } void x87_fldln2( struct x86_function *p ) { + DUMP(); emit_2ub(p, 0xd9, 0xed); } void x87_fwait( struct x86_function *p ) { + DUMP(); emit_1ub(p, 0x9b); } void x87_fnclex( struct x86_function *p ) { + DUMP(); emit_2ub(p, 0xdb, 0xe2); } @@ -855,49 +1088,55 @@ static void x87_arith_op( struct x86_function *p, struct x86_reg dst, struct x86 assert(0); } -void x87_fmul( struct x86_function *p, struct x86_reg dst, struct x86_reg arg ) +void x87_fmul( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { - x87_arith_op(p, dst, arg, + DUMP_RR( dst, src ); + x87_arith_op(p, dst, src, 0xd8, 0xc8, 0xdc, 0xc8, 4); } -void x87_fsub( struct x86_function *p, struct x86_reg dst, struct x86_reg arg ) +void x87_fsub( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { - x87_arith_op(p, dst, arg, + DUMP_RR( dst, src ); + x87_arith_op(p, dst, src, 0xd8, 0xe0, 0xdc, 0xe8, 4); } -void x87_fsubr( struct x86_function *p, struct x86_reg dst, struct x86_reg arg ) +void x87_fsubr( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { - x87_arith_op(p, dst, arg, + DUMP_RR( dst, src ); + x87_arith_op(p, dst, src, 0xd8, 0xe8, 0xdc, 0xe0, 5); } -void x87_fadd( struct x86_function *p, struct x86_reg dst, struct x86_reg arg ) +void x87_fadd( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { - x87_arith_op(p, dst, arg, + DUMP_RR( dst, src ); + x87_arith_op(p, dst, src, 0xd8, 0xc0, 0xdc, 0xc0, 0); } -void x87_fdiv( struct x86_function *p, struct x86_reg dst, struct x86_reg arg ) +void x87_fdiv( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { - x87_arith_op(p, dst, arg, + DUMP_RR( dst, src ); + x87_arith_op(p, dst, src, 0xd8, 0xf0, 0xdc, 0xf8, 6); } -void x87_fdivr( struct x86_function *p, struct x86_reg dst, struct x86_reg arg ) +void x87_fdivr( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { - x87_arith_op(p, dst, arg, + DUMP_RR( dst, src ); + x87_arith_op(p, dst, src, 0xd8, 0xf8, 0xdc, 0xf0, 7); @@ -905,6 +1144,7 @@ void x87_fdivr( struct x86_function *p, struct x86_reg dst, struct x86_reg arg ) void x87_fmulp( struct x86_function *p, struct x86_reg dst ) { + DUMP_R( dst ); assert(dst.file == file_x87); assert(dst.idx >= 1); emit_2ub(p, 0xde, 0xc8+dst.idx); @@ -912,6 +1152,7 @@ void x87_fmulp( struct x86_function *p, struct x86_reg dst ) void x87_fsubp( struct x86_function *p, struct x86_reg dst ) { + DUMP_R( dst ); assert(dst.file == file_x87); assert(dst.idx >= 1); emit_2ub(p, 0xde, 0xe8+dst.idx); @@ -919,6 +1160,7 @@ void x87_fsubp( struct x86_function *p, struct x86_reg dst ) void x87_fsubrp( struct x86_function *p, struct x86_reg dst ) { + DUMP_R( dst ); assert(dst.file == file_x87); assert(dst.idx >= 1); emit_2ub(p, 0xde, 0xe0+dst.idx); @@ -926,6 +1168,7 @@ void x87_fsubrp( struct x86_function *p, struct x86_reg dst ) void x87_faddp( struct x86_function *p, struct x86_reg dst ) { + DUMP_R( dst ); assert(dst.file == file_x87); assert(dst.idx >= 1); emit_2ub(p, 0xde, 0xc0+dst.idx); @@ -933,6 +1176,7 @@ void x87_faddp( struct x86_function *p, struct x86_reg dst ) void x87_fdivp( struct x86_function *p, struct x86_reg dst ) { + DUMP_R( dst ); assert(dst.file == file_x87); assert(dst.idx >= 1); emit_2ub(p, 0xde, 0xf8+dst.idx); @@ -940,6 +1184,7 @@ void x87_fdivp( struct x86_function *p, struct x86_reg dst ) void x87_fdivrp( struct x86_function *p, struct x86_reg dst ) { + DUMP_R( dst ); assert(dst.file == file_x87); assert(dst.idx >= 1); emit_2ub(p, 0xde, 0xf0+dst.idx); @@ -947,70 +1192,83 @@ void x87_fdivrp( struct x86_function *p, struct x86_reg dst ) void x87_fucom( struct x86_function *p, struct x86_reg arg ) { + DUMP_R( arg ); assert(arg.file == file_x87); emit_2ub(p, 0xdd, 0xe0+arg.idx); } void x87_fucomp( struct x86_function *p, struct x86_reg arg ) { + DUMP_R( arg ); assert(arg.file == file_x87); emit_2ub(p, 0xdd, 0xe8+arg.idx); } void x87_fucompp( struct x86_function *p ) { + DUMP(); emit_2ub(p, 0xda, 0xe9); } void x87_fxch( struct x86_function *p, struct x86_reg arg ) { + DUMP_R( arg ); assert(arg.file == file_x87); emit_2ub(p, 0xd9, 0xc8+arg.idx); } void x87_fabs( struct x86_function *p ) { + DUMP(); emit_2ub(p, 0xd9, 0xe1); } void x87_fchs( struct x86_function *p ) { + DUMP(); emit_2ub(p, 0xd9, 0xe0); } void x87_fcos( struct x86_function *p ) { + DUMP(); emit_2ub(p, 0xd9, 0xff); } void x87_fprndint( struct x86_function *p ) { + DUMP(); emit_2ub(p, 0xd9, 0xfc); } void x87_fscale( struct x86_function *p ) { + DUMP(); emit_2ub(p, 0xd9, 0xfd); } void x87_fsin( struct x86_function *p ) { + DUMP(); emit_2ub(p, 0xd9, 0xfe); } void x87_fsincos( struct x86_function *p ) { + DUMP(); emit_2ub(p, 0xd9, 0xfb); } void x87_fsqrt( struct x86_function *p ) { + DUMP(); emit_2ub(p, 0xd9, 0xfa); } void x87_fxtract( struct x86_function *p ) { + DUMP(); emit_2ub(p, 0xd9, 0xf4); } @@ -1020,6 +1278,7 @@ void x87_fxtract( struct x86_function *p ) */ void x87_f2xm1( struct x86_function *p ) { + DUMP(); emit_2ub(p, 0xd9, 0xf0); } @@ -1028,6 +1287,7 @@ void x87_f2xm1( struct x86_function *p ) */ void x87_fyl2x( struct x86_function *p ) { + DUMP(); emit_2ub(p, 0xd9, 0xf1); } @@ -1038,12 +1298,14 @@ void x87_fyl2x( struct x86_function *p ) */ void x87_fyl2xp1( struct x86_function *p ) { + DUMP(); emit_2ub(p, 0xd9, 0xf9); } void x87_fld( struct x86_function *p, struct x86_reg arg ) { + DUMP_R( arg ); if (arg.file == file_x87) emit_2ub(p, 0xd9, 0xc0 + arg.idx); else { @@ -1054,6 +1316,7 @@ void x87_fld( struct x86_function *p, struct x86_reg arg ) void x87_fst( struct x86_function *p, struct x86_reg dst ) { + DUMP_R( dst ); if (dst.file == file_x87) emit_2ub(p, 0xdd, 0xd0 + dst.idx); else { @@ -1064,6 +1327,7 @@ void x87_fst( struct x86_function *p, struct x86_reg dst ) void x87_fstp( struct x86_function *p, struct x86_reg dst ) { + DUMP_R( dst ); if (dst.file == file_x87) emit_2ub(p, 0xdd, 0xd8 + dst.idx); else { @@ -1074,6 +1338,7 @@ void x87_fstp( struct x86_function *p, struct x86_reg dst ) void x87_fcom( struct x86_function *p, struct x86_reg dst ) { + DUMP_R( dst ); if (dst.file == file_x87) emit_2ub(p, 0xd8, 0xd0 + dst.idx); else { @@ -1084,6 +1349,7 @@ void x87_fcom( struct x86_function *p, struct x86_reg dst ) void x87_fcomp( struct x86_function *p, struct x86_reg dst ) { + DUMP_R( dst ); if (dst.file == file_x87) emit_2ub(p, 0xd8, 0xd8 + dst.idx); else { @@ -1095,6 +1361,7 @@ void x87_fcomp( struct x86_function *p, struct x86_reg dst ) void x87_fnstsw( struct x86_function *p, struct x86_reg dst ) { + DUMP_R( dst ); assert(dst.file == file_REG32); if (dst.idx == reg_AX && @@ -1115,6 +1382,7 @@ void x87_fnstsw( struct x86_function *p, struct x86_reg dst ) void mmx_emms( struct x86_function *p ) { + DUMP(); assert(p->need_emms); emit_2ub(p, 0x0f, 0x77); p->need_emms = 0; @@ -1124,6 +1392,7 @@ void mmx_packssdw( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); assert(dst.file == file_MMX && (src.file == file_MMX || src.mod != mod_REG)); @@ -1137,6 +1406,7 @@ void mmx_packuswb( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); assert(dst.file == file_MMX && (src.file == file_MMX || src.mod != mod_REG)); @@ -1150,6 +1420,7 @@ void mmx_movd( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); p->need_emms = 1; emit_1ub(p, X86_TWOB); emit_op_modrm( p, 0x6e, 0x7e, dst, src ); @@ -1159,6 +1430,7 @@ void mmx_movq( struct x86_function *p, struct x86_reg dst, struct x86_reg src ) { + DUMP_RR( dst, src ); p->need_emms = 1; emit_1ub(p, X86_TWOB); emit_op_modrm( p, 0x6f, 0x7f, dst, src ); @@ -1186,18 +1458,25 @@ void x86_init_func( struct x86_function *p ) p->size = 0; p->store = NULL; p->csr = p->store; + DUMP_START(); } void x86_init_func_size( struct x86_function *p, unsigned code_size ) { p->size = code_size; p->store = rtasm_exec_malloc(code_size); + if (p->store == NULL) { + p->store = p->error_overflow; + } p->csr = p->store; + DUMP_START(); } void x86_release_func( struct x86_function *p ) { - rtasm_exec_free(p->store); + if (p->store && p->store != p->error_overflow) + rtasm_exec_free(p->store); + p->store = NULL; p->csr = NULL; p->size = 0; @@ -1206,9 +1485,14 @@ void x86_release_func( struct x86_function *p ) void (*x86_get_func( struct x86_function *p ))(void) { + DUMP_END(); if (DISASSEM && p->store) debug_printf("disassemble %p %p\n", p->store, p->csr); - return (void (*)(void)) p->store; + + if (p->store == p->error_overflow) + return (void (*)(void)) NULL; + else + return (void (*)(void)) p->store; } #else diff --git a/src/gallium/auxiliary/rtasm/rtasm_x86sse.h b/src/gallium/auxiliary/rtasm/rtasm_x86sse.h index 606b41eb35..eacaeeaf6f 100644 --- a/src/gallium/auxiliary/rtasm/rtasm_x86sse.h +++ b/src/gallium/auxiliary/rtasm/rtasm_x86sse.h @@ -43,6 +43,7 @@ struct x86_function { unsigned char *csr; unsigned stack_offset; int need_emms; + unsigned char error_overflow[4]; const char *fn; }; @@ -123,23 +124,23 @@ struct x86_reg x86_get_base_reg( struct x86_reg reg ); /* Labels, jumps and fixup: */ -unsigned char *x86_get_label( struct x86_function *p ); +int x86_get_label( struct x86_function *p ); void x86_jcc( struct x86_function *p, enum x86_cc cc, - unsigned char *label ); + int label ); -unsigned char *x86_jcc_forward( struct x86_function *p, +int x86_jcc_forward( struct x86_function *p, enum x86_cc cc ); -unsigned char *x86_jmp_forward( struct x86_function *p); +int x86_jmp_forward( struct x86_function *p); -unsigned char *x86_call_forward( struct x86_function *p); +int x86_call_forward( struct x86_function *p); void x86_fixup_fwd_jump( struct x86_function *p, - unsigned char *fixup ); + int fixup ); -void x86_jmp( struct x86_function *p, unsigned char *label ); +void x86_jmp( struct x86_function *p, int label ); /* void x86_call( struct x86_function *p, void (*label)() ); */ void x86_call( struct x86_function *p, struct x86_reg reg); @@ -165,6 +166,7 @@ void mmx_packuswb( struct x86_function *p, struct x86_reg dst, struct x86_reg sr void sse2_cvtps2dq( struct x86_function *p, struct x86_reg dst, struct x86_reg src ); void sse2_cvttps2dq( struct x86_function *p, struct x86_reg dst, struct x86_reg src ); +void sse2_cvtdq2ps( struct x86_function *p, struct x86_reg dst, struct x86_reg src ); void sse2_movd( struct x86_function *p, struct x86_reg dst, struct x86_reg src ); void sse2_packssdw( struct x86_function *p, struct x86_reg dst, struct x86_reg src ); void sse2_packsswb( struct x86_function *p, struct x86_reg dst, struct x86_reg src ); @@ -201,7 +203,10 @@ void sse_rsqrtps( struct x86_function *p, struct x86_reg dst, struct x86_reg src void sse_rsqrtss( struct x86_function *p, struct x86_reg dst, struct x86_reg src ); void sse_shufps( struct x86_function *p, struct x86_reg dest, struct x86_reg arg0, unsigned char shuf ); +void sse_unpckhps( struct x86_function *p, struct x86_reg dst, struct x86_reg src ); +void sse_unpcklps( struct x86_function *p, struct x86_reg dst, struct x86_reg src ); void sse_pmovmskb( struct x86_function *p, struct x86_reg dest, struct x86_reg src ); +void sse2_punpcklbw( struct x86_function *p, struct x86_reg dst, struct x86_reg src ); void x86_add( struct x86_function *p, struct x86_reg dst, struct x86_reg src ); void x86_and( struct x86_function *p, struct x86_reg dst, struct x86_reg src ); @@ -211,10 +216,12 @@ void x86_inc( struct x86_function *p, struct x86_reg reg ); void x86_lea( struct x86_function *p, struct x86_reg dst, struct x86_reg src ); void x86_mov( struct x86_function *p, struct x86_reg dst, struct x86_reg src ); void x86_mul( struct x86_function *p, struct x86_reg src ); +void x86_imul( struct x86_function *p, struct x86_reg dst, struct x86_reg src ); void x86_or( struct x86_function *p, struct x86_reg dst, struct x86_reg src ); void x86_pop( struct x86_function *p, struct x86_reg reg ); void x86_push( struct x86_function *p, struct x86_reg reg ); void x86_ret( struct x86_function *p ); +void x86_retw( struct x86_function *p, unsigned short imm ); void x86_sub( struct x86_function *p, struct x86_reg dst, struct x86_reg src ); void x86_test( struct x86_function *p, struct x86_reg dst, struct x86_reg src ); void x86_xor( struct x86_function *p, struct x86_reg dst, struct x86_reg src ); diff --git a/src/gallium/auxiliary/sct/sct.c b/src/gallium/auxiliary/sct/sct.c index 97ee5882a1..5e4126e014 100644 --- a/src/gallium/auxiliary/sct/sct.c +++ b/src/gallium/auxiliary/sct/sct.c @@ -209,6 +209,7 @@ remove_context_from_surface(struct sct_surface *si, } else { prev = curr; + next = curr->next; } } } diff --git a/src/gallium/auxiliary/tgsi/exec/Makefile b/src/gallium/auxiliary/tgsi/exec/Makefile new file mode 100644 index 0000000000..451911a354 --- /dev/null +++ b/src/gallium/auxiliary/tgsi/exec/Makefile @@ -0,0 +1,2 @@ +default: + cd .. ; make diff --git a/src/gallium/auxiliary/tgsi/exec/tgsi_exec.c b/src/gallium/auxiliary/tgsi/exec/tgsi_exec.c index 78e7dec569..826b432f09 100644 --- a/src/gallium/auxiliary/tgsi/exec/tgsi_exec.c +++ b/src/gallium/auxiliary/tgsi/exec/tgsi_exec.c @@ -88,6 +88,10 @@ #define TEMP_OUTPUT_C TGSI_EXEC_TEMP_OUTPUT_C #define TEMP_PRIMITIVE_I TGSI_EXEC_TEMP_PRIMITIVE_I #define TEMP_PRIMITIVE_C TGSI_EXEC_TEMP_PRIMITIVE_C +#define TEMP_3_I TGSI_EXEC_TEMP_THREE_I +#define TEMP_3_C TGSI_EXEC_TEMP_THREE_C +#define TEMP_HALF_I TGSI_EXEC_TEMP_HALF_I +#define TEMP_HALF_C TGSI_EXEC_TEMP_HALF_C #define TEMP_R0 TGSI_EXEC_TEMP_R0 #define FOR_EACH_CHANNEL(CHAN)\ @@ -262,6 +266,8 @@ tgsi_exec_machine_init( mach->Temps[TEMP_2_I].xyzw[TEMP_2_C].f[i] = 2.0f; mach->Temps[TEMP_128_I].xyzw[TEMP_128_C].f[i] = 128.0f; mach->Temps[TEMP_M128_I].xyzw[TEMP_M128_C].f[i] = -128.0f; + mach->Temps[TEMP_3_I].xyzw[TEMP_3_C].f[i] = 3.0f; + mach->Temps[TEMP_HALF_I].xyzw[TEMP_HALF_C].f[i] = 0.5f; } } @@ -287,10 +293,10 @@ micro_abs( union tgsi_exec_channel *dst, const union tgsi_exec_channel *src ) { - dst->f[0] = (float) fabs( (double) src->f[0] ); - dst->f[1] = (float) fabs( (double) src->f[1] ); - dst->f[2] = (float) fabs( (double) src->f[2] ); - dst->f[3] = (float) fabs( (double) src->f[3] ); + dst->f[0] = fabsf( src->f[0] ); + dst->f[1] = fabsf( src->f[1] ); + dst->f[2] = fabsf( src->f[2] ); + dst->f[3] = fabsf( src->f[3] ); } static void @@ -334,10 +340,10 @@ micro_ceil( union tgsi_exec_channel *dst, const union tgsi_exec_channel *src ) { - dst->f[0] = (float) ceil( (double) src->f[0] ); - dst->f[1] = (float) ceil( (double) src->f[1] ); - dst->f[2] = (float) ceil( (double) src->f[2] ); - dst->f[3] = (float) ceil( (double) src->f[3] ); + dst->f[0] = ceilf( src->f[0] ); + dst->f[1] = ceilf( src->f[1] ); + dst->f[2] = ceilf( src->f[2] ); + dst->f[3] = ceilf( src->f[3] ); } static void @@ -345,10 +351,10 @@ micro_cos( union tgsi_exec_channel *dst, const union tgsi_exec_channel *src ) { - dst->f[0] = (float) cos( (double) src->f[0] ); - dst->f[1] = (float) cos( (double) src->f[1] ); - dst->f[2] = (float) cos( (double) src->f[2] ); - dst->f[3] = (float) cos( (double) src->f[3] ); + dst->f[0] = cosf( src->f[0] ); + dst->f[1] = cosf( src->f[1] ); + dst->f[2] = cosf( src->f[2] ); + dst->f[3] = cosf( src->f[3] ); } static void @@ -430,10 +436,10 @@ micro_exp2( union tgsi_exec_channel *dst, const union tgsi_exec_channel *src) { - dst->f[0] = (float) pow( 2.0, (double) src->f[0] ); - dst->f[1] = (float) pow( 2.0, (double) src->f[1] ); - dst->f[2] = (float) pow( 2.0, (double) src->f[2] ); - dst->f[3] = (float) pow( 2.0, (double) src->f[3] ); + dst->f[0] = powf( 2.0f, src->f[0] ); + dst->f[1] = powf( 2.0f, src->f[1] ); + dst->f[2] = powf( 2.0f, src->f[2] ); + dst->f[3] = powf( 2.0f, src->f[3] ); } static void @@ -463,10 +469,10 @@ micro_flr( union tgsi_exec_channel *dst, const union tgsi_exec_channel *src ) { - dst->f[0] = (float) floor( (double) src->f[0] ); - dst->f[1] = (float) floor( (double) src->f[1] ); - dst->f[2] = (float) floor( (double) src->f[2] ); - dst->f[3] = (float) floor( (double) src->f[3] ); + dst->f[0] = floorf( src->f[0] ); + dst->f[1] = floorf( src->f[1] ); + dst->f[2] = floorf( src->f[2] ); + dst->f[3] = floorf( src->f[3] ); } static void @@ -474,10 +480,10 @@ micro_frc( union tgsi_exec_channel *dst, const union tgsi_exec_channel *src ) { - dst->f[0] = src->f[0] - (float) floor( (double) src->f[0] ); - dst->f[1] = src->f[1] - (float) floor( (double) src->f[1] ); - dst->f[2] = src->f[2] - (float) floor( (double) src->f[2] ); - dst->f[3] = src->f[3] - (float) floor( (double) src->f[3] ); + dst->f[0] = src->f[0] - floorf( src->f[0] ); + dst->f[1] = src->f[1] - floorf( src->f[1] ); + dst->f[2] = src->f[2] - floorf( src->f[2] ); + dst->f[3] = src->f[3] - floorf( src->f[3] ); } static void @@ -510,10 +516,24 @@ micro_lg2( union tgsi_exec_channel *dst, const union tgsi_exec_channel *src ) { - dst->f[0] = (float) log( (double) src->f[0] ) * 1.442695f; - dst->f[1] = (float) log( (double) src->f[1] ) * 1.442695f; - dst->f[2] = (float) log( (double) src->f[2] ) * 1.442695f; - dst->f[3] = (float) log( (double) src->f[3] ) * 1.442695f; + dst->f[0] = logf( src->f[0] ) * 1.442695f; + dst->f[1] = logf( src->f[1] ) * 1.442695f; + dst->f[2] = logf( src->f[2] ) * 1.442695f; + dst->f[3] = logf( src->f[3] ) * 1.442695f; +} + +static void +micro_le( + union tgsi_exec_channel *dst, + const union tgsi_exec_channel *src0, + const union tgsi_exec_channel *src1, + const union tgsi_exec_channel *src2, + const union tgsi_exec_channel *src3 ) +{ + dst->f[0] = src0->f[0] <= src1->f[0] ? src2->f[0] : src3->f[0]; + dst->f[1] = src0->f[1] <= src1->f[1] ? src2->f[1] : src3->f[1]; + dst->f[2] = src0->f[2] <= src1->f[2] ? src2->f[2] : src3->f[2]; + dst->f[3] = src0->f[3] <= src1->f[3] ? src2->f[3] : src3->f[3]; } static void @@ -764,10 +784,10 @@ micro_pow( const union tgsi_exec_channel *src0, const union tgsi_exec_channel *src1 ) { - dst->f[0] = (float) pow( (double) src0->f[0], (double) src1->f[0] ); - dst->f[1] = (float) pow( (double) src0->f[1], (double) src1->f[1] ); - dst->f[2] = (float) pow( (double) src0->f[2], (double) src1->f[2] ); - dst->f[3] = (float) pow( (double) src0->f[3], (double) src1->f[3] ); + dst->f[0] = powf( src0->f[0], src1->f[0] ); + dst->f[1] = powf( src0->f[1], src1->f[1] ); + dst->f[2] = powf( src0->f[2], src1->f[2] ); + dst->f[3] = powf( src0->f[3], src1->f[3] ); } static void @@ -775,10 +795,10 @@ micro_rnd( union tgsi_exec_channel *dst, const union tgsi_exec_channel *src ) { - dst->f[0] = (float) floor( (double) (src->f[0] + 0.5f) ); - dst->f[1] = (float) floor( (double) (src->f[1] + 0.5f) ); - dst->f[2] = (float) floor( (double) (src->f[2] + 0.5f) ); - dst->f[3] = (float) floor( (double) (src->f[3] + 0.5f) ); + dst->f[0] = floorf( src->f[0] + 0.5f ); + dst->f[1] = floorf( src->f[1] + 0.5f ); + dst->f[2] = floorf( src->f[2] + 0.5f ); + dst->f[3] = floorf( src->f[3] + 0.5f ); } static void @@ -833,20 +853,20 @@ micro_sin( union tgsi_exec_channel *dst, const union tgsi_exec_channel *src ) { - dst->f[0] = (float) sin( (double) src->f[0] ); - dst->f[1] = (float) sin( (double) src->f[1] ); - dst->f[2] = (float) sin( (double) src->f[2] ); - dst->f[3] = (float) sin( (double) src->f[3] ); + dst->f[0] = sinf( src->f[0] ); + dst->f[1] = sinf( src->f[1] ); + dst->f[2] = sinf( src->f[2] ); + dst->f[3] = sinf( src->f[3] ); } static void micro_sqrt( union tgsi_exec_channel *dst, const union tgsi_exec_channel *src ) { - dst->f[0] = (float) sqrt( (double) src->f[0] ); - dst->f[1] = (float) sqrt( (double) src->f[1] ); - dst->f[2] = (float) sqrt( (double) src->f[2] ); - dst->f[3] = (float) sqrt( (double) src->f[3] ); + dst->f[0] = sqrtf( src->f[0] ); + dst->f[1] = sqrtf( src->f[1] ); + dst->f[2] = sqrtf( src->f[2] ); + dst->f[3] = sqrtf( src->f[3] ); } static void @@ -1516,41 +1536,44 @@ exec_instruction( break; case TGSI_OPCODE_EXP: - debug_printf("TGSI: EXP opcode not implemented\n"); - /* from ARB_v_p: - tmp = ScalarLoad(op0); - result.x = 2^floor(tmp); - result.y = tmp - floor(tmp); - result.z = RoughApprox2ToX(tmp); - result.w = 1.0; - */ -#if 0 - /* something like this: */ FETCH( &r[0], 0, CHAN_X ); - micro_exp2( &r[0], &r[0] ); - FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { - STORE( &r[0], 0, chan_index ); + micro_flr( &r[1], &r[0] ); /* r1 = floor(r0) */ + if (IS_CHANNEL_ENABLED( *inst, CHAN_X )) { + micro_exp2( &r[2], &r[1] ); /* r2 = 2 ^ r1 */ + STORE( &r[2], 0, CHAN_X ); /* store r2 */ + } + if (IS_CHANNEL_ENABLED( *inst, CHAN_Y )) { + micro_sub( &r[2], &r[0], &r[1] ); /* r2 = r0 - r1 */ + STORE( &r[2], 0, CHAN_Y ); /* store r2 */ + } + if (IS_CHANNEL_ENABLED( *inst, CHAN_Z )) { + micro_exp2( &r[2], &r[0] ); /* r2 = 2 ^ r0 */ + STORE( &r[2], 0, CHAN_Z ); /* store r2 */ + } + if (IS_CHANNEL_ENABLED( *inst, CHAN_W )) { + STORE( &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], 0, CHAN_W ); } -#endif break; case TGSI_OPCODE_LOG: - debug_printf("TGSI: LOG opcode not implemented\n"); - /* from ARB_v_p: - tmp = fabs(ScalarLoad(op0)); - result.x = floor(log2(tmp)); - result.y = tmp / 2^(floor(log2(tmp))); - result.z = RoughApproxLog2(tmp); - result.w = 1.0; - */ -#if 0 - /* something like this: */ FETCH( &r[0], 0, CHAN_X ); - micro_lg2( &r[0], &r[0] ); - FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { - STORE( &r[0], 0, chan_index ); + micro_abs( &r[2], &r[0] ); /* r2 = abs(r0) */ + micro_lg2( &r[1], &r[2] ); /* r1 = lg2(r2) */ + micro_flr( &r[0], &r[1] ); /* r0 = floor(r1) */ + if (IS_CHANNEL_ENABLED( *inst, CHAN_X )) { + STORE( &r[0], 0, CHAN_X ); + } + if (IS_CHANNEL_ENABLED( *inst, CHAN_Y )) { + micro_exp2( &r[0], &r[0] ); /* r0 = 2 ^ r0 */ + micro_div( &r[0], &r[2], &r[0] ); /* r0 = r2 / r0 */ + STORE( &r[0], 0, CHAN_Y ); + } + if (IS_CHANNEL_ENABLED( *inst, CHAN_Z )) { + STORE( &r[1], 0, CHAN_Z ); + } + if (IS_CHANNEL_ENABLED( *inst, CHAN_W )) { + STORE( &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], 0, CHAN_W ); } -#endif break; case TGSI_OPCODE_MUL: @@ -1975,7 +1998,7 @@ exec_instruction( FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( &r[0], 0, chan_index ); FETCH( &r[1], 1, chan_index ); - micro_lt( &r[0], &r[0], &r[1], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C] ); + micro_le( &r[0], &r[0], &r[1], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C] ); STORE( &r[0], 0, chan_index ); } break; @@ -1992,7 +2015,7 @@ exec_instruction( FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( &r[0], 0, chan_index ); FETCH( &r[1], 1, chan_index ); - micro_ge( &r[0], &r[0], &r[1], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C] ); + micro_le( &r[0], &r[0], &r[1], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C] ); STORE( &r[0], 0, chan_index ); } break; diff --git a/src/gallium/auxiliary/tgsi/exec/tgsi_exec.h b/src/gallium/auxiliary/tgsi/exec/tgsi_exec.h index 45c49dd007..19bd78df3d 100644 --- a/src/gallium/auxiliary/tgsi/exec/tgsi_exec.h +++ b/src/gallium/auxiliary/tgsi/exec/tgsi_exec.h @@ -133,9 +133,15 @@ struct tgsi_exec_labels #define TGSI_EXEC_TEMP_PRIMITIVE_I 34 #define TGSI_EXEC_TEMP_PRIMITIVE_C 2 -#define TGSI_EXEC_TEMP_R0 35 +#define TGSI_EXEC_TEMP_THREE_I 34 +#define TGSI_EXEC_TEMP_THREE_C 3 -#define TGSI_EXEC_NUM_TEMPS (32 + 4) +#define TGSI_EXEC_TEMP_HALF_I 35 +#define TGSI_EXEC_TEMP_HALF_C 0 + +#define TGSI_EXEC_TEMP_R0 36 + +#define TGSI_EXEC_NUM_TEMPS (32 + 5) #define TGSI_EXEC_NUM_ADDRS 1 #define TGSI_EXEC_NUM_IMMEDIATES 256 @@ -166,7 +172,7 @@ struct tgsi_exec_machine float Imms[TGSI_EXEC_NUM_IMMEDIATES][4]; unsigned ImmLimit; - float (*Consts)[4]; + const float (*Consts)[4]; struct tgsi_exec_vector *Inputs; struct tgsi_exec_vector *Outputs; const struct tgsi_token *Tokens; diff --git a/src/gallium/auxiliary/tgsi/exec/tgsi_sse2.c b/src/gallium/auxiliary/tgsi/exec/tgsi_sse2.c index 4e80597b3f..8018bd7fa4 100755 --- a/src/gallium/auxiliary/tgsi/exec/tgsi_sse2.c +++ b/src/gallium/auxiliary/tgsi/exec/tgsi_sse2.c @@ -34,115 +34,14 @@ #include "rtasm/rtasm_x86sse.h" -#if defined(__i386__) || defined(__386__) +#ifdef PIPE_ARCH_X86 -#define DUMP_SSE 0 - -#if DUMP_SSE - -static void -_print_reg( - struct x86_reg reg ) -{ - if (reg.mod != mod_REG) - debug_printf( "[" ); - - switch( reg.file ) { - case file_REG32: - switch( reg.idx ) { - case reg_AX: - debug_printf( "EAX" ); - break; - case reg_CX: - debug_printf( "ECX" ); - break; - case reg_DX: - debug_printf( "EDX" ); - break; - case reg_BX: - debug_printf( "EBX" ); - break; - case reg_SP: - debug_printf( "ESP" ); - break; - case reg_BP: - debug_printf( "EBP" ); - break; - case reg_SI: - debug_printf( "ESI" ); - break; - case reg_DI: - debug_printf( "EDI" ); - break; - } - break; - case file_MMX: - assert( 0 ); - break; - case file_XMM: - debug_printf( "XMM%u", reg.idx ); - break; - case file_x87: - assert( 0 ); - break; - } - - if (reg.mod == mod_DISP8 || - reg.mod == mod_DISP32) - debug_printf("+%d", reg.disp); - - if (reg.mod != mod_REG) - debug_printf( "]" ); -} - -static void -_fill( - const char *op ) -{ - unsigned count = 10 - strlen( op ); - - while( count-- ) { - debug_printf( " " ); - } -} - -#define DUMP_START() debug_printf( "\nsse-dump start ----------------" ) -#define DUMP_END() debug_printf( "\nsse-dump end ----------------\n" ) -#define DUMP( OP ) debug_printf( "\n%s", OP ) -#define DUMP_I( OP, I ) do {\ - debug_printf( "\n%s", OP );\ - _fill( OP );\ - debug_printf( "%u", I ); } while( 0 ) -#define DUMP_R( OP, R0 ) do {\ - debug_printf( "\n%s", OP );\ - _fill( OP );\ - _print_reg( R0 ); } while( 0 ) -#define DUMP_RR( OP, R0, R1 ) do {\ - debug_printf( "\n%s", OP );\ - _fill( OP );\ - _print_reg( R0 );\ - debug_printf( ", " );\ - _print_reg( R1 ); } while( 0 ) -#define DUMP_RRI( OP, R0, R1, I ) do {\ - debug_printf( "\n%s", OP );\ - _fill( OP );\ - _print_reg( R0 );\ - debug_printf( ", " );\ - _print_reg( R1 );\ - debug_printf( ", " );\ - debug_printf( "%u", I ); } while( 0 ) - -#else - -#define DUMP_START() -#define DUMP_END() -#define DUMP( OP ) -#define DUMP_I( OP, I ) -#define DUMP_R( OP, R0 ) -#define DUMP_RR( OP, R0, R1 ) -#define DUMP_RRI( OP, R0, R1, I ) +/* for 1/sqrt() + * + * This costs about 100fps (close to 10%) in gears: + */ +#define HIGH_PRECISION 1 -#endif #define FOR_EACH_CHANNEL( CHAN )\ for( CHAN = 0; CHAN < 4; CHAN++ ) @@ -208,15 +107,9 @@ get_output_base( void ) static struct x86_reg get_temp_base( void ) { -#ifdef WIN32 return x86_make_reg( file_REG32, reg_BX ); -#else - return x86_make_reg( - file_REG32, - reg_SI ); -#endif } static struct x86_reg @@ -225,17 +118,28 @@ get_coef_base( void ) return get_output_base(); } +static struct x86_reg +get_immediate_base( void ) +{ + return x86_make_reg( + file_REG32, + reg_DI ); +} + + /** * Data access helpers. */ + static struct x86_reg -get_argument( - unsigned index ) +get_immediate( + unsigned vec, + unsigned chan ) { return x86_make_disp( - x86_make_reg( file_REG32, reg_SP ), - (index + 1) * 4 ); + get_immediate_base(), + (vec * 4 + chan) * 4 ); } static struct x86_reg @@ -289,200 +193,6 @@ get_coef( ((vec * 3 + member) * 4 + chan) * 4 ); } -/** - * X86 rtasm wrappers. - */ - -static void -emit_addps( - struct x86_function *func, - struct x86_reg dst, - struct x86_reg src ) -{ - DUMP_RR( "ADDPS", dst, src ); - sse_addps( func, dst, src ); -} - -static void -emit_andnps( - struct x86_function *func, - struct x86_reg dst, - struct x86_reg src ) -{ - DUMP_RR( "ANDNPS", dst, src ); - sse_andnps( func, dst, src ); -} - -static void -emit_andps( - struct x86_function *func, - struct x86_reg dst, - struct x86_reg src ) -{ - DUMP_RR( "ANDPS", dst, src ); - sse_andps( func, dst, src ); -} - -static void -emit_call( - struct x86_function *func, - void (* addr)() ) -{ - struct x86_reg ecx = x86_make_reg( file_REG32, reg_CX ); - - DUMP_I( "CALL", addr ); - x86_mov_reg_imm( func, ecx, (unsigned long) addr ); - x86_call( func, ecx ); -} - -static void -emit_cmpps( - struct x86_function *func, - struct x86_reg dst, - struct x86_reg src, - enum sse_cc cc ) -{ - DUMP_RRI( "CMPPS", dst, src, cc ); - sse_cmpps( func, dst, src, cc ); -} - -static void -emit_cvttps2dq( - struct x86_function *func, - struct x86_reg dst, - struct x86_reg src ) -{ - DUMP_RR( "CVTTPS2DQ", dst, src ); - sse2_cvttps2dq( func, dst, src ); -} - -static void -emit_maxps( - struct x86_function *func, - struct x86_reg dst, - struct x86_reg src ) -{ - DUMP_RR( "MAXPS", dst, src ); - sse_maxps( func, dst, src ); -} - -static void -emit_minps( - struct x86_function *func, - struct x86_reg dst, - struct x86_reg src ) -{ - DUMP_RR( "MINPS", dst, src ); - sse_minps( func, dst, src ); -} - -static void -emit_mov( - struct x86_function *func, - struct x86_reg dst, - struct x86_reg src ) -{ - DUMP_RR( "MOV", dst, src ); - x86_mov( func, dst, src ); -} - -static void -emit_movaps( - struct x86_function *func, - struct x86_reg dst, - struct x86_reg src ) -{ - DUMP_RR( "MOVAPS", dst, src ); - sse_movaps( func, dst, src ); -} - -static void -emit_movss( - struct x86_function *func, - struct x86_reg dst, - struct x86_reg src ) -{ - DUMP_RR( "MOVSS", dst, src ); - sse_movss( func, dst, src ); -} - -static void -emit_movups( - struct x86_function *func, - struct x86_reg dst, - struct x86_reg src ) -{ - DUMP_RR( "MOVUPS", dst, src ); - sse_movups( func, dst, src ); -} - -static void -emit_mulps( - struct x86_function *func, - struct x86_reg dst, - struct x86_reg src ) -{ - DUMP_RR( "MULPS", dst, src ); - sse_mulps( func, dst, src ); -} - -static void -emit_or( - struct x86_function *func, - struct x86_reg dst, - struct x86_reg src ) -{ - DUMP_RR( "OR", dst, src ); - x86_or( func, dst, src ); -} - -static void -emit_orps( - struct x86_function *func, - struct x86_reg dst, - struct x86_reg src ) -{ - DUMP_RR( "ORPS", dst, src ); - sse_orps( func, dst, src ); -} - -static void -emit_pmovmskb( - struct x86_function *func, - struct x86_reg dst, - struct x86_reg src ) -{ - DUMP_RR( "PMOVMSKB", dst, src ); - sse_pmovmskb( func, dst, src ); -} - -static void -emit_pop( - struct x86_function *func, - struct x86_reg dst ) -{ - DUMP_R( "POP", dst ); - x86_pop( func, dst ); -} - -static void -emit_push( - struct x86_function *func, - struct x86_reg dst ) -{ - DUMP_R( "PUSH", dst ); - x86_push( func, dst ); -} - -static void -emit_rcpps( - struct x86_function *func, - struct x86_reg dst, - struct x86_reg src ) -{ - DUMP_RR( "RCPPS", dst, src ); - sse2_rcpps( func, dst, src ); -} #ifdef WIN32 static void @@ -490,7 +200,6 @@ emit_retw( struct x86_function *func, unsigned size ) { - DUMP_I( "RET", size ); x86_retw( func, size ); } #else @@ -498,56 +207,21 @@ static void emit_ret( struct x86_function *func ) { - DUMP( "RET" ); x86_ret( func ); } #endif -static void -emit_rsqrtps( - struct x86_function *func, - struct x86_reg dst, - struct x86_reg src ) -{ - DUMP_RR( "RSQRTPS", dst, src ); - sse_rsqrtps( func, dst, src ); -} - -static void -emit_shufps( - struct x86_function *func, - struct x86_reg dst, - struct x86_reg src, - unsigned char shuf ) -{ - DUMP_RRI( "SHUFPS", dst, src, shuf ); - sse_shufps( func, dst, src, shuf ); -} - -static void -emit_subps( - struct x86_function *func, - struct x86_reg dst, - struct x86_reg src ) -{ - DUMP_RR( "SUBPS", dst, src ); - sse_subps( func, dst, src ); -} - -static void -emit_xorps( - struct x86_function *func, - struct x86_reg dst, - struct x86_reg src ) -{ - DUMP_RR( "XORPS", dst, src ); - sse_xorps( func, dst, src ); -} /** * Data fetch helpers. */ +/** + * Copy a shader constant to xmm register + * \param xmm the destination xmm register + * \param vec the src const buffer index + * \param chan src channel to fetch (X, Y, Z or W) + */ static void emit_const( struct x86_function *func, @@ -555,11 +229,11 @@ emit_const( unsigned vec, unsigned chan ) { - emit_movss( + sse_movss( func, make_xmm( xmm ), get_const( vec, chan ) ); - emit_shufps( + sse_shufps( func, make_xmm( xmm ), make_xmm( xmm ), @@ -567,18 +241,49 @@ emit_const( } static void +emit_immediate( + struct x86_function *func, + unsigned xmm, + unsigned vec, + unsigned chan ) +{ + sse_movss( + func, + make_xmm( xmm ), + get_immediate( vec, chan ) ); + sse_shufps( + func, + make_xmm( xmm ), + make_xmm( xmm ), + SHUF( 0, 0, 0, 0 ) ); +} + + +/** + * Copy a shader input to xmm register + * \param xmm the destination xmm register + * \param vec the src input attrib + * \param chan src channel to fetch (X, Y, Z or W) + */ +static void emit_inputf( struct x86_function *func, unsigned xmm, unsigned vec, unsigned chan ) { - emit_movups( + sse_movups( func, make_xmm( xmm ), get_input( vec, chan ) ); } +/** + * Store an xmm register to a shader output + * \param xmm the source xmm register + * \param vec the dest output attrib + * \param chan src dest channel to store (X, Y, Z or W) + */ static void emit_output( struct x86_function *func, @@ -586,12 +291,18 @@ emit_output( unsigned vec, unsigned chan ) { - emit_movups( + sse_movups( func, get_output( vec, chan ), make_xmm( xmm ) ); } +/** + * Copy a shader temporary to xmm register + * \param xmm the destination xmm register + * \param vec the src temp register + * \param chan src channel to fetch (X, Y, Z or W) + */ static void emit_tempf( struct x86_function *func, @@ -599,12 +310,19 @@ emit_tempf( unsigned vec, unsigned chan ) { - emit_movaps( + sse_movaps( func, make_xmm( xmm ), get_temp( vec, chan ) ); } +/** + * Load an xmm register with an input attrib coefficient (a0, dadx or dady) + * \param xmm the destination xmm register + * \param vec the src input/attribute coefficient index + * \param chan src channel to fetch (X, Y, Z or W) + * \param member 0=a0, 1=dadx, 2=dady + */ static void emit_coef( struct x86_function *func, @@ -613,11 +331,11 @@ emit_coef( unsigned chan, unsigned member ) { - emit_movss( + sse_movss( func, make_xmm( xmm ), get_coef( vec, chan, member ) ); - emit_shufps( + sse_shufps( func, make_xmm( xmm ), make_xmm( xmm ), @@ -635,7 +353,7 @@ emit_inputs( unsigned vec, unsigned chan ) { - emit_movups( + sse_movups( func, get_input( vec, chan ), make_xmm( xmm ) ); @@ -648,7 +366,7 @@ emit_temps( unsigned vec, unsigned chan ) { - emit_movaps( + sse_movaps( func, get_temp( vec, chan ), make_xmm( xmm ) ); @@ -725,41 +443,32 @@ static void emit_push_gp( struct x86_function *func ) { - emit_push( + x86_push( func, - get_const_base() ); - emit_push( + x86_make_reg( file_REG32, reg_AX) ); + x86_push( func, - get_input_base() ); - emit_push( + x86_make_reg( file_REG32, reg_CX) ); + x86_push( func, - get_output_base() ); - - /* It is important on non-win32 platforms that temp base is pushed last. - */ - emit_push( - func, - get_temp_base() ); + x86_make_reg( file_REG32, reg_DX) ); } static void -emit_pop_gp( +x86_pop_gp( struct x86_function *func ) { /* Restore GP registers in a reverse order. */ - emit_pop( - func, - get_temp_base() ); - emit_pop( + x86_pop( func, - get_output_base() ); - emit_pop( + x86_make_reg( file_REG32, reg_DX) ); + x86_pop( func, - get_input_base() ); - emit_pop( + x86_make_reg( file_REG32, reg_CX) ); + x86_pop( func, - get_const_base() ); + x86_make_reg( file_REG32, reg_AX) ); } static void @@ -768,7 +477,7 @@ emit_func_call_dst( unsigned xmm_dst, void (*code)() ) { - emit_movaps( + sse_movaps( func, get_temp( TEMP_R0, 0 ), make_xmm( xmm_dst ) ); @@ -776,20 +485,27 @@ emit_func_call_dst( emit_push_gp( func ); -#ifdef WIN32 - emit_push( - func, - get_temp( TEMP_R0, 0 ) ); + { + struct x86_reg ecx = x86_make_reg( file_REG32, reg_CX ); + + x86_lea( + func, + ecx, + get_temp( TEMP_R0, 0 ) ); + + x86_push( func, ecx ); + x86_mov_reg_imm( func, ecx, (unsigned long) code ); + x86_call( func, ecx ); +#ifndef WIN32 + x86_pop(func, ecx ); #endif + } - emit_call( - func, - code ); - emit_pop_gp( + x86_pop_gp( func ); - emit_movaps( + sse_movaps( func, make_xmm( xmm_dst ), get_temp( TEMP_R0, 0 ) ); @@ -802,7 +518,7 @@ emit_func_call_dst_src( unsigned xmm_src, void (*code)() ) { - emit_movaps( + sse_movaps( func, get_temp( TEMP_R0, 1 ), make_xmm( xmm_src ) ); @@ -822,7 +538,7 @@ emit_abs( struct x86_function *func, unsigned xmm ) { - emit_andps( + sse_andps( func, make_xmm( xmm ), get_temp( @@ -836,7 +552,7 @@ emit_add( unsigned xmm_dst, unsigned xmm_src ) { - emit_addps( + sse_addps( func, make_xmm( xmm_dst ), make_xmm( xmm_src ) ); @@ -846,18 +562,12 @@ static void XSTDCALL cos4f( float *store ) { -#ifdef WIN32 - store[0] = (float) cos( (double) store[0] ); - store[1] = (float) cos( (double) store[1] ); - store[2] = (float) cos( (double) store[2] ); - store[3] = (float) cos( (double) store[3] ); -#else - const unsigned X = TEMP_R0 * 16; + const unsigned X = 0; + store[X + 0] = cosf( store[X + 0] ); store[X + 1] = cosf( store[X + 1] ); store[X + 2] = cosf( store[X + 2] ); store[X + 3] = cosf( store[X + 3] ); -#endif } static void @@ -875,18 +585,12 @@ static void XSTDCALL ex24f( float *store ) { -#ifdef WIN32 - store[0] = (float) pow( 2.0, (double) store[0] ); - store[1] = (float) pow( 2.0, (double) store[1] ); - store[2] = (float) pow( 2.0, (double) store[2] ); - store[3] = (float) pow( 2.0, (double) store[3] ); -#else - const unsigned X = TEMP_R0 * 16; + const unsigned X = 0; + store[X + 0] = powf( 2.0f, store[X + 0] ); store[X + 1] = powf( 2.0f, store[X + 1] ); store[X + 2] = powf( 2.0f, store[X + 2] ); store[X + 3] = powf( 2.0f, store[X + 3] ); -#endif } static void @@ -905,7 +609,7 @@ emit_f2it( struct x86_function *func, unsigned xmm ) { - emit_cvttps2dq( + sse2_cvttps2dq( func, make_xmm( xmm ), make_xmm( xmm ) ); @@ -915,15 +619,12 @@ static void XSTDCALL flr4f( float *store ) { -#ifdef WIN32 const unsigned X = 0; -#else - const unsigned X = TEMP_R0 * 16; -#endif - store[X + 0] = (float) floor( (double) store[X + 0] ); - store[X + 1] = (float) floor( (double) store[X + 1] ); - store[X + 2] = (float) floor( (double) store[X + 2] ); - store[X + 3] = (float) floor( (double) store[X + 3] ); + + store[X + 0] = floorf( store[X + 0] ); + store[X + 1] = floorf( store[X + 1] ); + store[X + 2] = floorf( store[X + 2] ); + store[X + 3] = floorf( store[X + 3] ); } static void @@ -941,15 +642,12 @@ static void XSTDCALL frc4f( float *store ) { -#ifdef WIN32 const unsigned X = 0; -#else - const unsigned X = TEMP_R0 * 16; -#endif - store[X + 0] -= (float) floor( (double) store[X + 0] ); - store[X + 1] -= (float) floor( (double) store[X + 1] ); - store[X + 2] -= (float) floor( (double) store[X + 2] ); - store[X + 3] -= (float) floor( (double) store[X + 3] ); + + store[X + 0] -= floorf( store[X + 0] ); + store[X + 1] -= floorf( store[X + 1] ); + store[X + 2] -= floorf( store[X + 2] ); + store[X + 3] -= floorf( store[X + 3] ); } static void @@ -967,11 +665,8 @@ static void XSTDCALL lg24f( float *store ) { -#ifdef WIN32 const unsigned X = 0; -#else - const unsigned X = TEMP_R0 * 16; -#endif + store[X + 0] = LOG2( store[X + 0] ); store[X + 1] = LOG2( store[X + 1] ); store[X + 2] = LOG2( store[X + 2] ); @@ -995,7 +690,7 @@ emit_MOV( unsigned xmm_dst, unsigned xmm_src ) { - emit_movups( + sse_movups( func, make_xmm( xmm_dst ), make_xmm( xmm_src ) ); @@ -1006,7 +701,7 @@ emit_mul (struct x86_function *func, unsigned xmm_dst, unsigned xmm_src) { - emit_mulps( + sse_mulps( func, make_xmm( xmm_dst ), make_xmm( xmm_src ) ); @@ -1017,7 +712,7 @@ emit_neg( struct x86_function *func, unsigned xmm ) { - emit_xorps( + sse_xorps( func, make_xmm( xmm ), get_temp( @@ -1029,18 +724,12 @@ static void XSTDCALL pow4f( float *store ) { -#ifdef WIN32 - store[0] = (float) pow( (double) store[0], (double) store[4] ); - store[1] = (float) pow( (double) store[1], (double) store[5] ); - store[2] = (float) pow( (double) store[2], (double) store[6] ); - store[3] = (float) pow( (double) store[3], (double) store[7] ); -#else - const unsigned X = TEMP_R0 * 16; + const unsigned X = 0; + store[X + 0] = powf( store[X + 0], store[X + 4] ); store[X + 1] = powf( store[X + 1], store[X + 5] ); store[X + 2] = powf( store[X + 2], store[X + 6] ); store[X + 3] = powf( store[X + 3], store[X + 7] ); -#endif } static void @@ -1062,7 +751,11 @@ emit_rcp ( unsigned xmm_dst, unsigned xmm_src ) { - emit_rcpps( + /* On Intel CPUs at least, this is only accurate to 12 bits -- not + * good enough. Need to either emit a proper divide or use the + * iterative technique described below in emit_rsqrt(). + */ + sse2_rcpps( func, make_xmm( xmm_dst ), make_xmm( xmm_src ) ); @@ -1074,10 +767,44 @@ emit_rsqrt( unsigned xmm_dst, unsigned xmm_src ) { - emit_rsqrtps( +#if HIGH_PRECISION + /* Although rsqrtps() and rcpps() are low precision on some/all SSE + * implementations, it is possible to improve its precision at + * fairly low cost, using a newton/raphson step, as below: + * + * x1 = 2 * rcpps(a) - a * rcpps(a) * rcpps(a) + * x1 = 0.5 * rsqrtps(a) * [3.0 - (a * rsqrtps(a))* rsqrtps(a)] + * + * See: http://softwarecommunity.intel.com/articles/eng/1818.htm + */ + { + struct x86_reg dst = make_xmm( xmm_dst ); + struct x86_reg src = make_xmm( xmm_src ); + struct x86_reg tmp0 = make_xmm( 2 ); + struct x86_reg tmp1 = make_xmm( 3 ); + + assert( xmm_dst != xmm_src ); + assert( xmm_dst != 2 && xmm_dst != 3 ); + assert( xmm_src != 2 && xmm_src != 3 ); + + sse_movaps( func, dst, get_temp( TGSI_EXEC_TEMP_HALF_I, TGSI_EXEC_TEMP_HALF_C ) ); + sse_movaps( func, tmp0, get_temp( TGSI_EXEC_TEMP_THREE_I, TGSI_EXEC_TEMP_THREE_C ) ); + sse_rsqrtps( func, tmp1, src ); + sse_mulps( func, src, tmp1 ); + sse_mulps( func, dst, tmp1 ); + sse_mulps( func, src, tmp1 ); + sse_subps( func, tmp0, src ); + sse_mulps( func, dst, tmp0 ); + } +#else + /* On Intel CPUs at least, this is only accurate to 12 bits -- not + * good enough. + */ + sse_rsqrtps( func, make_xmm( xmm_dst ), make_xmm( xmm_src ) ); +#endif } static void @@ -1085,7 +812,7 @@ emit_setsign( struct x86_function *func, unsigned xmm ) { - emit_orps( + sse_orps( func, make_xmm( xmm ), get_temp( @@ -1097,18 +824,12 @@ static void XSTDCALL sin4f( float *store ) { -#ifdef WIN32 - store[0] = (float) sin( (double) store[0] ); - store[1] = (float) sin( (double) store[1] ); - store[2] = (float) sin( (double) store[2] ); - store[3] = (float) sin( (double) store[3] ); -#else - const unsigned X = TEMP_R0 * 16; + const unsigned X = 0; + store[X + 0] = sinf( store[X + 0] ); store[X + 1] = sinf( store[X + 1] ); store[X + 2] = sinf( store[X + 2] ); store[X + 3] = sinf( store[X + 3] ); -#endif } static void @@ -1127,7 +848,7 @@ emit_sub( unsigned xmm_dst, unsigned xmm_src ) { - emit_subps( + sse_subps( func, make_xmm( xmm_dst ), make_xmm( xmm_src ) ); @@ -1160,6 +881,14 @@ emit_fetch( swizzle ); break; + case TGSI_FILE_IMMEDIATE: + emit_immediate( + func, + xmm, + reg->SrcRegister.Index, + swizzle ); + break; + case TGSI_FILE_INPUT: emit_inputf( func, @@ -1328,16 +1057,16 @@ emit_kil( } } - emit_push( + x86_push( func, x86_make_reg( file_REG32, reg_AX ) ); - emit_push( + x86_push( func, x86_make_reg( file_REG32, reg_DX ) ); FOR_EACH_CHANNEL( chan_index ) { if( uniquemask & (1 << chan_index) ) { - emit_cmpps( + sse_cmpps( func, make_xmm( registers[chan_index] ), get_temp( @@ -1346,17 +1075,17 @@ emit_kil( cc_LessThan ); if( chan_index == firstchan ) { - emit_pmovmskb( + sse_pmovmskb( func, x86_make_reg( file_REG32, reg_AX ), make_xmm( registers[chan_index] ) ); } else { - emit_pmovmskb( + sse_pmovmskb( func, x86_make_reg( file_REG32, reg_DX ), make_xmm( registers[chan_index] ) ); - emit_or( + x86_or( func, x86_make_reg( file_REG32, reg_AX ), x86_make_reg( file_REG32, reg_DX ) ); @@ -1364,17 +1093,17 @@ emit_kil( } } - emit_or( + x86_or( func, get_temp( TGSI_EXEC_TEMP_KILMASK_I, TGSI_EXEC_TEMP_KILMASK_C ), x86_make_reg( file_REG32, reg_AX ) ); - emit_pop( + x86_pop( func, x86_make_reg( file_REG32, reg_DX ) ); - emit_pop( + x86_pop( func, x86_make_reg( file_REG32, reg_AX ) ); } @@ -1390,12 +1119,12 @@ emit_setcc( FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( func, *inst, 0, 0, chan_index ); FETCH( func, *inst, 1, 1, chan_index ); - emit_cmpps( + sse_cmpps( func, make_xmm( 0 ), make_xmm( 1 ), cc ); - emit_andps( + sse_andps( func, make_xmm( 0 ), get_temp( @@ -1416,22 +1145,22 @@ emit_cmp( FETCH( func, *inst, 0, 0, chan_index ); FETCH( func, *inst, 1, 1, chan_index ); FETCH( func, *inst, 2, 2, chan_index ); - emit_cmpps( + sse_cmpps( func, make_xmm( 0 ), get_temp( TGSI_EXEC_TEMP_00000000_I, TGSI_EXEC_TEMP_00000000_C ), cc_LessThan ); - emit_andps( + sse_andps( func, make_xmm( 1 ), make_xmm( 0 ) ); - emit_andnps( + sse_andnps( func, make_xmm( 0 ), make_xmm( 2 ) ); - emit_orps( + sse_orps( func, make_xmm( 0 ), make_xmm( 1 ) ); @@ -1448,11 +1177,16 @@ emit_instruction( switch( inst->Instruction.Opcode ) { case TGSI_OPCODE_ARL: +#if 0 + /* XXX this isn't working properly (see glean vertProg1 test) */ FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( func, *inst, 0, 0, chan_index ); emit_f2it( func, 0 ); STORE( func, *inst, 0, 0, chan_index ); } +#else + return 0; +#endif break; case TGSI_OPCODE_MOV: @@ -1482,7 +1216,7 @@ emit_instruction( IS_DST0_CHANNEL_ENABLED( *inst, CHAN_Z ) ) { if( IS_DST0_CHANNEL_ENABLED( *inst, CHAN_Y ) ) { FETCH( func, *inst, 0, 0, CHAN_X ); - emit_maxps( + sse_maxps( func, make_xmm( 0 ), get_temp( @@ -1491,21 +1225,26 @@ emit_instruction( STORE( func, *inst, 0, 0, CHAN_Y ); } if( IS_DST0_CHANNEL_ENABLED( *inst, CHAN_Z ) ) { + /* XMM[1] = SrcReg[0].yyyy */ FETCH( func, *inst, 1, 0, CHAN_Y ); - emit_maxps( + /* XMM[1] = max(XMM[1], 0) */ + sse_maxps( func, make_xmm( 1 ), get_temp( TGSI_EXEC_TEMP_00000000_I, TGSI_EXEC_TEMP_00000000_C ) ); + /* XMM[2] = SrcReg[0].wwww */ FETCH( func, *inst, 2, 0, CHAN_W ); - emit_minps( + /* XMM[2] = min(XMM[2], 128.0) */ + sse_minps( func, make_xmm( 2 ), get_temp( TGSI_EXEC_TEMP_128_I, TGSI_EXEC_TEMP_128_C ) ); - emit_maxps( + /* XMM[2] = max(XMM[2], -128.0) */ + sse_maxps( func, make_xmm( 2 ), get_temp( @@ -1513,16 +1252,16 @@ emit_instruction( TGSI_EXEC_TEMP_MINUS_128_C ) ); emit_pow( func, 1, 2 ); FETCH( func, *inst, 0, 0, CHAN_X ); - emit_xorps( + sse_xorps( func, make_xmm( 2 ), make_xmm( 2 ) ); - emit_cmpps( + sse_cmpps( func, make_xmm( 2 ), make_xmm( 0 ), cc_LessThanEqual ); - emit_andps( + sse_andps( func, make_xmm( 2 ), make_xmm( 1 ) ); @@ -1543,9 +1282,9 @@ emit_instruction( case TGSI_OPCODE_RSQ: /* TGSI_OPCODE_RECIPSQRT */ FETCH( func, *inst, 0, 0, CHAN_X ); - emit_rsqrt( func, 0, 0 ); + emit_rsqrt( func, 1, 0 ); FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) { - STORE( func, *inst, 0, 0, chan_index ); + STORE( func, *inst, 1, 0, chan_index ); } break; @@ -1644,7 +1383,7 @@ emit_instruction( FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( func, *inst, 0, 0, chan_index ); FETCH( func, *inst, 1, 1, chan_index ); - emit_minps( + sse_minps( func, make_xmm( 0 ), make_xmm( 1 ) ); @@ -1656,7 +1395,7 @@ emit_instruction( FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( func, *inst, 0, 0, chan_index ); FETCH( func, *inst, 1, 1, chan_index ); - emit_maxps( + sse_maxps( func, make_xmm( 0 ), make_xmm( 1 ) ); @@ -1997,7 +1736,6 @@ emit_instruction( break; case TGSI_OPCODE_RET: - case TGSI_OPCODE_END: #ifdef WIN32 emit_retw( func, 16 ); #else @@ -2005,6 +1743,9 @@ emit_instruction( #endif break; + case TGSI_OPCODE_END: + break; + case TGSI_OPCODE_SSG: return 0; break; @@ -2020,7 +1761,7 @@ emit_instruction( STORE( func, *inst, 0, 0, CHAN_X ); } IF_IS_DST0_CHANNEL_ENABLED( *inst, CHAN_Y ) { - FETCH( func, *inst, 0, 0, CHAN_Y ); + FETCH( func, *inst, 0, 0, CHAN_X ); emit_sin( func, 0 ); STORE( func, *inst, 0, 0, CHAN_Y ); } @@ -2236,149 +1977,289 @@ emit_declaration( } } -unsigned -tgsi_emit_sse2( - struct tgsi_token *tokens, - struct x86_function *func ) -{ - struct tgsi_parse_context parse; - unsigned ok = 1; - - DUMP_START(); - - func->csr = func->store; - - emit_mov( - func, - get_input_base(), - get_argument( 0 ) ); - emit_mov( - func, - get_output_base(), - get_argument( 1 ) ); - emit_mov( - func, - get_const_base(), - get_argument( 2 ) ); - emit_mov( - func, - get_temp_base(), - get_argument( 3 ) ); - - tgsi_parse_init( &parse, tokens ); - - while( !tgsi_parse_end_of_tokens( &parse ) && ok ) { - tgsi_parse_token( &parse ); - - switch( parse.FullToken.Token.Type ) { - case TGSI_TOKEN_TYPE_DECLARATION: - break; - - case TGSI_TOKEN_TYPE_INSTRUCTION: - ok = emit_instruction( - func, - &parse.FullToken.FullInstruction ); - - if (!ok) { - debug_printf("failed to translate tgsi opcode %d to SSE\n", - parse.FullToken.FullInstruction.Instruction.Opcode ); - } - break; - - case TGSI_TOKEN_TYPE_IMMEDIATE: - /* XXX implement this */ - ok = 0; - debug_printf("failed to emit immediate value to SSE\n"); - break; - - default: - assert( 0 ); - ok = 0; - break; - } +static void aos_to_soa( struct x86_function *func, + uint arg_aos, + uint arg_soa, + uint arg_num, + uint arg_stride ) +{ + struct x86_reg soa_input = x86_make_reg( file_REG32, reg_AX ); + struct x86_reg aos_input = x86_make_reg( file_REG32, reg_BX ); + struct x86_reg num_inputs = x86_make_reg( file_REG32, reg_CX ); + struct x86_reg stride = x86_make_reg( file_REG32, reg_DX ); + int inner_loop; + + + /* Save EBX */ + x86_push( func, x86_make_reg( file_REG32, reg_BX ) ); + + x86_mov( func, aos_input, x86_fn_arg( func, arg_aos ) ); + x86_mov( func, soa_input, x86_fn_arg( func, arg_soa ) ); + x86_mov( func, num_inputs, x86_fn_arg( func, arg_num ) ); + x86_mov( func, stride, x86_fn_arg( func, arg_stride ) ); + + /* do */ + inner_loop = x86_get_label( func ); + { + x86_push( func, aos_input ); + sse_movlps( func, make_xmm( 0 ), x86_make_disp( aos_input, 0 ) ); + sse_movlps( func, make_xmm( 3 ), x86_make_disp( aos_input, 8 ) ); + x86_add( func, aos_input, stride ); + sse_movhps( func, make_xmm( 0 ), x86_make_disp( aos_input, 0 ) ); + sse_movhps( func, make_xmm( 3 ), x86_make_disp( aos_input, 8 ) ); + x86_add( func, aos_input, stride ); + sse_movlps( func, make_xmm( 1 ), x86_make_disp( aos_input, 0 ) ); + sse_movlps( func, make_xmm( 4 ), x86_make_disp( aos_input, 8 ) ); + x86_add( func, aos_input, stride ); + sse_movhps( func, make_xmm( 1 ), x86_make_disp( aos_input, 0 ) ); + sse_movhps( func, make_xmm( 4 ), x86_make_disp( aos_input, 8 ) ); + x86_pop( func, aos_input ); + + sse_movaps( func, make_xmm( 2 ), make_xmm( 0 ) ); + sse_movaps( func, make_xmm( 5 ), make_xmm( 3 ) ); + sse_shufps( func, make_xmm( 0 ), make_xmm( 1 ), 0x88 ); + sse_shufps( func, make_xmm( 2 ), make_xmm( 1 ), 0xdd ); + sse_shufps( func, make_xmm( 3 ), make_xmm( 4 ), 0x88 ); + sse_shufps( func, make_xmm( 5 ), make_xmm( 4 ), 0xdd ); + + sse_movups( func, x86_make_disp( soa_input, 0 ), make_xmm( 0 ) ); + sse_movups( func, x86_make_disp( soa_input, 16 ), make_xmm( 2 ) ); + sse_movups( func, x86_make_disp( soa_input, 32 ), make_xmm( 3 ) ); + sse_movups( func, x86_make_disp( soa_input, 48 ), make_xmm( 5 ) ); + + /* Advance to next input */ + x86_lea( func, aos_input, x86_make_disp(aos_input, 16) ); + x86_lea( func, soa_input, x86_make_disp(soa_input, 64) ); } + /* while --num_inputs */ + x86_dec( func, num_inputs ); + x86_jcc( func, cc_NE, inner_loop ); + + /* Restore EBX */ + x86_pop( func, aos_input ); +} + +static void soa_to_aos( struct x86_function *func, uint aos, uint soa, uint num, uint stride ) +{ + struct x86_reg soa_output; + struct x86_reg aos_output; + struct x86_reg num_outputs; + struct x86_reg temp; + int inner_loop; + + soa_output = x86_make_reg( file_REG32, reg_AX ); + aos_output = x86_make_reg( file_REG32, reg_BX ); + num_outputs = x86_make_reg( file_REG32, reg_CX ); + temp = x86_make_reg( file_REG32, reg_DX ); + + /* Save EBX */ + x86_push( func, aos_output ); + + x86_mov( func, soa_output, x86_fn_arg( func, soa ) ); + x86_mov( func, aos_output, x86_fn_arg( func, aos ) ); + x86_mov( func, num_outputs, x86_fn_arg( func, num ) ); + + /* do */ + inner_loop = x86_get_label( func ); + { + sse_movups( func, make_xmm( 0 ), x86_make_disp( soa_output, 0 ) ); + sse_movups( func, make_xmm( 1 ), x86_make_disp( soa_output, 16 ) ); + sse_movups( func, make_xmm( 3 ), x86_make_disp( soa_output, 32 ) ); + sse_movups( func, make_xmm( 4 ), x86_make_disp( soa_output, 48 ) ); + + sse_movaps( func, make_xmm( 2 ), make_xmm( 0 ) ); + sse_movaps( func, make_xmm( 5 ), make_xmm( 3 ) ); + sse_unpcklps( func, make_xmm( 0 ), make_xmm( 1 ) ); + sse_unpckhps( func, make_xmm( 2 ), make_xmm( 1 ) ); + sse_unpcklps( func, make_xmm( 3 ), make_xmm( 4 ) ); + sse_unpckhps( func, make_xmm( 5 ), make_xmm( 4 ) ); + + x86_mov( func, temp, x86_fn_arg( func, stride ) ); + x86_push( func, aos_output ); + sse_movlps( func, x86_make_disp( aos_output, 0 ), make_xmm( 0 ) ); + sse_movlps( func, x86_make_disp( aos_output, 8 ), make_xmm( 3 ) ); + x86_add( func, aos_output, temp ); + sse_movhps( func, x86_make_disp( aos_output, 0 ), make_xmm( 0 ) ); + sse_movhps( func, x86_make_disp( aos_output, 8 ), make_xmm( 3 ) ); + x86_add( func, aos_output, temp ); + sse_movlps( func, x86_make_disp( aos_output, 0 ), make_xmm( 2 ) ); + sse_movlps( func, x86_make_disp( aos_output, 8 ), make_xmm( 5 ) ); + x86_add( func, aos_output, temp ); + sse_movhps( func, x86_make_disp( aos_output, 0 ), make_xmm( 2 ) ); + sse_movhps( func, x86_make_disp( aos_output, 8 ), make_xmm( 5 ) ); + x86_pop( func, aos_output ); + + /* Advance to next output */ + x86_lea( func, aos_output, x86_make_disp(aos_output, 16) ); + x86_lea( func, soa_output, x86_make_disp(soa_output, 64) ); + } + /* while --num_outputs */ + x86_dec( func, num_outputs ); + x86_jcc( func, cc_NE, inner_loop ); - tgsi_parse_free( &parse ); - - DUMP_END(); - - return ok; + /* Restore EBX */ + x86_pop( func, aos_output ); } /** - * Fragment shaders are responsible for interpolating shader inputs. Because on - * x86 we have only 4 GP registers, and here we have 5 shader arguments (input, - * output, const, temp and coef), the code is split into two phases -- - * DECLARATION and INSTRUCTION phase. - * GP register holding the output argument is aliased with the coeff argument, - * as outputs are not needed in the DECLARATION phase. + * Translate a TGSI vertex/fragment shader to SSE2 code. + * Slightly different things are done for vertex vs. fragment shaders. + * + * Note that fragment shaders are responsible for interpolating shader + * inputs. Because on x86 we have only 4 GP registers, and here we + * have 5 shader arguments (input, output, const, temp and coef), the + * code is split into two phases -- DECLARATION and INSTRUCTION phase. + * GP register holding the output argument is aliased with the coeff + * argument, as outputs are not needed in the DECLARATION phase. + * + * \param tokens the TGSI input shader + * \param func the output SSE code/function + * \param immediates buffer to place immediates, later passed to SSE func + * \param return 1 for success, 0 if translation failed */ unsigned -tgsi_emit_sse2_fs( - struct tgsi_token *tokens, - struct x86_function *func ) +tgsi_emit_sse2( + const struct tgsi_token *tokens, + struct x86_function *func, + float (*immediates)[4], + boolean do_swizzles ) { struct tgsi_parse_context parse; boolean instruction_phase = FALSE; unsigned ok = 1; - - DUMP_START(); + uint num_immediates = 0; func->csr = func->store; - /* DECLARATION phase, do not load output argument. */ - emit_mov( - func, - get_input_base(), - get_argument( 0 ) ); - emit_mov( - func, - get_const_base(), - get_argument( 2 ) ); - emit_mov( + tgsi_parse_init( &parse, tokens ); + + /* Can't just use EDI, EBX without save/restoring them: + */ + x86_push( func, - get_temp_base(), - get_argument( 3 ) ); - emit_mov( + get_immediate_base() ); + + x86_push( func, - get_coef_base(), - get_argument( 4 ) ); + get_temp_base() ); - tgsi_parse_init( &parse, tokens ); + + /* + * Different function args for vertex/fragment shaders: + */ + if (parse.FullHeader.Processor.Processor == TGSI_PROCESSOR_FRAGMENT) { + /* DECLARATION phase, do not load output argument. */ + x86_mov( + func, + get_input_base(), + x86_fn_arg( func, 1 ) ); + /* skipping outputs argument here */ + x86_mov( + func, + get_const_base(), + x86_fn_arg( func, 3 ) ); + x86_mov( + func, + get_temp_base(), + x86_fn_arg( func, 4 ) ); + x86_mov( + func, + get_coef_base(), + x86_fn_arg( func, 5 ) ); + x86_mov( + func, + get_immediate_base(), + x86_fn_arg( func, 6 ) ); + } + else { + assert(parse.FullHeader.Processor.Processor == TGSI_PROCESSOR_VERTEX); + + if (do_swizzles) + aos_to_soa( func, + 6, /* aos_input */ + 1, /* machine->input */ + 7, /* num_inputs */ + 8 ); /* input_stride */ + + x86_mov( + func, + get_input_base(), + x86_fn_arg( func, 1 ) ); + x86_mov( + func, + get_output_base(), + x86_fn_arg( func, 2 ) ); + x86_mov( + func, + get_const_base(), + x86_fn_arg( func, 3 ) ); + x86_mov( + func, + get_temp_base(), + x86_fn_arg( func, 4 ) ); + x86_mov( + func, + get_immediate_base(), + x86_fn_arg( func, 5 ) ); + } while( !tgsi_parse_end_of_tokens( &parse ) && ok ) { tgsi_parse_token( &parse ); switch( parse.FullToken.Token.Type ) { case TGSI_TOKEN_TYPE_DECLARATION: - emit_declaration( - func, - &parse.FullToken.FullDeclaration ); + if (parse.FullHeader.Processor.Processor == TGSI_PROCESSOR_FRAGMENT) { + emit_declaration( + func, + &parse.FullToken.FullDeclaration ); + } break; case TGSI_TOKEN_TYPE_INSTRUCTION: - if( !instruction_phase ) { - /* INSTRUCTION phase, overwrite coeff with output. */ - instruction_phase = TRUE; - emit_mov( - func, - get_output_base(), - get_argument( 1 ) ); + if (parse.FullHeader.Processor.Processor == TGSI_PROCESSOR_FRAGMENT) { + if( !instruction_phase ) { + /* INSTRUCTION phase, overwrite coeff with output. */ + instruction_phase = TRUE; + x86_mov( + func, + get_output_base(), + x86_fn_arg( func, 2 ) ); + } } + ok = emit_instruction( func, &parse.FullToken.FullInstruction ); if (!ok) { - debug_printf("failed to translate tgsi opcode %d to SSE\n", - parse.FullToken.FullInstruction.Instruction.Opcode ); + debug_printf("failed to translate tgsi opcode %d to SSE (%s)\n", + parse.FullToken.FullInstruction.Instruction.Opcode, + parse.FullHeader.Processor.Processor == TGSI_PROCESSOR_VERTEX ? + "vertex shader" : "fragment shader"); } break; case TGSI_TOKEN_TYPE_IMMEDIATE: - /* XXX implement this */ - ok = 0; - debug_printf("failed to emit immediate value to SSE\n"); + /* simply copy the immediate values into the next immediates[] slot */ + { + const uint size = parse.FullToken.FullImmediate.Immediate.Size - 1; + uint i; + assert(size <= 4); + assert(num_immediates < TGSI_EXEC_NUM_IMMEDIATES); + for( i = 0; i < size; i++ ) { + immediates[num_immediates][i] = + parse.FullToken.FullImmediate.u.ImmediateFloat32[i].Float; + } +#if 0 + debug_printf("SSE FS immediate[%d] = %f %f %f %f\n", + num_immediates, + immediates[num_immediates][0], + immediates[num_immediates][1], + immediates[num_immediates][2], + immediates[num_immediates][3]); +#endif + num_immediates++; + } break; default: @@ -2387,11 +2268,30 @@ tgsi_emit_sse2_fs( } } - tgsi_parse_free( &parse ); + if (parse.FullHeader.Processor.Processor == TGSI_PROCESSOR_VERTEX) { + if (do_swizzles) + soa_to_aos( func, 9, 2, 10, 11 ); + } - DUMP_END(); + /* Can't just use EBX, EDI without save/restoring them: + */ + x86_pop( + func, + get_temp_base() ); + + x86_pop( + func, + get_immediate_base() ); + +#ifdef WIN32 + emit_retw( func, 16 ); +#else + emit_ret( func ); +#endif + + tgsi_parse_free( &parse ); return ok; } -#endif /* i386 */ +#endif /* PIPE_ARCH_X86 */ diff --git a/src/gallium/auxiliary/tgsi/exec/tgsi_sse2.h b/src/gallium/auxiliary/tgsi/exec/tgsi_sse2.h index 63b8ef3911..e66d115283 100755 --- a/src/gallium/auxiliary/tgsi/exec/tgsi_sse2.h +++ b/src/gallium/auxiliary/tgsi/exec/tgsi_sse2.h @@ -10,13 +10,10 @@ struct x86_function; unsigned tgsi_emit_sse2( - struct tgsi_token *tokens, - struct x86_function *function ); - -unsigned -tgsi_emit_sse2_fs( - struct tgsi_token *tokens, - struct x86_function *function ); + const struct tgsi_token *tokens, + struct x86_function *function, + float (*immediates)[4], + boolean do_swizzles ); #if defined __cplusplus } diff --git a/src/gallium/auxiliary/tgsi/util/tgsi_dump.c b/src/gallium/auxiliary/tgsi/util/tgsi_dump.c index 7d292778ad..4c65ffd780 100644 --- a/src/gallium/auxiliary/tgsi/util/tgsi_dump.c +++ b/src/gallium/auxiliary/tgsi/util/tgsi_dump.c @@ -25,205 +25,36 @@ * **************************************************************************/ -#include <stdio.h> - #include "pipe/p_debug.h" #include "pipe/p_util.h" #include "pipe/p_shader_tokens.h" +#include "util/u_string.h" #include "tgsi_dump.h" #include "tgsi_parse.h" #include "tgsi_build.h" -struct gen_dump -{ - unsigned tabs; - void (* write)( - struct gen_dump *dump, - const void *data, - unsigned size ); -}; - -struct text_dump -{ - struct gen_dump base; - char *text; - unsigned length; - unsigned capacity; -}; - -static void -_text_dump_write( - struct gen_dump *dump, - const void *data, - unsigned size ) -{ - struct text_dump *td = (struct text_dump *) dump; - unsigned new_length = td->length + size; - - if( new_length >= td->capacity ) { - unsigned new_capacity = td->capacity; - - do { - if( new_capacity == 0 ) { - new_capacity = 256; - } - else { - new_capacity *= 2; - } - } while( new_length >= new_capacity ); - td->text = (char *) REALLOC( - td->text, - td->capacity, - new_capacity ); - td->capacity = new_capacity; - } - memcpy( - &td->text[td->length], - data, - size ); - td->length = new_length; - td->text[td->length] = '\0'; -} - -struct file_dump -{ - struct gen_dump base; - FILE *file; -}; - -static void -_file_dump_write( - struct gen_dump *dump, - const void *data, - unsigned size ) -{ - struct file_dump *fd = (struct file_dump *) dump; - -#if 0 - fwrite( data, 1, size, fd->file ); -#else - { - unsigned i; - - for (i = 0; i < size; i++ ) { - fprintf( fd->file, "%c", ((const char *) data)[i] ); - } - } -#endif -} - -static void -gen_dump_str( - struct gen_dump *dump, - const char *str ) -{ - unsigned i; - size_t len = strlen( str ); - - for (i = 0; i < len; i++) { - dump->write( dump, &str[i], 1 ); - if (str[i] == '\n') { - unsigned i; - - for (i = 0; i < dump->tabs; i++) { - dump->write( dump, " ", 4 ); - } - } - } -} - static void -gen_dump_chr( - struct gen_dump *dump, - const char chr ) -{ - dump->write( dump, &chr, 1 ); -} - -static void -gen_dump_uix( - struct gen_dump *dump, - const unsigned ui ) -{ - char str[36]; - - sprintf( str, "0x%x", ui ); - gen_dump_str( dump, str ); -} - -static void -gen_dump_uid( - struct gen_dump *dump, - const unsigned ui ) -{ - char str[16]; - - sprintf( str, "%u", ui ); - gen_dump_str( dump, str ); -} - -static void -gen_dump_sid( - struct gen_dump *dump, - const int si ) -{ - char str[16]; - - sprintf( str, "%d", si ); - gen_dump_str( dump, str ); -} - -static void -gen_dump_flt( - struct gen_dump *dump, - const float flt ) -{ - char str[48]; - - sprintf( str, "%10.4f", flt ); - gen_dump_str( dump, str ); -} - -static void -gen_dump_enum( - struct gen_dump *dump, +dump_enum( const unsigned e, const char **enums, const unsigned enums_count ) { if (e >= enums_count) { - gen_dump_uid( dump, e ); + debug_printf( "%u", e ); } else { - gen_dump_str( dump, enums[e] ); + debug_printf( "%s", enums[e] ); } } -static void -gen_dump_tab( - struct gen_dump *dump ) -{ - ++dump->tabs; -} - -static void -gen_dump_untab( - struct gen_dump *dump ) -{ - assert( dump->tabs > 0 ); - - --dump->tabs; -} - -#define TXT(S) gen_dump_str( dump, S ) -#define CHR(C) gen_dump_chr( dump, C ) -#define UIX(I) gen_dump_uix( dump, I ) -#define UID(I) gen_dump_uid( dump, I ) -#define SID(I) gen_dump_sid( dump, I ) -#define FLT(F) gen_dump_flt( dump, F ) -#define TAB() gen_dump_tab( dump ) -#define UNT() gen_dump_untab( dump ) -#define ENM(E,ENUMS) gen_dump_enum( dump, E, ENUMS, sizeof( ENUMS ) / sizeof( *ENUMS ) ) +#define EOL() debug_printf( "\n" ) +#define TXT(S) debug_printf( "%s", S ) +#define CHR(C) debug_printf( "%c", C ) +#define UIX(I) debug_printf( "0x%x", I ) +#define UID(I) debug_printf( "%u", I ) +#define SID(I) debug_printf( "%d", I ) +#define FLT(F) debug_printf( "%10.4f", F ) +#define ENM(E,ENUMS) dump_enum( E, ENUMS, sizeof( ENUMS ) / sizeof( *ENUMS ) ) static const char *TGSI_PROCESSOR_TYPES[] = { @@ -710,7 +541,6 @@ static const char *TGSI_MODULATES[] = static void dump_declaration_short( - struct gen_dump *dump, struct tgsi_full_declaration *decl ) { TXT( "\nDCL " ); @@ -746,23 +576,24 @@ dump_declaration_short( } } - if( decl->Declaration.Interpolate ) { + if (decl->Declaration.Semantic) { TXT( ", " ); - ENM( decl->Interpolation.Interpolate, TGSI_INTERPOLATES_SHORT ); + ENM( decl->Semantic.SemanticName, TGSI_SEMANTICS_SHORT ); + if (decl->Semantic.SemanticIndex != 0) { + CHR( '[' ); + UID( decl->Semantic.SemanticIndex ); + CHR( ']' ); + } } - if( decl->Declaration.Semantic ) { + if (decl->Declaration.Interpolate) { TXT( ", " ); - ENM( decl->Semantic.SemanticName, TGSI_SEMANTICS_SHORT ); - CHR( '[' ); - UID( decl->Semantic.SemanticIndex ); - CHR( ']' ); + ENM( decl->Interpolation.Interpolate, TGSI_INTERPOLATES_SHORT ); } } static void dump_declaration_verbose( - struct gen_dump *dump, struct tgsi_full_declaration *decl, unsigned ignored, unsigned deflt, @@ -800,7 +631,7 @@ dump_declaration_verbose( UIX( decl->Declaration.Padding ); } - CHR( '\n' ); + EOL(); switch( decl->Declaration.Declare ) { case TGSI_DECLARE_RANGE: TXT( "\nFirst: " ); @@ -819,7 +650,7 @@ dump_declaration_verbose( } if( decl->Declaration.Interpolate ) { - CHR( '\n' ); + EOL(); TXT( "\nInterpolate: " ); ENM( decl->Interpolation.Interpolate, TGSI_INTERPOLATES ); if( ignored ) { @@ -829,7 +660,7 @@ dump_declaration_verbose( } if( decl->Declaration.Semantic ) { - CHR( '\n' ); + EOL(); TXT( "\nSemanticName : " ); ENM( decl->Semantic.SemanticName, TGSI_SEMANTICS ); TXT( "\nSemanticIndex: " ); @@ -843,7 +674,6 @@ dump_declaration_verbose( static void dump_immediate_short( - struct gen_dump *dump, struct tgsi_full_immediate *imm ) { unsigned i; @@ -871,7 +701,6 @@ dump_immediate_short( static void dump_immediate_verbose( - struct gen_dump *dump, struct tgsi_full_immediate *imm, unsigned ignored ) { @@ -885,7 +714,7 @@ dump_immediate_verbose( } for( i = 0; i < imm->Immediate.Size - 1; i++ ) { - CHR( '\n' ); + EOL(); switch( imm->Immediate.DataType ) { case TGSI_IMM_FLOAT32: TXT( "\nFloat: " ); @@ -900,14 +729,13 @@ dump_immediate_verbose( static void dump_instruction_short( - struct gen_dump *dump, struct tgsi_full_instruction *inst, unsigned instno ) { unsigned i; boolean first_reg = TRUE; - CHR( '\n' ); + EOL(); UID( instno ); CHR( ':' ); ENM( inst->Instruction.Opcode, TGSI_OPCODES_SHORT ); @@ -939,6 +767,31 @@ dump_instruction_short( SID( dst->DstRegister.Index ); CHR( ']' ); + switch (dst->DstRegisterExtModulate.Modulate) { + case TGSI_MODULATE_1X: + break; + case TGSI_MODULATE_2X: + TXT( "_2X" ); + break; + case TGSI_MODULATE_4X: + TXT( "_4X" ); + break; + case TGSI_MODULATE_8X: + TXT( "_8X" ); + break; + case TGSI_MODULATE_HALF: + TXT( "_D2" ); + break; + case TGSI_MODULATE_QUARTER: + TXT( "_D4" ); + break; + case TGSI_MODULATE_EIGHTH: + TXT( "_D8" ); + break; + default: + assert( 0 ); + } + if( dst->DstRegister.WriteMask != TGSI_WRITEMASK_XYZW ) { CHR( '.' ); if( dst->DstRegister.WriteMask & TGSI_WRITEMASK_X ) { @@ -966,18 +819,18 @@ dump_instruction_short( } CHR( ' ' ); - if( src->SrcRegisterExtMod.Complement ) { - TXT( "(1 - " ); - } - if( src->SrcRegisterExtMod.Negate ) { - CHR( '-' ); - } - if( src->SrcRegisterExtMod.Absolute ) { + if (src->SrcRegisterExtMod.Negate) + TXT( "-(" ); + if (src->SrcRegisterExtMod.Absolute) CHR( '|' ); - } - if( src->SrcRegister.Negate ) { + if (src->SrcRegisterExtMod.Scale2X) + TXT( "2*(" ); + if (src->SrcRegisterExtMod.Bias) + CHR( '(' ); + if (src->SrcRegisterExtMod.Complement) + TXT( "1-(" ); + if (src->SrcRegister.Negate) CHR( '-' ); - } ENM( src->SrcRegister.File, TGSI_FILES_SHORT ); @@ -985,35 +838,37 @@ dump_instruction_short( SID( src->SrcRegister.Index ); CHR( ']' ); - if (src->SrcRegister.Extended) { - if (src->SrcRegisterExtSwz.ExtSwizzleX != TGSI_EXTSWIZZLE_X || - src->SrcRegisterExtSwz.ExtSwizzleY != TGSI_EXTSWIZZLE_Y || - src->SrcRegisterExtSwz.ExtSwizzleZ != TGSI_EXTSWIZZLE_Z || - src->SrcRegisterExtSwz.ExtSwizzleW != TGSI_EXTSWIZZLE_W) { - CHR( '.' ); - ENM( src->SrcRegisterExtSwz.ExtSwizzleX, TGSI_EXTSWIZZLES_SHORT ); - ENM( src->SrcRegisterExtSwz.ExtSwizzleY, TGSI_EXTSWIZZLES_SHORT ); - ENM( src->SrcRegisterExtSwz.ExtSwizzleZ, TGSI_EXTSWIZZLES_SHORT ); - ENM( src->SrcRegisterExtSwz.ExtSwizzleW, TGSI_EXTSWIZZLES_SHORT ); - } - } - else if( src->SrcRegister.SwizzleX != TGSI_SWIZZLE_X || - src->SrcRegister.SwizzleY != TGSI_SWIZZLE_Y || - src->SrcRegister.SwizzleZ != TGSI_SWIZZLE_Z || - src->SrcRegister.SwizzleW != TGSI_SWIZZLE_W ) { + if (src->SrcRegister.SwizzleX != TGSI_SWIZZLE_X || + src->SrcRegister.SwizzleY != TGSI_SWIZZLE_Y || + src->SrcRegister.SwizzleZ != TGSI_SWIZZLE_Z || + src->SrcRegister.SwizzleW != TGSI_SWIZZLE_W) { CHR( '.' ); ENM( src->SrcRegister.SwizzleX, TGSI_SWIZZLES_SHORT ); ENM( src->SrcRegister.SwizzleY, TGSI_SWIZZLES_SHORT ); ENM( src->SrcRegister.SwizzleZ, TGSI_SWIZZLES_SHORT ); ENM( src->SrcRegister.SwizzleW, TGSI_SWIZZLES_SHORT ); } + if (src->SrcRegisterExtSwz.ExtSwizzleX != TGSI_EXTSWIZZLE_X || + src->SrcRegisterExtSwz.ExtSwizzleY != TGSI_EXTSWIZZLE_Y || + src->SrcRegisterExtSwz.ExtSwizzleZ != TGSI_EXTSWIZZLE_Z || + src->SrcRegisterExtSwz.ExtSwizzleW != TGSI_EXTSWIZZLE_W) { + CHR( '.' ); + ENM( src->SrcRegisterExtSwz.ExtSwizzleX, TGSI_EXTSWIZZLES_SHORT ); + ENM( src->SrcRegisterExtSwz.ExtSwizzleY, TGSI_EXTSWIZZLES_SHORT ); + ENM( src->SrcRegisterExtSwz.ExtSwizzleZ, TGSI_EXTSWIZZLES_SHORT ); + ENM( src->SrcRegisterExtSwz.ExtSwizzleW, TGSI_EXTSWIZZLES_SHORT ); + } - if( src->SrcRegisterExtMod.Absolute ) { + if (src->SrcRegisterExtMod.Complement) + CHR( ')' ); + if (src->SrcRegisterExtMod.Bias) + TXT( ")-.5" ); + if (src->SrcRegisterExtMod.Scale2X) + CHR( ')' ); + if (src->SrcRegisterExtMod.Absolute) CHR( '|' ); - } - if( src->SrcRegisterExtMod.Complement ) { + if (src->SrcRegisterExtMod.Negate) CHR( ')' ); - } first_reg = FALSE; } @@ -1037,7 +892,6 @@ dump_instruction_short( static void dump_instruction_verbose( - struct gen_dump *dump, struct tgsi_full_instruction *inst, unsigned ignored, unsigned deflt, @@ -1065,7 +919,7 @@ dump_instruction_verbose( } if( deflt || tgsi_compare_instruction_ext_nv( inst->InstructionExtNv, fi->InstructionExtNv ) ) { - CHR( '\n' ); + EOL(); TXT( "\nType : " ); ENM( inst->InstructionExtNv.Type, TGSI_INSTRUCTION_EXTS ); if( deflt || fi->InstructionExtNv.Precision != inst->InstructionExtNv.Precision ) { @@ -1119,7 +973,7 @@ dump_instruction_verbose( } if( deflt || tgsi_compare_instruction_ext_label( inst->InstructionExtLabel, fi->InstructionExtLabel ) ) { - CHR( '\n' ); + EOL(); TXT( "\nType : " ); ENM( inst->InstructionExtLabel.Type, TGSI_INSTRUCTION_EXTS ); if( deflt || fi->InstructionExtLabel.Label != inst->InstructionExtLabel.Label ) { @@ -1137,7 +991,7 @@ dump_instruction_verbose( } if( deflt || tgsi_compare_instruction_ext_texture( inst->InstructionExtTexture, fi->InstructionExtTexture ) ) { - CHR( '\n' ); + EOL(); TXT( "\nType : " ); ENM( inst->InstructionExtTexture.Type, TGSI_INSTRUCTION_EXTS ); if( deflt || fi->InstructionExtTexture.Texture != inst->InstructionExtTexture.Texture ) { @@ -1158,7 +1012,7 @@ dump_instruction_verbose( struct tgsi_full_dst_register *dst = &inst->FullDstRegisters[i]; struct tgsi_full_dst_register *fd = &fi->FullDstRegisters[i]; - CHR( '\n' ); + EOL(); TXT( "\nFile : " ); ENM( dst->DstRegister.File, TGSI_FILES ); if( deflt || fd->DstRegister.WriteMask != dst->DstRegister.WriteMask ) { @@ -1189,7 +1043,7 @@ dump_instruction_verbose( } if( deflt || tgsi_compare_dst_register_ext_concode( dst->DstRegisterExtConcode, fd->DstRegisterExtConcode ) ) { - CHR( '\n' ); + EOL(); TXT( "\nType : " ); ENM( dst->DstRegisterExtConcode.Type, TGSI_DST_REGISTER_EXTS ); if( deflt || fd->DstRegisterExtConcode.CondMask != dst->DstRegisterExtConcode.CondMask ) { @@ -1227,7 +1081,7 @@ dump_instruction_verbose( } if( deflt || tgsi_compare_dst_register_ext_modulate( dst->DstRegisterExtModulate, fd->DstRegisterExtModulate ) ) { - CHR( '\n' ); + EOL(); TXT( "\nType : " ); ENM( dst->DstRegisterExtModulate.Type, TGSI_DST_REGISTER_EXTS ); if( deflt || fd->DstRegisterExtModulate.Modulate != dst->DstRegisterExtModulate.Modulate ) { @@ -1249,7 +1103,7 @@ dump_instruction_verbose( struct tgsi_full_src_register *src = &inst->FullSrcRegisters[i]; struct tgsi_full_src_register *fs = &fi->FullSrcRegisters[i]; - CHR( '\n' ); + EOL(); TXT( "\nFile : "); ENM( src->SrcRegister.File, TGSI_FILES ); if( deflt || fs->SrcRegister.SwizzleX != src->SrcRegister.SwizzleX ) { @@ -1294,7 +1148,7 @@ dump_instruction_verbose( } if( deflt || tgsi_compare_src_register_ext_swz( src->SrcRegisterExtSwz, fs->SrcRegisterExtSwz ) ) { - CHR( '\n' ); + EOL(); TXT( "\nType : " ); ENM( src->SrcRegisterExtSwz.Type, TGSI_SRC_REGISTER_EXTS ); if( deflt || fs->SrcRegisterExtSwz.ExtSwizzleX != src->SrcRegisterExtSwz.ExtSwizzleX ) { @@ -1340,7 +1194,7 @@ dump_instruction_verbose( } if( deflt || tgsi_compare_src_register_ext_mod( src->SrcRegisterExtMod, fs->SrcRegisterExtMod ) ) { - CHR( '\n' ); + EOL(); TXT( "\nType : " ); ENM( src->SrcRegisterExtMod.Type, TGSI_SRC_REGISTER_EXTS ); if( deflt || fs->SrcRegisterExtMod.Complement != src->SrcRegisterExtMod.Complement ) { @@ -1375,9 +1229,8 @@ dump_instruction_verbose( } } -static void -dump_gen( - struct gen_dump *dump, +void +tgsi_dump( const struct tgsi_token *tokens, unsigned flags ) { @@ -1389,18 +1242,17 @@ dump_gen( unsigned deflt = !(flags & TGSI_DUMP_NO_DEFAULT); unsigned instno = 0; - dump->tabs = 0; - - /* sanity check */ + /* sanity checks */ assert(strcmp(TGSI_OPCODES[TGSI_OPCODE_CONT], "OPCODE_CONT") == 0); + assert(strcmp(TGSI_OPCODES[TGSI_OPCODE_END], "OPCODE_END") == 0); + assert(strcmp(TGSI_OPCODES_SHORT[TGSI_OPCODE_END], "END") == 0); tgsi_parse_init( &parse, tokens ); TXT( "tgsi-dump begin -----------------" ); - CHR( '\n' ); + EOL(); ENM( parse.FullHeader.Processor.Processor, TGSI_PROCESSOR_TYPES_SHORT ); - CHR( ' ' ); UID( parse.FullVersion.Version.MajorVersion ); CHR( '.' ); UID( parse.FullVersion.Version.MinorVersion ); @@ -1410,7 +1262,7 @@ dump_gen( UID( parse.FullVersion.Version.MajorVersion ); TXT( "\nMinorVersion: " ); UID( parse.FullVersion.Version.MinorVersion ); - CHR( '\n' ); + EOL(); TXT( "\nHeaderSize: " ); UID( parse.FullHeader.Header.HeaderSize ); @@ -1418,7 +1270,7 @@ dump_gen( UID( parse.FullHeader.Header.BodySize ); TXT( "\nProcessor : " ); ENM( parse.FullHeader.Processor.Processor, TGSI_PROCESSOR_TYPES ); - CHR( '\n' ); + EOL(); } fi = tgsi_default_full_instruction(); @@ -1430,19 +1282,16 @@ dump_gen( switch( parse.FullToken.Token.Type ) { case TGSI_TOKEN_TYPE_DECLARATION: dump_declaration_short( - dump, &parse.FullToken.FullDeclaration ); break; case TGSI_TOKEN_TYPE_IMMEDIATE: dump_immediate_short( - dump, &parse.FullToken.FullImmediate ); break; case TGSI_TOKEN_TYPE_INSTRUCTION: dump_instruction_short( - dump, &parse.FullToken.FullInstruction, instno ); instno++; @@ -1467,7 +1316,6 @@ dump_gen( switch( parse.FullToken.Token.Type ) { case TGSI_TOKEN_TYPE_DECLARATION: dump_declaration_verbose( - dump, &parse.FullToken.FullDeclaration, ignored, deflt, @@ -1476,14 +1324,12 @@ dump_gen( case TGSI_TOKEN_TYPE_IMMEDIATE: dump_immediate_verbose( - dump, &parse.FullToken.FullImmediate, ignored ); break; case TGSI_TOKEN_TYPE_INSTRUCTION: dump_instruction_verbose( - dump, &parse.FullToken.FullInstruction, ignored, deflt, @@ -1494,7 +1340,7 @@ dump_gen( assert( 0 ); } - CHR( '\n' ); + EOL(); } } @@ -1502,86 +1348,3 @@ dump_gen( tgsi_parse_free( &parse ); } - - -static void -sanity_checks(void) -{ - assert(strcmp(TGSI_OPCODES[TGSI_OPCODE_END], "OPCODE_END") == 0); - assert(strcmp(TGSI_OPCODES_SHORT[TGSI_OPCODE_END], "END") == 0); -} - - -void -tgsi_dump( - const struct tgsi_token *tokens, - unsigned flags ) -{ - struct file_dump dump; - - sanity_checks(); - - dump.base.write = _file_dump_write; -#if 0 - { - static unsigned counter = 0; - char buffer[64]; - sprintf( buffer, "tgsi-dump-%.4u.txt", counter++ ); - dump.file = fopen( buffer, "wt" ); - } -#else - dump.file = stderr; -#endif - - dump_gen( - &dump.base, - tokens, - flags ); - -#if 0 - fclose( dump.file ); -#endif -} - -void -tgsi_dump_str( - char **str, - const struct tgsi_token *tokens, - unsigned flags ) -{ - struct text_dump dump; - - dump.base.write = _text_dump_write; - dump.text = NULL; - dump.length = 0; - dump.capacity = 0; - - dump_gen( - &dump.base, - tokens, - flags ); - - *str = dump.text; -} - - -void tgsi_debug_dump( struct tgsi_token *tokens ) -{ - char *str, *p; - - tgsi_dump_str( &str, tokens, 0 ); - - p = str; - while (p != NULL) - { - char *end = strchr( p, '\n' ); - if (end != NULL) - { - *end++ = '\0'; - } - debug_printf( "%s\n", p ); - p = end; - } - - FREE( str ); -} diff --git a/src/gallium/auxiliary/tgsi/util/tgsi_dump.h b/src/gallium/auxiliary/tgsi/util/tgsi_dump.h index 51d79a0362..beb0155d56 100644 --- a/src/gallium/auxiliary/tgsi/util/tgsi_dump.h +++ b/src/gallium/auxiliary/tgsi/util/tgsi_dump.h @@ -14,16 +14,6 @@ tgsi_dump( const struct tgsi_token *tokens, unsigned flags ); -void -tgsi_dump_str( - char **str, - const struct tgsi_token *tokens, - unsigned flags ); - -/* Dump to debug_printf() - */ -void tgsi_debug_dump( struct tgsi_token *tokens ); - #if defined __cplusplus } #endif diff --git a/src/gallium/auxiliary/tgsi/util/tgsi_parse.c b/src/gallium/auxiliary/tgsi/util/tgsi_parse.c index c3526cb71f..5c0b0bfd61 100644 --- a/src/gallium/auxiliary/tgsi/util/tgsi_parse.c +++ b/src/gallium/auxiliary/tgsi/util/tgsi_parse.c @@ -43,7 +43,7 @@ tgsi_full_token_free( union tgsi_full_token *full_token ) { if( full_token->Token.Type == TGSI_TOKEN_TYPE_IMMEDIATE ) { - FREE( full_token->FullImmediate.u.Pointer ); + FREE( (void *) full_token->FullImmediate.u.Pointer ); } } @@ -156,7 +156,7 @@ tgsi_parse_token( imm->u.Pointer = MALLOC( sizeof( struct tgsi_immediate_float32 ) * (imm->Immediate.Size - 1) ); for( i = 0; i < imm->Immediate.Size - 1; i++ ) { - next_token( ctx, &imm->u.ImmediateFloat32[i] ); + next_token( ctx, (struct tgsi_immediate_float32 *) &imm->u.ImmediateFloat32[i] ); } break; @@ -330,3 +330,18 @@ tgsi_num_tokens(const struct tgsi_token *tokens) } return 0; } + + +/** + * Make a new copy of a token array. + */ +struct tgsi_token * +tgsi_dup_tokens(const struct tgsi_token *tokens) +{ + unsigned n = tgsi_num_tokens(tokens); + unsigned bytes = n * sizeof(struct tgsi_token); + struct tgsi_token *new_tokens = (struct tgsi_token *) MALLOC(bytes); + if (new_tokens) + memcpy(new_tokens, tokens, bytes); + return new_tokens; +} diff --git a/src/gallium/auxiliary/tgsi/util/tgsi_parse.h b/src/gallium/auxiliary/tgsi/util/tgsi_parse.h index a98e88e343..4102101093 100644 --- a/src/gallium/auxiliary/tgsi/util/tgsi_parse.h +++ b/src/gallium/auxiliary/tgsi/util/tgsi_parse.h @@ -1,6 +1,35 @@ -#if !defined TGSI_PARSE_H +/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#ifndef TGSI_PARSE_H #define TGSI_PARSE_H +#include "pipe/p_shader_tokens.h" + #if defined __cplusplus extern "C" { #endif @@ -50,8 +79,8 @@ struct tgsi_full_immediate struct tgsi_immediate Immediate; union { - void *Pointer; - struct tgsi_immediate_float32 *ImmediateFloat32; + const void *Pointer; + const struct tgsi_immediate_float32 *ImmediateFloat32; } u; }; @@ -116,6 +145,8 @@ tgsi_parse_token( unsigned tgsi_num_tokens(const struct tgsi_token *tokens); +struct tgsi_token * +tgsi_dup_tokens(const struct tgsi_token *tokens); #if defined __cplusplus } diff --git a/src/gallium/auxiliary/tgsi/util/tgsi_scan.c b/src/gallium/auxiliary/tgsi/util/tgsi_scan.c index ea4a72967d..65650ed22a 100644 --- a/src/gallium/auxiliary/tgsi/util/tgsi_scan.c +++ b/src/gallium/auxiliary/tgsi/util/tgsi_scan.c @@ -103,18 +103,14 @@ tgsi_scan_shader(const struct tgsi_token *tokens, info->file_max[file] = MAX2(info->file_max[file], (int)i); if (file == TGSI_FILE_INPUT) { - info->input_semantic_name[info->num_inputs] - = (ubyte)fulldecl->Semantic.SemanticName; - info->input_semantic_index[info->num_inputs] - = (ubyte)fulldecl->Semantic.SemanticIndex; + info->input_semantic_name[i] = (ubyte)fulldecl->Semantic.SemanticName; + info->input_semantic_index[i] = (ubyte)fulldecl->Semantic.SemanticIndex; info->num_inputs++; } if (file == TGSI_FILE_OUTPUT) { - info->output_semantic_name[info->num_outputs] - = (ubyte)fulldecl->Semantic.SemanticName; - info->output_semantic_index[info->num_outputs] - = (ubyte)fulldecl->Semantic.SemanticIndex; + info->output_semantic_name[i] = (ubyte)fulldecl->Semantic.SemanticName; + info->output_semantic_index[i] = (ubyte)fulldecl->Semantic.SemanticIndex; info->num_outputs++; } @@ -137,6 +133,9 @@ tgsi_scan_shader(const struct tgsi_token *tokens, } } + assert( info->file_max[TGSI_FILE_INPUT] + 1 == info->num_inputs ); + assert( info->file_max[TGSI_FILE_OUTPUT] + 1 == info->num_outputs ); + info->uses_kill = (info->opcode_count[TGSI_OPCODE_KIL] || info->opcode_count[TGSI_OPCODE_KILP]); diff --git a/src/gallium/auxiliary/translate/Makefile b/src/gallium/auxiliary/translate/Makefile new file mode 100644 index 0000000000..ad2a5b705e --- /dev/null +++ b/src/gallium/auxiliary/translate/Makefile @@ -0,0 +1,15 @@ +TOP = ../../../.. +include $(TOP)/configs/current + +LIBNAME = translate + +C_SOURCES = \ + translate_generic.c \ + translate_sse.c \ + translate.c \ + translate_cache.c + +include ../../Makefile.template + +symlinks: + diff --git a/src/gallium/auxiliary/translate/SConscript b/src/gallium/auxiliary/translate/SConscript new file mode 100644 index 0000000000..9553a67537 --- /dev/null +++ b/src/gallium/auxiliary/translate/SConscript @@ -0,0 +1,12 @@ +Import('*') + +translate = env.ConvenienceLibrary( + target = 'translate', + source = [ + 'translate_generic.c', + 'translate_sse.c', + 'translate.c', + 'translate_cache.c', + ]) + +auxiliaries.insert(0, translate) diff --git a/src/mesa/state_tracker/st_texobj.h b/src/gallium/auxiliary/translate/translate.c index 3c66031018..b04bc6eefd 100644 --- a/src/mesa/state_tracker/st_texobj.h +++ b/src/gallium/auxiliary/translate/translate.c @@ -25,17 +25,24 @@ * **************************************************************************/ -/* - * Authors: - * Brian Paul - */ + /* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ -#ifndef ST_TEXOBJ_H -#define ST_TEXOBJ_H 1 +#include "pipe/p_util.h" +#include "pipe/p_state.h" +#include "translate.h" +struct translate *translate_create( const struct translate_key *key ) +{ + struct translate *translate = NULL; -extern struct pipe_texture_object * -create_texture_object(struct gl_texture_object *texObj); +#if defined(__i386__) || defined(__386__) || defined(i386) + translate = translate_sse2_create( key ); + if (translate) + return translate; +#endif - -#endif /* ST_TEXOBJ_H */ + return translate_generic_create( key ); +} diff --git a/src/gallium/auxiliary/translate/translate.h b/src/gallium/auxiliary/translate/translate.h new file mode 100644 index 0000000000..b8210af50c --- /dev/null +++ b/src/gallium/auxiliary/translate/translate.h @@ -0,0 +1,127 @@ +/* + * Copyright 2008 Tungsten Graphics, inc. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * on the rights to use, copy, modify, merge, publish, distribute, sub + * license, and/or sell copies of the Software, and to permit persons to whom + * the Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * TUNGSTEN GRAPHICS AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + */ + + +/** + * Vertex fetch/store/convert code. This functionality is used in two places: + * 1. Vertex fetch/convert - to grab vertex data from incoming vertex + * arrays and convert to format needed by vertex shaders. + * 2. Vertex store/emit - to convert simple float[][4] vertex attributes + * (which is the organization used throughout the draw/prim pipeline) to + * hardware-specific formats and emit into hardware vertex buffers. + * + * + * Authors: + * Keith Whitwell <keithw@tungstengraphics.com> + */ + +#ifndef _TRANSLATE_H +#define _TRANSLATE_H + + +#include "pipe/p_compiler.h" +#include "pipe/p_format.h" +#include "pipe/p_state.h" + +struct translate_element +{ + enum pipe_format input_format; + enum pipe_format output_format; + unsigned input_buffer; + unsigned input_offset; /* can't really reduce the size of these */ + unsigned output_offset; +}; + + +struct translate_key { + unsigned output_stride; + unsigned nr_elements; + struct translate_element element[PIPE_MAX_ATTRIBS]; +}; + + +struct translate { + struct translate_key key; + + void (*release)( struct translate * ); + + void (*set_buffer)( struct translate *, + unsigned i, + const void *ptr, + unsigned stride ); + + void (*run_elts)( struct translate *, + const unsigned *elts, + unsigned count, + void *output_buffer); + + void (*run)( struct translate *, + unsigned start, + unsigned count, + void *output_buffer); +}; + + + +#if 0 +struct translate_context *translate_context_create( void ); +void translate_context_destroy( struct translate_context * ); + +struct translate *translate_lookup_or_create( struct translate_context *tctx, + const struct translate_key *key ); +#endif + + +struct translate *translate_create( const struct translate_key *key ); + +static INLINE int translate_keysize( const struct translate_key *key ) +{ + return 2 * sizeof(int) + key->nr_elements * sizeof(struct translate_element); +} + +static INLINE int translate_key_compare( const struct translate_key *a, + const struct translate_key *b ) +{ + int keysize = translate_keysize(a); + return memcmp(a, b, keysize); +} + + +static INLINE void translate_key_sanitize( struct translate_key *a ) +{ + int keysize = translate_keysize(a); + char *ptr = (char *)a; + memset(ptr + keysize, 0, sizeof(*a) - keysize); +} + + +/******************************************************************************* + * Private: + */ +struct translate *translate_sse2_create( const struct translate_key *key ); + +struct translate *translate_generic_create( const struct translate_key *key ); + + +#endif diff --git a/src/gallium/auxiliary/translate/translate_cache.c b/src/gallium/auxiliary/translate/translate_cache.c new file mode 100644 index 0000000000..115dc9287e --- /dev/null +++ b/src/gallium/auxiliary/translate/translate_cache.c @@ -0,0 +1,102 @@ +/************************************************************************** + * + * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include "pipe/p_util.h" +#include "pipe/p_state.h" +#include "translate.h" +#include "translate_cache.h" + +#include "cso_cache/cso_cache.h" +#include "cso_cache/cso_hash.h" + +struct translate_cache { + struct cso_hash *hash; +}; + +struct translate_cache * translate_cache_create( void ) +{ + struct translate_cache *cache = MALLOC_STRUCT(translate_cache); + cache->hash = cso_hash_create(); + return cache; +} + + +static INLINE void delete_translates(struct translate_cache *cache) +{ + struct cso_hash *hash = cache->hash; + struct cso_hash_iter iter = cso_hash_first_node(hash); + while (!cso_hash_iter_is_null(iter)) { + struct translate *state = (struct translate*)cso_hash_iter_data(iter); + iter = cso_hash_iter_next(iter); + if (state) { + state->release(state); + } + } +} + +void translate_cache_destroy(struct translate_cache *cache) +{ + delete_translates(cache); + cso_hash_delete(cache->hash); + FREE(cache); +} + + +static INLINE unsigned translate_hash_key_size(struct translate_key *key) +{ + unsigned size = sizeof(struct translate_key) - + sizeof(struct translate_element) * (PIPE_MAX_ATTRIBS - key->nr_elements); + return size; +} + +static INLINE unsigned create_key(struct translate_key *key) +{ + unsigned hash_key; + unsigned size = translate_hash_key_size(key); + /*debug_printf("key size = %d, (els = %d)\n", + size, key->nr_elements);*/ + hash_key = cso_construct_key(key, size); + return hash_key; +} + +struct translate * translate_cache_find(struct translate_cache *cache, + struct translate_key *key) +{ + unsigned hash_key = create_key(key); + struct translate *translate = (struct translate*) + cso_hash_find_data_from_template(cache->hash, + hash_key, + key, sizeof(*key)); + + if (!translate) { + /* create/insert */ + translate = translate_create(key); + cso_hash_insert(cache->hash, hash_key, translate); + } + + return translate; +} diff --git a/src/gallium/auxiliary/translate/translate_cache.h b/src/gallium/auxiliary/translate/translate_cache.h new file mode 100644 index 0000000000..7dba871e57 --- /dev/null +++ b/src/gallium/auxiliary/translate/translate_cache.h @@ -0,0 +1,54 @@ +/* + * Copyright 2008 Tungsten Graphics, inc. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * on the rights to use, copy, modify, merge, publish, distribute, sub + * license, and/or sell copies of the Software, and to permit persons to whom + * the Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * TUNGSTEN GRAPHICS AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + */ +#ifndef _TRANSLATE_CACHE_H +#define _TRANSLATE_CACHE_H + + +/******************************************************************************* + * Translate cache. + * Simply used to cache created translates. Avoids unecessary creation of + * translate's if one suitable for a given translate_key has already been + * created. + * + * Note: this functionality depends and requires the CSO module. + */ +struct translate_cache; + +struct translate_key; +struct translate; + +struct translate_cache *translate_cache_create( void ); +void translate_cache_destroy(struct translate_cache *cache); + +/** + * Will try to find a translate structure matched by the given key. + * If such a structure doesn't exist in the cache the function + * will automatically create it, insert it in the cache and + * return the created version. + * + */ +struct translate *translate_cache_find(struct translate_cache *cache, + struct translate_key *key); + +#endif diff --git a/src/gallium/auxiliary/translate/translate_generic.c b/src/gallium/auxiliary/translate/translate_generic.c new file mode 100644 index 0000000000..402780ee53 --- /dev/null +++ b/src/gallium/auxiliary/translate/translate_generic.c @@ -0,0 +1,676 @@ +/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + + /* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#include "pipe/p_util.h" +#include "pipe/p_state.h" +#include "translate.h" + + +#define DRAW_DBG 0 + +typedef void (*fetch_func)(const void *ptr, float *attrib); +typedef void (*emit_func)(const float *attrib, void *ptr); + + + +struct translate_generic { + struct translate translate; + + struct { + fetch_func fetch; + unsigned buffer; + unsigned input_offset; + + emit_func emit; + unsigned output_offset; + + char *input_ptr; + unsigned input_stride; + + } attrib[PIPE_MAX_ATTRIBS]; + + unsigned nr_attrib; +}; + + +static struct translate_generic *translate_generic( struct translate *translate ) +{ + return (struct translate_generic *)translate; +} + +/** + * Fetch a float[4] vertex attribute from memory, doing format/type + * conversion as needed. + * + * This is probably needed/dupliocated elsewhere, eg format + * conversion, texture sampling etc. + */ +#define ATTRIB( NAME, SZ, TYPE, FROM, TO ) \ +static void \ +fetch_##NAME(const void *ptr, float *attrib) \ +{ \ + const float defaults[4] = { 0.0f,0.0f,0.0f,1.0f }; \ + unsigned i; \ + \ + for (i = 0; i < SZ; i++) { \ + attrib[i] = FROM(i); \ + } \ + \ + for (; i < 4; i++) { \ + attrib[i] = defaults[i]; \ + } \ +} \ + \ +static void \ +emit_##NAME(const float *attrib, void *ptr) \ +{ \ + unsigned i; \ + TYPE *out = (TYPE *)ptr; \ + \ + for (i = 0; i < SZ; i++) { \ + out[i] = TO(attrib[i]); \ + } \ +} + + +#define FROM_64_FLOAT(i) ((float) ((double *) ptr)[i]) +#define FROM_32_FLOAT(i) (((float *) ptr)[i]) + +#define FROM_8_USCALED(i) ((float) ((unsigned char *) ptr)[i]) +#define FROM_16_USCALED(i) ((float) ((unsigned short *) ptr)[i]) +#define FROM_32_USCALED(i) ((float) ((unsigned int *) ptr)[i]) + +#define FROM_8_SSCALED(i) ((float) ((char *) ptr)[i]) +#define FROM_16_SSCALED(i) ((float) ((short *) ptr)[i]) +#define FROM_32_SSCALED(i) ((float) ((int *) ptr)[i]) + +#define FROM_8_UNORM(i) ((float) ((unsigned char *) ptr)[i] / 255.0f) +#define FROM_16_UNORM(i) ((float) ((unsigned short *) ptr)[i] / 65535.0f) +#define FROM_32_UNORM(i) ((float) ((unsigned int *) ptr)[i] / 4294967295.0f) + +#define FROM_8_SNORM(i) ((float) ((char *) ptr)[i] / 127.0f) +#define FROM_16_SNORM(i) ((float) ((short *) ptr)[i] / 32767.0f) +#define FROM_32_SNORM(i) ((float) ((int *) ptr)[i] / 2147483647.0f) + +#define TO_64_FLOAT(x) ((double) x) +#define TO_32_FLOAT(x) (x) + +#define TO_8_USCALED(x) ((unsigned char) x) +#define TO_16_USCALED(x) ((unsigned short) x) +#define TO_32_USCALED(x) ((unsigned int) x) + +#define TO_8_SSCALED(x) ((char) x) +#define TO_16_SSCALED(x) ((short) x) +#define TO_32_SSCALED(x) ((int) x) + +#define TO_8_UNORM(x) ((unsigned char) (x * 255.0f)) +#define TO_16_UNORM(x) ((unsigned short) (x * 65535.0f)) +#define TO_32_UNORM(x) ((unsigned int) (x * 4294967295.0f)) + +#define TO_8_SNORM(x) ((char) (x * 127.0f)) +#define TO_16_SNORM(x) ((short) (x * 32767.0f)) +#define TO_32_SNORM(x) ((int) (x * 2147483647.0f)) + + + +ATTRIB( R64G64B64A64_FLOAT, 4, double, FROM_64_FLOAT, TO_64_FLOAT ) +ATTRIB( R64G64B64_FLOAT, 3, double, FROM_64_FLOAT, TO_64_FLOAT ) +ATTRIB( R64G64_FLOAT, 2, double, FROM_64_FLOAT, TO_64_FLOAT ) +ATTRIB( R64_FLOAT, 1, double, FROM_64_FLOAT, TO_64_FLOAT ) + +ATTRIB( R32G32B32A32_FLOAT, 4, float, FROM_32_FLOAT, TO_32_FLOAT ) +ATTRIB( R32G32B32_FLOAT, 3, float, FROM_32_FLOAT, TO_32_FLOAT ) +ATTRIB( R32G32_FLOAT, 2, float, FROM_32_FLOAT, TO_32_FLOAT ) +ATTRIB( R32_FLOAT, 1, float, FROM_32_FLOAT, TO_32_FLOAT ) + +ATTRIB( R32G32B32A32_USCALED, 4, unsigned, FROM_32_USCALED, TO_32_USCALED ) +ATTRIB( R32G32B32_USCALED, 3, unsigned, FROM_32_USCALED, TO_32_USCALED ) +ATTRIB( R32G32_USCALED, 2, unsigned, FROM_32_USCALED, TO_32_USCALED ) +ATTRIB( R32_USCALED, 1, unsigned, FROM_32_USCALED, TO_32_USCALED ) + +ATTRIB( R32G32B32A32_SSCALED, 4, int, FROM_32_SSCALED, TO_32_SSCALED ) +ATTRIB( R32G32B32_SSCALED, 3, int, FROM_32_SSCALED, TO_32_SSCALED ) +ATTRIB( R32G32_SSCALED, 2, int, FROM_32_SSCALED, TO_32_SSCALED ) +ATTRIB( R32_SSCALED, 1, int, FROM_32_SSCALED, TO_32_SSCALED ) + +ATTRIB( R32G32B32A32_UNORM, 4, unsigned, FROM_32_UNORM, TO_32_UNORM ) +ATTRIB( R32G32B32_UNORM, 3, unsigned, FROM_32_UNORM, TO_32_UNORM ) +ATTRIB( R32G32_UNORM, 2, unsigned, FROM_32_UNORM, TO_32_UNORM ) +ATTRIB( R32_UNORM, 1, unsigned, FROM_32_UNORM, TO_32_UNORM ) + +ATTRIB( R32G32B32A32_SNORM, 4, int, FROM_32_SNORM, TO_32_SNORM ) +ATTRIB( R32G32B32_SNORM, 3, int, FROM_32_SNORM, TO_32_SNORM ) +ATTRIB( R32G32_SNORM, 2, int, FROM_32_SNORM, TO_32_SNORM ) +ATTRIB( R32_SNORM, 1, int, FROM_32_SNORM, TO_32_SNORM ) + +ATTRIB( R16G16B16A16_USCALED, 4, ushort, FROM_16_USCALED, TO_16_USCALED ) +ATTRIB( R16G16B16_USCALED, 3, ushort, FROM_16_USCALED, TO_16_USCALED ) +ATTRIB( R16G16_USCALED, 2, ushort, FROM_16_USCALED, TO_16_USCALED ) +ATTRIB( R16_USCALED, 1, ushort, FROM_16_USCALED, TO_16_USCALED ) + +ATTRIB( R16G16B16A16_SSCALED, 4, short, FROM_16_SSCALED, TO_16_SSCALED ) +ATTRIB( R16G16B16_SSCALED, 3, short, FROM_16_SSCALED, TO_16_SSCALED ) +ATTRIB( R16G16_SSCALED, 2, short, FROM_16_SSCALED, TO_16_SSCALED ) +ATTRIB( R16_SSCALED, 1, short, FROM_16_SSCALED, TO_16_SSCALED ) + +ATTRIB( R16G16B16A16_UNORM, 4, ushort, FROM_16_UNORM, TO_16_UNORM ) +ATTRIB( R16G16B16_UNORM, 3, ushort, FROM_16_UNORM, TO_16_UNORM ) +ATTRIB( R16G16_UNORM, 2, ushort, FROM_16_UNORM, TO_16_UNORM ) +ATTRIB( R16_UNORM, 1, ushort, FROM_16_UNORM, TO_16_UNORM ) + +ATTRIB( R16G16B16A16_SNORM, 4, short, FROM_16_SNORM, TO_16_SNORM ) +ATTRIB( R16G16B16_SNORM, 3, short, FROM_16_SNORM, TO_16_SNORM ) +ATTRIB( R16G16_SNORM, 2, short, FROM_16_SNORM, TO_16_SNORM ) +ATTRIB( R16_SNORM, 1, short, FROM_16_SNORM, TO_16_SNORM ) + +ATTRIB( R8G8B8A8_USCALED, 4, ubyte, FROM_8_USCALED, TO_8_USCALED ) +ATTRIB( R8G8B8_USCALED, 3, ubyte, FROM_8_USCALED, TO_8_USCALED ) +ATTRIB( R8G8_USCALED, 2, ubyte, FROM_8_USCALED, TO_8_USCALED ) +ATTRIB( R8_USCALED, 1, ubyte, FROM_8_USCALED, TO_8_USCALED ) + +ATTRIB( R8G8B8A8_SSCALED, 4, char, FROM_8_SSCALED, TO_8_SSCALED ) +ATTRIB( R8G8B8_SSCALED, 3, char, FROM_8_SSCALED, TO_8_SSCALED ) +ATTRIB( R8G8_SSCALED, 2, char, FROM_8_SSCALED, TO_8_SSCALED ) +ATTRIB( R8_SSCALED, 1, char, FROM_8_SSCALED, TO_8_SSCALED ) + +ATTRIB( R8G8B8A8_UNORM, 4, ubyte, FROM_8_UNORM, TO_8_UNORM ) +ATTRIB( R8G8B8_UNORM, 3, ubyte, FROM_8_UNORM, TO_8_UNORM ) +ATTRIB( R8G8_UNORM, 2, ubyte, FROM_8_UNORM, TO_8_UNORM ) +ATTRIB( R8_UNORM, 1, ubyte, FROM_8_UNORM, TO_8_UNORM ) + +ATTRIB( R8G8B8A8_SNORM, 4, char, FROM_8_SNORM, TO_8_SNORM ) +ATTRIB( R8G8B8_SNORM, 3, char, FROM_8_SNORM, TO_8_SNORM ) +ATTRIB( R8G8_SNORM, 2, char, FROM_8_SNORM, TO_8_SNORM ) +ATTRIB( R8_SNORM, 1, char, FROM_8_SNORM, TO_8_SNORM ) + +ATTRIB( A8R8G8B8_UNORM, 4, ubyte, FROM_8_UNORM, TO_8_UNORM ) +//ATTRIB( R8G8B8A8_UNORM, 4, ubyte, FROM_8_UNORM, TO_8_UNORM ) + + + +static void +fetch_B8G8R8A8_UNORM(const void *ptr, float *attrib) +{ + attrib[2] = FROM_8_UNORM(0); + attrib[1] = FROM_8_UNORM(1); + attrib[0] = FROM_8_UNORM(2); + attrib[3] = FROM_8_UNORM(3); +} + +static void +emit_B8G8R8A8_UNORM( const float *attrib, void *ptr) +{ + ubyte *out = (ubyte *)ptr; + out[2] = TO_8_UNORM(attrib[0]); + out[1] = TO_8_UNORM(attrib[1]); + out[0] = TO_8_UNORM(attrib[2]); + out[3] = TO_8_UNORM(attrib[3]); +} + +static void +fetch_NULL( const void *ptr, float *attrib ) +{ + attrib[0] = 0; + attrib[1] = 0; + attrib[2] = 0; + attrib[3] = 1; +} + +static void +emit_NULL( const float *attrib, void *ptr ) +{ + /* do nothing is the only sensible option */ +} + +static fetch_func get_fetch_func( enum pipe_format format ) +{ + switch (format) { + case PIPE_FORMAT_R64_FLOAT: + return fetch_R64_FLOAT; + case PIPE_FORMAT_R64G64_FLOAT: + return fetch_R64G64_FLOAT; + case PIPE_FORMAT_R64G64B64_FLOAT: + return fetch_R64G64B64_FLOAT; + case PIPE_FORMAT_R64G64B64A64_FLOAT: + return fetch_R64G64B64A64_FLOAT; + + case PIPE_FORMAT_R32_FLOAT: + return fetch_R32_FLOAT; + case PIPE_FORMAT_R32G32_FLOAT: + return fetch_R32G32_FLOAT; + case PIPE_FORMAT_R32G32B32_FLOAT: + return fetch_R32G32B32_FLOAT; + case PIPE_FORMAT_R32G32B32A32_FLOAT: + return fetch_R32G32B32A32_FLOAT; + + case PIPE_FORMAT_R32_UNORM: + return fetch_R32_UNORM; + case PIPE_FORMAT_R32G32_UNORM: + return fetch_R32G32_UNORM; + case PIPE_FORMAT_R32G32B32_UNORM: + return fetch_R32G32B32_UNORM; + case PIPE_FORMAT_R32G32B32A32_UNORM: + return fetch_R32G32B32A32_UNORM; + + case PIPE_FORMAT_R32_USCALED: + return fetch_R32_USCALED; + case PIPE_FORMAT_R32G32_USCALED: + return fetch_R32G32_USCALED; + case PIPE_FORMAT_R32G32B32_USCALED: + return fetch_R32G32B32_USCALED; + case PIPE_FORMAT_R32G32B32A32_USCALED: + return fetch_R32G32B32A32_USCALED; + + case PIPE_FORMAT_R32_SNORM: + return fetch_R32_SNORM; + case PIPE_FORMAT_R32G32_SNORM: + return fetch_R32G32_SNORM; + case PIPE_FORMAT_R32G32B32_SNORM: + return fetch_R32G32B32_SNORM; + case PIPE_FORMAT_R32G32B32A32_SNORM: + return fetch_R32G32B32A32_SNORM; + + case PIPE_FORMAT_R32_SSCALED: + return fetch_R32_SSCALED; + case PIPE_FORMAT_R32G32_SSCALED: + return fetch_R32G32_SSCALED; + case PIPE_FORMAT_R32G32B32_SSCALED: + return fetch_R32G32B32_SSCALED; + case PIPE_FORMAT_R32G32B32A32_SSCALED: + return fetch_R32G32B32A32_SSCALED; + + case PIPE_FORMAT_R16_UNORM: + return fetch_R16_UNORM; + case PIPE_FORMAT_R16G16_UNORM: + return fetch_R16G16_UNORM; + case PIPE_FORMAT_R16G16B16_UNORM: + return fetch_R16G16B16_UNORM; + case PIPE_FORMAT_R16G16B16A16_UNORM: + return fetch_R16G16B16A16_UNORM; + + case PIPE_FORMAT_R16_USCALED: + return fetch_R16_USCALED; + case PIPE_FORMAT_R16G16_USCALED: + return fetch_R16G16_USCALED; + case PIPE_FORMAT_R16G16B16_USCALED: + return fetch_R16G16B16_USCALED; + case PIPE_FORMAT_R16G16B16A16_USCALED: + return fetch_R16G16B16A16_USCALED; + + case PIPE_FORMAT_R16_SNORM: + return fetch_R16_SNORM; + case PIPE_FORMAT_R16G16_SNORM: + return fetch_R16G16_SNORM; + case PIPE_FORMAT_R16G16B16_SNORM: + return fetch_R16G16B16_SNORM; + case PIPE_FORMAT_R16G16B16A16_SNORM: + return fetch_R16G16B16A16_SNORM; + + case PIPE_FORMAT_R16_SSCALED: + return fetch_R16_SSCALED; + case PIPE_FORMAT_R16G16_SSCALED: + return fetch_R16G16_SSCALED; + case PIPE_FORMAT_R16G16B16_SSCALED: + return fetch_R16G16B16_SSCALED; + case PIPE_FORMAT_R16G16B16A16_SSCALED: + return fetch_R16G16B16A16_SSCALED; + + case PIPE_FORMAT_R8_UNORM: + return fetch_R8_UNORM; + case PIPE_FORMAT_R8G8_UNORM: + return fetch_R8G8_UNORM; + case PIPE_FORMAT_R8G8B8_UNORM: + return fetch_R8G8B8_UNORM; + case PIPE_FORMAT_R8G8B8A8_UNORM: + return fetch_R8G8B8A8_UNORM; + + case PIPE_FORMAT_R8_USCALED: + return fetch_R8_USCALED; + case PIPE_FORMAT_R8G8_USCALED: + return fetch_R8G8_USCALED; + case PIPE_FORMAT_R8G8B8_USCALED: + return fetch_R8G8B8_USCALED; + case PIPE_FORMAT_R8G8B8A8_USCALED: + return fetch_R8G8B8A8_USCALED; + + case PIPE_FORMAT_R8_SNORM: + return fetch_R8_SNORM; + case PIPE_FORMAT_R8G8_SNORM: + return fetch_R8G8_SNORM; + case PIPE_FORMAT_R8G8B8_SNORM: + return fetch_R8G8B8_SNORM; + case PIPE_FORMAT_R8G8B8A8_SNORM: + return fetch_R8G8B8A8_SNORM; + + case PIPE_FORMAT_R8_SSCALED: + return fetch_R8_SSCALED; + case PIPE_FORMAT_R8G8_SSCALED: + return fetch_R8G8_SSCALED; + case PIPE_FORMAT_R8G8B8_SSCALED: + return fetch_R8G8B8_SSCALED; + case PIPE_FORMAT_R8G8B8A8_SSCALED: + return fetch_R8G8B8A8_SSCALED; + + case PIPE_FORMAT_A8R8G8B8_UNORM: + return fetch_A8R8G8B8_UNORM; + + case PIPE_FORMAT_B8G8R8A8_UNORM: + return fetch_B8G8R8A8_UNORM; + + default: + assert(0); + return fetch_NULL; + } +} + + + + +static emit_func get_emit_func( enum pipe_format format ) +{ + switch (format) { + case PIPE_FORMAT_R64_FLOAT: + return emit_R64_FLOAT; + case PIPE_FORMAT_R64G64_FLOAT: + return emit_R64G64_FLOAT; + case PIPE_FORMAT_R64G64B64_FLOAT: + return emit_R64G64B64_FLOAT; + case PIPE_FORMAT_R64G64B64A64_FLOAT: + return emit_R64G64B64A64_FLOAT; + + case PIPE_FORMAT_R32_FLOAT: + return emit_R32_FLOAT; + case PIPE_FORMAT_R32G32_FLOAT: + return emit_R32G32_FLOAT; + case PIPE_FORMAT_R32G32B32_FLOAT: + return emit_R32G32B32_FLOAT; + case PIPE_FORMAT_R32G32B32A32_FLOAT: + return emit_R32G32B32A32_FLOAT; + + case PIPE_FORMAT_R32_UNORM: + return emit_R32_UNORM; + case PIPE_FORMAT_R32G32_UNORM: + return emit_R32G32_UNORM; + case PIPE_FORMAT_R32G32B32_UNORM: + return emit_R32G32B32_UNORM; + case PIPE_FORMAT_R32G32B32A32_UNORM: + return emit_R32G32B32A32_UNORM; + + case PIPE_FORMAT_R32_USCALED: + return emit_R32_USCALED; + case PIPE_FORMAT_R32G32_USCALED: + return emit_R32G32_USCALED; + case PIPE_FORMAT_R32G32B32_USCALED: + return emit_R32G32B32_USCALED; + case PIPE_FORMAT_R32G32B32A32_USCALED: + return emit_R32G32B32A32_USCALED; + + case PIPE_FORMAT_R32_SNORM: + return emit_R32_SNORM; + case PIPE_FORMAT_R32G32_SNORM: + return emit_R32G32_SNORM; + case PIPE_FORMAT_R32G32B32_SNORM: + return emit_R32G32B32_SNORM; + case PIPE_FORMAT_R32G32B32A32_SNORM: + return emit_R32G32B32A32_SNORM; + + case PIPE_FORMAT_R32_SSCALED: + return emit_R32_SSCALED; + case PIPE_FORMAT_R32G32_SSCALED: + return emit_R32G32_SSCALED; + case PIPE_FORMAT_R32G32B32_SSCALED: + return emit_R32G32B32_SSCALED; + case PIPE_FORMAT_R32G32B32A32_SSCALED: + return emit_R32G32B32A32_SSCALED; + + case PIPE_FORMAT_R16_UNORM: + return emit_R16_UNORM; + case PIPE_FORMAT_R16G16_UNORM: + return emit_R16G16_UNORM; + case PIPE_FORMAT_R16G16B16_UNORM: + return emit_R16G16B16_UNORM; + case PIPE_FORMAT_R16G16B16A16_UNORM: + return emit_R16G16B16A16_UNORM; + + case PIPE_FORMAT_R16_USCALED: + return emit_R16_USCALED; + case PIPE_FORMAT_R16G16_USCALED: + return emit_R16G16_USCALED; + case PIPE_FORMAT_R16G16B16_USCALED: + return emit_R16G16B16_USCALED; + case PIPE_FORMAT_R16G16B16A16_USCALED: + return emit_R16G16B16A16_USCALED; + + case PIPE_FORMAT_R16_SNORM: + return emit_R16_SNORM; + case PIPE_FORMAT_R16G16_SNORM: + return emit_R16G16_SNORM; + case PIPE_FORMAT_R16G16B16_SNORM: + return emit_R16G16B16_SNORM; + case PIPE_FORMAT_R16G16B16A16_SNORM: + return emit_R16G16B16A16_SNORM; + + case PIPE_FORMAT_R16_SSCALED: + return emit_R16_SSCALED; + case PIPE_FORMAT_R16G16_SSCALED: + return emit_R16G16_SSCALED; + case PIPE_FORMAT_R16G16B16_SSCALED: + return emit_R16G16B16_SSCALED; + case PIPE_FORMAT_R16G16B16A16_SSCALED: + return emit_R16G16B16A16_SSCALED; + + case PIPE_FORMAT_R8_UNORM: + return emit_R8_UNORM; + case PIPE_FORMAT_R8G8_UNORM: + return emit_R8G8_UNORM; + case PIPE_FORMAT_R8G8B8_UNORM: + return emit_R8G8B8_UNORM; + case PIPE_FORMAT_R8G8B8A8_UNORM: + return emit_R8G8B8A8_UNORM; + + case PIPE_FORMAT_R8_USCALED: + return emit_R8_USCALED; + case PIPE_FORMAT_R8G8_USCALED: + return emit_R8G8_USCALED; + case PIPE_FORMAT_R8G8B8_USCALED: + return emit_R8G8B8_USCALED; + case PIPE_FORMAT_R8G8B8A8_USCALED: + return emit_R8G8B8A8_USCALED; + + case PIPE_FORMAT_R8_SNORM: + return emit_R8_SNORM; + case PIPE_FORMAT_R8G8_SNORM: + return emit_R8G8_SNORM; + case PIPE_FORMAT_R8G8B8_SNORM: + return emit_R8G8B8_SNORM; + case PIPE_FORMAT_R8G8B8A8_SNORM: + return emit_R8G8B8A8_SNORM; + + case PIPE_FORMAT_R8_SSCALED: + return emit_R8_SSCALED; + case PIPE_FORMAT_R8G8_SSCALED: + return emit_R8G8_SSCALED; + case PIPE_FORMAT_R8G8B8_SSCALED: + return emit_R8G8B8_SSCALED; + case PIPE_FORMAT_R8G8B8A8_SSCALED: + return emit_R8G8B8A8_SSCALED; + + case PIPE_FORMAT_A8R8G8B8_UNORM: + return emit_A8R8G8B8_UNORM; + + case PIPE_FORMAT_B8G8R8A8_UNORM: + return emit_B8G8R8A8_UNORM; + + default: + assert(0); + return emit_NULL; + } +} + + + +/** + * Fetch vertex attributes for 'count' vertices. + */ +static void generic_run_elts( struct translate *translate, + const unsigned *elts, + unsigned count, + void *output_buffer ) +{ + struct translate_generic *tg = translate_generic(translate); + char *vert = output_buffer; + unsigned nr_attrs = tg->nr_attrib; + unsigned attr; + unsigned i; + + /* loop over vertex attributes (vertex shader inputs) + */ + for (i = 0; i < count; i++) { + unsigned elt = *elts++; + + for (attr = 0; attr < nr_attrs; attr++) { + float data[4]; + + const char *src = (tg->attrib[attr].input_ptr + + tg->attrib[attr].input_stride * elt); + + char *dst = (vert + + tg->attrib[attr].output_offset); + + tg->attrib[attr].fetch( src, data ); + + if (0) debug_printf("vert %d/%d attr %d: %f %f %f %f\n", + i, elt, attr, data[0], data[1], data[2], data[3]); + + tg->attrib[attr].emit( data, dst ); + } + + vert += tg->translate.key.output_stride; + } +} + + + +static void generic_run( struct translate *translate, + unsigned start, + unsigned count, + void *output_buffer ) +{ + struct translate_generic *tg = translate_generic(translate); + char *vert = output_buffer; + unsigned nr_attrs = tg->nr_attrib; + unsigned attr; + unsigned i; + + /* loop over vertex attributes (vertex shader inputs) + */ + for (i = 0; i < count; i++) { + unsigned elt = start + i; + + for (attr = 0; attr < nr_attrs; attr++) { + float data[4]; + + const char *src = (tg->attrib[attr].input_ptr + + tg->attrib[attr].input_stride * elt); + + char *dst = (vert + + tg->attrib[attr].output_offset); + + tg->attrib[attr].fetch( src, data ); + + if (0) debug_printf("vert %d attr %d: %f %f %f %f\n", + i, attr, data[0], data[1], data[2], data[3]); + + tg->attrib[attr].emit( data, dst ); + } + + vert += tg->translate.key.output_stride; + } +} + + + +static void generic_set_buffer( struct translate *translate, + unsigned buf, + const void *ptr, + unsigned stride ) +{ + struct translate_generic *tg = translate_generic(translate); + unsigned i; + + for (i = 0; i < tg->nr_attrib; i++) { + if (tg->attrib[i].buffer == buf) { + tg->attrib[i].input_ptr = ((char *)ptr + + tg->attrib[i].input_offset); + tg->attrib[i].input_stride = stride; + } + } +} + + +static void generic_release( struct translate *translate ) +{ + /* Refcount? + */ + FREE(translate); +} + +struct translate *translate_generic_create( const struct translate_key *key ) +{ + struct translate_generic *tg = CALLOC_STRUCT(translate_generic); + unsigned i; + + if (tg == NULL) + return NULL; + + tg->translate.key = *key; + tg->translate.release = generic_release; + tg->translate.set_buffer = generic_set_buffer; + tg->translate.run_elts = generic_run_elts; + tg->translate.run = generic_run; + + for (i = 0; i < key->nr_elements; i++) { + + tg->attrib[i].fetch = get_fetch_func(key->element[i].input_format); + tg->attrib[i].buffer = key->element[i].input_buffer; + tg->attrib[i].input_offset = key->element[i].input_offset; + + tg->attrib[i].emit = get_emit_func(key->element[i].output_format); + tg->attrib[i].output_offset = key->element[i].output_offset; + + } + + tg->nr_attrib = key->nr_elements; + + + return &tg->translate; +} diff --git a/src/gallium/auxiliary/translate/translate_sse.c b/src/gallium/auxiliary/translate/translate_sse.c new file mode 100644 index 0000000000..a54ac5a82f --- /dev/null +++ b/src/gallium/auxiliary/translate/translate_sse.c @@ -0,0 +1,625 @@ +/* + * Copyright 2003 Tungsten Graphics, inc. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * on the rights to use, copy, modify, merge, publish, distribute, sub + * license, and/or sell copies of the Software, and to permit persons to whom + * the Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * TUNGSTEN GRAPHICS AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: + * Keith Whitwell <keithw@tungstengraphics.com> + */ + + +#include "pipe/p_compiler.h" +#include "pipe/p_util.h" +#include "util/u_simple_list.h" + +#include "translate.h" + + +#if defined(__i386__) || defined(__386__) || defined(i386) + +#include "rtasm/rtasm_cpu.h" +#include "rtasm/rtasm_x86sse.h" + + +#define X 0 +#define Y 1 +#define Z 2 +#define W 3 + + +#ifdef WIN32 +#define RTASM __cdecl +#else +#define RTASM +#endif + +typedef void (RTASM *run_func)( struct translate *translate, + unsigned start, + unsigned count, + void *output_buffer ); + +typedef void (RTASM *run_elts_func)( struct translate *translate, + const unsigned *elts, + unsigned count, + void *output_buffer ); + + + +struct translate_sse { + struct translate translate; + + struct x86_function linear_func; + struct x86_function elt_func; + struct x86_function *func; + + boolean loaded_identity; + boolean loaded_255; + boolean loaded_inv_255; + + float identity[4]; + float float_255[4]; + float inv_255[4]; + + struct { + char *input_ptr; + unsigned input_stride; + } attrib[PIPE_MAX_ATTRIBS]; + + run_func gen_run; + run_elts_func gen_run_elts; + +}; + +static int get_offset( const void *a, const void *b ) +{ + return (const char *)b - (const char *)a; +} + + + +static struct x86_reg get_identity( struct translate_sse *p ) +{ + struct x86_reg reg = x86_make_reg(file_XMM, 6); + + if (!p->loaded_identity) { + /* Nasty: + */ + struct x86_reg translateESI = x86_make_reg(file_REG32, reg_SI); + + p->loaded_identity = TRUE; + p->identity[0] = 0; + p->identity[1] = 0; + p->identity[2] = 0; + p->identity[3] = 1; + + sse_movups(p->func, reg, + x86_make_disp(translateESI, + get_offset(p, &p->identity[0]))); + } + + return reg; +} + +static struct x86_reg get_255( struct translate_sse *p ) +{ + struct x86_reg reg = x86_make_reg(file_XMM, 6); + + if (!p->loaded_255) { + struct x86_reg translateESI = x86_make_reg(file_REG32, reg_SI); + + p->loaded_255 = TRUE; + p->float_255[0] = + p->float_255[1] = + p->float_255[2] = + p->float_255[3] = 255.0f; + + sse_movups(p->func, reg, + x86_make_disp(translateESI, + get_offset(p, &p->float_255[0]))); + } + + return reg; + return x86_make_reg(file_XMM, 7); +} + +static struct x86_reg get_inv_255( struct translate_sse *p ) +{ + struct x86_reg reg = x86_make_reg(file_XMM, 5); + + if (!p->loaded_inv_255) { + struct x86_reg translateESI = x86_make_reg(file_REG32, reg_SI); + + p->loaded_inv_255 = TRUE; + p->inv_255[0] = + p->inv_255[1] = + p->inv_255[2] = + p->inv_255[3] = 1.0f / 255.0f; + + sse_movups(p->func, reg, + x86_make_disp(translateESI, + get_offset(p, &p->inv_255[0]))); + } + + return reg; +} + + +static void emit_load_R32G32B32A32( struct translate_sse *p, + struct x86_reg data, + struct x86_reg arg0 ) +{ + sse_movups(p->func, data, arg0); +} + +static void emit_load_R32G32B32( struct translate_sse *p, + struct x86_reg data, + struct x86_reg arg0 ) +{ + /* Have to jump through some hoops: + * + * c 0 0 0 + * c 0 0 1 + * 0 0 c 1 + * a b c 1 + */ + sse_movss(p->func, data, x86_make_disp(arg0, 8)); + sse_shufps(p->func, data, get_identity(p), SHUF(X,Y,Z,W) ); + sse_shufps(p->func, data, data, SHUF(Y,Z,X,W) ); + sse_movlps(p->func, data, arg0); +} + +static void emit_load_R32G32( struct translate_sse *p, + struct x86_reg data, + struct x86_reg arg0 ) +{ + /* 0 0 0 1 + * a b 0 1 + */ + sse_movups(p->func, data, get_identity(p) ); + sse_movlps(p->func, data, arg0); +} + + +static void emit_load_R32( struct translate_sse *p, + struct x86_reg data, + struct x86_reg arg0 ) +{ + /* a 0 0 0 + * a 0 0 1 + */ + sse_movss(p->func, data, arg0); + sse_orps(p->func, data, get_identity(p) ); +} + + +static void emit_load_R8G8B8A8_UNORM( struct translate_sse *p, + struct x86_reg data, + struct x86_reg src ) +{ + + /* Load and unpack twice: + */ + sse_movss(p->func, data, src); + sse2_punpcklbw(p->func, data, get_identity(p)); + sse2_punpcklbw(p->func, data, get_identity(p)); + + /* Convert to float: + */ + sse2_cvtdq2ps(p->func, data, data); + + + /* Scale by 1/255.0 + */ + sse_mulps(p->func, data, get_inv_255(p)); +} + + + + +static void emit_store_R32G32B32A32( struct translate_sse *p, + struct x86_reg dest, + struct x86_reg dataXMM ) +{ + sse_movups(p->func, dest, dataXMM); +} + +static void emit_store_R32G32B32( struct translate_sse *p, + struct x86_reg dest, + struct x86_reg dataXMM ) +{ + /* Emit two, shuffle, emit one. + */ + sse_movlps(p->func, dest, dataXMM); + sse_shufps(p->func, dataXMM, dataXMM, SHUF(Z,Z,Z,Z) ); /* NOTE! destructive */ + sse_movss(p->func, x86_make_disp(dest,8), dataXMM); +} + +static void emit_store_R32G32( struct translate_sse *p, + struct x86_reg dest, + struct x86_reg dataXMM ) +{ + sse_movlps(p->func, dest, dataXMM); +} + +static void emit_store_R32( struct translate_sse *p, + struct x86_reg dest, + struct x86_reg dataXMM ) +{ + sse_movss(p->func, dest, dataXMM); +} + + + +static void emit_store_R8G8B8A8_UNORM( struct translate_sse *p, + struct x86_reg dest, + struct x86_reg dataXMM ) +{ + /* Scale by 255.0 + */ + sse_mulps(p->func, dataXMM, get_255(p)); + + /* Pack and emit: + */ + sse2_cvtps2dq(p->func, dataXMM, dataXMM); + sse2_packssdw(p->func, dataXMM, dataXMM); + sse2_packuswb(p->func, dataXMM, dataXMM); + sse_movss(p->func, dest, dataXMM); +} + + + + + +static void get_src_ptr( struct translate_sse *p, + struct x86_reg srcEAX, + struct x86_reg translateREG, + struct x86_reg eltREG, + unsigned a ) +{ + struct x86_reg input_ptr = + x86_make_disp(translateREG, + get_offset(p, &p->attrib[a].input_ptr)); + + struct x86_reg input_stride = + x86_make_disp(translateREG, + get_offset(p, &p->attrib[a].input_stride)); + + /* Calculate pointer to current attrib: + */ + x86_mov(p->func, srcEAX, input_stride); + x86_imul(p->func, srcEAX, eltREG); + x86_add(p->func, srcEAX, input_ptr); +} + + +/* Extended swizzles? Maybe later. + */ +static void emit_swizzle( struct translate_sse *p, + struct x86_reg dest, + struct x86_reg src, + unsigned shuffle ) +{ + sse_shufps(p->func, dest, src, shuffle); +} + + +static boolean translate_attr( struct translate_sse *p, + const struct translate_element *a, + struct x86_reg srcECX, + struct x86_reg dstEAX) +{ + struct x86_reg dataXMM = x86_make_reg(file_XMM, 0); + + switch (a->input_format) { + case PIPE_FORMAT_R32_FLOAT: + emit_load_R32(p, dataXMM, srcECX); + break; + case PIPE_FORMAT_R32G32_FLOAT: + emit_load_R32G32(p, dataXMM, srcECX); + break; + case PIPE_FORMAT_R32G32B32_FLOAT: + emit_load_R32G32B32(p, dataXMM, srcECX); + break; + case PIPE_FORMAT_R32G32B32A32_FLOAT: + emit_load_R32G32B32A32(p, dataXMM, srcECX); + break; + case PIPE_FORMAT_B8G8R8A8_UNORM: + emit_load_R8G8B8A8_UNORM(p, dataXMM, srcECX); + emit_swizzle(p, dataXMM, dataXMM, SHUF(Z,Y,X,W)); + break; + case PIPE_FORMAT_R8G8B8A8_UNORM: + emit_load_R8G8B8A8_UNORM(p, dataXMM, srcECX); + break; + default: + return FALSE; + } + + switch (a->output_format) { + case PIPE_FORMAT_R32_FLOAT: + emit_store_R32(p, dstEAX, dataXMM); + break; + case PIPE_FORMAT_R32G32_FLOAT: + emit_store_R32G32(p, dstEAX, dataXMM); + break; + case PIPE_FORMAT_R32G32B32_FLOAT: + emit_store_R32G32B32(p, dstEAX, dataXMM); + break; + case PIPE_FORMAT_R32G32B32A32_FLOAT: + emit_store_R32G32B32A32(p, dstEAX, dataXMM); + break; + case PIPE_FORMAT_B8G8R8A8_UNORM: + emit_swizzle(p, dataXMM, dataXMM, SHUF(Z,Y,X,W)); + emit_store_R8G8B8A8_UNORM(p, dstEAX, dataXMM); + break; + case PIPE_FORMAT_R8G8B8A8_UNORM: + emit_store_R8G8B8A8_UNORM(p, dstEAX, dataXMM); + break; + default: + return FALSE; + } + + return TRUE; +} + +/* Build run( struct translate *translate, + * unsigned start, + * unsigned count, + * void *output_buffer ) + * or + * run_elts( struct translate *translate, + * unsigned *elts, + * unsigned count, + * void *output_buffer ) + * + * Lots of hardcoding + * + * EAX -- pointer to current output vertex + * ECX -- pointer to current attribute + * + */ +static boolean build_vertex_emit( struct translate_sse *p, + struct x86_function *func, + boolean linear ) +{ + struct x86_reg vertexECX = x86_make_reg(file_REG32, reg_AX); + struct x86_reg idxEBX = x86_make_reg(file_REG32, reg_BX); + struct x86_reg srcEAX = x86_make_reg(file_REG32, reg_CX); + struct x86_reg countEBP = x86_make_reg(file_REG32, reg_BP); + struct x86_reg translateESI = x86_make_reg(file_REG32, reg_SI); + int fixup, label; + unsigned j; + + p->func = func; + p->loaded_inv_255 = FALSE; + p->loaded_255 = FALSE; + p->loaded_identity = FALSE; + + x86_init_func(p->func); + + /* Push a few regs? + */ + x86_push(p->func, countEBP); + x86_push(p->func, translateESI); + x86_push(p->func, idxEBX); + + /* Get vertex count, compare to zero + */ + x86_xor(p->func, idxEBX, idxEBX); + x86_mov(p->func, countEBP, x86_fn_arg(p->func, 3)); + x86_cmp(p->func, countEBP, idxEBX); + fixup = x86_jcc_forward(p->func, cc_E); + + /* If linear, idx is the current element, otherwise it is a pointer + * to the current element. + */ + x86_mov(p->func, idxEBX, x86_fn_arg(p->func, 2)); + + /* Initialize destination register. + */ + x86_mov(p->func, vertexECX, x86_fn_arg(p->func, 4)); + + /* Move argument 1 (translate_sse pointer) into a reg: + */ + x86_mov(p->func, translateESI, x86_fn_arg(p->func, 1)); + + + /* always load, needed or not: + */ + + /* Note address for loop jump */ + label = x86_get_label(p->func); + + + for (j = 0; j < p->translate.key.nr_elements; j++) { + const struct translate_element *a = &p->translate.key.element[j]; + + struct x86_reg destEAX = x86_make_disp(vertexECX, + a->output_offset); + + /* Figure out source pointer address: + */ + if (linear) { + get_src_ptr(p, srcEAX, translateESI, idxEBX, j); + } + else { + get_src_ptr(p, srcEAX, translateESI, x86_deref(idxEBX), j); + } + + if (!translate_attr( p, a, x86_deref(srcEAX), destEAX )) + return FALSE; + } + + /* Next vertex: + */ + x86_lea(p->func, vertexECX, x86_make_disp(vertexECX, p->translate.key.output_stride)); + + /* Incr index + */ /* Emit code for each of the attributes. Currently routes + * everything through SSE registers, even when it might be more + * efficient to stick with regular old x86. No optimization or + * other tricks - enough new ground to cover here just getting + * things working. + */ + + if (linear) { + x86_inc(p->func, idxEBX); + } + else { + x86_lea(p->func, idxEBX, x86_make_disp(idxEBX, 4)); + } + + /* decr count, loop if not zero + */ + x86_dec(p->func, countEBP); + x86_test(p->func, countEBP, countEBP); + x86_jcc(p->func, cc_NZ, label); + + /* Exit mmx state? + */ + if (p->func->need_emms) + mmx_emms(p->func); + + /* Land forward jump here: + */ + x86_fixup_fwd_jump(p->func, fixup); + + /* Pop regs and return + */ + + x86_pop(p->func, idxEBX); + x86_pop(p->func, translateESI); + x86_pop(p->func, countEBP); + x86_ret(p->func); + + return TRUE; +} + + + + + + + +static void translate_sse_set_buffer( struct translate *translate, + unsigned buf, + const void *ptr, + unsigned stride ) +{ + struct translate_sse *p = (struct translate_sse *)translate; + unsigned i; + + for (i = 0; i < p->translate.key.nr_elements; i++) { + if (p->translate.key.element[i].input_buffer == buf) { + p->attrib[i].input_ptr = ((char *)ptr + + p->translate.key.element[i].input_offset); + p->attrib[i].input_stride = stride; + } + } +} + + +static void translate_sse_release( struct translate *translate ) +{ + struct translate_sse *p = (struct translate_sse *)translate; + + x86_release_func( &p->linear_func ); + x86_release_func( &p->elt_func ); + + FREE(p); +} + +static void translate_sse_run_elts( struct translate *translate, + const unsigned *elts, + unsigned count, + void *output_buffer ) +{ + struct translate_sse *p = (struct translate_sse *)translate; + + p->gen_run_elts( translate, + elts, + count, + output_buffer ); +} + +static void translate_sse_run( struct translate *translate, + unsigned start, + unsigned count, + void *output_buffer ) +{ + struct translate_sse *p = (struct translate_sse *)translate; + + p->gen_run( translate, + start, + count, + output_buffer ); +} + + +struct translate *translate_sse2_create( const struct translate_key *key ) +{ + struct translate_sse *p = NULL; + + if (!rtasm_cpu_has_sse() || !rtasm_cpu_has_sse2()) + goto fail; + + p = CALLOC_STRUCT( translate_sse ); + if (p == NULL) + goto fail; + + p->translate.key = *key; + p->translate.release = translate_sse_release; + p->translate.set_buffer = translate_sse_set_buffer; + p->translate.run_elts = translate_sse_run_elts; + p->translate.run = translate_sse_run; + + if (!build_vertex_emit(p, &p->linear_func, TRUE)) + goto fail; + + if (!build_vertex_emit(p, &p->elt_func, FALSE)) + goto fail; + + p->gen_run = (run_func)x86_get_func(&p->linear_func); + if (p->gen_run == NULL) + goto fail; + + p->gen_run_elts = (run_elts_func)x86_get_func(&p->elt_func); + if (p->gen_run_elts == NULL) + goto fail; + + return &p->translate; + + fail: + if (p) + translate_sse_release( &p->translate ); + + return NULL; +} + + + +#else + +void translate_create_sse( const struct translate_key *key ) +{ + return NULL; +} + +#endif diff --git a/src/gallium/auxiliary/util/Makefile b/src/gallium/auxiliary/util/Makefile index 9b6c2708b6..05bc43131a 100644 --- a/src/gallium/auxiliary/util/Makefile +++ b/src/gallium/auxiliary/util/Makefile @@ -14,7 +14,9 @@ C_SOURCES = \ u_hash_table.c \ u_mm.c \ u_simple_shaders.c \ - u_snprintf.c + u_snprintf.c \ + u_time.c \ + u_mm.c include ../../Makefile.template diff --git a/src/gallium/auxiliary/util/SConscript b/src/gallium/auxiliary/util/SConscript index 9b3929eb2d..d55d2c7081 100644 --- a/src/gallium/auxiliary/util/SConscript +++ b/src/gallium/auxiliary/util/SConscript @@ -15,6 +15,7 @@ util = env.ConvenienceLibrary( 'u_mm.c', 'u_simple_shaders.c', 'u_snprintf.c', + 'u_time.c', ]) auxiliaries.insert(0, util) diff --git a/src/gallium/auxiliary/util/p_debug.c b/src/gallium/auxiliary/util/p_debug.c index 090e3b7794..4ec1746662 100644 --- a/src/gallium/auxiliary/util/p_debug.c +++ b/src/gallium/auxiliary/util/p_debug.c @@ -26,9 +26,11 @@ **************************************************************************/ +#include "pipe/p_config.h" + #include <stdarg.h> -#ifdef WIN32 +#ifdef PIPE_SUBSYSTEM_WINDOWS_DISPLAY #include <windows.h> #include <winddi.h> #else @@ -39,9 +41,11 @@ #include "pipe/p_compiler.h" #include "pipe/p_util.h" #include "pipe/p_debug.h" +#include "pipe/p_format.h" +#include "util/u_string.h" -#ifdef WIN32 +#ifdef PIPE_SUBSYSTEM_WINDOWS_DISPLAY static INLINE void _EngDebugPrint(const char *format, ...) { @@ -55,13 +59,18 @@ _EngDebugPrint(const char *format, ...) void _debug_vprintf(const char *format, va_list ap) { -#ifdef WIN32 +#ifdef PIPE_SUBSYSTEM_WINDOWS_DISPLAY #ifndef WINCE /* EngDebugPrint does not handle float point arguments, so we need to use - * our own vsnprintf implementation */ - char buf[512 + 1]; - vsnprintf(buf, sizeof(buf), format, ap); - _EngDebugPrint("%s", buf); + * our own vsnprintf implementation. It is also very slow, so buffer until + * we find a newline. */ + static char buf[512 + 1] = {'\0'}; + size_t len = strlen(buf); + int ret = util_vsnprintf(buf + len, sizeof(buf) - len, format, ap); + if(ret > (int)(sizeof(buf) - len - 1) || strchr(buf + len, '\n')) { + _EngDebugPrint("%s", buf); + buf[0] = '\0'; + } #else /* TODO: Implement debug print for WINCE */ #endif @@ -95,7 +104,7 @@ void _debug_break(void) __asm("int3"); #elif (defined(__i386__) || defined(__386__)) && defined(__MSC__) _asm {int 3}; -#elif defined(WIN32) && !defined(WINCE) +#elif defined(PIPE_SUBSYSTEM_WINDOWS_DISPLAY) EngDebugBreak(); #else abort(); @@ -103,7 +112,7 @@ void _debug_break(void) } -#ifdef WIN32 +#ifdef PIPE_SUBSYSTEM_WINDOWS_DISPLAY static const char * find(const char *start, const char *end, char c) { @@ -144,7 +153,7 @@ const char * debug_get_option(const char *name, const char *dfault) { const char *result; -#ifdef WIN32 +#ifdef PIPE_SUBSYSTEM_WINDOWS_DISPLAY ULONG_PTR iFile = 0; const void *pMap = NULL; const char *sol, *eol, *sep; @@ -194,6 +203,8 @@ debug_get_bool_option(const char *name, boolean dfault) if(str == NULL) result = dfault; + else if(!strcmp(str, "n")) + result = FALSE; else if(!strcmp(str, "no")) result = FALSE; else if(!strcmp(str, "0")) @@ -245,57 +256,16 @@ debug_get_flags_option(const char *name, } -#if defined(WIN32) -ULONG_PTR debug_config_file = 0; -void *mapped_config_file = 0; - -enum { - eAssertAbortEn = 0x1, -}; - -/* Check for aborts enabled. */ -static unsigned abort_en(void) -{ - if (!mapped_config_file) - { - /* Open an 8 byte file for configuration data. */ - mapped_config_file = EngMapFile(L"\\??\\c:\\gaDebug.cfg", 8, &debug_config_file); - } - - /* A value of "0" (ascii) in the configuration file will clear the - * first 8 bits in the test byte. - * - * A value of "1" (ascii) in the configuration file will set the - * first bit in the test byte. - * - * A value of "2" (ascii) in the configuration file will set the - * second bit in the test byte. - * - * Currently the only interesting values are 0 and 1, which clear - * and set abort-on-assert behaviour respectively. - */ - return ((((char *)mapped_config_file)[0]) - 0x30) & eAssertAbortEn; -} -#else /* WIN32 */ -static unsigned abort_en(void) -{ - return !GETENV("GALLIUM_ABORT_ON_ASSERT"); -} -#endif - void _debug_assert_fail(const char *expr, const char *file, unsigned line, const char *function) { _debug_printf("%s:%u:%s: Assertion `%s' failed.\n", file, line, function, expr); - if (abort_en()) - { + if (debug_get_bool_option("GALLIUM_ABORT_ON_ASSERT", TRUE)) debug_break(); - } else - { + else _debug_printf("continuing...\n"); - } } @@ -311,7 +281,7 @@ debug_dump_enum(const struct debug_named_value *names, ++names; } - snprintf(rest, sizeof(rest), "0x%08lx", value); + util_snprintf(rest, sizeof(rest), "0x%08lx", value); return rest; } @@ -344,7 +314,7 @@ debug_dump_flags(const struct debug_named_value *names, else first = 0; - snprintf(rest, sizeof(rest), "0x%08lx", value); + util_snprintf(rest, sizeof(rest), "0x%08lx", value); strncat(output, rest, sizeof(output)); } @@ -355,3 +325,101 @@ debug_dump_flags(const struct debug_named_value *names, } + + + + +char *pf_sprint_name( char *str, enum pipe_format format ) +{ + strcpy( str, "PIPE_FORMAT_" ); + switch (pf_layout( format )) { + case PIPE_FORMAT_LAYOUT_RGBAZS: + { + pipe_format_rgbazs_t rgbazs = (pipe_format_rgbazs_t) format; + uint i; + uint scale = 1 << (pf_exp8( rgbazs ) * 3); + + for (i = 0; i < 4; i++) { + uint size = pf_size_xyzw( rgbazs, i ); + + if (size == 0) { + break; + } + switch (pf_swizzle_xyzw( rgbazs, i )) { + case PIPE_FORMAT_COMP_R: + strcat( str, "R" ); + break; + case PIPE_FORMAT_COMP_G: + strcat( str, "G" ); + break; + case PIPE_FORMAT_COMP_B: + strcat( str, "B" ); + break; + case PIPE_FORMAT_COMP_A: + strcat( str, "A" ); + break; + case PIPE_FORMAT_COMP_0: + strcat( str, "0" ); + break; + case PIPE_FORMAT_COMP_1: + strcat( str, "1" ); + break; + case PIPE_FORMAT_COMP_Z: + strcat( str, "Z" ); + break; + case PIPE_FORMAT_COMP_S: + strcat( str, "S" ); + break; + } + util_snprintf( &str[strlen( str )], 32, "%u", size * scale ); + } + if (i != 0) { + strcat( str, "_" ); + } + switch (pf_type( rgbazs )) { + case PIPE_FORMAT_TYPE_UNKNOWN: + strcat( str, "NONE" ); + break; + case PIPE_FORMAT_TYPE_FLOAT: + strcat( str, "FLOAT" ); + break; + case PIPE_FORMAT_TYPE_UNORM: + strcat( str, "UNORM" ); + break; + case PIPE_FORMAT_TYPE_SNORM: + strcat( str, "SNORM" ); + break; + case PIPE_FORMAT_TYPE_USCALED: + strcat( str, "USCALED" ); + break; + case PIPE_FORMAT_TYPE_SSCALED: + strcat( str, "SSCALED" ); + break; + } + } + break; + case PIPE_FORMAT_LAYOUT_YCBCR: + { + pipe_format_ycbcr_t ycbcr = (pipe_format_ycbcr_t) format; + + strcat( str, "YCBCR" ); + if (pf_rev( ycbcr )) { + strcat( str, "_REV" ); + } + } + break; + } + return str; +} + + +#ifdef DEBUG +void debug_print_format(const char *msg, unsigned fmt ) +{ + char fmtstr[80]; + + pf_sprint_name(fmtstr, (enum pipe_format)fmt); + + debug_printf("%s: %s\n", msg, fmtstr); +} +#endif diff --git a/src/gallium/auxiliary/util/p_debug_mem.c b/src/gallium/auxiliary/util/p_debug_mem.c index c160afe5b7..3b5e4fbaee 100644 --- a/src/gallium/auxiliary/util/p_debug_mem.c +++ b/src/gallium/auxiliary/util/p_debug_mem.c @@ -32,9 +32,13 @@ * @author José Fonseca <jrfonseca@tungstengraphics.com> */ -#ifdef WIN32 +#include "pipe/p_config.h" + +#if defined(PIPE_SUBSYSTEM_WINDOWS_DISPLAY) #include <windows.h> #include <winddi.h> +#elif defined(PIPE_SUBSYSTEM_WINDOWS_MINIPORT) +#include <wdm.h> #else #include <stdio.h> #include <stdlib.h> @@ -47,9 +51,12 @@ #define DEBUG_MEMORY_MAGIC 0x6e34090aU -#if defined(WIN32) && !defined(WINCE) +#if defined(PIPE_SUBSYSTEM_WINDOWS_DISPLAY) && !defined(WINCE) #define real_malloc(_size) EngAllocMem(0, _size, 'D3AG') #define real_free(_ptr) EngFreeMem(_ptr) +#elif defined(PIPE_SUBSYSTEM_WINDOWS_MINIPORT) +#define real_malloc(_size) ExAllocatePool(0, _size) +#define real_free(_ptr) ExFreePool(_ptr) #else #define real_malloc(_size) malloc(_size) #define real_free(_ptr) free(_ptr) @@ -70,8 +77,26 @@ struct debug_memory_header static struct list_head list = { &list, &list }; -static unsigned long start_no = 0; -static unsigned long end_no = 0; +static unsigned long last_no = 0; + + +static INLINE struct debug_memory_header * +header_from_data(void *data) +{ + if(data) + return (struct debug_memory_header *)((char *)data - sizeof(struct debug_memory_header)); + else + return NULL; +} + +static INLINE void * +data_from_header(struct debug_memory_header *hdr) +{ + if(hdr) + return (void *)((char *)hdr + sizeof(struct debug_memory_header)); + else + return NULL; +} void * @@ -84,7 +109,7 @@ debug_malloc(const char *file, unsigned line, const char *function, if(!hdr) return NULL; - hdr->no = end_no++; + hdr->no = last_no++; hdr->file = file; hdr->line = line; hdr->function = function; @@ -93,7 +118,7 @@ debug_malloc(const char *file, unsigned line, const char *function, LIST_ADDTAIL(&hdr->head, &list); - return (void *)((char *)hdr + sizeof(*hdr)); + return data_from_header(hdr); } void @@ -105,7 +130,7 @@ debug_free(const char *file, unsigned line, const char *function, if(!ptr) return; - hdr = (struct debug_memory_header *)((char *)ptr - sizeof(*hdr)); + hdr = header_from_data(ptr); if(hdr->magic != DEBUG_MEMORY_MAGIC) { debug_printf("%s:%u:%s: freeing bad or corrupted memory %p\n", file, line, function, @@ -134,27 +159,57 @@ void * debug_realloc(const char *file, unsigned line, const char *function, void *old_ptr, size_t old_size, size_t new_size ) { - void *new_ptr = NULL; - - if (new_size != 0) { - new_ptr = debug_malloc( file, line, function, new_size ); - - if( new_ptr && old_ptr ) - memcpy( new_ptr, old_ptr, old_size ); + struct debug_memory_header *old_hdr, *new_hdr; + void *new_ptr; + + if(!old_ptr) + return debug_malloc( file, line, function, new_size ); + + if(!new_size) { + debug_free( file, line, function, old_ptr ); + return NULL; + } + + old_hdr = header_from_data(old_ptr); + if(old_hdr->magic != DEBUG_MEMORY_MAGIC) { + debug_printf("%s:%u:%s: reallocating bad or corrupted memory %p\n", + file, line, function, + old_ptr); + debug_assert(0); + return NULL; } + + /* alloc new */ + new_hdr = real_malloc(sizeof(*new_hdr) + new_size); + if(!new_hdr) + return NULL; + new_hdr->no = old_hdr->no; + new_hdr->file = old_hdr->file; + new_hdr->line = old_hdr->line; + new_hdr->function = old_hdr->function; + new_hdr->size = new_size; + new_hdr->magic = DEBUG_MEMORY_MAGIC; + LIST_REPLACE(&old_hdr->head, &new_hdr->head); + + /* copy data */ + new_ptr = data_from_header(new_hdr); + memcpy( new_ptr, old_ptr, old_size < new_size ? old_size : new_size ); + + /* free old */ + old_hdr->magic = 0; + real_free(old_hdr); - debug_free( file, line, function, old_ptr ); return new_ptr; } -void -debug_memory_reset(void) +unsigned long +debug_memory_begin(void) { - start_no = end_no; + return last_no; } void -debug_memory_report(void) +debug_memory_end(unsigned long start_no) { struct list_head *entry; @@ -163,8 +218,9 @@ debug_memory_report(void) struct debug_memory_header *hdr; void *ptr; hdr = LIST_ENTRY(struct debug_memory_header, entry, head); - ptr = (void *)((char *)hdr + sizeof(*hdr)); - if(hdr->no >= start_no) + ptr = data_from_header(hdr); + if(start_no <= hdr->no && hdr->no < last_no || + last_no < start_no && (hdr->no < last_no || start_no <= hdr->no)) debug_printf("%s:%u:%s: %u bytes at %p not freed\n", hdr->file, hdr->line, hdr->function, hdr->size, ptr); diff --git a/src/gallium/auxiliary/util/p_tile.c b/src/gallium/auxiliary/util/p_tile.c index 520da5cecd..63e1cc6013 100644 --- a/src/gallium/auxiliary/util/p_tile.c +++ b/src/gallium/auxiliary/util/p_tile.c @@ -169,6 +169,52 @@ a8r8g8b8_put_tile_rgba(unsigned *dst, } +/*** PIPE_FORMAT_A8R8G8B8_UNORM ***/ + +static void +x8r8g8b8_get_tile_rgba(unsigned *src, + unsigned w, unsigned h, + float *p, + unsigned dst_stride) +{ + unsigned i, j; + + for (i = 0; i < h; i++) { + float *pRow = p; + for (j = 0; j < w; j++, pRow += 4) { + const unsigned pixel = *src++; + pRow[0] = UBYTE_TO_FLOAT((pixel >> 16) & 0xff); + pRow[1] = UBYTE_TO_FLOAT((pixel >> 8) & 0xff); + pRow[2] = UBYTE_TO_FLOAT((pixel >> 0) & 0xff); + pRow[3] = UBYTE_TO_FLOAT(0xff); + } + p += dst_stride; + } +} + + +static void +x8r8g8b8_put_tile_rgba(unsigned *dst, + unsigned w, unsigned h, + const float *p, + unsigned src_stride) +{ + unsigned i, j; + + for (i = 0; i < h; i++) { + const float *pRow = p; + for (j = 0; j < w; j++, pRow += 4) { + unsigned r, g, b; + UNCLAMPED_FLOAT_TO_UBYTE(r, pRow[0]); + UNCLAMPED_FLOAT_TO_UBYTE(g, pRow[1]); + UNCLAMPED_FLOAT_TO_UBYTE(b, pRow[2]); + *dst++ = (0xff << 24) | (r << 16) | (g << 8) | b; + } + p += src_stride; + } +} + + /*** PIPE_FORMAT_B8G8R8A8_UNORM ***/ static void @@ -339,7 +385,7 @@ z16_get_tile_rgba(ushort *src, -/*** PIPE_FORMAT_U_L8 ***/ +/*** PIPE_FORMAT_L8_UNORM ***/ static void l8_get_tile_rgba(ubyte *src, @@ -362,7 +408,7 @@ l8_get_tile_rgba(ubyte *src, } -/*** PIPE_FORMAT_U_A8 ***/ +/*** PIPE_FORMAT_A8_UNORM ***/ static void a8_get_tile_rgba(ubyte *src, @@ -430,7 +476,7 @@ r16g16b16a16_put_tile_rgba(short *dst, -/*** PIPE_FORMAT_U_I8 ***/ +/*** PIPE_FORMAT_I8_UNORM ***/ static void i8_get_tile_rgba(ubyte *src, @@ -453,7 +499,7 @@ i8_get_tile_rgba(ubyte *src, } -/*** PIPE_FORMAT_U_A8_L8 ***/ +/*** PIPE_FORMAT_A8L8_UNORM ***/ static void a8_l8_get_tile_rgba(ushort *src, @@ -647,6 +693,9 @@ pipe_get_tile_rgba(struct pipe_context *pipe, case PIPE_FORMAT_A8R8G8B8_UNORM: a8r8g8b8_get_tile_rgba((unsigned *) packed, w, h, p, dst_stride); break; + case PIPE_FORMAT_X8R8G8B8_UNORM: + x8r8g8b8_get_tile_rgba((unsigned *) packed, w, h, p, dst_stride); + break; case PIPE_FORMAT_B8G8R8A8_UNORM: b8g8r8a8_get_tile_rgba((unsigned *) packed, w, h, p, dst_stride); break; @@ -659,16 +708,16 @@ pipe_get_tile_rgba(struct pipe_context *pipe, case PIPE_FORMAT_R5G6B5_UNORM: r5g6b5_get_tile_rgba((ushort *) packed, w, h, p, dst_stride); break; - case PIPE_FORMAT_U_L8: + case PIPE_FORMAT_L8_UNORM: l8_get_tile_rgba((ubyte *) packed, w, h, p, dst_stride); break; - case PIPE_FORMAT_U_A8: + case PIPE_FORMAT_A8_UNORM: a8_get_tile_rgba((ubyte *) packed, w, h, p, dst_stride); break; - case PIPE_FORMAT_U_I8: + case PIPE_FORMAT_I8_UNORM: i8_get_tile_rgba((ubyte *) packed, w, h, p, dst_stride); break; - case PIPE_FORMAT_U_A8_L8: + case PIPE_FORMAT_A8L8_UNORM: a8_l8_get_tile_rgba((ushort *) packed, w, h, p, dst_stride); break; case PIPE_FORMAT_R16G16B16A16_SNORM: @@ -681,6 +730,7 @@ pipe_get_tile_rgba(struct pipe_context *pipe, z32_get_tile_rgba((unsigned *) packed, w, h, p, dst_stride); break; case PIPE_FORMAT_S8Z24_UNORM: + case PIPE_FORMAT_X8Z24_UNORM: s8z24_get_tile_rgba((unsigned *) packed, w, h, p, dst_stride); break; case PIPE_FORMAT_Z24S8_UNORM: @@ -723,6 +773,9 @@ pipe_put_tile_rgba(struct pipe_context *pipe, case PIPE_FORMAT_A8R8G8B8_UNORM: a8r8g8b8_put_tile_rgba((unsigned *) packed, w, h, p, src_stride); break; + case PIPE_FORMAT_X8R8G8B8_UNORM: + x8r8g8b8_put_tile_rgba((unsigned *) packed, w, h, p, src_stride); + break; case PIPE_FORMAT_B8G8R8A8_UNORM: b8g8r8a8_put_tile_rgba((unsigned *) packed, w, h, p, src_stride); break; @@ -734,16 +787,16 @@ pipe_put_tile_rgba(struct pipe_context *pipe, break; case PIPE_FORMAT_R8G8B8A8_UNORM: break; - case PIPE_FORMAT_U_L8: + case PIPE_FORMAT_L8_UNORM: /*l8_put_tile_rgba((ubyte *) packed, w, h, p, src_stride);*/ break; - case PIPE_FORMAT_U_A8: + case PIPE_FORMAT_A8_UNORM: /*a8_put_tile_rgba((ubyte *) packed, w, h, p, src_stride);*/ break; - case PIPE_FORMAT_U_I8: + case PIPE_FORMAT_I8_UNORM: /*i8_put_tile_rgba((ubyte *) packed, w, h, p, src_stride);*/ break; - case PIPE_FORMAT_U_A8_L8: + case PIPE_FORMAT_A8L8_UNORM: /*a8_l8_put_tile_rgba((ushort *) packed, w, h, p, src_stride);*/ break; case PIPE_FORMAT_R16G16B16A16_SNORM: @@ -756,6 +809,7 @@ pipe_put_tile_rgba(struct pipe_context *pipe, /*z32_put_tile_rgba((unsigned *) packed, w, h, p, src_stride);*/ break; case PIPE_FORMAT_S8Z24_UNORM: + case PIPE_FORMAT_X8Z24_UNORM: /*s8z24_put_tile_rgba((unsigned *) packed, w, h, p, src_stride);*/ break; case PIPE_FORMAT_Z24S8_UNORM: @@ -800,6 +854,7 @@ pipe_get_tile_z(struct pipe_context *pipe, } break; case PIPE_FORMAT_S8Z24_UNORM: + case PIPE_FORMAT_X8Z24_UNORM: { const uint *pSrc = (const uint *) pipe_surface_map(ps) + (y * ps->pitch + x); @@ -860,6 +915,7 @@ pipe_put_tile_z(struct pipe_context *pipe, } break; case PIPE_FORMAT_S8Z24_UNORM: + case PIPE_FORMAT_X8Z24_UNORM: { uint *pDest = (uint *) pipe_surface_map(ps) + (y * ps->pitch + x); for (i = 0; i < h; i++) { diff --git a/src/gallium/auxiliary/util/u_blit.c b/src/gallium/auxiliary/util/u_blit.c index eec5e600c9..568d62ced1 100644 --- a/src/gallium/auxiliary/util/u_blit.c +++ b/src/gallium/auxiliary/util/u_blit.c @@ -57,6 +57,7 @@ struct blit_state struct pipe_depth_stencil_alpha_state depthstencil; struct pipe_rasterizer_state rasterizer; struct pipe_sampler_state sampler; + struct pipe_viewport_state viewport; struct pipe_shader_state vert_shader; struct pipe_shader_state frag_shader; @@ -100,7 +101,7 @@ util_create_blit(struct pipe_context *pipe, struct cso_context *cso) memset(&ctx->rasterizer, 0, sizeof(ctx->rasterizer)); ctx->rasterizer.front_winding = PIPE_WINDING_CW; ctx->rasterizer.cull_mode = PIPE_WINDING_NONE; - ctx->rasterizer.bypass_clipping = 1; /* bypasses viewport too */ + ctx->rasterizer.bypass_clipping = 1; /*ctx->rasterizer.bypass_vs = 1;*/ /* samplers */ @@ -113,8 +114,7 @@ util_create_blit(struct pipe_context *pipe, struct cso_context *cso) ctx->sampler.mag_img_filter = 0; /* set later */ ctx->sampler.normalized_coords = 1; -#if 0 - /* viewport */ + /* viewport (identity, we setup vertices in wincoords) */ ctx->viewport.scale[0] = 1.0; ctx->viewport.scale[1] = 1.0; ctx->viewport.scale[2] = 1.0; @@ -123,7 +123,6 @@ util_create_blit(struct pipe_context *pipe, struct cso_context *cso) ctx->viewport.translate[1] = 0.0; ctx->viewport.translate[2] = 0.0; ctx->viewport.translate[3] = 0.0; -#endif /* vertex shader */ { @@ -265,6 +264,9 @@ util_blit_pixels(struct blit_state *ctx, dstY1 = tmp; } + assert(screen->is_format_supported(screen, src->format, PIPE_TEXTURE)); + assert(screen->is_format_supported(screen, dst->format, PIPE_SURFACE)); + /* * XXX for now we're always creating a temporary texture. * Strictly speaking that's not always needed. @@ -293,6 +295,8 @@ util_blit_pixels(struct blit_state *ctx, src, srcLeft, srcTop, /* src */ srcW, srcH); /* size */ + pipe->texture_update(pipe, tex, 0, 1 << 0); + /* save state (restored below) */ cso_save_blend(ctx->cso); cso_save_depth_stencil_alpha(ctx->cso); @@ -300,11 +304,15 @@ util_blit_pixels(struct blit_state *ctx, cso_save_samplers(ctx->cso); cso_save_sampler_textures(ctx->cso); cso_save_framebuffer(ctx->cso); + cso_save_fragment_shader(ctx->cso); + cso_save_vertex_shader(ctx->cso); + cso_save_viewport(ctx->cso); /* set misc state we care about */ cso_set_blend(ctx->cso, &ctx->blend); cso_set_depth_stencil_alpha(ctx->cso, &ctx->depthstencil); cso_set_rasterizer(ctx->cso, &ctx->rasterizer); + cso_set_viewport(ctx->cso, &ctx->viewport); /* sampler */ ctx->sampler.min_img_filter = filter; @@ -313,11 +321,11 @@ util_blit_pixels(struct blit_state *ctx, cso_single_sampler_done(ctx->cso); /* texture */ - pipe->set_sampler_textures(pipe, 1, &tex); + cso_set_sampler_textures(ctx->cso, 1, &tex); /* shaders */ - pipe->bind_fs_state(pipe, ctx->fs); - pipe->bind_vs_state(pipe, ctx->vs); + cso_set_fragment_shader_handle(ctx->cso, ctx->fs); + cso_set_vertex_shader_handle(ctx->cso, ctx->vs); /* drawing dest */ memset(&fb, 0, sizeof(fb)); @@ -344,6 +352,9 @@ util_blit_pixels(struct blit_state *ctx, cso_restore_samplers(ctx->cso); cso_restore_sampler_textures(ctx->cso); cso_restore_framebuffer(ctx->cso); + cso_restore_fragment_shader(ctx->cso); + cso_restore_vertex_shader(ctx->cso); + cso_restore_viewport(ctx->cso); /* free the texture */ pipe_surface_reference(&texSurf, NULL); diff --git a/src/gallium/auxiliary/util/u_blit.h b/src/gallium/auxiliary/util/u_blit.h index 61f1d9bb32..0ce9732e62 100644 --- a/src/gallium/auxiliary/util/u_blit.h +++ b/src/gallium/auxiliary/util/u_blit.h @@ -30,7 +30,11 @@ #define U_BLIT_H +#ifdef __cplusplus +extern "C" { +#endif + struct pipe_context; struct pipe_surface; struct cso_context; @@ -58,4 +62,8 @@ util_blit_pixels(struct blit_state *ctx, float z, uint filter); +#ifdef __cplusplus +} +#endif + #endif diff --git a/src/gallium/auxiliary/util/u_double_list.h b/src/gallium/auxiliary/util/u_double_list.h index 8cb10c9820..d108d92e52 100644 --- a/src/gallium/auxiliary/util/u_double_list.h +++ b/src/gallium/auxiliary/util/u_double_list.h @@ -70,6 +70,14 @@ struct list_head (__list)->prev = (__item); \ } while(0) +#define LIST_REPLACE(__from, __to) \ + do { \ + (__to)->prev = (__from)->prev; \ + (__to)->next = (__from)->next; \ + (__from)->next->prev = (__to); \ + (__from)->prev->next = (__to); \ + } while (0) + #define LIST_DEL(__item) \ do { \ (__item)->prev->next = (__item)->next; \ diff --git a/src/gallium/auxiliary/util/u_gen_mipmap.c b/src/gallium/auxiliary/util/u_gen_mipmap.c index 2fd214d22e..c53c512268 100644 --- a/src/gallium/auxiliary/util/u_gen_mipmap.c +++ b/src/gallium/auxiliary/util/u_gen_mipmap.c @@ -61,6 +61,7 @@ struct gen_mipmap_state struct pipe_depth_stencil_alpha_state depthstencil; struct pipe_rasterizer_state rasterizer; struct pipe_sampler_state sampler; + struct pipe_viewport_state viewport; struct pipe_shader_state vert_shader; struct pipe_shader_state frag_shader; @@ -474,7 +475,9 @@ format_to_type_comps(enum pipe_format pformat, { switch (pformat) { case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_X8R8G8B8_UNORM: case PIPE_FORMAT_B8G8R8A8_UNORM: + case PIPE_FORMAT_B8G8R8X8_UNORM: *datatype = UBYTE; *comps = 4; return; @@ -490,18 +493,21 @@ format_to_type_comps(enum pipe_format pformat, *datatype = USHORT_5_6_5; *comps = 3; return; - case PIPE_FORMAT_U_L8: - case PIPE_FORMAT_U_A8: - case PIPE_FORMAT_U_I8: + case PIPE_FORMAT_L8_UNORM: + case PIPE_FORMAT_A8_UNORM: + case PIPE_FORMAT_I8_UNORM: *datatype = UBYTE; *comps = 1; return; - case PIPE_FORMAT_U_A8_L8: + case PIPE_FORMAT_A8L8_UNORM: *datatype = UBYTE; *comps = 2; return; default: assert(0); + *datatype = UBYTE; + *comps = 0; + break; } } @@ -712,7 +718,7 @@ util_create_gen_mipmap(struct pipe_context *pipe, memset(&ctx->rasterizer, 0, sizeof(ctx->rasterizer)); ctx->rasterizer.front_winding = PIPE_WINDING_CW; ctx->rasterizer.cull_mode = PIPE_WINDING_NONE; - ctx->rasterizer.bypass_clipping = 1; /* bypasses viewport too */ + ctx->rasterizer.bypass_clipping = 1; /*ctx->rasterizer.bypass_vs = 1;*/ /* sampler state */ @@ -723,9 +729,7 @@ util_create_gen_mipmap(struct pipe_context *pipe, ctx->sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NEAREST; ctx->sampler.normalized_coords = 1; - -#if 0 - /* viewport */ + /* viewport state (identity, verts are in wincoords) */ ctx->viewport.scale[0] = 1.0; ctx->viewport.scale[1] = 1.0; ctx->viewport.scale[2] = 1.0; @@ -734,7 +738,6 @@ util_create_gen_mipmap(struct pipe_context *pipe, ctx->viewport.translate[1] = 0.0; ctx->viewport.translate[2] = 0.0; ctx->viewport.translate[3] = 0.0; -#endif /* vertex shader */ { @@ -775,23 +778,23 @@ set_vertex_data(struct gen_mipmap_state *ctx, float width, float height) { void *buf; - ctx->vertices[0][0][0] = -0.5f; /*x*/ - ctx->vertices[0][0][1] = -0.5f; /*y*/ + ctx->vertices[0][0][0] = 0.0f; /*x*/ + ctx->vertices[0][0][1] = 0.0f; /*y*/ ctx->vertices[0][1][0] = 0.0f; /*s*/ ctx->vertices[0][1][1] = 0.0f; /*t*/ - ctx->vertices[1][0][0] = width - 0.5f; /*x*/ - ctx->vertices[1][0][1] = -0.5f; /*y*/ - ctx->vertices[1][1][0] = 1.0f; /*s*/ - ctx->vertices[1][1][1] = 0.0f; /*t*/ + ctx->vertices[1][0][0] = width; + ctx->vertices[1][0][1] = 0.0f; + ctx->vertices[1][1][0] = 1.0f; + ctx->vertices[1][1][1] = 0.0f; - ctx->vertices[2][0][0] = width - 0.5f; - ctx->vertices[2][0][1] = height - 0.5f; + ctx->vertices[2][0][0] = width; + ctx->vertices[2][0][1] = height; ctx->vertices[2][1][0] = 1.0f; ctx->vertices[2][1][1] = 1.0f; - ctx->vertices[3][0][0] = -0.5f; - ctx->vertices[3][0][1] = height - 0.5f; + ctx->vertices[3][0][0] = 0.0f; + ctx->vertices[3][0][1] = height; ctx->vertices[3][1][0] = 0.0f; ctx->vertices[3][1][1] = 1.0f; @@ -825,26 +828,6 @@ util_destroy_gen_mipmap(struct gen_mipmap_state *ctx) } -#if 0 -static void -simple_viewport(struct pipe_context *pipe, uint width, uint height) -{ - struct pipe_viewport_state vp; - - vp.scale[0] = 0.5 * width; - vp.scale[1] = -0.5 * height; - vp.scale[2] = 1.0; - vp.scale[3] = 1.0; - vp.translate[0] = 0.5 * width; - vp.translate[1] = 0.5 * height; - vp.translate[2] = 0.0; - vp.translate[3] = 0.0; - - pipe->set_viewport_state(pipe, &vp); -} -#endif - - /** * Generate mipmap images. It's assumed all needed texture memory is * already allocated. @@ -880,17 +863,18 @@ util_gen_mipmap(struct gen_mipmap_state *ctx, cso_save_samplers(ctx->cso); cso_save_sampler_textures(ctx->cso); cso_save_framebuffer(ctx->cso); + cso_save_fragment_shader(ctx->cso); + cso_save_vertex_shader(ctx->cso); + cso_save_viewport(ctx->cso); /* bind our state */ cso_set_blend(ctx->cso, &ctx->blend); cso_set_depth_stencil_alpha(ctx->cso, &ctx->depthstencil); cso_set_rasterizer(ctx->cso, &ctx->rasterizer); + cso_set_viewport(ctx->cso, &ctx->viewport); - pipe->bind_vs_state(pipe, ctx->vs); - pipe->bind_fs_state(pipe, ctx->fs); -#if 0 - pipe->set_viewport_state(pipe, &ctx->viewport); -#endif + cso_set_fragment_shader_handle(ctx->cso, ctx->fs); + cso_set_vertex_shader_handle(ctx->cso, ctx->vs); /* init framebuffer state */ memset(&fb, 0, sizeof(fb)); @@ -926,11 +910,8 @@ util_gen_mipmap(struct gen_mipmap_state *ctx, ctx->sampler.lod_bias = (float) srcLevel; cso_single_sampler(ctx->cso, 0, &ctx->sampler); cso_single_sampler_done(ctx->cso); -#if 0 - simple_viewport(pipe, pt->width[dstLevel], pt->height[dstLevel]); -#endif - pipe->set_sampler_textures(pipe, 1, &pt); + cso_set_sampler_textures(ctx->cso, 1, &pt); /* quad coords in window coords (bypassing clipping, viewport mapping) */ set_vertex_data(ctx, @@ -954,4 +935,7 @@ util_gen_mipmap(struct gen_mipmap_state *ctx, cso_restore_samplers(ctx->cso); cso_restore_sampler_textures(ctx->cso); cso_restore_framebuffer(ctx->cso); + cso_restore_fragment_shader(ctx->cso); + cso_restore_vertex_shader(ctx->cso); + cso_restore_viewport(ctx->cso); } diff --git a/src/gallium/auxiliary/util/u_gen_mipmap.h b/src/gallium/auxiliary/util/u_gen_mipmap.h index bd9af54fb7..3277024f07 100644 --- a/src/gallium/auxiliary/util/u_gen_mipmap.h +++ b/src/gallium/auxiliary/util/u_gen_mipmap.h @@ -31,6 +31,11 @@ #include "pipe/p_state.h" +#ifdef __cplusplus +extern "C" { +#endif + + struct pipe_context; struct pipe_texture; struct cso_context; @@ -52,4 +57,9 @@ util_gen_mipmap(struct gen_mipmap_state *ctx, struct pipe_texture *pt, uint face, uint baseLevel, uint lastLevel, uint filter); + +#ifdef __cplusplus +} +#endif + #endif diff --git a/src/gallium/auxiliary/util/u_pack_color.h b/src/gallium/auxiliary/util/u_pack_color.h index cd13823985..0b917c005f 100644 --- a/src/gallium/auxiliary/util/u_pack_color.h +++ b/src/gallium/auxiliary/util/u_pack_color.h @@ -40,6 +40,94 @@ /** + * Pack ubyte R,G,B,A into dest pixel. + */ +static INLINE void +util_pack_color_ub(ubyte r, ubyte g, ubyte b, ubyte a, + enum pipe_format format, void *dest) +{ + switch (format) { + case PIPE_FORMAT_R8G8B8A8_UNORM: + { + uint *d = (uint *) dest; + *d = (r << 24) | (g << 16) | (b << 8) | a; + } + return; + case PIPE_FORMAT_R8G8B8X8_UNORM: + { + uint *d = (uint *) dest; + *d = (r << 24) | (g << 16) | (b << 8) | 0xff; + } + return; + case PIPE_FORMAT_A8R8G8B8_UNORM: + { + uint *d = (uint *) dest; + *d = (a << 24) | (r << 16) | (g << 8) | b; + } + return; + case PIPE_FORMAT_X8R8G8B8_UNORM: + { + uint *d = (uint *) dest; + *d = (0xff << 24) | (r << 16) | (g << 8) | b; + } + return; + case PIPE_FORMAT_B8G8R8A8_UNORM: + { + uint *d = (uint *) dest; + *d = (b << 24) | (g << 16) | (r << 8) | a; + } + return; + case PIPE_FORMAT_B8G8R8X8_UNORM: + { + uint *d = (uint *) dest; + *d = (b << 24) | (g << 16) | (r << 8) | 0xff; + } + return; + case PIPE_FORMAT_R5G6B5_UNORM: + { + ushort *d = (ushort *) dest; + *d = ((r & 0xf8) << 8) | ((g & 0xfc) << 3) | (b >> 3); + } + return; + case PIPE_FORMAT_A1R5G5B5_UNORM: + { + ushort *d = (ushort *) dest; + *d = ((a & 0x80) << 8) | ((r & 0xf8) << 7) | ((g & 0xf8) << 2) | (b >> 3); + } + return; + case PIPE_FORMAT_A4R4G4B4_UNORM: + { + ushort *d = (ushort *) dest; + *d = ((a & 0xf0) << 8) | ((r & 0xf0) << 4) | ((g & 0xf0) << 0) | (b >> 4); + } + return; + case PIPE_FORMAT_R32G32B32A32_FLOAT: + { + float *d = (float *) dest; + d[0] = (float)r / 255.0f; + d[1] = (float)g / 255.0f; + d[2] = (float)b / 255.0f; + d[3] = (float)a / 255.0f; + } + return; + case PIPE_FORMAT_R32G32B32_FLOAT: + { + float *d = (float *) dest; + d[0] = (float)r / 255.0f; + d[1] = (float)g / 255.0f; + d[2] = (float)b / 255.0f; + } + return; + + /* XXX lots more cases to add */ + default: + debug_print_format("gallium: unhandled format in util_pack_color_ub()", format); + assert(0); + } +} + + +/** * Note rgba outside [0,1] will be clamped for int pixel formats. */ static INLINE void @@ -62,24 +150,54 @@ util_pack_color(const float rgba[4], enum pipe_format format, void *dest) *d = (r << 24) | (g << 16) | (b << 8) | a; } return; + case PIPE_FORMAT_R8G8B8X8_UNORM: + { + uint *d = (uint *) dest; + *d = (r << 24) | (g << 16) | (b << 8) | 0xff; + } + return; case PIPE_FORMAT_A8R8G8B8_UNORM: { uint *d = (uint *) dest; *d = (a << 24) | (r << 16) | (g << 8) | b; } return; + case PIPE_FORMAT_X8R8G8B8_UNORM: + { + uint *d = (uint *) dest; + *d = (0xff << 24) | (r << 16) | (g << 8) | b; + } + return; case PIPE_FORMAT_B8G8R8A8_UNORM: { uint *d = (uint *) dest; *d = (b << 24) | (g << 16) | (r << 8) | a; } return; + case PIPE_FORMAT_B8G8R8X8_UNORM: + { + uint *d = (uint *) dest; + *d = (b << 24) | (g << 16) | (r << 8) | 0xff; + } + return; case PIPE_FORMAT_R5G6B5_UNORM: { ushort *d = (ushort *) dest; *d = ((r & 0xf8) << 8) | ((g & 0xfc) << 3) | (b >> 3); } return; + case PIPE_FORMAT_A1R5G5B5_UNORM: + { + ushort *d = (ushort *) dest; + *d = ((a & 0x80) << 8) | ((r & 0xf8) << 7) | ((g & 0xf8) << 2) | (b >> 3); + } + return; + case PIPE_FORMAT_A4R4G4B4_UNORM: + { + ushort *d = (ushort *) dest; + *d = ((a & 0xf0) << 8) | ((r & 0xf0) << 4) | ((g & 0xf0) << 0) | (b >> 4); + } + return; case PIPE_FORMAT_R32G32B32A32_FLOAT: { float *d = (float *) dest; @@ -99,7 +217,8 @@ util_pack_color(const float rgba[4], enum pipe_format format, void *dest) return; /* XXX lots more cases to add */ default: - debug_printf("gallium: unhandled format in util_pack_color()"); + debug_print_format("gallium: unhandled format in util_pack_color()", format); + assert(0); } } @@ -120,11 +239,14 @@ util_pack_z(enum pipe_format format, double z) else return (uint) (z * 0xffffffff); case PIPE_FORMAT_S8Z24_UNORM: + case PIPE_FORMAT_X8Z24_UNORM: return (uint) (z * 0xffffff); case PIPE_FORMAT_Z24S8_UNORM: + case PIPE_FORMAT_Z24X8_UNORM: return ((uint) (z * 0xffffff)) << 8; default: - debug_printf("gallium: unhandled fomrat in util_pack_z()"); + debug_print_format("gallium: unhandled format in util_pack_z()", format); + assert(0); return 0; } } diff --git a/src/gallium/auxiliary/util/u_snprintf.c b/src/gallium/auxiliary/util/u_snprintf.c index 48426abcb7..c4f4bbd30c 100644 --- a/src/gallium/auxiliary/util/u_snprintf.c +++ b/src/gallium/auxiliary/util/u_snprintf.c @@ -166,8 +166,8 @@ #include <config.h> #else #ifdef WIN32 -#define vsnprintf rpl_vsnprintf -#define snprintf rpl_snprintf +#define vsnprintf util_vsnprintf +#define snprintf util_snprintf #define HAVE_VSNPRINTF 0 #define HAVE_SNPRINTF 0 #define HAVE_VASPRINTF 1 /* not needed */ @@ -445,7 +445,7 @@ static UINTMAX_T myround(LDOUBLE); static LDOUBLE mypow10(int); int -rpl_vsnprintf(char *str, size_t size, const char *format, va_list args) +util_vsnprintf(char *str, size_t size, const char *format, va_list args) { LDOUBLE fvalue; INTMAX_T value; @@ -1404,7 +1404,7 @@ mymemcpy(void *dst, void *src, size_t len) #endif /* NEED_MYMEMCPY */ int -rpl_vasprintf(char **ret, const char *format, va_list ap) +util_vasprintf(char **ret, const char *format, va_list ap) { size_t size; int len; @@ -1422,10 +1422,10 @@ rpl_vasprintf(char **ret, const char *format, va_list ap) #if !HAVE_SNPRINTF #if HAVE_STDARG_H int -rpl_snprintf(char *str, size_t size, const char *format, ...) +util_snprintf(char *str, size_t size, const char *format, ...) #else int -rpl_snprintf(va_alist) va_dcl +util_snprintf(va_alist) va_dcl #endif /* HAVE_STDARG_H */ { #if !HAVE_STDARG_H @@ -1449,10 +1449,10 @@ rpl_snprintf(va_alist) va_dcl #if !HAVE_ASPRINTF #if HAVE_STDARG_H int -rpl_asprintf(char **ret, const char *format, ...) +util_asprintf(char **ret, const char *format, ...) #else int -rpl_asprintf(va_alist) va_dcl +util_asprintf(va_alist) va_dcl #endif /* HAVE_STDARG_H */ { #if !HAVE_STDARG_H diff --git a/src/gallium/auxiliary/util/u_string.h b/src/gallium/auxiliary/util/u_string.h new file mode 100644 index 0000000000..b99d4e8021 --- /dev/null +++ b/src/gallium/auxiliary/util/u_string.h @@ -0,0 +1,63 @@ +/************************************************************************** + * + * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +/** + * @file + * Platform independent functions for string manipulation. + * + * @author Jose Fonseca <jrfonseca@tungstengraphics.com> + */ + +#ifndef U_STRING_H_ +#define U_STRING_H_ + +#ifndef WIN32 +#include <stdio.h> +#endif +#include <stddef.h> +#include <stdarg.h> + + +#ifdef __cplusplus +extern "C" { +#endif + + +#ifdef WIN32 +int util_vsnprintf(char *, size_t, const char *, va_list); +int util_snprintf(char *str, size_t size, const char *format, ...); +#else +#define util_vsnprintf vsnprintf +#define util_snprintf snprintf +#endif + + +#ifdef __cplusplus +} +#endif + +#endif /* U_STRING_H_ */ diff --git a/src/gallium/auxiliary/util/u_time.c b/src/gallium/auxiliary/util/u_time.c new file mode 100644 index 0000000000..9b97050d51 --- /dev/null +++ b/src/gallium/auxiliary/util/u_time.c @@ -0,0 +1,162 @@ +/************************************************************************** + * + * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +/** + * @file + * OS independent time-manipulation functions. + * + * @author Jose Fonseca <jrfonseca@tungstengraphics.com> + */ + + +#include "util/u_time.h" + +#if defined(PIPE_OS_LINUX) +#include <sys/time.h> +#elif defined(PIPE_SUBSYSTEM_WINDOWS_DISPLAY) +#include <windows.h> +#include <winddi.h> +#elif defined(PIPE_SUBSYSTEM_WINDOWS_USER) +#include <windows.h> +#else +#error Unsupported OS +#endif + + +#if defined(PIPE_OS_WINDOWS) +static LONGLONG frequency = 0; +#if !defined(PIPE_SUBSYSTEM_WINDOWS_DISPLAY) +#define EngQueryPerformanceFrequency(p) QueryPerformanceFrequency((LARGE_INTEGER*)(p)) +#define EngQueryPerformanceCounter(p) QueryPerformanceCounter((LARGE_INTEGER*)(p)) +#endif +#endif + + +void +util_time_get(struct util_time *t) +{ +#if defined(PIPE_OS_LINUX) + gettimeofday(&t->tv, NULL); +#elif defined(PIPE_OS_WINDOWS) + EngQueryPerformanceCounter(&t->counter); +#endif +} + + +void +util_time_add(const struct util_time *t1, + int64_t usecs, + struct util_time *t2) +{ +#if defined(PIPE_OS_LINUX) + t2->tv.tv_sec = t1->tv.tv_sec + usecs / 1000000; + t2->tv.tv_usec = t1->tv.tv_usec + usecs % 1000000; +#elif defined(PIPE_OS_WINDOWS) + if(!frequency) + EngQueryPerformanceFrequency(&frequency); + t2->counter = t1->counter + (usecs * frequency + 999999LL)/1000000LL; +#endif +} + + +int64_t +util_time_diff(const struct util_time *t1, + const struct util_time *t2) +{ +#if defined(PIPE_OS_LINUX) + return (t2->tv.tv_usec - t1->tv.tv_usec) + + (t2->tv.tv_sec - t1->tv.tv_sec)*1000000; +#elif defined(PIPE_OS_WINDOWS) + if(!frequency) + EngQueryPerformanceFrequency(&frequency); + return (t2->counter - t1->counter)*1000000LL/frequency; +#endif +} + + +/** + * Compare two time values. + * + * Not publicly available because it does not take in account wrap-arounds. + * Use util_time_timeout instead. + */ +static INLINE int +util_time_compare(const struct util_time *t1, + const struct util_time *t2) +{ +#if defined(PIPE_OS_LINUX) + if (t1->tv.tv_sec < t2->tv.tv_sec) + return -1; + else if(t1->tv.tv_sec > t2->tv.tv_sec) + return 1; + else if (t1->tv.tv_usec < t2->tv.tv_usec) + return -1; + else if(t1->tv.tv_usec > t2->tv.tv_usec) + return 1; + else + return 0; +#elif defined(PIPE_OS_WINDOWS) + if (t1->counter < t2->counter) + return -1; + else if(t1->counter > t2->counter) + return 1; + else + return 0; +#endif +} + + +boolean +util_time_timeout(const struct util_time *start, + const struct util_time *end, + const struct util_time *curr) +{ + if(util_time_compare(start, end) <= 0) + return !(util_time_compare(start, curr) <= 0 && util_time_compare(curr, end) < 0); + else + return !(util_time_compare(start, curr) <= 0 || util_time_compare(curr, end) < 0); +} + + +#if defined(PIPE_OS_WINDOWS) +void util_time_sleep(unsigned usecs) +{ + LONGLONG start, curr, end; + + EngQueryPerformanceCounter(&start); + + if(!frequency) + EngQueryPerformanceFrequency(&frequency); + + end = start + (usecs * frequency + 999999LL)/1000000LL; + + do { + EngQueryPerformanceCounter(&curr); + } while(start <= curr && curr < end || + end < start && (curr < end || start <= curr)); +} +#endif diff --git a/src/gallium/auxiliary/util/u_time.h b/src/gallium/auxiliary/util/u_time.h new file mode 100644 index 0000000000..48ec7a4a96 --- /dev/null +++ b/src/gallium/auxiliary/util/u_time.h @@ -0,0 +1,101 @@ +/************************************************************************** + * + * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +/** + * @file + * OS independent time-manipulation functions. + * + * @author Jose Fonseca <jrfonseca@tungstengraphics.com> + */ + +#ifndef U_TIME_H_ +#define U_TIME_H_ + + +#include "pipe/p_config.h" + +#if defined(PIPE_OS_LINUX) +#include <time.h> /* timeval */ +#include <unistd.h> /* usleep */ +#endif + +#include "pipe/p_compiler.h" + + +#ifdef __cplusplus +extern "C" { +#endif + + +/** + * Time abstraction. + * + * Do not access this structure directly. Use the provided function instead. + */ +struct util_time +{ +#if defined(PIPE_OS_LINUX) + struct timeval tv; +#else + long long counter; +#endif +}; + + +void +util_time_get(struct util_time *t); + +void +util_time_add(const struct util_time *t1, + int64_t usecs, + struct util_time *t2); + +int64_t +util_time_diff(const struct util_time *t1, + const struct util_time *t2); + +/** + * Returns non-zero when the timeout expires. + */ +boolean +util_time_timeout(const struct util_time *start, + const struct util_time *end, + const struct util_time *curr); + +#if defined(PIPE_OS_LINUX) +#define util_time_sleep usleep +#else +void +util_time_sleep(unsigned usecs); +#endif + + +#ifdef __cplusplus +} +#endif + +#endif /* U_TIME_H_ */ diff --git a/src/gallium/drivers/cell/ppu/cell_context.h b/src/gallium/drivers/cell/ppu/cell_context.h index 7f656a9744..8df41c1d4c 100644 --- a/src/gallium/drivers/cell/ppu/cell_context.h +++ b/src/gallium/drivers/cell/ppu/cell_context.h @@ -104,7 +104,9 @@ struct cell_context uint num_textures; struct pipe_viewport_state viewport; struct pipe_vertex_buffer vertex_buffer[PIPE_MAX_ATTRIBS]; + uint num_vertex_buffers; struct pipe_vertex_element vertex_element[PIPE_MAX_ATTRIBS]; + uint num_vertex_elements; ubyte *cbuf_map[PIPE_MAX_COLOR_BUFS]; ubyte *zsbuf_map; diff --git a/src/gallium/drivers/cell/ppu/cell_draw_arrays.c b/src/gallium/drivers/cell/ppu/cell_draw_arrays.c index b896252f81..6e08cf6fe8 100644 --- a/src/gallium/drivers/cell/ppu/cell_draw_arrays.c +++ b/src/gallium/drivers/cell/ppu/cell_draw_arrays.c @@ -101,17 +101,6 @@ cell_draw_elements(struct pipe_context *pipe, struct draw_context *draw = sp->draw; unsigned i; - /* first, check that the primitive is not malformed. It is the - * state tracker's responsibility to do send only correctly formed - * primitives down. It currently isn't doing that though... - */ -#if 1 - count = draw_trim_prim( mode, count ); -#else - if (!draw_validate_prim( mode, count )) - assert(0); -#endif - if (sp->dirty) cell_update_derived( sp ); @@ -123,14 +112,12 @@ cell_draw_elements(struct pipe_context *pipe, /* * Map vertex buffers */ - for (i = 0; i < PIPE_MAX_ATTRIBS; i++) { - if (sp->vertex_buffer[i].buffer) { - void *buf = pipe->winsys->buffer_map(pipe->winsys, - sp->vertex_buffer[i].buffer, - PIPE_BUFFER_USAGE_CPU_READ); - cell_flush_buffer_range(sp, buf, sp->vertex_buffer[i].buffer->size); - draw_set_mapped_vertex_buffer(draw, i, buf); - } + for (i = 0; i < sp->num_vertex_buffers; i++) { + void *buf = pipe->winsys->buffer_map(pipe->winsys, + sp->vertex_buffer[i].buffer, + PIPE_BUFFER_USAGE_CPU_READ); + cell_flush_buffer_range(sp, buf, sp->vertex_buffer[i].buffer->size); + draw_set_mapped_vertex_buffer(draw, i, buf); } /* Map index buffer, if present */ if (indexBuffer) { @@ -151,11 +138,9 @@ cell_draw_elements(struct pipe_context *pipe, /* * unmap vertex/index buffers - will cause draw module to flush */ - for (i = 0; i < PIPE_MAX_ATTRIBS; i++) { - if (sp->vertex_buffer[i].buffer) { - draw_set_mapped_vertex_buffer(draw, i, NULL); - pipe->winsys->buffer_unmap(pipe->winsys, sp->vertex_buffer[i].buffer); - } + for (i = 0; i < sp->num_vertex_buffers; i++) { + draw_set_mapped_vertex_buffer(draw, i, NULL); + pipe->winsys->buffer_unmap(pipe->winsys, sp->vertex_buffer[i].buffer); } if (indexBuffer) { draw_set_mapped_element_buffer(draw, 0, NULL); diff --git a/src/gallium/drivers/cell/ppu/cell_state_shader.c b/src/gallium/drivers/cell/ppu/cell_state_shader.c index fb2e940348..c3a3fbd066 100644 --- a/src/gallium/drivers/cell/ppu/cell_state_shader.c +++ b/src/gallium/drivers/cell/ppu/cell_state_shader.c @@ -30,33 +30,50 @@ #include "pipe/p_inlines.h" #include "pipe/p_winsys.h" #include "draw/draw_context.h" -#if 0 -#include "pipe/p_shader_tokens.h" -#include "gallivm/gallivm.h" -#include "tgsi/util/tgsi_dump.h" -#include "tgsi/exec/tgsi_sse2.h" -#endif +#include "tgsi/util/tgsi_parse.h" #include "cell_context.h" #include "cell_state.h" + +/** cast wrapper */ +static INLINE struct cell_fragment_shader_state * +cell_fragment_shader_state(void *shader) +{ + return (struct pipe_shader_state *) shader; +} + + +/** cast wrapper */ +static INLINE struct cell_vertex_shader_state * +cell_vertex_shader_state(void *shader) +{ + return (struct pipe_shader_state *) shader; +} + + + static void * cell_create_fs_state(struct pipe_context *pipe, const struct pipe_shader_state *templ) { /*struct cell_context *cell = cell_context(pipe);*/ - struct cell_fragment_shader_state *state; + struct cell_fragment_shader_state *cfs; - state = CALLOC_STRUCT(cell_fragment_shader_state); - if (!state) + cfs = CALLOC_STRUCT(cell_fragment_shader_state); + if (!cfs) return NULL; - state->shader = *templ; + cfs->shader.tokens = tgsi_dup_tokens(templ->tokens); + if (!cfs->shader.tokens) { + FREE(cfs); + return NULL; + } - tgsi_scan_shader(templ->tokens, &state->info); + tgsi_scan_shader(templ->tokens, &cfs->info); - return state; + return cfs; } @@ -65,7 +82,7 @@ cell_bind_fs_state(struct pipe_context *pipe, void *fs) { struct cell_context *cell = cell_context(pipe); - cell->fs = (struct cell_fragment_shader_state *) fs; + cell->fs = cell_fragment_shader_state(fs); cell->dirty |= CELL_NEW_FS; } @@ -74,10 +91,10 @@ cell_bind_fs_state(struct pipe_context *pipe, void *fs) static void cell_delete_fs_state(struct pipe_context *pipe, void *fs) { - struct cell_fragment_shader_state *state = - (struct cell_fragment_shader_state *) fs; + struct cell_fragment_shader_state *cfs = cell_fragment_shader_state(fs); - FREE( state ); + FREE((void *) cfs->shader.tokens); + FREE(cfs); } @@ -86,22 +103,28 @@ cell_create_vs_state(struct pipe_context *pipe, const struct pipe_shader_state *templ) { struct cell_context *cell = cell_context(pipe); - struct cell_vertex_shader_state *state; + struct cell_vertex_shader_state *cvs; - state = CALLOC_STRUCT(cell_vertex_shader_state); - if (!state) + cvs = CALLOC_STRUCT(cell_vertex_shader_state); + if (!cvs) return NULL; - state->shader = *templ; - tgsi_scan_shader(templ->tokens, &state->info); + cvs->shader.tokens = tgsi_dup_tokens(templ->tokens); + if (!cvs->shader.tokens) { + FREE(cvs); + return NULL; + } - state->draw_data = draw_create_vertex_shader(cell->draw, &state->shader); - if (state->draw_data == NULL) { - FREE( state ); + tgsi_scan_shader(templ->tokens, &cvs->info); + + cvs->draw_data = draw_create_vertex_shader(cell->draw, &cvs->shader); + if (cvs->draw_data == NULL) { + FREE( (void *) cvs->shader.tokens ); + FREE( cvs ); return NULL; } - return state; + return cvs; } @@ -110,7 +133,7 @@ cell_bind_vs_state(struct pipe_context *pipe, void *vs) { struct cell_context *cell = cell_context(pipe); - cell->vs = (const struct cell_vertex_shader_state *) vs; + cell->vs = cell_vertex_shader_state(vs); draw_bind_vertex_shader(cell->draw, (cell->vs ? cell->vs->draw_data : NULL)); @@ -123,12 +146,11 @@ static void cell_delete_vs_state(struct pipe_context *pipe, void *vs) { struct cell_context *cell = cell_context(pipe); + struct cell_vertex_shader_state *cvs = cell_vertex_shader_state(vs); - struct cell_vertex_shader_state *state = - (struct cell_vertex_shader_state *) vs; - - draw_delete_vertex_shader(cell->draw, state->draw_data); - FREE( state ); + draw_delete_vertex_shader(cell->draw, cvs->draw_data); + FREE( (void *) cvs->shader.tokens ); + FREE( cvs ); } diff --git a/src/gallium/drivers/cell/ppu/cell_state_vertex.c b/src/gallium/drivers/cell/ppu/cell_state_vertex.c index 6c83b8dc72..114684c2a3 100644 --- a/src/gallium/drivers/cell/ppu/cell_state_vertex.c +++ b/src/gallium/drivers/cell/ppu/cell_state_vertex.c @@ -45,6 +45,7 @@ cell_set_vertex_elements(struct pipe_context *pipe, assert(count <= PIPE_MAX_ATTRIBS); memcpy(cell->vertex_element, elements, count * sizeof(elements[0])); + cell->num_vertex_elements = count; cell->dirty |= CELL_NEW_VERTEX; @@ -62,6 +63,7 @@ cell_set_vertex_buffers(struct pipe_context *pipe, assert(count <= PIPE_MAX_ATTRIBS); memcpy(cell->vertex_buffer, buffers, count * sizeof(buffers[0])); + cell->num_vertex_buffers = count; cell->dirty |= CELL_NEW_VERTEX; diff --git a/src/gallium/drivers/i915simple/i915_context.c b/src/gallium/drivers/i915simple/i915_context.c index fee33d82de..4bef21619c 100644 --- a/src/gallium/drivers/i915simple/i915_context.c +++ b/src/gallium/drivers/i915simple/i915_context.c @@ -65,14 +65,12 @@ i915_draw_elements( struct pipe_context *pipe, /* * Map vertex buffers */ - for (i = 0; i < PIPE_MAX_ATTRIBS; i++) { - if (i915->vertex_buffer[i].buffer) { - void *buf - = pipe->winsys->buffer_map(pipe->winsys, - i915->vertex_buffer[i].buffer, - PIPE_BUFFER_USAGE_CPU_READ); - draw_set_mapped_vertex_buffer(draw, i, buf); - } + for (i = 0; i < i915->num_vertex_buffers; i++) { + void *buf + = pipe->winsys->buffer_map(pipe->winsys, + i915->vertex_buffer[i].buffer, + PIPE_BUFFER_USAGE_CPU_READ); + draw_set_mapped_vertex_buffer(draw, i, buf); } /* Map index buffer, if present */ if (indexBuffer) { @@ -96,11 +94,9 @@ i915_draw_elements( struct pipe_context *pipe, /* * unmap vertex/index buffers */ - for (i = 0; i < PIPE_MAX_ATTRIBS; i++) { - if (i915->vertex_buffer[i].buffer) { - pipe->winsys->buffer_unmap(pipe->winsys, i915->vertex_buffer[i].buffer); - draw_set_mapped_vertex_buffer(draw, i, NULL); - } + for (i = 0; i < i915->num_vertex_buffers; i++) { + pipe->winsys->buffer_unmap(pipe->winsys, i915->vertex_buffer[i].buffer); + draw_set_mapped_vertex_buffer(draw, i, NULL); } if (indexBuffer) { pipe->winsys->buffer_unmap(pipe->winsys, indexBuffer); @@ -146,7 +142,7 @@ struct pipe_context *i915_create_context( struct pipe_screen *screen, */ i915->draw = draw_create(); assert(i915->draw); - if (GETENV("I915_VBUF")) { + if (!GETENV("I915_NO_VBUF")) { draw_set_rasterize_stage(i915->draw, i915_draw_vbuf_stage(i915)); } else { diff --git a/src/gallium/drivers/i915simple/i915_context.h b/src/gallium/drivers/i915simple/i915_context.h index 8e707ea574..38e6a34884 100644 --- a/src/gallium/drivers/i915simple/i915_context.h +++ b/src/gallium/drivers/i915simple/i915_context.h @@ -238,6 +238,8 @@ struct i915_context unsigned num_samplers;
unsigned num_textures;
+ unsigned num_vertex_elements;
+ unsigned num_vertex_buffers;
unsigned *batch_start;
diff --git a/src/gallium/drivers/i915simple/i915_debug_fp.c b/src/gallium/drivers/i915simple/i915_debug_fp.c index 37a3508fe1..c024a051a5 100644 --- a/src/gallium/drivers/i915simple/i915_debug_fp.c +++ b/src/gallium/drivers/i915simple/i915_debug_fp.c @@ -333,12 +333,11 @@ void i915_disassemble_program(struct debug_stream *stream, const unsigned * program, unsigned sz) { - unsigned size = program[0] & 0x1ff; unsigned i; PRINTF(stream, "\t\tBEGIN\n"); - assert(size + 2 == sz); + assert((program[0] & 0x1ff) + 2 == sz); program++; for (i = 1; i < sz; i += 3, program += 3) { diff --git a/src/gallium/drivers/i915simple/i915_fpc_translate.c b/src/gallium/drivers/i915simple/i915_fpc_translate.c index 7b4fca5db1..3ccf74c72c 100644 --- a/src/gallium/drivers/i915simple/i915_fpc_translate.c +++ b/src/gallium/drivers/i915simple/i915_fpc_translate.c @@ -33,6 +33,7 @@ #include "i915_fpc.h" #include "pipe/p_shader_tokens.h" +#include "util/u_string.h" #include "tgsi/util/tgsi_parse.h" #include "tgsi/util/tgsi_dump.h" @@ -122,7 +123,7 @@ i915_program_error(struct i915_fp_compile *p, const char *msg, ...) debug_printf("i915_program_error: "); va_start( args, msg ); - vsprintf( buffer, msg, args ); + util_vsnprintf( buffer, sizeof(buffer), msg, args ); va_end( args ); debug_printf(buffer); debug_printf("\n"); diff --git a/src/gallium/drivers/i915simple/i915_prim_emit.c b/src/gallium/drivers/i915simple/i915_prim_emit.c index b6fb0a6d88..9ffa460138 100644 --- a/src/gallium/drivers/i915simple/i915_prim_emit.c +++ b/src/gallium/drivers/i915simple/i915_prim_emit.c @@ -26,7 +26,7 @@ **************************************************************************/ -#include "draw/draw_private.h" +#include "draw/draw_pipe.h" #include "pipe/p_util.h" #include "i915_context.h" @@ -78,9 +78,6 @@ emit_hw_vertex( struct i915_context *i915, const uint j = vinfo->src_index[i]; const float *attrib = vertex->data[j]; switch (vinfo->emit[i]) { - case EMIT_OMIT: - /* no-op */ - break; case EMIT_1F: OUT_BATCH( fui(attrib[0]) ); count++; diff --git a/src/gallium/drivers/i915simple/i915_screen.c b/src/gallium/drivers/i915simple/i915_screen.c index 839b98c0ce..646cfd921d 100644 --- a/src/gallium/drivers/i915simple/i915_screen.c +++ b/src/gallium/drivers/i915simple/i915_screen.c @@ -28,6 +28,7 @@ #include "pipe/p_util.h" #include "pipe/p_winsys.h" +#include "util/u_string.h" #include "i915_reg.h" #include "i915_context.h" @@ -78,7 +79,7 @@ i915_get_name( struct pipe_screen *pscreen ) break; } - sprintf(buffer, "i915 (chipset: %s)", chipset); + util_snprintf(buffer, sizeof(buffer), "i915 (chipset: %s)", chipset); return buffer; } @@ -153,10 +154,10 @@ i915_is_format_supported( struct pipe_screen *screen, PIPE_FORMAT_R8G8B8A8_UNORM, PIPE_FORMAT_A8R8G8B8_UNORM, PIPE_FORMAT_R5G6B5_UNORM, - PIPE_FORMAT_U_L8, - PIPE_FORMAT_U_A8, - PIPE_FORMAT_U_I8, - PIPE_FORMAT_U_A8_L8, + PIPE_FORMAT_L8_UNORM, + PIPE_FORMAT_A8_UNORM, + PIPE_FORMAT_I8_UNORM, + PIPE_FORMAT_A8L8_UNORM, PIPE_FORMAT_YCBCR, PIPE_FORMAT_YCBCR_REV, PIPE_FORMAT_S8Z24_UNORM, @@ -181,6 +182,7 @@ i915_is_format_supported( struct pipe_screen *screen, break; default: assert(0); + return FALSE; } for (i = 0; list[i] != PIPE_FORMAT_NONE; i++) { diff --git a/src/gallium/drivers/i915simple/i915_state.c b/src/gallium/drivers/i915simple/i915_state.c index 4404bc4590..4adeb37e86 100644 --- a/src/gallium/drivers/i915simple/i915_state.c +++ b/src/gallium/drivers/i915simple/i915_state.c @@ -33,6 +33,7 @@ #include "pipe/p_winsys.h" #include "pipe/p_util.h" #include "pipe/p_inlines.h" +#include "tgsi/util/tgsi_parse.h" #include "i915_context.h" #include "i915_reg.h" @@ -436,7 +437,7 @@ i915_create_fs_state(struct pipe_context *pipe, if (!ifs) return NULL; - ifs->state = *templ; + ifs->state.tokens = tgsi_dup_tokens(templ->tokens); tgsi_scan_shader(templ->tokens, &ifs->info); @@ -465,6 +466,8 @@ void i915_delete_fs_state(struct pipe_context *pipe, void *shader) FREE(ifs->program); ifs->program_len = 0; + FREE(ifs->state.tokens); + FREE(ifs); } @@ -549,6 +552,9 @@ static void i915_set_sampler_textures(struct pipe_context *pipe, !memcmp(i915->texture, texture, num * sizeof(struct pipe_texture *))) return; + /* Fixes wrong texture in texobj with VBUF */ + draw_flush(i915->draw); + for (i = 0; i < num; i++) pipe_texture_reference((struct pipe_texture **) &i915->texture[i], texture[i]); @@ -694,6 +700,8 @@ static void i915_set_vertex_buffers(struct pipe_context *pipe, struct i915_context *i915 = i915_context(pipe); memcpy(i915->vertex_buffer, buffers, count * sizeof(buffers[0])); + i915->num_vertex_buffers = count; + /* pass-through to draw module */ draw_set_vertex_buffers(i915->draw, count, buffers); } @@ -703,6 +711,7 @@ static void i915_set_vertex_elements(struct pipe_context *pipe, const struct pipe_vertex_element *elements) { struct i915_context *i915 = i915_context(pipe); + i915->num_vertex_elements = count; /* pass-through to draw module */ draw_set_vertex_elements(i915->draw, count, elements); } diff --git a/src/gallium/drivers/i915simple/i915_state_sampler.c b/src/gallium/drivers/i915simple/i915_state_sampler.c index 84f6529a3a..982eec4a1b 100644 --- a/src/gallium/drivers/i915simple/i915_state_sampler.c +++ b/src/gallium/drivers/i915simple/i915_state_sampler.c @@ -131,13 +131,13 @@ static uint translate_texture_format(enum pipe_format pipeFormat) { switch (pipeFormat) { - case PIPE_FORMAT_U_L8: + case PIPE_FORMAT_L8_UNORM: return MAPSURF_8BIT | MT_8BIT_L8; - case PIPE_FORMAT_U_I8: + case PIPE_FORMAT_I8_UNORM: return MAPSURF_8BIT | MT_8BIT_I8; - case PIPE_FORMAT_U_A8: + case PIPE_FORMAT_A8_UNORM: return MAPSURF_8BIT | MT_8BIT_A8; - case PIPE_FORMAT_U_A8_L8: + case PIPE_FORMAT_A8L8_UNORM: return MAPSURF_16BIT | MT_16BIT_AY88; case PIPE_FORMAT_R5G6B5_UNORM: return MAPSURF_16BIT | MT_16BIT_RGB565; diff --git a/src/gallium/drivers/i965simple/brw_screen.c b/src/gallium/drivers/i965simple/brw_screen.c index 5be369fe52..b700f7e4f5 100644 --- a/src/gallium/drivers/i965simple/brw_screen.c +++ b/src/gallium/drivers/i965simple/brw_screen.c @@ -28,6 +28,7 @@ #include "pipe/p_util.h" #include "pipe/p_winsys.h" +#include "util/u_string.h" #include "brw_context.h" #include "brw_screen.h" @@ -66,7 +67,7 @@ brw_get_name( struct pipe_screen *screen ) break; } - sprintf(buffer, "i965 (chipset: %s)", chipset); + util_snprintf(buffer, sizeof(buffer), "i965 (chipset: %s)", chipset); return buffer; } @@ -140,13 +141,13 @@ brw_is_format_supported( struct pipe_screen *screen, #if 0 /* XXX: This is broken -- rewrite if still needed. */ static const unsigned tex_supported[] = { - PIPE_FORMAT_U_R8_G8_B8_A8, - PIPE_FORMAT_U_A8_R8_G8_B8, - PIPE_FORMAT_U_R5_G6_B5, - PIPE_FORMAT_U_L8, - PIPE_FORMAT_U_A8, - PIPE_FORMAT_U_I8, - PIPE_FORMAT_U_L8_A8, + PIPE_FORMAT_R8G8B8A8_UNORM, + PIPE_FORMAT_A8R8G8B8_UNORM, + PIPE_FORMAT_R5G6B5_UNORM, + PIPE_FORMAT_L8_UNORM, + PIPE_FORMAT_A8_UNORM, + PIPE_FORMAT_I8_UNORM, + PIPE_FORMAT_L8A8_UNORM, PIPE_FORMAT_YCBCR, PIPE_FORMAT_YCBCR_REV, PIPE_FORMAT_S8_Z24, @@ -156,16 +157,16 @@ brw_is_format_supported( struct pipe_screen *screen, /* Actually a lot more than this - add later: */ static const unsigned render_supported[] = { - PIPE_FORMAT_U_A8_R8_G8_B8, - PIPE_FORMAT_U_R5_G6_B5, + PIPE_FORMAT_A8R8G8B8_UNORM, + PIPE_FORMAT_R5G6B5_UNORM, }; /* */ static const unsigned z_stencil_supported[] = { - PIPE_FORMAT_U_Z16, - PIPE_FORMAT_U_Z32, - PIPE_FORMAT_S8_Z24, + PIPE_FORMAT_Z16_UNORM, + PIPE_FORMAT_Z32_UNORM, + PIPE_FORMAT_S8Z24_UNORM, }; switch (type) { diff --git a/src/gallium/drivers/i965simple/brw_state.c b/src/gallium/drivers/i965simple/brw_state.c index 376f1487b2..ac243b7e4f 100644 --- a/src/gallium/drivers/i965simple/brw_state.c +++ b/src/gallium/drivers/i965simple/brw_state.c @@ -35,6 +35,7 @@ #include "pipe/p_inlines.h"
#include "pipe/p_shader_tokens.h"
#include "tgsi/util/tgsi_dump.h"
+#include "tgsi/util/tgsi_parse.h"
#include "brw_context.h"
#include "brw_defines.h"
@@ -182,9 +183,7 @@ static void * brw_create_fs_state(struct pipe_context *pipe, {
struct brw_fragment_program *brw_fp = CALLOC_STRUCT(brw_fragment_program);
- /* XXX: Do I have to duplicate the tokens as well??
- */
- brw_fp->program = *shader;
+ brw_fp->program.tokens = tgsi_dup_tokens(shader->tokens);
brw_fp->id = brw_context(pipe)->program_id++;
tgsi_scan_shader(shader->tokens, &brw_fp->info);
@@ -210,7 +209,10 @@ static void brw_bind_fs_state(struct pipe_context *pipe, void *shader) static void brw_delete_fs_state(struct pipe_context *pipe, void *shader)
{
- FREE(shader);
+ struct brw_fragment_program *brw_fp = (struct brw_fragment_program *) shader;
+
+ FREE((void *) brw_fp->program.tokens);
+ FREE(brw_fp);
}
@@ -223,9 +225,7 @@ static void *brw_create_vs_state(struct pipe_context *pipe, {
struct brw_vertex_program *brw_vp = CALLOC_STRUCT(brw_vertex_program);
- /* XXX: Do I have to duplicate the tokens as well??
- */
- brw_vp->program = *shader;
+ brw_vp->program.tokens = tgsi_dup_tokens(shader->tokens);
brw_vp->id = brw_context(pipe)->program_id++;
tgsi_scan_shader(shader->tokens, &brw_vp->info);
@@ -251,7 +251,10 @@ static void brw_bind_vs_state(struct pipe_context *pipe, void *vs) static void brw_delete_vs_state(struct pipe_context *pipe, void *shader)
{
- FREE(shader);
+ struct brw_vertex_program *brw_vp = (struct brw_vertex_program *) shader;
+
+ FREE((void *) brw_vp->program.tokens);
+ FREE(brw_vp);
}
diff --git a/src/gallium/drivers/i965simple/brw_wm_surface_state.c b/src/gallium/drivers/i965simple/brw_wm_surface_state.c index 853c743ccf..69e56dc8bd 100644 --- a/src/gallium/drivers/i965simple/brw_wm_surface_state.c +++ b/src/gallium/drivers/i965simple/brw_wm_surface_state.c @@ -57,16 +57,16 @@ static unsigned translate_tex_target( enum pipe_texture_target target ) static unsigned translate_tex_format( enum pipe_format pipe_format ) { switch( pipe_format ) { - case PIPE_FORMAT_U_L8: + case PIPE_FORMAT_L8_UNORM: return BRW_SURFACEFORMAT_L8_UNORM; - case PIPE_FORMAT_U_I8: + case PIPE_FORMAT_I8_UNORM: return BRW_SURFACEFORMAT_I8_UNORM; - case PIPE_FORMAT_U_A8: + case PIPE_FORMAT_A8_UNORM: return BRW_SURFACEFORMAT_A8_UNORM; - case PIPE_FORMAT_U_A8_L8: + case PIPE_FORMAT_A8L8_UNORM: return BRW_SURFACEFORMAT_L8A8_UNORM; case PIPE_FORMAT_R8G8B8_UNORM: diff --git a/src/gallium/drivers/softpipe/Makefile b/src/gallium/drivers/softpipe/Makefile index f32db35d58..120bdfd9dd 100644 --- a/src/gallium/drivers/softpipe/Makefile +++ b/src/gallium/drivers/softpipe/Makefile @@ -17,7 +17,6 @@ C_SOURCES = \ sp_quad.c \ sp_quad_alpha_test.c \ sp_quad_blend.c \ - sp_quad_bufloop.c \ sp_quad_colormask.c \ sp_quad_coverage.c \ sp_quad_depth_test.c \ @@ -28,6 +27,7 @@ C_SOURCES = \ sp_quad_stencil.c \ sp_quad_stipple.c \ sp_screen.c \ + sp_setup.c \ sp_state_blend.c \ sp_state_clip.c \ sp_state_derived.c \ diff --git a/src/gallium/drivers/softpipe/SConscript b/src/gallium/drivers/softpipe/SConscript index 88c21ee3b0..c1f7daa8ab 100644 --- a/src/gallium/drivers/softpipe/SConscript +++ b/src/gallium/drivers/softpipe/SConscript @@ -14,9 +14,9 @@ softpipe = env.ConvenienceLibrary( 'sp_flush.c', 'sp_prim_setup.c', 'sp_prim_vbuf.c', + 'sp_setup.c', 'sp_quad_alpha_test.c', 'sp_quad_blend.c', - 'sp_quad_bufloop.c', 'sp_quad.c', 'sp_quad_colormask.c', 'sp_quad_coverage.c', diff --git a/src/gallium/drivers/softpipe/sp_clear.c b/src/gallium/drivers/softpipe/sp_clear.c index 39aed151c7..1236706891 100644 --- a/src/gallium/drivers/softpipe/sp_clear.c +++ b/src/gallium/drivers/softpipe/sp_clear.c @@ -49,6 +49,9 @@ softpipe_clear(struct pipe_context *pipe, struct pipe_surface *ps, struct softpipe_context *softpipe = softpipe_context(pipe); uint i; + if (softpipe->no_rast) + return; + #if 0 softpipe_update_derived(softpipe); /* not needed?? */ #endif diff --git a/src/gallium/drivers/softpipe/sp_context.c b/src/gallium/drivers/softpipe/sp_context.c index e298ed37c3..fe9cd8375e 100644 --- a/src/gallium/drivers/softpipe/sp_context.c +++ b/src/gallium/drivers/softpipe/sp_context.c @@ -98,7 +98,6 @@ static void softpipe_destroy( struct pipe_context *pipe ) softpipe->quad.stencil_test->destroy( softpipe->quad.stencil_test ); softpipe->quad.occlusion->destroy( softpipe->quad.occlusion ); softpipe->quad.coverage->destroy( softpipe->quad.coverage ); - softpipe->quad.bufloop->destroy( softpipe->quad.bufloop ); softpipe->quad.blend->destroy( softpipe->quad.blend ); softpipe->quad.colormask->destroy( softpipe->quad.colormask ); softpipe->quad.output->destroy( softpipe->quad.output ); @@ -123,12 +122,12 @@ static void softpipe_destroy( struct pipe_context *pipe ) struct pipe_context * softpipe_create( struct pipe_screen *screen, struct pipe_winsys *pipe_winsys, - struct softpipe_winsys *softpipe_winsys ) + void *unused ) { struct softpipe_context *softpipe = CALLOC_STRUCT(softpipe_context); uint i; -#if defined(__i386__) || defined(__386__) +#ifdef PIPE_ARCH_X86 softpipe->use_sse = GETENV( "GALLIUM_NOSSE" ) == NULL; #else softpipe->use_sse = FALSE; @@ -179,6 +178,8 @@ softpipe_create( struct pipe_screen *screen, softpipe->pipe.draw_arrays = softpipe_draw_arrays; softpipe->pipe.draw_elements = softpipe_draw_elements; + softpipe->pipe.set_edgeflags = softpipe_set_edgeflags; + softpipe->pipe.clear = softpipe_clear; softpipe->pipe.flush = softpipe_flush; @@ -207,13 +208,10 @@ softpipe_create( struct pipe_screen *screen, softpipe->quad.stencil_test = sp_quad_stencil_test_stage(softpipe); softpipe->quad.occlusion = sp_quad_occlusion_stage(softpipe); softpipe->quad.coverage = sp_quad_coverage_stage(softpipe); - softpipe->quad.bufloop = sp_quad_bufloop_stage(softpipe); softpipe->quad.blend = sp_quad_blend_stage(softpipe); softpipe->quad.colormask = sp_quad_colormask_stage(softpipe); softpipe->quad.output = sp_quad_output_stage(softpipe); - softpipe->winsys = softpipe_winsys; - /* * Create drawing context and plug our rendering stage into it. */ @@ -221,6 +219,9 @@ softpipe_create( struct pipe_screen *screen, assert(softpipe->draw); softpipe->setup = sp_draw_render_stage(softpipe); + if (GETENV( "SP_NO_RAST" ) != NULL) + softpipe->no_rast = TRUE; + if (GETENV( "SP_VBUF" ) != NULL) { sp_init_vbuf(softpipe); } diff --git a/src/gallium/drivers/softpipe/sp_context.h b/src/gallium/drivers/softpipe/sp_context.h index dc9d0e6d5d..62eabfb30e 100644 --- a/src/gallium/drivers/softpipe/sp_context.h +++ b/src/gallium/drivers/softpipe/sp_context.h @@ -57,8 +57,6 @@ struct sp_vertex_shader; struct softpipe_context { struct pipe_context pipe; /**< base class */ - struct softpipe_winsys *winsys; /**< window system interface */ - /* The most recent drawing state as set by the driver: */ @@ -83,6 +81,10 @@ struct softpipe_context { unsigned num_samplers; unsigned num_textures; + unsigned num_vertex_elements; + unsigned num_vertex_buffers; + + boolean no_rast; /* Counter for occlusion queries. Note this supports overlapping * queries. @@ -103,6 +105,8 @@ struct softpipe_context { int psize_slot; + unsigned reduced_api_prim; /**< PIPE_PRIM_POINTS, _LINES or _TRIANGLES */ + #if 0 /* Stipple derived state: */ @@ -124,7 +128,6 @@ struct softpipe_context { struct quad_stage *depth_test; struct quad_stage *occlusion; struct quad_stage *coverage; - struct quad_stage *bufloop; struct quad_stage *blend; struct quad_stage *colormask; struct quad_stage *output; @@ -138,8 +141,6 @@ struct softpipe_context { struct draw_stage *vbuf; struct softpipe_vbuf_render *vbuf_render; - uint current_cbuf; /**< current color buffer being written to */ - struct softpipe_tile_cache *cbuf_cache[PIPE_MAX_COLOR_BUFS]; struct softpipe_tile_cache *zsbuf_cache; diff --git a/src/gallium/drivers/softpipe/sp_draw_arrays.c b/src/gallium/drivers/softpipe/sp_draw_arrays.c index ab54050d3f..6c58f9909d 100644 --- a/src/gallium/drivers/softpipe/sp_draw_arrays.c +++ b/src/gallium/drivers/softpipe/sp_draw_arrays.c @@ -78,6 +78,20 @@ softpipe_unmap_constant_buffers(struct softpipe_context *sp) } +static unsigned reduced_prim[PIPE_PRIM_POLYGON + 1] = { + PIPE_PRIM_POINTS, + PIPE_PRIM_LINES, + PIPE_PRIM_LINES, + PIPE_PRIM_LINES, + PIPE_PRIM_TRIANGLES, + PIPE_PRIM_TRIANGLES, + PIPE_PRIM_TRIANGLES, + PIPE_PRIM_TRIANGLES, + PIPE_PRIM_TRIANGLES, + PIPE_PRIM_TRIANGLES +}; + + boolean softpipe_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, unsigned count) @@ -104,17 +118,7 @@ softpipe_draw_elements(struct pipe_context *pipe, struct draw_context *draw = sp->draw; unsigned i; - /* first, check that the primitive is not malformed. It is the - * state tracker's responsibility to do send only correctly formed - * primitives down. It currently isn't doing that though... - */ -#if 1 - count = draw_trim_prim( mode, count ); -#else - if (!draw_validate_prim( mode, count )) - assert(0); -#endif - + sp->reduced_api_prim = reduced_prim[mode]; if (sp->dirty) softpipe_update_derived( sp ); @@ -125,14 +129,12 @@ softpipe_draw_elements(struct pipe_context *pipe, /* * Map vertex buffers */ - for (i = 0; i < PIPE_MAX_ATTRIBS; i++) { - if (sp->vertex_buffer[i].buffer) { - void *buf - = pipe->winsys->buffer_map(pipe->winsys, - sp->vertex_buffer[i].buffer, - PIPE_BUFFER_USAGE_CPU_READ); - draw_set_mapped_vertex_buffer(draw, i, buf); - } + for (i = 0; i < sp->num_vertex_buffers; i++) { + void *buf + = pipe->winsys->buffer_map(pipe->winsys, + sp->vertex_buffer[i].buffer, + PIPE_BUFFER_USAGE_CPU_READ); + draw_set_mapped_vertex_buffer(draw, i, buf); } /* Map index buffer, if present */ if (indexBuffer) { @@ -153,11 +155,9 @@ softpipe_draw_elements(struct pipe_context *pipe, /* * unmap vertex/index buffers - will cause draw module to flush */ - for (i = 0; i < PIPE_MAX_ATTRIBS; i++) { - if (sp->vertex_buffer[i].buffer) { - draw_set_mapped_vertex_buffer(draw, i, NULL); - pipe->winsys->buffer_unmap(pipe->winsys, sp->vertex_buffer[i].buffer); - } + for (i = 0; i < sp->num_vertex_buffers; i++) { + draw_set_mapped_vertex_buffer(draw, i, NULL); + pipe->winsys->buffer_unmap(pipe->winsys, sp->vertex_buffer[i].buffer); } if (indexBuffer) { draw_set_mapped_element_buffer(draw, 0, NULL); @@ -170,3 +170,12 @@ softpipe_draw_elements(struct pipe_context *pipe, return TRUE; } + + +void +softpipe_set_edgeflags(struct pipe_context *pipe, const unsigned *edgeflags) +{ + struct softpipe_context *sp = softpipe_context(pipe); + draw_set_edgeflags(sp->draw, edgeflags); +} + diff --git a/src/gallium/drivers/softpipe/sp_fs_exec.c b/src/gallium/drivers/softpipe/sp_fs_exec.c index d5bd7a702f..0b199a2193 100644 --- a/src/gallium/drivers/softpipe/sp_fs_exec.c +++ b/src/gallium/drivers/softpipe/sp_fs_exec.c @@ -37,6 +37,7 @@ #include "pipe/p_util.h" #include "pipe/p_inlines.h" #include "tgsi/exec/tgsi_exec.h" +#include "tgsi/util/tgsi_parse.h" struct sp_exec_fragment_shader { struct sp_fragment_shader base; @@ -116,6 +117,7 @@ exec_run( const struct sp_fragment_shader *base, static void exec_delete( struct sp_fragment_shader *base ) { + FREE((void *) base->shader.tokens); FREE(base); } @@ -137,7 +139,8 @@ softpipe_create_fs_exec(struct softpipe_context *softpipe, if (!shader) return NULL; - shader->base.shader = *templ; + /* we need to keep a local copy of the tokens */ + shader->base.shader.tokens = tgsi_dup_tokens(templ->tokens); shader->base.prepare = exec_prepare; shader->base.run = exec_run; shader->base.delete = exec_delete; diff --git a/src/gallium/drivers/softpipe/sp_fs_llvm.c b/src/gallium/drivers/softpipe/sp_fs_llvm.c index 07d058155f..6e1d9280bb 100644 --- a/src/gallium/drivers/softpipe/sp_fs_llvm.c +++ b/src/gallium/drivers/softpipe/sp_fs_llvm.c @@ -99,19 +99,19 @@ shade_quad_llvm(struct quad_stage *qs, allvmrt(qss->stage.softpipe->fs->info.output_semantic_name[qss->colorOutSlot] == TGSI_SEMANTIC_COLOR); for (i = 0; i < QUAD_SIZE; ++i) { - quad->outputs.color[0][i] = dests[i][qss->colorOutSlot][0]; - quad->outputs.color[1][i] = dests[i][qss->colorOutSlot][1]; - quad->outputs.color[2][i] = dests[i][qss->colorOutSlot][2]; - quad->outputs.color[3][i] = dests[i][qss->colorOutSlot][3]; + quad->outputs.color[0][0][i] = dests[i][qss->colorOutSlot][0]; + quad->outputs.color[0][1][i] = dests[i][qss->colorOutSlot][1]; + quad->outputs.color[0][2][i] = dests[i][qss->colorOutSlot][2]; + quad->outputs.color[0][3][i] = dests[i][qss->colorOutSlot][3]; } } #if DLLVM for (int i = 0; i < QUAD_SIZE; ++i) { debug_printf("QLLVM%d(%d) [%f, %f, %f, %f]\n", i, qss->colorOutSlot, - quad->outputs.color[0][i], - quad->outputs.color[1][i], - quad->outputs.color[2][i], - quad->outputs.color[3][i]); + quad->outputs.color[0][0][i], + quad->outputs.color[0][1][i], + quad->outputs.color[0][2][i], + quad->outputs.color[0][3][i]); } #endif diff --git a/src/gallium/drivers/softpipe/sp_fs_sse.c b/src/gallium/drivers/softpipe/sp_fs_sse.c index 8095d662ee..55741cc1df 100644 --- a/src/gallium/drivers/softpipe/sp_fs_sse.c +++ b/src/gallium/drivers/softpipe/sp_fs_sse.c @@ -40,7 +40,7 @@ #include "tgsi/exec/tgsi_sse2.h" -#if defined(__i386__) || defined(__386__) +#ifdef PIPE_ARCH_X86 #include "rtasm/rtasm_x86sse.h" @@ -49,9 +49,10 @@ typedef void (XSTDCALL *codegen_function)( const struct tgsi_exec_vector *input, struct tgsi_exec_vector *output, - float (*constant)[4], + const float (*constant)[4], struct tgsi_exec_vector *temporary, - const struct tgsi_interp_coef *coef + const struct tgsi_interp_coef *coef, + float (*immediates)[4] //, const struct tgsi_exec_vector *quadPos ); @@ -60,14 +61,15 @@ struct sp_sse_fragment_shader { struct sp_fragment_shader base; struct x86_function sse2_program; codegen_function func; + float immediates[TGSI_EXEC_NUM_IMMEDIATES][4]; }; static void -fs_sse_prepare( struct sp_fragment_shader *base, - struct tgsi_exec_machine *machine, - struct tgsi_sampler *samplers ) +fs_sse_prepare( const struct sp_fragment_shader *base, + struct tgsi_exec_machine *machine, + struct tgsi_sampler *samplers ) { } @@ -78,9 +80,9 @@ fs_sse_prepare( struct sp_fragment_shader *base, * TODO: process >1 quad at a time */ static unsigned -fs_sse_run( struct sp_fragment_shader *base, - struct tgsi_exec_machine *machine, - struct quad_header *quad ) +fs_sse_run( const struct sp_fragment_shader *base, + struct tgsi_exec_machine *machine, + struct quad_header *quad ) { struct sp_sse_fragment_shader *shader = (struct sp_sse_fragment_shader *) base; @@ -96,7 +98,8 @@ fs_sse_run( struct sp_fragment_shader *base, machine->Outputs, machine->Consts, machine->Temps, - machine->InterpCoefs + machine->InterpCoefs, + shader->immediates // , &machine->QuadPos ); @@ -129,13 +132,18 @@ softpipe_create_fs_sse(struct softpipe_context *softpipe, x86_init_func( &shader->sse2_program ); - if (!tgsi_emit_sse2_fs( templ->tokens, &shader->sse2_program )) { + if (!tgsi_emit_sse2( templ->tokens, &shader->sse2_program, + shader->immediates, FALSE )) { FREE(shader); return NULL; } shader->func = (codegen_function) x86_get_func( &shader->sse2_program ); - assert(shader->func); + if (!shader->func) { + x86_release_func( &shader->sse2_program ); + FREE(shader); + return NULL; + } shader->base.shader = *templ; shader->base.prepare = fs_sse_prepare; diff --git a/src/gallium/drivers/softpipe/sp_headers.h b/src/gallium/drivers/softpipe/sp_headers.h index 9cf8222133..3d9ede69bb 100644 --- a/src/gallium/drivers/softpipe/sp_headers.h +++ b/src/gallium/drivers/softpipe/sp_headers.h @@ -31,6 +31,7 @@ #ifndef SP_HEADERS_H #define SP_HEADERS_H +#include "pipe/p_state.h" #include "tgsi/exec/tgsi_exec.h" #define PRIM_POINT 1 @@ -66,7 +67,8 @@ struct quad_header { unsigned prim:2; /**< PRIM_POINT, LINE, TRI */ struct { - float color[NUM_CHANNELS][QUAD_SIZE]; /* rrrr, gggg, bbbb, aaaa */ + /** colors in SOA format (rrrr, gggg, bbbb, aaaa) */ + float color[PIPE_MAX_COLOR_BUFS][NUM_CHANNELS][QUAD_SIZE]; float depth[QUAD_SIZE]; } outputs; diff --git a/src/gallium/drivers/softpipe/sp_prim_setup.c b/src/gallium/drivers/softpipe/sp_prim_setup.c index c7eb12b3bb..1cf9ffa632 100644 --- a/src/gallium/drivers/softpipe/sp_prim_setup.c +++ b/src/gallium/drivers/softpipe/sp_prim_setup.c @@ -26,7 +26,9 @@ **************************************************************************/ /** - * \brief Primitive rasterization/rendering (points, lines, triangles) + * \brief A draw stage that drives our triangle setup routines from + * within the draw pipeline. One of two ways to drive setup, the + * other being in sp_prim_vbuf.c. * * \author Keith Whitwell <keith@tungstengraphics.com> * \author Brian Paul @@ -34,29 +36,12 @@ #include "sp_context.h" -#include "sp_headers.h" -#include "sp_quad.h" +#include "sp_setup.h" #include "sp_state.h" #include "sp_prim_setup.h" -#include "draw/draw_private.h" +#include "draw/draw_pipe.h" #include "draw/draw_vertex.h" #include "pipe/p_util.h" -#include "pipe/p_shader_tokens.h" - -#define DEBUG_VERTS 0 -#define DEBUG_FRAGS 0 - -/** - * Triangle edge info - */ -struct edge { - float dx; /**< X(v1) - X(v0), used only during setup */ - float dy; /**< Y(v1) - Y(v0), used only during setup */ - float dxdy; /**< dx/dy */ - float sx, sy; /**< first sample point coord */ - int lines; /**< number of lines on this edge */ -}; - /** * Triangle setup info (derived from draw_stage). @@ -65,39 +50,7 @@ struct edge { struct setup_stage { struct draw_stage stage; /**< This must be first (base class) */ - struct softpipe_context *softpipe; - - /* Vertices are just an array of floats making up each attribute in - * turn. Currently fixed at 4 floats, but should change in time. - * Codegen will help cope with this. - */ - const struct vertex_header *vmax; - const struct vertex_header *vmid; - const struct vertex_header *vmin; - const struct vertex_header *vprovoke; - - struct edge ebot; - struct edge etop; - struct edge emaj; - - float oneoverarea; - - struct tgsi_interp_coef coef[PIPE_MAX_SHADER_INPUTS]; - struct tgsi_interp_coef posCoef; /* For Z, W */ - struct quad_header quad; - - struct { - int left[2]; /**< [0] = row0, [1] = row1 */ - int right[2]; - int y; - unsigned y_flags; - unsigned mask; /**< mask of MASK_BOTTOM/TOP_LEFT/RIGHT bits */ - } span; - -#if DEBUG_FRAGS - uint numFragsEmitted; /**< per primitive */ - uint numFragsWritten; /**< per primitive */ -#endif + struct setup_context *setup; }; @@ -111,1112 +64,49 @@ static INLINE struct setup_stage *setup_stage( struct draw_stage *stage ) } -/** - * Clip setup->quad against the scissor/surface bounds. - */ -static INLINE void -quad_clip(struct setup_stage *setup) -{ - const struct pipe_scissor_state *cliprect = &setup->softpipe->cliprect; - const int minx = (int) cliprect->minx; - const int maxx = (int) cliprect->maxx; - const int miny = (int) cliprect->miny; - const int maxy = (int) cliprect->maxy; - - if (setup->quad.x0 >= maxx || - setup->quad.y0 >= maxy || - setup->quad.x0 + 1 < minx || - setup->quad.y0 + 1 < miny) { - /* totally clipped */ - setup->quad.mask = 0x0; - return; - } - if (setup->quad.x0 < minx) - setup->quad.mask &= (MASK_BOTTOM_RIGHT | MASK_TOP_RIGHT); - if (setup->quad.y0 < miny) - setup->quad.mask &= (MASK_BOTTOM_LEFT | MASK_BOTTOM_RIGHT); - if (setup->quad.x0 == maxx - 1) - setup->quad.mask &= (MASK_BOTTOM_LEFT | MASK_TOP_LEFT); - if (setup->quad.y0 == maxy - 1) - setup->quad.mask &= (MASK_TOP_LEFT | MASK_TOP_RIGHT); -} - - -/** - * Emit a quad (pass to next stage) with clipping. - */ -static INLINE void -clip_emit_quad(struct setup_stage *setup) -{ - quad_clip(setup); - if (setup->quad.mask) { - struct softpipe_context *sp = setup->softpipe; - sp->quad.first->run(sp->quad.first, &setup->quad); - } -} - - -/** - * Emit a quad (pass to next stage). No clipping is done. - */ -static INLINE void -emit_quad( struct setup_stage *setup, int x, int y, unsigned mask ) -{ - struct softpipe_context *sp = setup->softpipe; - setup->quad.x0 = x; - setup->quad.y0 = y; - setup->quad.mask = mask; -#if DEBUG_FRAGS - if (mask & 1) setup->numFragsEmitted++; - if (mask & 2) setup->numFragsEmitted++; - if (mask & 4) setup->numFragsEmitted++; - if (mask & 8) setup->numFragsEmitted++; -#endif - sp->quad.first->run(sp->quad.first, &setup->quad); -#if DEBUG_FRAGS - mask = setup->quad.mask; - if (mask & 1) setup->numFragsWritten++; - if (mask & 2) setup->numFragsWritten++; - if (mask & 4) setup->numFragsWritten++; - if (mask & 8) setup->numFragsWritten++; -#endif -} - - -/** - * Given an X or Y coordinate, return the block/quad coordinate that it - * belongs to. - */ -static INLINE int block( int x ) -{ - return x & ~1; -} - - -/** - * Compute mask which indicates which pixels in the 2x2 quad are actually inside - * the triangle's bounds. - * - * this is pretty nasty... may need to rework flush_spans again to - * fix it, if possible. - */ -static unsigned calculate_mask( struct setup_stage *setup, int x ) -{ - unsigned mask = 0x0; - - if (x >= setup->span.left[0] && x < setup->span.right[0]) - mask |= MASK_TOP_LEFT; - - if (x >= setup->span.left[1] && x < setup->span.right[1]) - mask |= MASK_BOTTOM_LEFT; - - if (x+1 >= setup->span.left[0] && x+1 < setup->span.right[0]) - mask |= MASK_TOP_RIGHT; - - if (x+1 >= setup->span.left[1] && x+1 < setup->span.right[1]) - mask |= MASK_BOTTOM_RIGHT; - - return mask; -} - - -/** - * Render a horizontal span of quads - */ -static void flush_spans( struct setup_stage *setup ) -{ - int minleft, maxright; - int x; - - switch (setup->span.y_flags) { - case 0x3: - /* both odd and even lines written (both quad rows) */ - minleft = MIN2(setup->span.left[0], setup->span.left[1]); - maxright = MAX2(setup->span.right[0], setup->span.right[1]); - break; - - case 0x1: - /* only even line written (quad top row) */ - minleft = setup->span.left[0]; - maxright = setup->span.right[0]; - break; - - case 0x2: - /* only odd line written (quad bottom row) */ - minleft = setup->span.left[1]; - maxright = setup->span.right[1]; - break; - - default: - return; - } - - /* XXX this loop could be moved into the above switch cases and - * calculate_mask() could be simplified a bit... - */ - for (x = block(minleft); x <= block(maxright); x += 2) { - emit_quad( setup, x, setup->span.y, - calculate_mask( setup, x ) ); - } - - setup->span.y = 0; - setup->span.y_flags = 0; - setup->span.right[0] = 0; - setup->span.right[1] = 0; -} - -#if DEBUG_VERTS -static void print_vertex(const struct setup_stage *setup, - const struct vertex_header *v) -{ - int i; - debug_printf("Vertex: (%p)\n", v); - for (i = 0; i < setup->quad.nr_attrs; i++) { - debug_printf(" %d: %f %f %f %f\n", i, - v->data[i][0], v->data[i][1], v->data[i][2], v->data[i][3]); - } -} -#endif - -static boolean setup_sort_vertices( struct setup_stage *setup, - const struct prim_header *prim ) -{ - const struct vertex_header *v0 = prim->v[0]; - const struct vertex_header *v1 = prim->v[1]; - const struct vertex_header *v2 = prim->v[2]; - -#if DEBUG_VERTS - debug_printf("Triangle:\n"); - print_vertex(setup, v0); - print_vertex(setup, v1); - print_vertex(setup, v2); -#endif - - setup->vprovoke = v2; - - /* determine bottom to top order of vertices */ - { - float y0 = v0->data[0][1]; - float y1 = v1->data[0][1]; - float y2 = v2->data[0][1]; - if (y0 <= y1) { - if (y1 <= y2) { - /* y0<=y1<=y2 */ - setup->vmin = v0; - setup->vmid = v1; - setup->vmax = v2; - } - else if (y2 <= y0) { - /* y2<=y0<=y1 */ - setup->vmin = v2; - setup->vmid = v0; - setup->vmax = v1; - } - else { - /* y0<=y2<=y1 */ - setup->vmin = v0; - setup->vmid = v2; - setup->vmax = v1; - } - } - else { - if (y0 <= y2) { - /* y1<=y0<=y2 */ - setup->vmin = v1; - setup->vmid = v0; - setup->vmax = v2; - } - else if (y2 <= y1) { - /* y2<=y1<=y0 */ - setup->vmin = v2; - setup->vmid = v1; - setup->vmax = v0; - } - else { - /* y1<=y2<=y0 */ - setup->vmin = v1; - setup->vmid = v2; - setup->vmax = v0; - } - } - } - - setup->ebot.dx = setup->vmid->data[0][0] - setup->vmin->data[0][0]; - setup->ebot.dy = setup->vmid->data[0][1] - setup->vmin->data[0][1]; - setup->emaj.dx = setup->vmax->data[0][0] - setup->vmin->data[0][0]; - setup->emaj.dy = setup->vmax->data[0][1] - setup->vmin->data[0][1]; - setup->etop.dx = setup->vmax->data[0][0] - setup->vmid->data[0][0]; - setup->etop.dy = setup->vmax->data[0][1] - setup->vmid->data[0][1]; - - /* - * Compute triangle's area. Use 1/area to compute partial - * derivatives of attributes later. - * - * The area will be the same as prim->det, but the sign may be - * different depending on how the vertices get sorted above. - * - * To determine whether the primitive is front or back facing we - * use the prim->det value because its sign is correct. - */ - { - const float area = (setup->emaj.dx * setup->ebot.dy - - setup->ebot.dx * setup->emaj.dy); - - setup->oneoverarea = 1.0f / area; - /* - debug_printf("%s one-over-area %f area %f det %f\n", - __FUNCTION__, setup->oneoverarea, area, prim->det ); - */ - } - - /* We need to know if this is a front or back-facing triangle for: - * - the GLSL gl_FrontFacing fragment attribute (bool) - * - two-sided stencil test - */ - setup->quad.facing = (prim->det > 0.0) ^ (setup->softpipe->rasterizer->front_winding == PIPE_WINDING_CW); - - return TRUE; -} - - -/** - * Compute a0 for a constant-valued coefficient (GL_FLAT shading). - * The value value comes from vertex->data[slot][i]. - * The result will be put into setup->coef[slot].a0[i]. - * \param slot which attribute slot - * \param i which component of the slot (0..3) - */ -static void const_coeff( struct setup_stage *setup, - struct tgsi_interp_coef *coef, - uint vertSlot, uint i) -{ - assert(i <= 3); - - coef->dadx[i] = 0; - coef->dady[i] = 0; - /* need provoking vertex info! - */ - coef->a0[i] = setup->vprovoke->data[vertSlot][i]; -} - - -/** - * Compute a0, dadx and dady for a linearly interpolated coefficient, - * for a triangle. - */ -static void tri_linear_coeff( struct setup_stage *setup, - struct tgsi_interp_coef *coef, - uint vertSlot, uint i) -{ - float botda = setup->vmid->data[vertSlot][i] - setup->vmin->data[vertSlot][i]; - float majda = setup->vmax->data[vertSlot][i] - setup->vmin->data[vertSlot][i]; - float a = setup->ebot.dy * majda - botda * setup->emaj.dy; - float b = setup->emaj.dx * botda - majda * setup->ebot.dx; - float dadx = a * setup->oneoverarea; - float dady = b * setup->oneoverarea; - - assert(i <= 3); - - coef->dadx[i] = dadx; - coef->dady[i] = dady; - - /* calculate a0 as the value which would be sampled for the - * fragment at (0,0), taking into account that we want to sample at - * pixel centers, in other words (0.5, 0.5). - * - * this is neat but unfortunately not a good way to do things for - * triangles with very large values of dadx or dady as it will - * result in the subtraction and re-addition from a0 of a very - * large number, which means we'll end up loosing a lot of the - * fractional bits and precision from a0. the way to fix this is - * to define a0 as the sample at a pixel center somewhere near vmin - * instead - i'll switch to this later. - */ - coef->a0[i] = (setup->vmin->data[vertSlot][i] - - (dadx * (setup->vmin->data[0][0] - 0.5f) + - dady * (setup->vmin->data[0][1] - 0.5f))); - - /* - debug_printf("attr[%d].%c: %f dx:%f dy:%f\n", - slot, "xyzw"[i], - setup->coef[slot].a0[i], - setup->coef[slot].dadx[i], - setup->coef[slot].dady[i]); - */ -} - - -/** - * Compute a0, dadx and dady for a perspective-corrected interpolant, - * for a triangle. - * We basically multiply the vertex value by 1/w before computing - * the plane coefficients (a0, dadx, dady). - * Later, when we compute the value at a particular fragment position we'll - * divide the interpolated value by the interpolated W at that fragment. - */ -static void tri_persp_coeff( struct setup_stage *setup, - struct tgsi_interp_coef *coef, - uint vertSlot, uint i) -{ - /* premultiply by 1/w (v->data[0][3] is always W): - */ - float mina = setup->vmin->data[vertSlot][i] * setup->vmin->data[0][3]; - float mida = setup->vmid->data[vertSlot][i] * setup->vmid->data[0][3]; - float maxa = setup->vmax->data[vertSlot][i] * setup->vmax->data[0][3]; - float botda = mida - mina; - float majda = maxa - mina; - float a = setup->ebot.dy * majda - botda * setup->emaj.dy; - float b = setup->emaj.dx * botda - majda * setup->ebot.dx; - float dadx = a * setup->oneoverarea; - float dady = b * setup->oneoverarea; - - /* - debug_printf("tri persp %d,%d: %f %f %f\n", vertSlot, i, - setup->vmin->data[vertSlot][i], - setup->vmid->data[vertSlot][i], - setup->vmax->data[vertSlot][i] - ); - */ - assert(i <= 3); - - coef->dadx[i] = dadx; - coef->dady[i] = dady; - coef->a0[i] = (mina - - (dadx * (setup->vmin->data[0][0] - 0.5f) + - dady * (setup->vmin->data[0][1] - 0.5f))); -} - - -/** - * Special coefficient setup for gl_FragCoord. - * X and Y are trivial, though Y has to be inverted for OpenGL. - * Z and W are copied from posCoef which should have already been computed. - * We could do a bit less work if we'd examine gl_FragCoord's swizzle mask. - */ static void -setup_fragcoord_coeff(struct setup_stage *setup, uint slot) -{ - /*X*/ - setup->coef[slot].a0[0] = 0; - setup->coef[slot].dadx[0] = 1.0; - setup->coef[slot].dady[0] = 0.0; - /*Y*/ - if (setup->softpipe->rasterizer->origin_lower_left) { - /* y=0=bottom */ - const int winHeight = setup->softpipe->framebuffer.height; - setup->coef[slot].a0[1] = (float) (winHeight - 1); - setup->coef[slot].dady[1] = -1.0; - } - else { - /* y=0=top */ - setup->coef[slot].a0[1] = 0.0; - setup->coef[slot].dady[1] = 1.0; - } - setup->coef[slot].dadx[1] = 0.0; - /*Z*/ - setup->coef[slot].a0[2] = setup->posCoef.a0[2]; - setup->coef[slot].dadx[2] = setup->posCoef.dadx[2]; - setup->coef[slot].dady[2] = setup->posCoef.dady[2]; - /*W*/ - setup->coef[slot].a0[3] = setup->posCoef.a0[3]; - setup->coef[slot].dadx[3] = setup->posCoef.dadx[3]; - setup->coef[slot].dady[3] = setup->posCoef.dady[3]; -} - - - -/** - * Compute the setup->coef[] array dadx, dady, a0 values. - * Must be called after setup->vmin,vmid,vmax,vprovoke are initialized. - */ -static void setup_tri_coefficients( struct setup_stage *setup ) -{ - struct softpipe_context *softpipe = setup->softpipe; - const struct sp_fragment_shader *spfs = softpipe->fs; - const struct vertex_info *vinfo = softpipe_get_vertex_info(softpipe); - uint fragSlot; - - /* z and w are done by linear interpolation: - */ - tri_linear_coeff(setup, &setup->posCoef, 0, 2); - tri_linear_coeff(setup, &setup->posCoef, 0, 3); - - /* setup interpolation for all the remaining attributes: - */ - for (fragSlot = 0; fragSlot < spfs->info.num_inputs; fragSlot++) { - const uint vertSlot = vinfo->src_index[fragSlot]; - uint j; - - switch (vinfo->interp_mode[fragSlot]) { - case INTERP_CONSTANT: - for (j = 0; j < NUM_CHANNELS; j++) - const_coeff(setup, &setup->coef[fragSlot], vertSlot, j); - break; - case INTERP_LINEAR: - for (j = 0; j < NUM_CHANNELS; j++) - tri_linear_coeff(setup, &setup->coef[fragSlot], vertSlot, j); - break; - case INTERP_PERSPECTIVE: - for (j = 0; j < NUM_CHANNELS; j++) - tri_persp_coeff(setup, &setup->coef[fragSlot], vertSlot, j); - break; - case INTERP_POS: - setup_fragcoord_coeff(setup, fragSlot); - break; - default: - assert(0); - } - - if (spfs->info.input_semantic_name[fragSlot] == TGSI_SEMANTIC_FOG) { - /* FOG.y = front/back facing XXX fix this */ - setup->coef[fragSlot].a0[1] = 1.0f - setup->quad.facing; - setup->coef[fragSlot].dadx[1] = 0.0; - setup->coef[fragSlot].dady[1] = 0.0; - } - } -} - - - -static void setup_tri_edges( struct setup_stage *setup ) -{ - float vmin_x = setup->vmin->data[0][0] + 0.5f; - float vmid_x = setup->vmid->data[0][0] + 0.5f; - - float vmin_y = setup->vmin->data[0][1] - 0.5f; - float vmid_y = setup->vmid->data[0][1] - 0.5f; - float vmax_y = setup->vmax->data[0][1] - 0.5f; - - setup->emaj.sy = CEILF(vmin_y); - setup->emaj.lines = (int) CEILF(vmax_y - setup->emaj.sy); - setup->emaj.dxdy = setup->emaj.dx / setup->emaj.dy; - setup->emaj.sx = vmin_x + (setup->emaj.sy - vmin_y) * setup->emaj.dxdy; - - setup->etop.sy = CEILF(vmid_y); - setup->etop.lines = (int) CEILF(vmax_y - setup->etop.sy); - setup->etop.dxdy = setup->etop.dx / setup->etop.dy; - setup->etop.sx = vmid_x + (setup->etop.sy - vmid_y) * setup->etop.dxdy; - - setup->ebot.sy = CEILF(vmin_y); - setup->ebot.lines = (int) CEILF(vmid_y - setup->ebot.sy); - setup->ebot.dxdy = setup->ebot.dx / setup->ebot.dy; - setup->ebot.sx = vmin_x + (setup->ebot.sy - vmin_y) * setup->ebot.dxdy; -} - - -/** - * Render the upper or lower half of a triangle. - * Scissoring/cliprect is applied here too. - */ -static void subtriangle( struct setup_stage *setup, - struct edge *eleft, - struct edge *eright, - unsigned lines ) -{ - const struct pipe_scissor_state *cliprect = &setup->softpipe->cliprect; - const int minx = (int) cliprect->minx; - const int maxx = (int) cliprect->maxx; - const int miny = (int) cliprect->miny; - const int maxy = (int) cliprect->maxy; - int y, start_y, finish_y; - int sy = (int)eleft->sy; - - assert((int)eleft->sy == (int) eright->sy); - - /* clip top/bottom */ - start_y = sy; - finish_y = sy + lines; - - if (start_y < miny) - start_y = miny; - - if (finish_y > maxy) - finish_y = maxy; - - start_y -= sy; - finish_y -= sy; - - /* - debug_printf("%s %d %d\n", __FUNCTION__, start_y, finish_y); - */ - - for (y = start_y; y < finish_y; y++) { - - /* avoid accumulating adds as floats don't have the precision to - * accurately iterate large triangle edges that way. luckily we - * can just multiply these days. - * - * this is all drowned out by the attribute interpolation anyway. - */ - int left = (int)(eleft->sx + y * eleft->dxdy); - int right = (int)(eright->sx + y * eright->dxdy); - - /* clip left/right */ - if (left < minx) - left = minx; - if (right > maxx) - right = maxx; - - if (left < right) { - int _y = sy + y; - if (block(_y) != setup->span.y) { - flush_spans(setup); - setup->span.y = block(_y); - } - - setup->span.left[_y&1] = left; - setup->span.right[_y&1] = right; - setup->span.y_flags |= 1<<(_y&1); - } - } - - - /* save the values so that emaj can be restarted: - */ - eleft->sx += lines * eleft->dxdy; - eright->sx += lines * eright->dxdy; - eleft->sy += lines; - eright->sy += lines; -} - - -/** - * Do setup for triangle rasterization, then render the triangle. - */ -static void setup_tri( struct draw_stage *stage, - struct prim_header *prim ) +do_tri(struct draw_stage *stage, struct prim_header *prim) { struct setup_stage *setup = setup_stage( stage ); - /* - debug_printf("%s\n", __FUNCTION__ ); - */ - -#if DEBUG_FRAGS - setup->numFragsEmitted = 0; - setup->numFragsWritten = 0; -#endif - - setup_sort_vertices( setup, prim ); - setup_tri_coefficients( setup ); - setup_tri_edges( setup ); - - setup->quad.prim = PRIM_TRI; - - setup->span.y = 0; - setup->span.y_flags = 0; - setup->span.right[0] = 0; - setup->span.right[1] = 0; - /* setup->span.z_mode = tri_z_mode( setup->ctx ); */ - - /* init_constant_attribs( setup ); */ - - if (setup->oneoverarea < 0.0) { - /* emaj on left: - */ - subtriangle( setup, &setup->emaj, &setup->ebot, setup->ebot.lines ); - subtriangle( setup, &setup->emaj, &setup->etop, setup->etop.lines ); - } - else { - /* emaj on right: - */ - subtriangle( setup, &setup->ebot, &setup->emaj, setup->ebot.lines ); - subtriangle( setup, &setup->etop, &setup->emaj, setup->etop.lines ); - } - - flush_spans( setup ); - -#if DEBUG_FRAGS - printf("Tri: %u frags emitted, %u written\n", - setup->numFragsEmitted, - setup->numFragsWritten); -#endif -} - - - -/** - * Compute a0, dadx and dady for a linearly interpolated coefficient, - * for a line. - */ -static void -line_linear_coeff(struct setup_stage *setup, - struct tgsi_interp_coef *coef, - uint vertSlot, uint i) -{ - const float da = setup->vmax->data[vertSlot][i] - setup->vmin->data[vertSlot][i]; - const float dadx = da * setup->emaj.dx * setup->oneoverarea; - const float dady = da * setup->emaj.dy * setup->oneoverarea; - coef->dadx[i] = dadx; - coef->dady[i] = dady; - coef->a0[i] = (setup->vmin->data[vertSlot][i] - - (dadx * (setup->vmin->data[0][0] - 0.5f) + - dady * (setup->vmin->data[0][1] - 0.5f))); -} - - -/** - * Compute a0, dadx and dady for a perspective-corrected interpolant, - * for a line. - */ -static void -line_persp_coeff(struct setup_stage *setup, - struct tgsi_interp_coef *coef, - uint vertSlot, uint i) -{ - /* XXX double-check/verify this arithmetic */ - const float a0 = setup->vmin->data[vertSlot][i] * setup->vmin->data[0][3]; - const float a1 = setup->vmax->data[vertSlot][i] * setup->vmax->data[0][3]; - const float da = a1 - a0; - const float dadx = da * setup->emaj.dx * setup->oneoverarea; - const float dady = da * setup->emaj.dy * setup->oneoverarea; - coef->dadx[i] = dadx; - coef->dady[i] = dady; - coef->a0[i] = (setup->vmin->data[vertSlot][i] - - (dadx * (setup->vmin->data[0][0] - 0.5f) + - dady * (setup->vmin->data[0][1] - 0.5f))); -} - - -/** - * Compute the setup->coef[] array dadx, dady, a0 values. - * Must be called after setup->vmin,vmax are initialized. - */ -static INLINE void -setup_line_coefficients(struct setup_stage *setup, struct prim_header *prim) -{ - struct softpipe_context *softpipe = setup->softpipe; - const struct sp_fragment_shader *spfs = softpipe->fs; - const struct vertex_info *vinfo = softpipe_get_vertex_info(softpipe); - uint fragSlot; - - /* use setup->vmin, vmax to point to vertices */ - setup->vprovoke = prim->v[1]; - setup->vmin = prim->v[0]; - setup->vmax = prim->v[1]; - - setup->emaj.dx = setup->vmax->data[0][0] - setup->vmin->data[0][0]; - setup->emaj.dy = setup->vmax->data[0][1] - setup->vmin->data[0][1]; - /* NOTE: this is not really 1/area */ - setup->oneoverarea = 1.0f / (setup->emaj.dx * setup->emaj.dx + - setup->emaj.dy * setup->emaj.dy); - - /* z and w are done by linear interpolation: - */ - line_linear_coeff(setup, &setup->posCoef, 0, 2); - line_linear_coeff(setup, &setup->posCoef, 0, 3); - - /* setup interpolation for all the remaining attributes: - */ - for (fragSlot = 0; fragSlot < spfs->info.num_inputs; fragSlot++) { - const uint vertSlot = vinfo->src_index[fragSlot]; - uint j; - - switch (vinfo->interp_mode[fragSlot]) { - case INTERP_CONSTANT: - for (j = 0; j < NUM_CHANNELS; j++) - const_coeff(setup, &setup->coef[fragSlot], vertSlot, j); - break; - case INTERP_LINEAR: - for (j = 0; j < NUM_CHANNELS; j++) - line_linear_coeff(setup, &setup->coef[fragSlot], vertSlot, j); - break; - case INTERP_PERSPECTIVE: - for (j = 0; j < NUM_CHANNELS; j++) - line_persp_coeff(setup, &setup->coef[fragSlot], vertSlot, j); - break; - case INTERP_POS: - setup_fragcoord_coeff(setup, fragSlot); - break; - default: - assert(0); - } - - if (spfs->info.input_semantic_name[fragSlot] == TGSI_SEMANTIC_FOG) { - /* FOG.y = front/back facing XXX fix this */ - setup->coef[fragSlot].a0[1] = 1.0f - setup->quad.facing; - setup->coef[fragSlot].dadx[1] = 0.0; - setup->coef[fragSlot].dady[1] = 0.0; - } - } + setup_tri( setup->setup, + prim->v[0]->data, + prim->v[1]->data, + prim->v[2]->data ); } - -/** - * Plot a pixel in a line segment. - */ -static INLINE void -plot(struct setup_stage *setup, int x, int y) -{ - const int iy = y & 1; - const int ix = x & 1; - const int quadX = x - ix; - const int quadY = y - iy; - const int mask = (1 << ix) << (2 * iy); - - if (quadX != setup->quad.x0 || - quadY != setup->quad.y0) - { - /* flush prev quad, start new quad */ - - if (setup->quad.x0 != -1) - clip_emit_quad(setup); - - setup->quad.x0 = quadX; - setup->quad.y0 = quadY; - setup->quad.mask = 0x0; - } - - setup->quad.mask |= mask; -} - - -/** - * Do setup for line rasterization, then render the line. - * Single-pixel width, no stipple, etc. We rely on the 'draw' module - * to handle stippling and wide lines. - */ static void -setup_line(struct draw_stage *stage, struct prim_header *prim) +do_line(struct draw_stage *stage, struct prim_header *prim) { - const struct vertex_header *v0 = prim->v[0]; - const struct vertex_header *v1 = prim->v[1]; struct setup_stage *setup = setup_stage( stage ); - int x0 = (int) v0->data[0][0]; - int x1 = (int) v1->data[0][0]; - int y0 = (int) v0->data[0][1]; - int y1 = (int) v1->data[0][1]; - int dx = x1 - x0; - int dy = y1 - y0; - int xstep, ystep; - if (dx == 0 && dy == 0) - return; - - setup_line_coefficients(setup, prim); - - if (dx < 0) { - dx = -dx; /* make positive */ - xstep = -1; - } - else { - xstep = 1; - } - - if (dy < 0) { - dy = -dy; /* make positive */ - ystep = -1; - } - else { - ystep = 1; - } - - assert(dx >= 0); - assert(dy >= 0); - - setup->quad.x0 = setup->quad.y0 = -1; - setup->quad.mask = 0x0; - setup->quad.prim = PRIM_LINE; - /* XXX temporary: set coverage to 1.0 so the line appears - * if AA mode happens to be enabled. - */ - setup->quad.coverage[0] = - setup->quad.coverage[1] = - setup->quad.coverage[2] = - setup->quad.coverage[3] = 1.0; - - if (dx > dy) { - /*** X-major line ***/ - int i; - const int errorInc = dy + dy; - int error = errorInc - dx; - const int errorDec = error - dx; - - for (i = 0; i < dx; i++) { - plot(setup, x0, y0); - - x0 += xstep; - if (error < 0) { - error += errorInc; - } - else { - error += errorDec; - y0 += ystep; - } - } - } - else { - /*** Y-major line ***/ - int i; - const int errorInc = dx + dx; - int error = errorInc - dy; - const int errorDec = error - dy; - - for (i = 0; i < dy; i++) { - plot(setup, x0, y0); - - y0 += ystep; - if (error < 0) { - error += errorInc; - } - else { - error += errorDec; - x0 += xstep; - } - } - } - - /* draw final quad */ - if (setup->quad.mask) { - clip_emit_quad(setup); - } + setup_line( setup->setup, + prim->v[0]->data, + prim->v[1]->data ); } - static void -point_persp_coeff(struct setup_stage *setup, - const struct vertex_header *vert, - struct tgsi_interp_coef *coef, - uint vertSlot, uint i) -{ - assert(i <= 3); - coef->dadx[i] = 0.0F; - coef->dady[i] = 0.0F; - coef->a0[i] = vert->data[vertSlot][i] * vert->data[0][3]; -} - - -/** - * Do setup for point rasterization, then render the point. - * Round or square points... - * XXX could optimize a lot for 1-pixel points. - */ -static void -setup_point(struct draw_stage *stage, struct prim_header *prim) +do_point(struct draw_stage *stage, struct prim_header *prim) { struct setup_stage *setup = setup_stage( stage ); - struct softpipe_context *softpipe = setup->softpipe; - const struct sp_fragment_shader *spfs = softpipe->fs; - const struct vertex_header *v0 = prim->v[0]; - const int sizeAttr = setup->softpipe->psize_slot; - const float size - = sizeAttr > 0 ? v0->data[sizeAttr][0] - : setup->softpipe->rasterizer->point_size; - const float halfSize = 0.5F * size; - const boolean round = (boolean) setup->softpipe->rasterizer->point_smooth; - const float x = v0->data[0][0]; /* Note: data[0] is always position */ - const float y = v0->data[0][1]; - const struct vertex_info *vinfo = softpipe_get_vertex_info(softpipe); - uint fragSlot; - - /* For points, all interpolants are constant-valued. - * However, for point sprites, we'll need to setup texcoords appropriately. - * XXX: which coefficients are the texcoords??? - * We may do point sprites as textured quads... - * - * KW: We don't know which coefficients are texcoords - ultimately - * the choice of what interpolation mode to use for each attribute - * should be determined by the fragment program, using - * per-attribute declaration statements that include interpolation - * mode as a parameter. So either the fragment program will have - * to be adjusted for pointsprite vs normal point behaviour, or - * otherwise a special interpolation mode will have to be defined - * which matches the required behaviour for point sprites. But - - * the latter is not a feature of normal hardware, and as such - * probably should be ruled out on that basis. - */ - setup->vprovoke = prim->v[0]; - - /* setup Z, W */ - const_coeff(setup, &setup->posCoef, 0, 2); - const_coeff(setup, &setup->posCoef, 0, 3); - - for (fragSlot = 0; fragSlot < spfs->info.num_inputs; fragSlot++) { - const uint vertSlot = vinfo->src_index[fragSlot]; - uint j; - - switch (vinfo->interp_mode[fragSlot]) { - case INTERP_CONSTANT: - /* fall-through */ - case INTERP_LINEAR: - for (j = 0; j < NUM_CHANNELS; j++) - const_coeff(setup, &setup->coef[fragSlot], vertSlot, j); - break; - case INTERP_PERSPECTIVE: - for (j = 0; j < NUM_CHANNELS; j++) - point_persp_coeff(setup, setup->vprovoke, - &setup->coef[fragSlot], vertSlot, j); - break; - case INTERP_POS: - setup_fragcoord_coeff(setup, fragSlot); - break; - default: - assert(0); - } - - if (spfs->info.input_semantic_name[fragSlot] == TGSI_SEMANTIC_FOG) { - /* FOG.y = front/back facing XXX fix this */ - setup->coef[fragSlot].a0[1] = 1.0f - setup->quad.facing; - setup->coef[fragSlot].dadx[1] = 0.0; - setup->coef[fragSlot].dady[1] = 0.0; - } - } - - setup->quad.prim = PRIM_POINT; - - if (halfSize <= 0.5 && !round) { - /* special case for 1-pixel points */ - const int ix = ((int) x) & 1; - const int iy = ((int) y) & 1; - setup->quad.x0 = (int) x - ix; - setup->quad.y0 = (int) y - iy; - setup->quad.mask = (1 << ix) << (2 * iy); - clip_emit_quad(setup); - } - else { - if (round) { - /* rounded points */ - const int ixmin = block((int) (x - halfSize)); - const int ixmax = block((int) (x + halfSize)); - const int iymin = block((int) (y - halfSize)); - const int iymax = block((int) (y + halfSize)); - const float rmin = halfSize - 0.7071F; /* 0.7071 = sqrt(2)/2 */ - const float rmax = halfSize + 0.7071F; - const float rmin2 = MAX2(0.0F, rmin * rmin); - const float rmax2 = rmax * rmax; - const float cscale = 1.0F / (rmax2 - rmin2); - int ix, iy; - - for (iy = iymin; iy <= iymax; iy += 2) { - for (ix = ixmin; ix <= ixmax; ix += 2) { - float dx, dy, dist2, cover; - - setup->quad.mask = 0x0; - - dx = (ix + 0.5f) - x; - dy = (iy + 0.5f) - y; - dist2 = dx * dx + dy * dy; - if (dist2 <= rmax2) { - cover = 1.0F - (dist2 - rmin2) * cscale; - setup->quad.coverage[QUAD_TOP_LEFT] = MIN2(cover, 1.0f); - setup->quad.mask |= MASK_TOP_LEFT; - } - - dx = (ix + 1.5f) - x; - dy = (iy + 0.5f) - y; - dist2 = dx * dx + dy * dy; - if (dist2 <= rmax2) { - cover = 1.0F - (dist2 - rmin2) * cscale; - setup->quad.coverage[QUAD_TOP_RIGHT] = MIN2(cover, 1.0f); - setup->quad.mask |= MASK_TOP_RIGHT; - } - dx = (ix + 0.5f) - x; - dy = (iy + 1.5f) - y; - dist2 = dx * dx + dy * dy; - if (dist2 <= rmax2) { - cover = 1.0F - (dist2 - rmin2) * cscale; - setup->quad.coverage[QUAD_BOTTOM_LEFT] = MIN2(cover, 1.0f); - setup->quad.mask |= MASK_BOTTOM_LEFT; - } - - dx = (ix + 1.5f) - x; - dy = (iy + 1.5f) - y; - dist2 = dx * dx + dy * dy; - if (dist2 <= rmax2) { - cover = 1.0F - (dist2 - rmin2) * cscale; - setup->quad.coverage[QUAD_BOTTOM_RIGHT] = MIN2(cover, 1.0f); - setup->quad.mask |= MASK_BOTTOM_RIGHT; - } - - if (setup->quad.mask) { - setup->quad.x0 = ix; - setup->quad.y0 = iy; - clip_emit_quad(setup); - } - } - } - } - else { - /* square points */ - const int xmin = (int) (x + 0.75 - halfSize); - const int ymin = (int) (y + 0.25 - halfSize); - const int xmax = xmin + (int) size; - const int ymax = ymin + (int) size; - /* XXX could apply scissor to xmin,ymin,xmax,ymax now */ - const int ixmin = block(xmin); - const int ixmax = block(xmax - 1); - const int iymin = block(ymin); - const int iymax = block(ymax - 1); - int ix, iy; - - /* - debug_printf("(%f, %f) -> X:%d..%d Y:%d..%d\n", x, y, xmin, xmax,ymin,ymax); - */ - for (iy = iymin; iy <= iymax; iy += 2) { - uint rowMask = 0xf; - if (iy < ymin) { - /* above the top edge */ - rowMask &= (MASK_BOTTOM_LEFT | MASK_BOTTOM_RIGHT); - } - if (iy + 1 >= ymax) { - /* below the bottom edge */ - rowMask &= (MASK_TOP_LEFT | MASK_TOP_RIGHT); - } - - for (ix = ixmin; ix <= ixmax; ix += 2) { - uint mask = rowMask; - - if (ix < xmin) { - /* fragment is past left edge of point, turn off left bits */ - mask &= (MASK_BOTTOM_RIGHT | MASK_TOP_RIGHT); - } - if (ix + 1 >= xmax) { - /* past the right edge */ - mask &= (MASK_BOTTOM_LEFT | MASK_TOP_LEFT); - } - - setup->quad.mask = mask; - setup->quad.x0 = ix; - setup->quad.y0 = iy; - clip_emit_quad(setup); - } - } - } - } + setup_point( setup->setup, + prim->v[0]->data ); } + static void setup_begin( struct draw_stage *stage ) { struct setup_stage *setup = setup_stage(stage); - struct softpipe_context *sp = setup->softpipe; - const struct sp_fragment_shader *fs = setup->softpipe->fs; - uint i; - - if (sp->dirty) { - softpipe_update_derived(sp); - } - /* Mark surfaces as defined now */ - for (i = 0; i < sp->framebuffer.num_cbufs; i++){ - if (sp->framebuffer.cbufs[i]) { - sp->framebuffer.cbufs[i]->status = PIPE_SURFACE_STATUS_DEFINED; - } - } - if (sp->framebuffer.zsbuf) { - sp->framebuffer.zsbuf->status = PIPE_SURFACE_STATUS_DEFINED; - } + setup_prepare( setup->setup ); - setup->quad.nr_attrs = fs->info.num_inputs; - - sp->quad.first->begin(sp->quad.first); - - stage->point = setup_point; - stage->line = setup_line; - stage->tri = setup_tri; + stage->point = do_point; + stage->line = do_line; + stage->tri = do_tri; } @@ -1260,6 +150,8 @@ static void reset_stipple_counter( struct draw_stage *stage ) static void render_destroy( struct draw_stage *stage ) { + struct setup_stage *ssetup = setup_stage(stage); + setup_destroy_context(ssetup->setup); FREE( stage ); } @@ -1269,19 +161,29 @@ static void render_destroy( struct draw_stage *stage ) */ struct draw_stage *sp_draw_render_stage( struct softpipe_context *softpipe ) { - struct setup_stage *setup = CALLOC_STRUCT(setup_stage); + struct setup_stage *sstage = CALLOC_STRUCT(setup_stage); - setup->softpipe = softpipe; - setup->stage.draw = softpipe->draw; - setup->stage.point = setup_first_point; - setup->stage.line = setup_first_line; - setup->stage.tri = setup_first_tri; - setup->stage.flush = setup_flush; - setup->stage.reset_stipple_counter = reset_stipple_counter; - setup->stage.destroy = render_destroy; + sstage->setup = setup_create_context(softpipe); + sstage->stage.draw = softpipe->draw; + sstage->stage.point = setup_first_point; + sstage->stage.line = setup_first_line; + sstage->stage.tri = setup_first_tri; + sstage->stage.flush = setup_flush; + sstage->stage.reset_stipple_counter = reset_stipple_counter; + sstage->stage.destroy = render_destroy; - setup->quad.coef = setup->coef; - setup->quad.posCoef = &setup->posCoef; + return (struct draw_stage *)sstage; +} - return &setup->stage; +struct setup_context * +sp_draw_setup_context( struct draw_stage *stage ) +{ + struct setup_stage *ssetup = setup_stage(stage); + return ssetup->setup; +} + +void +sp_draw_flush( struct draw_stage *stage ) +{ + stage->flush( stage, 0 ); } diff --git a/src/gallium/drivers/softpipe/sp_prim_setup.h b/src/gallium/drivers/softpipe/sp_prim_setup.h index f3e8a79dd9..49bdd98ed8 100644 --- a/src/gallium/drivers/softpipe/sp_prim_setup.h +++ b/src/gallium/drivers/softpipe/sp_prim_setup.h @@ -69,6 +69,12 @@ typedef void (*vbuf_draw_func)( struct pipe_context *pipe, extern struct draw_stage * sp_draw_render_stage( struct softpipe_context *softpipe ); +extern struct setup_context * +sp_draw_setup_context( struct draw_stage * ); + +extern void +sp_draw_flush( struct draw_stage * ); + extern struct draw_stage * sp_draw_vbuf_stage( struct draw_context *draw_context, diff --git a/src/gallium/drivers/softpipe/sp_prim_vbuf.c b/src/gallium/drivers/softpipe/sp_prim_vbuf.c index d940718ed2..e063fe82ef 100644 --- a/src/gallium/drivers/softpipe/sp_prim_vbuf.c +++ b/src/gallium/drivers/softpipe/sp_prim_vbuf.c @@ -1,8 +1,8 @@ /************************************************************************** - * + * * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. * All Rights Reserved. - * + * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the * "Software"), to deal in the Software without restriction, including @@ -10,11 +10,11 @@ * distribute, sub license, and/or sell copies of the Software, and to * permit persons to whom the Software is furnished to do so, subject to * the following conditions: - * + * * The above copyright notice and this permission notice (including the * next paragraph) shall be included in all copies or substantial portions * of the Software. - * + * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. @@ -22,7 +22,7 @@ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * + * **************************************************************************/ /** @@ -30,7 +30,7 @@ * softpipe rendering pipeline. * Probably not desired in general, but useful for testing/debuggin. * Enabled/Disabled with SP_VBUF env var. - * + * * Authors * Brian Paul */ @@ -39,14 +39,16 @@ #include "sp_context.h" #include "sp_state.h" #include "sp_prim_vbuf.h" +#include "sp_prim_setup.h" +#include "sp_setup.h" #include "draw/draw_context.h" -#include "draw/draw_private.h" #include "draw/draw_vbuf.h" #define SP_MAX_VBUF_INDEXES 1024 #define SP_MAX_VBUF_SIZE 4096 +typedef const float (*cptrf4)[4]; /** * Subclass of vbuf_render. @@ -69,7 +71,6 @@ softpipe_vbuf_render(struct vbuf_render *vbr) } - static const struct vertex_info * sp_vbuf_get_vertex_info(struct vbuf_render *vbr) { @@ -91,7 +92,7 @@ sp_vbuf_allocate_vertices(struct vbuf_render *vbr, static void -sp_vbuf_release_vertices(struct vbuf_render *vbr, void *vertices, +sp_vbuf_release_vertices(struct vbuf_render *vbr, void *vertices, unsigned vertex_size, unsigned vertices_used) { struct softpipe_vbuf_render *cvbr = softpipe_vbuf_render(vbr); @@ -105,6 +106,16 @@ static boolean sp_vbuf_set_primitive(struct vbuf_render *vbr, unsigned prim) { struct softpipe_vbuf_render *cvbr = softpipe_vbuf_render(vbr); + + /* XXX: break this dependency - make setup_context live under + * softpipe, rename the old "setup" draw stage to something else. + */ + struct setup_context *setup_ctx = sp_draw_setup_context(cvbr->softpipe->setup); + + setup_prepare( setup_ctx ); + + + if (prim == PIPE_PRIM_TRIANGLES || prim == PIPE_PRIM_LINES || prim == PIPE_PRIM_POINTS) { @@ -114,31 +125,10 @@ sp_vbuf_set_primitive(struct vbuf_render *vbr, unsigned prim) else { return FALSE; } - + } -/** - * Recalculate prim's determinant. - * XXX is this needed? - */ -static void -calc_det(struct prim_header *header) -{ - /* Window coords: */ - const float *v0 = header->v[0]->data[0]; - const float *v1 = header->v[1]->data[0]; - const float *v2 = header->v[2]->data[0]; - - /* edge vectors e = v0 - v2, f = v1 - v2 */ - const float ex = v0[0] - v2[0]; - const float ey = v0[1] - v2[1]; - const float fx = v1[0] - v2[0]; - const float fy = v1[1] - v2[1]; - - /* det = cross(e,f).z */ - header->det = ex * fy - ey * fx; -} static void @@ -146,49 +136,59 @@ sp_vbuf_draw(struct vbuf_render *vbr, const ushort *indices, uint nr_indices) { struct softpipe_vbuf_render *cvbr = softpipe_vbuf_render(vbr); struct softpipe_context *softpipe = cvbr->softpipe; - struct draw_stage *setup = softpipe->setup; - struct prim_header prim; unsigned vertex_size = softpipe->vertex_info_vbuf.size * sizeof(float); unsigned i, j; void *vertex_buffer = cvbr->vertex_buffer; + cptrf4 v[3]; + + /* XXX: break this dependency - make setup_context live under + * softpipe, rename the old "setup" draw stage to something else. + */ + struct draw_stage *setup = softpipe->setup; + struct setup_context *setup_ctx = sp_draw_setup_context(softpipe->setup); - prim.det = 0; - prim.reset_line_stipple = 0; - prim.edgeflags = 0; - prim.pad = 0; switch (cvbr->prim) { case PIPE_PRIM_TRIANGLES: for (i = 0; i < nr_indices; i += 3) { - for (j = 0; j < 3; j++) - prim.v[j] = (struct vertex_header *)((char *)vertex_buffer + - indices[i+j] * vertex_size); - - calc_det(&prim); - setup->tri( setup, &prim ); + for (j = 0; j < 3; j++) + v[j] = (cptrf4)((char *)vertex_buffer + + indices[i+j] * vertex_size); + + setup_tri( setup_ctx, + v[0], + v[1], + v[2]); } break; case PIPE_PRIM_LINES: for (i = 0; i < nr_indices; i += 2) { - for (j = 0; j < 2; j++) - prim.v[j] = (struct vertex_header *)((char *)vertex_buffer + - indices[i+j] * vertex_size); - - setup->line( setup, &prim ); + for (j = 0; j < 2; j++) + v[j] = (cptrf4)((char *)vertex_buffer + + indices[i+j] * vertex_size); + + setup_line( setup_ctx, + v[0], + v[1] ); } break; case PIPE_PRIM_POINTS: for (i = 0; i < nr_indices; i++) { - prim.v[0] = (struct vertex_header *)((char *)vertex_buffer + - indices[i] * vertex_size); - setup->point( setup, &prim ); + v[0] = (cptrf4)((char *)vertex_buffer + + indices[i] * vertex_size); + + setup_point( setup_ctx, + v[0] ); } break; } - setup->flush( setup, 0 ); + /* XXX: why are we calling this??? If we had to call something, it + * would be a function in sp_setup.c: + */ + sp_draw_flush( setup ); } @@ -202,111 +202,123 @@ sp_vbuf_draw_arrays(struct vbuf_render *vbr, uint start, uint nr) struct softpipe_vbuf_render *cvbr = softpipe_vbuf_render(vbr); struct softpipe_context *softpipe = cvbr->softpipe; struct draw_stage *setup = softpipe->setup; - struct prim_header prim; const void *vertex_buffer = cvbr->vertex_buffer; const unsigned vertex_size = softpipe->vertex_info_vbuf.size * sizeof(float); unsigned i; - - prim.det = 0; - prim.reset_line_stipple = 0; - prim.edgeflags = 0; - prim.pad = 0; + struct setup_context *setup_ctx = sp_draw_setup_context(setup); + cptrf4 v[3]; #define VERTEX(I) \ - (struct vertex_header *) ((char *) vertex_buffer + (I) * vertex_size) + (cptrf4) ((char *) vertex_buffer + (I) * vertex_size) switch (cvbr->prim) { case PIPE_PRIM_POINTS: for (i = 0; i < nr; i++) { - prim.v[0] = VERTEX(i); - setup->point( setup, &prim ); + v[0] = VERTEX(i); + setup_point( setup_ctx, v[0] ); } break; case PIPE_PRIM_LINES: assert(nr % 2 == 0); for (i = 0; i < nr; i += 2) { - prim.v[0] = VERTEX(i); - prim.v[1] = VERTEX(i + 1); - setup->line( setup, &prim ); + v[0] = VERTEX(i); + v[1] = VERTEX(i + 1); + setup_line( setup_ctx, v[0], v[1] ); } break; case PIPE_PRIM_LINE_STRIP: for (i = 1; i < nr; i++) { - prim.v[0] = VERTEX(i - 1); - prim.v[1] = VERTEX(i); - setup->line( setup, &prim ); + v[0] = VERTEX(i - 1); + v[1] = VERTEX(i); + setup_line( setup_ctx, v[0], v[1] ); } break; case PIPE_PRIM_TRIANGLES: assert(nr % 3 == 0); for (i = 0; i < nr; i += 3) { - prim.v[0] = VERTEX(i + 0); - prim.v[1] = VERTEX(i + 1); - prim.v[2] = VERTEX(i + 2); - calc_det(&prim); - setup->tri( setup, &prim ); + v[0] = VERTEX(i + 0); + v[1] = VERTEX(i + 1); + v[2] = VERTEX(i + 2); + setup_tri( setup_ctx, + v[0], + v[1], + v[2] ); } break; case PIPE_PRIM_TRIANGLE_STRIP: assert(nr >= 3); for (i = 2; i < nr; i++) { - prim.v[0] = VERTEX(i - 2); - prim.v[1] = VERTEX(i - 1); - prim.v[2] = VERTEX(i); - calc_det(&prim); - setup->tri( setup, &prim ); + v[0] = VERTEX(i - 2); + v[1] = VERTEX(i - 1); + v[2] = VERTEX(i); + setup_tri( setup_ctx, + v[0], + v[1], + v[2] ); } break; case PIPE_PRIM_TRIANGLE_FAN: assert(nr >= 3); for (i = 2; i < nr; i++) { - prim.v[0] = VERTEX(0); - prim.v[1] = VERTEX(i - 1); - prim.v[2] = VERTEX(i); - calc_det(&prim); - setup->tri( setup, &prim ); + v[0] = VERTEX(0); + v[1] = VERTEX(i - 1); + v[2] = VERTEX(i); + setup_tri( setup_ctx, + v[0], + v[1], + v[2] ); } break; case PIPE_PRIM_QUADS: assert(nr % 4 == 0); for (i = 0; i < nr; i += 4) { - prim.v[0] = VERTEX(i + 0); - prim.v[1] = VERTEX(i + 1); - prim.v[2] = VERTEX(i + 2); - calc_det(&prim); - setup->tri( setup, &prim ); - - prim.v[0] = VERTEX(i + 0); - prim.v[1] = VERTEX(i + 2); - prim.v[2] = VERTEX(i + 3); - calc_det(&prim); - setup->tri( setup, &prim ); + v[0] = VERTEX(i + 0); + v[1] = VERTEX(i + 1); + v[2] = VERTEX(i + 2); + setup_tri( setup_ctx, + v[0], + v[1], + v[2] ); + + v[0] = VERTEX(i + 0); + v[1] = VERTEX(i + 2); + v[2] = VERTEX(i + 3); + setup_tri( setup_ctx, + v[0], + v[1], + v[2] ); } break; case PIPE_PRIM_QUAD_STRIP: assert(nr >= 4); for (i = 2; i < nr; i += 2) { - prim.v[0] = VERTEX(i - 2); - prim.v[1] = VERTEX(i); - prim.v[2] = VERTEX(i + 1); - calc_det(&prim); - setup->tri( setup, &prim ); - - prim.v[0] = VERTEX(i - 2); - prim.v[1] = VERTEX(i + 1); - prim.v[2] = VERTEX(i - 1); - calc_det(&prim); - setup->tri( setup, &prim ); + v[0] = VERTEX(i - 2); + v[1] = VERTEX(i); + v[2] = VERTEX(i + 1); + setup_tri( setup_ctx, + v[0], + v[1], + v[2] ); + + v[0] = VERTEX(i - 2); + v[1] = VERTEX(i + 1); + v[2] = VERTEX(i - 1); + setup_tri( setup_ctx, + v[0], + v[1], + v[2] ); } break; case PIPE_PRIM_POLYGON: /* draw as tri fan */ for (i = 2; i < nr; i++) { - prim.v[0] = VERTEX(0); - prim.v[1] = VERTEX(i - 1); - prim.v[2] = VERTEX(i); - calc_det(&prim); - setup->tri( setup, &prim ); + v[0] = VERTEX(0); + v[1] = VERTEX(i - 1); + v[2] = VERTEX(i); + setup_tri( setup_ctx, + v[0], + v[1], + v[2] ); } break; default: diff --git a/src/gallium/drivers/softpipe/sp_quad.c b/src/gallium/drivers/softpipe/sp_quad.c index 8603c1a367..bc83d78ea1 100644 --- a/src/gallium/drivers/softpipe/sp_quad.c +++ b/src/gallium/drivers/softpipe/sp_quad.c @@ -76,15 +76,6 @@ sp_build_quad_pipeline(struct softpipe_context *sp) sp_push_quad_first( sp, sp->quad.blend ); } - if (sp->framebuffer.num_cbufs == 1) { - /* the usual case: write to exactly one colorbuf */ - sp->current_cbuf = 0; - } - else { - /* insert bufloop stage */ - sp_push_quad_first( sp, sp->quad.bufloop ); - } - if (sp->depth_stencil->depth.occlusion_count) { sp_push_quad_first( sp, sp->quad.occlusion ); } diff --git a/src/gallium/drivers/softpipe/sp_quad.h b/src/gallium/drivers/softpipe/sp_quad.h index f1e0281764..08513cb95f 100644 --- a/src/gallium/drivers/softpipe/sp_quad.h +++ b/src/gallium/drivers/softpipe/sp_quad.h @@ -58,7 +58,6 @@ struct quad_stage *sp_quad_stencil_test_stage( struct softpipe_context *softpipe struct quad_stage *sp_quad_depth_test_stage( struct softpipe_context *softpipe ); struct quad_stage *sp_quad_occlusion_stage( struct softpipe_context *softpipe ); struct quad_stage *sp_quad_coverage_stage( struct softpipe_context *softpipe ); -struct quad_stage *sp_quad_bufloop_stage( struct softpipe_context *softpipe ); struct quad_stage *sp_quad_blend_stage( struct softpipe_context *softpipe ); struct quad_stage *sp_quad_colormask_stage( struct softpipe_context *softpipe ); struct quad_stage *sp_quad_output_stage( struct softpipe_context *softpipe ); diff --git a/src/gallium/drivers/softpipe/sp_quad_alpha_test.c b/src/gallium/drivers/softpipe/sp_quad_alpha_test.c index 318916fe30..7a42b08ef5 100644 --- a/src/gallium/drivers/softpipe/sp_quad_alpha_test.c +++ b/src/gallium/drivers/softpipe/sp_quad_alpha_test.c @@ -16,7 +16,8 @@ alpha_test_quad(struct quad_stage *qs, struct quad_header *quad) struct softpipe_context *softpipe = qs->softpipe; const float ref = softpipe->depth_stencil->alpha.ref; unsigned passMask = 0x0, j; - const float *aaaa = quad->outputs.color[3]; + const uint cbuf = 0; /* only output[0].alpha is tested */ + const float *aaaa = quad->outputs.color[cbuf][3]; switch (softpipe->depth_stencil->alpha.func) { case PIPE_FUNC_NEVER: @@ -25,7 +26,7 @@ alpha_test_quad(struct quad_stage *qs, struct quad_header *quad) case PIPE_FUNC_LESS: /* * If mask were an array [4] we could do this SIMD-style: - * passMask = (quad->outputs.color[3] <= vec4(ref)); + * passMask = (quad->outputs.color[0][3] <= vec4(ref)); */ for (j = 0; j < QUAD_SIZE; j++) { if (aaaa[j] < ref) { diff --git a/src/gallium/drivers/softpipe/sp_quad_blend.c b/src/gallium/drivers/softpipe/sp_quad_blend.c index 0b79cfa1ed..74c6bff84a 100644 --- a/src/gallium/drivers/softpipe/sp_quad_blend.c +++ b/src/gallium/drivers/softpipe/sp_quad_blend.c @@ -101,114 +101,119 @@ static void logicop_quad(struct quad_stage *qs, struct quad_header *quad) { struct softpipe_context *softpipe = qs->softpipe; - float dest[4][QUAD_SIZE]; - ubyte src[4][4], dst[4][4], res[4][4]; - uint *src4 = (uint *) src; - uint *dst4 = (uint *) dst; - uint *res4 = (uint *) res; - struct softpipe_cached_tile * - tile = sp_get_cached_tile(softpipe, - softpipe->cbuf_cache[softpipe->current_cbuf], - quad->x0, quad->y0); - float (*quadColor)[4] = quad->outputs.color; - uint i, j; - - /* get/swizzle dest colors */ - for (j = 0; j < QUAD_SIZE; j++) { - int x = (quad->x0 & (TILE_SIZE-1)) + (j & 1); - int y = (quad->y0 & (TILE_SIZE-1)) + (j >> 1); - for (i = 0; i < 4; i++) { - dest[i][j] = tile->data.color[y][x][i]; + uint cbuf; + + /* loop over colorbuffer outputs */ + for (cbuf = 0; cbuf < softpipe->framebuffer.num_cbufs; cbuf++) { + float dest[4][QUAD_SIZE]; + ubyte src[4][4], dst[4][4], res[4][4]; + uint *src4 = (uint *) src; + uint *dst4 = (uint *) dst; + uint *res4 = (uint *) res; + struct softpipe_cached_tile * + tile = sp_get_cached_tile(softpipe, + softpipe->cbuf_cache[cbuf], + quad->x0, quad->y0); + float (*quadColor)[4] = quad->outputs.color[cbuf]; + uint i, j; + + /* get/swizzle dest colors */ + for (j = 0; j < QUAD_SIZE; j++) { + int x = (quad->x0 & (TILE_SIZE-1)) + (j & 1); + int y = (quad->y0 & (TILE_SIZE-1)) + (j >> 1); + for (i = 0; i < 4; i++) { + dest[i][j] = tile->data.color[y][x][i]; + } } - } - /* convert to ubyte */ - for (j = 0; j < 4; j++) { /* loop over R,G,B,A channels */ - UNCLAMPED_FLOAT_TO_UBYTE(dst[j][0], dest[j][0]); /* P0 */ - UNCLAMPED_FLOAT_TO_UBYTE(dst[j][1], dest[j][1]); /* P1 */ - UNCLAMPED_FLOAT_TO_UBYTE(dst[j][2], dest[j][2]); /* P2 */ - UNCLAMPED_FLOAT_TO_UBYTE(dst[j][3], dest[j][3]); /* P3 */ - - UNCLAMPED_FLOAT_TO_UBYTE(src[j][0], quadColor[j][0]); /* P0 */ - UNCLAMPED_FLOAT_TO_UBYTE(src[j][1], quadColor[j][1]); /* P1 */ - UNCLAMPED_FLOAT_TO_UBYTE(src[j][2], quadColor[j][2]); /* P2 */ - UNCLAMPED_FLOAT_TO_UBYTE(src[j][3], quadColor[j][3]); /* P3 */ - } + /* convert to ubyte */ + for (j = 0; j < 4; j++) { /* loop over R,G,B,A channels */ + UNCLAMPED_FLOAT_TO_UBYTE(dst[j][0], dest[j][0]); /* P0 */ + UNCLAMPED_FLOAT_TO_UBYTE(dst[j][1], dest[j][1]); /* P1 */ + UNCLAMPED_FLOAT_TO_UBYTE(dst[j][2], dest[j][2]); /* P2 */ + UNCLAMPED_FLOAT_TO_UBYTE(dst[j][3], dest[j][3]); /* P3 */ + + UNCLAMPED_FLOAT_TO_UBYTE(src[j][0], quadColor[j][0]); /* P0 */ + UNCLAMPED_FLOAT_TO_UBYTE(src[j][1], quadColor[j][1]); /* P1 */ + UNCLAMPED_FLOAT_TO_UBYTE(src[j][2], quadColor[j][2]); /* P2 */ + UNCLAMPED_FLOAT_TO_UBYTE(src[j][3], quadColor[j][3]); /* P3 */ + } - switch (softpipe->blend->logicop_func) { - case PIPE_LOGICOP_CLEAR: - for (j = 0; j < 4; j++) - res4[j] = 0; - break; - case PIPE_LOGICOP_NOR: - for (j = 0; j < 4; j++) - res4[j] = ~(src4[j] | dst4[j]); - break; - case PIPE_LOGICOP_AND_INVERTED: - for (j = 0; j < 4; j++) - res4[j] = ~src4[j] & dst4[j]; - break; - case PIPE_LOGICOP_COPY_INVERTED: - for (j = 0; j < 4; j++) - res4[j] = ~src4[j]; - break; - case PIPE_LOGICOP_AND_REVERSE: - for (j = 0; j < 4; j++) - res4[j] = src4[j] & ~dst4[j]; - break; - case PIPE_LOGICOP_INVERT: - for (j = 0; j < 4; j++) - res4[j] = ~dst4[j]; - break; - case PIPE_LOGICOP_XOR: - for (j = 0; j < 4; j++) - res4[j] = dst4[j] ^ src4[j]; - break; - case PIPE_LOGICOP_NAND: - for (j = 0; j < 4; j++) - res4[j] = ~(src4[j] & dst4[j]); - break; - case PIPE_LOGICOP_AND: - for (j = 0; j < 4; j++) - res4[j] = src4[j] & dst4[j]; - break; - case PIPE_LOGICOP_EQUIV: - for (j = 0; j < 4; j++) - res4[j] = ~(src4[j] ^ dst4[j]); - break; - case PIPE_LOGICOP_NOOP: - for (j = 0; j < 4; j++) - res4[j] = dst4[j]; - break; - case PIPE_LOGICOP_OR_INVERTED: - for (j = 0; j < 4; j++) - res4[j] = ~src4[j] | dst4[j]; - break; - case PIPE_LOGICOP_COPY: - for (j = 0; j < 4; j++) - res4[j] = src4[j]; - break; - case PIPE_LOGICOP_OR_REVERSE: - for (j = 0; j < 4; j++) - res4[j] = src4[j] | ~dst4[j]; - break; - case PIPE_LOGICOP_OR: - for (j = 0; j < 4; j++) - res4[j] = src4[j] | dst4[j]; - break; - case PIPE_LOGICOP_SET: - for (j = 0; j < 4; j++) - res4[j] = ~0; - break; - default: - assert(0); - } + switch (softpipe->blend->logicop_func) { + case PIPE_LOGICOP_CLEAR: + for (j = 0; j < 4; j++) + res4[j] = 0; + break; + case PIPE_LOGICOP_NOR: + for (j = 0; j < 4; j++) + res4[j] = ~(src4[j] | dst4[j]); + break; + case PIPE_LOGICOP_AND_INVERTED: + for (j = 0; j < 4; j++) + res4[j] = ~src4[j] & dst4[j]; + break; + case PIPE_LOGICOP_COPY_INVERTED: + for (j = 0; j < 4; j++) + res4[j] = ~src4[j]; + break; + case PIPE_LOGICOP_AND_REVERSE: + for (j = 0; j < 4; j++) + res4[j] = src4[j] & ~dst4[j]; + break; + case PIPE_LOGICOP_INVERT: + for (j = 0; j < 4; j++) + res4[j] = ~dst4[j]; + break; + case PIPE_LOGICOP_XOR: + for (j = 0; j < 4; j++) + res4[j] = dst4[j] ^ src4[j]; + break; + case PIPE_LOGICOP_NAND: + for (j = 0; j < 4; j++) + res4[j] = ~(src4[j] & dst4[j]); + break; + case PIPE_LOGICOP_AND: + for (j = 0; j < 4; j++) + res4[j] = src4[j] & dst4[j]; + break; + case PIPE_LOGICOP_EQUIV: + for (j = 0; j < 4; j++) + res4[j] = ~(src4[j] ^ dst4[j]); + break; + case PIPE_LOGICOP_NOOP: + for (j = 0; j < 4; j++) + res4[j] = dst4[j]; + break; + case PIPE_LOGICOP_OR_INVERTED: + for (j = 0; j < 4; j++) + res4[j] = ~src4[j] | dst4[j]; + break; + case PIPE_LOGICOP_COPY: + for (j = 0; j < 4; j++) + res4[j] = src4[j]; + break; + case PIPE_LOGICOP_OR_REVERSE: + for (j = 0; j < 4; j++) + res4[j] = src4[j] | ~dst4[j]; + break; + case PIPE_LOGICOP_OR: + for (j = 0; j < 4; j++) + res4[j] = src4[j] | dst4[j]; + break; + case PIPE_LOGICOP_SET: + for (j = 0; j < 4; j++) + res4[j] = ~0; + break; + default: + assert(0); + } - for (j = 0; j < 4; j++) { - quadColor[j][0] = UBYTE_TO_FLOAT(res[j][0]); - quadColor[j][1] = UBYTE_TO_FLOAT(res[j][1]); - quadColor[j][2] = UBYTE_TO_FLOAT(res[j][2]); - quadColor[j][3] = UBYTE_TO_FLOAT(res[j][3]); + for (j = 0; j < 4; j++) { + quadColor[j][0] = UBYTE_TO_FLOAT(res[j][0]); + quadColor[j][1] = UBYTE_TO_FLOAT(res[j][1]); + quadColor[j][2] = UBYTE_TO_FLOAT(res[j][2]); + quadColor[j][3] = UBYTE_TO_FLOAT(res[j][3]); + } } /* pass quad to next stage */ @@ -221,504 +226,508 @@ logicop_quad(struct quad_stage *qs, struct quad_header *quad) static void blend_quad(struct quad_stage *qs, struct quad_header *quad) { - struct softpipe_context *softpipe = qs->softpipe; static const float zero[4] = { 0, 0, 0, 0 }; static const float one[4] = { 1, 1, 1, 1 }; - float source[4][QUAD_SIZE], dest[4][QUAD_SIZE]; - struct softpipe_cached_tile *tile - = sp_get_cached_tile(softpipe, - softpipe->cbuf_cache[softpipe->current_cbuf], - quad->x0, quad->y0); - float (*quadColor)[4] = quad->outputs.color; - uint i, j; + + struct softpipe_context *softpipe = qs->softpipe; + uint cbuf; if (softpipe->blend->logicop_enable) { logicop_quad(qs, quad); return; } - /* get/swizzle dest colors */ - for (j = 0; j < QUAD_SIZE; j++) { - int x = (quad->x0 & (TILE_SIZE-1)) + (j & 1); - int y = (quad->y0 & (TILE_SIZE-1)) + (j >> 1); - for (i = 0; i < 4; i++) { - dest[i][j] = tile->data.color[y][x][i]; + /* loop over colorbuffer outputs */ + for (cbuf = 0; cbuf < softpipe->framebuffer.num_cbufs; cbuf++) { + float source[4][QUAD_SIZE], dest[4][QUAD_SIZE]; + struct softpipe_cached_tile *tile + = sp_get_cached_tile(softpipe, + softpipe->cbuf_cache[cbuf], + quad->x0, quad->y0); + float (*quadColor)[4] = quad->outputs.color[cbuf]; + uint i, j; + + /* get/swizzle dest colors */ + for (j = 0; j < QUAD_SIZE; j++) { + int x = (quad->x0 & (TILE_SIZE-1)) + (j & 1); + int y = (quad->y0 & (TILE_SIZE-1)) + (j >> 1); + for (i = 0; i < 4; i++) { + dest[i][j] = tile->data.color[y][x][i]; + } } - } - /* - * Compute src/first term RGB - */ - switch (softpipe->blend->rgb_src_factor) { - case PIPE_BLENDFACTOR_ONE: - VEC4_COPY(source[0], quadColor[0]); /* R */ - VEC4_COPY(source[1], quadColor[1]); /* G */ - VEC4_COPY(source[2], quadColor[2]); /* B */ - break; - case PIPE_BLENDFACTOR_SRC_COLOR: - VEC4_MUL(source[0], quadColor[0], quadColor[0]); /* R */ - VEC4_MUL(source[1], quadColor[1], quadColor[1]); /* G */ - VEC4_MUL(source[2], quadColor[2], quadColor[2]); /* B */ - break; - case PIPE_BLENDFACTOR_SRC_ALPHA: - { - const float *alpha = quadColor[3]; - VEC4_MUL(source[0], quadColor[0], alpha); /* R */ - VEC4_MUL(source[1], quadColor[1], alpha); /* G */ - VEC4_MUL(source[2], quadColor[2], alpha); /* B */ - } - break; - case PIPE_BLENDFACTOR_DST_COLOR: - VEC4_MUL(source[0], quadColor[0], dest[0]); /* R */ - VEC4_MUL(source[1], quadColor[1], dest[1]); /* G */ - VEC4_MUL(source[2], quadColor[2], dest[2]); /* B */ - break; - case PIPE_BLENDFACTOR_DST_ALPHA: - { - const float *alpha = dest[3]; - VEC4_MUL(source[0], quadColor[0], alpha); /* R */ - VEC4_MUL(source[1], quadColor[1], alpha); /* G */ - VEC4_MUL(source[2], quadColor[2], alpha); /* B */ - } - break; - case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: - { - const float *alpha = quadColor[3]; - float diff[4]; - VEC4_SUB(diff, one, dest[3]); - VEC4_MIN(source[0], alpha, diff); /* R */ - VEC4_MIN(source[1], alpha, diff); /* G */ - VEC4_MIN(source[2], alpha, diff); /* B */ - } - break; - case PIPE_BLENDFACTOR_CONST_COLOR: - { - float comp[4]; - VEC4_SCALAR(comp, softpipe->blend_color.color[0]); /* R */ - VEC4_MUL(source[0], quadColor[0], comp); /* R */ - VEC4_SCALAR(comp, softpipe->blend_color.color[1]); /* G */ - VEC4_MUL(source[1], quadColor[1], comp); /* G */ - VEC4_SCALAR(comp, softpipe->blend_color.color[2]); /* B */ - VEC4_MUL(source[2], quadColor[2], comp); /* B */ - } - break; - case PIPE_BLENDFACTOR_CONST_ALPHA: - { - float alpha[4]; - VEC4_SCALAR(alpha, softpipe->blend_color.color[3]); - VEC4_MUL(source[0], quadColor[0], alpha); /* R */ - VEC4_MUL(source[1], quadColor[1], alpha); /* G */ - VEC4_MUL(source[2], quadColor[2], alpha); /* B */ - } - break; - case PIPE_BLENDFACTOR_SRC1_COLOR: - assert(0); /* to do */ - break; - case PIPE_BLENDFACTOR_SRC1_ALPHA: - assert(0); /* to do */ - break; - case PIPE_BLENDFACTOR_ZERO: - VEC4_COPY(source[0], zero); /* R */ - VEC4_COPY(source[1], zero); /* G */ - VEC4_COPY(source[2], zero); /* B */ - break; - case PIPE_BLENDFACTOR_INV_SRC_COLOR: - { - float inv_comp[4]; - VEC4_SUB(inv_comp, one, quadColor[0]); /* R */ - VEC4_MUL(source[0], quadColor[0], inv_comp); /* R */ - VEC4_SUB(inv_comp, one, quadColor[1]); /* G */ - VEC4_MUL(source[1], quadColor[1], inv_comp); /* G */ - VEC4_SUB(inv_comp, one, quadColor[2]); /* B */ - VEC4_MUL(source[2], quadColor[2], inv_comp); /* B */ - } - break; - case PIPE_BLENDFACTOR_INV_SRC_ALPHA: - { - float inv_alpha[4]; - VEC4_SUB(inv_alpha, one, quadColor[3]); - VEC4_MUL(source[0], quadColor[0], inv_alpha); /* R */ - VEC4_MUL(source[1], quadColor[1], inv_alpha); /* G */ - VEC4_MUL(source[2], quadColor[2], inv_alpha); /* B */ - } - break; - case PIPE_BLENDFACTOR_INV_DST_ALPHA: - { - float inv_alpha[4]; - VEC4_SUB(inv_alpha, one, dest[3]); - VEC4_MUL(source[0], quadColor[0], inv_alpha); /* R */ - VEC4_MUL(source[1], quadColor[1], inv_alpha); /* G */ - VEC4_MUL(source[2], quadColor[2], inv_alpha); /* B */ - } - break; - case PIPE_BLENDFACTOR_INV_DST_COLOR: - { - float inv_comp[4]; - VEC4_SUB(inv_comp, one, dest[0]); /* R */ - VEC4_MUL(source[0], quadColor[0], inv_comp); /* R */ - VEC4_SUB(inv_comp, one, dest[1]); /* G */ - VEC4_MUL(source[1], quadColor[1], inv_comp); /* G */ - VEC4_SUB(inv_comp, one, dest[2]); /* B */ - VEC4_MUL(source[2], quadColor[2], inv_comp); /* B */ - } - break; - case PIPE_BLENDFACTOR_INV_CONST_COLOR: - { - float inv_comp[4]; - /* R */ - VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[0]); - VEC4_MUL(source[0], quadColor[0], inv_comp); - /* G */ - VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[1]); - VEC4_MUL(source[1], quadColor[1], inv_comp); - /* B */ - VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[2]); - VEC4_MUL(source[2], quadColor[2], inv_comp); - } - break; - case PIPE_BLENDFACTOR_INV_CONST_ALPHA: - { - float inv_alpha[4]; - VEC4_SCALAR(inv_alpha, 1.0f - softpipe->blend_color.color[3]); - VEC4_MUL(source[0], quadColor[0], inv_alpha); /* R */ - VEC4_MUL(source[1], quadColor[1], inv_alpha); /* G */ - VEC4_MUL(source[2], quadColor[2], inv_alpha); /* B */ - } - break; - case PIPE_BLENDFACTOR_INV_SRC1_COLOR: - assert(0); /* to do */ - break; - case PIPE_BLENDFACTOR_INV_SRC1_ALPHA: - assert(0); /* to do */ - break; - default: - assert(0); - } - - /* - * Compute src/first term A - */ - switch (softpipe->blend->alpha_src_factor) { - case PIPE_BLENDFACTOR_ONE: - VEC4_COPY(source[3], quadColor[3]); /* A */ - break; - case PIPE_BLENDFACTOR_SRC_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_SRC_ALPHA: - { - const float *alpha = quadColor[3]; - VEC4_MUL(source[3], quadColor[3], alpha); /* A */ - } - break; - case PIPE_BLENDFACTOR_DST_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_DST_ALPHA: - VEC4_MUL(source[3], quadColor[3], dest[3]); /* A */ - break; - case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: - { - const float *alpha = quadColor[3]; - float diff[4]; - VEC4_SUB(diff, one, dest[3]); - VEC4_MIN(source[3], alpha, diff); /* A */ - } - break; - case PIPE_BLENDFACTOR_CONST_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_CONST_ALPHA: - { - float comp[4]; - VEC4_SCALAR(comp, softpipe->blend_color.color[3]); /* A */ - VEC4_MUL(source[3], quadColor[3], comp); /* A */ + /* + * Compute src/first term RGB + */ + switch (softpipe->blend->rgb_src_factor) { + case PIPE_BLENDFACTOR_ONE: + VEC4_COPY(source[0], quadColor[0]); /* R */ + VEC4_COPY(source[1], quadColor[1]); /* G */ + VEC4_COPY(source[2], quadColor[2]); /* B */ + break; + case PIPE_BLENDFACTOR_SRC_COLOR: + VEC4_MUL(source[0], quadColor[0], quadColor[0]); /* R */ + VEC4_MUL(source[1], quadColor[1], quadColor[1]); /* G */ + VEC4_MUL(source[2], quadColor[2], quadColor[2]); /* B */ + break; + case PIPE_BLENDFACTOR_SRC_ALPHA: + { + const float *alpha = quadColor[3]; + VEC4_MUL(source[0], quadColor[0], alpha); /* R */ + VEC4_MUL(source[1], quadColor[1], alpha); /* G */ + VEC4_MUL(source[2], quadColor[2], alpha); /* B */ + } + break; + case PIPE_BLENDFACTOR_DST_COLOR: + VEC4_MUL(source[0], quadColor[0], dest[0]); /* R */ + VEC4_MUL(source[1], quadColor[1], dest[1]); /* G */ + VEC4_MUL(source[2], quadColor[2], dest[2]); /* B */ + break; + case PIPE_BLENDFACTOR_DST_ALPHA: + { + const float *alpha = dest[3]; + VEC4_MUL(source[0], quadColor[0], alpha); /* R */ + VEC4_MUL(source[1], quadColor[1], alpha); /* G */ + VEC4_MUL(source[2], quadColor[2], alpha); /* B */ + } + break; + case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: + { + const float *alpha = quadColor[3]; + float diff[4], temp[4]; + VEC4_SUB(diff, one, dest[3]); + VEC4_MIN(temp, alpha, diff); + VEC4_MUL(source[0], quadColor[0], temp); /* R */ + VEC4_MUL(source[1], quadColor[1], temp); /* G */ + VEC4_MUL(source[2], quadColor[2], temp); /* B */ + } + break; + case PIPE_BLENDFACTOR_CONST_COLOR: + { + float comp[4]; + VEC4_SCALAR(comp, softpipe->blend_color.color[0]); /* R */ + VEC4_MUL(source[0], quadColor[0], comp); /* R */ + VEC4_SCALAR(comp, softpipe->blend_color.color[1]); /* G */ + VEC4_MUL(source[1], quadColor[1], comp); /* G */ + VEC4_SCALAR(comp, softpipe->blend_color.color[2]); /* B */ + VEC4_MUL(source[2], quadColor[2], comp); /* B */ + } + break; + case PIPE_BLENDFACTOR_CONST_ALPHA: + { + float alpha[4]; + VEC4_SCALAR(alpha, softpipe->blend_color.color[3]); + VEC4_MUL(source[0], quadColor[0], alpha); /* R */ + VEC4_MUL(source[1], quadColor[1], alpha); /* G */ + VEC4_MUL(source[2], quadColor[2], alpha); /* B */ + } + break; + case PIPE_BLENDFACTOR_SRC1_COLOR: + assert(0); /* to do */ + break; + case PIPE_BLENDFACTOR_SRC1_ALPHA: + assert(0); /* to do */ + break; + case PIPE_BLENDFACTOR_ZERO: + VEC4_COPY(source[0], zero); /* R */ + VEC4_COPY(source[1], zero); /* G */ + VEC4_COPY(source[2], zero); /* B */ + break; + case PIPE_BLENDFACTOR_INV_SRC_COLOR: + { + float inv_comp[4]; + VEC4_SUB(inv_comp, one, quadColor[0]); /* R */ + VEC4_MUL(source[0], quadColor[0], inv_comp); /* R */ + VEC4_SUB(inv_comp, one, quadColor[1]); /* G */ + VEC4_MUL(source[1], quadColor[1], inv_comp); /* G */ + VEC4_SUB(inv_comp, one, quadColor[2]); /* B */ + VEC4_MUL(source[2], quadColor[2], inv_comp); /* B */ + } + break; + case PIPE_BLENDFACTOR_INV_SRC_ALPHA: + { + float inv_alpha[4]; + VEC4_SUB(inv_alpha, one, quadColor[3]); + VEC4_MUL(source[0], quadColor[0], inv_alpha); /* R */ + VEC4_MUL(source[1], quadColor[1], inv_alpha); /* G */ + VEC4_MUL(source[2], quadColor[2], inv_alpha); /* B */ + } + break; + case PIPE_BLENDFACTOR_INV_DST_ALPHA: + { + float inv_alpha[4]; + VEC4_SUB(inv_alpha, one, dest[3]); + VEC4_MUL(source[0], quadColor[0], inv_alpha); /* R */ + VEC4_MUL(source[1], quadColor[1], inv_alpha); /* G */ + VEC4_MUL(source[2], quadColor[2], inv_alpha); /* B */ + } + break; + case PIPE_BLENDFACTOR_INV_DST_COLOR: + { + float inv_comp[4]; + VEC4_SUB(inv_comp, one, dest[0]); /* R */ + VEC4_MUL(source[0], quadColor[0], inv_comp); /* R */ + VEC4_SUB(inv_comp, one, dest[1]); /* G */ + VEC4_MUL(source[1], quadColor[1], inv_comp); /* G */ + VEC4_SUB(inv_comp, one, dest[2]); /* B */ + VEC4_MUL(source[2], quadColor[2], inv_comp); /* B */ + } + break; + case PIPE_BLENDFACTOR_INV_CONST_COLOR: + { + float inv_comp[4]; + /* R */ + VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[0]); + VEC4_MUL(source[0], quadColor[0], inv_comp); + /* G */ + VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[1]); + VEC4_MUL(source[1], quadColor[1], inv_comp); + /* B */ + VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[2]); + VEC4_MUL(source[2], quadColor[2], inv_comp); + } + break; + case PIPE_BLENDFACTOR_INV_CONST_ALPHA: + { + float inv_alpha[4]; + VEC4_SCALAR(inv_alpha, 1.0f - softpipe->blend_color.color[3]); + VEC4_MUL(source[0], quadColor[0], inv_alpha); /* R */ + VEC4_MUL(source[1], quadColor[1], inv_alpha); /* G */ + VEC4_MUL(source[2], quadColor[2], inv_alpha); /* B */ + } + break; + case PIPE_BLENDFACTOR_INV_SRC1_COLOR: + assert(0); /* to do */ + break; + case PIPE_BLENDFACTOR_INV_SRC1_ALPHA: + assert(0); /* to do */ + break; + default: + assert(0); } - break; - case PIPE_BLENDFACTOR_ZERO: - VEC4_COPY(source[3], zero); /* A */ - break; - case PIPE_BLENDFACTOR_INV_SRC_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_INV_SRC_ALPHA: - { - float inv_alpha[4]; - VEC4_SUB(inv_alpha, one, quadColor[3]); - VEC4_MUL(source[3], quadColor[3], inv_alpha); /* A */ - } - break; - case PIPE_BLENDFACTOR_INV_DST_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_INV_DST_ALPHA: - { - float inv_alpha[4]; - VEC4_SUB(inv_alpha, one, dest[3]); - VEC4_MUL(source[3], quadColor[3], inv_alpha); /* A */ - } - break; - case PIPE_BLENDFACTOR_INV_CONST_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_INV_CONST_ALPHA: - { - float inv_comp[4]; - /* A */ - VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[3]); - VEC4_MUL(source[3], quadColor[3], inv_comp); - } - break; - default: - assert(0); - } - - - /* - * Compute dest/second term RGB - */ - switch (softpipe->blend->rgb_dst_factor) { - case PIPE_BLENDFACTOR_ONE: - /* dest = dest * 1 NO-OP, leave dest as-is */ - break; - case PIPE_BLENDFACTOR_SRC_COLOR: - VEC4_MUL(dest[0], dest[0], quadColor[0]); /* R */ - VEC4_MUL(dest[1], dest[1], quadColor[1]); /* G */ - VEC4_MUL(dest[2], dest[2], quadColor[2]); /* B */ - break; - case PIPE_BLENDFACTOR_SRC_ALPHA: - VEC4_MUL(dest[0], dest[0], quadColor[3]); /* R * A */ - VEC4_MUL(dest[1], dest[1], quadColor[3]); /* G * A */ - VEC4_MUL(dest[2], dest[2], quadColor[3]); /* B * A */ - break; - case PIPE_BLENDFACTOR_DST_ALPHA: - VEC4_MUL(dest[0], dest[0], dest[3]); /* R * A */ - VEC4_MUL(dest[1], dest[1], dest[3]); /* G * A */ - VEC4_MUL(dest[2], dest[2], dest[3]); /* B * A */ - break; - case PIPE_BLENDFACTOR_DST_COLOR: - VEC4_MUL(dest[0], dest[0], dest[0]); /* R */ - VEC4_MUL(dest[1], dest[1], dest[1]); /* G */ - VEC4_MUL(dest[2], dest[2], dest[2]); /* B */ - break; - case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: - assert(0); /* illegal */ - break; - case PIPE_BLENDFACTOR_CONST_COLOR: - { - float comp[4]; - VEC4_SCALAR(comp, softpipe->blend_color.color[0]); /* R */ - VEC4_MUL(dest[0], dest[0], comp); /* R */ - VEC4_SCALAR(comp, softpipe->blend_color.color[1]); /* G */ - VEC4_MUL(dest[1], dest[1], comp); /* G */ - VEC4_SCALAR(comp, softpipe->blend_color.color[2]); /* B */ - VEC4_MUL(dest[2], dest[2], comp); /* B */ - } - break; - case PIPE_BLENDFACTOR_CONST_ALPHA: - { - float comp[4]; - VEC4_SCALAR(comp, softpipe->blend_color.color[3]); /* A */ - VEC4_MUL(dest[0], dest[0], comp); /* R */ - VEC4_MUL(dest[1], dest[1], comp); /* G */ - VEC4_MUL(dest[2], dest[2], comp); /* B */ - } - break; - case PIPE_BLENDFACTOR_ZERO: - VEC4_COPY(dest[0], zero); /* R */ - VEC4_COPY(dest[1], zero); /* G */ - VEC4_COPY(dest[2], zero); /* B */ - break; - case PIPE_BLENDFACTOR_SRC1_COLOR: - case PIPE_BLENDFACTOR_SRC1_ALPHA: - /* XXX what are these? */ - assert(0); - break; - case PIPE_BLENDFACTOR_INV_SRC_COLOR: - { - float inv_comp[4]; - VEC4_SUB(inv_comp, one, quadColor[0]); /* R */ - VEC4_MUL(dest[0], inv_comp, dest[0]); /* R */ - VEC4_SUB(inv_comp, one, quadColor[1]); /* G */ - VEC4_MUL(dest[1], inv_comp, dest[1]); /* G */ - VEC4_SUB(inv_comp, one, quadColor[2]); /* B */ - VEC4_MUL(dest[2], inv_comp, dest[2]); /* B */ - } - break; - case PIPE_BLENDFACTOR_INV_SRC_ALPHA: - { - float one_minus_alpha[QUAD_SIZE]; - VEC4_SUB(one_minus_alpha, one, quadColor[3]); - VEC4_MUL(dest[0], dest[0], one_minus_alpha); /* R */ - VEC4_MUL(dest[1], dest[1], one_minus_alpha); /* G */ - VEC4_MUL(dest[2], dest[2], one_minus_alpha); /* B */ - } - break; - case PIPE_BLENDFACTOR_INV_DST_ALPHA: - { - float inv_comp[4]; - VEC4_SUB(inv_comp, one, quadColor[3]); /* A */ - VEC4_MUL(dest[0], inv_comp, dest[0]); /* R */ - VEC4_MUL(dest[1], inv_comp, dest[1]); /* G */ - VEC4_MUL(dest[2], inv_comp, dest[2]); /* B */ - } - break; - case PIPE_BLENDFACTOR_INV_DST_COLOR: - { - float inv_comp[4]; - VEC4_SUB(inv_comp, one, dest[0]); /* R */ - VEC4_MUL(dest[0], dest[0], inv_comp); /* R */ - VEC4_SUB(inv_comp, one, dest[1]); /* G */ - VEC4_MUL(dest[1], dest[1], inv_comp); /* G */ - VEC4_SUB(inv_comp, one, dest[2]); /* B */ - VEC4_MUL(dest[2], dest[2], inv_comp); /* B */ - } - break; - case PIPE_BLENDFACTOR_INV_CONST_COLOR: - { - float inv_comp[4]; - /* R */ - VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[0]); - VEC4_MUL(dest[0], dest[0], inv_comp); - /* G */ - VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[1]); - VEC4_MUL(dest[1], dest[1], inv_comp); - /* B */ - VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[2]); - VEC4_MUL(dest[2], dest[2], inv_comp); - } - break; - case PIPE_BLENDFACTOR_INV_CONST_ALPHA: - { - float inv_comp[4]; - VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[3]); - VEC4_MUL(dest[0], dest[0], inv_comp); - VEC4_MUL(dest[1], dest[1], inv_comp); - VEC4_MUL(dest[2], dest[2], inv_comp); + + /* + * Compute src/first term A + */ + switch (softpipe->blend->alpha_src_factor) { + case PIPE_BLENDFACTOR_ONE: + VEC4_COPY(source[3], quadColor[3]); /* A */ + break; + case PIPE_BLENDFACTOR_SRC_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_SRC_ALPHA: + { + const float *alpha = quadColor[3]; + VEC4_MUL(source[3], quadColor[3], alpha); /* A */ + } + break; + case PIPE_BLENDFACTOR_DST_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_DST_ALPHA: + VEC4_MUL(source[3], quadColor[3], dest[3]); /* A */ + break; + case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: + /* multiply alpha by 1.0 */ + VEC4_COPY(source[3], quadColor[3]); /* A */ + break; + case PIPE_BLENDFACTOR_CONST_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_CONST_ALPHA: + { + float comp[4]; + VEC4_SCALAR(comp, softpipe->blend_color.color[3]); /* A */ + VEC4_MUL(source[3], quadColor[3], comp); /* A */ + } + break; + case PIPE_BLENDFACTOR_ZERO: + VEC4_COPY(source[3], zero); /* A */ + break; + case PIPE_BLENDFACTOR_INV_SRC_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_INV_SRC_ALPHA: + { + float inv_alpha[4]; + VEC4_SUB(inv_alpha, one, quadColor[3]); + VEC4_MUL(source[3], quadColor[3], inv_alpha); /* A */ + } + break; + case PIPE_BLENDFACTOR_INV_DST_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_INV_DST_ALPHA: + { + float inv_alpha[4]; + VEC4_SUB(inv_alpha, one, dest[3]); + VEC4_MUL(source[3], quadColor[3], inv_alpha); /* A */ + } + break; + case PIPE_BLENDFACTOR_INV_CONST_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_INV_CONST_ALPHA: + { + float inv_comp[4]; + /* A */ + VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[3]); + VEC4_MUL(source[3], quadColor[3], inv_comp); + } + break; + default: + assert(0); } - break; - case PIPE_BLENDFACTOR_INV_SRC1_COLOR: - case PIPE_BLENDFACTOR_INV_SRC1_ALPHA: - /* XXX what are these? */ - assert(0); - break; - default: - assert(0); - } - - /* - * Compute dest/second term A - */ - switch (softpipe->blend->alpha_dst_factor) { - case PIPE_BLENDFACTOR_ONE: - /* dest = dest * 1 NO-OP, leave dest as-is */ - break; - case PIPE_BLENDFACTOR_SRC_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_SRC_ALPHA: - VEC4_MUL(dest[3], dest[3], quadColor[3]); /* A * A */ - break; - case PIPE_BLENDFACTOR_DST_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_DST_ALPHA: - VEC4_MUL(dest[3], dest[3], dest[3]); /* A */ - break; - case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: - assert(0); /* illegal */ - break; - case PIPE_BLENDFACTOR_CONST_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_CONST_ALPHA: - { - float comp[4]; - VEC4_SCALAR(comp, softpipe->blend_color.color[3]); /* A */ - VEC4_MUL(dest[3], dest[3], comp); /* A */ + + + /* + * Compute dest/second term RGB + */ + switch (softpipe->blend->rgb_dst_factor) { + case PIPE_BLENDFACTOR_ONE: + /* dest = dest * 1 NO-OP, leave dest as-is */ + break; + case PIPE_BLENDFACTOR_SRC_COLOR: + VEC4_MUL(dest[0], dest[0], quadColor[0]); /* R */ + VEC4_MUL(dest[1], dest[1], quadColor[1]); /* G */ + VEC4_MUL(dest[2], dest[2], quadColor[2]); /* B */ + break; + case PIPE_BLENDFACTOR_SRC_ALPHA: + VEC4_MUL(dest[0], dest[0], quadColor[3]); /* R * A */ + VEC4_MUL(dest[1], dest[1], quadColor[3]); /* G * A */ + VEC4_MUL(dest[2], dest[2], quadColor[3]); /* B * A */ + break; + case PIPE_BLENDFACTOR_DST_ALPHA: + VEC4_MUL(dest[0], dest[0], dest[3]); /* R * A */ + VEC4_MUL(dest[1], dest[1], dest[3]); /* G * A */ + VEC4_MUL(dest[2], dest[2], dest[3]); /* B * A */ + break; + case PIPE_BLENDFACTOR_DST_COLOR: + VEC4_MUL(dest[0], dest[0], dest[0]); /* R */ + VEC4_MUL(dest[1], dest[1], dest[1]); /* G */ + VEC4_MUL(dest[2], dest[2], dest[2]); /* B */ + break; + case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: + assert(0); /* illegal */ + break; + case PIPE_BLENDFACTOR_CONST_COLOR: + { + float comp[4]; + VEC4_SCALAR(comp, softpipe->blend_color.color[0]); /* R */ + VEC4_MUL(dest[0], dest[0], comp); /* R */ + VEC4_SCALAR(comp, softpipe->blend_color.color[1]); /* G */ + VEC4_MUL(dest[1], dest[1], comp); /* G */ + VEC4_SCALAR(comp, softpipe->blend_color.color[2]); /* B */ + VEC4_MUL(dest[2], dest[2], comp); /* B */ + } + break; + case PIPE_BLENDFACTOR_CONST_ALPHA: + { + float comp[4]; + VEC4_SCALAR(comp, softpipe->blend_color.color[3]); /* A */ + VEC4_MUL(dest[0], dest[0], comp); /* R */ + VEC4_MUL(dest[1], dest[1], comp); /* G */ + VEC4_MUL(dest[2], dest[2], comp); /* B */ + } + break; + case PIPE_BLENDFACTOR_ZERO: + VEC4_COPY(dest[0], zero); /* R */ + VEC4_COPY(dest[1], zero); /* G */ + VEC4_COPY(dest[2], zero); /* B */ + break; + case PIPE_BLENDFACTOR_SRC1_COLOR: + case PIPE_BLENDFACTOR_SRC1_ALPHA: + /* XXX what are these? */ + assert(0); + break; + case PIPE_BLENDFACTOR_INV_SRC_COLOR: + { + float inv_comp[4]; + VEC4_SUB(inv_comp, one, quadColor[0]); /* R */ + VEC4_MUL(dest[0], inv_comp, dest[0]); /* R */ + VEC4_SUB(inv_comp, one, quadColor[1]); /* G */ + VEC4_MUL(dest[1], inv_comp, dest[1]); /* G */ + VEC4_SUB(inv_comp, one, quadColor[2]); /* B */ + VEC4_MUL(dest[2], inv_comp, dest[2]); /* B */ + } + break; + case PIPE_BLENDFACTOR_INV_SRC_ALPHA: + { + float one_minus_alpha[QUAD_SIZE]; + VEC4_SUB(one_minus_alpha, one, quadColor[3]); + VEC4_MUL(dest[0], dest[0], one_minus_alpha); /* R */ + VEC4_MUL(dest[1], dest[1], one_minus_alpha); /* G */ + VEC4_MUL(dest[2], dest[2], one_minus_alpha); /* B */ + } + break; + case PIPE_BLENDFACTOR_INV_DST_ALPHA: + { + float inv_comp[4]; + VEC4_SUB(inv_comp, one, dest[3]); /* A */ + VEC4_MUL(dest[0], inv_comp, dest[0]); /* R */ + VEC4_MUL(dest[1], inv_comp, dest[1]); /* G */ + VEC4_MUL(dest[2], inv_comp, dest[2]); /* B */ + } + break; + case PIPE_BLENDFACTOR_INV_DST_COLOR: + { + float inv_comp[4]; + VEC4_SUB(inv_comp, one, dest[0]); /* R */ + VEC4_MUL(dest[0], dest[0], inv_comp); /* R */ + VEC4_SUB(inv_comp, one, dest[1]); /* G */ + VEC4_MUL(dest[1], dest[1], inv_comp); /* G */ + VEC4_SUB(inv_comp, one, dest[2]); /* B */ + VEC4_MUL(dest[2], dest[2], inv_comp); /* B */ + } + break; + case PIPE_BLENDFACTOR_INV_CONST_COLOR: + { + float inv_comp[4]; + /* R */ + VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[0]); + VEC4_MUL(dest[0], dest[0], inv_comp); + /* G */ + VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[1]); + VEC4_MUL(dest[1], dest[1], inv_comp); + /* B */ + VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[2]); + VEC4_MUL(dest[2], dest[2], inv_comp); + } + break; + case PIPE_BLENDFACTOR_INV_CONST_ALPHA: + { + float inv_comp[4]; + VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[3]); + VEC4_MUL(dest[0], dest[0], inv_comp); + VEC4_MUL(dest[1], dest[1], inv_comp); + VEC4_MUL(dest[2], dest[2], inv_comp); + } + break; + case PIPE_BLENDFACTOR_INV_SRC1_COLOR: + case PIPE_BLENDFACTOR_INV_SRC1_ALPHA: + /* XXX what are these? */ + assert(0); + break; + default: + assert(0); } - break; - case PIPE_BLENDFACTOR_ZERO: - VEC4_COPY(dest[3], zero); /* A */ - break; - case PIPE_BLENDFACTOR_INV_SRC_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_INV_SRC_ALPHA: - { - float one_minus_alpha[QUAD_SIZE]; - VEC4_SUB(one_minus_alpha, one, quadColor[3]); - VEC4_MUL(dest[3], dest[3], one_minus_alpha); /* A */ + + /* + * Compute dest/second term A + */ + switch (softpipe->blend->alpha_dst_factor) { + case PIPE_BLENDFACTOR_ONE: + /* dest = dest * 1 NO-OP, leave dest as-is */ + break; + case PIPE_BLENDFACTOR_SRC_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_SRC_ALPHA: + VEC4_MUL(dest[3], dest[3], quadColor[3]); /* A * A */ + break; + case PIPE_BLENDFACTOR_DST_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_DST_ALPHA: + VEC4_MUL(dest[3], dest[3], dest[3]); /* A */ + break; + case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: + assert(0); /* illegal */ + break; + case PIPE_BLENDFACTOR_CONST_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_CONST_ALPHA: + { + float comp[4]; + VEC4_SCALAR(comp, softpipe->blend_color.color[3]); /* A */ + VEC4_MUL(dest[3], dest[3], comp); /* A */ + } + break; + case PIPE_BLENDFACTOR_ZERO: + VEC4_COPY(dest[3], zero); /* A */ + break; + case PIPE_BLENDFACTOR_INV_SRC_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_INV_SRC_ALPHA: + { + float one_minus_alpha[QUAD_SIZE]; + VEC4_SUB(one_minus_alpha, one, quadColor[3]); + VEC4_MUL(dest[3], dest[3], one_minus_alpha); /* A */ + } + break; + case PIPE_BLENDFACTOR_INV_DST_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_INV_DST_ALPHA: + { + float inv_comp[4]; + VEC4_SUB(inv_comp, one, dest[3]); /* A */ + VEC4_MUL(dest[3], inv_comp, dest[3]); /* A */ + } + break; + case PIPE_BLENDFACTOR_INV_CONST_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_INV_CONST_ALPHA: + { + float inv_comp[4]; + VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[3]); + VEC4_MUL(dest[3], dest[3], inv_comp); + } + break; + default: + assert(0); } - break; - case PIPE_BLENDFACTOR_INV_DST_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_INV_DST_ALPHA: - { - float inv_comp[4]; - VEC4_SUB(inv_comp, one, dest[3]); /* A */ - VEC4_MUL(dest[3], inv_comp, dest[3]); /* A */ + + /* + * Combine RGB terms + */ + switch (softpipe->blend->rgb_func) { + case PIPE_BLEND_ADD: + VEC4_ADD(quadColor[0], source[0], dest[0]); /* R */ + VEC4_ADD(quadColor[1], source[1], dest[1]); /* G */ + VEC4_ADD(quadColor[2], source[2], dest[2]); /* B */ + break; + case PIPE_BLEND_SUBTRACT: + VEC4_SUB(quadColor[0], source[0], dest[0]); /* R */ + VEC4_SUB(quadColor[1], source[1], dest[1]); /* G */ + VEC4_SUB(quadColor[2], source[2], dest[2]); /* B */ + break; + case PIPE_BLEND_REVERSE_SUBTRACT: + VEC4_SUB(quadColor[0], dest[0], source[0]); /* R */ + VEC4_SUB(quadColor[1], dest[1], source[1]); /* G */ + VEC4_SUB(quadColor[2], dest[2], source[2]); /* B */ + break; + case PIPE_BLEND_MIN: + VEC4_MIN(quadColor[0], source[0], dest[0]); /* R */ + VEC4_MIN(quadColor[1], source[1], dest[1]); /* G */ + VEC4_MIN(quadColor[2], source[2], dest[2]); /* B */ + break; + case PIPE_BLEND_MAX: + VEC4_MAX(quadColor[0], source[0], dest[0]); /* R */ + VEC4_MAX(quadColor[1], source[1], dest[1]); /* G */ + VEC4_MAX(quadColor[2], source[2], dest[2]); /* B */ + break; + default: + assert(0); } - break; - case PIPE_BLENDFACTOR_INV_CONST_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_INV_CONST_ALPHA: - { - float inv_comp[4]; - VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[3]); - VEC4_MUL(dest[3], dest[3], inv_comp); + + /* + * Combine A terms + */ + switch (softpipe->blend->alpha_func) { + case PIPE_BLEND_ADD: + VEC4_ADD(quadColor[3], source[3], dest[3]); /* A */ + break; + case PIPE_BLEND_SUBTRACT: + VEC4_SUB(quadColor[3], source[3], dest[3]); /* A */ + break; + case PIPE_BLEND_REVERSE_SUBTRACT: + VEC4_SUB(quadColor[3], dest[3], source[3]); /* A */ + break; + case PIPE_BLEND_MIN: + VEC4_MIN(quadColor[3], source[3], dest[3]); /* A */ + break; + case PIPE_BLEND_MAX: + VEC4_MAX(quadColor[3], source[3], dest[3]); /* A */ + break; + default: + assert(0); } - break; - default: - assert(0); - } - - /* - * Combine RGB terms - */ - switch (softpipe->blend->rgb_func) { - case PIPE_BLEND_ADD: - VEC4_ADD(quadColor[0], source[0], dest[0]); /* R */ - VEC4_ADD(quadColor[1], source[1], dest[1]); /* G */ - VEC4_ADD(quadColor[2], source[2], dest[2]); /* B */ - break; - case PIPE_BLEND_SUBTRACT: - VEC4_SUB(quadColor[0], source[0], dest[0]); /* R */ - VEC4_SUB(quadColor[1], source[1], dest[1]); /* G */ - VEC4_SUB(quadColor[2], source[2], dest[2]); /* B */ - break; - case PIPE_BLEND_REVERSE_SUBTRACT: - VEC4_SUB(quadColor[0], dest[0], source[0]); /* R */ - VEC4_SUB(quadColor[1], dest[1], source[1]); /* G */ - VEC4_SUB(quadColor[2], dest[2], source[2]); /* B */ - break; - case PIPE_BLEND_MIN: - VEC4_MIN(quadColor[0], source[0], dest[0]); /* R */ - VEC4_MIN(quadColor[1], source[1], dest[1]); /* G */ - VEC4_MIN(quadColor[2], source[2], dest[2]); /* B */ - break; - case PIPE_BLEND_MAX: - VEC4_MAX(quadColor[0], source[0], dest[0]); /* R */ - VEC4_MAX(quadColor[1], source[1], dest[1]); /* G */ - VEC4_MAX(quadColor[2], source[2], dest[2]); /* B */ - break; - default: - assert(0); - } - - /* - * Combine A terms - */ - switch (softpipe->blend->alpha_func) { - case PIPE_BLEND_ADD: - VEC4_ADD(quadColor[3], source[3], dest[3]); /* A */ - break; - case PIPE_BLEND_SUBTRACT: - VEC4_SUB(quadColor[3], source[3], dest[3]); /* A */ - break; - case PIPE_BLEND_REVERSE_SUBTRACT: - VEC4_SUB(quadColor[3], dest[3], source[3]); /* A */ - break; - case PIPE_BLEND_MIN: - VEC4_MIN(quadColor[3], source[3], dest[3]); /* A */ - break; - case PIPE_BLEND_MAX: - VEC4_MAX(quadColor[3], source[3], dest[3]); /* A */ - break; - default: - assert(0); - } - + + } /* cbuf loop */ + /* pass blended quad to next stage */ qs->next->run(qs->next, quad); } diff --git a/src/gallium/drivers/softpipe/sp_quad_bufloop.c b/src/gallium/drivers/softpipe/sp_quad_bufloop.c index 2ae4e22a7d..b3db428ef1 100644 --- a/src/gallium/drivers/softpipe/sp_quad_bufloop.c +++ b/src/gallium/drivers/softpipe/sp_quad_bufloop.c @@ -13,7 +13,7 @@ static void cbuf_loop_quad(struct quad_stage *qs, struct quad_header *quad) { struct softpipe_context *softpipe = qs->softpipe; - float tmp[4][QUAD_SIZE]; + float tmp[PIPE_MAX_COLOR_BUFS][4][QUAD_SIZE]; unsigned i; assert(sizeof(quad->outputs.color) == sizeof(tmp)); @@ -30,7 +30,9 @@ cbuf_loop_quad(struct quad_stage *qs, struct quad_header *quad) for (i = 0; i < softpipe->framebuffer.num_cbufs; i++) { /* set current cbuffer */ +#if 0 /* obsolete & going away */ softpipe->current_cbuf = i; +#endif /* pass blended quad to next stage */ qs->next->run(qs->next, quad); diff --git a/src/gallium/drivers/softpipe/sp_quad_colormask.c b/src/gallium/drivers/softpipe/sp_quad_colormask.c index 1f09d900ca..7fe080990b 100644 --- a/src/gallium/drivers/softpipe/sp_quad_colormask.c +++ b/src/gallium/drivers/softpipe/sp_quad_colormask.c @@ -47,38 +47,43 @@ static void colormask_quad(struct quad_stage *qs, struct quad_header *quad) { struct softpipe_context *softpipe = qs->softpipe; - float dest[4][QUAD_SIZE]; - struct softpipe_cached_tile *tile - = sp_get_cached_tile(softpipe, - softpipe->cbuf_cache[softpipe->current_cbuf], - quad->x0, quad->y0); - float (*quadColor)[4] = quad->outputs.color; - uint i, j; - - /* get/swizzle dest colors */ - for (j = 0; j < QUAD_SIZE; j++) { - int x = (quad->x0 & (TILE_SIZE-1)) + (j & 1); - int y = (quad->y0 & (TILE_SIZE-1)) + (j >> 1); - for (i = 0; i < 4; i++) { - dest[i][j] = tile->data.color[y][x][i]; + uint cbuf; + + /* loop over colorbuffer outputs */ + for (cbuf = 0; cbuf < softpipe->framebuffer.num_cbufs; cbuf++) { + float dest[4][QUAD_SIZE]; + struct softpipe_cached_tile *tile + = sp_get_cached_tile(softpipe, + softpipe->cbuf_cache[cbuf], + quad->x0, quad->y0); + float (*quadColor)[4] = quad->outputs.color[cbuf]; + uint i, j; + + /* get/swizzle dest colors */ + for (j = 0; j < QUAD_SIZE; j++) { + int x = (quad->x0 & (TILE_SIZE-1)) + (j & 1); + int y = (quad->y0 & (TILE_SIZE-1)) + (j >> 1); + for (i = 0; i < 4; i++) { + dest[i][j] = tile->data.color[y][x][i]; + } } - } - /* R */ - if (!(softpipe->blend->colormask & PIPE_MASK_R)) - COPY_4V(quadColor[0], dest[0]); + /* R */ + if (!(softpipe->blend->colormask & PIPE_MASK_R)) + COPY_4V(quadColor[0], dest[0]); - /* G */ - if (!(softpipe->blend->colormask & PIPE_MASK_G)) - COPY_4V(quadColor[1], dest[1]); + /* G */ + if (!(softpipe->blend->colormask & PIPE_MASK_G)) + COPY_4V(quadColor[1], dest[1]); - /* B */ - if (!(softpipe->blend->colormask & PIPE_MASK_B)) - COPY_4V(quadColor[2], dest[2]); + /* B */ + if (!(softpipe->blend->colormask & PIPE_MASK_B)) + COPY_4V(quadColor[2], dest[2]); - /* A */ - if (!(softpipe->blend->colormask & PIPE_MASK_A)) - COPY_4V(quadColor[3], dest[3]); + /* A */ + if (!(softpipe->blend->colormask & PIPE_MASK_A)) + COPY_4V(quadColor[3], dest[3]); + } /* pass quad to next stage */ qs->next->run(qs->next, quad); diff --git a/src/gallium/drivers/softpipe/sp_quad_coverage.c b/src/gallium/drivers/softpipe/sp_quad_coverage.c index b3d3fae22f..dd5ebb2296 100644 --- a/src/gallium/drivers/softpipe/sp_quad_coverage.c +++ b/src/gallium/drivers/softpipe/sp_quad_coverage.c @@ -50,12 +50,17 @@ coverage_quad(struct quad_stage *qs, struct quad_header *quad) if ((softpipe->rasterizer->poly_smooth && quad->prim == PRIM_TRI) || (softpipe->rasterizer->line_smooth && quad->prim == PRIM_LINE) || (softpipe->rasterizer->point_smooth && quad->prim == PRIM_POINT)) { - float (*quadColor)[4] = quad->outputs.color; - unsigned j; - for (j = 0; j < QUAD_SIZE; j++) { - assert(quad->coverage[j] >= 0.0); - assert(quad->coverage[j] <= 1.0); + uint cbuf; + + /* loop over colorbuffer outputs */ + for (cbuf = 0; cbuf < softpipe->framebuffer.num_cbufs; cbuf++) { + float (*quadColor)[4] = quad->outputs.color[cbuf]; + unsigned j; + for (j = 0; j < QUAD_SIZE; j++) { + assert(quad->coverage[j] >= 0.0); + assert(quad->coverage[j] <= 1.0); quadColor[3][j] *= quad->coverage[j]; + } } } diff --git a/src/gallium/drivers/softpipe/sp_quad_fs.c b/src/gallium/drivers/softpipe/sp_quad_fs.c index a73df31383..8c88c192f8 100644 --- a/src/gallium/drivers/softpipe/sp_quad_fs.c +++ b/src/gallium/drivers/softpipe/sp_quad_fs.c @@ -53,7 +53,6 @@ struct quad_shade_stage struct tgsi_sampler samplers[PIPE_MAX_SAMPLERS]; struct tgsi_exec_machine machine; struct tgsi_exec_vector *inputs, *outputs; - int colorOutSlot, depthOutSlot; }; @@ -77,7 +76,8 @@ shade_quad( struct quad_shade_stage *qss = quad_shade_stage( qs ); struct softpipe_context *softpipe = qs->softpipe; struct tgsi_exec_machine *machine = &qss->machine; - + boolean z_written; + /* Consts do not require 16 byte alignment. */ machine->Consts = softpipe->mapped_constants[PIPE_SHADER_FRAGMENT]; @@ -88,31 +88,37 @@ shade_quad( &qss->machine, quad ); - /* store result color */ - if (qss->colorOutSlot >= 0) { - /* XXX need to handle multiple color outputs someday */ - assert(qss->stage.softpipe->fs->info.output_semantic_name[qss->colorOutSlot] - == TGSI_SEMANTIC_COLOR); - memcpy( - quad->outputs.color, - &machine->Outputs[qss->colorOutSlot].xyzw[0].f[0], - sizeof( quad->outputs.color ) ); - } - - /* - * XXX the following code for updating quad->outputs.depth - * isn't really needed if we did early z testing. - */ - - /* store result Z */ - if (qss->depthOutSlot >= 0) { - /* output[slot] is new Z */ + /* store outputs */ + z_written = FALSE; + { + const ubyte *sem_name = softpipe->fs->info.output_semantic_name; + const ubyte *sem_index = softpipe->fs->info.output_semantic_index; + const uint n = qss->stage.softpipe->fs->info.num_outputs; uint i; - for (i = 0; i < 4; i++) { - quad->outputs.depth[i] = machine->Outputs[0].xyzw[2].f[i]; + for (i = 0; i < n; i++) { + switch (sem_name[i]) { + case TGSI_SEMANTIC_COLOR: + { + uint cbuf = sem_index[i]; + memcpy(quad->outputs.color[cbuf], + &machine->Outputs[i].xyzw[0].f[0], + sizeof(quad->outputs.color[0]) ); + } + break; + case TGSI_SEMANTIC_POSITION: + { + uint j; + for (j = 0; j < 4; j++) { + quad->outputs.depth[j] = machine->Outputs[0].xyzw[2].f[j]; + } + z_written = TRUE; + } + break; + } } } - else { + + if (!z_written) { /* compute Z values now, as in the quad earlyz stage */ /* XXX we should really only do this if the earlyz stage is not used */ const float fx = (float) quad->x0; @@ -149,20 +155,6 @@ static void shade_begin(struct quad_stage *qs) qss->samplers[i].texture = softpipe->texture[i]; } - /* find output slots for depth, color */ - qss->colorOutSlot = -1; - qss->depthOutSlot = -1; - for (i = 0; i < qss->stage.softpipe->fs->info.num_outputs; i++) { - switch (qss->stage.softpipe->fs->info.output_semantic_name[i]) { - case TGSI_SEMANTIC_POSITION: - qss->depthOutSlot = i; - break; - case TGSI_SEMANTIC_COLOR: - qss->colorOutSlot = i; - break; - } - } - softpipe->fs->prepare( softpipe->fs, &qss->machine, qss->samplers ); diff --git a/src/gallium/drivers/softpipe/sp_quad_output.c b/src/gallium/drivers/softpipe/sp_quad_output.c index cfe8f11808..40083138a4 100644 --- a/src/gallium/drivers/softpipe/sp_quad_output.c +++ b/src/gallium/drivers/softpipe/sp_quad_output.c @@ -34,31 +34,36 @@ /** - * Write quad to framebuffer, taking mask into account. - * - * Note that surfaces support only full quad reads and writes. + * Last step of quad processing: write quad colors to the framebuffer, + * taking mask into account. */ static void output_quad(struct quad_stage *qs, struct quad_header *quad) { - struct softpipe_context *softpipe = qs->softpipe; - struct softpipe_cached_tile *tile - = sp_get_cached_tile(softpipe, - softpipe->cbuf_cache[softpipe->current_cbuf], - quad->x0, quad->y0); /* in-tile pos: */ const int itx = quad->x0 % TILE_SIZE; const int ity = quad->y0 % TILE_SIZE; - float (*quadColor)[4] = quad->outputs.color; - int i, j; - /* get/swizzle dest colors */ - for (j = 0; j < QUAD_SIZE; j++) { - if (quad->mask & (1 << j)) { - int x = itx + (j & 1); - int y = ity + (j >> 1); - for (i = 0; i < 4; i++) { /* loop over color chans */ - tile->data.color[y][x][i] = quadColor[i][j]; + struct softpipe_context *softpipe = qs->softpipe; + uint cbuf; + + /* loop over colorbuffer outputs */ + for (cbuf = 0; cbuf < softpipe->framebuffer.num_cbufs; cbuf++) { + struct softpipe_cached_tile *tile + = sp_get_cached_tile(softpipe, + softpipe->cbuf_cache[cbuf], + quad->x0, quad->y0); + float (*quadColor)[4] = quad->outputs.color[cbuf]; + int i, j; + + /* get/swizzle dest colors */ + for (j = 0; j < QUAD_SIZE; j++) { + if (quad->mask & (1 << j)) { + int x = itx + (j & 1); + int y = ity + (j >> 1); + for (i = 0; i < 4; i++) { /* loop over color chans */ + tile->data.color[y][x][i] = quadColor[i][j]; + } } } } diff --git a/src/gallium/drivers/softpipe/sp_quad_stencil.c b/src/gallium/drivers/softpipe/sp_quad_stencil.c index 92a0da0083..b4c7e942fa 100644 --- a/src/gallium/drivers/softpipe/sp_quad_stencil.c +++ b/src/gallium/drivers/softpipe/sp_quad_stencil.c @@ -243,7 +243,7 @@ stencil_test_quad(struct quad_stage *qs, struct quad_header *quad) stencilVals[j] = tile->data.depth32[y][x] & 0xff; } break; - case PIPE_FORMAT_U_S8: + case PIPE_FORMAT_S8_UNORM: for (j = 0; j < QUAD_SIZE; j++) { int x = quad->x0 % TILE_SIZE + (j & 1); int y = quad->y0 % TILE_SIZE + (j >> 1); @@ -311,7 +311,7 @@ stencil_test_quad(struct quad_stage *qs, struct quad_header *quad) tile->data.depth32[y][x] = z24s8; } break; - case PIPE_FORMAT_U_S8: + case PIPE_FORMAT_S8_UNORM: for (j = 0; j < QUAD_SIZE; j++) { int x = quad->x0 % TILE_SIZE + (j & 1); int y = quad->y0 % TILE_SIZE + (j >> 1); diff --git a/src/gallium/drivers/softpipe/sp_screen.c b/src/gallium/drivers/softpipe/sp_screen.c index 1850a1ced3..7dacb1c461 100644 --- a/src/gallium/drivers/softpipe/sp_screen.c +++ b/src/gallium/drivers/softpipe/sp_screen.c @@ -68,7 +68,7 @@ softpipe_get_param(struct pipe_screen *screen, int param) case PIPE_CAP_POINT_SPRITE: return 1; case PIPE_CAP_MAX_RENDER_TARGETS: - return 1; + return PIPE_MAX_COLOR_BUFS; case PIPE_CAP_OCCLUSION_QUERY: return 1; case PIPE_CAP_TEXTURE_SHADOW_MAP: diff --git a/src/gallium/drivers/softpipe/sp_setup.c b/src/gallium/drivers/softpipe/sp_setup.c new file mode 100644 index 0000000000..543d86a5cb --- /dev/null +++ b/src/gallium/drivers/softpipe/sp_setup.c @@ -0,0 +1,1308 @@ +/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +/** + * \brief Primitive rasterization/rendering (points, lines, triangles) + * + * \author Keith Whitwell <keith@tungstengraphics.com> + * \author Brian Paul + */ + +#include "sp_setup.h" + +#include "sp_context.h" +#include "sp_headers.h" +#include "sp_quad.h" +#include "sp_state.h" +#include "sp_prim_setup.h" +#include "draw/draw_private.h" +#include "draw/draw_vertex.h" +#include "pipe/p_util.h" +#include "pipe/p_shader_tokens.h" + +#define DEBUG_VERTS 0 +#define DEBUG_FRAGS 0 + +/** + * Triangle edge info + */ +struct edge { + float dx; /**< X(v1) - X(v0), used only during setup */ + float dy; /**< Y(v1) - Y(v0), used only during setup */ + float dxdy; /**< dx/dy */ + float sx, sy; /**< first sample point coord */ + int lines; /**< number of lines on this edge */ +}; + + +/** + * Triangle setup info (derived from draw_stage). + * Also used for line drawing (taking some liberties). + */ +struct setup_context { + struct softpipe_context *softpipe; + + /* Vertices are just an array of floats making up each attribute in + * turn. Currently fixed at 4 floats, but should change in time. + * Codegen will help cope with this. + */ + const float (*vmax)[4]; + const float (*vmid)[4]; + const float (*vmin)[4]; + const float (*vprovoke)[4]; + + struct edge ebot; + struct edge etop; + struct edge emaj; + + float oneoverarea; + + struct tgsi_interp_coef coef[PIPE_MAX_SHADER_INPUTS]; + struct tgsi_interp_coef posCoef; /* For Z, W */ + struct quad_header quad; + + struct { + int left[2]; /**< [0] = row0, [1] = row1 */ + int right[2]; + int y; + unsigned y_flags; + unsigned mask; /**< mask of MASK_BOTTOM/TOP_LEFT/RIGHT bits */ + } span; + +#if DEBUG_FRAGS + uint numFragsEmitted; /**< per primitive */ + uint numFragsWritten; /**< per primitive */ +#endif + + unsigned winding; /* which winding to cull */ +}; + + + + + +static boolean cull_tri( struct setup_context *setup, + float det ) +{ + if (det != 0) + { + /* if (det < 0 then Z points toward camera and triangle is + * counter-clockwise winding. + */ + unsigned winding = (det < 0) ? PIPE_WINDING_CCW : PIPE_WINDING_CW; + + if ((winding & setup->winding) == 0) + return FALSE; + } + + /* Culled: + */ + return TRUE; +} + + + +/** + * Clip setup->quad against the scissor/surface bounds. + */ +static INLINE void +quad_clip(struct setup_context *setup) +{ + const struct pipe_scissor_state *cliprect = &setup->softpipe->cliprect; + const int minx = (int) cliprect->minx; + const int maxx = (int) cliprect->maxx; + const int miny = (int) cliprect->miny; + const int maxy = (int) cliprect->maxy; + + if (setup->quad.x0 >= maxx || + setup->quad.y0 >= maxy || + setup->quad.x0 + 1 < minx || + setup->quad.y0 + 1 < miny) { + /* totally clipped */ + setup->quad.mask = 0x0; + return; + } + if (setup->quad.x0 < minx) + setup->quad.mask &= (MASK_BOTTOM_RIGHT | MASK_TOP_RIGHT); + if (setup->quad.y0 < miny) + setup->quad.mask &= (MASK_BOTTOM_LEFT | MASK_BOTTOM_RIGHT); + if (setup->quad.x0 == maxx - 1) + setup->quad.mask &= (MASK_BOTTOM_LEFT | MASK_TOP_LEFT); + if (setup->quad.y0 == maxy - 1) + setup->quad.mask &= (MASK_TOP_LEFT | MASK_TOP_RIGHT); +} + + +/** + * Emit a quad (pass to next stage) with clipping. + */ +static INLINE void +clip_emit_quad(struct setup_context *setup) +{ + quad_clip(setup); + if (setup->quad.mask) { + struct softpipe_context *sp = setup->softpipe; + sp->quad.first->run(sp->quad.first, &setup->quad); + } +} + + +/** + * Emit a quad (pass to next stage). No clipping is done. + */ +static INLINE void +emit_quad( struct setup_context *setup, int x, int y, unsigned mask ) +{ + struct softpipe_context *sp = setup->softpipe; + setup->quad.x0 = x; + setup->quad.y0 = y; + setup->quad.mask = mask; +#if DEBUG_FRAGS + if (mask & 1) setup->numFragsEmitted++; + if (mask & 2) setup->numFragsEmitted++; + if (mask & 4) setup->numFragsEmitted++; + if (mask & 8) setup->numFragsEmitted++; +#endif + sp->quad.first->run(sp->quad.first, &setup->quad); +#if DEBUG_FRAGS + mask = setup->quad.mask; + if (mask & 1) setup->numFragsWritten++; + if (mask & 2) setup->numFragsWritten++; + if (mask & 4) setup->numFragsWritten++; + if (mask & 8) setup->numFragsWritten++; +#endif +} + + +/** + * Given an X or Y coordinate, return the block/quad coordinate that it + * belongs to. + */ +static INLINE int block( int x ) +{ + return x & ~1; +} + + +/** + * Render a horizontal span of quads + */ +static void flush_spans( struct setup_context *setup ) +{ + const int xleft0 = setup->span.left[0]; + const int xleft1 = setup->span.left[1]; + const int xright0 = setup->span.right[0]; + const int xright1 = setup->span.right[1]; + int minleft, maxright; + int x; + + switch (setup->span.y_flags) { + case 0x3: + /* both odd and even lines written (both quad rows) */ + minleft = block(MIN2(xleft0, xleft1)); + maxright = block(MAX2(xright0, xright1)); + for (x = minleft; x <= maxright; x += 2) { + /* determine which of the four pixels is inside the span bounds */ + uint mask = 0x0; + if (x >= xleft0 && x < xright0) + mask |= MASK_TOP_LEFT; + if (x >= xleft1 && x < xright1) + mask |= MASK_BOTTOM_LEFT; + if (x+1 >= xleft0 && x+1 < xright0) + mask |= MASK_TOP_RIGHT; + if (x+1 >= xleft1 && x+1 < xright1) + mask |= MASK_BOTTOM_RIGHT; + emit_quad( setup, x, setup->span.y, mask ); + } + break; + + case 0x1: + /* only even line written (quad top row) */ + minleft = block(xleft0); + maxright = block(xright0); + for (x = minleft; x <= maxright; x += 2) { + uint mask = 0x0; + if (x >= xleft0 && x < xright0) + mask |= MASK_TOP_LEFT; + if (x+1 >= xleft0 && x+1 < xright0) + mask |= MASK_TOP_RIGHT; + emit_quad( setup, x, setup->span.y, mask ); + } + break; + + case 0x2: + /* only odd line written (quad bottom row) */ + minleft = block(xleft1); + maxright = block(xright1); + for (x = minleft; x <= maxright; x += 2) { + uint mask = 0x0; + if (x >= xleft1 && x < xright1) + mask |= MASK_BOTTOM_LEFT; + if (x+1 >= xleft1 && x+1 < xright1) + mask |= MASK_BOTTOM_RIGHT; + emit_quad( setup, x, setup->span.y, mask ); + } + break; + + default: + return; + } + + setup->span.y = 0; + setup->span.y_flags = 0; + setup->span.right[0] = 0; + setup->span.right[1] = 0; +} + + +#if DEBUG_VERTS +static void print_vertex(const struct setup_context *setup, + const float (*v)[4]) +{ + int i; + debug_printf("Vertex: (%p)\n", v); + for (i = 0; i < setup->quad.nr_attrs; i++) { + debug_printf(" %d: %f %f %f %f\n", i, + v[i][0], v[i][1], v[i][2], v[i][3]); + } +} +#endif + +static boolean setup_sort_vertices( struct setup_context *setup, + float det, + const float (*v0)[4], + const float (*v1)[4], + const float (*v2)[4] ) +{ + setup->vprovoke = v2; + + /* determine bottom to top order of vertices */ + { + float y0 = v0[0][1]; + float y1 = v1[0][1]; + float y2 = v2[0][1]; + if (y0 <= y1) { + if (y1 <= y2) { + /* y0<=y1<=y2 */ + setup->vmin = v0; + setup->vmid = v1; + setup->vmax = v2; + } + else if (y2 <= y0) { + /* y2<=y0<=y1 */ + setup->vmin = v2; + setup->vmid = v0; + setup->vmax = v1; + } + else { + /* y0<=y2<=y1 */ + setup->vmin = v0; + setup->vmid = v2; + setup->vmax = v1; + } + } + else { + if (y0 <= y2) { + /* y1<=y0<=y2 */ + setup->vmin = v1; + setup->vmid = v0; + setup->vmax = v2; + } + else if (y2 <= y1) { + /* y2<=y1<=y0 */ + setup->vmin = v2; + setup->vmid = v1; + setup->vmax = v0; + } + else { + /* y1<=y2<=y0 */ + setup->vmin = v1; + setup->vmid = v2; + setup->vmax = v0; + } + } + } + + setup->ebot.dx = setup->vmid[0][0] - setup->vmin[0][0]; + setup->ebot.dy = setup->vmid[0][1] - setup->vmin[0][1]; + setup->emaj.dx = setup->vmax[0][0] - setup->vmin[0][0]; + setup->emaj.dy = setup->vmax[0][1] - setup->vmin[0][1]; + setup->etop.dx = setup->vmax[0][0] - setup->vmid[0][0]; + setup->etop.dy = setup->vmax[0][1] - setup->vmid[0][1]; + + /* + * Compute triangle's area. Use 1/area to compute partial + * derivatives of attributes later. + * + * The area will be the same as prim->det, but the sign may be + * different depending on how the vertices get sorted above. + * + * To determine whether the primitive is front or back facing we + * use the prim->det value because its sign is correct. + */ + { + const float area = (setup->emaj.dx * setup->ebot.dy - + setup->ebot.dx * setup->emaj.dy); + + setup->oneoverarea = 1.0f / area; + /* + debug_printf("%s one-over-area %f area %f det %f\n", + __FUNCTION__, setup->oneoverarea, area, det ); + */ + } + + /* We need to know if this is a front or back-facing triangle for: + * - the GLSL gl_FrontFacing fragment attribute (bool) + * - two-sided stencil test + */ + setup->quad.facing = (det > 0.0) ^ (setup->softpipe->rasterizer->front_winding == PIPE_WINDING_CW); + + return TRUE; +} + + +/** + * Compute a0 for a constant-valued coefficient (GL_FLAT shading). + * The value value comes from vertex[slot][i]. + * The result will be put into setup->coef[slot].a0[i]. + * \param slot which attribute slot + * \param i which component of the slot (0..3) + */ +static void const_coeff( struct setup_context *setup, + struct tgsi_interp_coef *coef, + uint vertSlot, uint i) +{ + assert(i <= 3); + + coef->dadx[i] = 0; + coef->dady[i] = 0; + + /* need provoking vertex info! + */ + coef->a0[i] = setup->vprovoke[vertSlot][i]; +} + + +/** + * Compute a0, dadx and dady for a linearly interpolated coefficient, + * for a triangle. + */ +static void tri_linear_coeff( struct setup_context *setup, + struct tgsi_interp_coef *coef, + uint vertSlot, uint i) +{ + float botda = setup->vmid[vertSlot][i] - setup->vmin[vertSlot][i]; + float majda = setup->vmax[vertSlot][i] - setup->vmin[vertSlot][i]; + float a = setup->ebot.dy * majda - botda * setup->emaj.dy; + float b = setup->emaj.dx * botda - majda * setup->ebot.dx; + float dadx = a * setup->oneoverarea; + float dady = b * setup->oneoverarea; + + assert(i <= 3); + + coef->dadx[i] = dadx; + coef->dady[i] = dady; + + /* calculate a0 as the value which would be sampled for the + * fragment at (0,0), taking into account that we want to sample at + * pixel centers, in other words (0.5, 0.5). + * + * this is neat but unfortunately not a good way to do things for + * triangles with very large values of dadx or dady as it will + * result in the subtraction and re-addition from a0 of a very + * large number, which means we'll end up loosing a lot of the + * fractional bits and precision from a0. the way to fix this is + * to define a0 as the sample at a pixel center somewhere near vmin + * instead - i'll switch to this later. + */ + coef->a0[i] = (setup->vmin[vertSlot][i] - + (dadx * (setup->vmin[0][0] - 0.5f) + + dady * (setup->vmin[0][1] - 0.5f))); + + /* + debug_printf("attr[%d].%c: %f dx:%f dy:%f\n", + slot, "xyzw"[i], + setup->coef[slot].a0[i], + setup->coef[slot].dadx[i], + setup->coef[slot].dady[i]); + */ +} + + +/** + * Compute a0, dadx and dady for a perspective-corrected interpolant, + * for a triangle. + * We basically multiply the vertex value by 1/w before computing + * the plane coefficients (a0, dadx, dady). + * Later, when we compute the value at a particular fragment position we'll + * divide the interpolated value by the interpolated W at that fragment. + */ +static void tri_persp_coeff( struct setup_context *setup, + struct tgsi_interp_coef *coef, + uint vertSlot, uint i) +{ + /* premultiply by 1/w (v[0][3] is always W): + */ + float mina = setup->vmin[vertSlot][i] * setup->vmin[0][3]; + float mida = setup->vmid[vertSlot][i] * setup->vmid[0][3]; + float maxa = setup->vmax[vertSlot][i] * setup->vmax[0][3]; + float botda = mida - mina; + float majda = maxa - mina; + float a = setup->ebot.dy * majda - botda * setup->emaj.dy; + float b = setup->emaj.dx * botda - majda * setup->ebot.dx; + float dadx = a * setup->oneoverarea; + float dady = b * setup->oneoverarea; + + /* + debug_printf("tri persp %d,%d: %f %f %f\n", vertSlot, i, + setup->vmin[vertSlot][i], + setup->vmid[vertSlot][i], + setup->vmax[vertSlot][i] + ); + */ + assert(i <= 3); + + coef->dadx[i] = dadx; + coef->dady[i] = dady; + coef->a0[i] = (mina - + (dadx * (setup->vmin[0][0] - 0.5f) + + dady * (setup->vmin[0][1] - 0.5f))); +} + + +/** + * Special coefficient setup for gl_FragCoord. + * X and Y are trivial, though Y has to be inverted for OpenGL. + * Z and W are copied from posCoef which should have already been computed. + * We could do a bit less work if we'd examine gl_FragCoord's swizzle mask. + */ +static void +setup_fragcoord_coeff(struct setup_context *setup, uint slot) +{ + /*X*/ + setup->coef[slot].a0[0] = 0; + setup->coef[slot].dadx[0] = 1.0; + setup->coef[slot].dady[0] = 0.0; + /*Y*/ + if (setup->softpipe->rasterizer->origin_lower_left) { + /* y=0=bottom */ + const int winHeight = setup->softpipe->framebuffer.height; + setup->coef[slot].a0[1] = (float) (winHeight - 1); + setup->coef[slot].dady[1] = -1.0; + } + else { + /* y=0=top */ + setup->coef[slot].a0[1] = 0.0; + setup->coef[slot].dady[1] = 1.0; + } + setup->coef[slot].dadx[1] = 0.0; + /*Z*/ + setup->coef[slot].a0[2] = setup->posCoef.a0[2]; + setup->coef[slot].dadx[2] = setup->posCoef.dadx[2]; + setup->coef[slot].dady[2] = setup->posCoef.dady[2]; + /*W*/ + setup->coef[slot].a0[3] = setup->posCoef.a0[3]; + setup->coef[slot].dadx[3] = setup->posCoef.dadx[3]; + setup->coef[slot].dady[3] = setup->posCoef.dady[3]; +} + + + +/** + * Compute the setup->coef[] array dadx, dady, a0 values. + * Must be called after setup->vmin,vmid,vmax,vprovoke are initialized. + */ +static void setup_tri_coefficients( struct setup_context *setup ) +{ + struct softpipe_context *softpipe = setup->softpipe; + const struct sp_fragment_shader *spfs = softpipe->fs; + const struct vertex_info *vinfo = softpipe_get_vertex_info(softpipe); + uint fragSlot; + + /* z and w are done by linear interpolation: + */ + tri_linear_coeff(setup, &setup->posCoef, 0, 2); + tri_linear_coeff(setup, &setup->posCoef, 0, 3); + + /* setup interpolation for all the remaining attributes: + */ + for (fragSlot = 0; fragSlot < spfs->info.num_inputs; fragSlot++) { + const uint vertSlot = vinfo->src_index[fragSlot]; + uint j; + + switch (vinfo->interp_mode[fragSlot]) { + case INTERP_CONSTANT: + for (j = 0; j < NUM_CHANNELS; j++) + const_coeff(setup, &setup->coef[fragSlot], vertSlot, j); + break; + case INTERP_LINEAR: + for (j = 0; j < NUM_CHANNELS; j++) + tri_linear_coeff(setup, &setup->coef[fragSlot], vertSlot, j); + break; + case INTERP_PERSPECTIVE: + for (j = 0; j < NUM_CHANNELS; j++) + tri_persp_coeff(setup, &setup->coef[fragSlot], vertSlot, j); + break; + case INTERP_POS: + setup_fragcoord_coeff(setup, fragSlot); + break; + default: + assert(0); + } + + if (spfs->info.input_semantic_name[fragSlot] == TGSI_SEMANTIC_FOG) { + /* FOG.y = front/back facing XXX fix this */ + setup->coef[fragSlot].a0[1] = 1.0f - setup->quad.facing; + setup->coef[fragSlot].dadx[1] = 0.0; + setup->coef[fragSlot].dady[1] = 0.0; + } + } +} + + + +static void setup_tri_edges( struct setup_context *setup ) +{ + float vmin_x = setup->vmin[0][0] + 0.5f; + float vmid_x = setup->vmid[0][0] + 0.5f; + + float vmin_y = setup->vmin[0][1] - 0.5f; + float vmid_y = setup->vmid[0][1] - 0.5f; + float vmax_y = setup->vmax[0][1] - 0.5f; + + setup->emaj.sy = CEILF(vmin_y); + setup->emaj.lines = (int) CEILF(vmax_y - setup->emaj.sy); + setup->emaj.dxdy = setup->emaj.dx / setup->emaj.dy; + setup->emaj.sx = vmin_x + (setup->emaj.sy - vmin_y) * setup->emaj.dxdy; + + setup->etop.sy = CEILF(vmid_y); + setup->etop.lines = (int) CEILF(vmax_y - setup->etop.sy); + setup->etop.dxdy = setup->etop.dx / setup->etop.dy; + setup->etop.sx = vmid_x + (setup->etop.sy - vmid_y) * setup->etop.dxdy; + + setup->ebot.sy = CEILF(vmin_y); + setup->ebot.lines = (int) CEILF(vmid_y - setup->ebot.sy); + setup->ebot.dxdy = setup->ebot.dx / setup->ebot.dy; + setup->ebot.sx = vmin_x + (setup->ebot.sy - vmin_y) * setup->ebot.dxdy; +} + + +/** + * Render the upper or lower half of a triangle. + * Scissoring/cliprect is applied here too. + */ +static void subtriangle( struct setup_context *setup, + struct edge *eleft, + struct edge *eright, + unsigned lines ) +{ + const struct pipe_scissor_state *cliprect = &setup->softpipe->cliprect; + const int minx = (int) cliprect->minx; + const int maxx = (int) cliprect->maxx; + const int miny = (int) cliprect->miny; + const int maxy = (int) cliprect->maxy; + int y, start_y, finish_y; + int sy = (int)eleft->sy; + + assert((int)eleft->sy == (int) eright->sy); + + /* clip top/bottom */ + start_y = sy; + finish_y = sy + lines; + + if (start_y < miny) + start_y = miny; + + if (finish_y > maxy) + finish_y = maxy; + + start_y -= sy; + finish_y -= sy; + + /* + debug_printf("%s %d %d\n", __FUNCTION__, start_y, finish_y); + */ + + for (y = start_y; y < finish_y; y++) { + + /* avoid accumulating adds as floats don't have the precision to + * accurately iterate large triangle edges that way. luckily we + * can just multiply these days. + * + * this is all drowned out by the attribute interpolation anyway. + */ + int left = (int)(eleft->sx + y * eleft->dxdy); + int right = (int)(eright->sx + y * eright->dxdy); + + /* clip left/right */ + if (left < minx) + left = minx; + if (right > maxx) + right = maxx; + + if (left < right) { + int _y = sy + y; + if (block(_y) != setup->span.y) { + flush_spans(setup); + setup->span.y = block(_y); + } + + setup->span.left[_y&1] = left; + setup->span.right[_y&1] = right; + setup->span.y_flags |= 1<<(_y&1); + } + } + + + /* save the values so that emaj can be restarted: + */ + eleft->sx += lines * eleft->dxdy; + eright->sx += lines * eright->dxdy; + eleft->sy += lines; + eright->sy += lines; +} + + +/** + * Recalculate prim's determinant. This is needed as we don't have + * get this information through the vbuf_render interface & we must + * calculate it here. + */ +static float +calc_det( const float (*v0)[4], + const float (*v1)[4], + const float (*v2)[4] ) +{ + /* edge vectors e = v0 - v2, f = v1 - v2 */ + const float ex = v0[0][0] - v2[0][0]; + const float ey = v0[0][1] - v2[0][1]; + const float fx = v1[0][0] - v2[0][0]; + const float fy = v1[0][1] - v2[0][1]; + + /* det = cross(e,f).z */ + return ex * fy - ey * fx; +} + + +/** + * Do setup for triangle rasterization, then render the triangle. + */ +void setup_tri( struct setup_context *setup, + const float (*v0)[4], + const float (*v1)[4], + const float (*v2)[4] ) +{ + float det; + +#if DEBUG_VERTS + debug_printf("Setup triangle:\n"); + print_vertex(setup, v0); + print_vertex(setup, v1); + print_vertex(setup, v2); +#endif + + if (setup->softpipe->no_rast) + return; + + det = calc_det(v0, v1, v2); + /* + debug_printf("%s\n", __FUNCTION__ ); + */ + +#if DEBUG_FRAGS + setup->numFragsEmitted = 0; + setup->numFragsWritten = 0; +#endif + + if (cull_tri( setup, det )) + return; + + setup_sort_vertices( setup, det, v0, v1, v2 ); + setup_tri_coefficients( setup ); + setup_tri_edges( setup ); + + setup->quad.prim = PRIM_TRI; + + setup->span.y = 0; + setup->span.y_flags = 0; + setup->span.right[0] = 0; + setup->span.right[1] = 0; + /* setup->span.z_mode = tri_z_mode( setup->ctx ); */ + + /* init_constant_attribs( setup ); */ + + if (setup->oneoverarea < 0.0) { + /* emaj on left: + */ + subtriangle( setup, &setup->emaj, &setup->ebot, setup->ebot.lines ); + subtriangle( setup, &setup->emaj, &setup->etop, setup->etop.lines ); + } + else { + /* emaj on right: + */ + subtriangle( setup, &setup->ebot, &setup->emaj, setup->ebot.lines ); + subtriangle( setup, &setup->etop, &setup->emaj, setup->etop.lines ); + } + + flush_spans( setup ); + +#if DEBUG_FRAGS + printf("Tri: %u frags emitted, %u written\n", + setup->numFragsEmitted, + setup->numFragsWritten); +#endif +} + + + +/** + * Compute a0, dadx and dady for a linearly interpolated coefficient, + * for a line. + */ +static void +line_linear_coeff(struct setup_context *setup, + struct tgsi_interp_coef *coef, + uint vertSlot, uint i) +{ + const float da = setup->vmax[vertSlot][i] - setup->vmin[vertSlot][i]; + const float dadx = da * setup->emaj.dx * setup->oneoverarea; + const float dady = da * setup->emaj.dy * setup->oneoverarea; + coef->dadx[i] = dadx; + coef->dady[i] = dady; + coef->a0[i] = (setup->vmin[vertSlot][i] - + (dadx * (setup->vmin[0][0] - 0.5f) + + dady * (setup->vmin[0][1] - 0.5f))); +} + + +/** + * Compute a0, dadx and dady for a perspective-corrected interpolant, + * for a line. + */ +static void +line_persp_coeff(struct setup_context *setup, + struct tgsi_interp_coef *coef, + uint vertSlot, uint i) +{ + /* XXX double-check/verify this arithmetic */ + const float a0 = setup->vmin[vertSlot][i] * setup->vmin[0][3]; + const float a1 = setup->vmax[vertSlot][i] * setup->vmax[0][3]; + const float da = a1 - a0; + const float dadx = da * setup->emaj.dx * setup->oneoverarea; + const float dady = da * setup->emaj.dy * setup->oneoverarea; + coef->dadx[i] = dadx; + coef->dady[i] = dady; + coef->a0[i] = (setup->vmin[vertSlot][i] - + (dadx * (setup->vmin[0][0] - 0.5f) + + dady * (setup->vmin[0][1] - 0.5f))); +} + + +/** + * Compute the setup->coef[] array dadx, dady, a0 values. + * Must be called after setup->vmin,vmax are initialized. + */ +static INLINE void +setup_line_coefficients(struct setup_context *setup, + const float (*v0)[4], + const float (*v1)[4]) +{ + struct softpipe_context *softpipe = setup->softpipe; + const struct sp_fragment_shader *spfs = softpipe->fs; + const struct vertex_info *vinfo = softpipe_get_vertex_info(softpipe); + uint fragSlot; + + /* use setup->vmin, vmax to point to vertices */ + setup->vprovoke = v1; + setup->vmin = v0; + setup->vmax = v1; + + setup->emaj.dx = setup->vmax[0][0] - setup->vmin[0][0]; + setup->emaj.dy = setup->vmax[0][1] - setup->vmin[0][1]; + /* NOTE: this is not really 1/area */ + setup->oneoverarea = 1.0f / (setup->emaj.dx * setup->emaj.dx + + setup->emaj.dy * setup->emaj.dy); + + /* z and w are done by linear interpolation: + */ + line_linear_coeff(setup, &setup->posCoef, 0, 2); + line_linear_coeff(setup, &setup->posCoef, 0, 3); + + /* setup interpolation for all the remaining attributes: + */ + for (fragSlot = 0; fragSlot < spfs->info.num_inputs; fragSlot++) { + const uint vertSlot = vinfo->src_index[fragSlot]; + uint j; + + switch (vinfo->interp_mode[fragSlot]) { + case INTERP_CONSTANT: + for (j = 0; j < NUM_CHANNELS; j++) + const_coeff(setup, &setup->coef[fragSlot], vertSlot, j); + break; + case INTERP_LINEAR: + for (j = 0; j < NUM_CHANNELS; j++) + line_linear_coeff(setup, &setup->coef[fragSlot], vertSlot, j); + break; + case INTERP_PERSPECTIVE: + for (j = 0; j < NUM_CHANNELS; j++) + line_persp_coeff(setup, &setup->coef[fragSlot], vertSlot, j); + break; + case INTERP_POS: + setup_fragcoord_coeff(setup, fragSlot); + break; + default: + assert(0); + } + + if (spfs->info.input_semantic_name[fragSlot] == TGSI_SEMANTIC_FOG) { + /* FOG.y = front/back facing XXX fix this */ + setup->coef[fragSlot].a0[1] = 1.0f - setup->quad.facing; + setup->coef[fragSlot].dadx[1] = 0.0; + setup->coef[fragSlot].dady[1] = 0.0; + } + } +} + + +/** + * Plot a pixel in a line segment. + */ +static INLINE void +plot(struct setup_context *setup, int x, int y) +{ + const int iy = y & 1; + const int ix = x & 1; + const int quadX = x - ix; + const int quadY = y - iy; + const int mask = (1 << ix) << (2 * iy); + + if (quadX != setup->quad.x0 || + quadY != setup->quad.y0) + { + /* flush prev quad, start new quad */ + + if (setup->quad.x0 != -1) + clip_emit_quad(setup); + + setup->quad.x0 = quadX; + setup->quad.y0 = quadY; + setup->quad.mask = 0x0; + } + + setup->quad.mask |= mask; +} + + +/** + * Do setup for line rasterization, then render the line. + * Single-pixel width, no stipple, etc. We rely on the 'draw' module + * to handle stippling and wide lines. + */ +void +setup_line(struct setup_context *setup, + const float (*v0)[4], + const float (*v1)[4]) +{ + int x0 = (int) v0[0][0]; + int x1 = (int) v1[0][0]; + int y0 = (int) v0[0][1]; + int y1 = (int) v1[0][1]; + int dx = x1 - x0; + int dy = y1 - y0; + int xstep, ystep; + +#if DEBUG_VERTS + debug_printf("Setup line:\n"); + print_vertex(setup, v0); + print_vertex(setup, v1); +#endif + + if (setup->softpipe->no_rast) + return; + + if (dx == 0 && dy == 0) + return; + + setup_line_coefficients(setup, v0, v1); + + if (dx < 0) { + dx = -dx; /* make positive */ + xstep = -1; + } + else { + xstep = 1; + } + + if (dy < 0) { + dy = -dy; /* make positive */ + ystep = -1; + } + else { + ystep = 1; + } + + assert(dx >= 0); + assert(dy >= 0); + + setup->quad.x0 = setup->quad.y0 = -1; + setup->quad.mask = 0x0; + setup->quad.prim = PRIM_LINE; + /* XXX temporary: set coverage to 1.0 so the line appears + * if AA mode happens to be enabled. + */ + setup->quad.coverage[0] = + setup->quad.coverage[1] = + setup->quad.coverage[2] = + setup->quad.coverage[3] = 1.0; + + if (dx > dy) { + /*** X-major line ***/ + int i; + const int errorInc = dy + dy; + int error = errorInc - dx; + const int errorDec = error - dx; + + for (i = 0; i < dx; i++) { + plot(setup, x0, y0); + + x0 += xstep; + if (error < 0) { + error += errorInc; + } + else { + error += errorDec; + y0 += ystep; + } + } + } + else { + /*** Y-major line ***/ + int i; + const int errorInc = dx + dx; + int error = errorInc - dy; + const int errorDec = error - dy; + + for (i = 0; i < dy; i++) { + plot(setup, x0, y0); + + y0 += ystep; + if (error < 0) { + error += errorInc; + } + else { + error += errorDec; + x0 += xstep; + } + } + } + + /* draw final quad */ + if (setup->quad.mask) { + clip_emit_quad(setup); + } +} + + +static void +point_persp_coeff(struct setup_context *setup, + const float (*vert)[4], + struct tgsi_interp_coef *coef, + uint vertSlot, uint i) +{ + assert(i <= 3); + coef->dadx[i] = 0.0F; + coef->dady[i] = 0.0F; + coef->a0[i] = vert[vertSlot][i] * vert[0][3]; +} + + +/** + * Do setup for point rasterization, then render the point. + * Round or square points... + * XXX could optimize a lot for 1-pixel points. + */ +void +setup_point( struct setup_context *setup, + const float (*v0)[4] ) +{ + struct softpipe_context *softpipe = setup->softpipe; + const struct sp_fragment_shader *spfs = softpipe->fs; + const int sizeAttr = setup->softpipe->psize_slot; + const float size + = sizeAttr > 0 ? v0[sizeAttr][0] + : setup->softpipe->rasterizer->point_size; + const float halfSize = 0.5F * size; + const boolean round = (boolean) setup->softpipe->rasterizer->point_smooth; + const float x = v0[0][0]; /* Note: data[0] is always position */ + const float y = v0[0][1]; + const struct vertex_info *vinfo = softpipe_get_vertex_info(softpipe); + uint fragSlot; + +#if DEBUG_VERTS + debug_printf("Setup point:\n"); + print_vertex(setup, v0); +#endif + + if (softpipe->no_rast) + return; + + /* For points, all interpolants are constant-valued. + * However, for point sprites, we'll need to setup texcoords appropriately. + * XXX: which coefficients are the texcoords??? + * We may do point sprites as textured quads... + * + * KW: We don't know which coefficients are texcoords - ultimately + * the choice of what interpolation mode to use for each attribute + * should be determined by the fragment program, using + * per-attribute declaration statements that include interpolation + * mode as a parameter. So either the fragment program will have + * to be adjusted for pointsprite vs normal point behaviour, or + * otherwise a special interpolation mode will have to be defined + * which matches the required behaviour for point sprites. But - + * the latter is not a feature of normal hardware, and as such + * probably should be ruled out on that basis. + */ + setup->vprovoke = v0; + + /* setup Z, W */ + const_coeff(setup, &setup->posCoef, 0, 2); + const_coeff(setup, &setup->posCoef, 0, 3); + + for (fragSlot = 0; fragSlot < spfs->info.num_inputs; fragSlot++) { + const uint vertSlot = vinfo->src_index[fragSlot]; + uint j; + + switch (vinfo->interp_mode[fragSlot]) { + case INTERP_CONSTANT: + /* fall-through */ + case INTERP_LINEAR: + for (j = 0; j < NUM_CHANNELS; j++) + const_coeff(setup, &setup->coef[fragSlot], vertSlot, j); + break; + case INTERP_PERSPECTIVE: + for (j = 0; j < NUM_CHANNELS; j++) + point_persp_coeff(setup, setup->vprovoke, + &setup->coef[fragSlot], vertSlot, j); + break; + case INTERP_POS: + setup_fragcoord_coeff(setup, fragSlot); + break; + default: + assert(0); + } + + if (spfs->info.input_semantic_name[fragSlot] == TGSI_SEMANTIC_FOG) { + /* FOG.y = front/back facing XXX fix this */ + setup->coef[fragSlot].a0[1] = 1.0f - setup->quad.facing; + setup->coef[fragSlot].dadx[1] = 0.0; + setup->coef[fragSlot].dady[1] = 0.0; + } + } + + setup->quad.prim = PRIM_POINT; + + if (halfSize <= 0.5 && !round) { + /* special case for 1-pixel points */ + const int ix = ((int) x) & 1; + const int iy = ((int) y) & 1; + setup->quad.x0 = (int) x - ix; + setup->quad.y0 = (int) y - iy; + setup->quad.mask = (1 << ix) << (2 * iy); + clip_emit_quad(setup); + } + else { + if (round) { + /* rounded points */ + const int ixmin = block((int) (x - halfSize)); + const int ixmax = block((int) (x + halfSize)); + const int iymin = block((int) (y - halfSize)); + const int iymax = block((int) (y + halfSize)); + const float rmin = halfSize - 0.7071F; /* 0.7071 = sqrt(2)/2 */ + const float rmax = halfSize + 0.7071F; + const float rmin2 = MAX2(0.0F, rmin * rmin); + const float rmax2 = rmax * rmax; + const float cscale = 1.0F / (rmax2 - rmin2); + int ix, iy; + + for (iy = iymin; iy <= iymax; iy += 2) { + for (ix = ixmin; ix <= ixmax; ix += 2) { + float dx, dy, dist2, cover; + + setup->quad.mask = 0x0; + + dx = (ix + 0.5f) - x; + dy = (iy + 0.5f) - y; + dist2 = dx * dx + dy * dy; + if (dist2 <= rmax2) { + cover = 1.0F - (dist2 - rmin2) * cscale; + setup->quad.coverage[QUAD_TOP_LEFT] = MIN2(cover, 1.0f); + setup->quad.mask |= MASK_TOP_LEFT; + } + + dx = (ix + 1.5f) - x; + dy = (iy + 0.5f) - y; + dist2 = dx * dx + dy * dy; + if (dist2 <= rmax2) { + cover = 1.0F - (dist2 - rmin2) * cscale; + setup->quad.coverage[QUAD_TOP_RIGHT] = MIN2(cover, 1.0f); + setup->quad.mask |= MASK_TOP_RIGHT; + } + + dx = (ix + 0.5f) - x; + dy = (iy + 1.5f) - y; + dist2 = dx * dx + dy * dy; + if (dist2 <= rmax2) { + cover = 1.0F - (dist2 - rmin2) * cscale; + setup->quad.coverage[QUAD_BOTTOM_LEFT] = MIN2(cover, 1.0f); + setup->quad.mask |= MASK_BOTTOM_LEFT; + } + + dx = (ix + 1.5f) - x; + dy = (iy + 1.5f) - y; + dist2 = dx * dx + dy * dy; + if (dist2 <= rmax2) { + cover = 1.0F - (dist2 - rmin2) * cscale; + setup->quad.coverage[QUAD_BOTTOM_RIGHT] = MIN2(cover, 1.0f); + setup->quad.mask |= MASK_BOTTOM_RIGHT; + } + + if (setup->quad.mask) { + setup->quad.x0 = ix; + setup->quad.y0 = iy; + clip_emit_quad(setup); + } + } + } + } + else { + /* square points */ + const int xmin = (int) (x + 0.75 - halfSize); + const int ymin = (int) (y + 0.25 - halfSize); + const int xmax = xmin + (int) size; + const int ymax = ymin + (int) size; + /* XXX could apply scissor to xmin,ymin,xmax,ymax now */ + const int ixmin = block(xmin); + const int ixmax = block(xmax - 1); + const int iymin = block(ymin); + const int iymax = block(ymax - 1); + int ix, iy; + + /* + debug_printf("(%f, %f) -> X:%d..%d Y:%d..%d\n", x, y, xmin, xmax,ymin,ymax); + */ + for (iy = iymin; iy <= iymax; iy += 2) { + uint rowMask = 0xf; + if (iy < ymin) { + /* above the top edge */ + rowMask &= (MASK_BOTTOM_LEFT | MASK_BOTTOM_RIGHT); + } + if (iy + 1 >= ymax) { + /* below the bottom edge */ + rowMask &= (MASK_TOP_LEFT | MASK_TOP_RIGHT); + } + + for (ix = ixmin; ix <= ixmax; ix += 2) { + uint mask = rowMask; + + if (ix < xmin) { + /* fragment is past left edge of point, turn off left bits */ + mask &= (MASK_BOTTOM_RIGHT | MASK_TOP_RIGHT); + } + if (ix + 1 >= xmax) { + /* past the right edge */ + mask &= (MASK_BOTTOM_LEFT | MASK_TOP_LEFT); + } + + setup->quad.mask = mask; + setup->quad.x0 = ix; + setup->quad.y0 = iy; + clip_emit_quad(setup); + } + } + } + } +} + +void setup_prepare( struct setup_context *setup ) +{ + struct softpipe_context *sp = setup->softpipe; + unsigned i; + + if (sp->dirty) { + softpipe_update_derived(sp); + } + + /* Mark surfaces as defined now */ + for (i = 0; i < sp->framebuffer.num_cbufs; i++){ + if (sp->framebuffer.cbufs[i]) { + sp->framebuffer.cbufs[i]->status = PIPE_SURFACE_STATUS_DEFINED; + } + } + if (sp->framebuffer.zsbuf) { + sp->framebuffer.zsbuf->status = PIPE_SURFACE_STATUS_DEFINED; + } + + { + const struct sp_fragment_shader *fs = setup->softpipe->fs; + setup->quad.nr_attrs = fs->info.num_inputs; + sp->quad.first->begin(sp->quad.first); + } + + if (sp->reduced_api_prim == PIPE_PRIM_TRIANGLES && + sp->rasterizer->fill_cw == PIPE_POLYGON_MODE_FILL && + sp->rasterizer->fill_ccw == PIPE_POLYGON_MODE_FILL) { + /* we'll do culling */ + setup->winding = sp->rasterizer->cull_mode; + } + else { + /* 'draw' will do culling */ + setup->winding = PIPE_WINDING_NONE; + } +} + + + +void setup_destroy_context( struct setup_context *setup ) +{ + FREE( setup ); +} + + +/** + * Create a new primitive setup/render stage. + */ +struct setup_context *setup_create_context( struct softpipe_context *softpipe ) +{ + struct setup_context *setup = CALLOC_STRUCT(setup_context); + + setup->softpipe = softpipe; + + setup->quad.coef = setup->coef; + setup->quad.posCoef = &setup->posCoef; + + return setup; +} diff --git a/src/gallium/drivers/softpipe/sp_setup.h b/src/gallium/drivers/softpipe/sp_setup.h new file mode 100644 index 0000000000..d54f334428 --- /dev/null +++ b/src/gallium/drivers/softpipe/sp_setup.h @@ -0,0 +1,53 @@ +/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ +#ifndef SP_SETUP_H +#define SP_SETUP_H + +struct setup_context; +struct softpipe_context; + +void +setup_tri( struct setup_context *setup, + const float (*v0)[4], + const float (*v1)[4], + const float (*v2)[4] ); + +void +setup_line(struct setup_context *setup, + const float (*v0)[4], + const float (*v1)[4]); + +void +setup_point( struct setup_context *setup, + const float (*v0)[4] ); + + +struct setup_context *setup_create_context( struct softpipe_context *softpipe ); +void setup_prepare( struct setup_context *setup ); +void setup_destroy_context( struct setup_context *setup ); + +#endif diff --git a/src/gallium/drivers/softpipe/sp_state.h b/src/gallium/drivers/softpipe/sp_state.h index 6e6501f5bc..452e51fa79 100644 --- a/src/gallium/drivers/softpipe/sp_state.h +++ b/src/gallium/drivers/softpipe/sp_state.h @@ -54,9 +54,11 @@ struct tgsi_sampler; struct tgsi_exec_machine; +struct vertex_info; -/** Subclass of pipe_shader_state (though it doesn't really need to be). +/** + * Subclass of pipe_shader_state (though it doesn't really need to be). * * This is starting to look an awful lot like a quad pipeline stage... */ @@ -80,11 +82,10 @@ struct sp_fragment_shader { void (*delete)( struct sp_fragment_shader * ); }; -struct vertex_info; /** Subclass of pipe_shader_state */ struct sp_vertex_shader { - struct pipe_shader_state shader; + struct pipe_shader_state shader; /* Note: this field not actually used */ struct draw_vertex_shader *draw_data; }; @@ -171,6 +172,9 @@ boolean softpipe_draw_elements(struct pipe_context *pipe, unsigned indexSize, unsigned mode, unsigned start, unsigned count); +void +softpipe_set_edgeflags(struct pipe_context *pipe, const unsigned *edgeflags); + void softpipe_map_surfaces(struct softpipe_context *sp); diff --git a/src/gallium/drivers/softpipe/sp_state_derived.c b/src/gallium/drivers/softpipe/sp_state_derived.c index 14abb20eeb..f10a1fa471 100644 --- a/src/gallium/drivers/softpipe/sp_state_derived.c +++ b/src/gallium/drivers/softpipe/sp_state_derived.c @@ -71,25 +71,16 @@ softpipe_get_vertex_info(struct softpipe_context *softpipe) * simply emit the whole post-xform vertex as-is: */ struct vertex_info *vinfo_vbuf = &softpipe->vertex_info_vbuf; -#if 0 - vinfo_vbuf->num_attribs = 0; - /* special-case to allow memcpy of whole vertex */ - draw_emit_vertex_attr(vinfo_vbuf, EMIT_ALL, INTERP_NONE, 0); - /* size in dwords or floats */ - vinfo_vbuf->size = 4 * draw_num_vs_outputs(softpipe->draw) - + sizeof(struct vertex_header) / 4; -#else - /* for pass-through mode, we need a more explicit list of attribs */ const uint num = draw_num_vs_outputs(softpipe->draw); uint i; + /* No longer any need to try and emit draw vertex_header info. + */ vinfo_vbuf->num_attribs = 0; - draw_emit_vertex_attr(vinfo_vbuf, EMIT_HEADER, INTERP_NONE, 0); for (i = 0; i < num; i++) { draw_emit_vertex_attr(vinfo_vbuf, EMIT_4F, INTERP_PERSPECTIVE, i); } draw_compute_vertex_size(vinfo_vbuf); -#endif } /* diff --git a/src/gallium/drivers/softpipe/sp_state_fs.c b/src/gallium/drivers/softpipe/sp_state_fs.c index 4eefd1d61f..24b91fbc79 100644 --- a/src/gallium/drivers/softpipe/sp_state_fs.c +++ b/src/gallium/drivers/softpipe/sp_state_fs.c @@ -83,6 +83,8 @@ void softpipe_delete_fs_state(struct pipe_context *pipe, void *fs) { struct sp_fragment_shader *state = fs; + + assert(fs != softpipe_context(pipe)->fs); state->delete( state ); } @@ -100,10 +102,7 @@ softpipe_create_vs_state(struct pipe_context *pipe, return NULL; } - state->shader = *templ; - - state->draw_data = draw_create_vertex_shader(softpipe->draw, - &state->shader); + state->draw_data = draw_create_vertex_shader(softpipe->draw, templ); if (state->draw_data == NULL) { FREE( state ); return NULL; diff --git a/src/gallium/drivers/softpipe/sp_state_vertex.c b/src/gallium/drivers/softpipe/sp_state_vertex.c index e0230e16a4..46b6991195 100644 --- a/src/gallium/drivers/softpipe/sp_state_vertex.c +++ b/src/gallium/drivers/softpipe/sp_state_vertex.c @@ -47,6 +47,7 @@ softpipe_set_vertex_elements(struct pipe_context *pipe, memcpy(softpipe->vertex_element, attribs, count * sizeof(struct pipe_vertex_element)); + softpipe->num_vertex_elements = count; softpipe->dirty |= SP_NEW_VERTEX; @@ -64,6 +65,7 @@ softpipe_set_vertex_buffers(struct pipe_context *pipe, assert(count <= PIPE_MAX_ATTRIBS); memcpy(softpipe->vertex_buffer, buffers, count * sizeof(buffers[0])); + softpipe->num_vertex_buffers = count; softpipe->dirty |= SP_NEW_VERTEX; diff --git a/src/gallium/drivers/softpipe/sp_tex_sample.c b/src/gallium/drivers/softpipe/sp_tex_sample.c index 34da6356d7..be0b57d9fa 100644 --- a/src/gallium/drivers/softpipe/sp_tex_sample.c +++ b/src/gallium/drivers/softpipe/sp_tex_sample.c @@ -651,7 +651,9 @@ shadow_compare(uint compare_func, k = 0; break; default: + k = 0; assert(0); + break; } rgba[0][j] = rgba[1][j] = rgba[2][j] = (float) k; @@ -1049,5 +1051,19 @@ sp_get_samples(struct tgsi_sampler *sampler, default: assert(0); } + +#if 0 /* DEBUG */ + { + int i; + printf("Sampled at %f, %f, %f:\n", s[0], t[0], p[0]); + for (i = 0; i < 4; i++) { + printf("Frag %d: %f %f %f %f\n", i, + rgba[0][i], + rgba[1][i], + rgba[2][i], + rgba[3][i]); + } + } +#endif } diff --git a/src/gallium/drivers/softpipe/sp_tile_cache.c b/src/gallium/drivers/softpipe/sp_tile_cache.c index 19f71887e7..1117c0ad4c 100644 --- a/src/gallium/drivers/softpipe/sp_tile_cache.c +++ b/src/gallium/drivers/softpipe/sp_tile_cache.c @@ -39,7 +39,7 @@ #include "sp_surface.h" #include "sp_tile_cache.h" -#define NUM_ENTRIES 30 +#define NUM_ENTRIES 32 /** XXX move these */ @@ -131,7 +131,7 @@ sp_destroy_tile_cache(struct softpipe_tile_cache *tc) uint pos; for (pos = 0; pos < NUM_ENTRIES; pos++) { - //assert(tc->entries[pos].x < 0); + /*assert(tc->entries[pos].x < 0);*/ } if (tc->surface) { pipe_surface_reference(&tc->surface, NULL); @@ -168,7 +168,7 @@ sp_tile_cache_set_surface(struct softpipe_tile_cache *tc, tc->depth_stencil = (ps->format == PIPE_FORMAT_S8Z24_UNORM || ps->format == PIPE_FORMAT_Z16_UNORM || ps->format == PIPE_FORMAT_Z32_UNORM || - ps->format == PIPE_FORMAT_U_S8); + ps->format == PIPE_FORMAT_S8_UNORM); } } @@ -332,8 +332,8 @@ sp_tile_cache_flush_clear(struct pipe_context *pipe, for (x = 0; x < w; x += TILE_SIZE) { if (is_clear_flag_set(tc->clear_flags, x, y)) { pipe_put_tile_raw(pipe, ps, - x, y, TILE_SIZE, TILE_SIZE, - tc->tile.data.color32, 0/*STRIDE*/); + x, y, TILE_SIZE, TILE_SIZE, + tc->tile.data.color32, 0/*STRIDE*/); /* do this? */ clear_clear_flag(tc->clear_flags, x, y); @@ -367,8 +367,8 @@ sp_flush_tile_cache(struct softpipe_context *softpipe, if (tile->x >= 0) { if (tc->depth_stencil) { pipe_put_tile_raw(pipe, ps, - tile->x, tile->y, TILE_SIZE, TILE_SIZE, - tile->data.depth32, 0/*STRIDE*/); + tile->x, tile->y, TILE_SIZE, TILE_SIZE, + tile->data.depth32, 0/*STRIDE*/); } else { pipe_put_tile_rgba(pipe, ps, @@ -385,7 +385,7 @@ sp_flush_tile_cache(struct softpipe_context *softpipe, #endif } else if (tc->texture) { - /* caching a texture, mark all entries as embpy */ + /* caching a texture, mark all entries as empty */ for (pos = 0; pos < NUM_ENTRIES; pos++) { tc->entries[pos].x = -1; } diff --git a/src/gallium/drivers/softpipe/sp_winsys.h b/src/gallium/drivers/softpipe/sp_winsys.h index 291825dfe2..4ab666486c 100644 --- a/src/gallium/drivers/softpipe/sp_winsys.h +++ b/src/gallium/drivers/softpipe/sp_winsys.h @@ -59,7 +59,7 @@ struct pipe_context; struct pipe_context *softpipe_create( struct pipe_screen *, struct pipe_winsys *, - struct softpipe_winsys * ); + void *unused ); struct pipe_screen * diff --git a/src/gallium/include/pipe/p_compiler.h b/src/gallium/include/pipe/p_compiler.h index 91f3d2ac2d..a4b772bc4f 100644 --- a/src/gallium/include/pipe/p_compiler.h +++ b/src/gallium/include/pipe/p_compiler.h @@ -29,6 +29,8 @@ #define P_COMPILER_H +#include "p_config.h" + #include <stdlib.h> #include <string.h> diff --git a/src/gallium/include/pipe/p_config.h b/src/gallium/include/pipe/p_config.h new file mode 100644 index 0000000000..5c030bdfff --- /dev/null +++ b/src/gallium/include/pipe/p_config.h @@ -0,0 +1,121 @@ +/************************************************************************** + * + * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +/** + * @file + * Gallium configuration defines. + * + * This header file sets several defines based on the compiler, processor + * architecture, and operating system being used. These defines should be used + * throughout the code to facilitate porting to new platforms. It is likely that + * this file is auto-generated by an autoconf-like tool at some point, as some + * things cannot be determined by existing defines alone. + * + * @author José Fonseca <jrfonseca@tungstengraphics.com> + */ + +#ifndef P_CONFIG_H_ +#define P_CONFIG_H_ + + +/* + * Compiler + */ + +#if defined(__GNUC__) +#define PIPE_CC_GCC +#endif + +#if defined(_MSC_VER) || defined(__MSC__) +#define PIPE_CC_MSVC +#endif + +#if defined(__ICL) +#define PIPE_CC_ICL +#endif + + +/* + * Processor architecture + */ + +#if defined(_X86_) || defined(__i386__) || defined(__386__) || defined(i386) +#define PIPE_ARCH_X86 +#endif + +#if 0 /* FIXME */ +#define PIPE_ARCH_X86_64 +#endif + +#if 0 /* FIXME */ +#define PIPE_ARCH_PPC +#endif + + +/* + * Operating system + */ + +#if defined(__linux__) +#define PIPE_OS_LINUX +#endif + +#if defined(_WIN32) || defined(WIN32) +#define PIPE_OS_WINDOWS +#endif + + +/* + * Subsystem + * + * XXX: There is no way to autodetect this. + */ + +#if defined(PIPE_OS_LINUX) +#define PIPE_SUBSYSTEM_DRI +#endif + +#if defined(PIPE_OS_WINDOWS) +#if !defined(PIPE_SUBSYSTEM_USER) && !defined(PIPE_SUBSYSTEM_KERNEL) +#error Neither PIPE_SUBSYSTEM_USER or PIPE_SUBSYSTEM_KERNEL defined. +#endif +#if defined(PIPE_SUBSYSTEM_KERNEL) +#define PIPE_SUBSYSTEM_WINDOWS_DISPLAY +#endif +#if 0 /* FIXME */ +#define PIPE_SUBSYSTEM_WINDOWS_MINIPORT +#endif +#if 0 /* FIXME */ +#define PIPE_SUBSYSTEM_WINDOWS_CE +#endif +#if defined(PIPE_SUBSYSTEM_USER) +#define PIPE_SUBSYSTEM_WINDOWS_USER +#endif +#endif /* PIPE_OS_WINDOWS */ + + +#endif /* P_CONFIG_H_ */ diff --git a/src/gallium/include/pipe/p_debug.h b/src/gallium/include/pipe/p_debug.h index 9e52ad9a37..7a7312ea12 100644 --- a/src/gallium/include/pipe/p_debug.h +++ b/src/gallium/include/pipe/p_debug.h @@ -59,7 +59,6 @@ extern "C" { #endif #endif - void _debug_vprintf(const char *format, va_list ap); @@ -103,16 +102,19 @@ debug_printf(const char *format, ...) #endif +#ifdef DEBUG /** * Dump a blob in hex to the same place that debug_printf sends its * messages. */ -#ifdef DEBUG -void debug_print_blob( const char *name, - const void *blob, - unsigned size ); +void debug_print_blob( const char *name, const void *blob, unsigned size ); + +/* Print a message along with a prettified format string + */ +void debug_print_format(const char *msg, unsigned fmt ); #else #define debug_print_blob(_name, _blob, _size) ((void)0) +#define debug_print_format(_msg, _fmt) ((void)0) #endif @@ -304,11 +306,11 @@ void * debug_realloc(const char *file, unsigned line, const char *function, void *old_ptr, size_t old_size, size_t new_size ); -void -debug_memory_reset(void); +unsigned long +debug_memory_begin(void); void -debug_memory_report(void); +debug_memory_end(unsigned long beginning); #ifdef __cplusplus diff --git a/src/gallium/include/pipe/p_defines.h b/src/gallium/include/pipe/p_defines.h index 586951d956..46a355daa2 100644 --- a/src/gallium/include/pipe/p_defines.h +++ b/src/gallium/include/pipe/p_defines.h @@ -192,7 +192,7 @@ enum pipe_texture_target { #define PIPE_BUFFER_USAGE_VERTEX (1 << 5) #define PIPE_BUFFER_USAGE_INDEX (1 << 6) #define PIPE_BUFFER_USAGE_CONSTANT (1 << 7) -/** Pipe driver custam usage flags should be greater or equal to this value */ +/** Pipe driver custom usage flags should be greater or equal to this value */ #define PIPE_BUFFER_USAGE_CUSTOM (1 << 16) @@ -268,7 +268,7 @@ enum pipe_texture_target { #define PIPE_CAP_MAX_POINT_WIDTH_AA 17 #define PIPE_CAP_MAX_TEXTURE_ANISOTROPY 18 #define PIPE_CAP_MAX_TEXTURE_LOD_BIAS 19 -#define PIPE_CAP_BITMAP_TEXCOORD_BIAS 20 + #ifdef __cplusplus } diff --git a/src/gallium/include/pipe/p_format.h b/src/gallium/include/pipe/p_format.h index 9e0f91f202..a4bd23c302 100644 --- a/src/gallium/include/pipe/p_format.h +++ b/src/gallium/include/pipe/p_format.h @@ -28,7 +28,7 @@ #ifndef PIPE_FORMAT_H #define PIPE_FORMAT_H -#include <stdio.h> /* for sprintf */ +#include "util/u_string.h" #include "p_compiler.h" #include "p_debug.h" @@ -163,18 +163,21 @@ static INLINE uint pf_get(pipe_format_rgbazs_t f, uint shift, uint mask) /** * Shorthand macro for common format swizzles. */ -#define _PIPE_FORMAT_R000 _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_R, PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_0 ) -#define _PIPE_FORMAT_RG00 _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_R, PIPE_FORMAT_COMP_G, PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_0 ) -#define _PIPE_FORMAT_RGB0 _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_R, PIPE_FORMAT_COMP_G, PIPE_FORMAT_COMP_B, PIPE_FORMAT_COMP_0 ) +#define _PIPE_FORMAT_R001 _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_R, PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_1 ) +#define _PIPE_FORMAT_RG01 _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_R, PIPE_FORMAT_COMP_G, PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_1 ) +#define _PIPE_FORMAT_RGB1 _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_R, PIPE_FORMAT_COMP_G, PIPE_FORMAT_COMP_B, PIPE_FORMAT_COMP_1 ) #define _PIPE_FORMAT_RGBA _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_R, PIPE_FORMAT_COMP_G, PIPE_FORMAT_COMP_B, PIPE_FORMAT_COMP_A ) #define _PIPE_FORMAT_ARGB _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_A, PIPE_FORMAT_COMP_R, PIPE_FORMAT_COMP_G, PIPE_FORMAT_COMP_B ) #define _PIPE_FORMAT_BGRA _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_B, PIPE_FORMAT_COMP_G, PIPE_FORMAT_COMP_R, PIPE_FORMAT_COMP_A ) +#define _PIPE_FORMAT_1RGB _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_1, PIPE_FORMAT_COMP_R, PIPE_FORMAT_COMP_G, PIPE_FORMAT_COMP_B ) +#define _PIPE_FORMAT_BGR1 _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_B, PIPE_FORMAT_COMP_G, PIPE_FORMAT_COMP_R, PIPE_FORMAT_COMP_1 ) #define _PIPE_FORMAT_0000 _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_0 ) #define _PIPE_FORMAT_000R _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_R ) #define _PIPE_FORMAT_RRR1 _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_R, PIPE_FORMAT_COMP_R, PIPE_FORMAT_COMP_R, PIPE_FORMAT_COMP_1 ) #define _PIPE_FORMAT_RRRR _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_R, PIPE_FORMAT_COMP_R, PIPE_FORMAT_COMP_R, PIPE_FORMAT_COMP_R ) #define _PIPE_FORMAT_RRRG _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_R, PIPE_FORMAT_COMP_R, PIPE_FORMAT_COMP_R, PIPE_FORMAT_COMP_G ) #define _PIPE_FORMAT_Z000 _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_Z, PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_0 ) +#define _PIPE_FORMAT_0Z00 _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_Z, PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_0 ) #define _PIPE_FORMAT_SZ00 _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_S, PIPE_FORMAT_COMP_Z, PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_0 ) #define _PIPE_FORMAT_ZS00 _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_Z, PIPE_FORMAT_COMP_S, PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_0 ) #define _PIPE_FORMAT_S000 _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_S, PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_0 ) @@ -229,14 +232,16 @@ static INLINE uint pf_rev(pipe_format_ycbcr_t f) enum pipe_format { PIPE_FORMAT_NONE = _PIPE_FORMAT_RGBAZS_1 ( _PIPE_FORMAT_0000, 0, 0, 0, 0, PIPE_FORMAT_TYPE_UNKNOWN ), PIPE_FORMAT_A8R8G8B8_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_ARGB, 1, 1, 1, 1, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_X8R8G8B8_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_1RGB, 1, 1, 1, 1, PIPE_FORMAT_TYPE_UNORM ), PIPE_FORMAT_B8G8R8A8_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_BGRA, 1, 1, 1, 1, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_B8G8R8X8_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_BGR1, 1, 1, 1, 1, PIPE_FORMAT_TYPE_UNORM ), PIPE_FORMAT_A1R5G5B5_UNORM = _PIPE_FORMAT_RGBAZS_1 ( _PIPE_FORMAT_ARGB, 1, 5, 5, 5, PIPE_FORMAT_TYPE_UNORM ), PIPE_FORMAT_A4R4G4B4_UNORM = _PIPE_FORMAT_RGBAZS_1 ( _PIPE_FORMAT_ARGB, 4, 4, 4, 4, PIPE_FORMAT_TYPE_UNORM ), - PIPE_FORMAT_R5G6B5_UNORM = _PIPE_FORMAT_RGBAZS_1 ( _PIPE_FORMAT_RGB0, 5, 6, 5, 0, PIPE_FORMAT_TYPE_UNORM ), - PIPE_FORMAT_U_L8 = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RRR1, 1, 1, 1, 1, PIPE_FORMAT_TYPE_UNORM ), /**< ubyte luminance */ - PIPE_FORMAT_U_A8 = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_000R, 1, 1, 1, 1, PIPE_FORMAT_TYPE_UNORM ), /**< ubyte alpha */ - PIPE_FORMAT_U_I8 = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RRRR, 1, 1, 1, 1, PIPE_FORMAT_TYPE_UNORM ), /**< ubyte intensity */ - PIPE_FORMAT_U_A8_L8 = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RRRG, 1, 1, 1, 1, PIPE_FORMAT_TYPE_UNORM ), /**< ubyte alpha, luminance */ + PIPE_FORMAT_R5G6B5_UNORM = _PIPE_FORMAT_RGBAZS_1 ( _PIPE_FORMAT_RGB1, 5, 6, 5, 0, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_L8_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RRR1, 1, 1, 1, 0, PIPE_FORMAT_TYPE_UNORM ), /**< ubyte luminance */ + PIPE_FORMAT_A8_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_000R, 0, 0, 0, 1, PIPE_FORMAT_TYPE_UNORM ), /**< ubyte alpha */ + PIPE_FORMAT_I8_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RRRR, 1, 1, 1, 1, PIPE_FORMAT_TYPE_UNORM ), /**< ubyte intensity */ + PIPE_FORMAT_A8L8_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RRRG, 1, 1, 1, 1, PIPE_FORMAT_TYPE_UNORM ), /**< ubyte alpha, luminance */ PIPE_FORMAT_YCBCR = _PIPE_FORMAT_YCBCR( 0 ), PIPE_FORMAT_YCBCR_REV = _PIPE_FORMAT_YCBCR( 1 ), PIPE_FORMAT_Z16_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_Z000, 2, 0, 0, 0, PIPE_FORMAT_TYPE_UNORM ), @@ -244,68 +249,75 @@ enum pipe_format { PIPE_FORMAT_Z32_FLOAT = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_Z000, 4, 0, 0, 0, PIPE_FORMAT_TYPE_FLOAT ), PIPE_FORMAT_S8Z24_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_SZ00, 1, 3, 0, 0, PIPE_FORMAT_TYPE_UNORM ), PIPE_FORMAT_Z24S8_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_ZS00, 3, 1, 0, 0, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_X8Z24_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_0Z00, 1, 3, 0, 0, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_Z24X8_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_Z000, 3, 1, 0, 0, PIPE_FORMAT_TYPE_UNORM ), PIPE_FORMAT_S8_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_S000, 1, 0, 0, 0, PIPE_FORMAT_TYPE_UNORM ), /**< ubyte stencil */ - PIPE_FORMAT_R64_FLOAT = _PIPE_FORMAT_RGBAZS_64( _PIPE_FORMAT_R000, 1, 0, 0, 0, PIPE_FORMAT_TYPE_FLOAT ), - PIPE_FORMAT_R64G64_FLOAT = _PIPE_FORMAT_RGBAZS_64( _PIPE_FORMAT_RG00, 1, 1, 0, 0, PIPE_FORMAT_TYPE_FLOAT ), - PIPE_FORMAT_R64G64B64_FLOAT = _PIPE_FORMAT_RGBAZS_64( _PIPE_FORMAT_RGB0, 1, 1, 1, 0, PIPE_FORMAT_TYPE_FLOAT ), + PIPE_FORMAT_R64_FLOAT = _PIPE_FORMAT_RGBAZS_64( _PIPE_FORMAT_R001, 1, 0, 0, 0, PIPE_FORMAT_TYPE_FLOAT ), + PIPE_FORMAT_R64G64_FLOAT = _PIPE_FORMAT_RGBAZS_64( _PIPE_FORMAT_RG01, 1, 1, 0, 0, PIPE_FORMAT_TYPE_FLOAT ), + PIPE_FORMAT_R64G64B64_FLOAT = _PIPE_FORMAT_RGBAZS_64( _PIPE_FORMAT_RGB1, 1, 1, 1, 0, PIPE_FORMAT_TYPE_FLOAT ), PIPE_FORMAT_R64G64B64A64_FLOAT = _PIPE_FORMAT_RGBAZS_64( _PIPE_FORMAT_RGBA, 1, 1, 1, 1, PIPE_FORMAT_TYPE_FLOAT ), - PIPE_FORMAT_R32_FLOAT = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 4, 0, 0, 0, PIPE_FORMAT_TYPE_FLOAT ), - PIPE_FORMAT_R32G32_FLOAT = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 4, 4, 0, 0, PIPE_FORMAT_TYPE_FLOAT ), - PIPE_FORMAT_R32G32B32_FLOAT = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 4, 4, 4, 0, PIPE_FORMAT_TYPE_FLOAT ), + PIPE_FORMAT_R32_FLOAT = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R001, 4, 0, 0, 0, PIPE_FORMAT_TYPE_FLOAT ), + PIPE_FORMAT_R32G32_FLOAT = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG01, 4, 4, 0, 0, PIPE_FORMAT_TYPE_FLOAT ), + PIPE_FORMAT_R32G32B32_FLOAT = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB1, 4, 4, 4, 0, PIPE_FORMAT_TYPE_FLOAT ), PIPE_FORMAT_R32G32B32A32_FLOAT = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 4, 4, 4, 4, PIPE_FORMAT_TYPE_FLOAT ), - PIPE_FORMAT_R32_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 4, 0, 0, 0, PIPE_FORMAT_TYPE_UNORM ), - PIPE_FORMAT_R32G32_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 4, 4, 0, 0, PIPE_FORMAT_TYPE_UNORM ), - PIPE_FORMAT_R32G32B32_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 4, 4, 4, 0, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_R32_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R001, 4, 0, 0, 0, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_R32G32_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG01, 4, 4, 0, 0, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_R32G32B32_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB1, 4, 4, 4, 0, PIPE_FORMAT_TYPE_UNORM ), PIPE_FORMAT_R32G32B32A32_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 4, 4, 4, 4, PIPE_FORMAT_TYPE_UNORM ), - PIPE_FORMAT_R32_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 4, 0, 0, 0, PIPE_FORMAT_TYPE_USCALED ), - PIPE_FORMAT_R32G32_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 4, 4, 0, 0, PIPE_FORMAT_TYPE_USCALED ), - PIPE_FORMAT_R32G32B32_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 4, 4, 4, 0, PIPE_FORMAT_TYPE_USCALED ), + PIPE_FORMAT_R32_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R001, 4, 0, 0, 0, PIPE_FORMAT_TYPE_USCALED ), + PIPE_FORMAT_R32G32_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG01, 4, 4, 0, 0, PIPE_FORMAT_TYPE_USCALED ), + PIPE_FORMAT_R32G32B32_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB1, 4, 4, 4, 0, PIPE_FORMAT_TYPE_USCALED ), PIPE_FORMAT_R32G32B32A32_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 4, 4, 4, 4, PIPE_FORMAT_TYPE_USCALED ), - PIPE_FORMAT_R32_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 4, 0, 0, 0, PIPE_FORMAT_TYPE_SNORM ), - PIPE_FORMAT_R32G32_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 4, 4, 0, 0, PIPE_FORMAT_TYPE_SNORM ), - PIPE_FORMAT_R32G32B32_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 4, 4, 4, 0, PIPE_FORMAT_TYPE_SNORM ), + PIPE_FORMAT_R32_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R001, 4, 0, 0, 0, PIPE_FORMAT_TYPE_SNORM ), + PIPE_FORMAT_R32G32_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG01, 4, 4, 0, 0, PIPE_FORMAT_TYPE_SNORM ), + PIPE_FORMAT_R32G32B32_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB1, 4, 4, 4, 0, PIPE_FORMAT_TYPE_SNORM ), PIPE_FORMAT_R32G32B32A32_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 4, 4, 4, 4, PIPE_FORMAT_TYPE_SNORM ), - PIPE_FORMAT_R32_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 4, 0, 0, 0, PIPE_FORMAT_TYPE_SSCALED ), - PIPE_FORMAT_R32G32_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 4, 4, 0, 0, PIPE_FORMAT_TYPE_SSCALED ), - PIPE_FORMAT_R32G32B32_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 4, 4, 4, 0, PIPE_FORMAT_TYPE_SSCALED ), + PIPE_FORMAT_R32_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R001, 4, 0, 0, 0, PIPE_FORMAT_TYPE_SSCALED ), + PIPE_FORMAT_R32G32_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG01, 4, 4, 0, 0, PIPE_FORMAT_TYPE_SSCALED ), + PIPE_FORMAT_R32G32B32_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB1, 4, 4, 4, 0, PIPE_FORMAT_TYPE_SSCALED ), PIPE_FORMAT_R32G32B32A32_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 4, 4, 4, 4, PIPE_FORMAT_TYPE_SSCALED ), - PIPE_FORMAT_R16_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 2, 0, 0, 0, PIPE_FORMAT_TYPE_UNORM ), - PIPE_FORMAT_R16G16_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 2, 2, 0, 0, PIPE_FORMAT_TYPE_UNORM ), - PIPE_FORMAT_R16G16B16_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 2, 2, 2, 0, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_R16_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R001, 2, 0, 0, 0, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_R16G16_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG01, 2, 2, 0, 0, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_R16G16B16_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB1, 2, 2, 2, 0, PIPE_FORMAT_TYPE_UNORM ), PIPE_FORMAT_R16G16B16A16_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 2, 2, 2, 2, PIPE_FORMAT_TYPE_UNORM ), - PIPE_FORMAT_R16_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 2, 0, 0, 0, PIPE_FORMAT_TYPE_USCALED ), - PIPE_FORMAT_R16G16_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 2, 2, 0, 0, PIPE_FORMAT_TYPE_USCALED ), - PIPE_FORMAT_R16G16B16_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 2, 2, 2, 0, PIPE_FORMAT_TYPE_USCALED ), + PIPE_FORMAT_R16_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R001, 2, 0, 0, 0, PIPE_FORMAT_TYPE_USCALED ), + PIPE_FORMAT_R16G16_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG01, 2, 2, 0, 0, PIPE_FORMAT_TYPE_USCALED ), + PIPE_FORMAT_R16G16B16_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB1, 2, 2, 2, 0, PIPE_FORMAT_TYPE_USCALED ), PIPE_FORMAT_R16G16B16A16_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 2, 2, 2, 2, PIPE_FORMAT_TYPE_USCALED ), - PIPE_FORMAT_R16_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 2, 0, 0, 0, PIPE_FORMAT_TYPE_SNORM ), - PIPE_FORMAT_R16G16_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 2, 2, 0, 0, PIPE_FORMAT_TYPE_SNORM ), - PIPE_FORMAT_R16G16B16_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 2, 2, 2, 0, PIPE_FORMAT_TYPE_SNORM ), + PIPE_FORMAT_R16_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R001, 2, 0, 0, 0, PIPE_FORMAT_TYPE_SNORM ), + PIPE_FORMAT_R16G16_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG01, 2, 2, 0, 0, PIPE_FORMAT_TYPE_SNORM ), + PIPE_FORMAT_R16G16B16_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB1, 2, 2, 2, 0, PIPE_FORMAT_TYPE_SNORM ), PIPE_FORMAT_R16G16B16A16_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 2, 2, 2, 2, PIPE_FORMAT_TYPE_SNORM ), - PIPE_FORMAT_R16_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 2, 0, 0, 0, PIPE_FORMAT_TYPE_SSCALED ), - PIPE_FORMAT_R16G16_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 2, 2, 0, 0, PIPE_FORMAT_TYPE_SSCALED ), - PIPE_FORMAT_R16G16B16_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 2, 2, 2, 0, PIPE_FORMAT_TYPE_SSCALED ), + PIPE_FORMAT_R16_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R001, 2, 0, 0, 0, PIPE_FORMAT_TYPE_SSCALED ), + PIPE_FORMAT_R16G16_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG01, 2, 2, 0, 0, PIPE_FORMAT_TYPE_SSCALED ), + PIPE_FORMAT_R16G16B16_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB1, 2, 2, 2, 0, PIPE_FORMAT_TYPE_SSCALED ), PIPE_FORMAT_R16G16B16A16_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 2, 2, 2, 2, PIPE_FORMAT_TYPE_SSCALED ), - PIPE_FORMAT_R8_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 1, 0, 0, 0, PIPE_FORMAT_TYPE_UNORM ), - PIPE_FORMAT_R8G8_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 1, 1, 0, 0, PIPE_FORMAT_TYPE_UNORM ), - PIPE_FORMAT_R8G8B8_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 1, 1, 1, 0, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_R8_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R001, 1, 0, 0, 0, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_R8G8_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG01, 1, 1, 0, 0, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_R8G8B8_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB1, 1, 1, 1, 0, PIPE_FORMAT_TYPE_UNORM ), PIPE_FORMAT_R8G8B8A8_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 1, 1, 1, 1, PIPE_FORMAT_TYPE_UNORM ), - PIPE_FORMAT_R8_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 1, 0, 0, 0, PIPE_FORMAT_TYPE_USCALED ), - PIPE_FORMAT_R8G8_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 1, 1, 0, 0, PIPE_FORMAT_TYPE_USCALED ), - PIPE_FORMAT_R8G8B8_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 1, 1, 1, 0, PIPE_FORMAT_TYPE_USCALED ), + PIPE_FORMAT_R8G8B8X8_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB1, 1, 1, 1, 1, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_R8_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R001, 1, 0, 0, 0, PIPE_FORMAT_TYPE_USCALED ), + PIPE_FORMAT_R8G8_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG01, 1, 1, 0, 0, PIPE_FORMAT_TYPE_USCALED ), + PIPE_FORMAT_R8G8B8_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB1, 1, 1, 1, 0, PIPE_FORMAT_TYPE_USCALED ), PIPE_FORMAT_R8G8B8A8_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 1, 1, 1, 1, PIPE_FORMAT_TYPE_USCALED ), - PIPE_FORMAT_R8_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 1, 0, 0, 0, PIPE_FORMAT_TYPE_SNORM ), - PIPE_FORMAT_R8G8_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 1, 1, 0, 0, PIPE_FORMAT_TYPE_SNORM ), - PIPE_FORMAT_R8G8B8_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 1, 1, 1, 0, PIPE_FORMAT_TYPE_SNORM ), + PIPE_FORMAT_R8G8B8X8_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB1, 1, 1, 1, 1, PIPE_FORMAT_TYPE_USCALED ), + PIPE_FORMAT_R8_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R001, 1, 0, 0, 0, PIPE_FORMAT_TYPE_SNORM ), + PIPE_FORMAT_R8G8_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG01, 1, 1, 0, 0, PIPE_FORMAT_TYPE_SNORM ), + PIPE_FORMAT_R8G8B8_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB1, 1, 1, 1, 0, PIPE_FORMAT_TYPE_SNORM ), PIPE_FORMAT_R8G8B8A8_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 1, 1, 1, 1, PIPE_FORMAT_TYPE_SNORM ), - PIPE_FORMAT_R8_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 1, 0, 0, 0, PIPE_FORMAT_TYPE_SSCALED ), - PIPE_FORMAT_R8G8_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 1, 1, 0, 0, PIPE_FORMAT_TYPE_SSCALED ), - PIPE_FORMAT_R8G8B8_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 1, 1, 1, 0, PIPE_FORMAT_TYPE_SSCALED ), + PIPE_FORMAT_R8G8B8X8_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB1, 1, 1, 1, 1, PIPE_FORMAT_TYPE_SNORM ), + PIPE_FORMAT_R8_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R001, 1, 0, 0, 0, PIPE_FORMAT_TYPE_SSCALED ), + PIPE_FORMAT_R8G8_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG01, 1, 1, 0, 0, PIPE_FORMAT_TYPE_SSCALED ), + PIPE_FORMAT_R8G8B8_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB1, 1, 1, 1, 0, PIPE_FORMAT_TYPE_SSCALED ), PIPE_FORMAT_R8G8B8A8_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 1, 1, 1, 1, PIPE_FORMAT_TYPE_SSCALED ), + PIPE_FORMAT_R8G8B8X8_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB1, 1, 1, 1, 1, PIPE_FORMAT_TYPE_SSCALED ), /* sRGB formats */ - PIPE_FORMAT_L8_SRGB = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RRR1, 1, 1, 1, 1, PIPE_FORMAT_TYPE_SRGB ), + PIPE_FORMAT_L8_SRGB = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RRR1, 1, 1, 1, 0, PIPE_FORMAT_TYPE_SRGB ), PIPE_FORMAT_A8_L8_SRGB = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RRRG, 1, 1, 1, 1, PIPE_FORMAT_TYPE_SRGB ), - PIPE_FORMAT_R8G8B8_SRGB = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 1, 1, 1, 0, PIPE_FORMAT_TYPE_SRGB ), + PIPE_FORMAT_R8G8B8_SRGB = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB1, 1, 1, 1, 0, PIPE_FORMAT_TYPE_SRGB ), PIPE_FORMAT_R8G8B8A8_SRGB = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 1, 1, 1, 1, PIPE_FORMAT_TYPE_SRGB ), + PIPE_FORMAT_R8G8B8X8_SRGB = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB1, 1, 1, 1, 1, PIPE_FORMAT_TYPE_SRGB ), /* compressed formats */ PIPE_FORMAT_DXT1_RGB = _PIPE_FORMAT_DXT( 1, 8, 8, 8, 0 ), @@ -316,97 +328,9 @@ enum pipe_format { /** - * Unsigned 8-bit stencil format. - * XXX should remove this, but S8_UNORM is a poor name - */ -#define PIPE_FORMAT_U_S8 PIPE_FORMAT_S8_UNORM - - -/** * Builds pipe format name from format token. */ -static INLINE char *pf_sprint_name( char *str, enum pipe_format format ) -{ - strcpy( str, "PIPE_FORMAT_" ); - switch (pf_layout( format )) { - case PIPE_FORMAT_LAYOUT_RGBAZS: - { - pipe_format_rgbazs_t rgbazs = (pipe_format_rgbazs_t) format; - uint i; - uint scale = 1 << (pf_exp8( rgbazs ) * 3); - - for (i = 0; i < 4; i++) { - uint size = pf_size_xyzw( rgbazs, i ); - - if (size == 0) { - break; - } - switch (pf_swizzle_xyzw( rgbazs, i )) { - case PIPE_FORMAT_COMP_R: - strcat( str, "R" ); - break; - case PIPE_FORMAT_COMP_G: - strcat( str, "G" ); - break; - case PIPE_FORMAT_COMP_B: - strcat( str, "B" ); - break; - case PIPE_FORMAT_COMP_A: - strcat( str, "A" ); - break; - case PIPE_FORMAT_COMP_0: - strcat( str, "0" ); - break; - case PIPE_FORMAT_COMP_1: - strcat( str, "1" ); - break; - case PIPE_FORMAT_COMP_Z: - strcat( str, "Z" ); - break; - case PIPE_FORMAT_COMP_S: - strcat( str, "S" ); - break; - } - sprintf( &str[strlen( str )], "%u", size * scale ); - } - if (i != 0) { - strcat( str, "_" ); - } - switch (pf_type( rgbazs )) { - case PIPE_FORMAT_TYPE_UNKNOWN: - strcat( str, "NONE" ); - break; - case PIPE_FORMAT_TYPE_FLOAT: - strcat( str, "FLOAT" ); - break; - case PIPE_FORMAT_TYPE_UNORM: - strcat( str, "UNORM" ); - break; - case PIPE_FORMAT_TYPE_SNORM: - strcat( str, "SNORM" ); - break; - case PIPE_FORMAT_TYPE_USCALED: - strcat( str, "USCALED" ); - break; - case PIPE_FORMAT_TYPE_SSCALED: - strcat( str, "SSCALED" ); - break; - } - } - break; - case PIPE_FORMAT_LAYOUT_YCBCR: - { - pipe_format_ycbcr_t ycbcr = (pipe_format_ycbcr_t) format; - - strcat( str, "YCBCR" ); - if (pf_rev( ycbcr )) { - strcat( str, "_REV" ); - } - } - break; - } - return str; -} +extern char *pf_sprint_name( char *str, enum pipe_format format ); /** * Return bits for a particular component. @@ -442,6 +366,8 @@ static INLINE uint pf_get_bits( enum pipe_format format ) switch (pf_layout(format)) { case PIPE_FORMAT_LAYOUT_RGBAZS: return + pf_get_component_bits( format, PIPE_FORMAT_COMP_0 ) + + pf_get_component_bits( format, PIPE_FORMAT_COMP_1 ) + pf_get_component_bits( format, PIPE_FORMAT_COMP_R ) + pf_get_component_bits( format, PIPE_FORMAT_COMP_G ) + pf_get_component_bits( format, PIPE_FORMAT_COMP_B ) + diff --git a/src/gallium/include/pipe/p_inlines.h b/src/gallium/include/pipe/p_inlines.h index 274f76a383..8eb604e73f 100644 --- a/src/gallium/include/pipe/p_inlines.h +++ b/src/gallium/include/pipe/p_inlines.h @@ -129,6 +129,56 @@ pipe_texture_release(struct pipe_texture **ptr) } +/** + * Convenience wrappers for winsys buffer functions. + */ + +static INLINE struct pipe_buffer * +pipe_buffer_create( struct pipe_context *pipe, + unsigned alignment, unsigned usage, unsigned size ) +{ + return pipe->winsys->buffer_create(pipe->winsys, alignment, usage, size); +} + +static INLINE struct pipe_buffer * +pipe_user_buffer_create( struct pipe_context *pipe, void *ptr, unsigned size ) +{ + return pipe->winsys->user_buffer_create(pipe->winsys, ptr, size); +} + +static INLINE void +pipe_buffer_destroy( struct pipe_context *pipe, struct pipe_buffer *buf ) +{ + pipe->winsys->buffer_destroy(pipe->winsys, buf); +} + +static INLINE void * +pipe_buffer_map(struct pipe_context *pipe, + struct pipe_buffer *buf, + unsigned usage) +{ + return pipe->winsys->buffer_map(pipe->winsys, buf, usage); +} + +static INLINE void +pipe_buffer_unmap(struct pipe_context *pipe, + struct pipe_buffer *buf) +{ + pipe->winsys->buffer_unmap(pipe->winsys, buf); +} + +/* XXX when we're using this everywhere, get rid of + * pipe_buffer_reference() above. + */ +static INLINE void +pipe_reference_buffer(struct pipe_context *pipe, + struct pipe_buffer **ptr, + struct pipe_buffer *buf) +{ + pipe_buffer_reference(pipe->winsys, ptr, buf); +} + + #ifdef __cplusplus } #endif diff --git a/src/gallium/include/pipe/p_screen.h b/src/gallium/include/pipe/p_screen.h index 35bd46d691..26ac99d287 100644 --- a/src/gallium/include/pipe/p_screen.h +++ b/src/gallium/include/pipe/p_screen.h @@ -58,25 +58,33 @@ struct pipe_screen { void (*destroy)( struct pipe_screen * ); - /* - * Capability queries - */ const char *(*get_name)( struct pipe_screen * ); const char *(*get_vendor)( struct pipe_screen * ); + /** + * Query an integer-valued capability/parameter/limit + * \param param one of PIPE_CAP_x + */ int (*get_param)( struct pipe_screen *, int param ); + /** + * Query a float-valued capability/parameter/limit + * \param param one of PIPE_CAP_x + */ float (*get_paramf)( struct pipe_screen *, int param ); - /**< type is one of PIPE_TEXTURE, PIPE_SURFACE */ + /** + * Check if the given pipe_format is supported as a texture or + * drawing surface. + * \param type one of PIPE_TEXTURE, PIPE_SURFACE + */ boolean (*is_format_supported)( struct pipe_screen *, enum pipe_format format, uint type ); - - /* - * Texture functions + /** + * Create a new texture object, using the given template info. */ struct pipe_texture * (*texture_create)(struct pipe_screen *, const struct pipe_texture *templat); diff --git a/src/gallium/include/pipe/p_state.h b/src/gallium/include/pipe/p_state.h index 2dc9a92186..912d84e7b9 100644 --- a/src/gallium/include/pipe/p_state.h +++ b/src/gallium/include/pipe/p_state.h @@ -51,16 +51,14 @@ extern "C" { /** * Implementation limits */ -#define PIPE_MAX_SAMPLERS 16 -#define PIPE_MAX_CLIP_PLANES 6 -#define PIPE_MAX_CONSTANT 32 -#define PIPE_MAX_ATTRIBS 32 -#define PIPE_ATTRIB_MAX 32 /* XXX obsolete - remove */ -#define PIPE_MAX_COLOR_BUFS 8 -#define PIPE_MAX_TEXTURE_LEVELS 16 -#define PIPE_MAX_FEEDBACK_ATTRIBS 16 -#define PIPE_MAX_SHADER_INPUTS 16 -#define PIPE_MAX_SHADER_OUTPUTS 16 +#define PIPE_MAX_ATTRIBS 32 +#define PIPE_MAX_CLIP_PLANES 6 +#define PIPE_MAX_COLOR_BUFS 8 +#define PIPE_MAX_CONSTANT 32 +#define PIPE_MAX_SAMPLERS 16 +#define PIPE_MAX_SHADER_INPUTS 16 +#define PIPE_MAX_SHADER_OUTPUTS 16 +#define PIPE_MAX_TEXTURE_LEVELS 16 /* fwd decls */ diff --git a/src/gallium/include/pipe/p_thread.h b/src/gallium/include/pipe/p_thread.h index 4325abc951..af9150aafb 100644 --- a/src/gallium/include/pipe/p_thread.h +++ b/src/gallium/include/pipe/p_thread.h @@ -114,6 +114,26 @@ typedef pthread_mutex_t _glthread_Mutex; #define _glthread_UNLOCK_MUTEX(name) \ (void) pthread_mutex_unlock(&(name)) +typedef pthread_cond_t _glthread_Cond; + +#define _glthread_DECLARE_STATIC_COND(name) \ + static _glthread_Cond name = PTHREAD_COND_INITIALIZER + +#define _glthread_INIT_COND(cond) \ + pthread_cond_init(&(cond), NULL) + +#define _glthread_DESTROY_COND(name) \ + pthread_cond_destroy(&(name)) + +#define _glthread_COND_WAIT(cond, mutex) \ + pthread_cond_wait(&(cond), &(mutex)) + +#define _glthread_COND_SIGNAL(cond) \ + pthread_cond_signal(&(cond)) + +#define _glthread_COND_BROADCAST(cond) \ + pthread_cond_broadcast(&(cond)) + #endif /* PTHREADS */ @@ -273,6 +293,20 @@ typedef unsigned _glthread_Mutex; #define _glthread_UNLOCK_MUTEX(name) (void) name +typedef unsigned _glthread_Cond; + +#define _glthread_DECLARE_STATIC_COND(name) static _glthread_Cond name = 0 + +#define _glthread_INIT_COND(name) (void) name + +#define _glthread_DESTROY_COND(name) (void) name + +#define _glthread_COND_WAIT(name, mutex) (void) name + +#define _glthread_COND_SIGNAL(name) (void) name + +#define _glthread_COND_BROADCAST(name) (void) name + #endif /* THREADS */ diff --git a/src/gallium/include/pipe/p_util.h b/src/gallium/include/pipe/p_util.h index 1bc00ac7a0..a66ab765f7 100644 --- a/src/gallium/include/pipe/p_util.h +++ b/src/gallium/include/pipe/p_util.h @@ -28,6 +28,7 @@ #ifndef P_UTIL_H #define P_UTIL_H +#include "p_config.h" #include "p_compiler.h" #include "p_debug.h" #include "p_pointer.h" @@ -40,7 +41,9 @@ extern "C" { #endif -#if defined(WIN32) && defined(DEBUG) /* memory debugging */ +#if defined(PIPE_SUBSYSTEM_WINDOWS_DISPLAY) && defined(DEBUG) + +/* memory debugging */ #include "p_debug.h" @@ -53,9 +56,7 @@ extern "C" { #define REALLOC( _ptr, _old_size, _size ) \ debug_realloc( __FILE__, __LINE__, __FUNCTION__, _ptr, _old_size, _size ) -#else - -#ifdef WIN32 +#elif defined(PIPE_SUBSYSTEM_WINDOWS_DISPLAY) void * __stdcall EngAllocMem( @@ -67,17 +68,33 @@ void __stdcall EngFreeMem( void *Mem ); -static INLINE void * -MALLOC( unsigned size ) -{ -#ifdef WINCE - /* TODO: Need to abstract this */ - return malloc( size ); +#define MALLOC( _size ) EngAllocMem( 0, _size, 'D3AG' ) +#define _FREE( _ptr ) EngFreeMem( _ptr ) + +#elif defined(PIPE_SUBSYSTEM_WINDOWS_MINIPORT) + +void * +ExAllocatePool( + unsigned long PoolType, + size_t NumberOfBytes); + +void +ExFreePool(void *P); + +#define MALLOC(_size) ExAllocatePool(0, _size) +#define _FREE(_ptr) ExFreePool(_ptr) + #else - return EngAllocMem( 0, size, 'D3AG' ); + +#define MALLOC( SIZE ) malloc( SIZE ) +#define CALLOC( COUNT, SIZE ) calloc( COUNT, SIZE ) +#define FREE( PTR ) free( PTR ) +#define REALLOC( OLDPTR, OLDSIZE, NEWSIZE ) realloc( OLDPTR, NEWSIZE ) + #endif -} + +#ifndef CALLOC static INLINE void * CALLOC( unsigned count, unsigned size ) { @@ -87,20 +104,19 @@ CALLOC( unsigned count, unsigned size ) } return ptr; } +#endif /* !CALLOC */ +#ifndef FREE static INLINE void FREE( void *ptr ) { if( ptr ) { -#ifdef WINCE - /* TODO: Need to abstract this */ - free( ptr ); -#else - EngFreeMem( ptr ); -#endif + _FREE( ptr ); } } +#endif /* !FREE */ +#ifndef REALLOC static INLINE void * REALLOC( void *old_ptr, unsigned old_size, unsigned new_size ) { @@ -117,19 +133,8 @@ REALLOC( void *old_ptr, unsigned old_size, unsigned new_size ) FREE( old_ptr ); return new_ptr; } +#endif /* !REALLOC */ -#else /* !WIN32 */ - -#define MALLOC( SIZE ) malloc( SIZE ) - -#define CALLOC( COUNT, SIZE ) calloc( COUNT, SIZE ) - -#define FREE( PTR ) free( PTR ) - -#define REALLOC( OLDPTR, OLDSIZE, NEWSIZE ) realloc( OLDPTR, NEWSIZE ) - -#endif /* !WIN32 */ -#endif /* !DEBUG */ #define MALLOC_STRUCT(T) (struct T *) MALLOC(sizeof(struct T)) @@ -138,14 +143,6 @@ REALLOC( void *old_ptr, unsigned old_size, unsigned new_size ) #define GETENV( X ) debug_get_option( X, NULL ) -#ifdef WIN32 -int rpl_vsnprintf(char *, size_t, const char *, va_list); -int rpl_snprintf(char *str, size_t size, const char *format, ...); -#define vsnprintf rpl_vsnprintf -#define snprintf rpl_snprintf -#endif - - /** * Return memory on given byte alignment */ @@ -413,6 +410,54 @@ extern void pipe_copy_rect(ubyte * dst, unsigned cpp, unsigned dst_pitch, int src_pitch, unsigned src_x, int src_y); + +#ifdef WIN32 + +#if !defined(_INC_MATH) || !defined(__cplusplus) + +static INLINE float cosf( float f ) +{ + return (float) cos( (double) f ); +} + +static INLINE float sinf( float f ) +{ + return (float) sin( (double) f ); +} + +static INLINE float ceilf( float f ) +{ + return (float) ceil( (double) f ); +} + +static INLINE float floorf( float f ) +{ + return (float) floor( (double) f ); +} + +static INLINE float powf( float f, float g ) +{ + return (float) pow( (double) f, (double) g ); +} + +static INLINE float sqrtf( float f ) +{ + return (float) sqrt( (double) f ); +} + +static INLINE float fabsf( float f ) +{ + return (float) fabs( (double) f ); +} + +static INLINE float logf( float f ) +{ + return (float) cos( (double) f ); +} +#endif /* _INC_MATH */ +#endif + + #ifdef __cplusplus } #endif diff --git a/src/gallium/include/pipe/p_winsys.h b/src/gallium/include/pipe/p_winsys.h index 8569cdcf12..3005ec2d94 100644 --- a/src/gallium/include/pipe/p_winsys.h +++ b/src/gallium/include/pipe/p_winsys.h @@ -63,13 +63,13 @@ struct pipe_surface; struct pipe_winsys { /** Returns name of this winsys interface */ - const char *(*get_name)( struct pipe_winsys *sws ); + const char *(*get_name)( struct pipe_winsys *ws ); /** * Do any special operations to ensure frontbuffer contents are * displayed, eg copy fake frontbuffer. */ - void (*flush_frontbuffer)( struct pipe_winsys *sws, + void (*flush_frontbuffer)( struct pipe_winsys *ws, struct pipe_surface *surf, void *context_private ); @@ -79,7 +79,8 @@ struct pipe_winsys /** * Allocate storage for a pipe_surface. - * Returns 0 if succeeds. + * \param flags XXX unused, remove someday + * \return 0 if succeeds. */ int (*surface_alloc_storage)(struct pipe_winsys *ws, struct pipe_surface *surf, @@ -104,7 +105,7 @@ struct pipe_winsys * alignment indicates the client's alignment requirements, eg for * SSE instructions. */ - struct pipe_buffer *(*buffer_create)( struct pipe_winsys *sws, + struct pipe_buffer *(*buffer_create)( struct pipe_winsys *ws, unsigned alignment, unsigned usage, unsigned size ); @@ -130,7 +131,7 @@ struct pipe_winsys * Note that ptr may be accessed at any time upto the time when the * buffer is destroyed, so the data must not be freed before then. */ - struct pipe_buffer *(*user_buffer_create)(struct pipe_winsys *sws, + struct pipe_buffer *(*user_buffer_create)(struct pipe_winsys *ws, void *ptr, unsigned bytes); @@ -138,45 +139,40 @@ struct pipe_winsys * Map the entire data store of a buffer object into the client's address. * flags is bitmask of PIPE_BUFFER_USAGE_CPU_READ/WRITE flags. */ - void *(*buffer_map)( struct pipe_winsys *sws, + void *(*buffer_map)( struct pipe_winsys *ws, struct pipe_buffer *buf, unsigned usage ); - void (*buffer_unmap)( struct pipe_winsys *sws, + void (*buffer_unmap)( struct pipe_winsys *ws, struct pipe_buffer *buf ); - void (*buffer_destroy)( struct pipe_winsys *sws, + void (*buffer_destroy)( struct pipe_winsys *ws, struct pipe_buffer *buf ); /** Set ptr = fence, with reference counting */ - void (*fence_reference)( struct pipe_winsys *sws, + void (*fence_reference)( struct pipe_winsys *ws, struct pipe_fence_handle **ptr, struct pipe_fence_handle *fence ); /** * Checks whether the fence has been signalled. - * - * The meaning of flag is pipe-driver specific. - * - * Returns zero if it has. + * \param flags driver-specific meaning + * \return zero on success. */ - int (*fence_signalled)( struct pipe_winsys *sws, + int (*fence_signalled)( struct pipe_winsys *ws, struct pipe_fence_handle *fence, unsigned flag ); /** * Wait for the fence to finish. - * - * The meaning of flag is pipe-driver specific. - * - * Returns zero on success. + * \param flags driver-specific meaning + * \return zero on success. */ - int (*fence_finish)( struct pipe_winsys *sws, + int (*fence_finish)( struct pipe_winsys *ws, struct pipe_fence_handle *fence, unsigned flag ); - }; diff --git a/src/gallium/winsys/xlib/fakeglx.c b/src/gallium/winsys/xlib/fakeglx.c index 902a755075..2c0075e934 100644 --- a/src/gallium/winsys/xlib/fakeglx.c +++ b/src/gallium/winsys/xlib/fakeglx.c @@ -1459,6 +1459,13 @@ Fake_glXMakeContextCurrent( Display *dpy, GLXDrawable draw, GLXDrawable read, GLXContext ctx ) { struct fake_glx_context *glxCtx = (struct fake_glx_context *) ctx; + static boolean firsttime = 1, no_rast = 0; + + if (firsttime) { + no_rast = getenv("SP_NO_RAST") != NULL; + firsttime = 0; + } + if (ctx && draw && read) { XMesaBuffer drawBuffer, readBuffer; @@ -1504,6 +1511,14 @@ Fake_glXMakeContextCurrent( Display *dpy, GLXDrawable draw, #endif } + if (no_rast && + MakeCurrent_PrevContext == ctx && + MakeCurrent_PrevDrawable == draw && + MakeCurrent_PrevReadable == read && + MakeCurrent_PrevDrawBuffer == drawBuffer && + MakeCurrent_PrevReadBuffer == readBuffer) + return True; + MakeCurrent_PrevContext = ctx; MakeCurrent_PrevDrawable = draw; MakeCurrent_PrevReadable = read; @@ -1674,6 +1689,15 @@ static void Fake_glXSwapBuffers( Display *dpy, GLXDrawable drawable ) { XMesaBuffer buffer = XMesaFindBuffer( dpy, drawable ); + static boolean firsttime = 1, no_rast = 0; + + if (firsttime) { + no_rast = getenv("SP_NO_RAST") != NULL; + firsttime = 0; + } + + if (no_rast) + return; if (buffer) { XMesaSwapBuffers(buffer); diff --git a/src/gallium/winsys/xlib/xm_api.c b/src/gallium/winsys/xlib/xm_api.c index a82d3c990e..26b722f343 100644 --- a/src/gallium/winsys/xlib/xm_api.c +++ b/src/gallium/winsys/xlib/xm_api.c @@ -110,6 +110,9 @@ int xmesa_check_for_xshm( XMesaDisplay *display ) int major, minor, ignore; Bool pixmaps; + if (getenv("SP_NO_RAST")) + return 0; + if (getenv("MESA_NOSHM")) { return 0; } @@ -814,9 +817,6 @@ XMesaContext XMesaCreateContext( XMesaVisual v, XMesaContext share_list ) mesaCtx->Const.CheckArrayBounds = GL_TRUE; #endif - /* finish up xmesa context initializations */ - c->st->haveFramebufferSurfaces = GL_TRUE; - return c; } diff --git a/src/gallium/winsys/xlib/xm_winsys.c b/src/gallium/winsys/xlib/xm_winsys.c index 9a20bdfb69..14c3892559 100644 --- a/src/gallium/winsys/xlib/xm_winsys.c +++ b/src/gallium/winsys/xlib/xm_winsys.c @@ -37,6 +37,7 @@ #include "xmesaP.h" #undef ASSERT +#undef Elements #include "pipe/p_winsys.h" #include "pipe/p_format.h" @@ -57,6 +58,7 @@ /** + * Subclass of pipe_buffer for Xlib winsys. * Low-level OS/window system memory buffer */ struct xm_buffer @@ -73,32 +75,22 @@ struct xm_buffer #endif }; -#if defined(USE_XSHM) && !defined(XFree86Server) -# define XSHM_ENABLED(b) ((b)->shm) -#else -# define XSHM_ENABLED(b) 0 -#endif +/** + * Subclass of pipe_surface for Xlib winsys + */ struct xmesa_surface { struct pipe_surface surface; int tileSize; + boolean no_swap; }; /** - * Derived from softpipe_winsys. - * We just need one extra field which indicates the pixel format to use for - * drawing surfaces so that we're compatible with the XVisual/window format. + * Subclass of pipe_winsys for Xlib winsys */ -struct xmesa_softpipe_winsys -{ - struct softpipe_winsys spws; - enum pipe_format pixelformat; -}; - - struct xmesa_pipe_winsys { struct pipe_winsys base; @@ -107,35 +99,147 @@ struct xmesa_pipe_winsys }; -static void alloc_shm_ximage(struct xm_buffer *b, struct xmesa_buffer *xmb, - unsigned width, unsigned height); /** Cast wrapper */ static INLINE struct xmesa_surface * xmesa_surface(struct pipe_surface *ps) { -// assert(0); return (struct xmesa_surface *) ps; } -/** cast wrapper */ -static INLINE struct xmesa_softpipe_winsys * -xmesa_softpipe_winsys(struct softpipe_winsys *spws) + +/** Cast wrapper */ +static INLINE struct xm_buffer * +xm_buffer( struct pipe_buffer *buf ) { - return (struct xmesa_softpipe_winsys *) spws; + return (struct xm_buffer *)buf; } + /** - * Turn the softpipe opaque buffer pointer into a dri_bufmgr opaque - * buffer pointer... + * X Shared Memory Image extension code */ -static INLINE struct xm_buffer * -xm_buffer( struct pipe_buffer *buf ) +#if defined(USE_XSHM) && !defined(XFree86Server) + +#define XSHM_ENABLED(b) ((b)->shm) + +static volatile int mesaXErrorFlag = 0; + +/** + * Catches potential Xlib errors. + */ +static int +mesaHandleXError(XMesaDisplay *dpy, XErrorEvent *event) { - return (struct xm_buffer *)buf; + (void) dpy; + (void) event; + mesaXErrorFlag = 1; + return 0; +} + + +static GLboolean alloc_shm(struct xm_buffer *buf, unsigned size) +{ + XShmSegmentInfo *const shminfo = & buf->shminfo; + + shminfo->shmid = shmget(IPC_PRIVATE, size, IPC_CREAT|0777); + if (shminfo->shmid < 0) { + return GL_FALSE; + } + + shminfo->shmaddr = (char *) shmat(shminfo->shmid, 0, 0); + if (shminfo->shmaddr == (char *) -1) { + shmctl(shminfo->shmid, IPC_RMID, 0); + return GL_FALSE; + } + + shminfo->readOnly = False; + return GL_TRUE; } +/** + * Allocate a shared memory XImage back buffer for the given XMesaBuffer. + */ +static void +alloc_shm_ximage(struct xm_buffer *b, struct xmesa_buffer *xmb, + unsigned width, unsigned height) +{ + /* + * We have to do a _lot_ of error checking here to be sure we can + * really use the XSHM extension. It seems different servers trigger + * errors at different points if the extension won't work. Therefore + * we have to be very careful... + */ +#if 0 + GC gc; +#endif + int (*old_handler)(XMesaDisplay *, XErrorEvent *); + + b->tempImage = XShmCreateImage(xmb->xm_visual->display, + xmb->xm_visual->visinfo->visual, + xmb->xm_visual->visinfo->depth, + ZPixmap, + NULL, + &b->shminfo, + width, height); + if (b->tempImage == NULL) { + b->shm = 0; + return; + } + + + mesaXErrorFlag = 0; + old_handler = XSetErrorHandler(mesaHandleXError); + /* This may trigger the X protocol error we're ready to catch: */ + XShmAttach(xmb->xm_visual->display, &b->shminfo); + XSync(xmb->xm_visual->display, False); + + if (mesaXErrorFlag) { + /* we are on a remote display, this error is normal, don't print it */ + XFlush(xmb->xm_visual->display); + mesaXErrorFlag = 0; + XDestroyImage(b->tempImage); + b->tempImage = NULL; + b->shm = 0; + (void) XSetErrorHandler(old_handler); + return; + } + + + /* Finally, try an XShmPutImage to be really sure the extension works */ +#if 0 + gc = XCreateGC(xmb->xm_visual->display, xmb->drawable, 0, NULL); + XShmPutImage(xmb->xm_visual->display, xmb->drawable, gc, + b->tempImage, 0, 0, 0, 0, 1, 1 /*one pixel*/, False); + XSync(xmb->xm_visual->display, False); + XFreeGC(xmb->xm_visual->display, gc); + (void) XSetErrorHandler(old_handler); + if (mesaXErrorFlag) { + XFlush(xmb->xm_visual->display); + mesaXErrorFlag = 0; + XDestroyImage(b->tempImage); + b->tempImage = NULL; + b->shm = 0; + return; + } +#endif +} + +#else + +#define XSHM_ENABLED(b) 0 + +static void +alloc_shm_ximage(struct xm_buffer *b, struct xmesa_buffer *xmb, + unsigned width, unsigned height) +{ + b->shm = 0; +} +#endif /* USE_XSHM */ + + + /* Most callbacks map direcly onto dri_bufmgr operations: */ @@ -252,6 +356,9 @@ xmesa_display_surface(XMesaBuffer b, const struct pipe_surface *surf) const struct xmesa_surface *xm_surf = xmesa_surface((struct pipe_surface *) surf); + if (xm_surf->no_swap) + return; + if (xm_surf->tileSize) { xmesa_display_surface_tiled(b, surf); return; @@ -292,10 +399,9 @@ xm_flush_frontbuffer(struct pipe_winsys *pws, struct pipe_surface *surf, void *context_private) { - /* The Xlib driver's front color surfaces are actually X Windows so - * this flush is a no-op. - * If we instead did front buffer rendering to a temporary XImage, - * this would be the place to copy the Ximage to the on-screen Window. + /* + * The front color buffer is actually just another XImage buffer. + * This function copies that XImage to the actual X Window. */ XMesaContext xmctx = (XMesaContext) context_private; xmesa_display_surface(xmctx->xm_buffer, surf); @@ -310,119 +416,6 @@ xm_get_name(struct pipe_winsys *pws) } -#if defined(USE_XSHM) && !defined(XFree86Server) -static volatile int mesaXErrorFlag = 0; - -/** - * Catches potential Xlib errors. - */ -static int -mesaHandleXError(XMesaDisplay *dpy, XErrorEvent *event) -{ - (void) dpy; - (void) event; - mesaXErrorFlag = 1; - return 0; -} - - -static GLboolean alloc_shm(struct xm_buffer *buf, unsigned size) -{ - XShmSegmentInfo *const shminfo = & buf->shminfo; - - shminfo->shmid = shmget(IPC_PRIVATE, size, IPC_CREAT|0777); - if (shminfo->shmid < 0) { - return GL_FALSE; - } - - shminfo->shmaddr = (char *) shmat(shminfo->shmid, 0, 0); - if (shminfo->shmaddr == (char *) -1) { - shmctl(shminfo->shmid, IPC_RMID, 0); - return GL_FALSE; - } - - shminfo->readOnly = False; - return GL_TRUE; -} - - -/** - * Allocate a shared memory XImage back buffer for the given XMesaBuffer. - */ -static void -alloc_shm_ximage(struct xm_buffer *b, struct xmesa_buffer *xmb, - unsigned width, unsigned height) -{ - /* - * We have to do a _lot_ of error checking here to be sure we can - * really use the XSHM extension. It seems different servers trigger - * errors at different points if the extension won't work. Therefore - * we have to be very careful... - */ -#if 0 - GC gc; -#endif - int (*old_handler)(XMesaDisplay *, XErrorEvent *); - - b->tempImage = XShmCreateImage(xmb->xm_visual->display, - xmb->xm_visual->visinfo->visual, - xmb->xm_visual->visinfo->depth, - ZPixmap, - NULL, - &b->shminfo, - width, height); - if (b->tempImage == NULL) { - b->shm = 0; - return; - } - - - mesaXErrorFlag = 0; - old_handler = XSetErrorHandler(mesaHandleXError); - /* This may trigger the X protocol error we're ready to catch: */ - XShmAttach(xmb->xm_visual->display, &b->shminfo); - XSync(xmb->xm_visual->display, False); - - if (mesaXErrorFlag) { - /* we are on a remote display, this error is normal, don't print it */ - XFlush(xmb->xm_visual->display); - mesaXErrorFlag = 0; - XDestroyImage(b->tempImage); - b->tempImage = NULL; - b->shm = 0; - (void) XSetErrorHandler(old_handler); - return; - } - - - /* Finally, try an XShmPutImage to be really sure the extension works */ -#if 0 - gc = XCreateGC(xmb->xm_visual->display, xmb->drawable, 0, NULL); - XShmPutImage(xmb->xm_visual->display, xmb->drawable, gc, - b->tempImage, 0, 0, 0, 0, 1, 1 /*one pixel*/, False); - XSync(xmb->xm_visual->display, False); - XFreeGC(xmb->xm_visual->display, gc); - (void) XSetErrorHandler(old_handler); - if (mesaXErrorFlag) { - XFlush(xmb->xm_visual->display); - mesaXErrorFlag = 0; - XDestroyImage(b->tempImage); - b->tempImage = NULL; - b->shm = 0; - return; - } -#endif -} -#else -static void -alloc_shm_ximage(struct xm_buffer *b, struct xmesa_buffer *xmb, - unsigned width, unsigned height) -{ - b->shm = 0; -} -#endif - - static struct pipe_buffer * xm_buffer_create(struct pipe_winsys *pws, unsigned alignment, @@ -522,13 +515,19 @@ xm_surface_alloc_storage(struct pipe_winsys *winsys, /** - * Called via pipe->surface_alloc() to create new surfaces (textures, - * renderbuffers, etc. + * Called via winsys->surface_alloc() to create new surfaces. */ static struct pipe_surface * xm_surface_alloc(struct pipe_winsys *ws) { struct xmesa_surface *xms = CALLOC_STRUCT(xmesa_surface); + static boolean no_swap = 0; + static boolean firsttime = 1; + + if (firsttime) { + no_swap = getenv("SP_NO_RAST") != NULL; + firsttime = 0; + } assert(ws); @@ -540,7 +539,9 @@ xm_surface_alloc(struct pipe_winsys *ws) xms->tileSize = 32; /** probably temporary */ } #endif - + + xms->no_swap = no_swap; + return &xms->surface; } @@ -598,10 +599,19 @@ xmesa_get_pipe_winsys_aub(struct xmesa_visual *xm_vis) { static struct xmesa_pipe_winsys *ws = NULL; - if (!ws && getenv("XM_AUB")) { + if (!ws) { ws = (struct xmesa_pipe_winsys *) xmesa_create_pipe_winsys_aub(); } - else if (!ws) { + return &ws->base; +} + + +static struct pipe_winsys * +xmesa_get_pipe_winsys(struct xmesa_visual *xm_vis) +{ + static struct xmesa_pipe_winsys *ws = NULL; + + if (!ws) { ws = CALLOC_STRUCT(xmesa_pipe_winsys); ws->xm_visual = xm_vis; @@ -632,45 +642,19 @@ xmesa_get_pipe_winsys_aub(struct xmesa_visual *xm_vis) } -/** - * Called via softpipe_winsys->is_format_supported(). - * This function is only called to test formats for front/back color surfaces. - * The winsys being queried will have been created at glXCreateContext - * time, with a pixel format corresponding to the context's visual. - */ -static boolean -xmesa_is_format_supported(struct softpipe_winsys *sws, - enum pipe_format format) -{ - struct xmesa_softpipe_winsys *xmws = xmesa_softpipe_winsys(sws); - return (format == xmws->pixelformat); -} - - -/** - * Return pointer to a softpipe_winsys object. - */ -static struct softpipe_winsys * -xmesa_get_softpipe_winsys(uint pixelformat) -{ - struct xmesa_softpipe_winsys *xmws - = CALLOC_STRUCT(xmesa_softpipe_winsys); - if (!xmws) - return NULL; - - xmws->spws.is_format_supported = xmesa_is_format_supported; - xmws->pixelformat = pixelformat; - - return &xmws->spws; -} - - struct pipe_context * xmesa_create_pipe_context(XMesaContext xmesa, uint pixelformat) { - struct pipe_winsys *pws = xmesa_get_pipe_winsys_aub(xmesa->xm_visual); + struct pipe_winsys *pws; struct pipe_context *pipe; + if (getenv("XM_AUB")) { + pws = xmesa_get_pipe_winsys_aub(xmesa->xm_visual); + } + else { + pws = xmesa_get_pipe_winsys(xmesa->xm_visual); + } + #ifdef GALLIUM_CELL if (!getenv("GALLIUM_NOCELL")) { struct cell_winsys *cws = cell_get_winsys(pixelformat); @@ -681,10 +665,9 @@ xmesa_create_pipe_context(XMesaContext xmesa, uint pixelformat) else #endif { - struct softpipe_winsys *spws = xmesa_get_softpipe_winsys(pixelformat); struct pipe_screen *screen = softpipe_create_screen(pws); - pipe = softpipe_create(screen, pws, spws); + pipe = softpipe_create(screen, pws, NULL); } if (pipe) diff --git a/src/glx/x11/indirect.c b/src/glx/x11/indirect.c index 87c523383b..fbb2a91956 100644 --- a/src/glx/x11/indirect.c +++ b/src/glx/x11/indirect.c @@ -7470,26 +7470,6 @@ __indirect_glGetProgramivARB(GLenum target, GLenum pname, GLint * params) #define X_GLrop_ProgramEnvParameter4dvARB 4185 void -__indirect_glProgramEnvParameter4dARB(GLenum target, GLuint index, GLdouble x, - GLdouble y, GLdouble z, GLdouble w) -{ - __GLXcontext *const gc = __glXGetCurrentContext(); - const GLuint cmdlen = 44; - emit_header(gc->pc, X_GLrop_ProgramEnvParameter4dvARB, cmdlen); - (void) memcpy((void *) (gc->pc + 4), (void *) (&target), 4); - (void) memcpy((void *) (gc->pc + 8), (void *) (&index), 4); - (void) memcpy((void *) (gc->pc + 12), (void *) (&x), 8); - (void) memcpy((void *) (gc->pc + 20), (void *) (&y), 8); - (void) memcpy((void *) (gc->pc + 28), (void *) (&z), 8); - (void) memcpy((void *) (gc->pc + 36), (void *) (&w), 8); - gc->pc += cmdlen; - if (__builtin_expect(gc->pc > gc->limit, 0)) { - (void) __glXFlushRenderBuffer(gc, gc->pc); - } -} - -#define X_GLrop_ProgramEnvParameter4dvARB 4185 -void __indirect_glProgramEnvParameter4dvARB(GLenum target, GLuint index, const GLdouble * params) { @@ -7507,26 +7487,6 @@ __indirect_glProgramEnvParameter4dvARB(GLenum target, GLuint index, #define X_GLrop_ProgramEnvParameter4fvARB 4184 void -__indirect_glProgramEnvParameter4fARB(GLenum target, GLuint index, GLfloat x, - GLfloat y, GLfloat z, GLfloat w) -{ - __GLXcontext *const gc = __glXGetCurrentContext(); - const GLuint cmdlen = 28; - emit_header(gc->pc, X_GLrop_ProgramEnvParameter4fvARB, cmdlen); - (void) memcpy((void *) (gc->pc + 4), (void *) (&target), 4); - (void) memcpy((void *) (gc->pc + 8), (void *) (&index), 4); - (void) memcpy((void *) (gc->pc + 12), (void *) (&x), 4); - (void) memcpy((void *) (gc->pc + 16), (void *) (&y), 4); - (void) memcpy((void *) (gc->pc + 20), (void *) (&z), 4); - (void) memcpy((void *) (gc->pc + 24), (void *) (&w), 4); - gc->pc += cmdlen; - if (__builtin_expect(gc->pc > gc->limit, 0)) { - (void) __glXFlushRenderBuffer(gc, gc->pc); - } -} - -#define X_GLrop_ProgramEnvParameter4fvARB 4184 -void __indirect_glProgramEnvParameter4fvARB(GLenum target, GLuint index, const GLfloat * params) { @@ -9104,80 +9064,6 @@ __indirect_glLoadProgramNV(GLenum target, GLuint id, GLsizei len, } } -#define X_GLrop_ProgramParameter4dvNV 4185 -void -__indirect_glProgramParameter4dNV(GLenum target, GLuint index, GLdouble x, - GLdouble y, GLdouble z, GLdouble w) -{ - __GLXcontext *const gc = __glXGetCurrentContext(); - const GLuint cmdlen = 44; - emit_header(gc->pc, X_GLrop_ProgramParameter4dvNV, cmdlen); - (void) memcpy((void *) (gc->pc + 4), (void *) (&target), 4); - (void) memcpy((void *) (gc->pc + 8), (void *) (&index), 4); - (void) memcpy((void *) (gc->pc + 12), (void *) (&x), 8); - (void) memcpy((void *) (gc->pc + 20), (void *) (&y), 8); - (void) memcpy((void *) (gc->pc + 28), (void *) (&z), 8); - (void) memcpy((void *) (gc->pc + 36), (void *) (&w), 8); - gc->pc += cmdlen; - if (__builtin_expect(gc->pc > gc->limit, 0)) { - (void) __glXFlushRenderBuffer(gc, gc->pc); - } -} - -#define X_GLrop_ProgramParameter4dvNV 4185 -void -__indirect_glProgramParameter4dvNV(GLenum target, GLuint index, - const GLdouble * params) -{ - __GLXcontext *const gc = __glXGetCurrentContext(); - const GLuint cmdlen = 44; - emit_header(gc->pc, X_GLrop_ProgramParameter4dvNV, cmdlen); - (void) memcpy((void *) (gc->pc + 4), (void *) (&target), 4); - (void) memcpy((void *) (gc->pc + 8), (void *) (&index), 4); - (void) memcpy((void *) (gc->pc + 12), (void *) (params), 32); - gc->pc += cmdlen; - if (__builtin_expect(gc->pc > gc->limit, 0)) { - (void) __glXFlushRenderBuffer(gc, gc->pc); - } -} - -#define X_GLrop_ProgramParameter4fvNV 4184 -void -__indirect_glProgramParameter4fNV(GLenum target, GLuint index, GLfloat x, - GLfloat y, GLfloat z, GLfloat w) -{ - __GLXcontext *const gc = __glXGetCurrentContext(); - const GLuint cmdlen = 28; - emit_header(gc->pc, X_GLrop_ProgramParameter4fvNV, cmdlen); - (void) memcpy((void *) (gc->pc + 4), (void *) (&target), 4); - (void) memcpy((void *) (gc->pc + 8), (void *) (&index), 4); - (void) memcpy((void *) (gc->pc + 12), (void *) (&x), 4); - (void) memcpy((void *) (gc->pc + 16), (void *) (&y), 4); - (void) memcpy((void *) (gc->pc + 20), (void *) (&z), 4); - (void) memcpy((void *) (gc->pc + 24), (void *) (&w), 4); - gc->pc += cmdlen; - if (__builtin_expect(gc->pc > gc->limit, 0)) { - (void) __glXFlushRenderBuffer(gc, gc->pc); - } -} - -#define X_GLrop_ProgramParameter4fvNV 4184 -void -__indirect_glProgramParameter4fvNV(GLenum target, GLuint index, - const GLfloat * params) -{ - __GLXcontext *const gc = __glXGetCurrentContext(); - const GLuint cmdlen = 28; - emit_header(gc->pc, X_GLrop_ProgramParameter4fvNV, cmdlen); - (void) memcpy((void *) (gc->pc + 4), (void *) (&target), 4); - (void) memcpy((void *) (gc->pc + 8), (void *) (&index), 4); - (void) memcpy((void *) (gc->pc + 12), (void *) (params), 16); - gc->pc += cmdlen; - if (__builtin_expect(gc->pc > gc->limit, 0)) { - (void) __glXFlushRenderBuffer(gc, gc->pc); - } -} - #define X_GLrop_ProgramParameters4dvNV 4187 void __indirect_glProgramParameters4dvNV(GLenum target, GLuint index, GLuint num, diff --git a/src/glx/x11/indirect.h b/src/glx/x11/indirect.h index 630062e233..f8c88b36bb 100644 --- a/src/glx/x11/indirect.h +++ b/src/glx/x11/indirect.h @@ -517,9 +517,7 @@ extern HIDDEN void __indirect_glGetProgramivARB(GLenum target, GLenum pname, GLi extern HIDDEN void __indirect_glGetVertexAttribdvARB(GLuint index, GLenum pname, GLdouble * params); extern HIDDEN void __indirect_glGetVertexAttribfvARB(GLuint index, GLenum pname, GLfloat * params); extern HIDDEN void __indirect_glGetVertexAttribivARB(GLuint index, GLenum pname, GLint * params); -extern HIDDEN void __indirect_glProgramEnvParameter4dARB(GLenum target, GLuint index, GLdouble x, GLdouble y, GLdouble z, GLdouble w); extern HIDDEN void __indirect_glProgramEnvParameter4dvARB(GLenum target, GLuint index, const GLdouble * params); -extern HIDDEN void __indirect_glProgramEnvParameter4fARB(GLenum target, GLuint index, GLfloat x, GLfloat y, GLfloat z, GLfloat w); extern HIDDEN void __indirect_glProgramEnvParameter4fvARB(GLenum target, GLuint index, const GLfloat * params); extern HIDDEN void __indirect_glProgramLocalParameter4dARB(GLenum target, GLuint index, GLdouble x, GLdouble y, GLdouble z, GLdouble w); extern HIDDEN void __indirect_glProgramLocalParameter4dvARB(GLenum target, GLuint index, const GLdouble * params); @@ -639,10 +637,6 @@ extern HIDDEN void __indirect_glGetVertexAttribfvNV(GLuint index, GLenum pname, extern HIDDEN void __indirect_glGetVertexAttribivNV(GLuint index, GLenum pname, GLint * params); extern HIDDEN GLboolean __indirect_glIsProgramNV(GLuint program); extern HIDDEN void __indirect_glLoadProgramNV(GLenum target, GLuint id, GLsizei len, const GLubyte * program); -extern HIDDEN void __indirect_glProgramParameter4dNV(GLenum target, GLuint index, GLdouble x, GLdouble y, GLdouble z, GLdouble w); -extern HIDDEN void __indirect_glProgramParameter4dvNV(GLenum target, GLuint index, const GLdouble * params); -extern HIDDEN void __indirect_glProgramParameter4fNV(GLenum target, GLuint index, GLfloat x, GLfloat y, GLfloat z, GLfloat w); -extern HIDDEN void __indirect_glProgramParameter4fvNV(GLenum target, GLuint index, const GLfloat * params); extern HIDDEN void __indirect_glProgramParameters4dvNV(GLenum target, GLuint index, GLuint num, const GLdouble * params); extern HIDDEN void __indirect_glProgramParameters4fvNV(GLenum target, GLuint index, GLuint num, const GLfloat * params); extern HIDDEN void __indirect_glRequestResidentProgramsNV(GLsizei n, const GLuint * ids); diff --git a/src/glx/x11/indirect_init.c b/src/glx/x11/indirect_init.c index aaa70c8796..479184337c 100644 --- a/src/glx/x11/indirect_init.c +++ b/src/glx/x11/indirect_init.c @@ -526,9 +526,7 @@ __GLapi * __glXNewIndirectAPI( void ) glAPI->GetVertexAttribdvARB = __indirect_glGetVertexAttribdvARB; glAPI->GetVertexAttribfvARB = __indirect_glGetVertexAttribfvARB; glAPI->GetVertexAttribivARB = __indirect_glGetVertexAttribivARB; - glAPI->ProgramEnvParameter4dARB = __indirect_glProgramEnvParameter4dARB; glAPI->ProgramEnvParameter4dvARB = __indirect_glProgramEnvParameter4dvARB; - glAPI->ProgramEnvParameter4fARB = __indirect_glProgramEnvParameter4fARB; glAPI->ProgramEnvParameter4fvARB = __indirect_glProgramEnvParameter4fvARB; glAPI->ProgramLocalParameter4dARB = __indirect_glProgramLocalParameter4dARB; glAPI->ProgramLocalParameter4dvARB = __indirect_glProgramLocalParameter4dvARB; @@ -681,10 +679,6 @@ __GLapi * __glXNewIndirectAPI( void ) glAPI->GetVertexAttribivNV = __indirect_glGetVertexAttribivNV; glAPI->IsProgramNV = __indirect_glIsProgramNV; glAPI->LoadProgramNV = __indirect_glLoadProgramNV; - glAPI->ProgramParameter4dNV = __indirect_glProgramParameter4dNV; - glAPI->ProgramParameter4dvNV = __indirect_glProgramParameter4dvNV; - glAPI->ProgramParameter4fNV = __indirect_glProgramParameter4fNV; - glAPI->ProgramParameter4fvNV = __indirect_glProgramParameter4fvNV; glAPI->ProgramParameters4dvNV = __indirect_glProgramParameters4dvNV; glAPI->ProgramParameters4fvNV = __indirect_glProgramParameters4fvNV; glAPI->RequestResidentProgramsNV = __indirect_glRequestResidentProgramsNV; diff --git a/src/mesa/SConscript b/src/mesa/SConscript index b816a21b38..6b99dacf2a 100644 --- a/src/mesa/SConscript +++ b/src/mesa/SConscript @@ -12,9 +12,19 @@ env.Append(CPPPATH = [ '#/src/mesa/main', ]) -env.Append(CFLAGS = [ - '-std=c99', -]) +if gcc: + env.Append(CFLAGS = [ + '-std=c99', + ]) + +# x86 assembly +if x86 and gcc: + env.Append(CPPDEFINES = [ + 'USE_X86_ASM', + 'USE_MMX_ASM', + 'USE_3DNOW_ASM', + 'USE_SSE_ASM', + ]) ####################################################################### @@ -184,6 +194,7 @@ SHADER_SOURCES = [ 'shader/prog_parameter.c', 'shader/prog_print.c', 'shader/prog_statevars.c', + 'shader/prog_uniform.c', 'shader/programopt.c', 'shader/shader_api.c', ] @@ -221,7 +232,6 @@ ASM_C_SOURCES = [ 'x86/x86.c', 'x86/3dnow.c', 'x86/sse.c', - 'x86/rtasm/x86sse.c', 'sparc/sparc.c', 'ppc/common_ppc.c', 'x86-64/x86-64.c', @@ -269,7 +279,7 @@ SPARC_API = [ 'sparc/glapi_sparc.S', ] -if x86: +if x86 and gcc: ASM_SOURCES = ASM_C_SOURCES + X86_SOURCES API_SOURCES = X86_API else: diff --git a/src/mesa/drivers/dri/common/extension_helper.h b/src/mesa/drivers/dri/common/extension_helper.h index 065c5d8dae..65e96657b8 100644 --- a/src/mesa/drivers/dri/common/extension_helper.h +++ b/src/mesa/drivers/dri/common/extension_helper.h @@ -40,13 +40,6 @@ static const char UniformMatrix3fvARB_names[] = ""; #endif -#if defined(need_GL_NV_vertex_program) -static const char ProgramParameter4fNV_names[] = - "iiffff\0" /* Parameter signature */ - "glProgramParameter4fNV\0" - ""; -#endif - #if defined(need_GL_VERSION_1_3) || defined(need_GL_ARB_multisample) static const char SampleCoverageARB_names[] = "fi\0" /* Parameter signature */ @@ -572,13 +565,6 @@ static const char MatrixIndexusvARB_names[] = ""; #endif -#if defined(need_GL_NV_vertex_program) -static const char ProgramParameter4dvNV_names[] = - "iip\0" /* Parameter signature */ - "glProgramParameter4dvNV\0" - ""; -#endif - #if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program) static const char DisableVertexAttribArrayARB_names[] = "i\0" /* Parameter signature */ @@ -967,13 +953,6 @@ static const char GenerateMipmapEXT_names[] = ""; #endif -#if defined(need_GL_NV_vertex_program) -static const char ProgramParameter4dNV_names[] = - "iidddd\0" /* Parameter signature */ - "glProgramParameter4dNV\0" - ""; -#endif - #if defined(need_GL_ATI_fragment_shader) static const char SetFragmentShaderConstantATI_names[] = "ip\0" /* Parameter signature */ @@ -1311,10 +1290,11 @@ static const char Color3fVertex3fSUN_names[] = ""; #endif -#if defined(need_GL_ARB_vertex_program) +#if defined(need_GL_ARB_vertex_program) || defined(need_GL_NV_vertex_program) static const char ProgramEnvParameter4fvARB_names[] = "iip\0" /* Parameter signature */ "glProgramEnvParameter4fvARB\0" + "glProgramParameter4fvNV\0" ""; #endif @@ -2035,13 +2015,6 @@ static const char WeightfvARB_names[] = ""; #endif -#if defined(need_GL_NV_vertex_program) -static const char ProgramParameter4fvNV_names[] = - "iip\0" /* Parameter signature */ - "glProgramParameter4fvNV\0" - ""; -#endif - #if defined(need_GL_MESA_window_pos) static const char WindowPos4fMESA_names[] = "ffff\0" /* Parameter signature */ @@ -2432,10 +2405,11 @@ static const char GetBufferPointervARB_names[] = ""; #endif -#if defined(need_GL_ARB_vertex_program) +#if defined(need_GL_ARB_vertex_program) || defined(need_GL_NV_vertex_program) static const char ProgramEnvParameter4fARB_names[] = "iiffff\0" /* Parameter signature */ "glProgramEnvParameter4fARB\0" + "glProgramParameter4fNV\0" ""; #endif @@ -2803,10 +2777,11 @@ static const char ReplacementCodePointerSUN_names[] = ""; #endif -#if defined(need_GL_ARB_vertex_program) +#if defined(need_GL_ARB_vertex_program) || defined(need_GL_NV_vertex_program) static const char ProgramEnvParameter4dARB_names[] = "iidddd\0" /* Parameter signature */ "glProgramEnvParameter4dARB\0" + "glProgramParameter4dNV\0" ""; #endif @@ -3660,10 +3635,11 @@ static const char GetColorTableParameteriv_names[] = ""; #endif -#if defined(need_GL_ARB_vertex_program) +#if defined(need_GL_ARB_vertex_program) || defined(need_GL_NV_vertex_program) static const char ProgramEnvParameter4dvARB_names[] = "iip\0" /* Parameter signature */ "glProgramEnvParameter4dvARB\0" + "glProgramParameter4dvNV\0" ""; #endif @@ -5748,12 +5724,10 @@ static const struct dri_extension_function GL_NV_vertex_array_range_functions[] #if defined(need_GL_NV_vertex_program) static const struct dri_extension_function GL_NV_vertex_program_functions[] = { - { ProgramParameter4fNV_names, ProgramParameter4fNV_remap_index, -1 }, { VertexAttrib4ubvNV_names, VertexAttrib4ubvNV_remap_index, -1 }, { VertexAttrib4svNV_names, VertexAttrib4svNV_remap_index, -1 }, { VertexAttribs1dvNV_names, VertexAttribs1dvNV_remap_index, -1 }, { VertexAttrib1fvNV_names, VertexAttrib1fvNV_remap_index, -1 }, - { ProgramParameter4dvNV_names, ProgramParameter4dvNV_remap_index, -1 }, { VertexAttrib4fNV_names, VertexAttrib4fNV_remap_index, -1 }, { VertexAttrib2dNV_names, VertexAttrib2dNV_remap_index, -1 }, { VertexAttrib4ubNV_names, VertexAttrib4ubNV_remap_index, -1 }, @@ -5761,7 +5735,7 @@ static const struct dri_extension_function GL_NV_vertex_program_functions[] = { { VertexAttribs4fvNV_names, VertexAttribs4fvNV_remap_index, -1 }, { VertexAttrib2sNV_names, VertexAttrib2sNV_remap_index, -1 }, { VertexAttribs3fvNV_names, VertexAttribs3fvNV_remap_index, -1 }, - { ProgramParameter4dNV_names, ProgramParameter4dNV_remap_index, -1 }, + { ProgramEnvParameter4fvARB_names, ProgramEnvParameter4fvARB_remap_index, -1 }, { LoadProgramNV_names, LoadProgramNV_remap_index, -1 }, { VertexAttrib4fvNV_names, VertexAttrib4fvNV_remap_index, -1 }, { VertexAttrib3fNV_names, VertexAttrib3fNV_remap_index, -1 }, @@ -5771,14 +5745,15 @@ static const struct dri_extension_function GL_NV_vertex_program_functions[] = { { VertexAttrib2fvNV_names, VertexAttrib2fvNV_remap_index, -1 }, { VertexAttrib2dvNV_names, VertexAttrib2dvNV_remap_index, -1 }, { VertexAttrib1dvNV_names, VertexAttrib1dvNV_remap_index, -1 }, - { ProgramParameter4fvNV_names, ProgramParameter4fvNV_remap_index, -1 }, { VertexAttrib1svNV_names, VertexAttrib1svNV_remap_index, -1 }, + { ProgramEnvParameter4fARB_names, ProgramEnvParameter4fARB_remap_index, -1 }, { VertexAttribs2svNV_names, VertexAttribs2svNV_remap_index, -1 }, { GetVertexAttribivNV_names, GetVertexAttribivNV_remap_index, -1 }, { GetVertexAttribfvNV_names, GetVertexAttribfvNV_remap_index, -1 }, { VertexAttrib2svNV_names, VertexAttrib2svNV_remap_index, -1 }, { VertexAttribs1fvNV_names, VertexAttribs1fvNV_remap_index, -1 }, { IsProgramNV_names, IsProgramNV_remap_index, -1 }, + { ProgramEnvParameter4dARB_names, ProgramEnvParameter4dARB_remap_index, -1 }, { VertexAttrib2fNV_names, VertexAttrib2fNV_remap_index, -1 }, { RequestResidentProgramsNV_names, RequestResidentProgramsNV_remap_index, -1 }, { ExecuteProgramNV_names, ExecuteProgramNV_remap_index, -1 }, @@ -5791,6 +5766,7 @@ static const struct dri_extension_function GL_NV_vertex_program_functions[] = { { GetProgramivNV_names, GetProgramivNV_remap_index, -1 }, { GetVertexAttribdvNV_names, GetVertexAttribdvNV_remap_index, -1 }, { VertexAttrib3fvNV_names, VertexAttrib3fvNV_remap_index, -1 }, + { ProgramEnvParameter4dvARB_names, ProgramEnvParameter4dvARB_remap_index, -1 }, { VertexAttribs2fvNV_names, VertexAttribs2fvNV_remap_index, -1 }, { DeleteProgramsNV_names, DeleteProgramsNV_remap_index, -1 }, { GetVertexAttribPointervNV_names, GetVertexAttribPointervNV_remap_index, -1 }, diff --git a/src/mesa/glapi/dispatch.h b/src/mesa/glapi/dispatch.h index db23e44396..7123156085 100644 --- a/src/mesa/glapi/dispatch.h +++ b/src/mesa/glapi/dispatch.h @@ -2077,18 +2077,6 @@ #define CALL_LoadProgramNV(disp, parameters) (*((disp)->LoadProgramNV)) parameters #define GET_LoadProgramNV(disp) ((disp)->LoadProgramNV) #define SET_LoadProgramNV(disp, fn) ((disp)->LoadProgramNV = fn) -#define CALL_ProgramParameter4dNV(disp, parameters) (*((disp)->ProgramParameter4dNV)) parameters -#define GET_ProgramParameter4dNV(disp) ((disp)->ProgramParameter4dNV) -#define SET_ProgramParameter4dNV(disp, fn) ((disp)->ProgramParameter4dNV = fn) -#define CALL_ProgramParameter4dvNV(disp, parameters) (*((disp)->ProgramParameter4dvNV)) parameters -#define GET_ProgramParameter4dvNV(disp) ((disp)->ProgramParameter4dvNV) -#define SET_ProgramParameter4dvNV(disp, fn) ((disp)->ProgramParameter4dvNV = fn) -#define CALL_ProgramParameter4fNV(disp, parameters) (*((disp)->ProgramParameter4fNV)) parameters -#define GET_ProgramParameter4fNV(disp) ((disp)->ProgramParameter4fNV) -#define SET_ProgramParameter4fNV(disp, fn) ((disp)->ProgramParameter4fNV = fn) -#define CALL_ProgramParameter4fvNV(disp, parameters) (*((disp)->ProgramParameter4fvNV)) parameters -#define GET_ProgramParameter4fvNV(disp) ((disp)->ProgramParameter4fvNV) -#define SET_ProgramParameter4fvNV(disp, fn) ((disp)->ProgramParameter4fvNV = fn) #define CALL_ProgramParameters4dvNV(disp, parameters) (*((disp)->ProgramParameters4dvNV)) parameters #define GET_ProgramParameters4dvNV(disp) ((disp)->ProgramParameters4dvNV) #define SET_ProgramParameters4dvNV(disp, fn) ((disp)->ProgramParameters4dvNV = fn) @@ -2383,7 +2371,7 @@ #else -#define driDispatchRemapTable_size 366 +#define driDispatchRemapTable_size 362 extern int driDispatchRemapTable[ driDispatchRemapTable_size ]; #define AttachShader_remap_index 0 @@ -2651,107 +2639,103 @@ extern int driDispatchRemapTable[ driDispatchRemapTable_size ]; #define GetVertexAttribivNV_remap_index 262 #define IsProgramNV_remap_index 263 #define LoadProgramNV_remap_index 264 -#define ProgramParameter4dNV_remap_index 265 -#define ProgramParameter4dvNV_remap_index 266 -#define ProgramParameter4fNV_remap_index 267 -#define ProgramParameter4fvNV_remap_index 268 -#define ProgramParameters4dvNV_remap_index 269 -#define ProgramParameters4fvNV_remap_index 270 -#define RequestResidentProgramsNV_remap_index 271 -#define TrackMatrixNV_remap_index 272 -#define VertexAttrib1dNV_remap_index 273 -#define VertexAttrib1dvNV_remap_index 274 -#define VertexAttrib1fNV_remap_index 275 -#define VertexAttrib1fvNV_remap_index 276 -#define VertexAttrib1sNV_remap_index 277 -#define VertexAttrib1svNV_remap_index 278 -#define VertexAttrib2dNV_remap_index 279 -#define VertexAttrib2dvNV_remap_index 280 -#define VertexAttrib2fNV_remap_index 281 -#define VertexAttrib2fvNV_remap_index 282 -#define VertexAttrib2sNV_remap_index 283 -#define VertexAttrib2svNV_remap_index 284 -#define VertexAttrib3dNV_remap_index 285 -#define VertexAttrib3dvNV_remap_index 286 -#define VertexAttrib3fNV_remap_index 287 -#define VertexAttrib3fvNV_remap_index 288 -#define VertexAttrib3sNV_remap_index 289 -#define VertexAttrib3svNV_remap_index 290 -#define VertexAttrib4dNV_remap_index 291 -#define VertexAttrib4dvNV_remap_index 292 -#define VertexAttrib4fNV_remap_index 293 -#define VertexAttrib4fvNV_remap_index 294 -#define VertexAttrib4sNV_remap_index 295 -#define VertexAttrib4svNV_remap_index 296 -#define VertexAttrib4ubNV_remap_index 297 -#define VertexAttrib4ubvNV_remap_index 298 -#define VertexAttribPointerNV_remap_index 299 -#define VertexAttribs1dvNV_remap_index 300 -#define VertexAttribs1fvNV_remap_index 301 -#define VertexAttribs1svNV_remap_index 302 -#define VertexAttribs2dvNV_remap_index 303 -#define VertexAttribs2fvNV_remap_index 304 -#define VertexAttribs2svNV_remap_index 305 -#define VertexAttribs3dvNV_remap_index 306 -#define VertexAttribs3fvNV_remap_index 307 -#define VertexAttribs3svNV_remap_index 308 -#define VertexAttribs4dvNV_remap_index 309 -#define VertexAttribs4fvNV_remap_index 310 -#define VertexAttribs4svNV_remap_index 311 -#define VertexAttribs4ubvNV_remap_index 312 -#define AlphaFragmentOp1ATI_remap_index 313 -#define AlphaFragmentOp2ATI_remap_index 314 -#define AlphaFragmentOp3ATI_remap_index 315 -#define BeginFragmentShaderATI_remap_index 316 -#define BindFragmentShaderATI_remap_index 317 -#define ColorFragmentOp1ATI_remap_index 318 -#define ColorFragmentOp2ATI_remap_index 319 -#define ColorFragmentOp3ATI_remap_index 320 -#define DeleteFragmentShaderATI_remap_index 321 -#define EndFragmentShaderATI_remap_index 322 -#define GenFragmentShadersATI_remap_index 323 -#define PassTexCoordATI_remap_index 324 -#define SampleMapATI_remap_index 325 -#define SetFragmentShaderConstantATI_remap_index 326 -#define PointParameteriNV_remap_index 327 -#define PointParameterivNV_remap_index 328 -#define ActiveStencilFaceEXT_remap_index 329 -#define BindVertexArrayAPPLE_remap_index 330 -#define DeleteVertexArraysAPPLE_remap_index 331 -#define GenVertexArraysAPPLE_remap_index 332 -#define IsVertexArrayAPPLE_remap_index 333 -#define GetProgramNamedParameterdvNV_remap_index 334 -#define GetProgramNamedParameterfvNV_remap_index 335 -#define ProgramNamedParameter4dNV_remap_index 336 -#define ProgramNamedParameter4dvNV_remap_index 337 -#define ProgramNamedParameter4fNV_remap_index 338 -#define ProgramNamedParameter4fvNV_remap_index 339 -#define DepthBoundsEXT_remap_index 340 -#define BlendEquationSeparateEXT_remap_index 341 -#define BindFramebufferEXT_remap_index 342 -#define BindRenderbufferEXT_remap_index 343 -#define CheckFramebufferStatusEXT_remap_index 344 -#define DeleteFramebuffersEXT_remap_index 345 -#define DeleteRenderbuffersEXT_remap_index 346 -#define FramebufferRenderbufferEXT_remap_index 347 -#define FramebufferTexture1DEXT_remap_index 348 -#define FramebufferTexture2DEXT_remap_index 349 -#define FramebufferTexture3DEXT_remap_index 350 -#define GenFramebuffersEXT_remap_index 351 -#define GenRenderbuffersEXT_remap_index 352 -#define GenerateMipmapEXT_remap_index 353 -#define GetFramebufferAttachmentParameterivEXT_remap_index 354 -#define GetRenderbufferParameterivEXT_remap_index 355 -#define IsFramebufferEXT_remap_index 356 -#define IsRenderbufferEXT_remap_index 357 -#define RenderbufferStorageEXT_remap_index 358 -#define BlitFramebufferEXT_remap_index 359 -#define FramebufferTextureLayerEXT_remap_index 360 -#define StencilFuncSeparateATI_remap_index 361 -#define ProgramEnvParameters4fvEXT_remap_index 362 -#define ProgramLocalParameters4fvEXT_remap_index 363 -#define GetQueryObjecti64vEXT_remap_index 364 -#define GetQueryObjectui64vEXT_remap_index 365 +#define ProgramParameters4dvNV_remap_index 265 +#define ProgramParameters4fvNV_remap_index 266 +#define RequestResidentProgramsNV_remap_index 267 +#define TrackMatrixNV_remap_index 268 +#define VertexAttrib1dNV_remap_index 269 +#define VertexAttrib1dvNV_remap_index 270 +#define VertexAttrib1fNV_remap_index 271 +#define VertexAttrib1fvNV_remap_index 272 +#define VertexAttrib1sNV_remap_index 273 +#define VertexAttrib1svNV_remap_index 274 +#define VertexAttrib2dNV_remap_index 275 +#define VertexAttrib2dvNV_remap_index 276 +#define VertexAttrib2fNV_remap_index 277 +#define VertexAttrib2fvNV_remap_index 278 +#define VertexAttrib2sNV_remap_index 279 +#define VertexAttrib2svNV_remap_index 280 +#define VertexAttrib3dNV_remap_index 281 +#define VertexAttrib3dvNV_remap_index 282 +#define VertexAttrib3fNV_remap_index 283 +#define VertexAttrib3fvNV_remap_index 284 +#define VertexAttrib3sNV_remap_index 285 +#define VertexAttrib3svNV_remap_index 286 +#define VertexAttrib4dNV_remap_index 287 +#define VertexAttrib4dvNV_remap_index 288 +#define VertexAttrib4fNV_remap_index 289 +#define VertexAttrib4fvNV_remap_index 290 +#define VertexAttrib4sNV_remap_index 291 +#define VertexAttrib4svNV_remap_index 292 +#define VertexAttrib4ubNV_remap_index 293 +#define VertexAttrib4ubvNV_remap_index 294 +#define VertexAttribPointerNV_remap_index 295 +#define VertexAttribs1dvNV_remap_index 296 +#define VertexAttribs1fvNV_remap_index 297 +#define VertexAttribs1svNV_remap_index 298 +#define VertexAttribs2dvNV_remap_index 299 +#define VertexAttribs2fvNV_remap_index 300 +#define VertexAttribs2svNV_remap_index 301 +#define VertexAttribs3dvNV_remap_index 302 +#define VertexAttribs3fvNV_remap_index 303 +#define VertexAttribs3svNV_remap_index 304 +#define VertexAttribs4dvNV_remap_index 305 +#define VertexAttribs4fvNV_remap_index 306 +#define VertexAttribs4svNV_remap_index 307 +#define VertexAttribs4ubvNV_remap_index 308 +#define AlphaFragmentOp1ATI_remap_index 309 +#define AlphaFragmentOp2ATI_remap_index 310 +#define AlphaFragmentOp3ATI_remap_index 311 +#define BeginFragmentShaderATI_remap_index 312 +#define BindFragmentShaderATI_remap_index 313 +#define ColorFragmentOp1ATI_remap_index 314 +#define ColorFragmentOp2ATI_remap_index 315 +#define ColorFragmentOp3ATI_remap_index 316 +#define DeleteFragmentShaderATI_remap_index 317 +#define EndFragmentShaderATI_remap_index 318 +#define GenFragmentShadersATI_remap_index 319 +#define PassTexCoordATI_remap_index 320 +#define SampleMapATI_remap_index 321 +#define SetFragmentShaderConstantATI_remap_index 322 +#define PointParameteriNV_remap_index 323 +#define PointParameterivNV_remap_index 324 +#define ActiveStencilFaceEXT_remap_index 325 +#define BindVertexArrayAPPLE_remap_index 326 +#define DeleteVertexArraysAPPLE_remap_index 327 +#define GenVertexArraysAPPLE_remap_index 328 +#define IsVertexArrayAPPLE_remap_index 329 +#define GetProgramNamedParameterdvNV_remap_index 330 +#define GetProgramNamedParameterfvNV_remap_index 331 +#define ProgramNamedParameter4dNV_remap_index 332 +#define ProgramNamedParameter4dvNV_remap_index 333 +#define ProgramNamedParameter4fNV_remap_index 334 +#define ProgramNamedParameter4fvNV_remap_index 335 +#define DepthBoundsEXT_remap_index 336 +#define BlendEquationSeparateEXT_remap_index 337 +#define BindFramebufferEXT_remap_index 338 +#define BindRenderbufferEXT_remap_index 339 +#define CheckFramebufferStatusEXT_remap_index 340 +#define DeleteFramebuffersEXT_remap_index 341 +#define DeleteRenderbuffersEXT_remap_index 342 +#define FramebufferRenderbufferEXT_remap_index 343 +#define FramebufferTexture1DEXT_remap_index 344 +#define FramebufferTexture2DEXT_remap_index 345 +#define FramebufferTexture3DEXT_remap_index 346 +#define GenFramebuffersEXT_remap_index 347 +#define GenRenderbuffersEXT_remap_index 348 +#define GenerateMipmapEXT_remap_index 349 +#define GetFramebufferAttachmentParameterivEXT_remap_index 350 +#define GetRenderbufferParameterivEXT_remap_index 351 +#define IsFramebufferEXT_remap_index 352 +#define IsRenderbufferEXT_remap_index 353 +#define RenderbufferStorageEXT_remap_index 354 +#define BlitFramebufferEXT_remap_index 355 +#define FramebufferTextureLayerEXT_remap_index 356 +#define StencilFuncSeparateATI_remap_index 357 +#define ProgramEnvParameters4fvEXT_remap_index 358 +#define ProgramLocalParameters4fvEXT_remap_index 359 +#define GetQueryObjecti64vEXT_remap_index 360 +#define GetQueryObjectui64vEXT_remap_index 361 #define CALL_AttachShader(disp, parameters) CALL_by_offset(disp, (void (GLAPIENTRYP)(GLuint, GLuint)), driDispatchRemapTable[AttachShader_remap_index], parameters) #define GET_AttachShader(disp) GET_by_offset(disp, driDispatchRemapTable[AttachShader_remap_index]) @@ -3548,18 +3532,6 @@ extern int driDispatchRemapTable[ driDispatchRemapTable_size ]; #define CALL_LoadProgramNV(disp, parameters) CALL_by_offset(disp, (void (GLAPIENTRYP)(GLenum, GLuint, GLsizei, const GLubyte *)), driDispatchRemapTable[LoadProgramNV_remap_index], parameters) #define GET_LoadProgramNV(disp) GET_by_offset(disp, driDispatchRemapTable[LoadProgramNV_remap_index]) #define SET_LoadProgramNV(disp, fn) SET_by_offset(disp, driDispatchRemapTable[LoadProgramNV_remap_index], fn) -#define CALL_ProgramParameter4dNV(disp, parameters) CALL_by_offset(disp, (void (GLAPIENTRYP)(GLenum, GLuint, GLdouble, GLdouble, GLdouble, GLdouble)), driDispatchRemapTable[ProgramParameter4dNV_remap_index], parameters) -#define GET_ProgramParameter4dNV(disp) GET_by_offset(disp, driDispatchRemapTable[ProgramParameter4dNV_remap_index]) -#define SET_ProgramParameter4dNV(disp, fn) SET_by_offset(disp, driDispatchRemapTable[ProgramParameter4dNV_remap_index], fn) -#define CALL_ProgramParameter4dvNV(disp, parameters) CALL_by_offset(disp, (void (GLAPIENTRYP)(GLenum, GLuint, const GLdouble *)), driDispatchRemapTable[ProgramParameter4dvNV_remap_index], parameters) -#define GET_ProgramParameter4dvNV(disp) GET_by_offset(disp, driDispatchRemapTable[ProgramParameter4dvNV_remap_index]) -#define SET_ProgramParameter4dvNV(disp, fn) SET_by_offset(disp, driDispatchRemapTable[ProgramParameter4dvNV_remap_index], fn) -#define CALL_ProgramParameter4fNV(disp, parameters) CALL_by_offset(disp, (void (GLAPIENTRYP)(GLenum, GLuint, GLfloat, GLfloat, GLfloat, GLfloat)), driDispatchRemapTable[ProgramParameter4fNV_remap_index], parameters) -#define GET_ProgramParameter4fNV(disp) GET_by_offset(disp, driDispatchRemapTable[ProgramParameter4fNV_remap_index]) -#define SET_ProgramParameter4fNV(disp, fn) SET_by_offset(disp, driDispatchRemapTable[ProgramParameter4fNV_remap_index], fn) -#define CALL_ProgramParameter4fvNV(disp, parameters) CALL_by_offset(disp, (void (GLAPIENTRYP)(GLenum, GLuint, const GLfloat *)), driDispatchRemapTable[ProgramParameter4fvNV_remap_index], parameters) -#define GET_ProgramParameter4fvNV(disp) GET_by_offset(disp, driDispatchRemapTable[ProgramParameter4fvNV_remap_index]) -#define SET_ProgramParameter4fvNV(disp, fn) SET_by_offset(disp, driDispatchRemapTable[ProgramParameter4fvNV_remap_index], fn) #define CALL_ProgramParameters4dvNV(disp, parameters) CALL_by_offset(disp, (void (GLAPIENTRYP)(GLenum, GLuint, GLuint, const GLdouble *)), driDispatchRemapTable[ProgramParameters4dvNV_remap_index], parameters) #define GET_ProgramParameters4dvNV(disp) GET_by_offset(disp, driDispatchRemapTable[ProgramParameters4dvNV_remap_index]) #define SET_ProgramParameters4dvNV(disp, fn) SET_by_offset(disp, driDispatchRemapTable[ProgramParameters4dvNV_remap_index], fn) diff --git a/src/mesa/glapi/gl_API.xml b/src/mesa/glapi/gl_API.xml index 3d47e6f2ce..ef4a309cd6 100644 --- a/src/mesa/glapi/gl_API.xml +++ b/src/mesa/glapi/gl_API.xml @@ -11275,7 +11275,7 @@ <glx rop="4183"/> </function> - <function name="ProgramParameter4dNV" offset="assign" vectorequiv="ProgramParameter4dvNV"> + <function name="ProgramParameter4dNV" alias="ProgramEnvParameter4dARB"> <param name="target" type="GLenum"/> <param name="index" type="GLuint"/> <param name="x" type="GLdouble"/> @@ -11284,14 +11284,13 @@ <param name="w" type="GLdouble"/> </function> - <function name="ProgramParameter4dvNV" offset="assign"> + <function name="ProgramParameter4dvNV" alias="ProgramEnvParameter4dvARB"> <param name="target" type="GLenum"/> <param name="index" type="GLuint"/> - <param name="params" type="const GLdouble *" count="4"/> - <glx rop="4185" doubles_in_order="true"/> + <param name="params" type="const GLdouble *"/> </function> - <function name="ProgramParameter4fNV" offset="assign" vectorequiv="ProgramParameter4fvNV"> + <function name="ProgramParameter4fNV" alias="ProgramEnvParameter4fARB"> <param name="target" type="GLenum"/> <param name="index" type="GLuint"/> <param name="x" type="GLfloat"/> @@ -11300,11 +11299,10 @@ <param name="w" type="GLfloat"/> </function> - <function name="ProgramParameter4fvNV" offset="assign"> + <function name="ProgramParameter4fvNV" alias="ProgramEnvParameter4fvARB"> <param name="target" type="GLenum"/> <param name="index" type="GLuint"/> - <param name="params" type="const GLfloat *" count="4"/> - <glx rop="4184"/> + <param name="params" type="const GLfloat *"/> </function> <function name="ProgramParameters4dvNV" offset="assign"> diff --git a/src/mesa/glapi/gl_x86_asm.py b/src/mesa/glapi/gl_x86_asm.py index 403e87261b..651cb03f14 100644 --- a/src/mesa/glapi/gl_x86_asm.py +++ b/src/mesa/glapi/gl_x86_asm.py @@ -51,7 +51,7 @@ class PrintGenericStubs(gl_XML.gl_print_base): def printRealHeader(self): print '#include "assyntax.h"' - print '#include "glapioffsets.h"' + print '#include "glapi/glapioffsets.h"' print '' print '#if defined(STDCALL_API)' print '# if defined(USE_MGL_NAMESPACE)' diff --git a/src/mesa/glapi/glapioffsets.h b/src/mesa/glapi/glapioffsets.h index 4799fd3076..43c265924b 100644 --- a/src/mesa/glapi/glapioffsets.h +++ b/src/mesa/glapi/glapioffsets.h @@ -705,108 +705,104 @@ #define _gloffset_GetVertexAttribivNV 670 #define _gloffset_IsProgramNV 671 #define _gloffset_LoadProgramNV 672 -#define _gloffset_ProgramParameter4dNV 673 -#define _gloffset_ProgramParameter4dvNV 674 -#define _gloffset_ProgramParameter4fNV 675 -#define _gloffset_ProgramParameter4fvNV 676 -#define _gloffset_ProgramParameters4dvNV 677 -#define _gloffset_ProgramParameters4fvNV 678 -#define _gloffset_RequestResidentProgramsNV 679 -#define _gloffset_TrackMatrixNV 680 -#define _gloffset_VertexAttrib1dNV 681 -#define _gloffset_VertexAttrib1dvNV 682 -#define _gloffset_VertexAttrib1fNV 683 -#define _gloffset_VertexAttrib1fvNV 684 -#define _gloffset_VertexAttrib1sNV 685 -#define _gloffset_VertexAttrib1svNV 686 -#define _gloffset_VertexAttrib2dNV 687 -#define _gloffset_VertexAttrib2dvNV 688 -#define _gloffset_VertexAttrib2fNV 689 -#define _gloffset_VertexAttrib2fvNV 690 -#define _gloffset_VertexAttrib2sNV 691 -#define _gloffset_VertexAttrib2svNV 692 -#define _gloffset_VertexAttrib3dNV 693 -#define _gloffset_VertexAttrib3dvNV 694 -#define _gloffset_VertexAttrib3fNV 695 -#define _gloffset_VertexAttrib3fvNV 696 -#define _gloffset_VertexAttrib3sNV 697 -#define _gloffset_VertexAttrib3svNV 698 -#define _gloffset_VertexAttrib4dNV 699 -#define _gloffset_VertexAttrib4dvNV 700 -#define _gloffset_VertexAttrib4fNV 701 -#define _gloffset_VertexAttrib4fvNV 702 -#define _gloffset_VertexAttrib4sNV 703 -#define _gloffset_VertexAttrib4svNV 704 -#define _gloffset_VertexAttrib4ubNV 705 -#define _gloffset_VertexAttrib4ubvNV 706 -#define _gloffset_VertexAttribPointerNV 707 -#define _gloffset_VertexAttribs1dvNV 708 -#define _gloffset_VertexAttribs1fvNV 709 -#define _gloffset_VertexAttribs1svNV 710 -#define _gloffset_VertexAttribs2dvNV 711 -#define _gloffset_VertexAttribs2fvNV 712 -#define _gloffset_VertexAttribs2svNV 713 -#define _gloffset_VertexAttribs3dvNV 714 -#define _gloffset_VertexAttribs3fvNV 715 -#define _gloffset_VertexAttribs3svNV 716 -#define _gloffset_VertexAttribs4dvNV 717 -#define _gloffset_VertexAttribs4fvNV 718 -#define _gloffset_VertexAttribs4svNV 719 -#define _gloffset_VertexAttribs4ubvNV 720 -#define _gloffset_AlphaFragmentOp1ATI 721 -#define _gloffset_AlphaFragmentOp2ATI 722 -#define _gloffset_AlphaFragmentOp3ATI 723 -#define _gloffset_BeginFragmentShaderATI 724 -#define _gloffset_BindFragmentShaderATI 725 -#define _gloffset_ColorFragmentOp1ATI 726 -#define _gloffset_ColorFragmentOp2ATI 727 -#define _gloffset_ColorFragmentOp3ATI 728 -#define _gloffset_DeleteFragmentShaderATI 729 -#define _gloffset_EndFragmentShaderATI 730 -#define _gloffset_GenFragmentShadersATI 731 -#define _gloffset_PassTexCoordATI 732 -#define _gloffset_SampleMapATI 733 -#define _gloffset_SetFragmentShaderConstantATI 734 -#define _gloffset_PointParameteriNV 735 -#define _gloffset_PointParameterivNV 736 -#define _gloffset_ActiveStencilFaceEXT 737 -#define _gloffset_BindVertexArrayAPPLE 738 -#define _gloffset_DeleteVertexArraysAPPLE 739 -#define _gloffset_GenVertexArraysAPPLE 740 -#define _gloffset_IsVertexArrayAPPLE 741 -#define _gloffset_GetProgramNamedParameterdvNV 742 -#define _gloffset_GetProgramNamedParameterfvNV 743 -#define _gloffset_ProgramNamedParameter4dNV 744 -#define _gloffset_ProgramNamedParameter4dvNV 745 -#define _gloffset_ProgramNamedParameter4fNV 746 -#define _gloffset_ProgramNamedParameter4fvNV 747 -#define _gloffset_DepthBoundsEXT 748 -#define _gloffset_BlendEquationSeparateEXT 749 -#define _gloffset_BindFramebufferEXT 750 -#define _gloffset_BindRenderbufferEXT 751 -#define _gloffset_CheckFramebufferStatusEXT 752 -#define _gloffset_DeleteFramebuffersEXT 753 -#define _gloffset_DeleteRenderbuffersEXT 754 -#define _gloffset_FramebufferRenderbufferEXT 755 -#define _gloffset_FramebufferTexture1DEXT 756 -#define _gloffset_FramebufferTexture2DEXT 757 -#define _gloffset_FramebufferTexture3DEXT 758 -#define _gloffset_GenFramebuffersEXT 759 -#define _gloffset_GenRenderbuffersEXT 760 -#define _gloffset_GenerateMipmapEXT 761 -#define _gloffset_GetFramebufferAttachmentParameterivEXT 762 -#define _gloffset_GetRenderbufferParameterivEXT 763 -#define _gloffset_IsFramebufferEXT 764 -#define _gloffset_IsRenderbufferEXT 765 -#define _gloffset_RenderbufferStorageEXT 766 -#define _gloffset_BlitFramebufferEXT 767 -#define _gloffset_FramebufferTextureLayerEXT 768 -#define _gloffset_StencilFuncSeparateATI 769 -#define _gloffset_ProgramEnvParameters4fvEXT 770 -#define _gloffset_ProgramLocalParameters4fvEXT 771 -#define _gloffset_GetQueryObjecti64vEXT 772 -#define _gloffset_GetQueryObjectui64vEXT 773 -#define _gloffset_FIRST_DYNAMIC 774 +#define _gloffset_ProgramParameters4dvNV 673 +#define _gloffset_ProgramParameters4fvNV 674 +#define _gloffset_RequestResidentProgramsNV 675 +#define _gloffset_TrackMatrixNV 676 +#define _gloffset_VertexAttrib1dNV 677 +#define _gloffset_VertexAttrib1dvNV 678 +#define _gloffset_VertexAttrib1fNV 679 +#define _gloffset_VertexAttrib1fvNV 680 +#define _gloffset_VertexAttrib1sNV 681 +#define _gloffset_VertexAttrib1svNV 682 +#define _gloffset_VertexAttrib2dNV 683 +#define _gloffset_VertexAttrib2dvNV 684 +#define _gloffset_VertexAttrib2fNV 685 +#define _gloffset_VertexAttrib2fvNV 686 +#define _gloffset_VertexAttrib2sNV 687 +#define _gloffset_VertexAttrib2svNV 688 +#define _gloffset_VertexAttrib3dNV 689 +#define _gloffset_VertexAttrib3dvNV 690 +#define _gloffset_VertexAttrib3fNV 691 +#define _gloffset_VertexAttrib3fvNV 692 +#define _gloffset_VertexAttrib3sNV 693 +#define _gloffset_VertexAttrib3svNV 694 +#define _gloffset_VertexAttrib4dNV 695 +#define _gloffset_VertexAttrib4dvNV 696 +#define _gloffset_VertexAttrib4fNV 697 +#define _gloffset_VertexAttrib4fvNV 698 +#define _gloffset_VertexAttrib4sNV 699 +#define _gloffset_VertexAttrib4svNV 700 +#define _gloffset_VertexAttrib4ubNV 701 +#define _gloffset_VertexAttrib4ubvNV 702 +#define _gloffset_VertexAttribPointerNV 703 +#define _gloffset_VertexAttribs1dvNV 704 +#define _gloffset_VertexAttribs1fvNV 705 +#define _gloffset_VertexAttribs1svNV 706 +#define _gloffset_VertexAttribs2dvNV 707 +#define _gloffset_VertexAttribs2fvNV 708 +#define _gloffset_VertexAttribs2svNV 709 +#define _gloffset_VertexAttribs3dvNV 710 +#define _gloffset_VertexAttribs3fvNV 711 +#define _gloffset_VertexAttribs3svNV 712 +#define _gloffset_VertexAttribs4dvNV 713 +#define _gloffset_VertexAttribs4fvNV 714 +#define _gloffset_VertexAttribs4svNV 715 +#define _gloffset_VertexAttribs4ubvNV 716 +#define _gloffset_AlphaFragmentOp1ATI 717 +#define _gloffset_AlphaFragmentOp2ATI 718 +#define _gloffset_AlphaFragmentOp3ATI 719 +#define _gloffset_BeginFragmentShaderATI 720 +#define _gloffset_BindFragmentShaderATI 721 +#define _gloffset_ColorFragmentOp1ATI 722 +#define _gloffset_ColorFragmentOp2ATI 723 +#define _gloffset_ColorFragmentOp3ATI 724 +#define _gloffset_DeleteFragmentShaderATI 725 +#define _gloffset_EndFragmentShaderATI 726 +#define _gloffset_GenFragmentShadersATI 727 +#define _gloffset_PassTexCoordATI 728 +#define _gloffset_SampleMapATI 729 +#define _gloffset_SetFragmentShaderConstantATI 730 +#define _gloffset_PointParameteriNV 731 +#define _gloffset_PointParameterivNV 732 +#define _gloffset_ActiveStencilFaceEXT 733 +#define _gloffset_BindVertexArrayAPPLE 734 +#define _gloffset_DeleteVertexArraysAPPLE 735 +#define _gloffset_GenVertexArraysAPPLE 736 +#define _gloffset_IsVertexArrayAPPLE 737 +#define _gloffset_GetProgramNamedParameterdvNV 738 +#define _gloffset_GetProgramNamedParameterfvNV 739 +#define _gloffset_ProgramNamedParameter4dNV 740 +#define _gloffset_ProgramNamedParameter4dvNV 741 +#define _gloffset_ProgramNamedParameter4fNV 742 +#define _gloffset_ProgramNamedParameter4fvNV 743 +#define _gloffset_DepthBoundsEXT 744 +#define _gloffset_BlendEquationSeparateEXT 745 +#define _gloffset_BindFramebufferEXT 746 +#define _gloffset_BindRenderbufferEXT 747 +#define _gloffset_CheckFramebufferStatusEXT 748 +#define _gloffset_DeleteFramebuffersEXT 749 +#define _gloffset_DeleteRenderbuffersEXT 750 +#define _gloffset_FramebufferRenderbufferEXT 751 +#define _gloffset_FramebufferTexture1DEXT 752 +#define _gloffset_FramebufferTexture2DEXT 753 +#define _gloffset_FramebufferTexture3DEXT 754 +#define _gloffset_GenFramebuffersEXT 755 +#define _gloffset_GenRenderbuffersEXT 756 +#define _gloffset_GenerateMipmapEXT 757 +#define _gloffset_GetFramebufferAttachmentParameterivEXT 758 +#define _gloffset_GetRenderbufferParameterivEXT 759 +#define _gloffset_IsFramebufferEXT 760 +#define _gloffset_IsRenderbufferEXT 761 +#define _gloffset_RenderbufferStorageEXT 762 +#define _gloffset_BlitFramebufferEXT 763 +#define _gloffset_FramebufferTextureLayerEXT 764 +#define _gloffset_StencilFuncSeparateATI 765 +#define _gloffset_ProgramEnvParameters4fvEXT 766 +#define _gloffset_ProgramLocalParameters4fvEXT 767 +#define _gloffset_GetQueryObjecti64vEXT 768 +#define _gloffset_GetQueryObjectui64vEXT 769 +#define _gloffset_FIRST_DYNAMIC 770 #else @@ -1075,10 +1071,6 @@ #define _gloffset_GetVertexAttribivNV driDispatchRemapTable[GetVertexAttribivNV_remap_index] #define _gloffset_IsProgramNV driDispatchRemapTable[IsProgramNV_remap_index] #define _gloffset_LoadProgramNV driDispatchRemapTable[LoadProgramNV_remap_index] -#define _gloffset_ProgramParameter4dNV driDispatchRemapTable[ProgramParameter4dNV_remap_index] -#define _gloffset_ProgramParameter4dvNV driDispatchRemapTable[ProgramParameter4dvNV_remap_index] -#define _gloffset_ProgramParameter4fNV driDispatchRemapTable[ProgramParameter4fNV_remap_index] -#define _gloffset_ProgramParameter4fvNV driDispatchRemapTable[ProgramParameter4fvNV_remap_index] #define _gloffset_ProgramParameters4dvNV driDispatchRemapTable[ProgramParameters4dvNV_remap_index] #define _gloffset_ProgramParameters4fvNV driDispatchRemapTable[ProgramParameters4fvNV_remap_index] #define _gloffset_RequestResidentProgramsNV driDispatchRemapTable[RequestResidentProgramsNV_remap_index] diff --git a/src/mesa/glapi/glapitable.h b/src/mesa/glapi/glapitable.h index 22c2dc2f69..48941f5590 100644 --- a/src/mesa/glapi/glapitable.h +++ b/src/mesa/glapi/glapitable.h @@ -714,107 +714,103 @@ struct _glapi_table void (GLAPIENTRYP GetVertexAttribivNV)(GLuint index, GLenum pname, GLint * params); /* 670 */ GLboolean (GLAPIENTRYP IsProgramNV)(GLuint program); /* 671 */ void (GLAPIENTRYP LoadProgramNV)(GLenum target, GLuint id, GLsizei len, const GLubyte * program); /* 672 */ - void (GLAPIENTRYP ProgramParameter4dNV)(GLenum target, GLuint index, GLdouble x, GLdouble y, GLdouble z, GLdouble w); /* 673 */ - void (GLAPIENTRYP ProgramParameter4dvNV)(GLenum target, GLuint index, const GLdouble * params); /* 674 */ - void (GLAPIENTRYP ProgramParameter4fNV)(GLenum target, GLuint index, GLfloat x, GLfloat y, GLfloat z, GLfloat w); /* 675 */ - void (GLAPIENTRYP ProgramParameter4fvNV)(GLenum target, GLuint index, const GLfloat * params); /* 676 */ - void (GLAPIENTRYP ProgramParameters4dvNV)(GLenum target, GLuint index, GLuint num, const GLdouble * params); /* 677 */ - void (GLAPIENTRYP ProgramParameters4fvNV)(GLenum target, GLuint index, GLuint num, const GLfloat * params); /* 678 */ - void (GLAPIENTRYP RequestResidentProgramsNV)(GLsizei n, const GLuint * ids); /* 679 */ - void (GLAPIENTRYP TrackMatrixNV)(GLenum target, GLuint address, GLenum matrix, GLenum transform); /* 680 */ - void (GLAPIENTRYP VertexAttrib1dNV)(GLuint index, GLdouble x); /* 681 */ - void (GLAPIENTRYP VertexAttrib1dvNV)(GLuint index, const GLdouble * v); /* 682 */ - void (GLAPIENTRYP VertexAttrib1fNV)(GLuint index, GLfloat x); /* 683 */ - void (GLAPIENTRYP VertexAttrib1fvNV)(GLuint index, const GLfloat * v); /* 684 */ - void (GLAPIENTRYP VertexAttrib1sNV)(GLuint index, GLshort x); /* 685 */ - void (GLAPIENTRYP VertexAttrib1svNV)(GLuint index, const GLshort * v); /* 686 */ - void (GLAPIENTRYP VertexAttrib2dNV)(GLuint index, GLdouble x, GLdouble y); /* 687 */ - void (GLAPIENTRYP VertexAttrib2dvNV)(GLuint index, const GLdouble * v); /* 688 */ - void (GLAPIENTRYP VertexAttrib2fNV)(GLuint index, GLfloat x, GLfloat y); /* 689 */ - void (GLAPIENTRYP VertexAttrib2fvNV)(GLuint index, const GLfloat * v); /* 690 */ - void (GLAPIENTRYP VertexAttrib2sNV)(GLuint index, GLshort x, GLshort y); /* 691 */ - void (GLAPIENTRYP VertexAttrib2svNV)(GLuint index, const GLshort * v); /* 692 */ - void (GLAPIENTRYP VertexAttrib3dNV)(GLuint index, GLdouble x, GLdouble y, GLdouble z); /* 693 */ - void (GLAPIENTRYP VertexAttrib3dvNV)(GLuint index, const GLdouble * v); /* 694 */ - void (GLAPIENTRYP VertexAttrib3fNV)(GLuint index, GLfloat x, GLfloat y, GLfloat z); /* 695 */ - void (GLAPIENTRYP VertexAttrib3fvNV)(GLuint index, const GLfloat * v); /* 696 */ - void (GLAPIENTRYP VertexAttrib3sNV)(GLuint index, GLshort x, GLshort y, GLshort z); /* 697 */ - void (GLAPIENTRYP VertexAttrib3svNV)(GLuint index, const GLshort * v); /* 698 */ - void (GLAPIENTRYP VertexAttrib4dNV)(GLuint index, GLdouble x, GLdouble y, GLdouble z, GLdouble w); /* 699 */ - void (GLAPIENTRYP VertexAttrib4dvNV)(GLuint index, const GLdouble * v); /* 700 */ - void (GLAPIENTRYP VertexAttrib4fNV)(GLuint index, GLfloat x, GLfloat y, GLfloat z, GLfloat w); /* 701 */ - void (GLAPIENTRYP VertexAttrib4fvNV)(GLuint index, const GLfloat * v); /* 702 */ - void (GLAPIENTRYP VertexAttrib4sNV)(GLuint index, GLshort x, GLshort y, GLshort z, GLshort w); /* 703 */ - void (GLAPIENTRYP VertexAttrib4svNV)(GLuint index, const GLshort * v); /* 704 */ - void (GLAPIENTRYP VertexAttrib4ubNV)(GLuint index, GLubyte x, GLubyte y, GLubyte z, GLubyte w); /* 705 */ - void (GLAPIENTRYP VertexAttrib4ubvNV)(GLuint index, const GLubyte * v); /* 706 */ - void (GLAPIENTRYP VertexAttribPointerNV)(GLuint index, GLint size, GLenum type, GLsizei stride, const GLvoid * pointer); /* 707 */ - void (GLAPIENTRYP VertexAttribs1dvNV)(GLuint index, GLsizei n, const GLdouble * v); /* 708 */ - void (GLAPIENTRYP VertexAttribs1fvNV)(GLuint index, GLsizei n, const GLfloat * v); /* 709 */ - void (GLAPIENTRYP VertexAttribs1svNV)(GLuint index, GLsizei n, const GLshort * v); /* 710 */ - void (GLAPIENTRYP VertexAttribs2dvNV)(GLuint index, GLsizei n, const GLdouble * v); /* 711 */ - void (GLAPIENTRYP VertexAttribs2fvNV)(GLuint index, GLsizei n, const GLfloat * v); /* 712 */ - void (GLAPIENTRYP VertexAttribs2svNV)(GLuint index, GLsizei n, const GLshort * v); /* 713 */ - void (GLAPIENTRYP VertexAttribs3dvNV)(GLuint index, GLsizei n, const GLdouble * v); /* 714 */ - void (GLAPIENTRYP VertexAttribs3fvNV)(GLuint index, GLsizei n, const GLfloat * v); /* 715 */ - void (GLAPIENTRYP VertexAttribs3svNV)(GLuint index, GLsizei n, const GLshort * v); /* 716 */ - void (GLAPIENTRYP VertexAttribs4dvNV)(GLuint index, GLsizei n, const GLdouble * v); /* 717 */ - void (GLAPIENTRYP VertexAttribs4fvNV)(GLuint index, GLsizei n, const GLfloat * v); /* 718 */ - void (GLAPIENTRYP VertexAttribs4svNV)(GLuint index, GLsizei n, const GLshort * v); /* 719 */ - void (GLAPIENTRYP VertexAttribs4ubvNV)(GLuint index, GLsizei n, const GLubyte * v); /* 720 */ - void (GLAPIENTRYP AlphaFragmentOp1ATI)(GLenum op, GLuint dst, GLuint dstMod, GLuint arg1, GLuint arg1Rep, GLuint arg1Mod); /* 721 */ - void (GLAPIENTRYP AlphaFragmentOp2ATI)(GLenum op, GLuint dst, GLuint dstMod, GLuint arg1, GLuint arg1Rep, GLuint arg1Mod, GLuint arg2, GLuint arg2Rep, GLuint arg2Mod); /* 722 */ - void (GLAPIENTRYP AlphaFragmentOp3ATI)(GLenum op, GLuint dst, GLuint dstMod, GLuint arg1, GLuint arg1Rep, GLuint arg1Mod, GLuint arg2, GLuint arg2Rep, GLuint arg2Mod, GLuint arg3, GLuint arg3Rep, GLuint arg3Mod); /* 723 */ - void (GLAPIENTRYP BeginFragmentShaderATI)(void); /* 724 */ - void (GLAPIENTRYP BindFragmentShaderATI)(GLuint id); /* 725 */ - void (GLAPIENTRYP ColorFragmentOp1ATI)(GLenum op, GLuint dst, GLuint dstMask, GLuint dstMod, GLuint arg1, GLuint arg1Rep, GLuint arg1Mod); /* 726 */ - void (GLAPIENTRYP ColorFragmentOp2ATI)(GLenum op, GLuint dst, GLuint dstMask, GLuint dstMod, GLuint arg1, GLuint arg1Rep, GLuint arg1Mod, GLuint arg2, GLuint arg2Rep, GLuint arg2Mod); /* 727 */ - void (GLAPIENTRYP ColorFragmentOp3ATI)(GLenum op, GLuint dst, GLuint dstMask, GLuint dstMod, GLuint arg1, GLuint arg1Rep, GLuint arg1Mod, GLuint arg2, GLuint arg2Rep, GLuint arg2Mod, GLuint arg3, GLuint arg3Rep, GLuint arg3Mod); /* 728 */ - void (GLAPIENTRYP DeleteFragmentShaderATI)(GLuint id); /* 729 */ - void (GLAPIENTRYP EndFragmentShaderATI)(void); /* 730 */ - GLuint (GLAPIENTRYP GenFragmentShadersATI)(GLuint range); /* 731 */ - void (GLAPIENTRYP PassTexCoordATI)(GLuint dst, GLuint coord, GLenum swizzle); /* 732 */ - void (GLAPIENTRYP SampleMapATI)(GLuint dst, GLuint interp, GLenum swizzle); /* 733 */ - void (GLAPIENTRYP SetFragmentShaderConstantATI)(GLuint dst, const GLfloat * value); /* 734 */ - void (GLAPIENTRYP PointParameteriNV)(GLenum pname, GLint param); /* 735 */ - void (GLAPIENTRYP PointParameterivNV)(GLenum pname, const GLint * params); /* 736 */ - void (GLAPIENTRYP ActiveStencilFaceEXT)(GLenum face); /* 737 */ - void (GLAPIENTRYP BindVertexArrayAPPLE)(GLuint array); /* 738 */ - void (GLAPIENTRYP DeleteVertexArraysAPPLE)(GLsizei n, const GLuint * arrays); /* 739 */ - void (GLAPIENTRYP GenVertexArraysAPPLE)(GLsizei n, GLuint * arrays); /* 740 */ - GLboolean (GLAPIENTRYP IsVertexArrayAPPLE)(GLuint array); /* 741 */ - void (GLAPIENTRYP GetProgramNamedParameterdvNV)(GLuint id, GLsizei len, const GLubyte * name, GLdouble * params); /* 742 */ - void (GLAPIENTRYP GetProgramNamedParameterfvNV)(GLuint id, GLsizei len, const GLubyte * name, GLfloat * params); /* 743 */ - void (GLAPIENTRYP ProgramNamedParameter4dNV)(GLuint id, GLsizei len, const GLubyte * name, GLdouble x, GLdouble y, GLdouble z, GLdouble w); /* 744 */ - void (GLAPIENTRYP ProgramNamedParameter4dvNV)(GLuint id, GLsizei len, const GLubyte * name, const GLdouble * v); /* 745 */ - void (GLAPIENTRYP ProgramNamedParameter4fNV)(GLuint id, GLsizei len, const GLubyte * name, GLfloat x, GLfloat y, GLfloat z, GLfloat w); /* 746 */ - void (GLAPIENTRYP ProgramNamedParameter4fvNV)(GLuint id, GLsizei len, const GLubyte * name, const GLfloat * v); /* 747 */ - void (GLAPIENTRYP DepthBoundsEXT)(GLclampd zmin, GLclampd zmax); /* 748 */ - void (GLAPIENTRYP BlendEquationSeparateEXT)(GLenum modeRGB, GLenum modeA); /* 749 */ - void (GLAPIENTRYP BindFramebufferEXT)(GLenum target, GLuint framebuffer); /* 750 */ - void (GLAPIENTRYP BindRenderbufferEXT)(GLenum target, GLuint renderbuffer); /* 751 */ - GLenum (GLAPIENTRYP CheckFramebufferStatusEXT)(GLenum target); /* 752 */ - void (GLAPIENTRYP DeleteFramebuffersEXT)(GLsizei n, const GLuint * framebuffers); /* 753 */ - void (GLAPIENTRYP DeleteRenderbuffersEXT)(GLsizei n, const GLuint * renderbuffers); /* 754 */ - void (GLAPIENTRYP FramebufferRenderbufferEXT)(GLenum target, GLenum attachment, GLenum renderbuffertarget, GLuint renderbuffer); /* 755 */ - void (GLAPIENTRYP FramebufferTexture1DEXT)(GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level); /* 756 */ - void (GLAPIENTRYP FramebufferTexture2DEXT)(GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level); /* 757 */ - void (GLAPIENTRYP FramebufferTexture3DEXT)(GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level, GLint zoffset); /* 758 */ - void (GLAPIENTRYP GenFramebuffersEXT)(GLsizei n, GLuint * framebuffers); /* 759 */ - void (GLAPIENTRYP GenRenderbuffersEXT)(GLsizei n, GLuint * renderbuffers); /* 760 */ - void (GLAPIENTRYP GenerateMipmapEXT)(GLenum target); /* 761 */ - void (GLAPIENTRYP GetFramebufferAttachmentParameterivEXT)(GLenum target, GLenum attachment, GLenum pname, GLint * params); /* 762 */ - void (GLAPIENTRYP GetRenderbufferParameterivEXT)(GLenum target, GLenum pname, GLint * params); /* 763 */ - GLboolean (GLAPIENTRYP IsFramebufferEXT)(GLuint framebuffer); /* 764 */ - GLboolean (GLAPIENTRYP IsRenderbufferEXT)(GLuint renderbuffer); /* 765 */ - void (GLAPIENTRYP RenderbufferStorageEXT)(GLenum target, GLenum internalformat, GLsizei width, GLsizei height); /* 766 */ - void (GLAPIENTRYP BlitFramebufferEXT)(GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter); /* 767 */ - void (GLAPIENTRYP FramebufferTextureLayerEXT)(GLenum target, GLenum attachment, GLuint texture, GLint level, GLint layer); /* 768 */ - void (GLAPIENTRYP StencilFuncSeparateATI)(GLenum frontfunc, GLenum backfunc, GLint ref, GLuint mask); /* 769 */ - void (GLAPIENTRYP ProgramEnvParameters4fvEXT)(GLenum target, GLuint index, GLsizei count, const GLfloat * params); /* 770 */ - void (GLAPIENTRYP ProgramLocalParameters4fvEXT)(GLenum target, GLuint index, GLsizei count, const GLfloat * params); /* 771 */ - void (GLAPIENTRYP GetQueryObjecti64vEXT)(GLuint id, GLenum pname, GLint64EXT * params); /* 772 */ - void (GLAPIENTRYP GetQueryObjectui64vEXT)(GLuint id, GLenum pname, GLuint64EXT * params); /* 773 */ + void (GLAPIENTRYP ProgramParameters4dvNV)(GLenum target, GLuint index, GLuint num, const GLdouble * params); /* 673 */ + void (GLAPIENTRYP ProgramParameters4fvNV)(GLenum target, GLuint index, GLuint num, const GLfloat * params); /* 674 */ + void (GLAPIENTRYP RequestResidentProgramsNV)(GLsizei n, const GLuint * ids); /* 675 */ + void (GLAPIENTRYP TrackMatrixNV)(GLenum target, GLuint address, GLenum matrix, GLenum transform); /* 676 */ + void (GLAPIENTRYP VertexAttrib1dNV)(GLuint index, GLdouble x); /* 677 */ + void (GLAPIENTRYP VertexAttrib1dvNV)(GLuint index, const GLdouble * v); /* 678 */ + void (GLAPIENTRYP VertexAttrib1fNV)(GLuint index, GLfloat x); /* 679 */ + void (GLAPIENTRYP VertexAttrib1fvNV)(GLuint index, const GLfloat * v); /* 680 */ + void (GLAPIENTRYP VertexAttrib1sNV)(GLuint index, GLshort x); /* 681 */ + void (GLAPIENTRYP VertexAttrib1svNV)(GLuint index, const GLshort * v); /* 682 */ + void (GLAPIENTRYP VertexAttrib2dNV)(GLuint index, GLdouble x, GLdouble y); /* 683 */ + void (GLAPIENTRYP VertexAttrib2dvNV)(GLuint index, const GLdouble * v); /* 684 */ + void (GLAPIENTRYP VertexAttrib2fNV)(GLuint index, GLfloat x, GLfloat y); /* 685 */ + void (GLAPIENTRYP VertexAttrib2fvNV)(GLuint index, const GLfloat * v); /* 686 */ + void (GLAPIENTRYP VertexAttrib2sNV)(GLuint index, GLshort x, GLshort y); /* 687 */ + void (GLAPIENTRYP VertexAttrib2svNV)(GLuint index, const GLshort * v); /* 688 */ + void (GLAPIENTRYP VertexAttrib3dNV)(GLuint index, GLdouble x, GLdouble y, GLdouble z); /* 689 */ + void (GLAPIENTRYP VertexAttrib3dvNV)(GLuint index, const GLdouble * v); /* 690 */ + void (GLAPIENTRYP VertexAttrib3fNV)(GLuint index, GLfloat x, GLfloat y, GLfloat z); /* 691 */ + void (GLAPIENTRYP VertexAttrib3fvNV)(GLuint index, const GLfloat * v); /* 692 */ + void (GLAPIENTRYP VertexAttrib3sNV)(GLuint index, GLshort x, GLshort y, GLshort z); /* 693 */ + void (GLAPIENTRYP VertexAttrib3svNV)(GLuint index, const GLshort * v); /* 694 */ + void (GLAPIENTRYP VertexAttrib4dNV)(GLuint index, GLdouble x, GLdouble y, GLdouble z, GLdouble w); /* 695 */ + void (GLAPIENTRYP VertexAttrib4dvNV)(GLuint index, const GLdouble * v); /* 696 */ + void (GLAPIENTRYP VertexAttrib4fNV)(GLuint index, GLfloat x, GLfloat y, GLfloat z, GLfloat w); /* 697 */ + void (GLAPIENTRYP VertexAttrib4fvNV)(GLuint index, const GLfloat * v); /* 698 */ + void (GLAPIENTRYP VertexAttrib4sNV)(GLuint index, GLshort x, GLshort y, GLshort z, GLshort w); /* 699 */ + void (GLAPIENTRYP VertexAttrib4svNV)(GLuint index, const GLshort * v); /* 700 */ + void (GLAPIENTRYP VertexAttrib4ubNV)(GLuint index, GLubyte x, GLubyte y, GLubyte z, GLubyte w); /* 701 */ + void (GLAPIENTRYP VertexAttrib4ubvNV)(GLuint index, const GLubyte * v); /* 702 */ + void (GLAPIENTRYP VertexAttribPointerNV)(GLuint index, GLint size, GLenum type, GLsizei stride, const GLvoid * pointer); /* 703 */ + void (GLAPIENTRYP VertexAttribs1dvNV)(GLuint index, GLsizei n, const GLdouble * v); /* 704 */ + void (GLAPIENTRYP VertexAttribs1fvNV)(GLuint index, GLsizei n, const GLfloat * v); /* 705 */ + void (GLAPIENTRYP VertexAttribs1svNV)(GLuint index, GLsizei n, const GLshort * v); /* 706 */ + void (GLAPIENTRYP VertexAttribs2dvNV)(GLuint index, GLsizei n, const GLdouble * v); /* 707 */ + void (GLAPIENTRYP VertexAttribs2fvNV)(GLuint index, GLsizei n, const GLfloat * v); /* 708 */ + void (GLAPIENTRYP VertexAttribs2svNV)(GLuint index, GLsizei n, const GLshort * v); /* 709 */ + void (GLAPIENTRYP VertexAttribs3dvNV)(GLuint index, GLsizei n, const GLdouble * v); /* 710 */ + void (GLAPIENTRYP VertexAttribs3fvNV)(GLuint index, GLsizei n, const GLfloat * v); /* 711 */ + void (GLAPIENTRYP VertexAttribs3svNV)(GLuint index, GLsizei n, const GLshort * v); /* 712 */ + void (GLAPIENTRYP VertexAttribs4dvNV)(GLuint index, GLsizei n, const GLdouble * v); /* 713 */ + void (GLAPIENTRYP VertexAttribs4fvNV)(GLuint index, GLsizei n, const GLfloat * v); /* 714 */ + void (GLAPIENTRYP VertexAttribs4svNV)(GLuint index, GLsizei n, const GLshort * v); /* 715 */ + void (GLAPIENTRYP VertexAttribs4ubvNV)(GLuint index, GLsizei n, const GLubyte * v); /* 716 */ + void (GLAPIENTRYP AlphaFragmentOp1ATI)(GLenum op, GLuint dst, GLuint dstMod, GLuint arg1, GLuint arg1Rep, GLuint arg1Mod); /* 717 */ + void (GLAPIENTRYP AlphaFragmentOp2ATI)(GLenum op, GLuint dst, GLuint dstMod, GLuint arg1, GLuint arg1Rep, GLuint arg1Mod, GLuint arg2, GLuint arg2Rep, GLuint arg2Mod); /* 718 */ + void (GLAPIENTRYP AlphaFragmentOp3ATI)(GLenum op, GLuint dst, GLuint dstMod, GLuint arg1, GLuint arg1Rep, GLuint arg1Mod, GLuint arg2, GLuint arg2Rep, GLuint arg2Mod, GLuint arg3, GLuint arg3Rep, GLuint arg3Mod); /* 719 */ + void (GLAPIENTRYP BeginFragmentShaderATI)(void); /* 720 */ + void (GLAPIENTRYP BindFragmentShaderATI)(GLuint id); /* 721 */ + void (GLAPIENTRYP ColorFragmentOp1ATI)(GLenum op, GLuint dst, GLuint dstMask, GLuint dstMod, GLuint arg1, GLuint arg1Rep, GLuint arg1Mod); /* 722 */ + void (GLAPIENTRYP ColorFragmentOp2ATI)(GLenum op, GLuint dst, GLuint dstMask, GLuint dstMod, GLuint arg1, GLuint arg1Rep, GLuint arg1Mod, GLuint arg2, GLuint arg2Rep, GLuint arg2Mod); /* 723 */ + void (GLAPIENTRYP ColorFragmentOp3ATI)(GLenum op, GLuint dst, GLuint dstMask, GLuint dstMod, GLuint arg1, GLuint arg1Rep, GLuint arg1Mod, GLuint arg2, GLuint arg2Rep, GLuint arg2Mod, GLuint arg3, GLuint arg3Rep, GLuint arg3Mod); /* 724 */ + void (GLAPIENTRYP DeleteFragmentShaderATI)(GLuint id); /* 725 */ + void (GLAPIENTRYP EndFragmentShaderATI)(void); /* 726 */ + GLuint (GLAPIENTRYP GenFragmentShadersATI)(GLuint range); /* 727 */ + void (GLAPIENTRYP PassTexCoordATI)(GLuint dst, GLuint coord, GLenum swizzle); /* 728 */ + void (GLAPIENTRYP SampleMapATI)(GLuint dst, GLuint interp, GLenum swizzle); /* 729 */ + void (GLAPIENTRYP SetFragmentShaderConstantATI)(GLuint dst, const GLfloat * value); /* 730 */ + void (GLAPIENTRYP PointParameteriNV)(GLenum pname, GLint param); /* 731 */ + void (GLAPIENTRYP PointParameterivNV)(GLenum pname, const GLint * params); /* 732 */ + void (GLAPIENTRYP ActiveStencilFaceEXT)(GLenum face); /* 733 */ + void (GLAPIENTRYP BindVertexArrayAPPLE)(GLuint array); /* 734 */ + void (GLAPIENTRYP DeleteVertexArraysAPPLE)(GLsizei n, const GLuint * arrays); /* 735 */ + void (GLAPIENTRYP GenVertexArraysAPPLE)(GLsizei n, GLuint * arrays); /* 736 */ + GLboolean (GLAPIENTRYP IsVertexArrayAPPLE)(GLuint array); /* 737 */ + void (GLAPIENTRYP GetProgramNamedParameterdvNV)(GLuint id, GLsizei len, const GLubyte * name, GLdouble * params); /* 738 */ + void (GLAPIENTRYP GetProgramNamedParameterfvNV)(GLuint id, GLsizei len, const GLubyte * name, GLfloat * params); /* 739 */ + void (GLAPIENTRYP ProgramNamedParameter4dNV)(GLuint id, GLsizei len, const GLubyte * name, GLdouble x, GLdouble y, GLdouble z, GLdouble w); /* 740 */ + void (GLAPIENTRYP ProgramNamedParameter4dvNV)(GLuint id, GLsizei len, const GLubyte * name, const GLdouble * v); /* 741 */ + void (GLAPIENTRYP ProgramNamedParameter4fNV)(GLuint id, GLsizei len, const GLubyte * name, GLfloat x, GLfloat y, GLfloat z, GLfloat w); /* 742 */ + void (GLAPIENTRYP ProgramNamedParameter4fvNV)(GLuint id, GLsizei len, const GLubyte * name, const GLfloat * v); /* 743 */ + void (GLAPIENTRYP DepthBoundsEXT)(GLclampd zmin, GLclampd zmax); /* 744 */ + void (GLAPIENTRYP BlendEquationSeparateEXT)(GLenum modeRGB, GLenum modeA); /* 745 */ + void (GLAPIENTRYP BindFramebufferEXT)(GLenum target, GLuint framebuffer); /* 746 */ + void (GLAPIENTRYP BindRenderbufferEXT)(GLenum target, GLuint renderbuffer); /* 747 */ + GLenum (GLAPIENTRYP CheckFramebufferStatusEXT)(GLenum target); /* 748 */ + void (GLAPIENTRYP DeleteFramebuffersEXT)(GLsizei n, const GLuint * framebuffers); /* 749 */ + void (GLAPIENTRYP DeleteRenderbuffersEXT)(GLsizei n, const GLuint * renderbuffers); /* 750 */ + void (GLAPIENTRYP FramebufferRenderbufferEXT)(GLenum target, GLenum attachment, GLenum renderbuffertarget, GLuint renderbuffer); /* 751 */ + void (GLAPIENTRYP FramebufferTexture1DEXT)(GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level); /* 752 */ + void (GLAPIENTRYP FramebufferTexture2DEXT)(GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level); /* 753 */ + void (GLAPIENTRYP FramebufferTexture3DEXT)(GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level, GLint zoffset); /* 754 */ + void (GLAPIENTRYP GenFramebuffersEXT)(GLsizei n, GLuint * framebuffers); /* 755 */ + void (GLAPIENTRYP GenRenderbuffersEXT)(GLsizei n, GLuint * renderbuffers); /* 756 */ + void (GLAPIENTRYP GenerateMipmapEXT)(GLenum target); /* 757 */ + void (GLAPIENTRYP GetFramebufferAttachmentParameterivEXT)(GLenum target, GLenum attachment, GLenum pname, GLint * params); /* 758 */ + void (GLAPIENTRYP GetRenderbufferParameterivEXT)(GLenum target, GLenum pname, GLint * params); /* 759 */ + GLboolean (GLAPIENTRYP IsFramebufferEXT)(GLuint framebuffer); /* 760 */ + GLboolean (GLAPIENTRYP IsRenderbufferEXT)(GLuint renderbuffer); /* 761 */ + void (GLAPIENTRYP RenderbufferStorageEXT)(GLenum target, GLenum internalformat, GLsizei width, GLsizei height); /* 762 */ + void (GLAPIENTRYP BlitFramebufferEXT)(GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter); /* 763 */ + void (GLAPIENTRYP FramebufferTextureLayerEXT)(GLenum target, GLenum attachment, GLuint texture, GLint level, GLint layer); /* 764 */ + void (GLAPIENTRYP StencilFuncSeparateATI)(GLenum frontfunc, GLenum backfunc, GLint ref, GLuint mask); /* 765 */ + void (GLAPIENTRYP ProgramEnvParameters4fvEXT)(GLenum target, GLuint index, GLsizei count, const GLfloat * params); /* 766 */ + void (GLAPIENTRYP ProgramLocalParameters4fvEXT)(GLenum target, GLuint index, GLsizei count, const GLfloat * params); /* 767 */ + void (GLAPIENTRYP GetQueryObjecti64vEXT)(GLuint id, GLenum pname, GLint64EXT * params); /* 768 */ + void (GLAPIENTRYP GetQueryObjectui64vEXT)(GLuint id, GLenum pname, GLuint64EXT * params); /* 769 */ }; #endif /* !defined( _GLAPI_TABLE_H_ ) */ diff --git a/src/mesa/glapi/glapitemp.h b/src/mesa/glapi/glapitemp.h index 6ded362c0e..09259f4704 100644 --- a/src/mesa/glapi/glapitemp.h +++ b/src/mesa/glapi/glapitemp.h @@ -3001,21 +3001,41 @@ KEYWORD1 void KEYWORD2 NAME(ProgramEnvParameter4dARB)(GLenum target, GLuint inde DISPATCH(ProgramEnvParameter4dARB, (target, index, x, y, z, w), (F, "glProgramEnvParameter4dARB(0x%x, %d, %f, %f, %f, %f);\n", target, index, x, y, z, w)); } +KEYWORD1 void KEYWORD2 NAME(ProgramParameter4dNV)(GLenum target, GLuint index, GLdouble x, GLdouble y, GLdouble z, GLdouble w) +{ + DISPATCH(ProgramEnvParameter4dARB, (target, index, x, y, z, w), (F, "glProgramParameter4dNV(0x%x, %d, %f, %f, %f, %f);\n", target, index, x, y, z, w)); +} + KEYWORD1 void KEYWORD2 NAME(ProgramEnvParameter4dvARB)(GLenum target, GLuint index, const GLdouble * params) { DISPATCH(ProgramEnvParameter4dvARB, (target, index, params), (F, "glProgramEnvParameter4dvARB(0x%x, %d, %p);\n", target, index, (const void *) params)); } +KEYWORD1 void KEYWORD2 NAME(ProgramParameter4dvNV)(GLenum target, GLuint index, const GLdouble * params) +{ + DISPATCH(ProgramEnvParameter4dvARB, (target, index, params), (F, "glProgramParameter4dvNV(0x%x, %d, %p);\n", target, index, (const void *) params)); +} + KEYWORD1 void KEYWORD2 NAME(ProgramEnvParameter4fARB)(GLenum target, GLuint index, GLfloat x, GLfloat y, GLfloat z, GLfloat w) { DISPATCH(ProgramEnvParameter4fARB, (target, index, x, y, z, w), (F, "glProgramEnvParameter4fARB(0x%x, %d, %f, %f, %f, %f);\n", target, index, x, y, z, w)); } +KEYWORD1 void KEYWORD2 NAME(ProgramParameter4fNV)(GLenum target, GLuint index, GLfloat x, GLfloat y, GLfloat z, GLfloat w) +{ + DISPATCH(ProgramEnvParameter4fARB, (target, index, x, y, z, w), (F, "glProgramParameter4fNV(0x%x, %d, %f, %f, %f, %f);\n", target, index, x, y, z, w)); +} + KEYWORD1 void KEYWORD2 NAME(ProgramEnvParameter4fvARB)(GLenum target, GLuint index, const GLfloat * params) { DISPATCH(ProgramEnvParameter4fvARB, (target, index, params), (F, "glProgramEnvParameter4fvARB(0x%x, %d, %p);\n", target, index, (const void *) params)); } +KEYWORD1 void KEYWORD2 NAME(ProgramParameter4fvNV)(GLenum target, GLuint index, const GLfloat * params) +{ + DISPATCH(ProgramEnvParameter4fvARB, (target, index, params), (F, "glProgramParameter4fvNV(0x%x, %d, %p);\n", target, index, (const void *) params)); +} + KEYWORD1 void KEYWORD2 NAME(ProgramLocalParameter4dARB)(GLenum target, GLuint index, GLdouble x, GLdouble y, GLdouble z, GLdouble w) { DISPATCH(ProgramLocalParameter4dARB, (target, index, x, y, z, w), (F, "glProgramLocalParameter4dARB(0x%x, %d, %f, %f, %f, %f);\n", target, index, x, y, z, w)); @@ -4942,26 +4962,6 @@ KEYWORD1 void KEYWORD2 NAME(LoadProgramNV)(GLenum target, GLuint id, GLsizei len DISPATCH(LoadProgramNV, (target, id, len, program), (F, "glLoadProgramNV(0x%x, %d, %d, %p);\n", target, id, len, (const void *) program)); } -KEYWORD1 void KEYWORD2 NAME(ProgramParameter4dNV)(GLenum target, GLuint index, GLdouble x, GLdouble y, GLdouble z, GLdouble w) -{ - DISPATCH(ProgramParameter4dNV, (target, index, x, y, z, w), (F, "glProgramParameter4dNV(0x%x, %d, %f, %f, %f, %f);\n", target, index, x, y, z, w)); -} - -KEYWORD1 void KEYWORD2 NAME(ProgramParameter4dvNV)(GLenum target, GLuint index, const GLdouble * params) -{ - DISPATCH(ProgramParameter4dvNV, (target, index, params), (F, "glProgramParameter4dvNV(0x%x, %d, %p);\n", target, index, (const void *) params)); -} - -KEYWORD1 void KEYWORD2 NAME(ProgramParameter4fNV)(GLenum target, GLuint index, GLfloat x, GLfloat y, GLfloat z, GLfloat w) -{ - DISPATCH(ProgramParameter4fNV, (target, index, x, y, z, w), (F, "glProgramParameter4fNV(0x%x, %d, %f, %f, %f, %f);\n", target, index, x, y, z, w)); -} - -KEYWORD1 void KEYWORD2 NAME(ProgramParameter4fvNV)(GLenum target, GLuint index, const GLfloat * params) -{ - DISPATCH(ProgramParameter4fvNV, (target, index, params), (F, "glProgramParameter4fvNV(0x%x, %d, %p);\n", target, index, (const void *) params)); -} - KEYWORD1 void KEYWORD2 NAME(ProgramParameters4dvNV)(GLenum target, GLuint index, GLuint num, const GLdouble * params) { DISPATCH(ProgramParameters4dvNV, (target, index, num, params), (F, "glProgramParameters4dvNV(0x%x, %d, %d, %p);\n", target, index, num, (const void *) params)); @@ -5272,37 +5272,37 @@ KEYWORD1 void KEYWORD2 NAME(PointParameterivNV)(GLenum pname, const GLint * para DISPATCH(PointParameterivNV, (pname, params), (F, "glPointParameterivNV(0x%x, %p);\n", pname, (const void *) params)); } -KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_737)(GLenum face); +KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_733)(GLenum face); -KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_737)(GLenum face) +KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_733)(GLenum face) { DISPATCH(ActiveStencilFaceEXT, (face), (F, "glActiveStencilFaceEXT(0x%x);\n", face)); } -KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_738)(GLuint array); +KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_734)(GLuint array); -KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_738)(GLuint array) +KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_734)(GLuint array) { DISPATCH(BindVertexArrayAPPLE, (array), (F, "glBindVertexArrayAPPLE(%d);\n", array)); } -KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_739)(GLsizei n, const GLuint * arrays); +KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_735)(GLsizei n, const GLuint * arrays); -KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_739)(GLsizei n, const GLuint * arrays) +KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_735)(GLsizei n, const GLuint * arrays) { DISPATCH(DeleteVertexArraysAPPLE, (n, arrays), (F, "glDeleteVertexArraysAPPLE(%d, %p);\n", n, (const void *) arrays)); } -KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_740)(GLsizei n, GLuint * arrays); +KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_736)(GLsizei n, GLuint * arrays); -KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_740)(GLsizei n, GLuint * arrays) +KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_736)(GLsizei n, GLuint * arrays) { DISPATCH(GenVertexArraysAPPLE, (n, arrays), (F, "glGenVertexArraysAPPLE(%d, %p);\n", n, (const void *) arrays)); } -KEYWORD1_ALT GLboolean KEYWORD2 NAME(_dispatch_stub_741)(GLuint array); +KEYWORD1_ALT GLboolean KEYWORD2 NAME(_dispatch_stub_737)(GLuint array); -KEYWORD1_ALT GLboolean KEYWORD2 NAME(_dispatch_stub_741)(GLuint array) +KEYWORD1_ALT GLboolean KEYWORD2 NAME(_dispatch_stub_737)(GLuint array) { RETURN_DISPATCH(IsVertexArrayAPPLE, (array), (F, "glIsVertexArrayAPPLE(%d);\n", array)); } @@ -5337,9 +5337,9 @@ KEYWORD1 void KEYWORD2 NAME(ProgramNamedParameter4fvNV)(GLuint id, GLsizei len, DISPATCH(ProgramNamedParameter4fvNV, (id, len, name, v), (F, "glProgramNamedParameter4fvNV(%d, %d, %p, %p);\n", id, len, (const void *) name, (const void *) v)); } -KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_748)(GLclampd zmin, GLclampd zmax); +KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_744)(GLclampd zmin, GLclampd zmax); -KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_748)(GLclampd zmin, GLclampd zmax) +KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_744)(GLclampd zmin, GLclampd zmax) { DISPATCH(DepthBoundsEXT, (zmin, zmax), (F, "glDepthBoundsEXT(%f, %f);\n", zmin, zmax)); } @@ -5349,9 +5349,9 @@ KEYWORD1 void KEYWORD2 NAME(BlendEquationSeparate)(GLenum modeRGB, GLenum modeA) DISPATCH(BlendEquationSeparateEXT, (modeRGB, modeA), (F, "glBlendEquationSeparate(0x%x, 0x%x);\n", modeRGB, modeA)); } -KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_749)(GLenum modeRGB, GLenum modeA); +KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_745)(GLenum modeRGB, GLenum modeA); -KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_749)(GLenum modeRGB, GLenum modeA) +KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_745)(GLenum modeRGB, GLenum modeA) { DISPATCH(BlendEquationSeparateEXT, (modeRGB, modeA), (F, "glBlendEquationSeparateEXT(0x%x, 0x%x);\n", modeRGB, modeA)); } @@ -5441,9 +5441,9 @@ KEYWORD1 void KEYWORD2 NAME(RenderbufferStorageEXT)(GLenum target, GLenum intern DISPATCH(RenderbufferStorageEXT, (target, internalformat, width, height), (F, "glRenderbufferStorageEXT(0x%x, 0x%x, %d, %d);\n", target, internalformat, width, height)); } -KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_767)(GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter); +KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_763)(GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter); -KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_767)(GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter) +KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_763)(GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter) { DISPATCH(BlitFramebufferEXT, (srcX0, srcY0, srcX1, srcY1, dstX0, dstY0, dstX1, dstY1, mask, filter), (F, "glBlitFramebufferEXT(%d, %d, %d, %d, %d, %d, %d, %d, %d, 0x%x);\n", srcX0, srcY0, srcX1, srcY1, dstX0, dstY0, dstX1, dstY1, mask, filter)); } @@ -5453,37 +5453,37 @@ KEYWORD1 void KEYWORD2 NAME(FramebufferTextureLayerEXT)(GLenum target, GLenum at DISPATCH(FramebufferTextureLayerEXT, (target, attachment, texture, level, layer), (F, "glFramebufferTextureLayerEXT(0x%x, 0x%x, %d, %d, %d);\n", target, attachment, texture, level, layer)); } -KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_769)(GLenum frontfunc, GLenum backfunc, GLint ref, GLuint mask); +KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_765)(GLenum frontfunc, GLenum backfunc, GLint ref, GLuint mask); -KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_769)(GLenum frontfunc, GLenum backfunc, GLint ref, GLuint mask) +KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_765)(GLenum frontfunc, GLenum backfunc, GLint ref, GLuint mask) { DISPATCH(StencilFuncSeparateATI, (frontfunc, backfunc, ref, mask), (F, "glStencilFuncSeparateATI(0x%x, 0x%x, %d, %d);\n", frontfunc, backfunc, ref, mask)); } -KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_770)(GLenum target, GLuint index, GLsizei count, const GLfloat * params); +KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_766)(GLenum target, GLuint index, GLsizei count, const GLfloat * params); -KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_770)(GLenum target, GLuint index, GLsizei count, const GLfloat * params) +KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_766)(GLenum target, GLuint index, GLsizei count, const GLfloat * params) { DISPATCH(ProgramEnvParameters4fvEXT, (target, index, count, params), (F, "glProgramEnvParameters4fvEXT(0x%x, %d, %d, %p);\n", target, index, count, (const void *) params)); } -KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_771)(GLenum target, GLuint index, GLsizei count, const GLfloat * params); +KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_767)(GLenum target, GLuint index, GLsizei count, const GLfloat * params); -KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_771)(GLenum target, GLuint index, GLsizei count, const GLfloat * params) +KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_767)(GLenum target, GLuint index, GLsizei count, const GLfloat * params) { DISPATCH(ProgramLocalParameters4fvEXT, (target, index, count, params), (F, "glProgramLocalParameters4fvEXT(0x%x, %d, %d, %p);\n", target, index, count, (const void *) params)); } -KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_772)(GLuint id, GLenum pname, GLint64EXT * params); +KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_768)(GLuint id, GLenum pname, GLint64EXT * params); -KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_772)(GLuint id, GLenum pname, GLint64EXT * params) +KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_768)(GLuint id, GLenum pname, GLint64EXT * params) { DISPATCH(GetQueryObjecti64vEXT, (id, pname, params), (F, "glGetQueryObjecti64vEXT(%d, 0x%x, %p);\n", id, pname, (const void *) params)); } -KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_773)(GLuint id, GLenum pname, GLuint64EXT * params); +KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_769)(GLuint id, GLenum pname, GLuint64EXT * params); -KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_773)(GLuint id, GLenum pname, GLuint64EXT * params) +KEYWORD1_ALT void KEYWORD2 NAME(_dispatch_stub_769)(GLuint id, GLenum pname, GLuint64EXT * params) { DISPATCH(GetQueryObjectui64vEXT, (id, pname, params), (F, "glGetQueryObjectui64vEXT(%d, 0x%x, %p);\n", id, pname, (const void *) params)); } @@ -6175,10 +6175,6 @@ static _glapi_proc DISPATCH_TABLE_NAME[] = { TABLE_ENTRY(GetVertexAttribivNV), TABLE_ENTRY(IsProgramNV), TABLE_ENTRY(LoadProgramNV), - TABLE_ENTRY(ProgramParameter4dNV), - TABLE_ENTRY(ProgramParameter4dvNV), - TABLE_ENTRY(ProgramParameter4fNV), - TABLE_ENTRY(ProgramParameter4fvNV), TABLE_ENTRY(ProgramParameters4dvNV), TABLE_ENTRY(ProgramParameters4fvNV), TABLE_ENTRY(RequestResidentProgramsNV), @@ -6239,19 +6235,19 @@ static _glapi_proc DISPATCH_TABLE_NAME[] = { TABLE_ENTRY(SetFragmentShaderConstantATI), TABLE_ENTRY(PointParameteriNV), TABLE_ENTRY(PointParameterivNV), + TABLE_ENTRY(_dispatch_stub_733), + TABLE_ENTRY(_dispatch_stub_734), + TABLE_ENTRY(_dispatch_stub_735), + TABLE_ENTRY(_dispatch_stub_736), TABLE_ENTRY(_dispatch_stub_737), - TABLE_ENTRY(_dispatch_stub_738), - TABLE_ENTRY(_dispatch_stub_739), - TABLE_ENTRY(_dispatch_stub_740), - TABLE_ENTRY(_dispatch_stub_741), TABLE_ENTRY(GetProgramNamedParameterdvNV), TABLE_ENTRY(GetProgramNamedParameterfvNV), TABLE_ENTRY(ProgramNamedParameter4dNV), TABLE_ENTRY(ProgramNamedParameter4dvNV), TABLE_ENTRY(ProgramNamedParameter4fNV), TABLE_ENTRY(ProgramNamedParameter4fvNV), - TABLE_ENTRY(_dispatch_stub_748), - TABLE_ENTRY(_dispatch_stub_749), + TABLE_ENTRY(_dispatch_stub_744), + TABLE_ENTRY(_dispatch_stub_745), TABLE_ENTRY(BindFramebufferEXT), TABLE_ENTRY(BindRenderbufferEXT), TABLE_ENTRY(CheckFramebufferStatusEXT), @@ -6269,13 +6265,13 @@ static _glapi_proc DISPATCH_TABLE_NAME[] = { TABLE_ENTRY(IsFramebufferEXT), TABLE_ENTRY(IsRenderbufferEXT), TABLE_ENTRY(RenderbufferStorageEXT), - TABLE_ENTRY(_dispatch_stub_767), + TABLE_ENTRY(_dispatch_stub_763), TABLE_ENTRY(FramebufferTextureLayerEXT), + TABLE_ENTRY(_dispatch_stub_765), + TABLE_ENTRY(_dispatch_stub_766), + TABLE_ENTRY(_dispatch_stub_767), + TABLE_ENTRY(_dispatch_stub_768), TABLE_ENTRY(_dispatch_stub_769), - TABLE_ENTRY(_dispatch_stub_770), - TABLE_ENTRY(_dispatch_stub_771), - TABLE_ENTRY(_dispatch_stub_772), - TABLE_ENTRY(_dispatch_stub_773), /* A whole bunch of no-op functions. These might be called * when someone tries to call a dynamically-registered * extension function without a current rendering context. @@ -6479,6 +6475,10 @@ static _glapi_proc UNUSED_TABLE_NAME[] = { TABLE_ENTRY(GetVertexAttribdv), TABLE_ENTRY(GetVertexAttribfv), TABLE_ENTRY(GetVertexAttribiv), + TABLE_ENTRY(ProgramParameter4dNV), + TABLE_ENTRY(ProgramParameter4dvNV), + TABLE_ENTRY(ProgramParameter4fNV), + TABLE_ENTRY(ProgramParameter4fvNV), TABLE_ENTRY(VertexAttrib1d), TABLE_ENTRY(VertexAttrib1dv), TABLE_ENTRY(VertexAttrib1f), diff --git a/src/mesa/glapi/glprocs.h b/src/mesa/glapi/glprocs.h index abc22a9a04..5a99d505ec 100644 --- a/src/mesa/glapi/glprocs.h +++ b/src/mesa/glapi/glprocs.h @@ -725,10 +725,6 @@ static const char gl_string_table[] = "glGetVertexAttribivNV\0" "glIsProgramNV\0" "glLoadProgramNV\0" - "glProgramParameter4dNV\0" - "glProgramParameter4dvNV\0" - "glProgramParameter4fNV\0" - "glProgramParameter4fvNV\0" "glProgramParameters4dvNV\0" "glProgramParameters4fvNV\0" "glRequestResidentProgramsNV\0" @@ -935,6 +931,10 @@ static const char gl_string_table[] = "glGetVertexAttribdv\0" "glGetVertexAttribfv\0" "glGetVertexAttribiv\0" + "glProgramParameter4dNV\0" + "glProgramParameter4dvNV\0" + "glProgramParameter4fNV\0" + "glProgramParameter4fvNV\0" "glVertexAttrib1d\0" "glVertexAttrib1dv\0" "glVertexAttrib1f\0" @@ -1138,19 +1138,19 @@ static const char gl_string_table[] = #define gl_dispatch_stub_654 mgl_dispatch_stub_654 #define gl_dispatch_stub_655 mgl_dispatch_stub_655 #define gl_dispatch_stub_656 mgl_dispatch_stub_656 +#define gl_dispatch_stub_733 mgl_dispatch_stub_733 +#define gl_dispatch_stub_734 mgl_dispatch_stub_734 +#define gl_dispatch_stub_735 mgl_dispatch_stub_735 +#define gl_dispatch_stub_736 mgl_dispatch_stub_736 #define gl_dispatch_stub_737 mgl_dispatch_stub_737 -#define gl_dispatch_stub_738 mgl_dispatch_stub_738 -#define gl_dispatch_stub_739 mgl_dispatch_stub_739 -#define gl_dispatch_stub_740 mgl_dispatch_stub_740 -#define gl_dispatch_stub_741 mgl_dispatch_stub_741 -#define gl_dispatch_stub_748 mgl_dispatch_stub_748 -#define gl_dispatch_stub_749 mgl_dispatch_stub_749 +#define gl_dispatch_stub_744 mgl_dispatch_stub_744 +#define gl_dispatch_stub_745 mgl_dispatch_stub_745 +#define gl_dispatch_stub_763 mgl_dispatch_stub_763 +#define gl_dispatch_stub_765 mgl_dispatch_stub_765 +#define gl_dispatch_stub_766 mgl_dispatch_stub_766 #define gl_dispatch_stub_767 mgl_dispatch_stub_767 +#define gl_dispatch_stub_768 mgl_dispatch_stub_768 #define gl_dispatch_stub_769 mgl_dispatch_stub_769 -#define gl_dispatch_stub_770 mgl_dispatch_stub_770 -#define gl_dispatch_stub_771 mgl_dispatch_stub_771 -#define gl_dispatch_stub_772 mgl_dispatch_stub_772 -#define gl_dispatch_stub_773 mgl_dispatch_stub_773 #endif /* USE_MGL_NAMESPACE */ @@ -1189,19 +1189,19 @@ extern void gl_dispatch_stub_653(void); extern void gl_dispatch_stub_654(void); extern void gl_dispatch_stub_655(void); extern void gl_dispatch_stub_656(void); +extern void gl_dispatch_stub_733(void); +extern void gl_dispatch_stub_734(void); +extern void gl_dispatch_stub_735(void); +extern void gl_dispatch_stub_736(void); extern void gl_dispatch_stub_737(void); -extern void gl_dispatch_stub_738(void); -extern void gl_dispatch_stub_739(void); -extern void gl_dispatch_stub_740(void); -extern void gl_dispatch_stub_741(void); -extern void gl_dispatch_stub_748(void); -extern void gl_dispatch_stub_749(void); +extern void gl_dispatch_stub_744(void); +extern void gl_dispatch_stub_745(void); +extern void gl_dispatch_stub_763(void); +extern void gl_dispatch_stub_765(void); +extern void gl_dispatch_stub_766(void); extern void gl_dispatch_stub_767(void); +extern void gl_dispatch_stub_768(void); extern void gl_dispatch_stub_769(void); -extern void gl_dispatch_stub_770(void); -extern void gl_dispatch_stub_771(void); -extern void gl_dispatch_stub_772(void); -extern void gl_dispatch_stub_773(void); #endif /* defined(NEED_FUNCTION_POINTER) || defined(GLX_INDIRECT_RENDERING) */ static const glprocs_table_t static_functions[] = { @@ -1878,216 +1878,216 @@ static const glprocs_table_t static_functions[] = { NAME_FUNC_OFFSET(11280, glGetVertexAttribivNV, glGetVertexAttribivNV, NULL, _gloffset_GetVertexAttribivNV), NAME_FUNC_OFFSET(11302, glIsProgramNV, glIsProgramNV, NULL, _gloffset_IsProgramNV), NAME_FUNC_OFFSET(11316, glLoadProgramNV, glLoadProgramNV, NULL, _gloffset_LoadProgramNV), - NAME_FUNC_OFFSET(11332, glProgramParameter4dNV, glProgramParameter4dNV, NULL, _gloffset_ProgramParameter4dNV), - NAME_FUNC_OFFSET(11355, glProgramParameter4dvNV, glProgramParameter4dvNV, NULL, _gloffset_ProgramParameter4dvNV), - NAME_FUNC_OFFSET(11379, glProgramParameter4fNV, glProgramParameter4fNV, NULL, _gloffset_ProgramParameter4fNV), - NAME_FUNC_OFFSET(11402, glProgramParameter4fvNV, glProgramParameter4fvNV, NULL, _gloffset_ProgramParameter4fvNV), - NAME_FUNC_OFFSET(11426, glProgramParameters4dvNV, glProgramParameters4dvNV, NULL, _gloffset_ProgramParameters4dvNV), - NAME_FUNC_OFFSET(11451, glProgramParameters4fvNV, glProgramParameters4fvNV, NULL, _gloffset_ProgramParameters4fvNV), - NAME_FUNC_OFFSET(11476, glRequestResidentProgramsNV, glRequestResidentProgramsNV, NULL, _gloffset_RequestResidentProgramsNV), - NAME_FUNC_OFFSET(11504, glTrackMatrixNV, glTrackMatrixNV, NULL, _gloffset_TrackMatrixNV), - NAME_FUNC_OFFSET(11520, glVertexAttrib1dNV, glVertexAttrib1dNV, NULL, _gloffset_VertexAttrib1dNV), - NAME_FUNC_OFFSET(11539, glVertexAttrib1dvNV, glVertexAttrib1dvNV, NULL, _gloffset_VertexAttrib1dvNV), - NAME_FUNC_OFFSET(11559, glVertexAttrib1fNV, glVertexAttrib1fNV, NULL, _gloffset_VertexAttrib1fNV), - NAME_FUNC_OFFSET(11578, glVertexAttrib1fvNV, glVertexAttrib1fvNV, NULL, _gloffset_VertexAttrib1fvNV), - NAME_FUNC_OFFSET(11598, glVertexAttrib1sNV, glVertexAttrib1sNV, NULL, _gloffset_VertexAttrib1sNV), - NAME_FUNC_OFFSET(11617, glVertexAttrib1svNV, glVertexAttrib1svNV, NULL, _gloffset_VertexAttrib1svNV), - NAME_FUNC_OFFSET(11637, glVertexAttrib2dNV, glVertexAttrib2dNV, NULL, _gloffset_VertexAttrib2dNV), - NAME_FUNC_OFFSET(11656, glVertexAttrib2dvNV, glVertexAttrib2dvNV, NULL, _gloffset_VertexAttrib2dvNV), - NAME_FUNC_OFFSET(11676, glVertexAttrib2fNV, glVertexAttrib2fNV, NULL, _gloffset_VertexAttrib2fNV), - NAME_FUNC_OFFSET(11695, glVertexAttrib2fvNV, glVertexAttrib2fvNV, NULL, _gloffset_VertexAttrib2fvNV), - NAME_FUNC_OFFSET(11715, glVertexAttrib2sNV, glVertexAttrib2sNV, NULL, _gloffset_VertexAttrib2sNV), - NAME_FUNC_OFFSET(11734, glVertexAttrib2svNV, glVertexAttrib2svNV, NULL, _gloffset_VertexAttrib2svNV), - NAME_FUNC_OFFSET(11754, glVertexAttrib3dNV, glVertexAttrib3dNV, NULL, _gloffset_VertexAttrib3dNV), - NAME_FUNC_OFFSET(11773, glVertexAttrib3dvNV, glVertexAttrib3dvNV, NULL, _gloffset_VertexAttrib3dvNV), - NAME_FUNC_OFFSET(11793, glVertexAttrib3fNV, glVertexAttrib3fNV, NULL, _gloffset_VertexAttrib3fNV), - NAME_FUNC_OFFSET(11812, glVertexAttrib3fvNV, glVertexAttrib3fvNV, NULL, _gloffset_VertexAttrib3fvNV), - NAME_FUNC_OFFSET(11832, glVertexAttrib3sNV, glVertexAttrib3sNV, NULL, _gloffset_VertexAttrib3sNV), - NAME_FUNC_OFFSET(11851, glVertexAttrib3svNV, glVertexAttrib3svNV, NULL, _gloffset_VertexAttrib3svNV), - NAME_FUNC_OFFSET(11871, glVertexAttrib4dNV, glVertexAttrib4dNV, NULL, _gloffset_VertexAttrib4dNV), - NAME_FUNC_OFFSET(11890, glVertexAttrib4dvNV, glVertexAttrib4dvNV, NULL, _gloffset_VertexAttrib4dvNV), - NAME_FUNC_OFFSET(11910, glVertexAttrib4fNV, glVertexAttrib4fNV, NULL, _gloffset_VertexAttrib4fNV), - NAME_FUNC_OFFSET(11929, glVertexAttrib4fvNV, glVertexAttrib4fvNV, NULL, _gloffset_VertexAttrib4fvNV), - NAME_FUNC_OFFSET(11949, glVertexAttrib4sNV, glVertexAttrib4sNV, NULL, _gloffset_VertexAttrib4sNV), - NAME_FUNC_OFFSET(11968, glVertexAttrib4svNV, glVertexAttrib4svNV, NULL, _gloffset_VertexAttrib4svNV), - NAME_FUNC_OFFSET(11988, glVertexAttrib4ubNV, glVertexAttrib4ubNV, NULL, _gloffset_VertexAttrib4ubNV), - NAME_FUNC_OFFSET(12008, glVertexAttrib4ubvNV, glVertexAttrib4ubvNV, NULL, _gloffset_VertexAttrib4ubvNV), - NAME_FUNC_OFFSET(12029, glVertexAttribPointerNV, glVertexAttribPointerNV, NULL, _gloffset_VertexAttribPointerNV), - NAME_FUNC_OFFSET(12053, glVertexAttribs1dvNV, glVertexAttribs1dvNV, NULL, _gloffset_VertexAttribs1dvNV), - NAME_FUNC_OFFSET(12074, glVertexAttribs1fvNV, glVertexAttribs1fvNV, NULL, _gloffset_VertexAttribs1fvNV), - NAME_FUNC_OFFSET(12095, glVertexAttribs1svNV, glVertexAttribs1svNV, NULL, _gloffset_VertexAttribs1svNV), - NAME_FUNC_OFFSET(12116, glVertexAttribs2dvNV, glVertexAttribs2dvNV, NULL, _gloffset_VertexAttribs2dvNV), - NAME_FUNC_OFFSET(12137, glVertexAttribs2fvNV, glVertexAttribs2fvNV, NULL, _gloffset_VertexAttribs2fvNV), - NAME_FUNC_OFFSET(12158, glVertexAttribs2svNV, glVertexAttribs2svNV, NULL, _gloffset_VertexAttribs2svNV), - NAME_FUNC_OFFSET(12179, glVertexAttribs3dvNV, glVertexAttribs3dvNV, NULL, _gloffset_VertexAttribs3dvNV), - NAME_FUNC_OFFSET(12200, glVertexAttribs3fvNV, glVertexAttribs3fvNV, NULL, _gloffset_VertexAttribs3fvNV), - NAME_FUNC_OFFSET(12221, glVertexAttribs3svNV, glVertexAttribs3svNV, NULL, _gloffset_VertexAttribs3svNV), - NAME_FUNC_OFFSET(12242, glVertexAttribs4dvNV, glVertexAttribs4dvNV, NULL, _gloffset_VertexAttribs4dvNV), - NAME_FUNC_OFFSET(12263, glVertexAttribs4fvNV, glVertexAttribs4fvNV, NULL, _gloffset_VertexAttribs4fvNV), - NAME_FUNC_OFFSET(12284, glVertexAttribs4svNV, glVertexAttribs4svNV, NULL, _gloffset_VertexAttribs4svNV), - NAME_FUNC_OFFSET(12305, glVertexAttribs4ubvNV, glVertexAttribs4ubvNV, NULL, _gloffset_VertexAttribs4ubvNV), - NAME_FUNC_OFFSET(12327, glAlphaFragmentOp1ATI, glAlphaFragmentOp1ATI, NULL, _gloffset_AlphaFragmentOp1ATI), - NAME_FUNC_OFFSET(12349, glAlphaFragmentOp2ATI, glAlphaFragmentOp2ATI, NULL, _gloffset_AlphaFragmentOp2ATI), - NAME_FUNC_OFFSET(12371, glAlphaFragmentOp3ATI, glAlphaFragmentOp3ATI, NULL, _gloffset_AlphaFragmentOp3ATI), - NAME_FUNC_OFFSET(12393, glBeginFragmentShaderATI, glBeginFragmentShaderATI, NULL, _gloffset_BeginFragmentShaderATI), - NAME_FUNC_OFFSET(12418, glBindFragmentShaderATI, glBindFragmentShaderATI, NULL, _gloffset_BindFragmentShaderATI), - NAME_FUNC_OFFSET(12442, glColorFragmentOp1ATI, glColorFragmentOp1ATI, NULL, _gloffset_ColorFragmentOp1ATI), - NAME_FUNC_OFFSET(12464, glColorFragmentOp2ATI, glColorFragmentOp2ATI, NULL, _gloffset_ColorFragmentOp2ATI), - NAME_FUNC_OFFSET(12486, glColorFragmentOp3ATI, glColorFragmentOp3ATI, NULL, _gloffset_ColorFragmentOp3ATI), - NAME_FUNC_OFFSET(12508, glDeleteFragmentShaderATI, glDeleteFragmentShaderATI, NULL, _gloffset_DeleteFragmentShaderATI), - NAME_FUNC_OFFSET(12534, glEndFragmentShaderATI, glEndFragmentShaderATI, NULL, _gloffset_EndFragmentShaderATI), - NAME_FUNC_OFFSET(12557, glGenFragmentShadersATI, glGenFragmentShadersATI, NULL, _gloffset_GenFragmentShadersATI), - NAME_FUNC_OFFSET(12581, glPassTexCoordATI, glPassTexCoordATI, NULL, _gloffset_PassTexCoordATI), - NAME_FUNC_OFFSET(12599, glSampleMapATI, glSampleMapATI, NULL, _gloffset_SampleMapATI), - NAME_FUNC_OFFSET(12614, glSetFragmentShaderConstantATI, glSetFragmentShaderConstantATI, NULL, _gloffset_SetFragmentShaderConstantATI), - NAME_FUNC_OFFSET(12645, glPointParameteriNV, glPointParameteriNV, NULL, _gloffset_PointParameteriNV), - NAME_FUNC_OFFSET(12665, glPointParameterivNV, glPointParameterivNV, NULL, _gloffset_PointParameterivNV), - NAME_FUNC_OFFSET(12686, gl_dispatch_stub_737, gl_dispatch_stub_737, NULL, _gloffset_ActiveStencilFaceEXT), - NAME_FUNC_OFFSET(12709, gl_dispatch_stub_738, gl_dispatch_stub_738, NULL, _gloffset_BindVertexArrayAPPLE), - NAME_FUNC_OFFSET(12732, gl_dispatch_stub_739, gl_dispatch_stub_739, NULL, _gloffset_DeleteVertexArraysAPPLE), - NAME_FUNC_OFFSET(12758, gl_dispatch_stub_740, gl_dispatch_stub_740, NULL, _gloffset_GenVertexArraysAPPLE), - NAME_FUNC_OFFSET(12781, gl_dispatch_stub_741, gl_dispatch_stub_741, NULL, _gloffset_IsVertexArrayAPPLE), - NAME_FUNC_OFFSET(12802, glGetProgramNamedParameterdvNV, glGetProgramNamedParameterdvNV, NULL, _gloffset_GetProgramNamedParameterdvNV), - NAME_FUNC_OFFSET(12833, glGetProgramNamedParameterfvNV, glGetProgramNamedParameterfvNV, NULL, _gloffset_GetProgramNamedParameterfvNV), - NAME_FUNC_OFFSET(12864, glProgramNamedParameter4dNV, glProgramNamedParameter4dNV, NULL, _gloffset_ProgramNamedParameter4dNV), - NAME_FUNC_OFFSET(12892, glProgramNamedParameter4dvNV, glProgramNamedParameter4dvNV, NULL, _gloffset_ProgramNamedParameter4dvNV), - NAME_FUNC_OFFSET(12921, glProgramNamedParameter4fNV, glProgramNamedParameter4fNV, NULL, _gloffset_ProgramNamedParameter4fNV), - NAME_FUNC_OFFSET(12949, glProgramNamedParameter4fvNV, glProgramNamedParameter4fvNV, NULL, _gloffset_ProgramNamedParameter4fvNV), - NAME_FUNC_OFFSET(12978, gl_dispatch_stub_748, gl_dispatch_stub_748, NULL, _gloffset_DepthBoundsEXT), - NAME_FUNC_OFFSET(12995, gl_dispatch_stub_749, gl_dispatch_stub_749, NULL, _gloffset_BlendEquationSeparateEXT), - NAME_FUNC_OFFSET(13022, glBindFramebufferEXT, glBindFramebufferEXT, NULL, _gloffset_BindFramebufferEXT), - NAME_FUNC_OFFSET(13043, glBindRenderbufferEXT, glBindRenderbufferEXT, NULL, _gloffset_BindRenderbufferEXT), - NAME_FUNC_OFFSET(13065, glCheckFramebufferStatusEXT, glCheckFramebufferStatusEXT, NULL, _gloffset_CheckFramebufferStatusEXT), - NAME_FUNC_OFFSET(13093, glDeleteFramebuffersEXT, glDeleteFramebuffersEXT, NULL, _gloffset_DeleteFramebuffersEXT), - NAME_FUNC_OFFSET(13117, glDeleteRenderbuffersEXT, glDeleteRenderbuffersEXT, NULL, _gloffset_DeleteRenderbuffersEXT), - NAME_FUNC_OFFSET(13142, glFramebufferRenderbufferEXT, glFramebufferRenderbufferEXT, NULL, _gloffset_FramebufferRenderbufferEXT), - NAME_FUNC_OFFSET(13171, glFramebufferTexture1DEXT, glFramebufferTexture1DEXT, NULL, _gloffset_FramebufferTexture1DEXT), - NAME_FUNC_OFFSET(13197, glFramebufferTexture2DEXT, glFramebufferTexture2DEXT, NULL, _gloffset_FramebufferTexture2DEXT), - NAME_FUNC_OFFSET(13223, glFramebufferTexture3DEXT, glFramebufferTexture3DEXT, NULL, _gloffset_FramebufferTexture3DEXT), - NAME_FUNC_OFFSET(13249, glGenFramebuffersEXT, glGenFramebuffersEXT, NULL, _gloffset_GenFramebuffersEXT), - NAME_FUNC_OFFSET(13270, glGenRenderbuffersEXT, glGenRenderbuffersEXT, NULL, _gloffset_GenRenderbuffersEXT), - NAME_FUNC_OFFSET(13292, glGenerateMipmapEXT, glGenerateMipmapEXT, NULL, _gloffset_GenerateMipmapEXT), - NAME_FUNC_OFFSET(13312, glGetFramebufferAttachmentParameterivEXT, glGetFramebufferAttachmentParameterivEXT, NULL, _gloffset_GetFramebufferAttachmentParameterivEXT), - NAME_FUNC_OFFSET(13353, glGetRenderbufferParameterivEXT, glGetRenderbufferParameterivEXT, NULL, _gloffset_GetRenderbufferParameterivEXT), - NAME_FUNC_OFFSET(13385, glIsFramebufferEXT, glIsFramebufferEXT, NULL, _gloffset_IsFramebufferEXT), - NAME_FUNC_OFFSET(13404, glIsRenderbufferEXT, glIsRenderbufferEXT, NULL, _gloffset_IsRenderbufferEXT), - NAME_FUNC_OFFSET(13424, glRenderbufferStorageEXT, glRenderbufferStorageEXT, NULL, _gloffset_RenderbufferStorageEXT), - NAME_FUNC_OFFSET(13449, gl_dispatch_stub_767, gl_dispatch_stub_767, NULL, _gloffset_BlitFramebufferEXT), - NAME_FUNC_OFFSET(13470, glFramebufferTextureLayerEXT, glFramebufferTextureLayerEXT, NULL, _gloffset_FramebufferTextureLayerEXT), - NAME_FUNC_OFFSET(13499, gl_dispatch_stub_769, gl_dispatch_stub_769, NULL, _gloffset_StencilFuncSeparateATI), - NAME_FUNC_OFFSET(13524, gl_dispatch_stub_770, gl_dispatch_stub_770, NULL, _gloffset_ProgramEnvParameters4fvEXT), - NAME_FUNC_OFFSET(13553, gl_dispatch_stub_771, gl_dispatch_stub_771, NULL, _gloffset_ProgramLocalParameters4fvEXT), - NAME_FUNC_OFFSET(13584, gl_dispatch_stub_772, gl_dispatch_stub_772, NULL, _gloffset_GetQueryObjecti64vEXT), - NAME_FUNC_OFFSET(13608, gl_dispatch_stub_773, gl_dispatch_stub_773, NULL, _gloffset_GetQueryObjectui64vEXT), - NAME_FUNC_OFFSET(13633, glArrayElement, glArrayElement, NULL, _gloffset_ArrayElement), - NAME_FUNC_OFFSET(13651, glBindTexture, glBindTexture, NULL, _gloffset_BindTexture), - NAME_FUNC_OFFSET(13668, glDrawArrays, glDrawArrays, NULL, _gloffset_DrawArrays), - NAME_FUNC_OFFSET(13684, glAreTexturesResident, glAreTexturesResidentEXT, glAreTexturesResidentEXT, _gloffset_AreTexturesResident), - NAME_FUNC_OFFSET(13709, glCopyTexImage1D, glCopyTexImage1D, NULL, _gloffset_CopyTexImage1D), - NAME_FUNC_OFFSET(13729, glCopyTexImage2D, glCopyTexImage2D, NULL, _gloffset_CopyTexImage2D), - NAME_FUNC_OFFSET(13749, glCopyTexSubImage1D, glCopyTexSubImage1D, NULL, _gloffset_CopyTexSubImage1D), - NAME_FUNC_OFFSET(13772, glCopyTexSubImage2D, glCopyTexSubImage2D, NULL, _gloffset_CopyTexSubImage2D), - NAME_FUNC_OFFSET(13795, glDeleteTextures, glDeleteTexturesEXT, glDeleteTexturesEXT, _gloffset_DeleteTextures), - NAME_FUNC_OFFSET(13815, glGenTextures, glGenTexturesEXT, glGenTexturesEXT, _gloffset_GenTextures), - NAME_FUNC_OFFSET(13832, glGetPointerv, glGetPointerv, NULL, _gloffset_GetPointerv), - NAME_FUNC_OFFSET(13849, glIsTexture, glIsTextureEXT, glIsTextureEXT, _gloffset_IsTexture), - NAME_FUNC_OFFSET(13864, glPrioritizeTextures, glPrioritizeTextures, NULL, _gloffset_PrioritizeTextures), - NAME_FUNC_OFFSET(13888, glTexSubImage1D, glTexSubImage1D, NULL, _gloffset_TexSubImage1D), - NAME_FUNC_OFFSET(13907, glTexSubImage2D, glTexSubImage2D, NULL, _gloffset_TexSubImage2D), - NAME_FUNC_OFFSET(13926, glBlendColor, glBlendColor, NULL, _gloffset_BlendColor), - NAME_FUNC_OFFSET(13942, glBlendEquation, glBlendEquation, NULL, _gloffset_BlendEquation), - NAME_FUNC_OFFSET(13961, glDrawRangeElements, glDrawRangeElements, NULL, _gloffset_DrawRangeElements), - NAME_FUNC_OFFSET(13984, glColorTable, glColorTable, NULL, _gloffset_ColorTable), - NAME_FUNC_OFFSET(14000, glColorTable, glColorTable, NULL, _gloffset_ColorTable), - NAME_FUNC_OFFSET(14016, glColorTableParameterfv, glColorTableParameterfv, NULL, _gloffset_ColorTableParameterfv), - NAME_FUNC_OFFSET(14043, glColorTableParameteriv, glColorTableParameteriv, NULL, _gloffset_ColorTableParameteriv), - NAME_FUNC_OFFSET(14070, glCopyColorTable, glCopyColorTable, NULL, _gloffset_CopyColorTable), - NAME_FUNC_OFFSET(14090, glGetColorTable, glGetColorTableEXT, glGetColorTableEXT, _gloffset_GetColorTable), - NAME_FUNC_OFFSET(14109, glGetColorTable, glGetColorTableEXT, glGetColorTableEXT, _gloffset_GetColorTable), - NAME_FUNC_OFFSET(14128, glGetColorTableParameterfv, glGetColorTableParameterfvEXT, glGetColorTableParameterfvEXT, _gloffset_GetColorTableParameterfv), - NAME_FUNC_OFFSET(14158, glGetColorTableParameterfv, glGetColorTableParameterfvEXT, glGetColorTableParameterfvEXT, _gloffset_GetColorTableParameterfv), - NAME_FUNC_OFFSET(14188, glGetColorTableParameteriv, glGetColorTableParameterivEXT, glGetColorTableParameterivEXT, _gloffset_GetColorTableParameteriv), - NAME_FUNC_OFFSET(14218, glGetColorTableParameteriv, glGetColorTableParameterivEXT, glGetColorTableParameterivEXT, _gloffset_GetColorTableParameteriv), - NAME_FUNC_OFFSET(14248, glColorSubTable, glColorSubTable, NULL, _gloffset_ColorSubTable), - NAME_FUNC_OFFSET(14267, glCopyColorSubTable, glCopyColorSubTable, NULL, _gloffset_CopyColorSubTable), - NAME_FUNC_OFFSET(14290, glConvolutionFilter1D, glConvolutionFilter1D, NULL, _gloffset_ConvolutionFilter1D), - NAME_FUNC_OFFSET(14315, glConvolutionFilter2D, glConvolutionFilter2D, NULL, _gloffset_ConvolutionFilter2D), - NAME_FUNC_OFFSET(14340, glConvolutionParameterf, glConvolutionParameterf, NULL, _gloffset_ConvolutionParameterf), - NAME_FUNC_OFFSET(14367, glConvolutionParameterfv, glConvolutionParameterfv, NULL, _gloffset_ConvolutionParameterfv), - NAME_FUNC_OFFSET(14395, glConvolutionParameteri, glConvolutionParameteri, NULL, _gloffset_ConvolutionParameteri), - NAME_FUNC_OFFSET(14422, glConvolutionParameteriv, glConvolutionParameteriv, NULL, _gloffset_ConvolutionParameteriv), - NAME_FUNC_OFFSET(14450, glCopyConvolutionFilter1D, glCopyConvolutionFilter1D, NULL, _gloffset_CopyConvolutionFilter1D), - NAME_FUNC_OFFSET(14479, glCopyConvolutionFilter2D, glCopyConvolutionFilter2D, NULL, _gloffset_CopyConvolutionFilter2D), - NAME_FUNC_OFFSET(14508, glGetConvolutionFilter, gl_dispatch_stub_356, gl_dispatch_stub_356, _gloffset_GetConvolutionFilter), - NAME_FUNC_OFFSET(14534, glGetConvolutionParameterfv, gl_dispatch_stub_357, gl_dispatch_stub_357, _gloffset_GetConvolutionParameterfv), - NAME_FUNC_OFFSET(14565, glGetConvolutionParameteriv, gl_dispatch_stub_358, gl_dispatch_stub_358, _gloffset_GetConvolutionParameteriv), - NAME_FUNC_OFFSET(14596, glGetSeparableFilter, gl_dispatch_stub_359, gl_dispatch_stub_359, _gloffset_GetSeparableFilter), - NAME_FUNC_OFFSET(14620, glSeparableFilter2D, glSeparableFilter2D, NULL, _gloffset_SeparableFilter2D), - NAME_FUNC_OFFSET(14643, glGetHistogram, gl_dispatch_stub_361, gl_dispatch_stub_361, _gloffset_GetHistogram), - NAME_FUNC_OFFSET(14661, glGetHistogramParameterfv, gl_dispatch_stub_362, gl_dispatch_stub_362, _gloffset_GetHistogramParameterfv), - NAME_FUNC_OFFSET(14690, glGetHistogramParameteriv, gl_dispatch_stub_363, gl_dispatch_stub_363, _gloffset_GetHistogramParameteriv), - NAME_FUNC_OFFSET(14719, glGetMinmax, gl_dispatch_stub_364, gl_dispatch_stub_364, _gloffset_GetMinmax), - NAME_FUNC_OFFSET(14734, glGetMinmaxParameterfv, gl_dispatch_stub_365, gl_dispatch_stub_365, _gloffset_GetMinmaxParameterfv), - NAME_FUNC_OFFSET(14760, glGetMinmaxParameteriv, gl_dispatch_stub_366, gl_dispatch_stub_366, _gloffset_GetMinmaxParameteriv), - NAME_FUNC_OFFSET(14786, glHistogram, glHistogram, NULL, _gloffset_Histogram), - NAME_FUNC_OFFSET(14801, glMinmax, glMinmax, NULL, _gloffset_Minmax), - NAME_FUNC_OFFSET(14813, glResetHistogram, glResetHistogram, NULL, _gloffset_ResetHistogram), - NAME_FUNC_OFFSET(14833, glResetMinmax, glResetMinmax, NULL, _gloffset_ResetMinmax), - NAME_FUNC_OFFSET(14850, glTexImage3D, glTexImage3D, NULL, _gloffset_TexImage3D), - NAME_FUNC_OFFSET(14866, glTexSubImage3D, glTexSubImage3D, NULL, _gloffset_TexSubImage3D), - NAME_FUNC_OFFSET(14885, glCopyTexSubImage3D, glCopyTexSubImage3D, NULL, _gloffset_CopyTexSubImage3D), - NAME_FUNC_OFFSET(14908, glActiveTextureARB, glActiveTextureARB, NULL, _gloffset_ActiveTextureARB), - NAME_FUNC_OFFSET(14924, glClientActiveTextureARB, glClientActiveTextureARB, NULL, _gloffset_ClientActiveTextureARB), - NAME_FUNC_OFFSET(14946, glMultiTexCoord1dARB, glMultiTexCoord1dARB, NULL, _gloffset_MultiTexCoord1dARB), - NAME_FUNC_OFFSET(14964, glMultiTexCoord1dvARB, glMultiTexCoord1dvARB, NULL, _gloffset_MultiTexCoord1dvARB), - NAME_FUNC_OFFSET(14983, glMultiTexCoord1fARB, glMultiTexCoord1fARB, NULL, _gloffset_MultiTexCoord1fARB), - NAME_FUNC_OFFSET(15001, glMultiTexCoord1fvARB, glMultiTexCoord1fvARB, NULL, _gloffset_MultiTexCoord1fvARB), - NAME_FUNC_OFFSET(15020, glMultiTexCoord1iARB, glMultiTexCoord1iARB, NULL, _gloffset_MultiTexCoord1iARB), - NAME_FUNC_OFFSET(15038, glMultiTexCoord1ivARB, glMultiTexCoord1ivARB, NULL, _gloffset_MultiTexCoord1ivARB), - NAME_FUNC_OFFSET(15057, glMultiTexCoord1sARB, glMultiTexCoord1sARB, NULL, _gloffset_MultiTexCoord1sARB), - NAME_FUNC_OFFSET(15075, glMultiTexCoord1svARB, glMultiTexCoord1svARB, NULL, _gloffset_MultiTexCoord1svARB), - NAME_FUNC_OFFSET(15094, glMultiTexCoord2dARB, glMultiTexCoord2dARB, NULL, _gloffset_MultiTexCoord2dARB), - NAME_FUNC_OFFSET(15112, glMultiTexCoord2dvARB, glMultiTexCoord2dvARB, NULL, _gloffset_MultiTexCoord2dvARB), - NAME_FUNC_OFFSET(15131, glMultiTexCoord2fARB, glMultiTexCoord2fARB, NULL, _gloffset_MultiTexCoord2fARB), - NAME_FUNC_OFFSET(15149, glMultiTexCoord2fvARB, glMultiTexCoord2fvARB, NULL, _gloffset_MultiTexCoord2fvARB), - NAME_FUNC_OFFSET(15168, glMultiTexCoord2iARB, glMultiTexCoord2iARB, NULL, _gloffset_MultiTexCoord2iARB), - NAME_FUNC_OFFSET(15186, glMultiTexCoord2ivARB, glMultiTexCoord2ivARB, NULL, _gloffset_MultiTexCoord2ivARB), - NAME_FUNC_OFFSET(15205, glMultiTexCoord2sARB, glMultiTexCoord2sARB, NULL, _gloffset_MultiTexCoord2sARB), - NAME_FUNC_OFFSET(15223, glMultiTexCoord2svARB, glMultiTexCoord2svARB, NULL, _gloffset_MultiTexCoord2svARB), - NAME_FUNC_OFFSET(15242, glMultiTexCoord3dARB, glMultiTexCoord3dARB, NULL, _gloffset_MultiTexCoord3dARB), - NAME_FUNC_OFFSET(15260, glMultiTexCoord3dvARB, glMultiTexCoord3dvARB, NULL, _gloffset_MultiTexCoord3dvARB), - NAME_FUNC_OFFSET(15279, glMultiTexCoord3fARB, glMultiTexCoord3fARB, NULL, _gloffset_MultiTexCoord3fARB), - NAME_FUNC_OFFSET(15297, glMultiTexCoord3fvARB, glMultiTexCoord3fvARB, NULL, _gloffset_MultiTexCoord3fvARB), - NAME_FUNC_OFFSET(15316, glMultiTexCoord3iARB, glMultiTexCoord3iARB, NULL, _gloffset_MultiTexCoord3iARB), - NAME_FUNC_OFFSET(15334, glMultiTexCoord3ivARB, glMultiTexCoord3ivARB, NULL, _gloffset_MultiTexCoord3ivARB), - NAME_FUNC_OFFSET(15353, glMultiTexCoord3sARB, glMultiTexCoord3sARB, NULL, _gloffset_MultiTexCoord3sARB), - NAME_FUNC_OFFSET(15371, glMultiTexCoord3svARB, glMultiTexCoord3svARB, NULL, _gloffset_MultiTexCoord3svARB), - NAME_FUNC_OFFSET(15390, glMultiTexCoord4dARB, glMultiTexCoord4dARB, NULL, _gloffset_MultiTexCoord4dARB), - NAME_FUNC_OFFSET(15408, glMultiTexCoord4dvARB, glMultiTexCoord4dvARB, NULL, _gloffset_MultiTexCoord4dvARB), - NAME_FUNC_OFFSET(15427, glMultiTexCoord4fARB, glMultiTexCoord4fARB, NULL, _gloffset_MultiTexCoord4fARB), - NAME_FUNC_OFFSET(15445, glMultiTexCoord4fvARB, glMultiTexCoord4fvARB, NULL, _gloffset_MultiTexCoord4fvARB), - NAME_FUNC_OFFSET(15464, glMultiTexCoord4iARB, glMultiTexCoord4iARB, NULL, _gloffset_MultiTexCoord4iARB), - NAME_FUNC_OFFSET(15482, glMultiTexCoord4ivARB, glMultiTexCoord4ivARB, NULL, _gloffset_MultiTexCoord4ivARB), - NAME_FUNC_OFFSET(15501, glMultiTexCoord4sARB, glMultiTexCoord4sARB, NULL, _gloffset_MultiTexCoord4sARB), - NAME_FUNC_OFFSET(15519, glMultiTexCoord4svARB, glMultiTexCoord4svARB, NULL, _gloffset_MultiTexCoord4svARB), - NAME_FUNC_OFFSET(15538, glStencilOpSeparate, glStencilOpSeparate, NULL, _gloffset_StencilOpSeparate), - NAME_FUNC_OFFSET(15561, glLoadTransposeMatrixdARB, glLoadTransposeMatrixdARB, NULL, _gloffset_LoadTransposeMatrixdARB), - NAME_FUNC_OFFSET(15584, glLoadTransposeMatrixfARB, glLoadTransposeMatrixfARB, NULL, _gloffset_LoadTransposeMatrixfARB), - NAME_FUNC_OFFSET(15607, glMultTransposeMatrixdARB, glMultTransposeMatrixdARB, NULL, _gloffset_MultTransposeMatrixdARB), - NAME_FUNC_OFFSET(15630, glMultTransposeMatrixfARB, glMultTransposeMatrixfARB, NULL, _gloffset_MultTransposeMatrixfARB), - NAME_FUNC_OFFSET(15653, glSampleCoverageARB, glSampleCoverageARB, NULL, _gloffset_SampleCoverageARB), - NAME_FUNC_OFFSET(15670, glCompressedTexImage1DARB, glCompressedTexImage1DARB, NULL, _gloffset_CompressedTexImage1DARB), - NAME_FUNC_OFFSET(15693, glCompressedTexImage2DARB, glCompressedTexImage2DARB, NULL, _gloffset_CompressedTexImage2DARB), - NAME_FUNC_OFFSET(15716, glCompressedTexImage3DARB, glCompressedTexImage3DARB, NULL, _gloffset_CompressedTexImage3DARB), - NAME_FUNC_OFFSET(15739, glCompressedTexSubImage1DARB, glCompressedTexSubImage1DARB, NULL, _gloffset_CompressedTexSubImage1DARB), - NAME_FUNC_OFFSET(15765, glCompressedTexSubImage2DARB, glCompressedTexSubImage2DARB, NULL, _gloffset_CompressedTexSubImage2DARB), - NAME_FUNC_OFFSET(15791, glCompressedTexSubImage3DARB, glCompressedTexSubImage3DARB, NULL, _gloffset_CompressedTexSubImage3DARB), - NAME_FUNC_OFFSET(15817, glGetCompressedTexImageARB, glGetCompressedTexImageARB, NULL, _gloffset_GetCompressedTexImageARB), - NAME_FUNC_OFFSET(15841, glDisableVertexAttribArrayARB, glDisableVertexAttribArrayARB, NULL, _gloffset_DisableVertexAttribArrayARB), - NAME_FUNC_OFFSET(15868, glEnableVertexAttribArrayARB, glEnableVertexAttribArrayARB, NULL, _gloffset_EnableVertexAttribArrayARB), - NAME_FUNC_OFFSET(15894, glGetVertexAttribdvARB, glGetVertexAttribdvARB, NULL, _gloffset_GetVertexAttribdvARB), - NAME_FUNC_OFFSET(15914, glGetVertexAttribfvARB, glGetVertexAttribfvARB, NULL, _gloffset_GetVertexAttribfvARB), - NAME_FUNC_OFFSET(15934, glGetVertexAttribivARB, glGetVertexAttribivARB, NULL, _gloffset_GetVertexAttribivARB), + NAME_FUNC_OFFSET(11332, glProgramParameters4dvNV, glProgramParameters4dvNV, NULL, _gloffset_ProgramParameters4dvNV), + NAME_FUNC_OFFSET(11357, glProgramParameters4fvNV, glProgramParameters4fvNV, NULL, _gloffset_ProgramParameters4fvNV), + NAME_FUNC_OFFSET(11382, glRequestResidentProgramsNV, glRequestResidentProgramsNV, NULL, _gloffset_RequestResidentProgramsNV), + NAME_FUNC_OFFSET(11410, glTrackMatrixNV, glTrackMatrixNV, NULL, _gloffset_TrackMatrixNV), + NAME_FUNC_OFFSET(11426, glVertexAttrib1dNV, glVertexAttrib1dNV, NULL, _gloffset_VertexAttrib1dNV), + NAME_FUNC_OFFSET(11445, glVertexAttrib1dvNV, glVertexAttrib1dvNV, NULL, _gloffset_VertexAttrib1dvNV), + NAME_FUNC_OFFSET(11465, glVertexAttrib1fNV, glVertexAttrib1fNV, NULL, _gloffset_VertexAttrib1fNV), + NAME_FUNC_OFFSET(11484, glVertexAttrib1fvNV, glVertexAttrib1fvNV, NULL, _gloffset_VertexAttrib1fvNV), + NAME_FUNC_OFFSET(11504, glVertexAttrib1sNV, glVertexAttrib1sNV, NULL, _gloffset_VertexAttrib1sNV), + NAME_FUNC_OFFSET(11523, glVertexAttrib1svNV, glVertexAttrib1svNV, NULL, _gloffset_VertexAttrib1svNV), + NAME_FUNC_OFFSET(11543, glVertexAttrib2dNV, glVertexAttrib2dNV, NULL, _gloffset_VertexAttrib2dNV), + NAME_FUNC_OFFSET(11562, glVertexAttrib2dvNV, glVertexAttrib2dvNV, NULL, _gloffset_VertexAttrib2dvNV), + NAME_FUNC_OFFSET(11582, glVertexAttrib2fNV, glVertexAttrib2fNV, NULL, _gloffset_VertexAttrib2fNV), + NAME_FUNC_OFFSET(11601, glVertexAttrib2fvNV, glVertexAttrib2fvNV, NULL, _gloffset_VertexAttrib2fvNV), + NAME_FUNC_OFFSET(11621, glVertexAttrib2sNV, glVertexAttrib2sNV, NULL, _gloffset_VertexAttrib2sNV), + NAME_FUNC_OFFSET(11640, glVertexAttrib2svNV, glVertexAttrib2svNV, NULL, _gloffset_VertexAttrib2svNV), + NAME_FUNC_OFFSET(11660, glVertexAttrib3dNV, glVertexAttrib3dNV, NULL, _gloffset_VertexAttrib3dNV), + NAME_FUNC_OFFSET(11679, glVertexAttrib3dvNV, glVertexAttrib3dvNV, NULL, _gloffset_VertexAttrib3dvNV), + NAME_FUNC_OFFSET(11699, glVertexAttrib3fNV, glVertexAttrib3fNV, NULL, _gloffset_VertexAttrib3fNV), + NAME_FUNC_OFFSET(11718, glVertexAttrib3fvNV, glVertexAttrib3fvNV, NULL, _gloffset_VertexAttrib3fvNV), + NAME_FUNC_OFFSET(11738, glVertexAttrib3sNV, glVertexAttrib3sNV, NULL, _gloffset_VertexAttrib3sNV), + NAME_FUNC_OFFSET(11757, glVertexAttrib3svNV, glVertexAttrib3svNV, NULL, _gloffset_VertexAttrib3svNV), + NAME_FUNC_OFFSET(11777, glVertexAttrib4dNV, glVertexAttrib4dNV, NULL, _gloffset_VertexAttrib4dNV), + NAME_FUNC_OFFSET(11796, glVertexAttrib4dvNV, glVertexAttrib4dvNV, NULL, _gloffset_VertexAttrib4dvNV), + NAME_FUNC_OFFSET(11816, glVertexAttrib4fNV, glVertexAttrib4fNV, NULL, _gloffset_VertexAttrib4fNV), + NAME_FUNC_OFFSET(11835, glVertexAttrib4fvNV, glVertexAttrib4fvNV, NULL, _gloffset_VertexAttrib4fvNV), + NAME_FUNC_OFFSET(11855, glVertexAttrib4sNV, glVertexAttrib4sNV, NULL, _gloffset_VertexAttrib4sNV), + NAME_FUNC_OFFSET(11874, glVertexAttrib4svNV, glVertexAttrib4svNV, NULL, _gloffset_VertexAttrib4svNV), + NAME_FUNC_OFFSET(11894, glVertexAttrib4ubNV, glVertexAttrib4ubNV, NULL, _gloffset_VertexAttrib4ubNV), + NAME_FUNC_OFFSET(11914, glVertexAttrib4ubvNV, glVertexAttrib4ubvNV, NULL, _gloffset_VertexAttrib4ubvNV), + NAME_FUNC_OFFSET(11935, glVertexAttribPointerNV, glVertexAttribPointerNV, NULL, _gloffset_VertexAttribPointerNV), + NAME_FUNC_OFFSET(11959, glVertexAttribs1dvNV, glVertexAttribs1dvNV, NULL, _gloffset_VertexAttribs1dvNV), + NAME_FUNC_OFFSET(11980, glVertexAttribs1fvNV, glVertexAttribs1fvNV, NULL, _gloffset_VertexAttribs1fvNV), + NAME_FUNC_OFFSET(12001, glVertexAttribs1svNV, glVertexAttribs1svNV, NULL, _gloffset_VertexAttribs1svNV), + NAME_FUNC_OFFSET(12022, glVertexAttribs2dvNV, glVertexAttribs2dvNV, NULL, _gloffset_VertexAttribs2dvNV), + NAME_FUNC_OFFSET(12043, glVertexAttribs2fvNV, glVertexAttribs2fvNV, NULL, _gloffset_VertexAttribs2fvNV), + NAME_FUNC_OFFSET(12064, glVertexAttribs2svNV, glVertexAttribs2svNV, NULL, _gloffset_VertexAttribs2svNV), + NAME_FUNC_OFFSET(12085, glVertexAttribs3dvNV, glVertexAttribs3dvNV, NULL, _gloffset_VertexAttribs3dvNV), + NAME_FUNC_OFFSET(12106, glVertexAttribs3fvNV, glVertexAttribs3fvNV, NULL, _gloffset_VertexAttribs3fvNV), + NAME_FUNC_OFFSET(12127, glVertexAttribs3svNV, glVertexAttribs3svNV, NULL, _gloffset_VertexAttribs3svNV), + NAME_FUNC_OFFSET(12148, glVertexAttribs4dvNV, glVertexAttribs4dvNV, NULL, _gloffset_VertexAttribs4dvNV), + NAME_FUNC_OFFSET(12169, glVertexAttribs4fvNV, glVertexAttribs4fvNV, NULL, _gloffset_VertexAttribs4fvNV), + NAME_FUNC_OFFSET(12190, glVertexAttribs4svNV, glVertexAttribs4svNV, NULL, _gloffset_VertexAttribs4svNV), + NAME_FUNC_OFFSET(12211, glVertexAttribs4ubvNV, glVertexAttribs4ubvNV, NULL, _gloffset_VertexAttribs4ubvNV), + NAME_FUNC_OFFSET(12233, glAlphaFragmentOp1ATI, glAlphaFragmentOp1ATI, NULL, _gloffset_AlphaFragmentOp1ATI), + NAME_FUNC_OFFSET(12255, glAlphaFragmentOp2ATI, glAlphaFragmentOp2ATI, NULL, _gloffset_AlphaFragmentOp2ATI), + NAME_FUNC_OFFSET(12277, glAlphaFragmentOp3ATI, glAlphaFragmentOp3ATI, NULL, _gloffset_AlphaFragmentOp3ATI), + NAME_FUNC_OFFSET(12299, glBeginFragmentShaderATI, glBeginFragmentShaderATI, NULL, _gloffset_BeginFragmentShaderATI), + NAME_FUNC_OFFSET(12324, glBindFragmentShaderATI, glBindFragmentShaderATI, NULL, _gloffset_BindFragmentShaderATI), + NAME_FUNC_OFFSET(12348, glColorFragmentOp1ATI, glColorFragmentOp1ATI, NULL, _gloffset_ColorFragmentOp1ATI), + NAME_FUNC_OFFSET(12370, glColorFragmentOp2ATI, glColorFragmentOp2ATI, NULL, _gloffset_ColorFragmentOp2ATI), + NAME_FUNC_OFFSET(12392, glColorFragmentOp3ATI, glColorFragmentOp3ATI, NULL, _gloffset_ColorFragmentOp3ATI), + NAME_FUNC_OFFSET(12414, glDeleteFragmentShaderATI, glDeleteFragmentShaderATI, NULL, _gloffset_DeleteFragmentShaderATI), + NAME_FUNC_OFFSET(12440, glEndFragmentShaderATI, glEndFragmentShaderATI, NULL, _gloffset_EndFragmentShaderATI), + NAME_FUNC_OFFSET(12463, glGenFragmentShadersATI, glGenFragmentShadersATI, NULL, _gloffset_GenFragmentShadersATI), + NAME_FUNC_OFFSET(12487, glPassTexCoordATI, glPassTexCoordATI, NULL, _gloffset_PassTexCoordATI), + NAME_FUNC_OFFSET(12505, glSampleMapATI, glSampleMapATI, NULL, _gloffset_SampleMapATI), + NAME_FUNC_OFFSET(12520, glSetFragmentShaderConstantATI, glSetFragmentShaderConstantATI, NULL, _gloffset_SetFragmentShaderConstantATI), + NAME_FUNC_OFFSET(12551, glPointParameteriNV, glPointParameteriNV, NULL, _gloffset_PointParameteriNV), + NAME_FUNC_OFFSET(12571, glPointParameterivNV, glPointParameterivNV, NULL, _gloffset_PointParameterivNV), + NAME_FUNC_OFFSET(12592, gl_dispatch_stub_733, gl_dispatch_stub_733, NULL, _gloffset_ActiveStencilFaceEXT), + NAME_FUNC_OFFSET(12615, gl_dispatch_stub_734, gl_dispatch_stub_734, NULL, _gloffset_BindVertexArrayAPPLE), + NAME_FUNC_OFFSET(12638, gl_dispatch_stub_735, gl_dispatch_stub_735, NULL, _gloffset_DeleteVertexArraysAPPLE), + NAME_FUNC_OFFSET(12664, gl_dispatch_stub_736, gl_dispatch_stub_736, NULL, _gloffset_GenVertexArraysAPPLE), + NAME_FUNC_OFFSET(12687, gl_dispatch_stub_737, gl_dispatch_stub_737, NULL, _gloffset_IsVertexArrayAPPLE), + NAME_FUNC_OFFSET(12708, glGetProgramNamedParameterdvNV, glGetProgramNamedParameterdvNV, NULL, _gloffset_GetProgramNamedParameterdvNV), + NAME_FUNC_OFFSET(12739, glGetProgramNamedParameterfvNV, glGetProgramNamedParameterfvNV, NULL, _gloffset_GetProgramNamedParameterfvNV), + NAME_FUNC_OFFSET(12770, glProgramNamedParameter4dNV, glProgramNamedParameter4dNV, NULL, _gloffset_ProgramNamedParameter4dNV), + NAME_FUNC_OFFSET(12798, glProgramNamedParameter4dvNV, glProgramNamedParameter4dvNV, NULL, _gloffset_ProgramNamedParameter4dvNV), + NAME_FUNC_OFFSET(12827, glProgramNamedParameter4fNV, glProgramNamedParameter4fNV, NULL, _gloffset_ProgramNamedParameter4fNV), + NAME_FUNC_OFFSET(12855, glProgramNamedParameter4fvNV, glProgramNamedParameter4fvNV, NULL, _gloffset_ProgramNamedParameter4fvNV), + NAME_FUNC_OFFSET(12884, gl_dispatch_stub_744, gl_dispatch_stub_744, NULL, _gloffset_DepthBoundsEXT), + NAME_FUNC_OFFSET(12901, gl_dispatch_stub_745, gl_dispatch_stub_745, NULL, _gloffset_BlendEquationSeparateEXT), + NAME_FUNC_OFFSET(12928, glBindFramebufferEXT, glBindFramebufferEXT, NULL, _gloffset_BindFramebufferEXT), + NAME_FUNC_OFFSET(12949, glBindRenderbufferEXT, glBindRenderbufferEXT, NULL, _gloffset_BindRenderbufferEXT), + NAME_FUNC_OFFSET(12971, glCheckFramebufferStatusEXT, glCheckFramebufferStatusEXT, NULL, _gloffset_CheckFramebufferStatusEXT), + NAME_FUNC_OFFSET(12999, glDeleteFramebuffersEXT, glDeleteFramebuffersEXT, NULL, _gloffset_DeleteFramebuffersEXT), + NAME_FUNC_OFFSET(13023, glDeleteRenderbuffersEXT, glDeleteRenderbuffersEXT, NULL, _gloffset_DeleteRenderbuffersEXT), + NAME_FUNC_OFFSET(13048, glFramebufferRenderbufferEXT, glFramebufferRenderbufferEXT, NULL, _gloffset_FramebufferRenderbufferEXT), + NAME_FUNC_OFFSET(13077, glFramebufferTexture1DEXT, glFramebufferTexture1DEXT, NULL, _gloffset_FramebufferTexture1DEXT), + NAME_FUNC_OFFSET(13103, glFramebufferTexture2DEXT, glFramebufferTexture2DEXT, NULL, _gloffset_FramebufferTexture2DEXT), + NAME_FUNC_OFFSET(13129, glFramebufferTexture3DEXT, glFramebufferTexture3DEXT, NULL, _gloffset_FramebufferTexture3DEXT), + NAME_FUNC_OFFSET(13155, glGenFramebuffersEXT, glGenFramebuffersEXT, NULL, _gloffset_GenFramebuffersEXT), + NAME_FUNC_OFFSET(13176, glGenRenderbuffersEXT, glGenRenderbuffersEXT, NULL, _gloffset_GenRenderbuffersEXT), + NAME_FUNC_OFFSET(13198, glGenerateMipmapEXT, glGenerateMipmapEXT, NULL, _gloffset_GenerateMipmapEXT), + NAME_FUNC_OFFSET(13218, glGetFramebufferAttachmentParameterivEXT, glGetFramebufferAttachmentParameterivEXT, NULL, _gloffset_GetFramebufferAttachmentParameterivEXT), + NAME_FUNC_OFFSET(13259, glGetRenderbufferParameterivEXT, glGetRenderbufferParameterivEXT, NULL, _gloffset_GetRenderbufferParameterivEXT), + NAME_FUNC_OFFSET(13291, glIsFramebufferEXT, glIsFramebufferEXT, NULL, _gloffset_IsFramebufferEXT), + NAME_FUNC_OFFSET(13310, glIsRenderbufferEXT, glIsRenderbufferEXT, NULL, _gloffset_IsRenderbufferEXT), + NAME_FUNC_OFFSET(13330, glRenderbufferStorageEXT, glRenderbufferStorageEXT, NULL, _gloffset_RenderbufferStorageEXT), + NAME_FUNC_OFFSET(13355, gl_dispatch_stub_763, gl_dispatch_stub_763, NULL, _gloffset_BlitFramebufferEXT), + NAME_FUNC_OFFSET(13376, glFramebufferTextureLayerEXT, glFramebufferTextureLayerEXT, NULL, _gloffset_FramebufferTextureLayerEXT), + NAME_FUNC_OFFSET(13405, gl_dispatch_stub_765, gl_dispatch_stub_765, NULL, _gloffset_StencilFuncSeparateATI), + NAME_FUNC_OFFSET(13430, gl_dispatch_stub_766, gl_dispatch_stub_766, NULL, _gloffset_ProgramEnvParameters4fvEXT), + NAME_FUNC_OFFSET(13459, gl_dispatch_stub_767, gl_dispatch_stub_767, NULL, _gloffset_ProgramLocalParameters4fvEXT), + NAME_FUNC_OFFSET(13490, gl_dispatch_stub_768, gl_dispatch_stub_768, NULL, _gloffset_GetQueryObjecti64vEXT), + NAME_FUNC_OFFSET(13514, gl_dispatch_stub_769, gl_dispatch_stub_769, NULL, _gloffset_GetQueryObjectui64vEXT), + NAME_FUNC_OFFSET(13539, glArrayElement, glArrayElement, NULL, _gloffset_ArrayElement), + NAME_FUNC_OFFSET(13557, glBindTexture, glBindTexture, NULL, _gloffset_BindTexture), + NAME_FUNC_OFFSET(13574, glDrawArrays, glDrawArrays, NULL, _gloffset_DrawArrays), + NAME_FUNC_OFFSET(13590, glAreTexturesResident, glAreTexturesResidentEXT, glAreTexturesResidentEXT, _gloffset_AreTexturesResident), + NAME_FUNC_OFFSET(13615, glCopyTexImage1D, glCopyTexImage1D, NULL, _gloffset_CopyTexImage1D), + NAME_FUNC_OFFSET(13635, glCopyTexImage2D, glCopyTexImage2D, NULL, _gloffset_CopyTexImage2D), + NAME_FUNC_OFFSET(13655, glCopyTexSubImage1D, glCopyTexSubImage1D, NULL, _gloffset_CopyTexSubImage1D), + NAME_FUNC_OFFSET(13678, glCopyTexSubImage2D, glCopyTexSubImage2D, NULL, _gloffset_CopyTexSubImage2D), + NAME_FUNC_OFFSET(13701, glDeleteTextures, glDeleteTexturesEXT, glDeleteTexturesEXT, _gloffset_DeleteTextures), + NAME_FUNC_OFFSET(13721, glGenTextures, glGenTexturesEXT, glGenTexturesEXT, _gloffset_GenTextures), + NAME_FUNC_OFFSET(13738, glGetPointerv, glGetPointerv, NULL, _gloffset_GetPointerv), + NAME_FUNC_OFFSET(13755, glIsTexture, glIsTextureEXT, glIsTextureEXT, _gloffset_IsTexture), + NAME_FUNC_OFFSET(13770, glPrioritizeTextures, glPrioritizeTextures, NULL, _gloffset_PrioritizeTextures), + NAME_FUNC_OFFSET(13794, glTexSubImage1D, glTexSubImage1D, NULL, _gloffset_TexSubImage1D), + NAME_FUNC_OFFSET(13813, glTexSubImage2D, glTexSubImage2D, NULL, _gloffset_TexSubImage2D), + NAME_FUNC_OFFSET(13832, glBlendColor, glBlendColor, NULL, _gloffset_BlendColor), + NAME_FUNC_OFFSET(13848, glBlendEquation, glBlendEquation, NULL, _gloffset_BlendEquation), + NAME_FUNC_OFFSET(13867, glDrawRangeElements, glDrawRangeElements, NULL, _gloffset_DrawRangeElements), + NAME_FUNC_OFFSET(13890, glColorTable, glColorTable, NULL, _gloffset_ColorTable), + NAME_FUNC_OFFSET(13906, glColorTable, glColorTable, NULL, _gloffset_ColorTable), + NAME_FUNC_OFFSET(13922, glColorTableParameterfv, glColorTableParameterfv, NULL, _gloffset_ColorTableParameterfv), + NAME_FUNC_OFFSET(13949, glColorTableParameteriv, glColorTableParameteriv, NULL, _gloffset_ColorTableParameteriv), + NAME_FUNC_OFFSET(13976, glCopyColorTable, glCopyColorTable, NULL, _gloffset_CopyColorTable), + NAME_FUNC_OFFSET(13996, glGetColorTable, glGetColorTableEXT, glGetColorTableEXT, _gloffset_GetColorTable), + NAME_FUNC_OFFSET(14015, glGetColorTable, glGetColorTableEXT, glGetColorTableEXT, _gloffset_GetColorTable), + NAME_FUNC_OFFSET(14034, glGetColorTableParameterfv, glGetColorTableParameterfvEXT, glGetColorTableParameterfvEXT, _gloffset_GetColorTableParameterfv), + NAME_FUNC_OFFSET(14064, glGetColorTableParameterfv, glGetColorTableParameterfvEXT, glGetColorTableParameterfvEXT, _gloffset_GetColorTableParameterfv), + NAME_FUNC_OFFSET(14094, glGetColorTableParameteriv, glGetColorTableParameterivEXT, glGetColorTableParameterivEXT, _gloffset_GetColorTableParameteriv), + NAME_FUNC_OFFSET(14124, glGetColorTableParameteriv, glGetColorTableParameterivEXT, glGetColorTableParameterivEXT, _gloffset_GetColorTableParameteriv), + NAME_FUNC_OFFSET(14154, glColorSubTable, glColorSubTable, NULL, _gloffset_ColorSubTable), + NAME_FUNC_OFFSET(14173, glCopyColorSubTable, glCopyColorSubTable, NULL, _gloffset_CopyColorSubTable), + NAME_FUNC_OFFSET(14196, glConvolutionFilter1D, glConvolutionFilter1D, NULL, _gloffset_ConvolutionFilter1D), + NAME_FUNC_OFFSET(14221, glConvolutionFilter2D, glConvolutionFilter2D, NULL, _gloffset_ConvolutionFilter2D), + NAME_FUNC_OFFSET(14246, glConvolutionParameterf, glConvolutionParameterf, NULL, _gloffset_ConvolutionParameterf), + NAME_FUNC_OFFSET(14273, glConvolutionParameterfv, glConvolutionParameterfv, NULL, _gloffset_ConvolutionParameterfv), + NAME_FUNC_OFFSET(14301, glConvolutionParameteri, glConvolutionParameteri, NULL, _gloffset_ConvolutionParameteri), + NAME_FUNC_OFFSET(14328, glConvolutionParameteriv, glConvolutionParameteriv, NULL, _gloffset_ConvolutionParameteriv), + NAME_FUNC_OFFSET(14356, glCopyConvolutionFilter1D, glCopyConvolutionFilter1D, NULL, _gloffset_CopyConvolutionFilter1D), + NAME_FUNC_OFFSET(14385, glCopyConvolutionFilter2D, glCopyConvolutionFilter2D, NULL, _gloffset_CopyConvolutionFilter2D), + NAME_FUNC_OFFSET(14414, glGetConvolutionFilter, gl_dispatch_stub_356, gl_dispatch_stub_356, _gloffset_GetConvolutionFilter), + NAME_FUNC_OFFSET(14440, glGetConvolutionParameterfv, gl_dispatch_stub_357, gl_dispatch_stub_357, _gloffset_GetConvolutionParameterfv), + NAME_FUNC_OFFSET(14471, glGetConvolutionParameteriv, gl_dispatch_stub_358, gl_dispatch_stub_358, _gloffset_GetConvolutionParameteriv), + NAME_FUNC_OFFSET(14502, glGetSeparableFilter, gl_dispatch_stub_359, gl_dispatch_stub_359, _gloffset_GetSeparableFilter), + NAME_FUNC_OFFSET(14526, glSeparableFilter2D, glSeparableFilter2D, NULL, _gloffset_SeparableFilter2D), + NAME_FUNC_OFFSET(14549, glGetHistogram, gl_dispatch_stub_361, gl_dispatch_stub_361, _gloffset_GetHistogram), + NAME_FUNC_OFFSET(14567, glGetHistogramParameterfv, gl_dispatch_stub_362, gl_dispatch_stub_362, _gloffset_GetHistogramParameterfv), + NAME_FUNC_OFFSET(14596, glGetHistogramParameteriv, gl_dispatch_stub_363, gl_dispatch_stub_363, _gloffset_GetHistogramParameteriv), + NAME_FUNC_OFFSET(14625, glGetMinmax, gl_dispatch_stub_364, gl_dispatch_stub_364, _gloffset_GetMinmax), + NAME_FUNC_OFFSET(14640, glGetMinmaxParameterfv, gl_dispatch_stub_365, gl_dispatch_stub_365, _gloffset_GetMinmaxParameterfv), + NAME_FUNC_OFFSET(14666, glGetMinmaxParameteriv, gl_dispatch_stub_366, gl_dispatch_stub_366, _gloffset_GetMinmaxParameteriv), + NAME_FUNC_OFFSET(14692, glHistogram, glHistogram, NULL, _gloffset_Histogram), + NAME_FUNC_OFFSET(14707, glMinmax, glMinmax, NULL, _gloffset_Minmax), + NAME_FUNC_OFFSET(14719, glResetHistogram, glResetHistogram, NULL, _gloffset_ResetHistogram), + NAME_FUNC_OFFSET(14739, glResetMinmax, glResetMinmax, NULL, _gloffset_ResetMinmax), + NAME_FUNC_OFFSET(14756, glTexImage3D, glTexImage3D, NULL, _gloffset_TexImage3D), + NAME_FUNC_OFFSET(14772, glTexSubImage3D, glTexSubImage3D, NULL, _gloffset_TexSubImage3D), + NAME_FUNC_OFFSET(14791, glCopyTexSubImage3D, glCopyTexSubImage3D, NULL, _gloffset_CopyTexSubImage3D), + NAME_FUNC_OFFSET(14814, glActiveTextureARB, glActiveTextureARB, NULL, _gloffset_ActiveTextureARB), + NAME_FUNC_OFFSET(14830, glClientActiveTextureARB, glClientActiveTextureARB, NULL, _gloffset_ClientActiveTextureARB), + NAME_FUNC_OFFSET(14852, glMultiTexCoord1dARB, glMultiTexCoord1dARB, NULL, _gloffset_MultiTexCoord1dARB), + NAME_FUNC_OFFSET(14870, glMultiTexCoord1dvARB, glMultiTexCoord1dvARB, NULL, _gloffset_MultiTexCoord1dvARB), + NAME_FUNC_OFFSET(14889, glMultiTexCoord1fARB, glMultiTexCoord1fARB, NULL, _gloffset_MultiTexCoord1fARB), + NAME_FUNC_OFFSET(14907, glMultiTexCoord1fvARB, glMultiTexCoord1fvARB, NULL, _gloffset_MultiTexCoord1fvARB), + NAME_FUNC_OFFSET(14926, glMultiTexCoord1iARB, glMultiTexCoord1iARB, NULL, _gloffset_MultiTexCoord1iARB), + NAME_FUNC_OFFSET(14944, glMultiTexCoord1ivARB, glMultiTexCoord1ivARB, NULL, _gloffset_MultiTexCoord1ivARB), + NAME_FUNC_OFFSET(14963, glMultiTexCoord1sARB, glMultiTexCoord1sARB, NULL, _gloffset_MultiTexCoord1sARB), + NAME_FUNC_OFFSET(14981, glMultiTexCoord1svARB, glMultiTexCoord1svARB, NULL, _gloffset_MultiTexCoord1svARB), + NAME_FUNC_OFFSET(15000, glMultiTexCoord2dARB, glMultiTexCoord2dARB, NULL, _gloffset_MultiTexCoord2dARB), + NAME_FUNC_OFFSET(15018, glMultiTexCoord2dvARB, glMultiTexCoord2dvARB, NULL, _gloffset_MultiTexCoord2dvARB), + NAME_FUNC_OFFSET(15037, glMultiTexCoord2fARB, glMultiTexCoord2fARB, NULL, _gloffset_MultiTexCoord2fARB), + NAME_FUNC_OFFSET(15055, glMultiTexCoord2fvARB, glMultiTexCoord2fvARB, NULL, _gloffset_MultiTexCoord2fvARB), + NAME_FUNC_OFFSET(15074, glMultiTexCoord2iARB, glMultiTexCoord2iARB, NULL, _gloffset_MultiTexCoord2iARB), + NAME_FUNC_OFFSET(15092, glMultiTexCoord2ivARB, glMultiTexCoord2ivARB, NULL, _gloffset_MultiTexCoord2ivARB), + NAME_FUNC_OFFSET(15111, glMultiTexCoord2sARB, glMultiTexCoord2sARB, NULL, _gloffset_MultiTexCoord2sARB), + NAME_FUNC_OFFSET(15129, glMultiTexCoord2svARB, glMultiTexCoord2svARB, NULL, _gloffset_MultiTexCoord2svARB), + NAME_FUNC_OFFSET(15148, glMultiTexCoord3dARB, glMultiTexCoord3dARB, NULL, _gloffset_MultiTexCoord3dARB), + NAME_FUNC_OFFSET(15166, glMultiTexCoord3dvARB, glMultiTexCoord3dvARB, NULL, _gloffset_MultiTexCoord3dvARB), + NAME_FUNC_OFFSET(15185, glMultiTexCoord3fARB, glMultiTexCoord3fARB, NULL, _gloffset_MultiTexCoord3fARB), + NAME_FUNC_OFFSET(15203, glMultiTexCoord3fvARB, glMultiTexCoord3fvARB, NULL, _gloffset_MultiTexCoord3fvARB), + NAME_FUNC_OFFSET(15222, glMultiTexCoord3iARB, glMultiTexCoord3iARB, NULL, _gloffset_MultiTexCoord3iARB), + NAME_FUNC_OFFSET(15240, glMultiTexCoord3ivARB, glMultiTexCoord3ivARB, NULL, _gloffset_MultiTexCoord3ivARB), + NAME_FUNC_OFFSET(15259, glMultiTexCoord3sARB, glMultiTexCoord3sARB, NULL, _gloffset_MultiTexCoord3sARB), + NAME_FUNC_OFFSET(15277, glMultiTexCoord3svARB, glMultiTexCoord3svARB, NULL, _gloffset_MultiTexCoord3svARB), + NAME_FUNC_OFFSET(15296, glMultiTexCoord4dARB, glMultiTexCoord4dARB, NULL, _gloffset_MultiTexCoord4dARB), + NAME_FUNC_OFFSET(15314, glMultiTexCoord4dvARB, glMultiTexCoord4dvARB, NULL, _gloffset_MultiTexCoord4dvARB), + NAME_FUNC_OFFSET(15333, glMultiTexCoord4fARB, glMultiTexCoord4fARB, NULL, _gloffset_MultiTexCoord4fARB), + NAME_FUNC_OFFSET(15351, glMultiTexCoord4fvARB, glMultiTexCoord4fvARB, NULL, _gloffset_MultiTexCoord4fvARB), + NAME_FUNC_OFFSET(15370, glMultiTexCoord4iARB, glMultiTexCoord4iARB, NULL, _gloffset_MultiTexCoord4iARB), + NAME_FUNC_OFFSET(15388, glMultiTexCoord4ivARB, glMultiTexCoord4ivARB, NULL, _gloffset_MultiTexCoord4ivARB), + NAME_FUNC_OFFSET(15407, glMultiTexCoord4sARB, glMultiTexCoord4sARB, NULL, _gloffset_MultiTexCoord4sARB), + NAME_FUNC_OFFSET(15425, glMultiTexCoord4svARB, glMultiTexCoord4svARB, NULL, _gloffset_MultiTexCoord4svARB), + NAME_FUNC_OFFSET(15444, glStencilOpSeparate, glStencilOpSeparate, NULL, _gloffset_StencilOpSeparate), + NAME_FUNC_OFFSET(15467, glLoadTransposeMatrixdARB, glLoadTransposeMatrixdARB, NULL, _gloffset_LoadTransposeMatrixdARB), + NAME_FUNC_OFFSET(15490, glLoadTransposeMatrixfARB, glLoadTransposeMatrixfARB, NULL, _gloffset_LoadTransposeMatrixfARB), + NAME_FUNC_OFFSET(15513, glMultTransposeMatrixdARB, glMultTransposeMatrixdARB, NULL, _gloffset_MultTransposeMatrixdARB), + NAME_FUNC_OFFSET(15536, glMultTransposeMatrixfARB, glMultTransposeMatrixfARB, NULL, _gloffset_MultTransposeMatrixfARB), + NAME_FUNC_OFFSET(15559, glSampleCoverageARB, glSampleCoverageARB, NULL, _gloffset_SampleCoverageARB), + NAME_FUNC_OFFSET(15576, glCompressedTexImage1DARB, glCompressedTexImage1DARB, NULL, _gloffset_CompressedTexImage1DARB), + NAME_FUNC_OFFSET(15599, glCompressedTexImage2DARB, glCompressedTexImage2DARB, NULL, _gloffset_CompressedTexImage2DARB), + NAME_FUNC_OFFSET(15622, glCompressedTexImage3DARB, glCompressedTexImage3DARB, NULL, _gloffset_CompressedTexImage3DARB), + NAME_FUNC_OFFSET(15645, glCompressedTexSubImage1DARB, glCompressedTexSubImage1DARB, NULL, _gloffset_CompressedTexSubImage1DARB), + NAME_FUNC_OFFSET(15671, glCompressedTexSubImage2DARB, glCompressedTexSubImage2DARB, NULL, _gloffset_CompressedTexSubImage2DARB), + NAME_FUNC_OFFSET(15697, glCompressedTexSubImage3DARB, glCompressedTexSubImage3DARB, NULL, _gloffset_CompressedTexSubImage3DARB), + NAME_FUNC_OFFSET(15723, glGetCompressedTexImageARB, glGetCompressedTexImageARB, NULL, _gloffset_GetCompressedTexImageARB), + NAME_FUNC_OFFSET(15747, glDisableVertexAttribArrayARB, glDisableVertexAttribArrayARB, NULL, _gloffset_DisableVertexAttribArrayARB), + NAME_FUNC_OFFSET(15774, glEnableVertexAttribArrayARB, glEnableVertexAttribArrayARB, NULL, _gloffset_EnableVertexAttribArrayARB), + NAME_FUNC_OFFSET(15800, glGetVertexAttribdvARB, glGetVertexAttribdvARB, NULL, _gloffset_GetVertexAttribdvARB), + NAME_FUNC_OFFSET(15820, glGetVertexAttribfvARB, glGetVertexAttribfvARB, NULL, _gloffset_GetVertexAttribfvARB), + NAME_FUNC_OFFSET(15840, glGetVertexAttribivARB, glGetVertexAttribivARB, NULL, _gloffset_GetVertexAttribivARB), + NAME_FUNC_OFFSET(15860, glProgramEnvParameter4dARB, glProgramEnvParameter4dARB, NULL, _gloffset_ProgramEnvParameter4dARB), + NAME_FUNC_OFFSET(15883, glProgramEnvParameter4dvARB, glProgramEnvParameter4dvARB, NULL, _gloffset_ProgramEnvParameter4dvARB), + NAME_FUNC_OFFSET(15907, glProgramEnvParameter4fARB, glProgramEnvParameter4fARB, NULL, _gloffset_ProgramEnvParameter4fARB), + NAME_FUNC_OFFSET(15930, glProgramEnvParameter4fvARB, glProgramEnvParameter4fvARB, NULL, _gloffset_ProgramEnvParameter4fvARB), NAME_FUNC_OFFSET(15954, glVertexAttrib1dARB, glVertexAttrib1dARB, NULL, _gloffset_VertexAttrib1dARB), NAME_FUNC_OFFSET(15971, glVertexAttrib1dvARB, glVertexAttrib1dvARB, NULL, _gloffset_VertexAttrib1dvARB), NAME_FUNC_OFFSET(15989, glVertexAttrib1fARB, glVertexAttrib1fARB, NULL, _gloffset_VertexAttrib1fARB), @@ -2252,8 +2252,8 @@ static const glprocs_table_t static_functions[] = { NAME_FUNC_OFFSET(18694, glIsProgramNV, glIsProgramNV, NULL, _gloffset_IsProgramNV), NAME_FUNC_OFFSET(18709, glPointParameteriNV, glPointParameteriNV, NULL, _gloffset_PointParameteriNV), NAME_FUNC_OFFSET(18727, glPointParameterivNV, glPointParameterivNV, NULL, _gloffset_PointParameterivNV), - NAME_FUNC_OFFSET(18746, gl_dispatch_stub_749, gl_dispatch_stub_749, NULL, _gloffset_BlendEquationSeparateEXT), - NAME_FUNC_OFFSET(18770, gl_dispatch_stub_749, gl_dispatch_stub_749, NULL, _gloffset_BlendEquationSeparateEXT), + NAME_FUNC_OFFSET(18746, gl_dispatch_stub_745, gl_dispatch_stub_745, NULL, _gloffset_BlendEquationSeparateEXT), + NAME_FUNC_OFFSET(18770, gl_dispatch_stub_745, gl_dispatch_stub_745, NULL, _gloffset_BlendEquationSeparateEXT), NAME_FUNC_OFFSET(-1, NULL, NULL, NULL, 0) }; diff --git a/src/mesa/main/bufferobj.c b/src/mesa/main/bufferobj.c index e762eb3b63..dc0307feb5 100644 --- a/src/mesa/main/bufferobj.c +++ b/src/mesa/main/bufferobj.c @@ -551,8 +551,8 @@ _mesa_map_drawpix_pbo(GLcontext *ctx, * \sa _mesa_unmap_bitmap_pbo */ void -_mesa_unmap_drapix_pbo(GLcontext *ctx, - const struct gl_pixelstore_attrib *unpack) +_mesa_unmap_drawpix_pbo(GLcontext *ctx, + const struct gl_pixelstore_attrib *unpack) { if (unpack->BufferObj->Name) { ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_UNPACK_BUFFER_EXT, diff --git a/src/mesa/main/bufferobj.h b/src/mesa/main/bufferobj.h index 46525f08ae..024e5a8c3c 100644 --- a/src/mesa/main/bufferobj.h +++ b/src/mesa/main/bufferobj.h @@ -101,8 +101,8 @@ _mesa_map_drawpix_pbo(GLcontext *ctx, const GLvoid *pixels); extern void -_mesa_unmap_drapix_pbo(GLcontext *ctx, - const struct gl_pixelstore_attrib *unpack); +_mesa_unmap_drawpix_pbo(GLcontext *ctx, + const struct gl_pixelstore_attrib *unpack); extern void * diff --git a/src/mesa/main/config.h b/src/mesa/main/config.h index 8c64248845..232e698f50 100644 --- a/src/mesa/main/config.h +++ b/src/mesa/main/config.h @@ -163,25 +163,6 @@ /** For GL_EXT_texture_lod_bias (typically MAX_TEXTURE_LEVELS - 1) */ #define MAX_TEXTURE_LOD_BIAS 11.0 -/** For GL_NV_vertex_program */ -/*@{*/ -#define MAX_NV_VERTEX_PROGRAM_INSTRUCTIONS 128 -#define MAX_NV_VERTEX_PROGRAM_TEMPS 12 -#define MAX_NV_VERTEX_PROGRAM_PARAMS 128 /* KW: power of two */ -#define MAX_NV_VERTEX_PROGRAM_INPUTS 16 -#define MAX_NV_VERTEX_PROGRAM_OUTPUTS 15 -/*@}*/ - -/** For GL_NV_fragment_program */ -/*@{*/ -#define MAX_NV_FRAGMENT_PROGRAM_INSTRUCTIONS 1024 /* 72 for GL_ARB_f_p */ -#define MAX_NV_FRAGMENT_PROGRAM_TEMPS 96 -#define MAX_NV_FRAGMENT_PROGRAM_PARAMS 64 -#define MAX_NV_FRAGMENT_PROGRAM_INPUTS 12 -#define MAX_NV_FRAGMENT_PROGRAM_OUTPUTS 3 -#define MAX_NV_FRAGMENT_PROGRAM_WRITE_ONLYS 2 -/*@}*/ - /** For GL_ARB_vertex_program */ /*@{*/ #define MAX_VERTEX_PROGRAM_ADDRESS_REGS 1 @@ -210,6 +191,26 @@ #define MAX_SAMPLERS 8 /*@}*/ +/** For GL_NV_vertex_program */ +/*@{*/ +#define MAX_NV_VERTEX_PROGRAM_INSTRUCTIONS 128 +#define MAX_NV_VERTEX_PROGRAM_TEMPS 12 +#define MAX_NV_VERTEX_PROGRAM_PARAMS MAX_PROGRAM_ENV_PARAMS +#define MAX_NV_VERTEX_PROGRAM_INPUTS 16 +#define MAX_NV_VERTEX_PROGRAM_OUTPUTS 15 +/*@}*/ + +/** For GL_NV_fragment_program */ +/*@{*/ +#define MAX_NV_FRAGMENT_PROGRAM_INSTRUCTIONS 1024 /* 72 for GL_ARB_f_p */ +#define MAX_NV_FRAGMENT_PROGRAM_TEMPS 96 +#define MAX_NV_FRAGMENT_PROGRAM_PARAMS 64 +#define MAX_NV_FRAGMENT_PROGRAM_INPUTS 12 +#define MAX_NV_FRAGMENT_PROGRAM_OUTPUTS 3 +#define MAX_NV_FRAGMENT_PROGRAM_WRITE_ONLYS 2 +/*@}*/ + + /** For GL_ARB_vertex_shader */ /*@{*/ #define MAX_VERTEX_ATTRIBS 16 diff --git a/src/mesa/main/context.c b/src/mesa/main/context.c index d06644f65d..2158eb6873 100644..100755 --- a/src/mesa/main/context.c +++ b/src/mesa/main/context.c @@ -150,8 +150,6 @@ int MESA_DEBUG_FLAGS = 0; /* ubyte -> float conversion */ GLfloat _mesa_ubyte_to_float_color_tab[256]; -static void -free_shared_state( GLcontext *ctx, struct gl_shared_state *ss ); /** @@ -423,12 +421,14 @@ alloc_shared_state( GLcontext *ctx ) #endif #if FEATURE_ARB_vertex_program - ss->DefaultVertexProgram = ctx->Driver.NewProgram(ctx, GL_VERTEX_PROGRAM_ARB, 0); + ss->DefaultVertexProgram = (struct gl_vertex_program *) + ctx->Driver.NewProgram(ctx, GL_VERTEX_PROGRAM_ARB, 0); if (!ss->DefaultVertexProgram) goto cleanup; #endif #if FEATURE_ARB_fragment_program - ss->DefaultFragmentProgram = ctx->Driver.NewProgram(ctx, GL_FRAGMENT_PROGRAM_ARB, 0); + ss->DefaultFragmentProgram = (struct gl_fragment_program *) + ctx->Driver.NewProgram(ctx, GL_FRAGMENT_PROGRAM_ARB, 0); if (!ss->DefaultFragmentProgram) goto cleanup; #endif @@ -513,12 +513,10 @@ alloc_shared_state( GLcontext *ctx ) _mesa_DeleteHashTable(ss->Programs); #endif #if FEATURE_ARB_vertex_program - if (ss->DefaultVertexProgram) - ctx->Driver.DeleteProgram(ctx, ss->DefaultVertexProgram); + _mesa_reference_vertprog(ctx, &ss->DefaultVertexProgram, NULL); #endif #if FEATURE_ARB_fragment_program - if (ss->DefaultFragmentProgram) - ctx->Driver.DeleteProgram(ctx, ss->DefaultFragmentProgram); + _mesa_reference_fragprog(ctx, &ss->DefaultFragmentProgram, NULL); #endif #if FEATURE_ATI_fragment_shader if (ss->DefaultFragmentShader) @@ -595,6 +593,8 @@ delete_program_cb(GLuint id, void *data, void *userData) { struct gl_program *prog = (struct gl_program *) data; GLcontext *ctx = (GLcontext *) userData; + ASSERT(prog->RefCount == 1); /* should only be referenced by hash table */ + prog->RefCount = 0; /* now going away */ ctx->Driver.DeleteProgram(ctx, prog); } @@ -633,6 +633,21 @@ delete_arrayobj_cb(GLuint id, void *data, void *userData) } /** + * Callback for freeing shader program data. Call it before delete_shader_cb + * to avoid memory access error. + */ +static void +free_shader_program_data_cb(GLuint id, void *data, void *userData) +{ + GLcontext *ctx = (GLcontext *) userData; + struct gl_shader_program *shProg = (struct gl_shader_program *) data; + + if (shProg->Type == GL_SHADER_PROGRAM_MESA) { + _mesa_free_shader_program_data(ctx, shProg); + } +} + +/** * Callback for deleting shader and shader programs objects. * Called by _mesa_HashDeleteAll(). */ @@ -673,32 +688,21 @@ free_shared_state( GLcontext *ctx, struct gl_shared_state *ss ) _mesa_HashDeleteAll(ss->DisplayList, delete_displaylist_cb, ctx); _mesa_DeleteHashTable(ss->DisplayList); - /* - * Free texture objects - */ - ASSERT(ctx->Driver.DeleteTexture); - /* the default textures */ - ctx->Driver.DeleteTexture(ctx, ss->Default1D); - ctx->Driver.DeleteTexture(ctx, ss->Default2D); - ctx->Driver.DeleteTexture(ctx, ss->Default3D); - ctx->Driver.DeleteTexture(ctx, ss->DefaultCubeMap); - ctx->Driver.DeleteTexture(ctx, ss->DefaultRect); - ctx->Driver.DeleteTexture(ctx, ss->Default1DArray); - ctx->Driver.DeleteTexture(ctx, ss->Default2DArray); - - /* all other textures */ - _mesa_HashDeleteAll(ss->TexObjects, delete_texture_cb, ctx); - _mesa_DeleteHashTable(ss->TexObjects); +#if FEATURE_ARB_shader_objects + _mesa_HashWalk(ss->ShaderObjects, free_shader_program_data_cb, ctx); + _mesa_HashDeleteAll(ss->ShaderObjects, delete_shader_cb, ctx); + _mesa_DeleteHashTable(ss->ShaderObjects); +#endif #if defined(FEATURE_NV_vertex_program) || defined(FEATURE_NV_fragment_program) _mesa_HashDeleteAll(ss->Programs, delete_program_cb, ctx); _mesa_DeleteHashTable(ss->Programs); #endif #if FEATURE_ARB_vertex_program - ctx->Driver.DeleteProgram(ctx, ss->DefaultVertexProgram); + _mesa_reference_vertprog(ctx, &ss->DefaultVertexProgram, NULL); #endif #if FEATURE_ARB_fragment_program - ctx->Driver.DeleteProgram(ctx, ss->DefaultFragmentProgram); + _mesa_reference_fragprog(ctx, &ss->DefaultFragmentProgram, NULL); #endif #if FEATURE_ATI_fragment_shader @@ -715,16 +719,28 @@ free_shared_state( GLcontext *ctx, struct gl_shared_state *ss ) _mesa_HashDeleteAll(ss->ArrayObjects, delete_arrayobj_cb, ctx); _mesa_DeleteHashTable(ss->ArrayObjects); -#if FEATURE_ARB_shader_objects - _mesa_HashDeleteAll(ss->ShaderObjects, delete_shader_cb, ctx); - _mesa_DeleteHashTable(ss->ShaderObjects); -#endif - #if FEATURE_EXT_framebuffer_object _mesa_DeleteHashTable(ss->FrameBuffers); _mesa_DeleteHashTable(ss->RenderBuffers); #endif + /* + * Free texture objects (after FBOs since some textures might have + * been bound to FBOs). + */ + ASSERT(ctx->Driver.DeleteTexture); + /* the default textures */ + ctx->Driver.DeleteTexture(ctx, ss->Default1D); + ctx->Driver.DeleteTexture(ctx, ss->Default2D); + ctx->Driver.DeleteTexture(ctx, ss->Default3D); + ctx->Driver.DeleteTexture(ctx, ss->DefaultCubeMap); + ctx->Driver.DeleteTexture(ctx, ss->DefaultRect); + ctx->Driver.DeleteTexture(ctx, ss->Default1DArray); + ctx->Driver.DeleteTexture(ctx, ss->Default2DArray); + /* all other textures */ + _mesa_HashDeleteAll(ss->TexObjects, delete_texture_cb, ctx); + _mesa_DeleteHashTable(ss->TexObjects); + _glthread_DESTROY_MUTEX(ss->Mutex); _mesa_free(ss); @@ -1190,6 +1206,14 @@ _mesa_free_context_data( GLcontext *ctx ) _mesa_unreference_framebuffer(&ctx->ReadBuffer); } + _mesa_reference_vertprog(ctx, &ctx->VertexProgram.Current, NULL); + _mesa_reference_vertprog(ctx, &ctx->VertexProgram._Current, NULL); + _mesa_reference_vertprog(ctx, &ctx->VertexProgram._TnlProgram, NULL); + + _mesa_reference_fragprog(ctx, &ctx->FragmentProgram.Current, NULL); + _mesa_reference_fragprog(ctx, &ctx->FragmentProgram._Current, NULL); + _mesa_reference_fragprog(ctx, &ctx->FragmentProgram._TexEnvProgram, NULL); + _mesa_free_lighting_data( ctx ); _mesa_free_eval_data( ctx ); _mesa_free_texture_data( ctx ); @@ -1504,15 +1528,19 @@ _mesa_make_current( GLcontext *newCtx, GLframebuffer *drawBuffer, * or not bound to a user-created FBO. */ if (!newCtx->DrawBuffer || newCtx->DrawBuffer->Name == 0) { - _mesa_reference_framebuffer(&newCtx->DrawBuffer, drawBuffer); /* fix up the fb fields - these will end up wrong otherwise - if the DRIdrawable changes, and everything relies on them. - This is a bit messy (same as needed in _mesa_BindFramebufferEXT) */ + * if the DRIdrawable changes, and everything relies on them. + * This is a bit messy (same as needed in _mesa_BindFramebufferEXT) + */ int i; GLenum buffers[MAX_DRAW_BUFFERS]; + + _mesa_reference_framebuffer(&newCtx->DrawBuffer, drawBuffer); + for(i = 0; i < newCtx->Const.MaxDrawBuffers; i++) { buffers[i] = newCtx->Color.DrawBuffer[i]; } + _mesa_drawbuffers(newCtx, newCtx->Const.MaxDrawBuffers, buffers, NULL); } if (!newCtx->ReadBuffer || newCtx->ReadBuffer->Name == 0) { diff --git a/src/mesa/main/dlist.c b/src/mesa/main/dlist.c index 293ee5fa34..13ebd4dd0d 100644 --- a/src/mesa/main/dlist.c +++ b/src/mesa/main/dlist.c @@ -303,7 +303,6 @@ typedef enum OPCODE_EXECUTE_PROGRAM_NV, OPCODE_REQUEST_RESIDENT_PROGRAMS_NV, OPCODE_LOAD_PROGRAM_NV, - OPCODE_PROGRAM_PARAMETER4F_NV, OPCODE_TRACK_MATRIX_NV, /* GL_NV_fragment_program */ OPCODE_PROGRAM_LOCAL_PARAMETER_ARB, @@ -4248,77 +4247,111 @@ save_BindProgramNV(GLenum target, GLuint id) CALL_BindProgramNV(ctx->Exec, (target, id)); } } -#endif /* FEATURE_NV_vertex_program || FEATURE_ARB_vertex_program || FEATURE_ARB_fragment_program */ -#if FEATURE_NV_vertex_program static void GLAPIENTRY -save_ExecuteProgramNV(GLenum target, GLuint id, const GLfloat *params) +save_ProgramEnvParameter4fARB(GLenum target, GLuint index, + GLfloat x, GLfloat y, GLfloat z, GLfloat w) { GET_CURRENT_CONTEXT(ctx); Node *n; ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); - n = ALLOC_INSTRUCTION(ctx, OPCODE_EXECUTE_PROGRAM_NV, 6); + n = ALLOC_INSTRUCTION(ctx, OPCODE_PROGRAM_ENV_PARAMETER_ARB, 6); if (n) { n[1].e = target; - n[2].ui = id; - n[3].f = params[0]; - n[4].f = params[1]; - n[5].f = params[2]; - n[6].f = params[3]; + n[2].ui = index; + n[3].f = x; + n[4].f = y; + n[5].f = z; + n[6].f = w; } if (ctx->ExecuteFlag) { - CALL_ExecuteProgramNV(ctx->Exec, (target, id, params)); + CALL_ProgramEnvParameter4fARB(ctx->Exec, (target, index, x, y, z, w)); } } static void GLAPIENTRY -save_ProgramParameter4fNV(GLenum target, GLuint index, - GLfloat x, GLfloat y, GLfloat z, GLfloat w) +save_ProgramEnvParameter4fvARB(GLenum target, GLuint index, + const GLfloat *params) +{ + save_ProgramEnvParameter4fARB(target, index, params[0], params[1], + params[2], params[3]); +} + + +static void GLAPIENTRY +save_ProgramEnvParameters4fvEXT(GLenum target, GLuint index, GLsizei count, + const GLfloat * params) { GET_CURRENT_CONTEXT(ctx); Node *n; ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); - n = ALLOC_INSTRUCTION(ctx, OPCODE_PROGRAM_PARAMETER4F_NV, 6); - if (n) { - n[1].e = target; - n[2].ui = index; - n[3].f = x; - n[4].f = y; - n[5].f = z; - n[6].f = w; + + if (count > 0) { + GLint i; + const GLfloat * p = params; + + for (i = 0 ; i < count ; i++) { + n = ALLOC_INSTRUCTION(ctx, OPCODE_PROGRAM_ENV_PARAMETER_ARB, 6); + if (n) { + n[1].e = target; + n[2].ui = index; + n[3].f = p[0]; + n[4].f = p[1]; + n[5].f = p[2]; + n[6].f = p[3]; + p += 4; + } + } } + if (ctx->ExecuteFlag) { - CALL_ProgramParameter4fNV(ctx->Exec, (target, index, x, y, z, w)); + CALL_ProgramEnvParameters4fvEXT(ctx->Exec, (target, index, count, params)); } } static void GLAPIENTRY -save_ProgramParameter4fvNV(GLenum target, GLuint index, - const GLfloat *params) +save_ProgramEnvParameter4dARB(GLenum target, GLuint index, + GLdouble x, GLdouble y, GLdouble z, GLdouble w) { - save_ProgramParameter4fNV(target, index, params[0], params[1], - params[2], params[3]); + save_ProgramEnvParameter4fARB(target, index, + (GLfloat) x, + (GLfloat) y, (GLfloat) z, (GLfloat) w); } static void GLAPIENTRY -save_ProgramParameter4dNV(GLenum target, GLuint index, - GLdouble x, GLdouble y, GLdouble z, GLdouble w) +save_ProgramEnvParameter4dvARB(GLenum target, GLuint index, + const GLdouble *params) { - save_ProgramParameter4fNV(target, index, (GLfloat) x, (GLfloat) y, - (GLfloat) z, (GLfloat) w); + save_ProgramEnvParameter4fARB(target, index, + (GLfloat) params[0], + (GLfloat) params[1], + (GLfloat) params[2], (GLfloat) params[3]); } +#endif /* FEATURE_ARB_vertex_program || FEATURE_ARB_fragment_program || FEATURE_NV_vertex_program */ +#if FEATURE_NV_vertex_program static void GLAPIENTRY -save_ProgramParameter4dvNV(GLenum target, GLuint index, - const GLdouble *params) +save_ExecuteProgramNV(GLenum target, GLuint id, const GLfloat *params) { - save_ProgramParameter4fNV(target, index, (GLfloat) params[0], - (GLfloat) params[1], (GLfloat) params[2], - (GLfloat) params[3]); + GET_CURRENT_CONTEXT(ctx); + Node *n; + ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); + n = ALLOC_INSTRUCTION(ctx, OPCODE_EXECUTE_PROGRAM_NV, 6); + if (n) { + n[1].e = target; + n[2].ui = id; + n[3].f = params[0]; + n[4].f = params[1]; + n[5].f = params[2]; + n[6].f = params[3]; + } + if (ctx->ExecuteFlag) { + CALL_ExecuteProgramNV(ctx->Exec, (target, id, params)); + } } @@ -4328,7 +4361,7 @@ save_ProgramParameters4dvNV(GLenum target, GLuint index, { GLuint i; for (i = 0; i < num; i++) { - save_ProgramParameter4dvNV(target, index + i, params + 4 * i); + save_ProgramEnvParameter4dvARB(target, index + i, params + 4 * i); } } @@ -4339,7 +4372,7 @@ save_ProgramParameters4fvNV(GLenum target, GLuint index, { GLuint i; for (i = 0; i < num; i++) { - save_ProgramParameter4fvNV(target, index + i, params + 4 * i); + save_ProgramEnvParameter4fvARB(target, index + i, params + 4 * i); } } @@ -4667,90 +4700,6 @@ save_ProgramStringARB(GLenum target, GLenum format, GLsizei len, } } - -static void GLAPIENTRY -save_ProgramEnvParameter4fARB(GLenum target, GLuint index, - GLfloat x, GLfloat y, GLfloat z, GLfloat w) -{ - GET_CURRENT_CONTEXT(ctx); - Node *n; - ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); - n = ALLOC_INSTRUCTION(ctx, OPCODE_PROGRAM_ENV_PARAMETER_ARB, 6); - if (n) { - n[1].e = target; - n[2].ui = index; - n[3].f = x; - n[4].f = y; - n[5].f = z; - n[6].f = w; - } - if (ctx->ExecuteFlag) { - CALL_ProgramEnvParameter4fARB(ctx->Exec, (target, index, x, y, z, w)); - } -} - - -static void GLAPIENTRY -save_ProgramEnvParameter4fvARB(GLenum target, GLuint index, - const GLfloat *params) -{ - save_ProgramEnvParameter4fARB(target, index, params[0], params[1], - params[2], params[3]); -} - - -static void GLAPIENTRY -save_ProgramEnvParameters4fvEXT(GLenum target, GLuint index, GLsizei count, - const GLfloat * params) -{ - GET_CURRENT_CONTEXT(ctx); - Node *n; - ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx); - - if (count > 0) { - GLint i; - const GLfloat * p = params; - - for (i = 0 ; i < count ; i++) { - n = ALLOC_INSTRUCTION(ctx, OPCODE_PROGRAM_ENV_PARAMETER_ARB, 6); - if (n) { - n[1].e = target; - n[2].ui = index; - n[3].f = p[0]; - n[4].f = p[1]; - n[5].f = p[2]; - n[6].f = p[3]; - p += 4; - } - } - } - - if (ctx->ExecuteFlag) { - CALL_ProgramEnvParameters4fvEXT(ctx->Exec, (target, index, count, params)); - } -} - - -static void GLAPIENTRY -save_ProgramEnvParameter4dARB(GLenum target, GLuint index, - GLdouble x, GLdouble y, GLdouble z, GLdouble w) -{ - save_ProgramEnvParameter4fARB(target, index, - (GLfloat) x, - (GLfloat) y, (GLfloat) z, (GLfloat) w); -} - - -static void GLAPIENTRY -save_ProgramEnvParameter4dvARB(GLenum target, GLuint index, - const GLdouble *params) -{ - save_ProgramEnvParameter4fARB(target, index, - (GLfloat) params[0], - (GLfloat) params[1], - (GLfloat) params[2], (GLfloat) params[3]); -} - #endif /* FEATURE_ARB_vertex_program || FEATURE_ARB_fragment_program */ @@ -6425,10 +6374,6 @@ execute_list(GLcontext *ctx, GLuint list) CALL_LoadProgramNV(ctx->Exec, (n[1].e, n[2].ui, n[3].i, (const GLubyte *) n[4].data)); break; - case OPCODE_PROGRAM_PARAMETER4F_NV: - CALL_ProgramParameter4fNV(ctx->Exec, (n[1].e, n[2].ui, n[3].f, - n[4].f, n[5].f, n[6].f)); - break; case OPCODE_TRACK_MATRIX_NV: CALL_TrackMatrixNV(ctx->Exec, (n[1].e, n[2].ui, n[3].e, n[4].e)); break; @@ -6459,6 +6404,8 @@ execute_list(GLcontext *ctx, GLuint list) CALL_ProgramStringARB(ctx->Exec, (n[1].e, n[2].e, n[3].i, n[4].data)); break; +#endif +#if FEATURE_ARB_vertex_program || FEATURE_ARB_fragment_program || FEATURE_NV_vertex_program case OPCODE_PROGRAM_ENV_PARAMETER_ARB: CALL_ProgramEnvParameter4fARB(ctx->Exec, (n[1].e, n[2].ui, n[3].f, n[4].f, n[5].f, @@ -8014,10 +7961,10 @@ _mesa_init_dlist_table(struct _glapi_table *table) SET_GetVertexAttribPointervNV(table, _mesa_GetVertexAttribPointervNV); SET_IsProgramNV(table, _mesa_IsProgramARB); SET_LoadProgramNV(table, save_LoadProgramNV); - SET_ProgramParameter4dNV(table, save_ProgramParameter4dNV); - SET_ProgramParameter4dvNV(table, save_ProgramParameter4dvNV); - SET_ProgramParameter4fNV(table, save_ProgramParameter4fNV); - SET_ProgramParameter4fvNV(table, save_ProgramParameter4fvNV); + SET_ProgramEnvParameter4dARB(table, save_ProgramEnvParameter4dARB); + SET_ProgramEnvParameter4dvARB(table, save_ProgramEnvParameter4dvARB); + SET_ProgramEnvParameter4fARB(table, save_ProgramEnvParameter4fARB); + SET_ProgramEnvParameter4fvARB(table, save_ProgramEnvParameter4fvARB); SET_ProgramParameters4dvNV(table, save_ProgramParameters4dvNV); SET_ProgramParameters4fvNV(table, save_ProgramParameters4fvNV); SET_TrackMatrixNV(table, save_TrackMatrixNV); diff --git a/src/mesa/main/drawpix.c b/src/mesa/main/drawpix.c index 0f64f1c1c0..fa422bb3c7 100644 --- a/src/mesa/main/drawpix.c +++ b/src/mesa/main/drawpix.c @@ -377,8 +377,9 @@ _mesa_Bitmap( GLsizei width, GLsizei height, if (ctx->RenderMode == GL_RENDER) { /* Truncate, to satisfy conformance tests (matches SGI's OpenGL). */ - GLint x = IFLOOR(ctx->Current.RasterPos[0] - xorig); - GLint y = IFLOOR(ctx->Current.RasterPos[1] - yorig); + const GLfloat epsilon = 0.0001; + GLint x = IFLOOR(ctx->Current.RasterPos[0] + epsilon - xorig); + GLint y = IFLOOR(ctx->Current.RasterPos[1] + epsilon - yorig); if (ctx->Unpack.BufferObj->Name) { /* unpack from PBO */ diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c index f648d081c0..810af9e33e 100644 --- a/src/mesa/main/ffvertex_prog.c +++ b/src/mesa/main/ffvertex_prog.c @@ -946,35 +946,33 @@ static void build_lighting( struct tnl_program *p ) _bfc1 = _bfc0; } - /* If no lights, still need to emit the scenecolor. */ - { - struct ureg res0 = register_output( p, VERT_RESULT_COL0 ); - emit_op1(p, OPCODE_MOV, res0, 0, _col0); - } + { + struct ureg res0 = register_output( p, VERT_RESULT_COL0 ); + emit_op1(p, OPCODE_MOV, res0, 0, _col0); + } - if (separate) { - struct ureg res1 = register_output( p, VERT_RESULT_COL1 ); - emit_op1(p, OPCODE_MOV, res1, 0, _col1); - } + if (separate) { + struct ureg res1 = register_output( p, VERT_RESULT_COL1 ); + emit_op1(p, OPCODE_MOV, res1, 0, _col1); + } - if (twoside) { - struct ureg res0 = register_output( p, VERT_RESULT_BFC0 ); - emit_op1(p, OPCODE_MOV, res0, 0, _bfc0); - } + if (twoside) { + struct ureg res0 = register_output( p, VERT_RESULT_BFC0 ); + emit_op1(p, OPCODE_MOV, res0, 0, _bfc0); + } - if (twoside && separate) { - struct ureg res1 = register_output( p, VERT_RESULT_BFC1 ); - emit_op1(p, OPCODE_MOV, res1, 0, _bfc1); - } + if (twoside && separate) { + struct ureg res1 = register_output( p, VERT_RESULT_BFC1 ); + emit_op1(p, OPCODE_MOV, res1, 0, _bfc1); + } if (nr_lights == 0) { release_temps(p); return; } - for (i = 0; i < MAX_LIGHTS; i++) { if (p->state->unit[i].light_enabled) { struct ureg half = undef; @@ -1006,7 +1004,7 @@ static void build_lighting( struct tnl_program *p ) VPpli = get_temp(p); half = get_temp(p); - /* Calulate VPpli vector + /* Calculate VPpli vector */ emit_op2(p, OPCODE_SUB, VPpli, 0, Ppli, V); @@ -1017,15 +1015,13 @@ static void build_lighting( struct tnl_program *p ) emit_op1(p, OPCODE_RSQ, dist, 0, dist); emit_op2(p, OPCODE_MUL, VPpli, 0, VPpli, dist); - - /* Calculate attenuation: + /* Calculate attenuation: */ if (!p->state->unit[i].light_spotcutoff_is_180 || p->state->unit[i].light_attenuated) { att = calculate_light_attenuation(p, i, VPpli, dist); } - - + /* Calculate viewer direction, or use infinite viewer: */ if (p->state->light_local_viewer) { @@ -1047,7 +1043,6 @@ static void build_lighting( struct tnl_program *p ) emit_op2(p, OPCODE_DP3, dots, WRITEMASK_X, normal, VPpli); emit_op2(p, OPCODE_DP3, dots, WRITEMASK_Y, normal, half); - /* Front face lighting: */ { diff --git a/src/mesa/main/imports.c b/src/mesa/main/imports.c index d8d35af15e..d798f80e25 100644 --- a/src/mesa/main/imports.c +++ b/src/mesa/main/imports.c @@ -542,26 +542,24 @@ int _mesa_ffs(int i) { #if (defined(_WIN32) && !defined(__MINGW32__) ) || defined(__IBMC__) || defined(__IBMCPP__) - register int bit = 0; - if (i != 0) { - if ((i & 0xffff) == 0) { - bit += 16; - i >>= 16; - } - if ((i & 0xff) == 0) { - bit += 8; - i >>= 8; - } - if ((i & 0xf) == 0) { - bit += 4; - i >>= 4; - } - while ((i & 1) == 0) { - bit++; - i >>= 1; - } + register int bit = 1; + if ((i & 0xffff) == 0) { + bit += 16; + i >>= 16; + } + if ((i & 0xff) == 0) { + bit += 8; + i >>= 8; + } + if ((i & 0xf) == 0) { + bit += 4; + i >>= 4; + } + if ((i & 0x3) == 0) { + bit += 2; + i >>= 2; } - return bit; + return (i) ? (bit + ((i + 1) & 0x01)) : 0; #else return ffs(i); #endif diff --git a/src/mesa/main/mtypes.h b/src/mesa/main/mtypes.h index 50b22d25bf..463142fe39 100644 --- a/src/mesa/main/mtypes.h +++ b/src/mesa/main/mtypes.h @@ -2203,10 +2203,10 @@ struct gl_shared_state /*@{*/ struct _mesa_HashTable *Programs; /**< All vertex/fragment programs */ #if FEATURE_ARB_vertex_program - struct gl_program *DefaultVertexProgram; + struct gl_vertex_program *DefaultVertexProgram; #endif #if FEATURE_ARB_fragment_program - struct gl_program *DefaultFragmentProgram; + struct gl_fragment_program *DefaultFragmentProgram; #endif /*@}*/ diff --git a/src/mesa/main/pixel.c b/src/mesa/main/pixel.c index eb4fd6e7c9..0e9915dd38 100644 --- a/src/mesa/main/pixel.c +++ b/src/mesa/main/pixel.c @@ -1341,6 +1341,21 @@ _mesa_scale_and_bias_depth(const GLcontext *ctx, GLuint n, } +void +_mesa_scale_and_bias_depth_uint(const GLcontext *ctx, GLuint n, + GLuint depthValues[]) +{ + const GLdouble max = (double) 0xffffffff; + const GLdouble scale = ctx->Pixel.DepthScale; + const GLdouble bias = ctx->Pixel.DepthBias * max; + GLuint i; + for (i = 0; i < n; i++) { + GLdouble d = (GLdouble) depthValues[i] * scale + bias; + d = CLAMP(d, 0.0, max); + depthValues[i] = (GLuint) d; + } +} + /**********************************************************************/ /***** State Management *****/ diff --git a/src/mesa/main/pixel.h b/src/mesa/main/pixel.h index 09155cfd70..3ba5b6689a 100644 --- a/src/mesa/main/pixel.h +++ b/src/mesa/main/pixel.h @@ -116,6 +116,9 @@ extern void _mesa_scale_and_bias_depth(const GLcontext *ctx, GLuint n, GLfloat depthValues[]); +extern void +_mesa_scale_and_bias_depth_uint(const GLcontext *ctx, GLuint n, + GLuint depthValues[]); extern void _mesa_update_pixel( GLcontext *ctx, GLuint newstate ); diff --git a/src/mesa/main/state.c b/src/mesa/main/state.c index 0b1c56fdd5..f8cb943e64 100644 --- a/src/mesa/main/state.c +++ b/src/mesa/main/state.c @@ -536,10 +536,10 @@ _mesa_init_exec_table(struct _glapi_table *exec) SET_GetVertexAttribPointervNV(exec, _mesa_GetVertexAttribPointervNV); SET_IsProgramNV(exec, _mesa_IsProgramARB); SET_LoadProgramNV(exec, _mesa_LoadProgramNV); - SET_ProgramParameter4dNV(exec, _mesa_ProgramParameter4dNV); - SET_ProgramParameter4dvNV(exec, _mesa_ProgramParameter4dvNV); - SET_ProgramParameter4fNV(exec, _mesa_ProgramParameter4fNV); - SET_ProgramParameter4fvNV(exec, _mesa_ProgramParameter4fvNV); + SET_ProgramEnvParameter4dARB(exec, _mesa_ProgramEnvParameter4dARB); /* alias to ProgramParameter4dNV */ + SET_ProgramEnvParameter4dvARB(exec, _mesa_ProgramEnvParameter4dvARB); /* alias to ProgramParameter4dvNV */ + SET_ProgramEnvParameter4fARB(exec, _mesa_ProgramEnvParameter4fARB); /* alias to ProgramParameter4fNV */ + SET_ProgramEnvParameter4fvARB(exec, _mesa_ProgramEnvParameter4fvARB); /* alias to ProgramParameter4fvNV */ SET_ProgramParameters4dvNV(exec, _mesa_ProgramParameters4dvNV); SET_ProgramParameters4fvNV(exec, _mesa_ProgramParameters4fvNV); SET_TrackMatrixNV(exec, _mesa_TrackMatrixNV); @@ -982,16 +982,20 @@ update_program(GLcontext *ctx) #endif if (shProg && shProg->LinkStatus && shProg->FragmentProgram) { /* user-defined fragment shader */ - ctx->FragmentProgram._Current = shProg->FragmentProgram; + _mesa_reference_fragprog(ctx, &ctx->FragmentProgram._Current, + shProg->FragmentProgram); } else if (ctx->FragmentProgram._Enabled) { /* use user-defined fragment program */ - ctx->FragmentProgram._Current = ctx->FragmentProgram.Current; + _mesa_reference_fragprog(ctx, &ctx->FragmentProgram._Current, + ctx->FragmentProgram.Current); } else if (ctx->FragmentProgram._MaintainTexEnvProgram) { /* fragment program generated from fixed-function state */ - ctx->FragmentProgram._Current = _mesa_get_fixed_func_fragment_program(ctx); - ctx->FragmentProgram._TexEnvProgram = ctx->FragmentProgram._Current; + _mesa_reference_fragprog(ctx, &ctx->FragmentProgram._Current, + _mesa_get_fixed_func_fragment_program(ctx)); + _mesa_reference_fragprog(ctx, &ctx->FragmentProgram._TexEnvProgram, + ctx->FragmentProgram._Current); /* XXX get rid of this confusing stuff someday? */ ctx->FragmentProgram._Active = ctx->FragmentProgram._Enabled; @@ -1000,7 +1004,7 @@ update_program(GLcontext *ctx) } else { /* no fragment program */ - ctx->FragmentProgram._Current = NULL; + _mesa_reference_fragprog(ctx, &ctx->FragmentProgram._Current, NULL); } if (ctx->FragmentProgram._Current != prevFP && ctx->Driver.BindProgram) { @@ -1013,29 +1017,33 @@ update_program(GLcontext *ctx) **/ #if 1 /* XXX get rid of this someday? */ - ctx->VertexProgram._TnlProgram = NULL; + _mesa_reference_vertprog(ctx, &ctx->VertexProgram._TnlProgram, NULL); #endif if (shProg && shProg->LinkStatus && shProg->VertexProgram) { /* user-defined vertex shader */ - ctx->VertexProgram._Current = shProg->VertexProgram; + _mesa_reference_vertprog(ctx, &ctx->VertexProgram._Current, + shProg->VertexProgram); } else if (ctx->VertexProgram._Enabled) { /* use user-defined vertex program */ - ctx->VertexProgram._Current = ctx->VertexProgram.Current; + _mesa_reference_vertprog(ctx, &ctx->VertexProgram._Current, + ctx->VertexProgram.Current); } else if (ctx->VertexProgram._MaintainTnlProgram) { /* vertex program generated from fixed-function state */ - ctx->VertexProgram._Current = _mesa_get_fixed_func_vertex_program(ctx); - ctx->VertexProgram._TnlProgram = ctx->VertexProgram._Current; + _mesa_reference_vertprog(ctx, &ctx->VertexProgram._Current, + _mesa_get_fixed_func_vertex_program(ctx)); + _mesa_reference_vertprog(ctx, &ctx->VertexProgram._TnlProgram, + ctx->VertexProgram._Current); } else { /* no vertex program / used fixed-function code */ - ctx->VertexProgram._Current = NULL; + _mesa_reference_vertprog(ctx, &ctx->VertexProgram._Current, NULL); } if (ctx->VertexProgram._Current != prevVP && ctx->Driver.BindProgram) { ctx->Driver.BindProgram(ctx, GL_VERTEX_PROGRAM_ARB, - (struct gl_program *) ctx->VertexProgram._Current); + (struct gl_program *) ctx->VertexProgram._Current); } } diff --git a/src/mesa/main/texstate.c b/src/mesa/main/texstate.c index cb7da39b51..84acfbd92c 100644 --- a/src/mesa/main/texstate.c +++ b/src/mesa/main/texstate.c @@ -238,6 +238,9 @@ calculate_derived_texenv( struct gl_tex_env_combine_state *state, return; } + if (mode == GL_REPLACE_EXT) + mode = GL_REPLACE; + switch (mode) { case GL_REPLACE: case GL_MODULATE: @@ -340,7 +343,9 @@ _mesa_TexEnvfv( GLenum target, GLenum pname, const GLfloat *param ) switch (pname) { case GL_TEXTURE_ENV_MODE: { - const GLenum mode = (GLenum) (GLint) *param; + GLenum mode = (GLenum) (GLint) *param; + if (mode == GL_REPLACE_EXT) + mode = GL_REPLACE; if (texUnit->EnvMode == mode) return; if (mode == GL_MODULATE || diff --git a/src/mesa/shader/arbprogram.c b/src/mesa/shader/arbprogram.c index 4b5f63ef92..1656dc9450 100644 --- a/src/mesa/shader/arbprogram.c +++ b/src/mesa/shader/arbprogram.c @@ -229,7 +229,7 @@ _mesa_ProgramStringARB(GLenum target, GLenum format, GLsizei len, struct gl_vertex_program *prog = ctx->VertexProgram.Current; _mesa_parse_arb_vertex_program(ctx, target, string, len, prog); - if (ctx->Driver.ProgramStringNotify) + if (ctx->Program.ErrorPos == -1 && ctx->Driver.ProgramStringNotify) ctx->Driver.ProgramStringNotify( ctx, target, &prog->Base ); } else if (target == GL_FRAGMENT_PROGRAM_ARB @@ -237,7 +237,7 @@ _mesa_ProgramStringARB(GLenum target, GLenum format, GLsizei len, struct gl_fragment_program *prog = ctx->FragmentProgram.Current; _mesa_parse_arb_fragment_program(ctx, target, string, len, prog); - if (ctx->Driver.ProgramStringNotify) + if (ctx->Program.ErrorPos == -1 && ctx->Driver.ProgramStringNotify) ctx->Driver.ProgramStringNotify( ctx, target, &prog->Base ); } else { @@ -247,6 +247,12 @@ _mesa_ProgramStringARB(GLenum target, GLenum format, GLsizei len, } +/** + * Set a program env parameter register. + * \note Called from the GL API dispatcher. + * Note, this function is also used by the GL_NV_vertex_program extension + * (alias to ProgramParameterdNV) + */ void GLAPIENTRY _mesa_ProgramEnvParameter4dARB(GLenum target, GLuint index, GLdouble x, GLdouble y, GLdouble z, GLdouble w) @@ -256,6 +262,12 @@ _mesa_ProgramEnvParameter4dARB(GLenum target, GLuint index, } +/** + * Set a program env parameter register. + * \note Called from the GL API dispatcher. + * Note, this function is also used by the GL_NV_vertex_program extension + * (alias to ProgramParameterdvNV) + */ void GLAPIENTRY _mesa_ProgramEnvParameter4dvARB(GLenum target, GLuint index, const GLdouble *params) @@ -266,6 +278,12 @@ _mesa_ProgramEnvParameter4dvARB(GLenum target, GLuint index, } +/** + * Set a program env parameter register. + * \note Called from the GL API dispatcher. + * Note, this function is also used by the GL_NV_vertex_program extension + * (alias to ProgramParameterfNV) + */ void GLAPIENTRY _mesa_ProgramEnvParameter4fARB(GLenum target, GLuint index, GLfloat x, GLfloat y, GLfloat z, GLfloat w) @@ -283,8 +301,8 @@ _mesa_ProgramEnvParameter4fARB(GLenum target, GLuint index, } ASSIGN_4V(ctx->FragmentProgram.Parameters[index], x, y, z, w); } - else if (target == GL_VERTEX_PROGRAM_ARB - && ctx->Extensions.ARB_vertex_program) { + else if (target == GL_VERTEX_PROGRAM_ARB /* == GL_VERTEX_PROGRAM_NV */ + && (ctx->Extensions.ARB_vertex_program || ctx->Extensions.NV_vertex_program)) { if (index >= ctx->Const.VertexProgram.MaxEnvParams) { _mesa_error(ctx, GL_INVALID_VALUE, "glProgramEnvParameter(index)"); return; @@ -297,7 +315,12 @@ _mesa_ProgramEnvParameter4fARB(GLenum target, GLuint index, } } - +/** + * Set a program env parameter register. + * \note Called from the GL API dispatcher. + * Note, this function is also used by the GL_NV_vertex_program extension + * (alias to ProgramParameterfvNV) + */ void GLAPIENTRY _mesa_ProgramEnvParameter4fvARB(GLenum target, GLuint index, const GLfloat *params) diff --git a/src/mesa/shader/arbprogram.syn b/src/mesa/shader/arbprogram.syn index ba17d638a3..b12c6a0eda 100644 --- a/src/mesa/shader/arbprogram.syn +++ b/src/mesa/shader/arbprogram.syn @@ -1942,10 +1942,10 @@ stateTexEnvProperty fragment program <optLegacyTexUnitNum> ::= "" | "[" <legacyTexUnitNum> "]" - -NOTE: <optLegaceTexUnitNum> is not optional. */ optLegacyTexUnitNum + optLegacyTexUnitNum_1 .or .true .emit 0x00; +optLegacyTexUnitNum_1 lbracket_ne .and legacyTexUnitNum .and rbracket; /* diff --git a/src/mesa/shader/arbprogram_syn.h b/src/mesa/shader/arbprogram_syn.h index e0c901ea8c..d95a5dede4 100644 --- a/src/mesa/shader/arbprogram_syn.h +++ b/src/mesa/shader/arbprogram_syn.h @@ -926,6 +926,8 @@ "stateTexEnvProperty\n" " \"color\" .emit TEX_ENV_COLOR;\n" "optLegacyTexUnitNum\n" +" optLegacyTexUnitNum_1 .or .true .emit 0x00;\n" +"optLegacyTexUnitNum_1\n" " lbracket_ne .and legacyTexUnitNum .and rbracket;\n" "legacyTexUnitNum\n" " integer;\n" diff --git a/src/mesa/shader/atifragshader.c b/src/mesa/shader/atifragshader.c index 4727c1a436..854c911874 100644 --- a/src/mesa/shader/atifragshader.c +++ b/src/mesa/shader/atifragshader.c @@ -440,7 +440,7 @@ _mesa_PassTexCoordATI(GLuint dst, GLuint coord, GLenum swizzle) _mesa_error(ctx, GL_INVALID_OPERATION, "glPassTexCoordATI(coord)"); return; } - if ((swizzle < GL_SWIZZLE_STR_ATI) && (swizzle > GL_SWIZZLE_STQ_DQ_ATI)) { + if (!(swizzle >= GL_SWIZZLE_STR_ATI) && (swizzle <= GL_SWIZZLE_STQ_DQ_ATI)) { _mesa_error(ctx, GL_INVALID_ENUM, "glPassTexCoordATI(swizzle)"); return; } @@ -513,7 +513,7 @@ _mesa_SampleMapATI(GLuint dst, GLuint interp, GLenum swizzle) _mesa_error(ctx, GL_INVALID_OPERATION, "glSampleMapATI(interp)"); return; } - if ((swizzle < GL_SWIZZLE_STR_ATI) && (swizzle > GL_SWIZZLE_STQ_DQ_ATI)) { + if (!(swizzle >= GL_SWIZZLE_STR_ATI) && (swizzle <= GL_SWIZZLE_STQ_DQ_ATI)) { _mesa_error(ctx, GL_INVALID_ENUM, "glSampleMapATI(swizzle)"); return; } diff --git a/src/mesa/shader/nvprogram.c b/src/mesa/shader/nvprogram.c index 337784bfe2..409c61cdc1 100644 --- a/src/mesa/shader/nvprogram.c +++ b/src/mesa/shader/nvprogram.c @@ -576,75 +576,6 @@ _mesa_LoadProgramNV(GLenum target, GLuint id, GLsizei len, /** - * Set a program parameter register. - * \note Called from the GL API dispatcher. - */ -void GLAPIENTRY -_mesa_ProgramParameter4dNV(GLenum target, GLuint index, - GLdouble x, GLdouble y, GLdouble z, GLdouble w) -{ - _mesa_ProgramParameter4fNV(target, index, - (GLfloat)x, (GLfloat)y, (GLfloat)z, (GLfloat)w); -} - - -/** - * Set a program parameter register. - * \note Called from the GL API dispatcher. - */ -void GLAPIENTRY -_mesa_ProgramParameter4dvNV(GLenum target, GLuint index, - const GLdouble *params) -{ - _mesa_ProgramParameter4fNV(target, index, - (GLfloat)params[0], (GLfloat)params[1], - (GLfloat)params[2], (GLfloat)params[3]); -} - - -/** - * Set a program parameter register. - * \note Called from the GL API dispatcher. - */ -void GLAPIENTRY -_mesa_ProgramParameter4fNV(GLenum target, GLuint index, - GLfloat x, GLfloat y, GLfloat z, GLfloat w) -{ - GET_CURRENT_CONTEXT(ctx); - ASSERT_OUTSIDE_BEGIN_END(ctx); - - if (target == GL_VERTEX_PROGRAM_NV && ctx->Extensions.NV_vertex_program) { - if (index < MAX_NV_VERTEX_PROGRAM_PARAMS) { - FLUSH_VERTICES(ctx, _NEW_PROGRAM); - ASSIGN_4V(ctx->VertexProgram.Parameters[index], x, y, z, w); - } - else { - _mesa_error(ctx, GL_INVALID_VALUE, "glProgramParameterNV(index)"); - return; - } - } - else { - _mesa_error(ctx, GL_INVALID_ENUM, "glProgramParameterNV"); - return; - } -} - - -/** - * Set a program parameter register. - * \note Called from the GL API dispatcher. - */ -void GLAPIENTRY -_mesa_ProgramParameter4fvNV(GLenum target, GLuint index, - const GLfloat *params) -{ - _mesa_ProgramParameter4fNV(target, index, - params[0], params[1], params[2], params[3]); -} - - - -/** * Set a sequence of program parameter registers. * \note Called from the GL API dispatcher. */ diff --git a/src/mesa/shader/nvprogram.h b/src/mesa/shader/nvprogram.h index dcea7727e0..bfac165b5e 100644 --- a/src/mesa/shader/nvprogram.h +++ b/src/mesa/shader/nvprogram.h @@ -70,18 +70,6 @@ extern void GLAPIENTRY _mesa_LoadProgramNV(GLenum target, GLuint id, GLsizei len, const GLubyte *program); extern void GLAPIENTRY -_mesa_ProgramParameter4dNV(GLenum target, GLuint index, GLdouble x, GLdouble y, GLdouble z, GLdouble w); - -extern void GLAPIENTRY -_mesa_ProgramParameter4dvNV(GLenum target, GLuint index, const GLdouble *params); - -extern void GLAPIENTRY -_mesa_ProgramParameter4fNV(GLenum target, GLuint index, GLfloat x, GLfloat y, GLfloat z, GLfloat w); - -extern void GLAPIENTRY -_mesa_ProgramParameter4fvNV(GLenum target, GLuint index, const GLfloat *params); - -extern void GLAPIENTRY _mesa_ProgramParameters4dvNV(GLenum target, GLuint index, GLuint num, const GLdouble *params); extern void GLAPIENTRY diff --git a/src/mesa/shader/prog_cache.c b/src/mesa/shader/prog_cache.c index dd0241ef24..36a25377c5 100644 --- a/src/mesa/shader/prog_cache.c +++ b/src/mesa/shader/prog_cache.c @@ -30,6 +30,7 @@ #include "main/mtypes.h" #include "main/imports.h" #include "shader/prog_cache.h" +#include "shader/program.h" struct cache_item @@ -109,7 +110,7 @@ clear_cache(GLcontext *ctx, struct gl_program_cache *cache) for (c = cache->items[i]; c; c = next) { next = c->next; _mesa_free(c->key); - ctx->Driver.DeleteProgram(ctx, c->program); + _mesa_reference_program(ctx, &c->program, NULL); _mesa_free(c); } cache->items[i] = NULL; @@ -177,7 +178,7 @@ _mesa_program_cache_insert(GLcontext *ctx, c->key = _mesa_malloc(keysize); memcpy(c->key, key, keysize); - c->program = program; + c->program = program; /* no refcount change */ if (cache->n_items > cache->size * 1.5) { if (cache->size < 1000) diff --git a/src/mesa/shader/prog_execute.c b/src/mesa/shader/prog_execute.c index bd64b57eb9..8ce2ca3964 100644 --- a/src/mesa/shader/prog_execute.c +++ b/src/mesa/shader/prog_execute.c @@ -1,6 +1,6 @@ /* * Mesa 3-D graphics library - * Version: 6.5.3 + * Version: 7.0.3 * * Copyright (C) 1999-2007 Brian Paul All Rights Reserved. * @@ -46,9 +46,6 @@ #include "shader/slang/slang_library_noise.h" -/* See comments below for info about this */ -#define LAMBDA_ZERO 1 - /* debug predicate */ #define DEBUG_PROG 0 @@ -92,6 +89,7 @@ get_register_pointer(const struct prog_src_register *source, else { const struct gl_program_parameter_list *params; ASSERT(source->File == PROGRAM_LOCAL_PARAM || + source->File == PROGRAM_CONSTANT || source->File == PROGRAM_STATE_VAR); params = machine->CurProgram->Parameters; if (reg < 0 || reg >= params->NumParameters) @@ -303,6 +301,36 @@ fetch_vector1(const struct prog_src_register *source, /** + * Fetch texel from texture. Use partial derivatives when possible. + */ +static INLINE void +fetch_texel(GLcontext *ctx, + const struct gl_program_machine *machine, + const struct prog_instruction *inst, + const GLfloat texcoord[4], GLfloat lodBias, + GLfloat color[4]) +{ + const GLuint unit = machine->Samplers[inst->TexSrcUnit]; + + /* Note: we only have the right derivatives for fragment input attribs. + */ + if (machine->NumDeriv > 0 && + inst->SrcReg[0].File == PROGRAM_INPUT && + inst->SrcReg[0].Index == FRAG_ATTRIB_TEX0 + inst->TexSrcUnit) { + /* simple texture fetch for which we should have derivatives */ + GLuint attr = inst->SrcReg[0].Index; + machine->FetchTexelDeriv(ctx, texcoord, + machine->DerivX[attr], + machine->DerivY[attr], + lodBias, unit, color); + } + else { + machine->FetchTexelLod(ctx, texcoord, lodBias, unit, color); + } +} + + +/** * Test value against zero and return GT, LT, EQ or UN if NaN. */ static INLINE GLuint @@ -1306,33 +1334,18 @@ _mesa_execute_program(GLcontext * ctx, } break; case OPCODE_TEX: /* Both ARB and NV frag prog */ - /* Texel lookup */ + /* Simple texel lookup */ { - /* Note: only use the precomputed lambda value when we're - * sampling texture unit [K] with texcoord[K]. - * Otherwise, the lambda value may have no relation to the - * instruction's texcoord or texture image. Using the wrong - * lambda is usually bad news. - * The rest of the time, just use zero (until we get a more - * sophisticated way of computing lambda). - */ - const GLuint unit = machine->Samplers[inst->TexSrcUnit]; - GLfloat coord[4], color[4], lambda; -#if 0 - if (inst->SrcReg[0].File == PROGRAM_INPUT && - inst->SrcReg[0].Index == FRAG_ATTRIB_TEX0 + unit) - lambda = span->array->lambda[unit][column]; - else -#endif - lambda = 0.0; - fetch_vector4(&inst->SrcReg[0], machine, coord); - machine->FetchTexelLod(ctx, coord, lambda, unit, color); + GLfloat texcoord[4], color[4]; + fetch_vector4(&inst->SrcReg[0], machine, texcoord); + + fetch_texel(ctx, machine, inst, texcoord, 0.0, color); + if (DEBUG_PROG) { - printf("TEX (%g, %g, %g, %g) = texture[%d][%g, %g, %g, %g], " - "lod %f\n", + printf("TEX (%g, %g, %g, %g) = texture[%d][%g, %g, %g, %g]\n", color[0], color[1], color[2], color[3], - unit, - coord[0], coord[1], coord[2], coord[3], lambda); + inst->TexSrcUnit, + texcoord[0], texcoord[1], texcoord[2], texcoord[3]); } store_vector4(inst, machine, color); } @@ -1342,21 +1355,18 @@ _mesa_execute_program(GLcontext * ctx, { const struct gl_texture_unit *texUnit = &ctx->Texture.Unit[inst->TexSrcUnit]; - GLfloat coord[4], color[4], lambda, bias; -#if 0 - if (inst->SrcReg[0].File == PROGRAM_INPUT && - inst->SrcReg[0].Index == FRAG_ATTRIB_TEX0 + inst->TexSrcUnit) - lambda = span->array->lambda[inst->TexSrcUnit][column]; - else -#endif - lambda = 0.0; - fetch_vector4(&inst->SrcReg[0], machine, coord); - /* coord[3] is the bias to add to lambda */ - bias = texUnit->LodBias + coord[3]; - if (texUnit->_Current) - bias += texUnit->_Current->LodBias; - machine->FetchTexelLod(ctx, coord, lambda + bias, - inst->TexSrcUnit, color); + GLfloat texcoord[4], color[4], lodBias; + + fetch_vector4(&inst->SrcReg[0], machine, texcoord); + + /* texcoord[3] is the bias to add to lambda */ + lodBias = texUnit->LodBias + texcoord[3]; + if (texUnit->_Current) { + lodBias += texUnit->_Current->LodBias; + } + + fetch_texel(ctx, machine, inst, texcoord, lodBias, color); + store_vector4(inst, machine, color); } break; @@ -1368,6 +1378,7 @@ _mesa_execute_program(GLcontext * ctx, fetch_vector4(&inst->SrcReg[1], machine, dtdx); fetch_vector4(&inst->SrcReg[2], machine, dtdy); machine->FetchTexelDeriv(ctx, texcoord, dtdx, dtdy, + 0.0, /* lodBias */ inst->TexSrcUnit, color); store_vector4(inst, machine, color); } @@ -1375,15 +1386,8 @@ _mesa_execute_program(GLcontext * ctx, case OPCODE_TXP: /* GL_ARB_fragment_program only */ /* Texture lookup w/ projective divide */ { - const GLuint unit = machine->Samplers[inst->TexSrcUnit]; - GLfloat texcoord[4], color[4], lambda; -#if 0 - if (inst->SrcReg[0].File == PROGRAM_INPUT && - inst->SrcReg[0].Index == FRAG_ATTRIB_TEX0 + inst->TexSrcUnit) - lambda = span->array->lambda[unit][column]; - else -#endif - lambda = 0.0; + GLfloat texcoord[4], color[4]; + fetch_vector4(&inst->SrcReg[0], machine, texcoord); /* Not so sure about this test - if texcoord[3] is * zero, we'd probably be fine except for an ASSERT in @@ -1394,21 +1398,19 @@ _mesa_execute_program(GLcontext * ctx, texcoord[1] /= texcoord[3]; texcoord[2] /= texcoord[3]; } - machine->FetchTexelLod(ctx, texcoord, lambda, unit, color); + + fetch_texel(ctx, machine, inst, texcoord, 0.0, color); + store_vector4(inst, machine, color); } break; case OPCODE_TXP_NV: /* GL_NV_fragment_program only */ - /* Texture lookup w/ projective divide */ + /* Texture lookup w/ projective divide, as above, but do not + * do the divide by w if sampling from a cube map. + */ { - GLfloat texcoord[4], color[4], lambda; -#if 0 - if (inst->SrcReg[0].File == PROGRAM_INPUT && - inst->SrcReg[0].Index == FRAG_ATTRIB_TEX0 + inst->TexSrcUnit) - lambda = span->array->lambda[inst->TexSrcUnit][column]; - else -#endif - lambda = 0.0; + GLfloat texcoord[4], color[4]; + fetch_vector4(&inst->SrcReg[0], machine, texcoord); if (inst->TexSrcTarget != TEXTURE_CUBE_INDEX && texcoord[3] != 0.0) { @@ -1416,8 +1418,9 @@ _mesa_execute_program(GLcontext * ctx, texcoord[1] /= texcoord[3]; texcoord[2] /= texcoord[3]; } - machine->FetchTexelLod(ctx, texcoord, lambda, - inst->TexSrcUnit, color); + + fetch_texel(ctx, machine, inst, texcoord, 0.0, color); + store_vector4(inst, machine, color); } break; @@ -1517,10 +1520,9 @@ _mesa_execute_program(GLcontext * ctx, case OPCODE_END: return GL_TRUE; default: - _mesa_problem(ctx, "Bad opcode %d in _mesa_exec_fragment_program", + _mesa_problem(ctx, "Bad opcode %d in _mesa_execute_program", inst->Opcode); return GL_TRUE; /* return value doesn't matter */ - } numExec++; diff --git a/src/mesa/shader/prog_execute.h b/src/mesa/shader/prog_execute.h index af77d4278b..18b13e11a4 100644 --- a/src/mesa/shader/prog_execute.h +++ b/src/mesa/shader/prog_execute.h @@ -1,6 +1,6 @@ /* * Mesa 3-D graphics library - * Version: 6.5.3 + * Version: 7.0.3 * * Copyright (C) 1999-2007 Brian Paul All Rights Reserved. * @@ -32,6 +32,7 @@ typedef void (*FetchTexelLodFunc)(GLcontext *ctx, const GLfloat texcoord[4], typedef void (*FetchTexelDerivFunc)(GLcontext *ctx, const GLfloat texcoord[4], const GLfloat texdx[4], const GLfloat texdy[4], + GLfloat lodBias, GLuint unit, GLfloat color[4]); diff --git a/src/mesa/shader/prog_instruction.c b/src/mesa/shader/prog_instruction.c index 066129037a..bea5d0551e 100644 --- a/src/mesa/shader/prog_instruction.c +++ b/src/mesa/shader/prog_instruction.c @@ -119,6 +119,23 @@ _mesa_copy_instructions(struct prog_instruction *dest, /** + * Free an array of instructions + */ +void +_mesa_free_instructions(struct prog_instruction *inst, GLuint count) +{ + GLuint i; + for (i = 0; i < count; i++) { + if (inst[i].Data) + _mesa_free(inst[i].Data); + if (inst[i].Comment) + _mesa_free((char *) inst[i].Comment); + } + _mesa_free(inst); +} + + +/** * Basic info about each instruction */ struct instruction_info diff --git a/src/mesa/shader/prog_instruction.h b/src/mesa/shader/prog_instruction.h index e8a2407ea8..711166f9dd 100644 --- a/src/mesa/shader/prog_instruction.h +++ b/src/mesa/shader/prog_instruction.h @@ -439,6 +439,9 @@ extern struct prog_instruction * _mesa_copy_instructions(struct prog_instruction *dest, const struct prog_instruction *src, GLuint n); +extern void +_mesa_free_instructions(struct prog_instruction *inst, GLuint count); + extern GLuint _mesa_num_inst_src_regs(gl_inst_opcode opcode); diff --git a/src/mesa/shader/prog_parameter.c b/src/mesa/shader/prog_parameter.c index 4eb7313bec..152bd79f69 100644 --- a/src/mesa/shader/prog_parameter.c +++ b/src/mesa/shader/prog_parameter.c @@ -40,8 +40,7 @@ struct gl_program_parameter_list * _mesa_new_parameter_list(void) { - return (struct gl_program_parameter_list *) - _mesa_calloc(sizeof(struct gl_program_parameter_list)); + return CALLOC_STRUCT(gl_program_parameter_list); } @@ -284,25 +283,31 @@ _mesa_add_uniform(struct gl_program_parameter_list *paramList, * \param name uniform's name * \param datatype GL_SAMPLER_2D, GL_SAMPLER_2D_RECT_ARB, etc. * \param index the sampler number (as seen in TEX instructions) + * \return sampler index (starting at zero) or -1 if error */ GLint _mesa_add_sampler(struct gl_program_parameter_list *paramList, - const char *name, GLenum datatype, GLuint index) + const char *name, GLenum datatype) { GLint i = _mesa_lookup_parameter_index(paramList, -1, name); if (i >= 0 && paramList->Parameters[i].Type == PROGRAM_SAMPLER) { ASSERT(paramList->Parameters[i].Size == 1); ASSERT(paramList->Parameters[i].DataType == datatype); - ASSERT(paramList->ParameterValues[i][0] == index); /* already in list */ - return i; + return (GLint) paramList->ParameterValues[i][0]; } else { - GLfloat indexf = index; const GLint size = 1; /* a sampler is basically a texture unit number */ - i = _mesa_add_parameter(paramList, PROGRAM_SAMPLER, name, - size, datatype, &indexf, NULL); - return i; + GLfloat value; + GLint numSamplers = 0; + for (i = 0; i < paramList->NumParameters; i++) { + if (paramList->Parameters[i].Type == PROGRAM_SAMPLER) + numSamplers++; + } + value = (GLfloat) numSamplers; + (void) _mesa_add_parameter(paramList, PROGRAM_SAMPLER, name, + size, datatype, &value, NULL); + return numSamplers; } } @@ -603,6 +608,8 @@ _mesa_clone_parameter_list(const struct gl_program_parameter_list *list) } } + clone->StateFlags = list->StateFlags; + return clone; } diff --git a/src/mesa/shader/prog_parameter.h b/src/mesa/shader/prog_parameter.h index d6cc03448c..105f6f24de 100644 --- a/src/mesa/shader/prog_parameter.h +++ b/src/mesa/shader/prog_parameter.h @@ -114,7 +114,7 @@ _mesa_add_uniform(struct gl_program_parameter_list *paramList, extern GLint _mesa_add_sampler(struct gl_program_parameter_list *paramList, - const char *name, GLenum datatype, GLuint index); + const char *name, GLenum datatype); extern GLint _mesa_add_varying(struct gl_program_parameter_list *paramList, diff --git a/src/mesa/shader/prog_print.c b/src/mesa/shader/prog_print.c index 1145f0c7a4..1c35ce3fec 100644 --- a/src/mesa/shader/prog_print.c +++ b/src/mesa/shader/prog_print.c @@ -99,7 +99,23 @@ arb_input_attrib_string(GLint index, GLenum progType) "vertex.texcoord[4]", "vertex.texcoord[5]", "vertex.texcoord[6]", - "vertex.texcoord[7]" + "vertex.texcoord[7]", + "vertex.attrib[0]", + "vertex.attrib[1]", + "vertex.attrib[2]", + "vertex.attrib[3]", + "vertex.attrib[4]", + "vertex.attrib[5]", + "vertex.attrib[6]", + "vertex.attrib[7]", + "vertex.attrib[8]", + "vertex.attrib[9]", + "vertex.attrib[10]", + "vertex.attrib[11]", + "vertex.attrib[12]", + "vertex.attrib[13]", + "vertex.attrib[14]", + "vertex.attrib[15]" }; const char *fragAttribs[] = { "fragment.position", @@ -751,9 +767,9 @@ _mesa_print_program_parameters(GLcontext *ctx, const struct gl_program *prog) _mesa_printf("%d ", prog->SamplerUnits[i]); } _mesa_printf("]\n"); - + _mesa_load_state_parameters(ctx, prog->Parameters); - + #if 0 _mesa_printf("Local Params:\n"); for (i = 0; i < MAX_PROGRAM_LOCAL_PARAMS; i++){ diff --git a/src/mesa/shader/prog_statevars.c b/src/mesa/shader/prog_statevars.c index 6cc490bc77..ba3c988445 100644 --- a/src/mesa/shader/prog_statevars.c +++ b/src/mesa/shader/prog_statevars.c @@ -1,6 +1,6 @@ /* * Mesa 3-D graphics library - * Version: 7.0 + * Version: 7.1 * * Copyright (C) 1999-2007 Brian Paul All Rights Reserved. * @@ -181,7 +181,7 @@ _mesa_fetch_state(GLcontext *ctx, const gl_state_index state[], ctx->Light.Material.Attrib[MAT_ATTRIB_FRONT_AMBIENT+face][i]; } /* [3] = material alpha */ - value[3] = ctx->Light.Material.Attrib[MAT_ATTRIB_FRONT_DIFFUSE+face][3]; + value[3] = ctx->Light.Material.Attrib[MAT_ATTRIB_FRONT_AMBIENT+face][3]; return; case STATE_DIFFUSE: for (i = 0; i < 3; i++) { @@ -197,7 +197,7 @@ _mesa_fetch_state(GLcontext *ctx, const gl_state_index state[], ctx->Light.Material.Attrib[MAT_ATTRIB_FRONT_SPECULAR+face][i]; } /* [3] = material alpha */ - value[3] = ctx->Light.Material.Attrib[MAT_ATTRIB_FRONT_DIFFUSE+face][3]; + value[3] = ctx->Light.Material.Attrib[MAT_ATTRIB_FRONT_SPECULAR+face][3]; return; default: _mesa_problem(ctx, "Invalid lightprod state in fetch_state"); @@ -253,7 +253,8 @@ _mesa_fetch_state(GLcontext *ctx, const gl_state_index state[], value[0] = ctx->Fog.Density; value[1] = ctx->Fog.Start; value[2] = ctx->Fog.End; - value[3] = 1.0F / (ctx->Fog.End - ctx->Fog.Start); + value[3] = (ctx->Fog.End == ctx->Fog.Start) + ? 1.0 : 1.0F / (ctx->Fog.End - ctx->Fog.Start); return; case STATE_CLIPPLANE: { @@ -353,7 +354,7 @@ _mesa_fetch_state(GLcontext *ctx, const gl_state_index state[], value[0] = ctx->Viewport.Near; /* near */ value[1] = ctx->Viewport.Far; /* far */ value[2] = ctx->Viewport.Far - ctx->Viewport.Near; /* far - near */ - value[3] = 0; + value[3] = 1.0; return; case STATE_FRAGMENT_PROGRAM: { @@ -424,8 +425,9 @@ _mesa_fetch_state(GLcontext *ctx, const gl_state_index state[], * exp: 2^-(density/ln(2) * fogcoord) * exp2: 2^-((density/(ln(2)^2) * fogcoord)^2) */ - value[0] = -1.0F / (ctx->Fog.End - ctx->Fog.Start); - value[1] = ctx->Fog.End / (ctx->Fog.End - ctx->Fog.Start); + value[0] = (ctx->Fog.End == ctx->Fog.Start) + ? 1.0 : -1.0F / (ctx->Fog.End - ctx->Fog.Start); + value[1] = ctx->Fog.End * -value[0]; value[2] = ctx->Fog.Density * ONE_DIV_LN2; value[3] = ctx->Fog.Density * ONE_DIV_SQRT_LN2; return; @@ -450,6 +452,12 @@ _mesa_fetch_state(GLcontext *ctx, const gl_state_index state[], value[2] = ctx->Pixel.BlueBias; value[3] = ctx->Pixel.AlphaBias; break; + case STATE_PCM_SCALE: + COPY_4V(value, ctx->Pixel.PostColorMatrixScale); + break; + case STATE_PCM_BIAS: + COPY_4V(value, ctx->Pixel.PostColorMatrixBias); + break; default: /* unknown state indexes are silently ignored * should be handled by the driver. @@ -697,6 +705,12 @@ append_token(char *dst, gl_state_index k) case STATE_PT_BIAS: append(dst, "PTbias"); break; + case STATE_PCM_SCALE: + append(dst, "PCMscale"); + break; + case STATE_PCM_BIAS: + append(dst, "PCMbias"); + break; default: ; } diff --git a/src/mesa/shader/prog_statevars.h b/src/mesa/shader/prog_statevars.h index 9fedcfab8a..d12142055f 100644 --- a/src/mesa/shader/prog_statevars.h +++ b/src/mesa/shader/prog_statevars.h @@ -111,6 +111,8 @@ typedef enum gl_state_index_ { STATE_SPOT_DIR_NORMALIZED, /* pre-normalized spot dir */ STATE_PT_SCALE, /**< Pixel transfer RGBA scale */ STATE_PT_BIAS, /**< Pixel transfer RGBA bias */ + STATE_PCM_SCALE, /**< Post color matrix RGBA scale */ + STATE_PCM_BIAS, /**< Post color matrix RGBA bias */ STATE_INTERNAL_DRIVER /* first available state index for drivers (must be last) */ } gl_state_index; diff --git a/src/mesa/shader/program.c b/src/mesa/shader/program.c index 3069b04836..a0817a91ec 100644 --- a/src/mesa/shader/program.c +++ b/src/mesa/shader/program.c @@ -60,9 +60,9 @@ _mesa_init_program(GLcontext *ctx) ctx->VertexProgram.Enabled = GL_FALSE; ctx->VertexProgram.PointSizeEnabled = GL_FALSE; ctx->VertexProgram.TwoSideEnabled = GL_FALSE; - ctx->VertexProgram.Current = (struct gl_vertex_program *) ctx->Shared->DefaultVertexProgram; + _mesa_reference_vertprog(ctx, &ctx->VertexProgram.Current, + ctx->Shared->DefaultVertexProgram); assert(ctx->VertexProgram.Current); - ctx->VertexProgram.Current->Base.RefCount++; for (i = 0; i < MAX_NV_VERTEX_PROGRAM_PARAMS / 4; i++) { ctx->VertexProgram.TrackMatrix[i] = GL_NONE; ctx->VertexProgram.TrackMatrixTransform[i] = GL_IDENTITY_NV; @@ -72,9 +72,9 @@ _mesa_init_program(GLcontext *ctx) #if FEATURE_NV_fragment_program || FEATURE_ARB_fragment_program ctx->FragmentProgram.Enabled = GL_FALSE; - ctx->FragmentProgram.Current = (struct gl_fragment_program *) ctx->Shared->DefaultFragmentProgram; + _mesa_reference_fragprog(ctx, &ctx->FragmentProgram.Current, + ctx->Shared->DefaultFragmentProgram); assert(ctx->FragmentProgram.Current); - ctx->FragmentProgram.Current->Base.RefCount++; ctx->FragmentProgram.Cache = _mesa_new_program_cache(); #endif @@ -96,19 +96,11 @@ void _mesa_free_program_data(GLcontext *ctx) { #if FEATURE_NV_vertex_program || FEATURE_ARB_vertex_program - if (ctx->VertexProgram.Current) { - ctx->VertexProgram.Current->Base.RefCount--; - if (ctx->VertexProgram.Current->Base.RefCount <= 0) - ctx->Driver.DeleteProgram(ctx, &(ctx->VertexProgram.Current->Base)); - } + _mesa_reference_vertprog(ctx, &ctx->VertexProgram.Current, NULL); _mesa_delete_program_cache(ctx, ctx->VertexProgram.Cache); #endif #if FEATURE_NV_fragment_program || FEATURE_ARB_fragment_program - if (ctx->FragmentProgram.Current) { - ctx->FragmentProgram.Current->Base.RefCount--; - if (ctx->FragmentProgram.Current->Base.RefCount <= 0) - ctx->Driver.DeleteProgram(ctx, &(ctx->FragmentProgram.Current->Base)); - } + _mesa_reference_fragprog(ctx, &ctx->FragmentProgram.Current, NULL); _mesa_delete_program_cache(ctx, ctx->FragmentProgram.Cache); #endif /* XXX probably move this stuff */ @@ -124,8 +116,6 @@ _mesa_free_program_data(GLcontext *ctx) } - - /** * Set the vertex/fragment program error state (position and error string). * This is generally called from within the parsers. @@ -280,6 +270,7 @@ _mesa_delete_program(GLcontext *ctx, struct gl_program *prog) { (void) ctx; ASSERT(prog); + ASSERT(prog->RefCount==0); if (prog == &_mesa_DummyProgram) return; @@ -287,16 +278,7 @@ _mesa_delete_program(GLcontext *ctx, struct gl_program *prog) if (prog->String) _mesa_free(prog->String); - if (prog->Instructions) { - GLuint i; - for (i = 0; i < prog->NumInstructions; i++) { - if (prog->Instructions[i].Data) - _mesa_free(prog->Instructions[i].Data); - if (prog->Instructions[i].Comment) - _mesa_free((char *) prog->Instructions[i].Comment); - } - _mesa_free(prog->Instructions); - } + _mesa_free_instructions(prog->Instructions, prog->NumInstructions); if (prog->Parameters) { _mesa_free_parameter_list(prog->Parameters); @@ -335,6 +317,59 @@ _mesa_lookup_program(GLcontext *ctx, GLuint id) /** + * Reference counting for vertex/fragment programs + */ +void +_mesa_reference_program(GLcontext *ctx, + struct gl_program **ptr, + struct gl_program *prog) +{ + assert(ptr); + if (*ptr && prog) { + /* sanity check */ + ASSERT((*ptr)->Target == prog->Target); + } + if (*ptr == prog) { + return; /* no change */ + } + if (*ptr) { + GLboolean deleteFlag; + + /*_glthread_LOCK_MUTEX((*ptr)->Mutex);*/ +#if 0 + printf("Program %p %u 0x%x Refcount-- to %d\n", + *ptr, (*ptr)->Id, (*ptr)->Target, (*ptr)->RefCount - 1); +#endif + ASSERT((*ptr)->RefCount > 0); + (*ptr)->RefCount--; + + deleteFlag = ((*ptr)->RefCount == 0); + /*_glthread_UNLOCK_MUTEX((*ptr)->Mutex);*/ + + if (deleteFlag) { + ASSERT(ctx); + ctx->Driver.DeleteProgram(ctx, *ptr); + } + + *ptr = NULL; + } + + assert(!*ptr); + if (prog) { + /*_glthread_LOCK_MUTEX(prog->Mutex);*/ + prog->RefCount++; +#if 0 + printf("Program %p %u 0x%x Refcount++ to %d\n", + prog, prog->Id, prog->Target, prog->RefCount); +#endif + /*_glthread_UNLOCK_MUTEX(prog->Mutex);*/ + } + + *ptr = prog; +} + + +/** * Return a copy of a program. * XXX Problem here if the program object is actually OO-derivation * made by a device driver. @@ -349,12 +384,13 @@ _mesa_clone_program(GLcontext *ctx, const struct gl_program *prog) return NULL; assert(clone->Target == prog->Target); + assert(clone->RefCount == 1); + clone->String = (GLubyte *) _mesa_strdup((char *) prog->String); - clone->RefCount = 1; clone->Format = prog->Format; clone->Instructions = _mesa_alloc_instructions(prog->NumInstructions); if (!clone->Instructions) { - ctx->Driver.DeleteProgram(ctx, clone); + _mesa_reference_program(ctx, &clone, NULL); return NULL; } _mesa_copy_instructions(clone->Instructions, prog->Instructions, @@ -415,6 +451,55 @@ _mesa_clone_program(GLcontext *ctx, const struct gl_program *prog) } +/** + * Insert 'count' NOP instructions at 'start' in the given program. + * Adjust branch targets accordingly. + */ +GLboolean +_mesa_insert_instructions(struct gl_program *prog, GLuint start, GLuint count) +{ + const GLuint origLen = prog->NumInstructions; + const GLuint newLen = origLen + count; + struct prog_instruction *newInst; + GLuint i; + + /* adjust branches */ + for (i = 0; i < prog->NumInstructions; i++) { + struct prog_instruction *inst = prog->Instructions + i; + if (inst->BranchTarget > 0) { + if (inst->BranchTarget >= start) { + inst->BranchTarget += count; + } + } + } + + /* Alloc storage for new instructions */ + newInst = _mesa_alloc_instructions(newLen); + if (!newInst) { + return GL_FALSE; + } + + /* Copy 'start' instructions into new instruction buffer */ + _mesa_copy_instructions(newInst, prog->Instructions, start); + + /* init the new instructions */ + _mesa_init_instructions(newInst + start, count); + + /* Copy the remaining/tail instructions to new inst buffer */ + _mesa_copy_instructions(newInst + start + count, + prog->Instructions + start, + origLen - start); + + /* free old instructions */ + _mesa_free_instructions(prog->Instructions, origLen); + + /* install new instructions */ + prog->Instructions = newInst; + prog->NumInstructions = newLen; + + return GL_TRUE; +} + /** * Search instructions for registers that match (oldFile, oldIndex), @@ -517,7 +602,7 @@ _mesa_combine_programs(GLcontext *ctx, if ((progA->OutputsWritten & (1 << FRAG_RESULT_COLR)) && (progB->InputsRead & (1 << FRAG_ATTRIB_COL0))) { GLint tempReg = _mesa_find_free_register(newProg, PROGRAM_TEMPORARY); - if (!tempReg) { + if (tempReg < 0) { _mesa_problem(ctx, "No free temp regs found in " "_mesa_combine_programs(), using 31"); tempReg = 31; @@ -664,9 +749,9 @@ _mesa_BindProgram(GLenum target, GLuint id) /* Bind a default program */ newProg = NULL; if (target == GL_VERTEX_PROGRAM_ARB) /* == GL_VERTEX_PROGRAM_NV */ - newProg = ctx->Shared->DefaultVertexProgram; + newProg = &ctx->Shared->DefaultVertexProgram->Base; else - newProg = ctx->Shared->DefaultFragmentProgram; + newProg = &ctx->Shared->DefaultFragmentProgram->Base; } else { /* Bind a user program */ @@ -694,26 +779,16 @@ _mesa_BindProgram(GLenum target, GLuint id) return; } - /* unbind/delete oldProg */ - if (curProg->Id != 0) { - /* decrement refcount on previously bound fragment program */ - curProg->RefCount--; - /* and delete if refcount goes below one */ - if (curProg->RefCount <= 0) { - /* the program ID was already removed from the hash table */ - ctx->Driver.DeleteProgram(ctx, curProg); - } - } - /* bind newProg */ if (target == GL_VERTEX_PROGRAM_ARB) { /* == GL_VERTEX_PROGRAM_NV */ - ctx->VertexProgram.Current = (struct gl_vertex_program *) newProg; + _mesa_reference_vertprog(ctx, &ctx->VertexProgram.Current, + (struct gl_vertex_program *) newProg); } else if (target == GL_FRAGMENT_PROGRAM_NV || target == GL_FRAGMENT_PROGRAM_ARB) { - ctx->FragmentProgram.Current = (struct gl_fragment_program *) newProg; + _mesa_reference_fragprog(ctx, &ctx->FragmentProgram.Current, + (struct gl_fragment_program *) newProg); } - newProg->RefCount++; /* Never null pointers */ ASSERT(ctx->VertexProgram.Current); @@ -771,10 +846,7 @@ _mesa_DeletePrograms(GLsizei n, const GLuint *ids) } /* The ID is immediately available for re-use now */ _mesa_HashRemove(ctx->Shared->Programs, ids[i]); - prog->RefCount--; - if (prog->RefCount <= 0) { - ctx->Driver.DeleteProgram(ctx, prog); - } + _mesa_reference_program(ctx, &prog, NULL); } } } diff --git a/src/mesa/shader/program.h b/src/mesa/shader/program.h index 4b7297e4c6..08fe576afc 100644 --- a/src/mesa/shader/program.h +++ b/src/mesa/shader/program.h @@ -83,10 +83,34 @@ _mesa_delete_program(GLcontext *ctx, struct gl_program *prog); extern struct gl_program * _mesa_lookup_program(GLcontext *ctx, GLuint id); +extern void +_mesa_reference_program(GLcontext *ctx, + struct gl_program **ptr, + struct gl_program *prog); + +static INLINE void +_mesa_reference_vertprog(GLcontext *ctx, + struct gl_vertex_program **ptr, + struct gl_vertex_program *prog) +{ + _mesa_reference_program(ctx, (struct gl_program **) ptr, + (struct gl_program *) prog); +} + +static INLINE void +_mesa_reference_fragprog(GLcontext *ctx, + struct gl_fragment_program **ptr, + struct gl_fragment_program *prog) +{ + _mesa_reference_program(ctx, (struct gl_program **) ptr, + (struct gl_program *) prog); +} extern struct gl_program * _mesa_clone_program(GLcontext *ctx, const struct gl_program *prog); +extern GLboolean +_mesa_insert_instructions(struct gl_program *prog, GLuint start, GLuint count); extern struct gl_program * _mesa_combine_programs(GLcontext *ctx, diff --git a/src/mesa/shader/programopt.c b/src/mesa/shader/programopt.c index 9eeb71db1b..7d560c74a5 100644 --- a/src/mesa/shader/programopt.c +++ b/src/mesa/shader/programopt.c @@ -35,6 +35,7 @@ #include "context.h" #include "prog_parameter.h" #include "prog_statevars.h" +#include "program.h" #include "programopt.h" #include "prog_instruction.h" @@ -102,7 +103,7 @@ _mesa_insert_mvp_code(GLcontext *ctx, struct gl_vertex_program *vprog) _mesa_copy_instructions (newInst + 4, vprog->Base.Instructions, origLen); /* free old instructions */ - _mesa_free(vprog->Base.Instructions); + _mesa_free_instructions(vprog->Base.Instructions, origLen); /* install new instructions */ vprog->Base.Instructions = newInst; @@ -274,7 +275,7 @@ _mesa_append_fog_code(GLcontext *ctx, struct gl_fragment_program *fprog) inst++; /* free old instructions */ - _mesa_free(fprog->Base.Instructions); + _mesa_free_instructions(fprog->Base.Instructions, origLen); /* install new instructions */ fprog->Base.Instructions = newInst; @@ -364,3 +365,94 @@ _mesa_count_texture_instructions(struct gl_program *prog) } } + +/** + * Scan/rewrite program to remove reads of varying (output) registers. + * In GLSL vertex shaders, varying vars can be read and written. + * Normally, vertex varying vars are implemented as output registers. + * On some hardware, trying to read an output register causes trouble. + * So, rewrite the program to use a temporary register in this case. + */ +void +_mesa_remove_varying_reads(struct gl_program *prog) +{ + GLuint i; + GLint outputMap[VERT_RESULT_MAX]; + GLuint numVaryingReads = 0; + + assert(prog->Target == GL_VERTEX_PROGRAM_ARB); + + for (i = 0; i < VERT_RESULT_MAX; i++) + outputMap[i] = -1; + + /* look for instructions which read from varying vars */ + for (i = 0; i < prog->NumInstructions; i++) { + struct prog_instruction *inst = prog->Instructions + i; + const GLuint numSrc = _mesa_num_inst_src_regs(inst->Opcode); + GLuint j; + for (j = 0; j < numSrc; j++) { + if (inst->SrcReg[j].File == PROGRAM_VARYING) { + /* replace the read with a temp reg */ + const GLuint var = inst->SrcReg[j].Index; + if (outputMap[var] == -1) { + numVaryingReads++; + outputMap[var] = _mesa_find_free_register(prog, + PROGRAM_TEMPORARY); + } + inst->SrcReg[j].File = PROGRAM_TEMPORARY; + inst->SrcReg[j].Index = outputMap[var]; + } + } + } + + if (numVaryingReads == 0) + return; /* nothing to be done */ + + /* look for instructions which write to the varying vars identified above */ + for (i = 0; i < prog->NumInstructions; i++) { + struct prog_instruction *inst = prog->Instructions + i; + const GLuint numSrc = _mesa_num_inst_src_regs(inst->Opcode); + GLuint j; + for (j = 0; j < numSrc; j++) { + if (inst->DstReg.File == PROGRAM_VARYING && + outputMap[inst->DstReg.Index] >= 0) { + /* change inst to write to the temp reg, instead of the varying */ + inst->DstReg.File = PROGRAM_TEMPORARY; + inst->DstReg.Index = outputMap[inst->DstReg.Index]; + } + } + } + + /* insert new instructions to copy the temp vars to the varying vars */ + { + struct prog_instruction *inst; + GLint endPos, var; + + /* Look for END instruction and insert the new varying writes */ + endPos = -1; + for (i = 0; i < prog->NumInstructions; i++) { + struct prog_instruction *inst = prog->Instructions + i; + if (inst->Opcode == OPCODE_END) { + endPos = i; + _mesa_insert_instructions(prog, i, numVaryingReads); + break; + } + } + + assert(endPos >= 0); + + /* insert new MOV instructions here */ + inst = prog->Instructions + endPos; + for (var = 0; var < VERT_RESULT_MAX; var++) { + if (outputMap[var] >= 0) { + /* MOV VAR[var], TEMP[tmp]; */ + inst->Opcode = OPCODE_MOV; + inst->DstReg.File = PROGRAM_VARYING; + inst->DstReg.Index = var; + inst->SrcReg[0].File = PROGRAM_TEMPORARY; + inst->SrcReg[0].Index = outputMap[var]; + inst++; + } + } + } +} diff --git a/src/mesa/shader/programopt.h b/src/mesa/shader/programopt.h index ce63644bbf..47ff2f0c7b 100644 --- a/src/mesa/shader/programopt.h +++ b/src/mesa/shader/programopt.h @@ -39,5 +39,7 @@ _mesa_count_texture_indirections(struct gl_program *prog); extern void _mesa_count_texture_instructions(struct gl_program *prog); +extern void +_mesa_remove_varying_reads(struct gl_program *prog); #endif /* PROGRAMOPT_H */ diff --git a/src/mesa/shader/shader_api.c b/src/mesa/shader/shader_api.c index 9c419c9903..24ab7568d6 100644 --- a/src/mesa/shader/shader_api.c +++ b/src/mesa/shader/shader_api.c @@ -75,23 +75,8 @@ void _mesa_clear_shader_program_data(GLcontext *ctx, struct gl_shader_program *shProg) { - if (shProg->VertexProgram) { - /* Set ptr to NULL since the param list is shared with the - * original/unlinked program. - */ - shProg->VertexProgram->Base.Parameters = NULL; - ctx->Driver.DeleteProgram(ctx, &shProg->VertexProgram->Base); - shProg->VertexProgram = NULL; - } - - if (shProg->FragmentProgram) { - /* Set ptr to NULL since the param list is shared with the - * original/unlinked program. - */ - shProg->FragmentProgram->Base.Parameters = NULL; - ctx->Driver.DeleteProgram(ctx, &shProg->FragmentProgram->Base); - shProg->FragmentProgram = NULL; - } + _mesa_reference_vertprog(ctx, &shProg->VertexProgram, NULL); + _mesa_reference_fragprog(ctx, &shProg->FragmentProgram, NULL); if (shProg->Uniforms) { _mesa_free_uniform_list(shProg->Uniforms); @@ -128,10 +113,17 @@ _mesa_free_shader_program_data(GLcontext *ctx, for (i = 0; i < shProg->NumShaders; i++) { _mesa_reference_shader(ctx, &shProg->Shaders[i], NULL); } + shProg->NumShaders = 0; + if (shProg->Shaders) { _mesa_free(shProg->Shaders); shProg->Shaders = NULL; } + + if (shProg->InfoLog) { + _mesa_free(shProg->InfoLog); + shProg->InfoLog = NULL; + } } @@ -142,10 +134,7 @@ void _mesa_free_shader_program(GLcontext *ctx, struct gl_shader_program *shProg) { _mesa_free_shader_program_data(ctx, shProg); - if (shProg->Shaders) { - _mesa_free(shProg->Shaders); - shProg->Shaders = NULL; - } + _mesa_free(shProg); } @@ -174,8 +163,10 @@ _mesa_reference_shader_program(GLcontext *ctx, ASSERT(old->RefCount > 0); old->RefCount--; - /*printf("SHPROG DECR %p (%d) to %d\n", - (void*) old, old->Name, old->RefCount);*/ +#if 0 + printf("ShaderProgram %p ID=%u RefCount-- to %d\n", + (void *) old, old->Name, old->RefCount); +#endif deleteFlag = (old->RefCount == 0); if (deleteFlag) { @@ -189,8 +180,10 @@ _mesa_reference_shader_program(GLcontext *ctx, if (shProg) { shProg->RefCount++; - /*printf("SHPROG INCR %p (%d) to %d\n", - (void*) shProg, shProg->Name, shProg->RefCount);*/ +#if 0 + printf("ShaderProgram %p ID=%u RefCount++ to %d\n", + (void *) shProg, shProg->Name, shProg->RefCount); +#endif *ptr = shProg; } } @@ -245,10 +238,8 @@ _mesa_free_shader(GLcontext *ctx, struct gl_shader *sh) _mesa_free((void *) sh->Source); if (sh->InfoLog) _mesa_free(sh->InfoLog); - for (i = 0; i < sh->NumPrograms; i++) { - assert(sh->Programs[i]); - ctx->Driver.DeleteProgram(ctx, sh->Programs[i]); - } + for (i = 0; i < sh->NumPrograms; i++) + _mesa_reference_program(ctx, &sh->Programs[i], NULL); if (sh->Programs) _mesa_free(sh->Programs); _mesa_free(sh); @@ -378,7 +369,7 @@ _mesa_attach_shader(GLcontext *ctx, GLuint program, GLuint shader) struct gl_shader_program *shProg = _mesa_lookup_shader_program(ctx, program); struct gl_shader *sh = _mesa_lookup_shader(ctx, shader); - const GLuint n = shProg->NumShaders; + GLuint n; GLuint i; if (!shProg || !sh) { @@ -387,6 +378,8 @@ _mesa_attach_shader(GLcontext *ctx, GLuint program, GLuint shader) return; } + n = shProg->NumShaders; + for (i = 0; i < n; i++) { if (shProg->Shaders[i] == sh) { /* already attached */ @@ -579,7 +572,7 @@ _mesa_detach_shader(GLcontext *ctx, GLuint program, GLuint shader) { struct gl_shader_program *shProg = _mesa_lookup_shader_program(ctx, program); - const GLuint n = shProg->NumShaders; + GLuint n; GLuint i, j; if (!shProg) { @@ -588,6 +581,8 @@ _mesa_detach_shader(GLcontext *ctx, GLuint program, GLuint shader) return; } + n = shProg->NumShaders; + for (i = 0; i < n; i++) { if (shProg->Shaders[i]->Name == shader) { /* found it */ @@ -899,7 +894,7 @@ _mesa_get_uniformfv(GLcontext *ctx, GLuint program, GLint location, if (shProg) { if (location < shProg->Uniforms->NumUniforms) { GLint progPos, i; - const struct gl_program *prog; + const struct gl_program *prog = NULL; progPos = shProg->Uniforms->Uniforms[location].VertPos; if (progPos >= 0) { @@ -912,8 +907,11 @@ _mesa_get_uniformfv(GLcontext *ctx, GLuint program, GLint location, } } - for (i = 0; i < prog->Parameters->Parameters[progPos].Size; i++) { - params[i] = prog->Parameters->ParameterValues[progPos][i]; + ASSERT(prog); + if (prog) { + for (i = 0; i < prog->Parameters->Parameters[progPos].Size; i++) { + params[i] = prog->Parameters->ParameterValues[progPos][i]; + } } } else { @@ -1010,6 +1008,8 @@ _mesa_link_program(GLcontext *ctx, GLuint program) return; } + FLUSH_VERTICES(ctx, _NEW_PROGRAM); + _slang_link(ctx, program, shProg); } @@ -1155,6 +1155,10 @@ _mesa_uniform(GLcontext *ctx, GLint location, GLsizei count, return; } + if (location == -1) + return; /* The standard specifies this as a no-op */ + + if (location < 0 || location >= (GLint) shProg->Uniforms->NumUniforms) { _mesa_error(ctx, GL_INVALID_VALUE, "glUniform(location)"); return; @@ -1250,11 +1254,16 @@ _mesa_uniform_matrix(GLcontext *ctx, GLint cols, GLint rows, GLboolean transpose, const GLfloat *values) { struct gl_shader_program *shProg = ctx->Shader.CurrentProgram; + if (!shProg || !shProg->LinkStatus) { _mesa_error(ctx, GL_INVALID_OPERATION, "glUniformMatrix(program not linked)"); return; } + + if (location == -1) + return; /* The standard specifies this as a no-op */ + if (location < 0 || location >= shProg->Uniforms->NumUniforms) { _mesa_error(ctx, GL_INVALID_VALUE, "glUniformMatrix(location)"); return; diff --git a/src/mesa/shader/slang/library/Makefile b/src/mesa/shader/slang/library/Makefile index dc67b59088..0e03fac2ee 100644 --- a/src/mesa/shader/slang/library/Makefile +++ b/src/mesa/shader/slang/library/Makefile @@ -17,7 +17,7 @@ LIB_DEP = $(TOP)/$(LIB_DIR)/$(GL_LIB_NAME) default: syntax builtin clean: - rm -f syn_to_c gc_to_bin *_syn.h *_gc.h + -rm -f syn_to_c gc_to_bin *_syn.h *_gc.h syntax: slang_pp_directives_syn.h slang_pp_expression_syn.h slang_shader_syn.h slang_pp_version_syn.h diff --git a/src/mesa/shader/slang/library/slang_common_builtin.gc b/src/mesa/shader/slang/library/slang_common_builtin.gc index 45cf1c6fd0..26080cf26c 100644 --- a/src/mesa/shader/slang/library/slang_common_builtin.gc +++ b/src/mesa/shader/slang/library/slang_common_builtin.gc @@ -343,6 +343,7 @@ vec4 asin(const vec4 v) __retVal.x = asin(v.x); __retVal.y = asin(v.y); __retVal.z = asin(v.z); + __retVal.w = asin(v.w); } float acos(const float x) diff --git a/src/mesa/shader/slang/library/slang_common_builtin_gc.h b/src/mesa/shader/slang/library/slang_common_builtin_gc.h index a896c5a287..c8d1ffc174 100644 --- a/src/mesa/shader/slang/library/slang_common_builtin_gc.h +++ b/src/mesa/shader/slang/library/slang_common_builtin_gc.h @@ -181,47 +181,48 @@ 0,1,0,12,0,97,115,105,110,0,1,1,0,12,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,97, 115,105,110,0,18,118,0,59,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,97,115,105, 110,0,18,118,0,59,121,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,97,115,105,110,0, -18,118,0,59,122,0,0,0,20,0,0,1,0,9,0,97,99,111,115,0,1,1,0,9,120,0,0,0,1,3,2,1,9,1,104,97,108,102, -80,105,0,2,17,51,0,49,52,49,53,57,50,54,0,0,17,48,0,53,0,0,48,0,0,9,18,95,95,114,101,116,86,97,108, -0,18,104,97,108,102,80,105,0,58,97,115,105,110,0,18,120,0,0,0,47,20,0,0,1,0,10,0,97,99,111,115,0,1, -1,0,10,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,97,99,111,115,0,18,118,0,59,120, -0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,97,99,111,115,0,18,118,0,59,121,0,0,0,20, -0,0,1,0,11,0,97,99,111,115,0,1,1,0,11,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58, -97,99,111,115,0,18,118,0,59,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,97,99, -111,115,0,18,118,0,59,121,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,97,99,111,115, -0,18,118,0,59,122,0,0,0,20,0,0,1,0,12,0,97,99,111,115,0,1,1,0,12,118,0,0,0,1,9,18,95,95,114,101, -116,86,97,108,0,59,120,0,58,97,99,111,115,0,18,118,0,59,120,0,0,0,20,0,9,18,95,95,114,101,116,86, -97,108,0,59,121,0,58,97,99,111,115,0,18,118,0,59,121,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0, -59,122,0,58,97,99,111,115,0,18,118,0,59,122,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,119,0, -58,97,99,111,115,0,18,118,0,59,119,0,0,0,20,0,0,1,0,9,0,97,116,97,110,0,1,1,0,9,120,0,0,0,1,9,18, -95,95,114,101,116,86,97,108,0,58,97,115,105,110,0,18,120,0,58,105,110,118,101,114,115,101,115,113, -114,116,0,18,120,0,18,120,0,48,17,49,0,48,0,0,46,0,0,48,0,0,20,0,0,1,0,10,0,97,116,97,110,0,1,1,0, -10,121,95,111,118,101,114,95,120,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,97,116,97, -110,0,18,121,95,111,118,101,114,95,120,0,59,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59, -121,0,58,97,116,97,110,0,18,121,95,111,118,101,114,95,120,0,59,121,0,0,0,20,0,0,1,0,11,0,97,116,97, -110,0,1,1,0,11,121,95,111,118,101,114,95,120,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0, -58,97,116,97,110,0,18,121,95,111,118,101,114,95,120,0,59,120,0,0,0,20,0,9,18,95,95,114,101,116,86, -97,108,0,59,121,0,58,97,116,97,110,0,18,121,95,111,118,101,114,95,120,0,59,121,0,0,0,20,0,9,18,95, -95,114,101,116,86,97,108,0,59,122,0,58,97,116,97,110,0,18,121,95,111,118,101,114,95,120,0,59,122,0, -0,0,20,0,0,1,0,12,0,97,116,97,110,0,1,1,0,12,121,95,111,118,101,114,95,120,0,0,0,1,9,18,95,95,114, -101,116,86,97,108,0,59,120,0,58,97,116,97,110,0,18,121,95,111,118,101,114,95,120,0,59,120,0,0,0,20, -0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,97,116,97,110,0,18,121,95,111,118,101,114,95,120, -0,59,121,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,97,116,97,110,0,18,121,95,111, -118,101,114,95,120,0,59,122,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,119,0,58,97,116,97, -110,0,18,121,95,111,118,101,114,95,120,0,59,119,0,0,0,20,0,0,1,0,9,0,97,116,97,110,0,1,1,0,9,121,0, -0,1,1,0,9,120,0,0,0,1,10,18,120,0,17,48,0,48,0,0,38,0,8,17,48,0,48,0,0,0,9,14,0,3,2,0,9,1,122,0,2, -58,97,116,97,110,0,18,121,0,18,120,0,49,0,0,0,0,10,18,120,0,17,48,0,48,0,0,40,0,2,10,18,121,0,17, -48,0,48,0,0,40,0,8,18,122,0,17,51,0,49,52,49,53,57,51,0,0,47,0,9,14,0,8,18,122,0,17,51,0,49,52,49, -53,57,51,0,0,46,0,0,9,14,0,8,18,122,0,0,0,1,0,10,0,97,116,97,110,0,1,1,0,10,117,0,0,1,1,0,10,118,0, -0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,97,116,97,110,0,18,117,0,59,120,0,0,18,118,0, -59,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,97,116,97,110,0,18,117,0,59,121,0, -0,18,118,0,59,121,0,0,0,20,0,0,1,0,11,0,97,116,97,110,0,1,1,0,11,117,0,0,1,1,0,11,118,0,0,0,1,9,18, -95,95,114,101,116,86,97,108,0,59,120,0,58,97,116,97,110,0,18,117,0,59,120,0,0,18,118,0,59,120,0,0, -0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,97,116,97,110,0,18,117,0,59,121,0,0,18,118,0, -59,121,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,97,116,97,110,0,18,117,0,59,122,0, -0,18,118,0,59,122,0,0,0,20,0,0,1,0,12,0,97,116,97,110,0,1,1,0,12,117,0,0,1,1,0,12,118,0,0,0,1,9,18, -95,95,114,101,116,86,97,108,0,59,120,0,58,97,116,97,110,0,18,117,0,59,120,0,0,18,118,0,59,120,0,0, -0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,97,116,97,110,0,18,117,0,59,121,0,0,18,118,0, +18,118,0,59,122,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,119,0,58,97,115,105,110,0,18,118, +0,59,119,0,0,0,20,0,0,1,0,9,0,97,99,111,115,0,1,1,0,9,120,0,0,0,1,3,2,1,9,1,104,97,108,102,80,105, +0,2,17,51,0,49,52,49,53,57,50,54,0,0,17,48,0,53,0,0,48,0,0,9,18,95,95,114,101,116,86,97,108,0,18, +104,97,108,102,80,105,0,58,97,115,105,110,0,18,120,0,0,0,47,20,0,0,1,0,10,0,97,99,111,115,0,1,1,0, +10,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,97,99,111,115,0,18,118,0,59,120,0,0, +0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,97,99,111,115,0,18,118,0,59,121,0,0,0,20,0,0, +1,0,11,0,97,99,111,115,0,1,1,0,11,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,97,99, +111,115,0,18,118,0,59,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,97,99,111,115, +0,18,118,0,59,121,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,97,99,111,115,0,18,118, +0,59,122,0,0,0,20,0,0,1,0,12,0,97,99,111,115,0,1,1,0,12,118,0,0,0,1,9,18,95,95,114,101,116,86,97, +108,0,59,120,0,58,97,99,111,115,0,18,118,0,59,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59, +121,0,58,97,99,111,115,0,18,118,0,59,121,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58, +97,99,111,115,0,18,118,0,59,122,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,119,0,58,97,99, +111,115,0,18,118,0,59,119,0,0,0,20,0,0,1,0,9,0,97,116,97,110,0,1,1,0,9,120,0,0,0,1,9,18,95,95,114, +101,116,86,97,108,0,58,97,115,105,110,0,18,120,0,58,105,110,118,101,114,115,101,115,113,114,116,0, +18,120,0,18,120,0,48,17,49,0,48,0,0,46,0,0,48,0,0,20,0,0,1,0,10,0,97,116,97,110,0,1,1,0,10,121,95, +111,118,101,114,95,120,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,97,116,97,110,0,18, +121,95,111,118,101,114,95,120,0,59,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58, +97,116,97,110,0,18,121,95,111,118,101,114,95,120,0,59,121,0,0,0,20,0,0,1,0,11,0,97,116,97,110,0,1, +1,0,11,121,95,111,118,101,114,95,120,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,97,116, +97,110,0,18,121,95,111,118,101,114,95,120,0,59,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0, +59,121,0,58,97,116,97,110,0,18,121,95,111,118,101,114,95,120,0,59,121,0,0,0,20,0,9,18,95,95,114, +101,116,86,97,108,0,59,122,0,58,97,116,97,110,0,18,121,95,111,118,101,114,95,120,0,59,122,0,0,0,20, +0,0,1,0,12,0,97,116,97,110,0,1,1,0,12,121,95,111,118,101,114,95,120,0,0,0,1,9,18,95,95,114,101,116, +86,97,108,0,59,120,0,58,97,116,97,110,0,18,121,95,111,118,101,114,95,120,0,59,120,0,0,0,20,0,9,18, +95,95,114,101,116,86,97,108,0,59,121,0,58,97,116,97,110,0,18,121,95,111,118,101,114,95,120,0,59, +121,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,97,116,97,110,0,18,121,95,111,118, +101,114,95,120,0,59,122,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,119,0,58,97,116,97,110,0, +18,121,95,111,118,101,114,95,120,0,59,119,0,0,0,20,0,0,1,0,9,0,97,116,97,110,0,1,1,0,9,121,0,0,1,1, +0,9,120,0,0,0,1,10,18,120,0,17,48,0,48,0,0,38,0,8,17,48,0,48,0,0,0,9,14,0,3,2,0,9,1,122,0,2,58,97, +116,97,110,0,18,121,0,18,120,0,49,0,0,0,0,10,18,120,0,17,48,0,48,0,0,40,0,2,10,18,121,0,17,48,0,48, +0,0,40,0,8,18,122,0,17,51,0,49,52,49,53,57,51,0,0,47,0,9,14,0,8,18,122,0,17,51,0,49,52,49,53,57,51, +0,0,46,0,0,9,14,0,8,18,122,0,0,0,1,0,10,0,97,116,97,110,0,1,1,0,10,117,0,0,1,1,0,10,118,0,0,0,1,9, +18,95,95,114,101,116,86,97,108,0,59,120,0,58,97,116,97,110,0,18,117,0,59,120,0,0,18,118,0,59,120,0, +0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,97,116,97,110,0,18,117,0,59,121,0,0,18,118, +0,59,121,0,0,0,20,0,0,1,0,11,0,97,116,97,110,0,1,1,0,11,117,0,0,1,1,0,11,118,0,0,0,1,9,18,95,95, +114,101,116,86,97,108,0,59,120,0,58,97,116,97,110,0,18,117,0,59,120,0,0,18,118,0,59,120,0,0,0,20,0, +9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,97,116,97,110,0,18,117,0,59,121,0,0,18,118,0,59,121, +0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,97,116,97,110,0,18,117,0,59,122,0,0,18, +118,0,59,122,0,0,0,20,0,0,1,0,12,0,97,116,97,110,0,1,1,0,12,117,0,0,1,1,0,12,118,0,0,0,1,9,18,95, +95,114,101,116,86,97,108,0,59,120,0,58,97,116,97,110,0,18,117,0,59,120,0,0,18,118,0,59,120,0,0,0, +20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,97,116,97,110,0,18,117,0,59,121,0,0,18,118,0, 59,121,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,97,116,97,110,0,18,117,0,59,122,0, 0,18,118,0,59,122,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,119,0,58,97,116,97,110,0,18,117, 0,59,119,0,0,18,118,0,59,119,0,0,0,20,0,0,1,0,9,0,112,111,119,0,1,1,0,9,97,0,0,1,1,0,9,98,0,0,0,1, diff --git a/src/mesa/shader/slang/library/slang_core.gc b/src/mesa/shader/slang/library/slang_core.gc index 2e7ebc347a..ffa57f74c2 100644 --- a/src/mesa/shader/slang/library/slang_core.gc +++ b/src/mesa/shader/slang/library/slang_core.gc @@ -1545,15 +1545,8 @@ mat2 __operator - (const mat2 m, const mat2 n) mat2 __operator * (const mat2 m, const mat2 n) { - vec2 mRow0, mRow1; - mRow0.x = m[0].x; - mRow0.y = m[1].x; - mRow1.x = m[0].y; - mRow1.y = m[1].y; - __retVal[0].x = dot(mRow0, n[0]); - __retVal[1].x = dot(mRow0, n[1]); - __retVal[0].y = dot(mRow1, n[0]); - __retVal[1].y = dot(mRow1, n[1]); + __retVal[0] = m[0] * n[0].xx + m[1] * n[0].yy; + __retVal[1] = m[0] * n[1].xx + m[1] * n[1].yy; } mat2 __operator / (const mat2 m, const mat2 n) @@ -1581,34 +1574,9 @@ mat3 __operator - (const mat3 m, const mat3 n) mat3 __operator * (const mat3 m, const mat3 n) { - // sub-blocks to reduce register usage - { - vec3 mRow0; - mRow0.x = m[0].x; - mRow0.y = m[1].x; - mRow0.z = m[2].x; - __retVal[0].x = dot(mRow0, n[0]); - __retVal[1].x = dot(mRow0, n[1]); - __retVal[2].x = dot(mRow0, n[2]); - } - { - vec3 mRow1; - mRow1.x = m[0].y; - mRow1.y = m[1].y; - mRow1.z = m[2].y; - __retVal[0].y = dot(mRow1, n[0]); - __retVal[1].y = dot(mRow1, n[1]); - __retVal[2].y = dot(mRow1, n[2]); - } - { - vec3 mRow2; - mRow2.x = m[0].z; - mRow2.y = m[1].z; - mRow2.z = m[2].z; - __retVal[0].z = dot(mRow2, n[0]); - __retVal[1].z = dot(mRow2, n[1]); - __retVal[2].z = dot(mRow2, n[2]); - } + __retVal[0] = m[0] * n[0].xxx + m[1] * n[0].yyy + m[2] * n[0].zzz; + __retVal[1] = m[0] * n[1].xxx + m[1] * n[1].yyy + m[2] * n[1].zzz; + __retVal[2] = m[0] * n[2].xxx + m[1] * n[2].yyy + m[2] * n[2].zzz; } mat3 __operator / (const mat3 m, const mat3 n) @@ -1639,51 +1607,10 @@ mat4 __operator - (const mat4 m, const mat4 n) mat4 __operator * (const mat4 m, const mat4 n) { - // sub-blocks to reduce temporary usage - { - vec4 mRow0; - mRow0.x = m[0].x; - mRow0.y = m[1].x; - mRow0.z = m[2].x; - mRow0.w = m[3].x; - __retVal[0].x = dot(mRow0, n[0]); - __retVal[1].x = dot(mRow0, n[1]); - __retVal[2].x = dot(mRow0, n[2]); - __retVal[3].x = dot(mRow0, n[3]); - } - { - vec4 mRow1; - mRow1.x = m[0].y; - mRow1.y = m[1].y; - mRow1.z = m[2].y; - mRow1.w = m[3].y; - __retVal[0].y = dot(mRow1, n[0]); - __retVal[1].y = dot(mRow1, n[1]); - __retVal[2].y = dot(mRow1, n[2]); - __retVal[3].y = dot(mRow1, n[3]); - } - { - vec4 mRow2; - mRow2.x = m[0].z; - mRow2.y = m[1].z; - mRow2.z = m[2].z; - mRow2.w = m[3].z; - __retVal[0].z = dot(mRow2, n[0]); - __retVal[1].z = dot(mRow2, n[1]); - __retVal[2].z = dot(mRow2, n[2]); - __retVal[3].z = dot(mRow2, n[3]); - } - { - vec4 mRow3; - mRow3.x = m[0].w; - mRow3.y = m[1].w; - mRow3.z = m[2].w; - mRow3.w = m[3].w; - __retVal[0].w = dot(mRow3, n[0]); - __retVal[1].w = dot(mRow3, n[1]); - __retVal[2].w = dot(mRow3, n[2]); - __retVal[3].w = dot(mRow3, n[3]); - } + __retVal[0] = m[0] * n[0].xxxx + m[1] * n[0].yyyy + m[2] * n[0].zzzz + m[3] * n[0].wwww; + __retVal[1] = m[0] * n[1].xxxx + m[1] * n[1].yyyy + m[2] * n[1].zzzz + m[3] * n[1].wwww; + __retVal[2] = m[0] * n[2].xxxx + m[1] * n[2].yyyy + m[2] * n[2].zzzz + m[3] * n[2].wwww; + __retVal[3] = m[0] * n[3].xxxx + m[1] * n[3].yyyy + m[2] * n[3].zzzz + m[3] * n[3].wwww; } mat4 __operator / (const mat4 m, const mat4 n) @@ -1877,13 +1804,8 @@ mat4 __operator / (const mat4 m, const float b) vec2 __operator * (const mat2 m, const vec2 v) { - vec2 r0, r1; - r0.x = m[0].x; - r0.y = m[1].x; - r1.x = m[0].y; - r1.y = m[1].y; - __retVal.x = dot(r0, v); - __retVal.y = dot(r1, v); + __retVal = m[0] * v.xx + + m[1] * v.yy; } vec2 __operator * (const vec2 v, const mat2 m) @@ -1894,27 +1816,9 @@ vec2 __operator * (const vec2 v, const mat2 m) vec3 __operator * (const mat3 m, const vec3 v) { - { - vec3 r0; - r0.x = m[0].x; - r0.y = m[1].x; - r0.z = m[2].x; - __asm vec3_dot __retVal.x, r0, v; - } - { - vec3 r1; - r1.x = m[0].y; - r1.y = m[1].y; - r1.z = m[2].y; - __asm vec3_dot __retVal.y, r1, v; - } - { - vec3 r2; - r2.x = m[0].z; - r2.y = m[1].z; - r2.z = m[2].z; - __asm vec3_dot __retVal.z, r2, v; - } + __retVal = m[0] * v.xxx + + m[1] * v.yyy + + m[2] * v.zzz; } vec3 __operator * (const vec3 v, const mat3 m) @@ -1926,44 +1830,14 @@ vec3 __operator * (const vec3 v, const mat3 m) vec4 __operator * (const mat4 m, const vec4 v) { - // extract rows, then do dot product - { - vec4 r0; - r0.x = m[0].x; - r0.y = m[1].x; - r0.z = m[2].x; - r0.w = m[3].x; - __asm vec4_dot __retVal.x, r0, v; - } - { - vec4 r1; - r1.x = m[0].y; - r1.y = m[1].y; - r1.z = m[2].y; - r1.w = m[3].y; - __asm vec4_dot __retVal.y, r1, v; - } - { - vec4 r2; - r2.x = m[0].z; - r2.y = m[1].z; - r2.z = m[2].z; - r2.w = m[3].z; - __asm vec4_dot __retVal.z, r2, v; - } - { - vec4 r3; - r3.x = m[0].w; - r3.y = m[1].w; - r3.z = m[2].w; - r3.w = m[3].w; - __asm vec4_dot __retVal.w, r3, v; - } + __retVal = m[0] * v.xxxx + + m[1] * v.yyyy + + m[2] * v.zzzz + + m[3] * v.wwww; } vec4 __operator * (const vec4 v, const mat4 m) { - //mm __retVal.x = dot(v, m[0]); __retVal.y = dot(v, m[1]); __retVal.z = dot(v, m[2]); diff --git a/src/mesa/shader/slang/library/slang_core_gc.h b/src/mesa/shader/slang/library/slang_core_gc.h index f2b4fd6464..1604d5a034 100644 --- a/src/mesa/shader/slang/library/slang_core_gc.h +++ b/src/mesa/shader/slang/library/slang_core_gc.h @@ -453,151 +453,113 @@ 97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,46,20,0,0,1,0,13,2,27,1,1,0, 13,109,0,0,1,1,0,13,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0, 57,18,110,0,16,8,48,0,57,47,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10, -49,0,57,18,110,0,16,10,49,0,57,47,20,0,0,1,0,13,2,21,1,1,0,13,109,0,0,1,1,0,13,110,0,0,0,1,3,2,0, -10,1,109,82,111,119,48,0,0,1,1,109,82,111,119,49,0,0,0,9,18,109,82,111,119,48,0,59,120,0,18,109,0, -16,8,48,0,57,59,120,0,20,0,9,18,109,82,111,119,48,0,59,121,0,18,109,0,16,10,49,0,57,59,120,0,20,0, -9,18,109,82,111,119,49,0,59,120,0,18,109,0,16,8,48,0,57,59,121,0,20,0,9,18,109,82,111,119,49,0,59, -121,0,18,109,0,16,10,49,0,57,59,121,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,120, -0,58,100,111,116,0,18,109,82,111,119,48,0,0,18,110,0,16,8,48,0,57,0,0,20,0,9,18,95,95,114,101,116, -86,97,108,0,16,10,49,0,57,59,120,0,58,100,111,116,0,18,109,82,111,119,48,0,0,18,110,0,16,10,49,0, -57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,121,0,58,100,111,116,0,18,109,82, -111,119,49,0,0,18,110,0,16,8,48,0,57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59, -121,0,58,100,111,116,0,18,109,82,111,119,49,0,0,18,110,0,16,10,49,0,57,0,0,20,0,0,1,0,13,2,22,1,1, -0,13,109,0,0,1,1,0,13,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48, +49,0,57,18,110,0,16,10,49,0,57,47,20,0,0,1,0,13,2,21,1,1,0,13,109,0,0,1,1,0,13,110,0,0,0,1,9,18,95, +95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,59,120,120,0, +48,18,109,0,16,10,49,0,57,18,110,0,16,8,48,0,57,59,121,121,0,48,46,20,0,9,18,95,95,114,101,116,86, +97,108,0,16,10,49,0,57,18,109,0,16,8,48,0,57,18,110,0,16,10,49,0,57,59,120,120,0,48,18,109,0,16,10, +49,0,57,18,110,0,16,10,49,0,57,59,121,121,0,48,46,20,0,0,1,0,13,2,22,1,1,0,13,109,0,0,1,1,0,13,110, +0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0, +57,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,110,0,16,10, +49,0,57,49,20,0,0,1,0,14,2,26,1,1,0,14,109,0,0,1,1,0,14,110,0,0,0,1,9,18,95,95,114,101,116,86,97, +108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,46,20,0,9,18,95,95,114,101,116,86, +97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,46,20,0,9,18,95,95,114,101, +116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,46,20,0,0,1,0,14,2,27, +1,1,0,14,109,0,0,1,1,0,14,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16, +8,48,0,57,18,110,0,16,8,48,0,57,47,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0, +16,10,49,0,57,18,110,0,16,10,49,0,57,47,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18, +109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,47,20,0,0,1,0,14,2,21,1,1,0,14,109,0,0,1,1,0,14,110,0,0, +0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,59, +120,120,120,0,48,18,109,0,16,10,49,0,57,18,110,0,16,8,48,0,57,59,121,121,121,0,48,46,18,109,0,16, +10,50,0,57,18,110,0,16,8,48,0,57,59,122,122,122,0,48,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16, +10,49,0,57,18,109,0,16,8,48,0,57,18,110,0,16,10,49,0,57,59,120,120,120,0,48,18,109,0,16,10,49,0,57, +18,110,0,16,10,49,0,57,59,121,121,121,0,48,46,18,109,0,16,10,50,0,57,18,110,0,16,10,49,0,57,59,122, +122,122,0,48,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,8,48,0,57,18,110, +0,16,10,50,0,57,59,120,120,120,0,48,18,109,0,16,10,49,0,57,18,110,0,16,10,50,0,57,59,121,121,121,0, +48,46,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,59,122,122,122,0,48,46,20,0,0,1,0,14,2,22,1,1, +0,14,109,0,0,1,1,0,14,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48, 0,57,18,110,0,16,8,48,0,57,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10, -49,0,57,18,110,0,16,10,49,0,57,49,20,0,0,1,0,14,2,26,1,1,0,14,109,0,0,1,1,0,14,110,0,0,0,1,9,18,95, -95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,46,20,0,9,18, -95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,46,20,0, -9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,46, -20,0,0,1,0,14,2,27,1,1,0,14,109,0,0,1,1,0,14,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8, -48,0,57,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,47,20,0,9,18,95,95,114,101,116,86,97,108,0,16, -10,49,0,57,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,47,20,0,9,18,95,95,114,101,116,86,97,108, -0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,47,20,0,0,1,0,14,2,21,1,1,0,14,109,0, -0,1,1,0,14,110,0,0,0,1,2,3,2,0,11,1,109,82,111,119,48,0,0,0,9,18,109,82,111,119,48,0,59,120,0,18, -109,0,16,8,48,0,57,59,120,0,20,0,9,18,109,82,111,119,48,0,59,121,0,18,109,0,16,10,49,0,57,59,120,0, -20,0,9,18,109,82,111,119,48,0,59,122,0,18,109,0,16,10,50,0,57,59,120,0,20,0,9,18,95,95,114,101,116, -86,97,108,0,16,8,48,0,57,59,120,0,58,100,111,116,0,18,109,82,111,119,48,0,0,18,110,0,16,8,48,0,57, -0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,120,0,58,100,111,116,0,18,109,82,111, -119,48,0,0,18,110,0,16,10,49,0,57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,59,120, -0,58,100,111,116,0,18,109,82,111,119,48,0,0,18,110,0,16,10,50,0,57,0,0,20,0,0,2,3,2,0,11,1,109,82, -111,119,49,0,0,0,9,18,109,82,111,119,49,0,59,120,0,18,109,0,16,8,48,0,57,59,121,0,20,0,9,18,109,82, -111,119,49,0,59,121,0,18,109,0,16,10,49,0,57,59,121,0,20,0,9,18,109,82,111,119,49,0,59,122,0,18, -109,0,16,10,50,0,57,59,121,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,121,0,58,100, -111,116,0,18,109,82,111,119,49,0,0,18,110,0,16,8,48,0,57,0,0,20,0,9,18,95,95,114,101,116,86,97,108, -0,16,10,49,0,57,59,121,0,58,100,111,116,0,18,109,82,111,119,49,0,0,18,110,0,16,10,49,0,57,0,0,20,0, -9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,59,121,0,58,100,111,116,0,18,109,82,111,119,49,0, -0,18,110,0,16,10,50,0,57,0,0,20,0,0,2,3,2,0,11,1,109,82,111,119,50,0,0,0,9,18,109,82,111,119,50,0, -59,120,0,18,109,0,16,8,48,0,57,59,122,0,20,0,9,18,109,82,111,119,50,0,59,121,0,18,109,0,16,10,49,0, -57,59,122,0,20,0,9,18,109,82,111,119,50,0,59,122,0,18,109,0,16,10,50,0,57,59,122,0,20,0,9,18,95,95, -114,101,116,86,97,108,0,16,8,48,0,57,59,122,0,58,100,111,116,0,18,109,82,111,119,50,0,0,18,110,0, -16,8,48,0,57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,122,0,58,100,111,116,0, -18,109,82,111,119,50,0,0,18,110,0,16,10,49,0,57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10, -50,0,57,59,122,0,58,100,111,116,0,18,109,82,111,119,50,0,0,18,110,0,16,10,50,0,57,0,0,20,0,0,0,1,0, -14,2,22,1,1,0,14,109,0,0,1,1,0,14,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18, -109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57, -18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50, -0,57,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,49,20,0,0,1,0,15,2,26,1,1,0,15,109,0,0,1,1,0,15, -110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,110,0,16,8,48, -0,57,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,110,0,16, -10,49,0,57,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,110, -0,16,10,50,0,57,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,109,0,16,10,51,0,57,18, -110,0,16,10,51,0,57,46,20,0,0,1,0,15,2,27,1,1,0,15,109,0,0,1,1,0,15,110,0,0,0,1,9,18,95,95,114,101, -116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,47,20,0,9,18,95,95,114, -101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,47,20,0,9,18,95,95, -114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,47,20,0,9,18, -95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,47,20,0, -0,1,0,15,2,21,1,1,0,15,109,0,0,1,1,0,15,110,0,0,0,1,2,3,2,0,12,1,109,82,111,119,48,0,0,0,9,18,109, -82,111,119,48,0,59,120,0,18,109,0,16,8,48,0,57,59,120,0,20,0,9,18,109,82,111,119,48,0,59,121,0,18, -109,0,16,10,49,0,57,59,120,0,20,0,9,18,109,82,111,119,48,0,59,122,0,18,109,0,16,10,50,0,57,59,120, -0,20,0,9,18,109,82,111,119,48,0,59,119,0,18,109,0,16,10,51,0,57,59,120,0,20,0,9,18,95,95,114,101, -116,86,97,108,0,16,8,48,0,57,59,120,0,58,100,111,116,0,18,109,82,111,119,48,0,0,18,110,0,16,8,48,0, -57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,120,0,58,100,111,116,0,18,109,82, -111,119,48,0,0,18,110,0,16,10,49,0,57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,59, -120,0,58,100,111,116,0,18,109,82,111,119,48,0,0,18,110,0,16,10,50,0,57,0,0,20,0,9,18,95,95,114,101, -116,86,97,108,0,16,10,51,0,57,59,120,0,58,100,111,116,0,18,109,82,111,119,48,0,0,18,110,0,16,10,51, -0,57,0,0,20,0,0,2,3,2,0,12,1,109,82,111,119,49,0,0,0,9,18,109,82,111,119,49,0,59,120,0,18,109,0,16, -8,48,0,57,59,121,0,20,0,9,18,109,82,111,119,49,0,59,121,0,18,109,0,16,10,49,0,57,59,121,0,20,0,9, -18,109,82,111,119,49,0,59,122,0,18,109,0,16,10,50,0,57,59,121,0,20,0,9,18,109,82,111,119,49,0,59, -119,0,18,109,0,16,10,51,0,57,59,121,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,121, -0,58,100,111,116,0,18,109,82,111,119,49,0,0,18,110,0,16,8,48,0,57,0,0,20,0,9,18,95,95,114,101,116, -86,97,108,0,16,10,49,0,57,59,121,0,58,100,111,116,0,18,109,82,111,119,49,0,0,18,110,0,16,10,49,0, -57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,59,121,0,58,100,111,116,0,18,109,82, -111,119,49,0,0,18,110,0,16,10,50,0,57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,59, -121,0,58,100,111,116,0,18,109,82,111,119,49,0,0,18,110,0,16,10,51,0,57,0,0,20,0,0,2,3,2,0,12,1,109, -82,111,119,50,0,0,0,9,18,109,82,111,119,50,0,59,120,0,18,109,0,16,8,48,0,57,59,122,0,20,0,9,18,109, -82,111,119,50,0,59,121,0,18,109,0,16,10,49,0,57,59,122,0,20,0,9,18,109,82,111,119,50,0,59,122,0,18, -109,0,16,10,50,0,57,59,122,0,20,0,9,18,109,82,111,119,50,0,59,119,0,18,109,0,16,10,51,0,57,59,122, -0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,122,0,58,100,111,116,0,18,109,82,111,119, -50,0,0,18,110,0,16,8,48,0,57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,122,0,58, -100,111,116,0,18,109,82,111,119,50,0,0,18,110,0,16,10,49,0,57,0,0,20,0,9,18,95,95,114,101,116,86, -97,108,0,16,10,50,0,57,59,122,0,58,100,111,116,0,18,109,82,111,119,50,0,0,18,110,0,16,10,50,0,57,0, -0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,59,122,0,58,100,111,116,0,18,109,82,111, -119,50,0,0,18,110,0,16,10,51,0,57,0,0,20,0,0,2,3,2,0,12,1,109,82,111,119,51,0,0,0,9,18,109,82,111, -119,51,0,59,120,0,18,109,0,16,8,48,0,57,59,119,0,20,0,9,18,109,82,111,119,51,0,59,121,0,18,109,0, -16,10,49,0,57,59,119,0,20,0,9,18,109,82,111,119,51,0,59,122,0,18,109,0,16,10,50,0,57,59,119,0,20,0, -9,18,109,82,111,119,51,0,59,119,0,18,109,0,16,10,51,0,57,59,119,0,20,0,9,18,95,95,114,101,116,86, -97,108,0,16,8,48,0,57,59,119,0,58,100,111,116,0,18,109,82,111,119,51,0,0,18,110,0,16,8,48,0,57,0,0, -20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,119,0,58,100,111,116,0,18,109,82,111,119, -51,0,0,18,110,0,16,10,49,0,57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,59,119,0, -58,100,111,116,0,18,109,82,111,119,51,0,0,18,110,0,16,10,50,0,57,0,0,20,0,9,18,95,95,114,101,116, -86,97,108,0,16,10,51,0,57,59,119,0,58,100,111,116,0,18,109,82,111,119,51,0,0,18,110,0,16,10,51,0, -57,0,0,20,0,0,0,1,0,15,2,22,1,1,0,15,109,0,0,1,1,0,15,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108, -0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,49,20,0,9,18,95,95,114,101,116,86,97, -108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,49,20,0,9,18,95,95,114,101,116, -86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,49,20,0,9,18,95,95,114,101, -116,86,97,108,0,16,10,51,0,57,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,49,20,0,0,1,0,13,2,26, -1,1,0,9,97,0,0,1,1,0,13,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110, -0,16,8,48,0,57,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49, -0,57,46,20,0,0,1,0,13,2,26,1,1,0,13,109,0,0,1,1,0,9,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0, -16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0, -57,18,109,0,16,10,49,0,57,18,98,0,46,20,0,0,1,0,13,2,27,1,1,0,9,97,0,0,1,1,0,13,110,0,0,0,1,9,18, -95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48,0,57,47,20,0,9,18,95,95,114, -101,116,86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,47,20,0,0,1,0,13,2,27,1,1,0,13, -109,0,0,1,1,0,9,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57, -18,98,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,47, -20,0,0,1,0,13,2,21,1,1,0,9,97,0,0,1,1,0,13,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48, +49,0,57,18,110,0,16,10,49,0,57,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0, +16,10,50,0,57,18,110,0,16,10,50,0,57,49,20,0,0,1,0,15,2,26,1,1,0,15,109,0,0,1,1,0,15,110,0,0,0,1,9, +18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,46,20,0, +9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,46, +20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0, +57,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,109,0,16,10,51,0,57,18,110,0,16,10, +51,0,57,46,20,0,0,1,0,15,2,27,1,1,0,15,109,0,0,1,1,0,15,110,0,0,0,1,9,18,95,95,114,101,116,86,97, +108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,47,20,0,9,18,95,95,114,101,116,86, +97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,47,20,0,9,18,95,95,114,101, +116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,47,20,0,9,18,95,95,114, +101,116,86,97,108,0,16,10,51,0,57,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,47,20,0,0,1,0,15,2, +21,1,1,0,15,109,0,0,1,1,0,15,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0, +16,8,48,0,57,18,110,0,16,8,48,0,57,59,120,120,120,120,0,48,18,109,0,16,10,49,0,57,18,110,0,16,8,48, +0,57,59,121,121,121,121,0,48,46,18,109,0,16,10,50,0,57,18,110,0,16,8,48,0,57,59,122,122,122,122,0, +48,46,18,109,0,16,10,51,0,57,18,110,0,16,8,48,0,57,59,119,119,119,119,0,48,46,20,0,9,18,95,95,114, +101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,8,48,0,57,18,110,0,16,10,49,0,57,59,120,120,120,120, +0,48,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,59,121,121,121,121,0,48,46,18,109,0,16,10,50,0, +57,18,110,0,16,10,49,0,57,59,122,122,122,122,0,48,46,18,109,0,16,10,51,0,57,18,110,0,16,10,49,0,57, +59,119,119,119,119,0,48,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,8,48, +0,57,18,110,0,16,10,50,0,57,59,120,120,120,120,0,48,18,109,0,16,10,49,0,57,18,110,0,16,10,50,0,57, +59,121,121,121,121,0,48,46,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,59,122,122,122,122,0,48, +46,18,109,0,16,10,51,0,57,18,110,0,16,10,50,0,57,59,119,119,119,119,0,48,46,20,0,9,18,95,95,114, +101,116,86,97,108,0,16,10,51,0,57,18,109,0,16,8,48,0,57,18,110,0,16,10,51,0,57,59,120,120,120,120, +0,48,18,109,0,16,10,49,0,57,18,110,0,16,10,51,0,57,59,121,121,121,121,0,48,46,18,109,0,16,10,50,0, +57,18,110,0,16,10,51,0,57,59,122,122,122,122,0,48,46,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57, +59,119,119,119,119,0,48,46,20,0,0,1,0,15,2,22,1,1,0,15,109,0,0,1,1,0,15,110,0,0,0,1,9,18,95,95,114, +101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,49,20,0,9,18,95,95, +114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,49,20,0,9,18, +95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,49,20,0, +9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,49, +20,0,0,1,0,13,2,26,1,1,0,9,97,0,0,1,1,0,13,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48, +0,57,18,97,0,18,110,0,16,8,48,0,57,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97, +0,18,110,0,16,10,49,0,57,46,20,0,0,1,0,13,2,26,1,1,0,13,109,0,0,1,1,0,9,98,0,0,0,1,9,18,95,95,114, +101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,0,46,20,0,9,18,95,95,114,101,116,86, +97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,46,20,0,0,1,0,13,2,27,1,1,0,9,97,0,0,1,1,0, +13,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48,0,57,47,20, +0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,47,20,0,0,1,0,13, +2,27,1,1,0,13,109,0,0,1,1,0,9,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0, +16,8,48,0,57,18,98,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0, +57,18,98,0,47,20,0,0,1,0,13,2,21,1,1,0,9,97,0,0,1,1,0,13,110,0,0,0,1,9,18,95,95,114,101,116,86,97, +108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10, +49,0,57,18,97,0,18,110,0,16,10,49,0,57,48,20,0,0,1,0,13,2,21,1,1,0,13,109,0,0,1,1,0,9,98,0,0,0,1,9, +18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,0,48,20,0,9,18,95,95,114, +101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,48,20,0,0,1,0,13,2,22,1,1,0,9,97, +0,0,1,1,0,13,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48, +0,57,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,49,20, +0,0,1,0,13,2,22,1,1,0,13,109,0,0,1,1,0,9,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0, +57,18,109,0,16,8,48,0,57,18,98,0,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0, +16,10,49,0,57,18,98,0,49,20,0,0,1,0,14,2,26,1,1,0,9,97,0,0,1,1,0,14,110,0,0,0,1,9,18,95,95,114,101, +116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48,0,57,46,20,0,9,18,95,95,114,101,116,86,97, +108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16, +10,50,0,57,18,97,0,18,110,0,16,10,50,0,57,46,20,0,0,1,0,14,2,26,1,1,0,14,109,0,0,1,1,0,9,98,0,0,0, +1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,0,46,20,0,9,18,95,95, +114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,46,20,0,9,18,95,95,114,101, +116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,98,0,46,20,0,0,1,0,14,2,27,1,1,0,9,97,0,0, +1,1,0,14,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48,0,57, +47,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,47,20,0,9, +18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,97,0,18,110,0,16,10,50,0,57,47,20,0,0,1,0,14,2, +27,1,1,0,14,109,0,0,1,1,0,9,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16, +8,48,0,57,18,98,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57, +18,98,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,98,0,47, +20,0,0,1,0,14,2,21,1,1,0,9,97,0,0,1,1,0,14,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48, 0,57,18,97,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97, -0,18,110,0,16,10,49,0,57,48,20,0,0,1,0,13,2,21,1,1,0,13,109,0,0,1,1,0,9,98,0,0,0,1,9,18,95,95,114, -101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,0,48,20,0,9,18,95,95,114,101,116,86, -97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,48,20,0,0,1,0,13,2,22,1,1,0,9,97,0,0,1,1,0, -13,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48,0,57,49,20, -0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,49,20,0,0,1,0,13, -2,22,1,1,0,13,109,0,0,1,1,0,9,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0, -16,8,48,0,57,18,98,0,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0, -57,18,98,0,49,20,0,0,1,0,14,2,26,1,1,0,9,97,0,0,1,1,0,14,110,0,0,0,1,9,18,95,95,114,101,116,86,97, -108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48,0,57,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10, -49,0,57,18,97,0,18,110,0,16,10,49,0,57,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18, -97,0,18,110,0,16,10,50,0,57,46,20,0,0,1,0,14,2,26,1,1,0,14,109,0,0,1,1,0,9,98,0,0,0,1,9,18,95,95, -114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,0,46,20,0,9,18,95,95,114,101,116, -86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,46,20,0,9,18,95,95,114,101,116,86,97,108, -0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,98,0,46,20,0,0,1,0,14,2,27,1,1,0,9,97,0,0,1,1,0,14,110,0, -0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48,0,57,47,20,0,9,18, -95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,47,20,0,9,18,95,95,114, -101,116,86,97,108,0,16,10,50,0,57,18,97,0,18,110,0,16,10,50,0,57,47,20,0,0,1,0,14,2,27,1,1,0,14, -109,0,0,1,1,0,9,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57, -18,98,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,47, -20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,98,0,47,20,0,0,1,0, -14,2,21,1,1,0,9,97,0,0,1,1,0,14,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,97, -0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,18,110,0, -16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,97,0,18,110,0,16,10,50,0, -57,48,20,0,0,1,0,14,2,21,1,1,0,14,109,0,0,1,1,0,9,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16, -8,48,0,57,18,109,0,16,8,48,0,57,18,98,0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57, -18,109,0,16,10,49,0,57,18,98,0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0, -16,10,50,0,57,18,98,0,48,20,0,0,1,0,14,2,22,1,1,0,9,97,0,0,1,1,0,14,110,0,0,0,1,9,18,95,95,114,101, -116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48,0,57,49,20,0,9,18,95,95,114,101,116,86,97, -108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16, -10,50,0,57,18,97,0,18,110,0,16,10,50,0,57,49,20,0,0,1,0,14,2,22,1,1,0,14,109,0,0,1,1,0,9,98,0,0,0, -1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,0,49,20,0,9,18,95,95, -114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,49,20,0,9,18,95,95,114,101, -116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,98,0,49,20,0,0,1,0,15,2,26,1,1,0,9,97,0,0, -1,1,0,15,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48,0,57, -46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,46,20,0,9, -18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,97,0,18,110,0,16,10,50,0,57,46,20,0,9,18,95,95, -114,101,116,86,97,108,0,16,10,51,0,57,18,97,0,18,110,0,16,10,51,0,57,46,20,0,0,1,0,15,2,26,1,1,0, -15,109,0,0,1,1,0,9,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0, -57,18,98,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0, -46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,98,0,46,20,0,9, -18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,109,0,16,10,51,0,57,18,98,0,46,20,0,0,1,0,15,2, +0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,97,0,18,110,0, +16,10,50,0,57,48,20,0,0,1,0,14,2,21,1,1,0,14,109,0,0,1,1,0,9,98,0,0,0,1,9,18,95,95,114,101,116,86, +97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16, +10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57, +18,109,0,16,10,50,0,57,18,98,0,48,20,0,0,1,0,14,2,22,1,1,0,9,97,0,0,1,1,0,14,110,0,0,0,1,9,18,95, +95,114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48,0,57,49,20,0,9,18,95,95,114,101, +116,86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,49,20,0,9,18,95,95,114,101,116,86,97, +108,0,16,10,50,0,57,18,97,0,18,110,0,16,10,50,0,57,49,20,0,0,1,0,14,2,22,1,1,0,14,109,0,0,1,1,0,9, +98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,0,49,20,0,9, +18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,49,20,0,9,18,95,95, +114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,98,0,49,20,0,0,1,0,15,2,26,1,1,0,9, +97,0,0,1,1,0,15,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8, +48,0,57,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,46, +20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,97,0,18,110,0,16,10,50,0,57,46,20,0,9,18, +95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,97,0,18,110,0,16,10,51,0,57,46,20,0,0,1,0,15,2,26,1, +1,0,15,109,0,0,1,1,0,9,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48, +0,57,18,98,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98, +0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,98,0,46,20,0, +9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,109,0,16,10,51,0,57,18,98,0,46,20,0,0,1,0,15,2, 27,1,1,0,9,97,0,0,1,1,0,15,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18, 110,0,16,8,48,0,57,47,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10, 49,0,57,47,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,97,0,18,110,0,16,10,50,0,57,47, @@ -622,244 +584,221 @@ 97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,0,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16, 10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57, 18,109,0,16,10,50,0,57,18,98,0,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,109,0, -16,10,51,0,57,18,98,0,49,20,0,0,1,0,10,2,21,1,1,0,13,109,0,0,1,1,0,10,118,0,0,0,1,3,2,0,10,1,114, -48,0,0,1,1,114,49,0,0,0,9,18,114,48,0,59,120,0,18,109,0,16,8,48,0,57,59,120,0,20,0,9,18,114,48,0, -59,121,0,18,109,0,16,10,49,0,57,59,120,0,20,0,9,18,114,49,0,59,120,0,18,109,0,16,8,48,0,57,59,121, -0,20,0,9,18,114,49,0,59,121,0,18,109,0,16,10,49,0,57,59,121,0,20,0,9,18,95,95,114,101,116,86,97, -108,0,59,120,0,58,100,111,116,0,18,114,48,0,0,18,118,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0, -59,121,0,58,100,111,116,0,18,114,49,0,0,18,118,0,0,0,20,0,0,1,0,10,2,21,1,1,0,10,118,0,0,1,1,0,13, -109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,100,111,116,0,18,118,0,0,18,109,0,16,8, -48,0,57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,100,111,116,0,18,118,0,0,18,109,0, -16,10,49,0,57,0,0,20,0,0,1,0,11,2,21,1,1,0,14,109,0,0,1,1,0,11,118,0,0,0,1,2,3,2,0,11,1,114,48,0,0, -0,9,18,114,48,0,59,120,0,18,109,0,16,8,48,0,57,59,120,0,20,0,9,18,114,48,0,59,121,0,18,109,0,16,10, -49,0,57,59,120,0,20,0,9,18,114,48,0,59,122,0,18,109,0,16,10,50,0,57,59,120,0,20,0,4,118,101,99,51, -95,100,111,116,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,114,48,0,0,18,118,0,0,0,0,2,3,2,0, -11,1,114,49,0,0,0,9,18,114,49,0,59,120,0,18,109,0,16,8,48,0,57,59,121,0,20,0,9,18,114,49,0,59,121, -0,18,109,0,16,10,49,0,57,59,121,0,20,0,9,18,114,49,0,59,122,0,18,109,0,16,10,50,0,57,59,121,0,20,0, -4,118,101,99,51,95,100,111,116,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,114,49,0,0,18,118, -0,0,0,0,2,3,2,0,11,1,114,50,0,0,0,9,18,114,50,0,59,120,0,18,109,0,16,8,48,0,57,59,122,0,20,0,9,18, -114,50,0,59,121,0,18,109,0,16,10,49,0,57,59,122,0,20,0,9,18,114,50,0,59,122,0,18,109,0,16,10,50,0, -57,59,122,0,20,0,4,118,101,99,51,95,100,111,116,0,18,95,95,114,101,116,86,97,108,0,59,122,0,0,18, -114,50,0,0,18,118,0,0,0,0,0,1,0,11,2,21,1,1,0,11,118,0,0,1,1,0,14,109,0,0,0,1,9,18,95,95,114,101, -116,86,97,108,0,59,120,0,58,100,111,116,0,18,118,0,0,18,109,0,16,8,48,0,57,0,0,20,0,9,18,95,95,114, -101,116,86,97,108,0,59,121,0,58,100,111,116,0,18,118,0,0,18,109,0,16,10,49,0,57,0,0,20,0,9,18,95, -95,114,101,116,86,97,108,0,59,122,0,58,100,111,116,0,18,118,0,0,18,109,0,16,10,50,0,57,0,0,20,0,0, -1,0,12,2,21,1,1,0,15,109,0,0,1,1,0,12,118,0,0,0,1,2,3,2,0,12,1,114,48,0,0,0,9,18,114,48,0,59,120,0, -18,109,0,16,8,48,0,57,59,120,0,20,0,9,18,114,48,0,59,121,0,18,109,0,16,10,49,0,57,59,120,0,20,0,9, -18,114,48,0,59,122,0,18,109,0,16,10,50,0,57,59,120,0,20,0,9,18,114,48,0,59,119,0,18,109,0,16,10,51, -0,57,59,120,0,20,0,4,118,101,99,52,95,100,111,116,0,18,95,95,114,101,116,86,97,108,0,59,115,0,0,18, -114,48,0,0,18,118,0,0,0,0,2,3,2,0,12,1,114,49,0,0,0,9,18,114,49,0,59,120,0,18,109,0,16,8,48,0,57, -59,121,0,20,0,9,18,114,49,0,59,121,0,18,109,0,16,10,49,0,57,59,121,0,20,0,9,18,114,49,0,59,122,0, -18,109,0,16,10,50,0,57,59,121,0,20,0,9,18,114,49,0,59,119,0,18,109,0,16,10,51,0,57,59,121,0,20,0,4, -118,101,99,52,95,100,111,116,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,114,49,0,0,18,118,0, -0,0,0,2,3,2,0,12,1,114,50,0,0,0,9,18,114,50,0,59,120,0,18,109,0,16,8,48,0,57,59,122,0,20,0,9,18, -114,50,0,59,121,0,18,109,0,16,10,49,0,57,59,122,0,20,0,9,18,114,50,0,59,122,0,18,109,0,16,10,50,0, -57,59,122,0,20,0,9,18,114,50,0,59,119,0,18,109,0,16,10,51,0,57,59,122,0,20,0,4,118,101,99,52,95, -100,111,116,0,18,95,95,114,101,116,86,97,108,0,59,122,0,0,18,114,50,0,0,18,118,0,0,0,0,2,3,2,0,12, -1,114,51,0,0,0,9,18,114,51,0,59,120,0,18,109,0,16,8,48,0,57,59,119,0,20,0,9,18,114,51,0,59,121,0, -18,109,0,16,10,49,0,57,59,119,0,20,0,9,18,114,51,0,59,122,0,18,109,0,16,10,50,0,57,59,119,0,20,0,9, -18,114,51,0,59,119,0,18,109,0,16,10,51,0,57,59,119,0,20,0,4,118,101,99,52,95,100,111,116,0,18,95, -95,114,101,116,86,97,108,0,59,119,0,0,18,114,51,0,0,18,118,0,0,0,0,0,1,0,12,2,21,1,1,0,12,118,0,0, -1,1,0,15,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,100,111,116,0,18,118,0,0,18, -109,0,16,8,48,0,57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,100,111,116,0,18,118,0, -0,18,109,0,16,10,49,0,57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,100,111,116,0,18, -118,0,0,18,109,0,16,10,50,0,57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,119,0,58,100,111,116, -0,18,118,0,0,18,109,0,16,10,51,0,57,0,0,20,0,0,1,0,0,2,1,1,0,2,13,109,0,0,1,1,0,13,110,0,0,0,1,9, -18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,21,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57, -21,0,0,1,0,0,2,2,1,0,2,13,109,0,0,1,1,0,13,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0, -57,22,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,22,0,0,1,0,0,2,3,1,0,2,13,109,0,0,1,1,0,13, -110,0,0,0,1,9,18,109,0,18,109,0,18,110,0,48,20,0,0,1,0,0,2,4,1,0,2,13,109,0,0,1,1,0,13,110,0,0,0,1, -9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,24,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57, -24,0,0,1,0,0,2,1,1,0,2,14,109,0,0,1,1,0,14,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0, -57,21,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,21,0,9,18,109,0,16,10,50,0,57,18,110,0,16, -10,50,0,57,21,0,0,1,0,0,2,2,1,0,2,14,109,0,0,1,1,0,14,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0, -16,8,48,0,57,22,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,22,0,9,18,109,0,16,10,50,0,57,18, -110,0,16,10,50,0,57,22,0,0,1,0,0,2,3,1,0,2,14,109,0,0,1,1,0,14,110,0,0,0,1,9,18,109,0,18,109,0,18, -110,0,48,20,0,0,1,0,0,2,4,1,0,2,14,109,0,0,1,1,0,14,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0, -16,8,48,0,57,24,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,24,0,9,18,109,0,16,10,50,0,57,18, -110,0,16,10,50,0,57,24,0,0,1,0,0,2,1,1,0,2,15,109,0,0,1,1,0,15,110,0,0,0,1,9,18,109,0,16,8,48,0,57, -18,110,0,16,8,48,0,57,21,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,21,0,9,18,109,0,16,10, -50,0,57,18,110,0,16,10,50,0,57,21,0,9,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,21,0,0,1,0,0,2, -2,1,0,2,15,109,0,0,1,1,0,15,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,22,0,9,18, +16,10,51,0,57,18,98,0,49,20,0,0,1,0,10,2,21,1,1,0,13,109,0,0,1,1,0,10,118,0,0,0,1,9,18,95,95,114, +101,116,86,97,108,0,18,109,0,16,8,48,0,57,18,118,0,59,120,120,0,48,18,109,0,16,10,49,0,57,18,118,0, +59,121,121,0,48,46,20,0,0,1,0,10,2,21,1,1,0,10,118,0,0,1,1,0,13,109,0,0,0,1,9,18,95,95,114,101,116, +86,97,108,0,59,120,0,58,100,111,116,0,18,118,0,0,18,109,0,16,8,48,0,57,0,0,20,0,9,18,95,95,114,101, +116,86,97,108,0,59,121,0,58,100,111,116,0,18,118,0,0,18,109,0,16,10,49,0,57,0,0,20,0,0,1,0,11,2,21, +1,1,0,14,109,0,0,1,1,0,11,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,109,0,16,8,48,0,57,18, +118,0,59,120,120,120,0,48,18,109,0,16,10,49,0,57,18,118,0,59,121,121,121,0,48,46,18,109,0,16,10,50, +0,57,18,118,0,59,122,122,122,0,48,46,20,0,0,1,0,11,2,21,1,1,0,11,118,0,0,1,1,0,14,109,0,0,0,1,9,18, +95,95,114,101,116,86,97,108,0,59,120,0,58,100,111,116,0,18,118,0,0,18,109,0,16,8,48,0,57,0,0,20,0, +9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,100,111,116,0,18,118,0,0,18,109,0,16,10,49,0,57,0,0, +20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,100,111,116,0,18,118,0,0,18,109,0,16,10,50,0, +57,0,0,20,0,0,1,0,12,2,21,1,1,0,15,109,0,0,1,1,0,12,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0, +18,109,0,16,8,48,0,57,18,118,0,59,120,120,120,120,0,48,18,109,0,16,10,49,0,57,18,118,0,59,121,121, +121,121,0,48,46,18,109,0,16,10,50,0,57,18,118,0,59,122,122,122,122,0,48,46,18,109,0,16,10,51,0,57, +18,118,0,59,119,119,119,119,0,48,46,20,0,0,1,0,12,2,21,1,1,0,12,118,0,0,1,1,0,15,109,0,0,0,1,9,18, +95,95,114,101,116,86,97,108,0,59,120,0,58,100,111,116,0,18,118,0,0,18,109,0,16,8,48,0,57,0,0,20,0, +9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,100,111,116,0,18,118,0,0,18,109,0,16,10,49,0,57,0,0, +20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,100,111,116,0,18,118,0,0,18,109,0,16,10,50,0, +57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,119,0,58,100,111,116,0,18,118,0,0,18,109,0,16,10, +51,0,57,0,0,20,0,0,1,0,0,2,1,1,0,2,13,109,0,0,1,1,0,13,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110, +0,16,8,48,0,57,21,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,21,0,0,1,0,0,2,2,1,0,2,13,109, +0,0,1,1,0,13,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,22,0,9,18,109,0,16,10,49,0, +57,18,110,0,16,10,49,0,57,22,0,0,1,0,0,2,3,1,0,2,13,109,0,0,1,1,0,13,110,0,0,0,1,9,18,109,0,18,109, +0,18,110,0,48,20,0,0,1,0,0,2,4,1,0,2,13,109,0,0,1,1,0,13,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18, +110,0,16,8,48,0,57,24,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,24,0,0,1,0,0,2,1,1,0,2,14, +109,0,0,1,1,0,14,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,21,0,9,18,109,0,16,10, +49,0,57,18,110,0,16,10,49,0,57,21,0,9,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,21,0,0,1,0,0,2, +2,1,0,2,14,109,0,0,1,1,0,14,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,22,0,9,18, 109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,22,0,9,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,22, -0,9,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,22,0,0,1,0,0,2,3,1,0,2,15,109,0,0,1,1,0,15,110,0, -0,0,1,9,18,109,0,18,109,0,18,110,0,48,20,0,0,1,0,0,2,4,1,0,2,15,109,0,0,1,1,0,15,110,0,0,0,1,9,18, -109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,24,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,24,0, -9,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,24,0,9,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0, -57,24,0,0,1,0,0,2,1,1,0,2,13,109,0,0,1,1,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,21,0,9,18, -109,0,16,10,49,0,57,18,97,0,21,0,0,1,0,0,2,2,1,0,2,13,109,0,0,1,1,0,9,97,0,0,0,1,9,18,109,0,16,8, -48,0,57,18,97,0,22,0,9,18,109,0,16,10,49,0,57,18,97,0,22,0,0,1,0,0,2,3,1,0,2,13,109,0,0,1,1,0,9,97, -0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,23,0,9,18,109,0,16,10,49,0,57,18,97,0,23,0,0,1,0,0,2,4,1,0, -2,13,109,0,0,1,1,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,24,0,9,18,109,0,16,10,49,0,57,18, -97,0,24,0,0,1,0,0,2,1,1,0,2,14,109,0,0,1,1,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,21,0,9, -18,109,0,16,10,49,0,57,18,97,0,21,0,9,18,109,0,16,10,50,0,57,18,97,0,21,0,0,1,0,0,2,2,1,0,2,14,109, -0,0,1,1,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,22,0,9,18,109,0,16,10,49,0,57,18,97,0,22,0, -9,18,109,0,16,10,50,0,57,18,97,0,22,0,0,1,0,0,2,3,1,0,2,14,109,0,0,1,1,0,9,97,0,0,0,1,9,18,109,0, -16,8,48,0,57,18,97,0,23,0,9,18,109,0,16,10,49,0,57,18,97,0,23,0,9,18,109,0,16,10,50,0,57,18,97,0, -23,0,0,1,0,0,2,4,1,0,2,14,109,0,0,1,1,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,24,0,9,18,109, -0,16,10,49,0,57,18,97,0,24,0,9,18,109,0,16,10,50,0,57,18,97,0,24,0,0,1,0,0,2,1,1,0,2,15,109,0,0,1, -1,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,21,0,9,18,109,0,16,10,49,0,57,18,97,0,21,0,9,18, -109,0,16,10,50,0,57,18,97,0,21,0,9,18,109,0,16,10,51,0,57,18,97,0,21,0,0,1,0,0,2,2,1,0,2,15,109,0, -0,1,1,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,22,0,9,18,109,0,16,10,49,0,57,18,97,0,22,0,9, -18,109,0,16,10,50,0,57,18,97,0,22,0,9,18,109,0,16,10,51,0,57,18,97,0,22,0,0,1,0,0,2,3,1,0,2,15,109, -0,0,1,1,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,23,0,9,18,109,0,16,10,49,0,57,18,97,0,23,0, -9,18,109,0,16,10,50,0,57,18,97,0,23,0,9,18,109,0,16,10,51,0,57,18,97,0,23,0,0,1,0,0,2,4,1,0,2,15, -109,0,0,1,1,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,24,0,9,18,109,0,16,10,49,0,57,18,97,0, -24,0,9,18,109,0,16,10,50,0,57,18,97,0,24,0,9,18,109,0,16,10,51,0,57,18,97,0,24,0,0,1,0,0,2,3,1,0,2, -10,118,0,0,1,1,0,13,109,0,0,0,1,9,18,118,0,18,118,0,18,109,0,48,20,0,0,1,0,0,2,3,1,0,2,11,118,0,0, -1,1,0,14,109,0,0,0,1,9,18,118,0,18,118,0,18,109,0,48,20,0,0,1,0,0,2,3,1,0,2,12,118,0,0,1,1,0,15, -109,0,0,0,1,9,18,118,0,18,118,0,18,109,0,48,20,0,0,1,0,5,2,25,1,0,2,5,97,0,0,0,1,9,18,97,0,18,97,0, -16,10,49,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,18,97,0,20,0,0,1,0,6,2,25,1,0,2,6,118,0,0,0, -1,9,18,118,0,18,118,0,58,105,118,101,99,50,0,16,10,49,0,0,0,47,20,0,9,18,95,95,114,101,116,86,97, -108,0,18,118,0,20,0,0,1,0,7,2,25,1,0,2,7,118,0,0,0,1,9,18,118,0,18,118,0,58,105,118,101,99,51,0,16, -10,49,0,0,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,0,1,0,8,2,25,1,0,2,8,118,0,0, -0,1,9,18,118,0,18,118,0,58,105,118,101,99,52,0,16,10,49,0,0,0,47,20,0,9,18,95,95,114,101,116,86,97, -108,0,18,118,0,20,0,0,1,0,9,2,25,1,0,2,9,97,0,0,0,1,9,18,97,0,18,97,0,17,49,0,48,0,0,47,20,0,9,18, -95,95,114,101,116,86,97,108,0,18,97,0,20,0,0,1,0,10,2,25,1,0,2,10,118,0,0,0,1,9,18,118,0,18,118,0, -58,118,101,99,50,0,17,49,0,48,0,0,0,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,0,1, -0,11,2,25,1,0,2,11,118,0,0,0,1,9,18,118,0,18,118,0,58,118,101,99,51,0,17,49,0,48,0,0,0,0,47,20,0,9, -18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,0,1,0,12,2,25,1,0,2,12,118,0,0,0,1,9,18,118,0,18, -118,0,58,118,101,99,52,0,17,49,0,48,0,0,0,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20, -0,0,1,0,13,2,25,1,0,2,13,109,0,0,0,1,9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0,57,58,118,101,99, -50,0,17,49,0,48,0,0,0,0,47,20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49,0,57,58,118,101,99,50,0, -17,49,0,48,0,0,0,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,18,109,0,20,0,0,1,0,14,2,25,1,0,2,14, -109,0,0,0,1,9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0,57,58,118,101,99,51,0,17,49,0,48,0,0,0,0,47, -20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49,0,57,58,118,101,99,51,0,17,49,0,48,0,0,0,0,47,20,0, -9,18,109,0,16,10,50,0,57,18,109,0,16,10,50,0,57,58,118,101,99,51,0,17,49,0,48,0,0,0,0,47,20,0,9,18, -95,95,114,101,116,86,97,108,0,18,109,0,20,0,0,1,0,15,2,25,1,0,2,15,109,0,0,0,1,9,18,109,0,16,8,48, -0,57,18,109,0,16,8,48,0,57,58,118,101,99,52,0,17,49,0,48,0,0,0,0,47,20,0,9,18,109,0,16,10,49,0,57, -18,109,0,16,10,49,0,57,58,118,101,99,52,0,17,49,0,48,0,0,0,0,47,20,0,9,18,109,0,16,10,50,0,57,18, -109,0,16,10,50,0,57,58,118,101,99,52,0,17,49,0,48,0,0,0,0,47,20,0,9,18,109,0,16,10,51,0,57,18,109, -0,16,10,51,0,57,58,118,101,99,52,0,17,49,0,48,0,0,0,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0, -18,109,0,20,0,0,1,0,5,2,24,1,0,2,5,97,0,0,0,1,9,18,97,0,18,97,0,16,10,49,0,46,20,0,9,18,95,95,114, -101,116,86,97,108,0,18,97,0,20,0,0,1,0,6,2,24,1,0,2,6,118,0,0,0,1,9,18,118,0,18,118,0,58,105,118, -101,99,50,0,16,10,49,0,0,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,0,1,0,7,2,24,1, -0,2,7,118,0,0,0,1,9,18,118,0,18,118,0,58,105,118,101,99,51,0,16,10,49,0,0,0,46,20,0,9,18,95,95,114, -101,116,86,97,108,0,18,118,0,20,0,0,1,0,8,2,24,1,0,2,8,118,0,0,0,1,9,18,118,0,18,118,0,58,105,118, -101,99,52,0,16,10,49,0,0,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,0,1,0,9,2,24,1, -0,2,9,97,0,0,0,1,9,18,97,0,18,97,0,17,49,0,48,0,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,18,97, -0,20,0,0,1,0,10,2,24,1,0,2,10,118,0,0,0,1,9,18,118,0,18,118,0,58,118,101,99,50,0,17,49,0,48,0,0,0, -0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,0,1,0,11,2,24,1,0,2,11,118,0,0,0,1,9,18, -118,0,18,118,0,58,118,101,99,51,0,17,49,0,48,0,0,0,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,18, -118,0,20,0,0,1,0,12,2,24,1,0,2,12,118,0,0,0,1,9,18,118,0,18,118,0,58,118,101,99,52,0,17,49,0,48,0, -0,0,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,0,1,0,13,2,24,1,0,2,13,109,0,0,0,1, -9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0,57,58,118,101,99,50,0,17,49,0,48,0,0,0,0,46,20,0,9,18, -109,0,16,10,49,0,57,18,109,0,16,10,49,0,57,58,118,101,99,50,0,17,49,0,48,0,0,0,0,46,20,0,9,18,95, -95,114,101,116,86,97,108,0,18,109,0,20,0,0,1,0,14,2,24,1,0,2,14,109,0,0,0,1,9,18,109,0,16,8,48,0, -57,18,109,0,16,8,48,0,57,58,118,101,99,51,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,49,0,57,18, -109,0,16,10,49,0,57,58,118,101,99,51,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,50,0,57,18,109, -0,16,10,50,0,57,58,118,101,99,51,0,17,49,0,48,0,0,0,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0, -18,109,0,20,0,0,1,0,15,2,24,1,0,2,15,109,0,0,0,1,9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0,57,58, -118,101,99,52,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49,0,57,58,118, -101,99,52,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,50,0,57,18,109,0,16,10,50,0,57,58,118,101, -99,52,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,51,0,57,18,109,0,16,10,51,0,57,58,118,101,99, -52,0,17,49,0,48,0,0,0,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,18,109,0,20,0,0,1,0,5,0,95,95, -112,111,115,116,68,101,99,114,0,1,0,2,5,97,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,97,0,20,0, -9,18,97,0,18,97,0,16,10,49,0,47,20,0,0,1,0,6,0,95,95,112,111,115,116,68,101,99,114,0,1,0,2,6,118,0, -0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,9,18,118,0,18,118,0,58,105,118,101,99,50,0, -16,10,49,0,0,0,47,20,0,0,1,0,7,0,95,95,112,111,115,116,68,101,99,114,0,1,0,2,7,118,0,0,0,1,9,18,95, -95,114,101,116,86,97,108,0,18,118,0,20,0,9,18,118,0,18,118,0,58,105,118,101,99,51,0,16,10,49,0,0,0, -47,20,0,0,1,0,8,0,95,95,112,111,115,116,68,101,99,114,0,1,0,2,8,118,0,0,0,1,9,18,95,95,114,101,116, -86,97,108,0,18,118,0,20,0,9,18,118,0,18,118,0,58,105,118,101,99,52,0,16,10,49,0,0,0,47,20,0,0,1,0, -9,0,95,95,112,111,115,116,68,101,99,114,0,1,0,2,9,97,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18, -97,0,20,0,9,18,97,0,18,97,0,17,49,0,48,0,0,47,20,0,0,1,0,10,0,95,95,112,111,115,116,68,101,99,114, -0,1,0,2,10,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,9,18,118,0,18,118,0,58,118, -101,99,50,0,17,49,0,48,0,0,0,0,47,20,0,0,1,0,11,0,95,95,112,111,115,116,68,101,99,114,0,1,0,2,11, -118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,9,18,118,0,18,118,0,58,118,101,99,51, -0,17,49,0,48,0,0,0,0,47,20,0,0,1,0,12,0,95,95,112,111,115,116,68,101,99,114,0,1,0,2,12,118,0,0,0,1, -9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,9,18,118,0,18,118,0,58,118,101,99,52,0,17,49,0,48, -0,0,0,0,47,20,0,0,1,0,13,0,95,95,112,111,115,116,68,101,99,114,0,1,0,2,13,109,0,0,0,1,9,18,95,95, -114,101,116,86,97,108,0,18,109,0,20,0,9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0,57,58,118,101,99, -50,0,17,49,0,48,0,0,0,0,47,20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49,0,57,58,118,101,99,50,0, -17,49,0,48,0,0,0,0,47,20,0,0,1,0,14,0,95,95,112,111,115,116,68,101,99,114,0,1,0,2,14,109,0,0,0,1,9, -18,95,95,114,101,116,86,97,108,0,18,109,0,20,0,9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0,57,58, -118,101,99,51,0,17,49,0,48,0,0,0,0,47,20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49,0,57,58,118, -101,99,51,0,17,49,0,48,0,0,0,0,47,20,0,9,18,109,0,16,10,50,0,57,18,109,0,16,10,50,0,57,58,118,101, -99,51,0,17,49,0,48,0,0,0,0,47,20,0,0,1,0,15,0,95,95,112,111,115,116,68,101,99,114,0,1,0,2,15,109,0, -0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,109,0,20,0,9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0, +0,0,1,0,0,2,3,1,0,2,14,109,0,0,1,1,0,14,110,0,0,0,1,9,18,109,0,18,109,0,18,110,0,48,20,0,0,1,0,0,2, +4,1,0,2,14,109,0,0,1,1,0,14,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,24,0,9,18, +109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,24,0,9,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,24, +0,0,1,0,0,2,1,1,0,2,15,109,0,0,1,1,0,15,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57, +21,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,21,0,9,18,109,0,16,10,50,0,57,18,110,0,16,10, +50,0,57,21,0,9,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,21,0,0,1,0,0,2,2,1,0,2,15,109,0,0,1,1, +0,15,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,22,0,9,18,109,0,16,10,49,0,57,18, +110,0,16,10,49,0,57,22,0,9,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,22,0,9,18,109,0,16,10,51, +0,57,18,110,0,16,10,51,0,57,22,0,0,1,0,0,2,3,1,0,2,15,109,0,0,1,1,0,15,110,0,0,0,1,9,18,109,0,18, +109,0,18,110,0,48,20,0,0,1,0,0,2,4,1,0,2,15,109,0,0,1,1,0,15,110,0,0,0,1,9,18,109,0,16,8,48,0,57, +18,110,0,16,8,48,0,57,24,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,24,0,9,18,109,0,16,10, +50,0,57,18,110,0,16,10,50,0,57,24,0,9,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,24,0,0,1,0,0,2, +1,1,0,2,13,109,0,0,1,1,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,21,0,9,18,109,0,16,10,49,0, +57,18,97,0,21,0,0,1,0,0,2,2,1,0,2,13,109,0,0,1,1,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,22, +0,9,18,109,0,16,10,49,0,57,18,97,0,22,0,0,1,0,0,2,3,1,0,2,13,109,0,0,1,1,0,9,97,0,0,0,1,9,18,109,0, +16,8,48,0,57,18,97,0,23,0,9,18,109,0,16,10,49,0,57,18,97,0,23,0,0,1,0,0,2,4,1,0,2,13,109,0,0,1,1,0, +9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,24,0,9,18,109,0,16,10,49,0,57,18,97,0,24,0,0,1,0,0,2, +1,1,0,2,14,109,0,0,1,1,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,21,0,9,18,109,0,16,10,49,0, +57,18,97,0,21,0,9,18,109,0,16,10,50,0,57,18,97,0,21,0,0,1,0,0,2,2,1,0,2,14,109,0,0,1,1,0,9,97,0,0, +0,1,9,18,109,0,16,8,48,0,57,18,97,0,22,0,9,18,109,0,16,10,49,0,57,18,97,0,22,0,9,18,109,0,16,10,50, +0,57,18,97,0,22,0,0,1,0,0,2,3,1,0,2,14,109,0,0,1,1,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0, +23,0,9,18,109,0,16,10,49,0,57,18,97,0,23,0,9,18,109,0,16,10,50,0,57,18,97,0,23,0,0,1,0,0,2,4,1,0,2, +14,109,0,0,1,1,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,24,0,9,18,109,0,16,10,49,0,57,18,97, +0,24,0,9,18,109,0,16,10,50,0,57,18,97,0,24,0,0,1,0,0,2,1,1,0,2,15,109,0,0,1,1,0,9,97,0,0,0,1,9,18, +109,0,16,8,48,0,57,18,97,0,21,0,9,18,109,0,16,10,49,0,57,18,97,0,21,0,9,18,109,0,16,10,50,0,57,18, +97,0,21,0,9,18,109,0,16,10,51,0,57,18,97,0,21,0,0,1,0,0,2,2,1,0,2,15,109,0,0,1,1,0,9,97,0,0,0,1,9, +18,109,0,16,8,48,0,57,18,97,0,22,0,9,18,109,0,16,10,49,0,57,18,97,0,22,0,9,18,109,0,16,10,50,0,57, +18,97,0,22,0,9,18,109,0,16,10,51,0,57,18,97,0,22,0,0,1,0,0,2,3,1,0,2,15,109,0,0,1,1,0,9,97,0,0,0,1, +9,18,109,0,16,8,48,0,57,18,97,0,23,0,9,18,109,0,16,10,49,0,57,18,97,0,23,0,9,18,109,0,16,10,50,0, +57,18,97,0,23,0,9,18,109,0,16,10,51,0,57,18,97,0,23,0,0,1,0,0,2,4,1,0,2,15,109,0,0,1,1,0,9,97,0,0, +0,1,9,18,109,0,16,8,48,0,57,18,97,0,24,0,9,18,109,0,16,10,49,0,57,18,97,0,24,0,9,18,109,0,16,10,50, +0,57,18,97,0,24,0,9,18,109,0,16,10,51,0,57,18,97,0,24,0,0,1,0,0,2,3,1,0,2,10,118,0,0,1,1,0,13,109, +0,0,0,1,9,18,118,0,18,118,0,18,109,0,48,20,0,0,1,0,0,2,3,1,0,2,11,118,0,0,1,1,0,14,109,0,0,0,1,9, +18,118,0,18,118,0,18,109,0,48,20,0,0,1,0,0,2,3,1,0,2,12,118,0,0,1,1,0,15,109,0,0,0,1,9,18,118,0,18, +118,0,18,109,0,48,20,0,0,1,0,5,2,25,1,0,2,5,97,0,0,0,1,9,18,97,0,18,97,0,16,10,49,0,47,20,0,9,18, +95,95,114,101,116,86,97,108,0,18,97,0,20,0,0,1,0,6,2,25,1,0,2,6,118,0,0,0,1,9,18,118,0,18,118,0,58, +105,118,101,99,50,0,16,10,49,0,0,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,0,1,0, +7,2,25,1,0,2,7,118,0,0,0,1,9,18,118,0,18,118,0,58,105,118,101,99,51,0,16,10,49,0,0,0,47,20,0,9,18, +95,95,114,101,116,86,97,108,0,18,118,0,20,0,0,1,0,8,2,25,1,0,2,8,118,0,0,0,1,9,18,118,0,18,118,0, +58,105,118,101,99,52,0,16,10,49,0,0,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,0,1, +0,9,2,25,1,0,2,9,97,0,0,0,1,9,18,97,0,18,97,0,17,49,0,48,0,0,47,20,0,9,18,95,95,114,101,116,86,97, +108,0,18,97,0,20,0,0,1,0,10,2,25,1,0,2,10,118,0,0,0,1,9,18,118,0,18,118,0,58,118,101,99,50,0,17,49, +0,48,0,0,0,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,0,1,0,11,2,25,1,0,2,11,118,0, +0,0,1,9,18,118,0,18,118,0,58,118,101,99,51,0,17,49,0,48,0,0,0,0,47,20,0,9,18,95,95,114,101,116,86, +97,108,0,18,118,0,20,0,0,1,0,12,2,25,1,0,2,12,118,0,0,0,1,9,18,118,0,18,118,0,58,118,101,99,52,0, +17,49,0,48,0,0,0,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,0,1,0,13,2,25,1,0,2,13, +109,0,0,0,1,9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0,57,58,118,101,99,50,0,17,49,0,48,0,0,0,0,47, +20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49,0,57,58,118,101,99,50,0,17,49,0,48,0,0,0,0,47,20,0, +9,18,95,95,114,101,116,86,97,108,0,18,109,0,20,0,0,1,0,14,2,25,1,0,2,14,109,0,0,0,1,9,18,109,0,16, +8,48,0,57,18,109,0,16,8,48,0,57,58,118,101,99,51,0,17,49,0,48,0,0,0,0,47,20,0,9,18,109,0,16,10,49, +0,57,18,109,0,16,10,49,0,57,58,118,101,99,51,0,17,49,0,48,0,0,0,0,47,20,0,9,18,109,0,16,10,50,0,57, +18,109,0,16,10,50,0,57,58,118,101,99,51,0,17,49,0,48,0,0,0,0,47,20,0,9,18,95,95,114,101,116,86,97, +108,0,18,109,0,20,0,0,1,0,15,2,25,1,0,2,15,109,0,0,0,1,9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0, 57,58,118,101,99,52,0,17,49,0,48,0,0,0,0,47,20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49,0,57, 58,118,101,99,52,0,17,49,0,48,0,0,0,0,47,20,0,9,18,109,0,16,10,50,0,57,18,109,0,16,10,50,0,57,58, 118,101,99,52,0,17,49,0,48,0,0,0,0,47,20,0,9,18,109,0,16,10,51,0,57,18,109,0,16,10,51,0,57,58,118, -101,99,52,0,17,49,0,48,0,0,0,0,47,20,0,0,1,0,9,0,95,95,112,111,115,116,73,110,99,114,0,1,0,2,9,97, -0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,97,0,20,0,9,18,97,0,18,97,0,16,10,49,0,46,20,0,0,1,0, -10,0,95,95,112,111,115,116,73,110,99,114,0,1,0,2,10,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0, -18,118,0,20,0,9,18,118,0,18,118,0,58,118,101,99,50,0,17,49,0,48,0,0,0,0,46,20,0,0,1,0,11,0,95,95, -112,111,115,116,73,110,99,114,0,1,0,2,11,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0, -20,0,9,18,118,0,18,118,0,58,118,101,99,51,0,17,49,0,48,0,0,0,0,46,20,0,0,1,0,12,0,95,95,112,111, -115,116,73,110,99,114,0,1,0,2,12,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,9,18, -118,0,18,118,0,58,118,101,99,52,0,17,49,0,48,0,0,0,0,46,20,0,0,1,0,5,0,95,95,112,111,115,116,73, -110,99,114,0,1,0,2,5,97,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,97,0,20,0,9,18,97,0,18,97,0, -16,10,49,0,46,20,0,0,1,0,6,0,95,95,112,111,115,116,73,110,99,114,0,1,0,2,6,118,0,0,0,1,9,18,95,95, -114,101,116,86,97,108,0,18,118,0,20,0,9,18,118,0,18,118,0,58,105,118,101,99,50,0,16,10,49,0,0,0,46, -20,0,0,1,0,7,0,95,95,112,111,115,116,73,110,99,114,0,1,0,2,7,118,0,0,0,1,9,18,95,95,114,101,116,86, -97,108,0,18,118,0,20,0,9,18,118,0,18,118,0,58,105,118,101,99,51,0,16,10,49,0,0,0,46,20,0,0,1,0,8,0, -95,95,112,111,115,116,73,110,99,114,0,1,0,2,8,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18, -118,0,20,0,9,18,118,0,18,118,0,58,105,118,101,99,51,0,16,10,49,0,0,0,46,20,0,0,1,0,13,0,95,95,112, -111,115,116,73,110,99,114,0,1,0,2,13,109,0,0,0,1,3,2,0,13,1,110,0,2,18,109,0,0,0,9,18,109,0,16,8, +101,99,52,0,17,49,0,48,0,0,0,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,18,109,0,20,0,0,1,0,5,2, +24,1,0,2,5,97,0,0,0,1,9,18,97,0,18,97,0,16,10,49,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,18, +97,0,20,0,0,1,0,6,2,24,1,0,2,6,118,0,0,0,1,9,18,118,0,18,118,0,58,105,118,101,99,50,0,16,10,49,0,0, +0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,0,1,0,7,2,24,1,0,2,7,118,0,0,0,1,9,18, +118,0,18,118,0,58,105,118,101,99,51,0,16,10,49,0,0,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,18, +118,0,20,0,0,1,0,8,2,24,1,0,2,8,118,0,0,0,1,9,18,118,0,18,118,0,58,105,118,101,99,52,0,16,10,49,0, +0,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,0,1,0,9,2,24,1,0,2,9,97,0,0,0,1,9,18, +97,0,18,97,0,17,49,0,48,0,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,18,97,0,20,0,0,1,0,10,2,24, +1,0,2,10,118,0,0,0,1,9,18,118,0,18,118,0,58,118,101,99,50,0,17,49,0,48,0,0,0,0,46,20,0,9,18,95,95, +114,101,116,86,97,108,0,18,118,0,20,0,0,1,0,11,2,24,1,0,2,11,118,0,0,0,1,9,18,118,0,18,118,0,58, +118,101,99,51,0,17,49,0,48,0,0,0,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,0,1,0, +12,2,24,1,0,2,12,118,0,0,0,1,9,18,118,0,18,118,0,58,118,101,99,52,0,17,49,0,48,0,0,0,0,46,20,0,9, +18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,0,1,0,13,2,24,1,0,2,13,109,0,0,0,1,9,18,109,0,16,8, 48,0,57,18,109,0,16,8,48,0,57,58,118,101,99,50,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,49,0, -57,18,109,0,16,10,49,0,57,58,118,101,99,50,0,17,49,0,48,0,0,0,0,46,20,0,8,18,110,0,0,0,1,0,14,0,95, -95,112,111,115,116,73,110,99,114,0,1,0,2,14,109,0,0,0,1,3,2,0,14,1,110,0,2,18,109,0,0,0,9,18,109,0, -16,8,48,0,57,18,109,0,16,8,48,0,57,58,118,101,99,51,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10, -49,0,57,18,109,0,16,10,49,0,57,58,118,101,99,51,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,50,0, -57,18,109,0,16,10,50,0,57,58,118,101,99,51,0,17,49,0,48,0,0,0,0,46,20,0,8,18,110,0,0,0,1,0,15,0,95, -95,112,111,115,116,73,110,99,114,0,1,0,2,15,109,0,0,0,1,3,2,0,15,1,110,0,2,18,109,0,0,0,9,18,109,0, -16,8,48,0,57,18,109,0,16,8,48,0,57,58,118,101,99,52,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10, -49,0,57,18,109,0,16,10,49,0,57,58,118,101,99,52,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,50,0, -57,18,109,0,16,10,50,0,57,58,118,101,99,52,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,51,0,57, -18,109,0,16,10,51,0,57,58,118,101,99,52,0,17,49,0,48,0,0,0,0,46,20,0,8,18,110,0,0,0,1,0,1,2,15,1,1, -0,9,97,0,0,1,1,0,9,98,0,0,0,1,4,118,101,99,52,95,115,103,116,0,18,95,95,114,101,116,86,97,108,0,59, -120,0,0,18,98,0,0,18,97,0,0,0,0,1,0,1,2,15,1,1,0,5,97,0,0,1,1,0,5,98,0,0,0,1,8,58,102,108,111,97, -116,0,18,97,0,0,0,58,102,108,111,97,116,0,18,98,0,0,0,40,0,0,1,0,1,2,16,1,1,0,9,97,0,0,1,1,0,9,98, -0,0,0,1,3,2,0,1,1,99,0,0,0,4,102,108,111,97,116,95,108,101,115,115,0,18,99,0,0,18,98,0,0,18,97,0,0, -0,8,18,99,0,0,0,1,0,1,2,16,1,1,0,5,97,0,0,1,1,0,5,98,0,0,0,1,8,58,102,108,111,97,116,0,18,97,0,0,0, -58,102,108,111,97,116,0,18,98,0,0,0,41,0,0,1,0,1,2,18,1,1,0,9,97,0,0,1,1,0,9,98,0,0,0,1,3,2,0,1,1, -103,0,0,1,1,101,0,0,0,4,102,108,111,97,116,95,108,101,115,115,0,18,103,0,0,18,98,0,0,18,97,0,0,0,4, -102,108,111,97,116,95,101,113,117,97,108,0,18,101,0,0,18,97,0,0,18,98,0,0,0,8,18,103,0,18,101,0,32, -0,0,1,0,1,2,18,1,1,0,5,97,0,0,1,1,0,5,98,0,0,0,1,8,58,102,108,111,97,116,0,18,97,0,0,0,58,102,108, -111,97,116,0,18,98,0,0,0,43,0,0,1,0,1,2,17,1,1,0,9,97,0,0,1,1,0,9,98,0,0,0,1,3,2,0,1,1,103,0,0,1,1, -101,0,0,0,4,102,108,111,97,116,95,108,101,115,115,0,18,103,0,0,18,97,0,0,18,98,0,0,0,4,102,108,111, -97,116,95,101,113,117,97,108,0,18,101,0,0,18,97,0,0,18,98,0,0,0,8,18,103,0,18,101,0,32,0,0,1,0,1,2, -17,1,1,0,5,97,0,0,1,1,0,5,98,0,0,0,1,8,58,102,108,111,97,116,0,18,97,0,0,0,58,102,108,111,97,116,0, -18,98,0,0,0,42,0,0,1,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,9,102,0,0,0,1,4,102,108,111,97, -116,95,112,114,105,110,116,0,18,102,0,0,0,0,1,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,5,105, -0,0,0,1,4,105,110,116,95,112,114,105,110,116,0,18,105,0,0,0,0,1,0,0,0,112,114,105,110,116,77,69,83, -65,0,1,1,0,1,98,0,0,0,1,4,98,111,111,108,95,112,114,105,110,116,0,18,98,0,0,0,0,1,0,0,0,112,114, -105,110,116,77,69,83,65,0,1,1,0,10,118,0,0,0,1,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59, -120,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,121,0,0,0,0,0,1,0,0,0,112,114,105, -110,116,77,69,83,65,0,1,1,0,11,118,0,0,0,1,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,120, -0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,121,0,0,0,0,9,58,112,114,105,110,116,77, -69,83,65,0,18,118,0,59,122,0,0,0,0,0,1,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,12,118,0,0,0, -1,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,120,0,0,0,0,9,58,112,114,105,110,116,77,69,83, -65,0,18,118,0,59,121,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,122,0,0,0,0,9,58, -112,114,105,110,116,77,69,83,65,0,18,118,0,59,119,0,0,0,0,0,1,0,0,0,112,114,105,110,116,77,69,83, -65,0,1,1,0,6,118,0,0,0,1,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,120,0,0,0,0,9,58,112, -114,105,110,116,77,69,83,65,0,18,118,0,59,121,0,0,0,0,0,1,0,0,0,112,114,105,110,116,77,69,83,65,0, -1,1,0,7,118,0,0,0,1,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,120,0,0,0,0,9,58,112,114, -105,110,116,77,69,83,65,0,18,118,0,59,121,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0, -59,122,0,0,0,0,0,1,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,8,118,0,0,0,1,9,58,112,114,105, -110,116,77,69,83,65,0,18,118,0,59,120,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59, -121,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,122,0,0,0,0,9,58,112,114,105,110, -116,77,69,83,65,0,18,118,0,59,119,0,0,0,0,0,1,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,2,118, -0,0,0,1,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,120,0,0,0,0,9,58,112,114,105,110,116,77, -69,83,65,0,18,118,0,59,121,0,0,0,0,0,1,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,3,118,0,0,0,1, -9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,120,0,0,0,0,9,58,112,114,105,110,116,77,69,83, -65,0,18,118,0,59,121,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,122,0,0,0,0,0,1,0, -0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,4,118,0,0,0,1,9,58,112,114,105,110,116,77,69,83,65,0, -18,118,0,59,120,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,121,0,0,0,0,9,58,112, -114,105,110,116,77,69,83,65,0,18,118,0,59,122,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18, -118,0,59,119,0,0,0,0,0,1,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,13,109,0,0,0,1,9,58,112,114, -105,110,116,77,69,83,65,0,18,109,0,16,8,48,0,57,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18, -109,0,16,10,49,0,57,0,0,0,0,1,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,14,109,0,0,0,1,9,58, -112,114,105,110,116,77,69,83,65,0,18,109,0,16,8,48,0,57,0,0,0,9,58,112,114,105,110,116,77,69,83,65, -0,18,109,0,16,10,49,0,57,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,109,0,16,10,50,0,57,0,0,0, -0,1,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,15,109,0,0,0,1,9,58,112,114,105,110,116,77,69,83, -65,0,18,109,0,16,8,48,0,57,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,109,0,16,10,49,0,57,0,0, -0,9,58,112,114,105,110,116,77,69,83,65,0,18,109,0,16,10,50,0,57,0,0,0,9,58,112,114,105,110,116,77, -69,83,65,0,18,109,0,16,10,51,0,57,0,0,0,0,1,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,16,101,0, -0,0,1,4,105,110,116,95,112,114,105,110,116,0,18,101,0,0,0,0,1,0,0,0,112,114,105,110,116,77,69,83, -65,0,1,1,0,17,101,0,0,0,1,4,105,110,116,95,112,114,105,110,116,0,18,101,0,0,0,0,1,0,0,0,112,114, -105,110,116,77,69,83,65,0,1,1,0,18,101,0,0,0,1,4,105,110,116,95,112,114,105,110,116,0,18,101,0,0,0, -0,1,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,19,101,0,0,0,1,4,105,110,116,95,112,114,105,110, -116,0,18,101,0,0,0,0,1,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,20,101,0,0,0,1,4,105,110,116, -95,112,114,105,110,116,0,18,101,0,0,0,0,1,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,21,101,0,0, -0,1,4,105,110,116,95,112,114,105,110,116,0,18,101,0,0,0,0,0 +57,18,109,0,16,10,49,0,57,58,118,101,99,50,0,17,49,0,48,0,0,0,0,46,20,0,9,18,95,95,114,101,116,86, +97,108,0,18,109,0,20,0,0,1,0,14,2,24,1,0,2,14,109,0,0,0,1,9,18,109,0,16,8,48,0,57,18,109,0,16,8,48, +0,57,58,118,101,99,51,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49,0,57, +58,118,101,99,51,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,50,0,57,18,109,0,16,10,50,0,57,58, +118,101,99,51,0,17,49,0,48,0,0,0,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,18,109,0,20,0,0,1,0, +15,2,24,1,0,2,15,109,0,0,0,1,9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0,57,58,118,101,99,52,0,17, +49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49,0,57,58,118,101,99,52,0,17,49,0, +48,0,0,0,0,46,20,0,9,18,109,0,16,10,50,0,57,18,109,0,16,10,50,0,57,58,118,101,99,52,0,17,49,0,48,0, +0,0,0,46,20,0,9,18,109,0,16,10,51,0,57,18,109,0,16,10,51,0,57,58,118,101,99,52,0,17,49,0,48,0,0,0, +0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,18,109,0,20,0,0,1,0,5,0,95,95,112,111,115,116,68,101, +99,114,0,1,0,2,5,97,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,97,0,20,0,9,18,97,0,18,97,0,16, +10,49,0,47,20,0,0,1,0,6,0,95,95,112,111,115,116,68,101,99,114,0,1,0,2,6,118,0,0,0,1,9,18,95,95,114, +101,116,86,97,108,0,18,118,0,20,0,9,18,118,0,18,118,0,58,105,118,101,99,50,0,16,10,49,0,0,0,47,20, +0,0,1,0,7,0,95,95,112,111,115,116,68,101,99,114,0,1,0,2,7,118,0,0,0,1,9,18,95,95,114,101,116,86,97, +108,0,18,118,0,20,0,9,18,118,0,18,118,0,58,105,118,101,99,51,0,16,10,49,0,0,0,47,20,0,0,1,0,8,0,95, +95,112,111,115,116,68,101,99,114,0,1,0,2,8,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0, +20,0,9,18,118,0,18,118,0,58,105,118,101,99,52,0,16,10,49,0,0,0,47,20,0,0,1,0,9,0,95,95,112,111,115, +116,68,101,99,114,0,1,0,2,9,97,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,97,0,20,0,9,18,97,0, +18,97,0,17,49,0,48,0,0,47,20,0,0,1,0,10,0,95,95,112,111,115,116,68,101,99,114,0,1,0,2,10,118,0,0,0, +1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,9,18,118,0,18,118,0,58,118,101,99,50,0,17,49,0, +48,0,0,0,0,47,20,0,0,1,0,11,0,95,95,112,111,115,116,68,101,99,114,0,1,0,2,11,118,0,0,0,1,9,18,95, +95,114,101,116,86,97,108,0,18,118,0,20,0,9,18,118,0,18,118,0,58,118,101,99,51,0,17,49,0,48,0,0,0,0, +47,20,0,0,1,0,12,0,95,95,112,111,115,116,68,101,99,114,0,1,0,2,12,118,0,0,0,1,9,18,95,95,114,101, +116,86,97,108,0,18,118,0,20,0,9,18,118,0,18,118,0,58,118,101,99,52,0,17,49,0,48,0,0,0,0,47,20,0,0, +1,0,13,0,95,95,112,111,115,116,68,101,99,114,0,1,0,2,13,109,0,0,0,1,9,18,95,95,114,101,116,86,97, +108,0,18,109,0,20,0,9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0,57,58,118,101,99,50,0,17,49,0,48,0, +0,0,0,47,20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49,0,57,58,118,101,99,50,0,17,49,0,48,0,0,0, +0,47,20,0,0,1,0,14,0,95,95,112,111,115,116,68,101,99,114,0,1,0,2,14,109,0,0,0,1,9,18,95,95,114,101, +116,86,97,108,0,18,109,0,20,0,9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0,57,58,118,101,99,51,0,17, +49,0,48,0,0,0,0,47,20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49,0,57,58,118,101,99,51,0,17,49,0, +48,0,0,0,0,47,20,0,9,18,109,0,16,10,50,0,57,18,109,0,16,10,50,0,57,58,118,101,99,51,0,17,49,0,48,0, +0,0,0,47,20,0,0,1,0,15,0,95,95,112,111,115,116,68,101,99,114,0,1,0,2,15,109,0,0,0,1,9,18,95,95,114, +101,116,86,97,108,0,18,109,0,20,0,9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0,57,58,118,101,99,52,0, +17,49,0,48,0,0,0,0,47,20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49,0,57,58,118,101,99,52,0,17, +49,0,48,0,0,0,0,47,20,0,9,18,109,0,16,10,50,0,57,18,109,0,16,10,50,0,57,58,118,101,99,52,0,17,49,0, +48,0,0,0,0,47,20,0,9,18,109,0,16,10,51,0,57,18,109,0,16,10,51,0,57,58,118,101,99,52,0,17,49,0,48,0, +0,0,0,47,20,0,0,1,0,9,0,95,95,112,111,115,116,73,110,99,114,0,1,0,2,9,97,0,0,0,1,9,18,95,95,114, +101,116,86,97,108,0,18,97,0,20,0,9,18,97,0,18,97,0,16,10,49,0,46,20,0,0,1,0,10,0,95,95,112,111,115, +116,73,110,99,114,0,1,0,2,10,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,9,18,118, +0,18,118,0,58,118,101,99,50,0,17,49,0,48,0,0,0,0,46,20,0,0,1,0,11,0,95,95,112,111,115,116,73,110, +99,114,0,1,0,2,11,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,9,18,118,0,18,118,0, +58,118,101,99,51,0,17,49,0,48,0,0,0,0,46,20,0,0,1,0,12,0,95,95,112,111,115,116,73,110,99,114,0,1,0, +2,12,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,9,18,118,0,18,118,0,58,118,101, +99,52,0,17,49,0,48,0,0,0,0,46,20,0,0,1,0,5,0,95,95,112,111,115,116,73,110,99,114,0,1,0,2,5,97,0,0, +0,1,9,18,95,95,114,101,116,86,97,108,0,18,97,0,20,0,9,18,97,0,18,97,0,16,10,49,0,46,20,0,0,1,0,6,0, +95,95,112,111,115,116,73,110,99,114,0,1,0,2,6,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18, +118,0,20,0,9,18,118,0,18,118,0,58,105,118,101,99,50,0,16,10,49,0,0,0,46,20,0,0,1,0,7,0,95,95,112, +111,115,116,73,110,99,114,0,1,0,2,7,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,9, +18,118,0,18,118,0,58,105,118,101,99,51,0,16,10,49,0,0,0,46,20,0,0,1,0,8,0,95,95,112,111,115,116,73, +110,99,114,0,1,0,2,8,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,9,18,118,0,18, +118,0,58,105,118,101,99,51,0,16,10,49,0,0,0,46,20,0,0,1,0,13,0,95,95,112,111,115,116,73,110,99,114, +0,1,0,2,13,109,0,0,0,1,3,2,0,13,1,110,0,2,18,109,0,0,0,9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0, +57,58,118,101,99,50,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49,0,57, +58,118,101,99,50,0,17,49,0,48,0,0,0,0,46,20,0,8,18,110,0,0,0,1,0,14,0,95,95,112,111,115,116,73,110, +99,114,0,1,0,2,14,109,0,0,0,1,3,2,0,14,1,110,0,2,18,109,0,0,0,9,18,109,0,16,8,48,0,57,18,109,0,16, +8,48,0,57,58,118,101,99,51,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49, +0,57,58,118,101,99,51,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,50,0,57,18,109,0,16,10,50,0,57, +58,118,101,99,51,0,17,49,0,48,0,0,0,0,46,20,0,8,18,110,0,0,0,1,0,15,0,95,95,112,111,115,116,73,110, +99,114,0,1,0,2,15,109,0,0,0,1,3,2,0,15,1,110,0,2,18,109,0,0,0,9,18,109,0,16,8,48,0,57,18,109,0,16, +8,48,0,57,58,118,101,99,52,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49, +0,57,58,118,101,99,52,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,50,0,57,18,109,0,16,10,50,0,57, +58,118,101,99,52,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,51,0,57,18,109,0,16,10,51,0,57,58, +118,101,99,52,0,17,49,0,48,0,0,0,0,46,20,0,8,18,110,0,0,0,1,0,1,2,15,1,1,0,9,97,0,0,1,1,0,9,98,0,0, +0,1,4,118,101,99,52,95,115,103,116,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,98,0,0,18,97,0, +0,0,0,1,0,1,2,15,1,1,0,5,97,0,0,1,1,0,5,98,0,0,0,1,8,58,102,108,111,97,116,0,18,97,0,0,0,58,102, +108,111,97,116,0,18,98,0,0,0,40,0,0,1,0,1,2,16,1,1,0,9,97,0,0,1,1,0,9,98,0,0,0,1,3,2,0,1,1,99,0,0, +0,4,102,108,111,97,116,95,108,101,115,115,0,18,99,0,0,18,98,0,0,18,97,0,0,0,8,18,99,0,0,0,1,0,1,2, +16,1,1,0,5,97,0,0,1,1,0,5,98,0,0,0,1,8,58,102,108,111,97,116,0,18,97,0,0,0,58,102,108,111,97,116,0, +18,98,0,0,0,41,0,0,1,0,1,2,18,1,1,0,9,97,0,0,1,1,0,9,98,0,0,0,1,3,2,0,1,1,103,0,0,1,1,101,0,0,0,4, +102,108,111,97,116,95,108,101,115,115,0,18,103,0,0,18,98,0,0,18,97,0,0,0,4,102,108,111,97,116,95, +101,113,117,97,108,0,18,101,0,0,18,97,0,0,18,98,0,0,0,8,18,103,0,18,101,0,32,0,0,1,0,1,2,18,1,1,0, +5,97,0,0,1,1,0,5,98,0,0,0,1,8,58,102,108,111,97,116,0,18,97,0,0,0,58,102,108,111,97,116,0,18,98,0, +0,0,43,0,0,1,0,1,2,17,1,1,0,9,97,0,0,1,1,0,9,98,0,0,0,1,3,2,0,1,1,103,0,0,1,1,101,0,0,0,4,102,108, +111,97,116,95,108,101,115,115,0,18,103,0,0,18,97,0,0,18,98,0,0,0,4,102,108,111,97,116,95,101,113, +117,97,108,0,18,101,0,0,18,97,0,0,18,98,0,0,0,8,18,103,0,18,101,0,32,0,0,1,0,1,2,17,1,1,0,5,97,0,0, +1,1,0,5,98,0,0,0,1,8,58,102,108,111,97,116,0,18,97,0,0,0,58,102,108,111,97,116,0,18,98,0,0,0,42,0, +0,1,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,9,102,0,0,0,1,4,102,108,111,97,116,95,112,114, +105,110,116,0,18,102,0,0,0,0,1,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,5,105,0,0,0,1,4,105, +110,116,95,112,114,105,110,116,0,18,105,0,0,0,0,1,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,1, +98,0,0,0,1,4,98,111,111,108,95,112,114,105,110,116,0,18,98,0,0,0,0,1,0,0,0,112,114,105,110,116,77, +69,83,65,0,1,1,0,10,118,0,0,0,1,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,120,0,0,0,0,9, +58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,121,0,0,0,0,0,1,0,0,0,112,114,105,110,116,77,69, +83,65,0,1,1,0,11,118,0,0,0,1,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,120,0,0,0,0,9,58, +112,114,105,110,116,77,69,83,65,0,18,118,0,59,121,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0, +18,118,0,59,122,0,0,0,0,0,1,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,12,118,0,0,0,1,9,58,112, +114,105,110,116,77,69,83,65,0,18,118,0,59,120,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18, +118,0,59,121,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,122,0,0,0,0,9,58,112,114, +105,110,116,77,69,83,65,0,18,118,0,59,119,0,0,0,0,0,1,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1, +0,6,118,0,0,0,1,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,120,0,0,0,0,9,58,112,114,105, +110,116,77,69,83,65,0,18,118,0,59,121,0,0,0,0,0,1,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,7, +118,0,0,0,1,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,120,0,0,0,0,9,58,112,114,105,110, +116,77,69,83,65,0,18,118,0,59,121,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,122,0, +0,0,0,0,1,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,8,118,0,0,0,1,9,58,112,114,105,110,116,77, +69,83,65,0,18,118,0,59,120,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,121,0,0,0,0, +9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,122,0,0,0,0,9,58,112,114,105,110,116,77,69,83, +65,0,18,118,0,59,119,0,0,0,0,0,1,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,2,118,0,0,0,1,9,58, +112,114,105,110,116,77,69,83,65,0,18,118,0,59,120,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0, +18,118,0,59,121,0,0,0,0,0,1,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,3,118,0,0,0,1,9,58,112, +114,105,110,116,77,69,83,65,0,18,118,0,59,120,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18, +118,0,59,121,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,122,0,0,0,0,0,1,0,0,0,112, +114,105,110,116,77,69,83,65,0,1,1,0,4,118,0,0,0,1,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0, +59,120,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,121,0,0,0,0,9,58,112,114,105,110, +116,77,69,83,65,0,18,118,0,59,122,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,119,0, +0,0,0,0,1,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,13,109,0,0,0,1,9,58,112,114,105,110,116,77, +69,83,65,0,18,109,0,16,8,48,0,57,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,109,0,16,10,49,0, +57,0,0,0,0,1,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,14,109,0,0,0,1,9,58,112,114,105,110,116, +77,69,83,65,0,18,109,0,16,8,48,0,57,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,109,0,16,10,49, +0,57,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,109,0,16,10,50,0,57,0,0,0,0,1,0,0,0,112,114, +105,110,116,77,69,83,65,0,1,1,0,15,109,0,0,0,1,9,58,112,114,105,110,116,77,69,83,65,0,18,109,0,16, +8,48,0,57,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,109,0,16,10,49,0,57,0,0,0,9,58,112,114, +105,110,116,77,69,83,65,0,18,109,0,16,10,50,0,57,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18, +109,0,16,10,51,0,57,0,0,0,0,1,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,16,101,0,0,0,1,4,105, +110,116,95,112,114,105,110,116,0,18,101,0,0,0,0,1,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,17, +101,0,0,0,1,4,105,110,116,95,112,114,105,110,116,0,18,101,0,0,0,0,1,0,0,0,112,114,105,110,116,77, +69,83,65,0,1,1,0,18,101,0,0,0,1,4,105,110,116,95,112,114,105,110,116,0,18,101,0,0,0,0,1,0,0,0,112, +114,105,110,116,77,69,83,65,0,1,1,0,19,101,0,0,0,1,4,105,110,116,95,112,114,105,110,116,0,18,101,0, +0,0,0,1,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,20,101,0,0,0,1,4,105,110,116,95,112,114,105, +110,116,0,18,101,0,0,0,0,1,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,21,101,0,0,0,1,4,105,110, +116,95,112,114,105,110,116,0,18,101,0,0,0,0,0 diff --git a/src/mesa/shader/slang/library/slang_version_syn.h b/src/mesa/shader/slang/library/slang_version_syn.h deleted file mode 100755 index 3b94d85927..0000000000 --- a/src/mesa/shader/slang/library/slang_version_syn.h +++ /dev/null @@ -1,64 +0,0 @@ -".syntax version_directive;\n" -"version_directive\n" -" version_directive_1 .and .loop version_directive_2;\n" -"version_directive_1\n" -" prior_optional_spaces .and optional_version_directive .and .true .emit $;\n" -"version_directive_2\n" -" prior_optional_spaces .and version_directive_body .and .true .emit $;\n" -"optional_version_directive\n" -" version_directive_body .or .true .emit 10 .emit 1;\n" -"version_directive_body\n" -" '#' .and optional_space .and \"version\" .and space .and version_number .and optional_space .and\n" -" new_line;\n" -"version_number\n" -" version_number_110;\n" -"version_number_110\n" -" leading_zeroes .and \"110\" .emit 10 .emit 1;\n" -"leading_zeroes\n" -" .loop zero;\n" -"zero\n" -" '0';\n" -"space\n" -" single_space .and .loop single_space;\n" -"optional_space\n" -" .loop single_space;\n" -"single_space\n" -" ' ' .or '\\t';\n" -"prior_optional_spaces\n" -" .loop prior_space;\n" -"prior_space\n" -" c_style_comment_block .or cpp_style_comment_block .or space .or new_line;\n" -"c_style_comment_block\n" -" '/' .and '*' .and c_style_comment_rest;\n" -"c_style_comment_rest\n" -" .loop c_style_comment_char_no_star .and c_style_comment_rest_1;\n" -"c_style_comment_rest_1\n" -" c_style_comment_end .or c_style_comment_rest_2;\n" -"c_style_comment_rest_2\n" -" '*' .and c_style_comment_rest;\n" -"c_style_comment_char_no_star\n" -" '\\x2B'-'\\xFF' .or '\\x01'-'\\x29';\n" -"c_style_comment_end\n" -" '*' .and '/';\n" -"cpp_style_comment_block\n" -" '/' .and '/' .and cpp_style_comment_block_1;\n" -"cpp_style_comment_block_1\n" -" cpp_style_comment_block_2 .or cpp_style_comment_block_3;\n" -"cpp_style_comment_block_2\n" -" .loop cpp_style_comment_char .and new_line;\n" -"cpp_style_comment_block_3\n" -" .loop cpp_style_comment_char;\n" -"cpp_style_comment_char\n" -" '\\x0E'-'\\xFF' .or '\\x01'-'\\x09' .or '\\x0B'-'\\x0C';\n" -"new_line\n" -" cr_lf .or lf_cr .or '\\n' .or '\\r';\n" -"cr_lf\n" -" '\\r' .and '\\n';\n" -"lf_cr\n" -" '\\n' .and '\\r';\n" -".string __string_filter;\n" -"__string_filter\n" -" .loop __identifier_char;\n" -"__identifier_char\n" -" 'a'-'z' .or 'A'-'Z' .or '_' .or '0'-'9';\n" -"" diff --git a/src/mesa/shader/slang/slang_builtin.c b/src/mesa/shader/slang/slang_builtin.c index 1081d8ff8d..f0f59aa87a 100644 --- a/src/mesa/shader/slang/slang_builtin.c +++ b/src/mesa/shader/slang/slang_builtin.c @@ -250,7 +250,7 @@ lookup_statevar(const char *var, GLint index1, GLint index2, const char *field, } } else if (strcmp(var, "gl_FrontLightModelProduct") == 0) { - if (strcmp(field, "ambient") == 0) { + if (strcmp(field, "sceneColor") == 0) { tokens[0] = STATE_LIGHTMODEL_SCENECOLOR; tokens[1] = 0; } @@ -259,7 +259,7 @@ lookup_statevar(const char *var, GLint index1, GLint index2, const char *field, } } else if (strcmp(var, "gl_BackLightModelProduct") == 0) { - if (strcmp(field, "ambient") == 0) { + if (strcmp(field, "sceneColor") == 0) { tokens[0] = STATE_LIGHTMODEL_SCENECOLOR; tokens[1] = 1; } @@ -397,6 +397,8 @@ lookup_statevar(const char *var, GLint index1, GLint index2, const char *field, * var.field * var[i].field * var[i][j] + * + * \return -1 upon error, else position in paramList of the state var/data */ GLint _slang_alloc_statevar(slang_ir_node *n, @@ -414,9 +416,13 @@ _slang_alloc_statevar(slang_ir_node *n, if (n->Opcode == IR_ELEMENT) { /* XXX can only handle constant indexes for now */ - assert(n->Children[1]->Opcode == IR_FLOAT); - index1 = (GLint) n->Children[1]->Value[0]; - n = n->Children[0]; + if (n->Children[1]->Opcode == IR_FLOAT) { + index1 = (GLint) n->Children[1]->Value[0]; + n = n->Children[0]; + } + else { + return -1; + } } if (n->Opcode == IR_ELEMENT) { diff --git a/src/mesa/shader/slang/slang_codegen.c b/src/mesa/shader/slang/slang_codegen.c index 0de2b0a57e..d19d5a0abb 100644 --- a/src/mesa/shader/slang/slang_codegen.c +++ b/src/mesa/shader/slang/slang_codegen.c @@ -495,6 +495,9 @@ new_node0(slang_ir_opcode op) } +/** + * Create sequence of two nodes. + */ static slang_ir_node * new_seq(slang_ir_node *left, slang_ir_node *right) { @@ -535,10 +538,10 @@ new_not(slang_ir_node *n) /** - * Inlined subroutine. + * Non-inlined function call. */ static slang_ir_node * -new_inlined_function_call(slang_ir_node *code, slang_label *name) +new_function_call(slang_ir_node *code, slang_label *name) { slang_ir_node *n = new_node1(IR_CALL, code); assert(name); @@ -1226,7 +1229,7 @@ _slang_gen_function_call(slang_assemble_ctx *A, slang_function *fun, else { callOper = inlined; } - callOper->type = SLANG_OPER_INLINED_CALL; + callOper->type = SLANG_OPER_NON_INLINED_CALL; callOper->fun = fun; callOper->label = _slang_label_new_unique((char*) fun->header.a_name); } @@ -1844,6 +1847,7 @@ _slang_gen_var_decl(slang_assemble_ctx *A, slang_variable *var) n->Store->File = PROGRAM_TEMPORARY; n->Store->Size = _slang_sizeof_type_specifier(&n->Var->type.specifier); + A->program->NumTemporaries++; assert(n->Store->Size > 0); } return n; @@ -2060,7 +2064,8 @@ _slang_gen_declaration(slang_assemble_ctx *A, slang_operation *oper) } /* XXX make copy of this initializer? */ rhs = _slang_gen_operation(A, &oper->children[0]); - assert(rhs); + if (!rhs) + return NULL; /* must have found an error */ init = new_node2(IR_MOVE, var, rhs); /*assert(rhs->Opcode != IR_SEQ);*/ n = new_seq(varDecl, init); @@ -2347,7 +2352,8 @@ _slang_gen_field(slang_assemble_ctx * A, slang_operation *oper) return n; } else if ( ti.spec.type == SLANG_SPEC_FLOAT - || ti.spec.type == SLANG_SPEC_INT) { + || ti.spec.type == SLANG_SPEC_INT + || ti.spec.type == SLANG_SPEC_BOOL) { const GLuint rows = 1; slang_swizzle swz; slang_ir_node *n; @@ -2509,49 +2515,6 @@ _slang_gen_subscript(slang_assemble_ctx * A, slang_operation *oper) /** - * Look for expressions such as: gl_ModelviewMatrix * gl_Vertex - * and replace with this: gl_Vertex * gl_ModelviewMatrixTranpose - * Since matrices are stored in column-major order, the second form of - * multiplication is much more efficient (just 4 dot products). - */ -static void -_slang_check_matmul_optimization(slang_assemble_ctx *A, slang_operation *oper) -{ - static const struct { - const char *orig; - const char *tranpose; - } matrices[] = { - {"gl_ModelViewMatrix", "gl_ModelViewMatrixTranspose"}, - {"gl_ProjectionMatrix", "gl_ProjectionMatrixTranspose"}, - {"gl_ModelViewProjectionMatrix", "gl_ModelViewProjectionMatrixTranspose"}, - {"gl_TextureMatrix", "gl_TextureMatrixTranspose"}, - {"gl_NormalMatrix", "__NormalMatrixTranspose"}, - { NULL, NULL } - }; - - assert(oper->type == SLANG_OPER_MULTIPLY); - if (oper->children[0].type == SLANG_OPER_IDENTIFIER) { - GLuint i; - for (i = 0; matrices[i].orig; i++) { - if (oper->children[0].a_id - == slang_atom_pool_atom(A->atoms, matrices[i].orig)) { - /* - _mesa_printf("Replace %s with %s\n", - matrices[i].orig, matrices[i].tranpose); - */ - assert(oper->children[0].type == SLANG_OPER_IDENTIFIER); - oper->children[0].a_id - = slang_atom_pool_atom(A->atoms, matrices[i].tranpose); - /* finally, swap the operands */ - _slang_operation_swap(&oper->children[0], &oper->children[1]); - return; - } - } - } -} - - -/** * Generate IR tree for a slang_operation (AST node) */ static slang_ir_node * @@ -2589,7 +2552,7 @@ _slang_gen_operation(slang_assemble_ctx * A, slang_operation *oper) _slang_free_ir_tree(tree); return NULL; /* error must have occured */ } - tree = tree ? new_seq(tree, n) : n; + tree = new_seq(tree, n); } #if 00 @@ -2684,7 +2647,6 @@ _slang_gen_operation(slang_assemble_ctx * A, slang_operation *oper) { slang_ir_node *n; assert(oper->num_children == 2); - _slang_check_matmul_optimization(A, oper); n = _slang_gen_function_call_name(A, "*", oper, NULL); return n; } @@ -2817,17 +2779,17 @@ _slang_gen_operation(slang_assemble_ctx * A, slang_operation *oper) return n; } - case SLANG_OPER_INLINED_CALL: + case SLANG_OPER_NON_INLINED_CALL: case SLANG_OPER_SEQUENCE: { slang_ir_node *tree = NULL; GLuint i; for (i = 0; i < oper->num_children; i++) { slang_ir_node *n = _slang_gen_operation(A, &oper->children[i]); - tree = tree ? new_seq(tree, n) : n; + tree = new_seq(tree, n); } - if (oper->type == SLANG_OPER_INLINED_CALL) { - tree = new_inlined_function_call(tree, oper->label); + if (oper->type == SLANG_OPER_NON_INLINED_CALL) { + tree = new_function_call(tree, oper->label); } return tree; } @@ -2879,17 +2841,7 @@ _slang_codegen_global_variable(slang_assemble_ctx *A, slang_variable *var, * store->Index = sampler number (0..7, typically) * store->Size = texture type index (1D, 2D, 3D, cube, etc) */ -#if 0 - GLint samplerUniform - = _mesa_add_sampler(prog->Parameters, varName, datatype); -#elif 0 - GLint samplerUniform - = _mesa_add_sampler(prog->Samplers, varName, datatype); - (void) _mesa_add_sampler(prog->Parameters, varName, datatype); /* dummy entry */ -#else - const GLint sampNum = A->numSamplers++; - _mesa_add_sampler(prog->Parameters, varName, datatype, sampNum); -#endif + GLint sampNum = _mesa_add_sampler(prog->Parameters, varName, datatype); store = _slang_new_ir_storage(PROGRAM_SAMPLER, sampNum, texIndex); if (dbg) printf("SAMPLER "); } diff --git a/src/mesa/shader/slang/slang_compile.c b/src/mesa/shader/slang/slang_compile.c index bfb9ca4db6..46b5c54bbe 100644 --- a/src/mesa/shader/slang/slang_compile.c +++ b/src/mesa/shader/slang/slang_compile.c @@ -31,6 +31,8 @@ #include "main/imports.h" #include "main/context.h" #include "shader/program.h" +#include "shader/programopt.h" +#include "shader/prog_print.h" #include "shader/prog_parameter.h" #include "shader/grammar/grammar_mesa.h" #include "slang_codegen.h" @@ -1618,7 +1620,6 @@ parse_init_declarator(slang_parse_ctx * C, slang_output_ctx * O, A.program = O->program; A.vartable = O->vartable; A.curFuncEndLabel = NULL; - A.numSamplers = 0; if (!_slang_codegen_global_variable(&A, var, C->type)) return 0; } @@ -1641,7 +1642,6 @@ parse_init_declarator(slang_parse_ctx * C, slang_output_ctx * O, A.space.funcs = O->funs; A.space.structs = O->structs; A.space.vars = O->vars; - A.numSamplers = 0; if (!initialize_global(&A, var)) return 0; } @@ -1775,7 +1775,6 @@ parse_function(slang_parse_ctx * C, slang_output_ctx * O, int definition, A.program = O->program; A.vartable = O->vartable; A.log = C->L; - A.numSamplers = 0; _slang_codegen_function(&A, *parsed_func_ret); } @@ -2155,9 +2154,6 @@ _slang_compile(GLcontext *ctx, struct gl_shader *shader) shader->Programs[0]->Parameters = _mesa_new_parameter_list(); shader->Programs[0]->Varying = _mesa_new_parameter_list(); shader->Programs[0]->Attributes = _mesa_new_parameter_list(); -#if 0 - shader->Programs[0]->Samplers = _mesa_new_parameter_list(); -#endif } slang_info_log_construct(&info_log); @@ -2186,6 +2182,19 @@ _slang_compile(GLcontext *ctx, struct gl_shader *shader) _slang_delete_mempool((slang_mempool *) ctx->Shader.MemPool); ctx->Shader.MemPool = NULL; + if (shader->Type == GL_VERTEX_SHADER) { + /* remove any reads of varying (output) registers */ +#if 0 + printf("Pre-remove output reads:\n"); + _mesa_print_program(shader->Programs[0]); +#endif + _mesa_remove_varying_reads(shader->Programs[0]); +#if 0 + printf("Post-remove output reads:\n"); + _mesa_print_program(shader->Programs[0]); +#endif + } + return success; } diff --git a/src/mesa/shader/slang/slang_compile_operation.h b/src/mesa/shader/slang/slang_compile_operation.h index d497b6f66f..d5cbe779a6 100644 --- a/src/mesa/shader/slang/slang_compile_operation.h +++ b/src/mesa/shader/slang/slang_compile_operation.h @@ -93,7 +93,7 @@ typedef enum slang_operation_type_ SLANG_OPER_NOT, /* "!" [expr] */ SLANG_OPER_SUBSCRIPT, /* [expr] "[" [expr] "]" */ SLANG_OPER_CALL, /* [func name] [param] [param] [...] */ - SLANG_OPER_INLINED_CALL, /* inlined function call */ + SLANG_OPER_NON_INLINED_CALL, /* a real function call */ SLANG_OPER_FIELD, /* i.e.: ".next" or ".xzy" or ".xxx" etc */ SLANG_OPER_POSTINCREMENT, /* [var] "++" */ SLANG_OPER_POSTDECREMENT /* [var] "--" */ diff --git a/src/mesa/shader/slang/slang_emit.c b/src/mesa/shader/slang/slang_emit.c index 382b13ff05..ff63e05dd2 100644 --- a/src/mesa/shader/slang/slang_emit.c +++ b/src/mesa/shader/slang/slang_emit.c @@ -1,6 +1,6 @@ /* * Mesa 3-D graphics library - * Version: 6.5.3 + * Version: 7.0.3 * * Copyright (C) 2005-2007 Brian Paul All Rights Reserved. * @@ -79,7 +79,7 @@ new_subroutine(slang_emit_info *emitInfo, GLuint *id) _mesa_realloc(emitInfo->Subroutines, n * sizeof(struct gl_program), (n + 1) * sizeof(struct gl_program)); - emitInfo->Subroutines[n] = _mesa_new_program(ctx, emitInfo->prog->Target, 0); + emitInfo->Subroutines[n] = ctx->Driver.NewProgram(ctx, emitInfo->prog->Target, 0); emitInfo->Subroutines[n]->Parameters = emitInfo->prog->Parameters; emitInfo->NumSubroutines++; *id = n; @@ -869,12 +869,18 @@ emit_return(slang_emit_info *emitInfo, slang_ir_node *n) static struct prog_instruction * emit_kill(slang_emit_info *emitInfo) { + struct gl_fragment_program *fp; struct prog_instruction *inst; /* NV-KILL - discard fragment depending on condition code. * Note that ARB-KILL depends on sign of vector operand. */ inst = new_instruction(emitInfo, OPCODE_KIL_NV); inst->DstReg.CondMask = COND_TR; /* always branch */ + + assert(emitInfo->prog->Target == GL_FRAGMENT_PROGRAM_ARB); + fp = (struct gl_fragment_program *) emitInfo->prog; + fp->UsesKill = GL_TRUE; + return inst; } @@ -1500,6 +1506,10 @@ emit_struct_field(slang_emit_info *emitInfo, slang_ir_node *n) { if (n->Store->File == PROGRAM_STATE_VAR) { n->Store->Index = _slang_alloc_statevar(n, emitInfo->prog->Parameters); + if (n->Store->Index < 0) { + slang_info_log_error(emitInfo->log, "Error parsing state variable"); + return NULL; + } } else { GLint offset = n->FieldOffset / 4; @@ -1787,7 +1797,7 @@ _slang_resolve_subroutines(slang_emit_info *emitInfo) sub->NumInstructions); /* delete subroutine code */ sub->Parameters = NULL; /* prevent double-free */ - _mesa_delete_program(ctx, sub); + _mesa_reference_program(ctx, &emitInfo->Subroutines[i], NULL); } /* free subroutine list */ diff --git a/src/mesa/shader/slang/slang_link.c b/src/mesa/shader/slang/slang_link.c index addff20421..80cd4b6df6 100644 --- a/src/mesa/shader/slang/slang_link.c +++ b/src/mesa/shader/slang/slang_link.c @@ -369,12 +369,9 @@ link_error(struct gl_shader_program *shProg, const char *msg) * 2. Varying vars in the two shaders are combined so their locations * agree between the vertex and fragment stages. They're treated as * vertex program output attribs and as fragment program input attribs. - * 3. Uniform vars (including state references, constants, etc) from the - * vertex and fragment shaders are merged into one group. Recall that - * GLSL uniforms are shared by all linked shaders. - * 4. The vertex and fragment programs are cloned and modified to update - * src/dst register references so they use the new, linked uniform/ - * varying storage locations. + * 3. The vertex and fragment programs are cloned and modified to update + * src/dst register references so they use the new, linked varying + * storage locations. */ void _slang_link(GLcontext *ctx, @@ -409,20 +406,20 @@ _slang_link(GLcontext *ctx, * Make copies of the vertex/fragment programs now since we'll be * changing src/dst registers after merging the uniforms and varying vars. */ + _mesa_reference_vertprog(ctx, &shProg->VertexProgram, NULL); if (vertProg) { - shProg->VertexProgram - = vertex_program(_mesa_clone_program(ctx, &vertProg->Base)); - } - else { - shProg->VertexProgram = NULL; + struct gl_vertex_program *linked_vprog = + vertex_program(_mesa_clone_program(ctx, &vertProg->Base)); + shProg->VertexProgram = linked_vprog; /* refcount OK */ + ASSERT(shProg->VertexProgram->Base.RefCount == 1); } + _mesa_reference_fragprog(ctx, &shProg->FragmentProgram, NULL); if (fragProg) { - shProg->FragmentProgram - = fragment_program(_mesa_clone_program(ctx, &fragProg->Base)); - } - else { - shProg->FragmentProgram = NULL; + struct gl_fragment_program *linked_fprog = + fragment_program(_mesa_clone_program(ctx, &fragProg->Base)); + shProg->FragmentProgram = linked_fprog; /* refcount OK */ + ASSERT(shProg->FragmentProgram->Base.RefCount == 1); } /* link varying vars */ @@ -440,18 +437,6 @@ _slang_link(GLcontext *ctx, /*_mesa_print_uniforms(shProg->Uniforms);*/ if (shProg->VertexProgram) { - /* Rather than cloning the parameter list here, just share it. - * We need to be careful _mesa_clear_shader_program_data() in - * to avoid double-freeing. - */ - shProg->VertexProgram->Base.Parameters = vertProg->Base.Parameters; - } - if (shProg->FragmentProgram) { - /* see comment just above */ - shProg->FragmentProgram->Base.Parameters = fragProg->Base.Parameters; - } - - if (shProg->VertexProgram) { if (!_slang_resolve_attributes(shProg, &shProg->VertexProgram->Base)) { /*goto cleanup;*/ _mesa_problem(ctx, "_slang_resolve_attributes() failed"); diff --git a/src/mesa/shader/slang/slang_preprocess.c b/src/mesa/shader/slang/slang_preprocess.c index 076e982f8f..1645afcc18 100644 --- a/src/mesa/shader/slang/slang_preprocess.c +++ b/src/mesa/shader/slang/slang_preprocess.c @@ -483,6 +483,7 @@ pp_cond_stack_reevaluate (pp_cond_stack *self) typedef struct { GLboolean MESA_shader_debug; /* GL_MESA_shader_debug enable */ + GLboolean ARB_texture_rectangle; /* GL_ARB_texture_rectangle enable */ } pp_ext; /* @@ -498,6 +499,7 @@ static GLvoid pp_ext_init (pp_ext *self) { pp_ext_disable_all (self); + self->ARB_texture_rectangle = GL_TRUE; /* Other initialization code goes here. */ } @@ -506,6 +508,8 @@ pp_ext_set (pp_ext *self, const char *name, GLboolean enable) { if (_mesa_strcmp (name, "MESA_shader_debug") == 0) self->MESA_shader_debug = enable; + else if (_mesa_strcmp (name, "GL_ARB_texture_rectangle") == 0) + self->ARB_texture_rectangle = enable; /* Next extension name tests go here. */ else return GL_FALSE; diff --git a/src/mesa/shader/slang/slang_typeinfo.h b/src/mesa/shader/slang/slang_typeinfo.h index ad5aa3e195..587331e8b1 100644 --- a/src/mesa/shader/slang/slang_typeinfo.h +++ b/src/mesa/shader/slang/slang_typeinfo.h @@ -65,7 +65,6 @@ typedef struct slang_assemble_ctx_ struct slang_label_ *curFuncEndLabel; struct slang_ir_node_ *CurLoop; struct slang_function_ *CurFunction; - GLuint numSamplers; } slang_assemble_ctx; diff --git a/src/mesa/sources b/src/mesa/sources index a4e2026e4b..5c7ff50cc6 100644 --- a/src/mesa/sources +++ b/src/mesa/sources @@ -245,7 +245,6 @@ ASM_C_SOURCES = \ x86/x86.c \ x86/3dnow.c \ x86/sse.c \ - x86/rtasm/x86sse.c \ sparc/sparc.c \ ppc/common_ppc.c \ x86-64/x86-64.c diff --git a/src/mesa/sparc/glapi_sparc.S b/src/mesa/sparc/glapi_sparc.S index b214bef54d..3d1d55fd76 100644 --- a/src/mesa/sparc/glapi_sparc.S +++ b/src/mesa/sparc/glapi_sparc.S @@ -738,10 +738,6 @@ __glapi_sparc_icache_flush: /* %o0 = insn_addr */ .globl glGetVertexAttribivNV ; .type glGetVertexAttribivNV,#function .globl glIsProgramNV ; .type glIsProgramNV,#function .globl glLoadProgramNV ; .type glLoadProgramNV,#function - .globl glProgramParameter4dNV ; .type glProgramParameter4dNV,#function - .globl glProgramParameter4dvNV ; .type glProgramParameter4dvNV,#function - .globl glProgramParameter4fNV ; .type glProgramParameter4fNV,#function - .globl glProgramParameter4fvNV ; .type glProgramParameter4fvNV,#function .globl glProgramParameters4dvNV ; .type glProgramParameters4dvNV,#function .globl glProgramParameters4fvNV ; .type glProgramParameters4fvNV,#function .globl glRequestResidentProgramsNV ; .type glRequestResidentProgramsNV,#function @@ -802,19 +798,19 @@ __glapi_sparc_icache_flush: /* %o0 = insn_addr */ .globl glSetFragmentShaderConstantATI ; .type glSetFragmentShaderConstantATI,#function .globl glPointParameteriNV ; .type glPointParameteriNV,#function .globl glPointParameterivNV ; .type glPointParameterivNV,#function + .globl gl_dispatch_stub_733 ; .type gl_dispatch_stub_733,#function + .globl gl_dispatch_stub_734 ; .type gl_dispatch_stub_734,#function + .globl gl_dispatch_stub_735 ; .type gl_dispatch_stub_735,#function + .globl gl_dispatch_stub_736 ; .type gl_dispatch_stub_736,#function .globl gl_dispatch_stub_737 ; .type gl_dispatch_stub_737,#function - .globl gl_dispatch_stub_738 ; .type gl_dispatch_stub_738,#function - .globl gl_dispatch_stub_739 ; .type gl_dispatch_stub_739,#function - .globl gl_dispatch_stub_740 ; .type gl_dispatch_stub_740,#function - .globl gl_dispatch_stub_741 ; .type gl_dispatch_stub_741,#function .globl glGetProgramNamedParameterdvNV ; .type glGetProgramNamedParameterdvNV,#function .globl glGetProgramNamedParameterfvNV ; .type glGetProgramNamedParameterfvNV,#function .globl glProgramNamedParameter4dNV ; .type glProgramNamedParameter4dNV,#function .globl glProgramNamedParameter4dvNV ; .type glProgramNamedParameter4dvNV,#function .globl glProgramNamedParameter4fNV ; .type glProgramNamedParameter4fNV,#function .globl glProgramNamedParameter4fvNV ; .type glProgramNamedParameter4fvNV,#function - .globl gl_dispatch_stub_748 ; .type gl_dispatch_stub_748,#function - .globl gl_dispatch_stub_749 ; .type gl_dispatch_stub_749,#function + .globl gl_dispatch_stub_744 ; .type gl_dispatch_stub_744,#function + .globl gl_dispatch_stub_745 ; .type gl_dispatch_stub_745,#function .globl glBindFramebufferEXT ; .type glBindFramebufferEXT,#function .globl glBindRenderbufferEXT ; .type glBindRenderbufferEXT,#function .globl glCheckFramebufferStatusEXT ; .type glCheckFramebufferStatusEXT,#function @@ -832,13 +828,13 @@ __glapi_sparc_icache_flush: /* %o0 = insn_addr */ .globl glIsFramebufferEXT ; .type glIsFramebufferEXT,#function .globl glIsRenderbufferEXT ; .type glIsRenderbufferEXT,#function .globl glRenderbufferStorageEXT ; .type glRenderbufferStorageEXT,#function - .globl gl_dispatch_stub_767 ; .type gl_dispatch_stub_767,#function + .globl gl_dispatch_stub_763 ; .type gl_dispatch_stub_763,#function .globl glFramebufferTextureLayerEXT ; .type glFramebufferTextureLayerEXT,#function + .globl gl_dispatch_stub_765 ; .type gl_dispatch_stub_765,#function + .globl gl_dispatch_stub_766 ; .type gl_dispatch_stub_766,#function + .globl gl_dispatch_stub_767 ; .type gl_dispatch_stub_767,#function + .globl gl_dispatch_stub_768 ; .type gl_dispatch_stub_768,#function .globl gl_dispatch_stub_769 ; .type gl_dispatch_stub_769,#function - .globl gl_dispatch_stub_770 ; .type gl_dispatch_stub_770,#function - .globl gl_dispatch_stub_771 ; .type gl_dispatch_stub_771,#function - .globl gl_dispatch_stub_772 ; .type gl_dispatch_stub_772,#function - .globl gl_dispatch_stub_773 ; .type gl_dispatch_stub_773,#function .globl _mesa_sparc_glapi_begin ; .type _mesa_sparc_glapi_begin,#function _mesa_sparc_glapi_begin: @@ -1515,10 +1511,6 @@ _mesa_sparc_glapi_begin: GL_STUB(glGetVertexAttribivNV, _gloffset_GetVertexAttribivNV) GL_STUB(glIsProgramNV, _gloffset_IsProgramNV) GL_STUB(glLoadProgramNV, _gloffset_LoadProgramNV) - GL_STUB(glProgramParameter4dNV, _gloffset_ProgramParameter4dNV) - GL_STUB(glProgramParameter4dvNV, _gloffset_ProgramParameter4dvNV) - GL_STUB(glProgramParameter4fNV, _gloffset_ProgramParameter4fNV) - GL_STUB(glProgramParameter4fvNV, _gloffset_ProgramParameter4fvNV) GL_STUB(glProgramParameters4dvNV, _gloffset_ProgramParameters4dvNV) GL_STUB(glProgramParameters4fvNV, _gloffset_ProgramParameters4fvNV) GL_STUB(glRequestResidentProgramsNV, _gloffset_RequestResidentProgramsNV) @@ -1579,19 +1571,19 @@ _mesa_sparc_glapi_begin: GL_STUB(glSetFragmentShaderConstantATI, _gloffset_SetFragmentShaderConstantATI) GL_STUB(glPointParameteriNV, _gloffset_PointParameteriNV) GL_STUB(glPointParameterivNV, _gloffset_PointParameterivNV) + GL_STUB(gl_dispatch_stub_733, _gloffset__dispatch_stub_733) + GL_STUB(gl_dispatch_stub_734, _gloffset__dispatch_stub_734) + GL_STUB(gl_dispatch_stub_735, _gloffset__dispatch_stub_735) + GL_STUB(gl_dispatch_stub_736, _gloffset__dispatch_stub_736) GL_STUB(gl_dispatch_stub_737, _gloffset__dispatch_stub_737) - GL_STUB(gl_dispatch_stub_738, _gloffset__dispatch_stub_738) - GL_STUB(gl_dispatch_stub_739, _gloffset__dispatch_stub_739) - GL_STUB(gl_dispatch_stub_740, _gloffset__dispatch_stub_740) - GL_STUB(gl_dispatch_stub_741, _gloffset__dispatch_stub_741) GL_STUB(glGetProgramNamedParameterdvNV, _gloffset_GetProgramNamedParameterdvNV) GL_STUB(glGetProgramNamedParameterfvNV, _gloffset_GetProgramNamedParameterfvNV) GL_STUB(glProgramNamedParameter4dNV, _gloffset_ProgramNamedParameter4dNV) GL_STUB(glProgramNamedParameter4dvNV, _gloffset_ProgramNamedParameter4dvNV) GL_STUB(glProgramNamedParameter4fNV, _gloffset_ProgramNamedParameter4fNV) GL_STUB(glProgramNamedParameter4fvNV, _gloffset_ProgramNamedParameter4fvNV) - GL_STUB(gl_dispatch_stub_748, _gloffset__dispatch_stub_748) - GL_STUB(gl_dispatch_stub_749, _gloffset__dispatch_stub_749) + GL_STUB(gl_dispatch_stub_744, _gloffset__dispatch_stub_744) + GL_STUB(gl_dispatch_stub_745, _gloffset__dispatch_stub_745) GL_STUB(glBindFramebufferEXT, _gloffset_BindFramebufferEXT) GL_STUB(glBindRenderbufferEXT, _gloffset_BindRenderbufferEXT) GL_STUB(glCheckFramebufferStatusEXT, _gloffset_CheckFramebufferStatusEXT) @@ -1609,13 +1601,13 @@ _mesa_sparc_glapi_begin: GL_STUB(glIsFramebufferEXT, _gloffset_IsFramebufferEXT) GL_STUB(glIsRenderbufferEXT, _gloffset_IsRenderbufferEXT) GL_STUB(glRenderbufferStorageEXT, _gloffset_RenderbufferStorageEXT) - GL_STUB(gl_dispatch_stub_767, _gloffset__dispatch_stub_767) + GL_STUB(gl_dispatch_stub_763, _gloffset__dispatch_stub_763) GL_STUB(glFramebufferTextureLayerEXT, _gloffset_FramebufferTextureLayerEXT) + GL_STUB(gl_dispatch_stub_765, _gloffset__dispatch_stub_765) + GL_STUB(gl_dispatch_stub_766, _gloffset__dispatch_stub_766) + GL_STUB(gl_dispatch_stub_767, _gloffset__dispatch_stub_767) + GL_STUB(gl_dispatch_stub_768, _gloffset__dispatch_stub_768) GL_STUB(gl_dispatch_stub_769, _gloffset__dispatch_stub_769) - GL_STUB(gl_dispatch_stub_770, _gloffset__dispatch_stub_770) - GL_STUB(gl_dispatch_stub_771, _gloffset__dispatch_stub_771) - GL_STUB(gl_dispatch_stub_772, _gloffset__dispatch_stub_772) - GL_STUB(gl_dispatch_stub_773, _gloffset__dispatch_stub_773) .globl _mesa_sparc_glapi_end ; .type _mesa_sparc_glapi_end,#function _mesa_sparc_glapi_end: @@ -1710,6 +1702,10 @@ _mesa_sparc_glapi_end: .globl glGetVertexAttribdv ; .type glGetVertexAttribdv,#function ; glGetVertexAttribdv = glGetVertexAttribdvARB .globl glGetVertexAttribfv ; .type glGetVertexAttribfv,#function ; glGetVertexAttribfv = glGetVertexAttribfvARB .globl glGetVertexAttribiv ; .type glGetVertexAttribiv,#function ; glGetVertexAttribiv = glGetVertexAttribivARB + .globl glProgramParameter4dNV ; .type glProgramParameter4dNV,#function ; glProgramParameter4dNV = glProgramEnvParameter4dARB + .globl glProgramParameter4dvNV ; .type glProgramParameter4dvNV,#function ; glProgramParameter4dvNV = glProgramEnvParameter4dvARB + .globl glProgramParameter4fNV ; .type glProgramParameter4fNV,#function ; glProgramParameter4fNV = glProgramEnvParameter4fARB + .globl glProgramParameter4fvNV ; .type glProgramParameter4fvNV,#function ; glProgramParameter4fvNV = glProgramEnvParameter4fvARB .globl glVertexAttrib1d ; .type glVertexAttrib1d,#function ; glVertexAttrib1d = glVertexAttrib1dARB .globl glVertexAttrib1dv ; .type glVertexAttrib1dv,#function ; glVertexAttrib1dv = glVertexAttrib1dvARB .globl glVertexAttrib1f ; .type glVertexAttrib1f,#function ; glVertexAttrib1f = glVertexAttrib1fARB diff --git a/src/mesa/state_tracker/st_atom.h b/src/mesa/state_tracker/st_atom.h index 3a63e2dec0..c6c6eba812 100644 --- a/src/mesa/state_tracker/st_atom.h +++ b/src/mesa/state_tracker/st_atom.h @@ -44,20 +44,20 @@ void st_destroy_atoms( struct st_context *st ); void st_validate_state( struct st_context *st ); -const struct st_tracked_state st_update_framebuffer; -const struct st_tracked_state st_update_clip; -const struct st_tracked_state st_update_depth_stencil_alpha; -const struct st_tracked_state st_update_shader; -const struct st_tracked_state st_update_rasterizer; -const struct st_tracked_state st_update_polygon_stipple; -const struct st_tracked_state st_update_viewport; -const struct st_tracked_state st_update_scissor; -const struct st_tracked_state st_update_blend; -const struct st_tracked_state st_update_sampler; -const struct st_tracked_state st_update_texture; -const struct st_tracked_state st_update_fs_constants; -const struct st_tracked_state st_update_vs_constants; -const struct st_tracked_state st_update_pixel_transfer; +extern const struct st_tracked_state st_update_framebuffer; +extern const struct st_tracked_state st_update_clip; +extern const struct st_tracked_state st_update_depth_stencil_alpha; +extern const struct st_tracked_state st_update_shader; +extern const struct st_tracked_state st_update_rasterizer; +extern const struct st_tracked_state st_update_polygon_stipple; +extern const struct st_tracked_state st_update_viewport; +extern const struct st_tracked_state st_update_scissor; +extern const struct st_tracked_state st_update_blend; +extern const struct st_tracked_state st_update_sampler; +extern const struct st_tracked_state st_update_texture; +extern const struct st_tracked_state st_update_fs_constants; +extern const struct st_tracked_state st_update_vs_constants; +extern const struct st_tracked_state st_update_pixel_transfer; uint st_compare_func_to_pipe(GLenum func); diff --git a/src/mesa/state_tracker/st_atom_blend.c b/src/mesa/state_tracker/st_atom_blend.c index 2a0e92245c..35c09c3e08 100644 --- a/src/mesa/state_tracker/st_atom_blend.c +++ b/src/mesa/state_tracker/st_atom_blend.c @@ -223,15 +223,10 @@ update_blend( struct st_context *st ) const struct st_tracked_state st_update_blend = { - .name = "st_update_blend", - .dirty = { - .mesa = (_NEW_COLOR), /* XXX _NEW_BLEND someday? */ - .st = 0, + "st_update_blend", /* name */ + { /* dirty */ + (_NEW_COLOR), /* XXX _NEW_BLEND someday? */ /* mesa */ + 0, /* st */ }, - .update = update_blend + update_blend, /* update */ }; - - - - - diff --git a/src/mesa/state_tracker/st_atom_clip.c b/src/mesa/state_tracker/st_atom_clip.c index a6f0568660..23d709b814 100644 --- a/src/mesa/state_tracker/st_atom_clip.c +++ b/src/mesa/state_tracker/st_atom_clip.c @@ -62,15 +62,10 @@ static void update_clip( struct st_context *st ) const struct st_tracked_state st_update_clip = { - .name = "st_update_clip", - .dirty = { - .mesa = (_NEW_TRANSFORM), - .st = 0, + "st_update_clip", /* name */ + { /* dirty */ + (_NEW_TRANSFORM), /* mesa */ + 0, /* st */ }, - .update = update_clip + update_clip /* update */ }; - - - - - diff --git a/src/mesa/state_tracker/st_atom_constbuf.c b/src/mesa/state_tracker/st_atom_constbuf.c index 03093579e1..2856e0f0e0 100644 --- a/src/mesa/state_tracker/st_atom_constbuf.c +++ b/src/mesa/state_tracker/st_atom_constbuf.c @@ -36,7 +36,6 @@ #include "pipe/p_context.h" #include "pipe/p_defines.h" -#include "pipe/p_winsys.h" #include "pipe/p_inlines.h" #include "st_context.h" @@ -54,7 +53,7 @@ void st_upload_constants( struct st_context *st, struct gl_program_parameter_list *params, unsigned id) { - struct pipe_winsys *ws = st->pipe->winsys; + struct pipe_context *pipe = st->pipe; struct pipe_constant_buffer *cbuf = &st->state.constants[id]; assert(id == PIPE_SHADER_VERTEX || id == PIPE_SHADER_FRAGMENT); @@ -74,8 +73,8 @@ void st_upload_constants( struct st_context *st, /* We always need to get a new buffer, to keep the drivers simple and * avoid gratuitous rendering synchronization. */ - pipe_buffer_reference( ws, &cbuf->buffer, NULL ); - cbuf->buffer = ws->buffer_create( ws, 1, PIPE_BUFFER_USAGE_CONSTANT, + pipe_reference_buffer(pipe, &cbuf->buffer, NULL ); + cbuf->buffer = pipe_buffer_create(pipe, 1, PIPE_BUFFER_USAGE_CONSTANT, paramBytes ); if (0) @@ -87,9 +86,10 @@ void st_upload_constants( struct st_context *st, /* load Mesa constants into the constant buffer */ if (cbuf->buffer) { - memcpy(ws->buffer_map(ws, cbuf->buffer, PIPE_BUFFER_USAGE_CPU_WRITE), - params->ParameterValues, paramBytes); - ws->buffer_unmap(ws, cbuf->buffer); + void *map = pipe_buffer_map(pipe, cbuf->buffer, + PIPE_BUFFER_USAGE_CPU_WRITE); + memcpy(map, params->ParameterValues, paramBytes); + pipe_buffer_unmap(pipe, cbuf->buffer); } cbuf->size = paramBytes; @@ -113,12 +113,12 @@ static void update_vs_constants(struct st_context *st ) } const struct st_tracked_state st_update_vs_constants = { - .name = "st_update_vs_constants", - .dirty = { - .mesa = 0, /* set dynamically above */ - .st = ST_NEW_VERTEX_PROGRAM, + "st_update_vs_constants", /* name */ + { /* dirty */ + 0, /* set dynamically above */ /* mesa */ + ST_NEW_VERTEX_PROGRAM, /* st */ }, - .update = update_vs_constants + update_vs_constants /* update */ }; /* Fragment shader: @@ -132,11 +132,11 @@ static void update_fs_constants(struct st_context *st ) } const struct st_tracked_state st_update_fs_constants = { - .name = "st_update_fs_constants", - .dirty = { - .mesa = 0, /* set dynamically above */ - .st = ST_NEW_FRAGMENT_PROGRAM, + "st_update_fs_constants", /* name */ + { /* dirty */ + 0, /* set dynamically above */ /* mesa */ + ST_NEW_FRAGMENT_PROGRAM, /* st */ }, - .update = update_fs_constants + update_fs_constants /* update */ }; diff --git a/src/mesa/state_tracker/st_atom_depth.c b/src/mesa/state_tracker/st_atom_depth.c index ef467582c0..0e791ceb20 100644 --- a/src/mesa/state_tracker/st_atom_depth.c +++ b/src/mesa/state_tracker/st_atom_depth.c @@ -142,10 +142,10 @@ update_depth_stencil_alpha(struct st_context *st) const struct st_tracked_state st_update_depth_stencil_alpha = { - .name = "st_update_depth_stencil", - .dirty = { - .mesa = (_NEW_DEPTH|_NEW_STENCIL|_NEW_COLOR), - .st = 0, + "st_update_depth_stencil", /* name */ + { /* dirty */ + (_NEW_DEPTH|_NEW_STENCIL|_NEW_COLOR), /* mesa */ + 0, /* st */ }, - .update = update_depth_stencil_alpha + update_depth_stencil_alpha /* update */ }; diff --git a/src/mesa/state_tracker/st_atom_fixedfunction.c b/src/mesa/state_tracker/st_atom_fixedfunction.c index 3f137e1633..165567af70 100644 --- a/src/mesa/state_tracker/st_atom_fixedfunction.c +++ b/src/mesa/state_tracker/st_atom_fixedfunction.c @@ -55,12 +55,12 @@ static void update_tnl( struct st_context *st ) const struct st_tracked_state st_update_tnl = { - .name = "st_update_tnl", - .dirty = { - .mesa = TNL_FIXED_FUNCTION_STATE_FLAGS, - .st = 0 + "st_update_tnl", /* name */ + { /* dirty */ + TNL_FIXED_FUNCTION_STATE_FLAGS, /* mesa */ + 0 /* st */ }, - .update = update_tnl + update_tnl /* update */ }; diff --git a/src/mesa/state_tracker/st_atom_framebuffer.c b/src/mesa/state_tracker/st_atom_framebuffer.c index 02573af8f0..0a6974d8a7 100644 --- a/src/mesa/state_tracker/st_atom_framebuffer.c +++ b/src/mesa/state_tracker/st_atom_framebuffer.c @@ -48,7 +48,7 @@ update_framebuffer_state( struct st_context *st ) struct pipe_framebuffer_state *framebuffer = &st->state.framebuffer; struct gl_framebuffer *fb = st->ctx->DrawBuffer; struct st_renderbuffer *strb; - GLuint i; + GLuint i, j; memset(framebuffer, 0, sizeof(*framebuffer)); @@ -58,11 +58,14 @@ update_framebuffer_state( struct st_context *st ) /* Examine Mesa's ctx->DrawBuffer->_ColorDrawBuffers state * to determine which surfaces to draw to */ - framebuffer->num_cbufs = fb->_NumColorDrawBuffers[0]; - for (i = 0; i < framebuffer->num_cbufs; i++) { - strb = st_renderbuffer(fb->_ColorDrawBuffers[0][i]); - assert(strb->surface); - framebuffer->cbufs[i] = strb->surface; + framebuffer->num_cbufs = 0; + for (j = 0; j < MAX_DRAW_BUFFERS; j++) { + for (i = 0; i < fb->_NumColorDrawBuffers[j]; i++) { + strb = st_renderbuffer(fb->_ColorDrawBuffers[j][i]); + assert(strb->surface); + framebuffer->cbufs[framebuffer->num_cbufs] = strb->surface; + framebuffer->num_cbufs++; + } } strb = st_renderbuffer(fb->Attachment[BUFFER_DEPTH].Renderbuffer); @@ -81,15 +84,23 @@ update_framebuffer_state( struct st_context *st ) } cso_set_framebuffer(st->cso_context, framebuffer); + + if (fb->_ColorDrawBufferMask[0] & BUFFER_BIT_FRONT_LEFT) { + if (st->frontbuffer_status == FRONT_STATUS_COPY_OF_BACK) { + /* XXX copy back buf to front? */ + } + /* we're assuming we'll really draw to the front buffer */ + st->frontbuffer_status = FRONT_STATUS_DIRTY; + } } const struct st_tracked_state st_update_framebuffer = { - .name = "st_update_framebuffer", - .dirty = { - .mesa = _NEW_BUFFERS, - .st = 0, + "st_update_framebuffer", /* name */ + { /* dirty */ + _NEW_BUFFERS, /* mesa */ + 0, /* st */ }, - .update = update_framebuffer_state + update_framebuffer_state /* update */ }; diff --git a/src/mesa/state_tracker/st_atom_pixeltransfer.c b/src/mesa/state_tracker/st_atom_pixeltransfer.c index 6410e7cb24..999c148449 100644 --- a/src/mesa/state_tracker/st_atom_pixeltransfer.c +++ b/src/mesa/state_tracker/st_atom_pixeltransfer.c @@ -42,14 +42,22 @@ #include "shader/prog_print.h" #include "st_context.h" +#include "st_format.h" #include "st_program.h" +#include "st_texture.h" +#include "pipe/p_screen.h" +#include "pipe/p_context.h" +#include "pipe/p_inlines.h" +#include "util/u_pack_color.h" struct state_key { GLuint scaleAndBias:1; GLuint colorMatrix:1; + GLuint colorMatrixPostScaleBias:1; + GLuint pixelMaps:1; #if 0 GLfloat Maps[3][256][4]; @@ -80,6 +88,9 @@ is_identity(const GLfloat m[16]) static void make_state_key(GLcontext *ctx, struct state_key *key) { + static const GLfloat zero[4] = { 0.0f, 0.0f, 0.0f, 0.0f }; + static const GLfloat one[4] = { 1.0f, 1.0f, 1.0f, 1.0f }; + memset(key, 0, sizeof(*key)); if (ctx->Pixel.RedBias != 0.0 || ctx->Pixel.RedScale != 1.0 || @@ -92,9 +103,76 @@ make_state_key(GLcontext *ctx, struct state_key *key) if (!is_identity(ctx->ColorMatrixStack.Top->m)) { key->colorMatrix = 1; } + + if (!TEST_EQ_4V(ctx->Pixel.PostColorMatrixScale, one) || + !TEST_EQ_4V(ctx->Pixel.PostColorMatrixBias, zero)) { + key->colorMatrixPostScaleBias = 1; + } + + key->pixelMaps = ctx->Pixel.MapColorFlag; +} + + +static struct pipe_texture * +create_color_map_texture(GLcontext *ctx) +{ + struct pipe_context *pipe = ctx->st->pipe; + struct pipe_texture *pt; + enum pipe_format format; + const uint texSize = 256; /* simple, and usually perfect */ + + /* find an RGBA texture format */ + format = st_choose_format(pipe, GL_RGBA, PIPE_TEXTURE); + + /* create texture for color map/table */ + pt = st_texture_create(ctx->st, PIPE_TEXTURE_2D, format, 0, + texSize, texSize, 1, 0); + return pt; } +/** + * Update the pixelmap texture with the contents of the R/G/B/A pixel maps. + */ +static void +load_color_map_texture(GLcontext *ctx, struct pipe_texture *pt) +{ + struct pipe_context *pipe = ctx->st->pipe; + struct pipe_screen *screen = pipe->screen; + struct pipe_surface *surface; + const GLuint rSize = ctx->PixelMaps.RtoR.Size; + const GLuint gSize = ctx->PixelMaps.GtoG.Size; + const GLuint bSize = ctx->PixelMaps.BtoB.Size; + const GLuint aSize = ctx->PixelMaps.AtoA.Size; + const uint texSize = pt->width[0]; + uint *dest; + uint i, j; + + surface = screen->get_tex_surface(screen, pt, 0, 0, 0); + dest = (uint *) pipe_surface_map(surface); + + /* Pack four 1D maps into a 2D texture: + * R map is placed horizontally, indexed by S, in channel 0 + * G map is placed vertically, indexed by T, in channel 1 + * B map is placed horizontally, indexed by S, in channel 2 + * A map is placed vertically, indexed by T, in channel 3 + */ + for (i = 0; i < texSize; i++) { + for (j = 0; j < texSize; j++) { + int k = (i * texSize + j); + ubyte r = ctx->PixelMaps.RtoR.Map8[j * rSize / texSize]; + ubyte g = ctx->PixelMaps.GtoG.Map8[i * gSize / texSize]; + ubyte b = ctx->PixelMaps.BtoB.Map8[j * bSize / texSize]; + ubyte a = ctx->PixelMaps.AtoA.Map8[i * aSize / texSize]; + util_pack_color_ub(r, g, b, a, pt->format, dest + k); + } + } + + pipe_surface_unmap(surface); + pipe_surface_reference(&surface, NULL); + pipe->texture_update(pipe, pt, 0, 0x1); +} + #define MAX_INST 100 @@ -105,6 +183,7 @@ make_state_key(GLcontext *ctx, struct state_key *key) static struct gl_fragment_program * get_pixel_transfer_program(GLcontext *ctx, const struct state_key *key) { + struct st_context *st = ctx->st; struct prog_instruction inst[MAX_INST]; struct gl_program_parameter_list *params; struct gl_fragment_program *fp; @@ -118,7 +197,10 @@ get_pixel_transfer_program(GLcontext *ctx, const struct state_key *key) params = _mesa_new_parameter_list(); - /* TEX colorTemp, fragment.texcoord[0], texture[0], 2D; */ + /* + * Get initial pixel color from the texture. + * TEX colorTemp, fragment.texcoord[0], texture[0], 2D; + */ _mesa_init_instructions(inst + ic, 1); inst[ic].Opcode = OPCODE_TEX; inst[ic].DstReg.File = PROGRAM_TEMPORARY; @@ -132,7 +214,6 @@ get_pixel_transfer_program(GLcontext *ctx, const struct state_key *key) fp->Base.OutputsWritten = (1 << FRAG_RESULT_COLR); fp->Base.SamplersUsed = 0x1; /* sampler 0 (bit 0) is used */ - /* MAD colorTemp, colorTemp, scale, bias; */ if (key->scaleAndBias) { static const gl_state_index scale_state[STATE_LENGTH] = { STATE_INTERNAL, STATE_PT_SCALE, 0, 0, 0 }; @@ -153,6 +234,7 @@ get_pixel_transfer_program(GLcontext *ctx, const struct state_key *key) scale_p = _mesa_add_state_reference(params, scale_state); bias_p = _mesa_add_state_reference(params, bias_state); + /* MAD colorTemp, colorTemp, scale, bias; */ _mesa_init_instructions(inst + ic, 1); inst[ic].Opcode = OPCODE_MAD; inst[ic].DstReg.File = PROGRAM_TEMPORARY; @@ -166,6 +248,56 @@ get_pixel_transfer_program(GLcontext *ctx, const struct state_key *key) ic++; } + if (key->pixelMaps) { + const GLuint temp = 1; + + /* create the colormap/texture now if not already done */ + if (!st->pixel_xfer.pixelmap_texture) { + st->pixel_xfer.pixelmap_texture = create_color_map_texture(ctx); + } + + /* with a little effort, we can do four pixel map look-ups with + * two TEX instructions: + */ + + /* TEX temp.rg, colorTemp.rgba, texture[1], 2D; */ + _mesa_init_instructions(inst + ic, 1); + inst[ic].Opcode = OPCODE_TEX; + inst[ic].DstReg.File = PROGRAM_TEMPORARY; + inst[ic].DstReg.Index = temp; + inst[ic].DstReg.WriteMask = WRITEMASK_XY; /* write R,G */ + inst[ic].SrcReg[0].File = PROGRAM_TEMPORARY; + inst[ic].SrcReg[0].Index = colorTemp; + inst[ic].TexSrcUnit = 1; + inst[ic].TexSrcTarget = TEXTURE_2D_INDEX; + ic++; + + /* TEX temp.ba, colorTemp.baba, texture[1], 2D; */ + _mesa_init_instructions(inst + ic, 1); + inst[ic].Opcode = OPCODE_TEX; + inst[ic].DstReg.File = PROGRAM_TEMPORARY; + inst[ic].DstReg.Index = temp; + inst[ic].DstReg.WriteMask = WRITEMASK_ZW; /* write B,A */ + inst[ic].SrcReg[0].File = PROGRAM_TEMPORARY; + inst[ic].SrcReg[0].Index = colorTemp; + inst[ic].SrcReg[0].Swizzle = MAKE_SWIZZLE4(SWIZZLE_Z, SWIZZLE_W, + SWIZZLE_Z, SWIZZLE_W); + inst[ic].TexSrcUnit = 1; + inst[ic].TexSrcTarget = TEXTURE_2D_INDEX; + ic++; + + /* MOV colorTemp, temp; */ + _mesa_init_instructions(inst + ic, 1); + inst[ic].Opcode = OPCODE_MOV; + inst[ic].DstReg.File = PROGRAM_TEMPORARY; + inst[ic].DstReg.Index = colorTemp; + inst[ic].SrcReg[0].File = PROGRAM_TEMPORARY; + inst[ic].SrcReg[0].Index = temp; + ic++; + + fp->Base.SamplersUsed |= (1 << 1); /* sampler 1 is used */ + } + if (key->colorMatrix) { static const gl_state_index row0_state[STATE_LENGTH] = { STATE_COLOR_MATRIX, 0, 0, 0, 0 }; @@ -182,64 +314,85 @@ get_pixel_transfer_program(GLcontext *ctx, const struct state_key *key) GLint row3_p = _mesa_add_state_reference(params, row3_state); const GLuint temp = 1; - /* MOV temp, colorTemp; */ - _mesa_init_instructions(inst + ic, 1); - inst[ic].Opcode = OPCODE_MOV; - inst[ic].DstReg.File = PROGRAM_TEMPORARY; - inst[ic].DstReg.Index = temp; - inst[ic].SrcReg[0].File = PROGRAM_TEMPORARY; - inst[ic].SrcReg[0].Index = colorTemp; - ic++; - - /* XXX reimplement in terms of MUL/MAD (see t_vp_build.c) */ - - /* DP4 colorTemp.x, temp, matrow0; */ + /* DP4 temp.x, colorTemp, matrow0; */ _mesa_init_instructions(inst + ic, 1); inst[ic].Opcode = OPCODE_DP4; inst[ic].DstReg.File = PROGRAM_TEMPORARY; - inst[ic].DstReg.Index = colorTemp; + inst[ic].DstReg.Index = temp; inst[ic].DstReg.WriteMask = WRITEMASK_X; inst[ic].SrcReg[0].File = PROGRAM_TEMPORARY; - inst[ic].SrcReg[0].Index = temp; + inst[ic].SrcReg[0].Index = colorTemp; inst[ic].SrcReg[1].File = PROGRAM_STATE_VAR; inst[ic].SrcReg[1].Index = row0_p; ic++; - /* DP4 colorTemp.y, temp, matrow1; */ + /* DP4 temp.y, colorTemp, matrow1; */ _mesa_init_instructions(inst + ic, 1); inst[ic].Opcode = OPCODE_DP4; inst[ic].DstReg.File = PROGRAM_TEMPORARY; - inst[ic].DstReg.Index = colorTemp; + inst[ic].DstReg.Index = temp; inst[ic].DstReg.WriteMask = WRITEMASK_Y; inst[ic].SrcReg[0].File = PROGRAM_TEMPORARY; - inst[ic].SrcReg[0].Index = temp; + inst[ic].SrcReg[0].Index = colorTemp; inst[ic].SrcReg[1].File = PROGRAM_STATE_VAR; inst[ic].SrcReg[1].Index = row1_p; ic++; - /* DP4 colorTemp.z, temp, matrow2; */ + /* DP4 temp.z, colorTemp, matrow2; */ _mesa_init_instructions(inst + ic, 1); inst[ic].Opcode = OPCODE_DP4; inst[ic].DstReg.File = PROGRAM_TEMPORARY; - inst[ic].DstReg.Index = colorTemp; + inst[ic].DstReg.Index = temp; inst[ic].DstReg.WriteMask = WRITEMASK_Z; inst[ic].SrcReg[0].File = PROGRAM_TEMPORARY; - inst[ic].SrcReg[0].Index = temp; + inst[ic].SrcReg[0].Index = colorTemp; inst[ic].SrcReg[1].File = PROGRAM_STATE_VAR; inst[ic].SrcReg[1].Index = row2_p; ic++; - /* DP4 colorTemp.w, temp, matrow3; */ + /* DP4 temp.w, colorTemp, matrow3; */ _mesa_init_instructions(inst + ic, 1); inst[ic].Opcode = OPCODE_DP4; inst[ic].DstReg.File = PROGRAM_TEMPORARY; - inst[ic].DstReg.Index =colorTemp; + inst[ic].DstReg.Index = temp; inst[ic].DstReg.WriteMask = WRITEMASK_W; inst[ic].SrcReg[0].File = PROGRAM_TEMPORARY; - inst[ic].SrcReg[0].Index = temp; + inst[ic].SrcReg[0].Index = colorTemp; inst[ic].SrcReg[1].File = PROGRAM_STATE_VAR; inst[ic].SrcReg[1].Index = row3_p; ic++; + + /* MOV colorTemp, temp; */ + _mesa_init_instructions(inst + ic, 1); + inst[ic].Opcode = OPCODE_MOV; + inst[ic].DstReg.File = PROGRAM_TEMPORARY; + inst[ic].DstReg.Index = colorTemp; + inst[ic].SrcReg[0].File = PROGRAM_TEMPORARY; + inst[ic].SrcReg[0].Index = temp; + ic++; + } + + if (key->colorMatrixPostScaleBias) { + static const gl_state_index scale_state[STATE_LENGTH] = + { STATE_INTERNAL, STATE_PT_SCALE, 0, 0, 0 }; + static const gl_state_index bias_state[STATE_LENGTH] = + { STATE_INTERNAL, STATE_PT_BIAS, 0, 0, 0 }; + GLint scale_param, bias_param; + + scale_param = _mesa_add_state_reference(params, scale_state); + bias_param = _mesa_add_state_reference(params, bias_state); + + _mesa_init_instructions(inst + ic, 1); + inst[ic].Opcode = OPCODE_MAD; + inst[ic].DstReg.File = PROGRAM_TEMPORARY; + inst[ic].DstReg.Index = colorTemp; + inst[ic].SrcReg[0].File = PROGRAM_TEMPORARY; + inst[ic].SrcReg[0].Index = colorTemp; + inst[ic].SrcReg[1].File = PROGRAM_STATE_VAR; + inst[ic].SrcReg[1].Index = scale_param; + inst[ic].SrcReg[2].File = PROGRAM_STATE_VAR; + inst[ic].SrcReg[2].Index = bias_param; + ic++; } /* Modify last instruction's dst reg to write to result.color */ @@ -285,6 +438,7 @@ get_pixel_transfer_program(GLcontext *ctx, const struct state_key *key) static void update_pixel_transfer(struct st_context *st) { + GLcontext *ctx = st->ctx; struct state_key key; struct gl_fragment_program *fp; @@ -298,16 +452,21 @@ update_pixel_transfer(struct st_context *st) &key, sizeof(key), &fp->Base); } + if (ctx->Pixel.MapColorFlag) { + load_color_map_texture(ctx, st->pixel_xfer.pixelmap_texture); + } + st->pixel_xfer.pixelmap_enabled = ctx->Pixel.MapColorFlag; + st->pixel_xfer.program = (struct st_fragment_program *) fp; } const struct st_tracked_state st_update_pixel_transfer = { - .name = "st_update_pixel_transfer", - .dirty = { - .mesa = _NEW_PIXEL | _NEW_COLOR_MATRIX, - .st = 0, + "st_update_pixel_transfer", /* name */ + { /* dirty */ + _NEW_PIXEL | _NEW_COLOR_MATRIX, /* mesa */ + 0, /* st */ }, - .update = update_pixel_transfer + update_pixel_transfer /* update */ }; diff --git a/src/mesa/state_tracker/st_atom_rasterizer.c b/src/mesa/state_tracker/st_atom_rasterizer.c index bb14cf9045..87a91d56d0 100644 --- a/src/mesa/state_tracker/st_atom_rasterizer.c +++ b/src/mesa/state_tracker/st_atom_rasterizer.c @@ -267,11 +267,11 @@ static void update_raster_state( struct st_context *st ) } const struct st_tracked_state st_update_rasterizer = { - .name = "st_update_rasterizer", - .dirty = { - .mesa = (_NEW_LIGHT | _NEW_POLYGON | _NEW_LINE | _NEW_SCISSOR | - _NEW_POINT | _NEW_BUFFERS | _NEW_MULTISAMPLE), - .st = 0, + "st_update_rasterizer", /* name */ + { /* dirty */ + (_NEW_LIGHT | _NEW_POLYGON | _NEW_LINE | _NEW_SCISSOR | /* mesa */ + _NEW_POINT | _NEW_BUFFERS | _NEW_MULTISAMPLE), + 0, /* st */ }, - .update = update_raster_state + update_raster_state /* update */ }; diff --git a/src/mesa/state_tracker/st_atom_sampler.c b/src/mesa/state_tracker/st_atom_sampler.c index 5dd242ac66..10283d31a1 100644 --- a/src/mesa/state_tracker/st_atom_sampler.c +++ b/src/mesa/state_tracker/st_atom_sampler.c @@ -32,11 +32,14 @@ */ +#include "main/macros.h" + #include "st_context.h" #include "st_atom.h" #include "st_program.h" #include "pipe/p_context.h" #include "pipe/p_defines.h" + #include "cso_cache/cso_context.h" @@ -147,17 +150,14 @@ update_samplers(struct st_context *st) sampler->normalized_coords = 1; sampler->lod_bias = st->ctx->Texture.Unit[su].LodBias; -#if 1 - sampler->min_lod = (texobj->MinLod) < 0.0 ? 0.0 : texobj->MinLod; - sampler->max_lod = texobj->MaxLod; -#else - /* min/max lod should really be as follows (untested). - * Also, calculate_first_last_level() needs to be overhauled - * since today's hardware had real support for LOD clamping. - */ - sampler->min_lod = MAX2(texobj->BaseLevel, texobj->MinLod); - sampler->max_lod = MIN2(texobj->MaxLevel, texobj->MaxLod); -#endif + sampler->min_lod = MAX2(0.0f, texobj->MinLod); + sampler->max_lod = MIN2(texobj->MaxLevel - texobj->BaseLevel, + texobj->MaxLod); + + sampler->border_color[0] = texobj->BorderColor[RCOMP]; + sampler->border_color[1] = texobj->BorderColor[GCOMP]; + sampler->border_color[2] = texobj->BorderColor[BCOMP]; + sampler->border_color[3] = texobj->BorderColor[ACOMP]; sampler->max_anisotropy = texobj->MaxAnisotropy; if (sampler->max_anisotropy > 1.0) { @@ -188,15 +188,10 @@ update_samplers(struct st_context *st) const struct st_tracked_state st_update_sampler = { - .name = "st_update_sampler", - .dirty = { - .mesa = _NEW_TEXTURE, - .st = 0, + "st_update_sampler", /* name */ + { /* dirty */ + _NEW_TEXTURE, /* mesa */ + 0, /* st */ }, - .update = update_samplers + update_samplers /* update */ }; - - - - - diff --git a/src/mesa/state_tracker/st_atom_scissor.c b/src/mesa/state_tracker/st_atom_scissor.c index 59601e91a1..f5db492403 100644 --- a/src/mesa/state_tracker/st_atom_scissor.c +++ b/src/mesa/state_tracker/st_atom_scissor.c @@ -83,15 +83,10 @@ update_scissor( struct st_context *st ) const struct st_tracked_state st_update_scissor = { - .name = "st_update_scissor", - .dirty = { - .mesa = (_NEW_SCISSOR | _NEW_BUFFERS), - .st = 0, + "st_update_scissor", /* name */ + { /* dirty */ + (_NEW_SCISSOR | _NEW_BUFFERS), /* mesa */ + 0, /* st */ }, - .update = update_scissor + update_scissor /* update */ }; - - - - - diff --git a/src/mesa/state_tracker/st_atom_shader.c b/src/mesa/state_tracker/st_atom_shader.c index ec39026eb3..7745591afb 100644 --- a/src/mesa/state_tracker/st_atom_shader.c +++ b/src/mesa/state_tracker/st_atom_shader.c @@ -39,6 +39,7 @@ #include "main/imports.h" #include "main/mtypes.h" +#include "shader/program.h" #include "pipe/p_context.h" #include "pipe/p_shader_tokens.h" @@ -264,27 +265,29 @@ update_linkage( struct st_context *st ) */ assert(st->ctx->VertexProgram._Current); stvp = st_vertex_program(st->ctx->VertexProgram._Current); + assert(stvp->Base.Base.Target == GL_VERTEX_PROGRAM_ARB); assert(st->ctx->FragmentProgram._Current); stfp = st_fragment_program(st->ctx->FragmentProgram._Current); + assert(stfp->Base.Base.Target == GL_FRAGMENT_PROGRAM_ARB); xvp = find_translated_vp(st, stvp, stfp); - st->vp = stvp; - st->fp = stfp; + st_reference_vertprog(st, &st->vp, stvp); + st_reference_fragprog(st, &st->fp, stfp); - st->pipe->bind_vs_state(st->pipe, stvp->driver_shader); - st->pipe->bind_fs_state(st->pipe, stfp->driver_shader); + cso_set_vertex_shader_handle(st->cso_context, stvp->driver_shader); + cso_set_fragment_shader_handle(st->cso_context, stfp->driver_shader); st->vertex_result_to_slot = xvp->output_to_slot; } const struct st_tracked_state st_update_shader = { - .name = "st_update_shader", - .dirty = { - .mesa = 0, - .st = ST_NEW_VERTEX_PROGRAM | ST_NEW_FRAGMENT_PROGRAM + "st_update_shader", /* name */ + { /* dirty */ + 0, /* mesa */ + ST_NEW_VERTEX_PROGRAM | ST_NEW_FRAGMENT_PROGRAM /* st */ }, - .update = update_linkage + update_linkage /* update */ }; diff --git a/src/mesa/state_tracker/st_atom_stipple.c b/src/mesa/state_tracker/st_atom_stipple.c index c91214059a..f395930ab4 100644 --- a/src/mesa/state_tracker/st_atom_stipple.c +++ b/src/mesa/state_tracker/st_atom_stipple.c @@ -54,10 +54,10 @@ update_stipple( struct st_context *st ) const struct st_tracked_state st_update_polygon_stipple = { - .name = "st_update_polygon_stipple", - .dirty = { - .mesa = (_NEW_POLYGONSTIPPLE), - .st = 0, + "st_update_polygon_stipple", /* name */ + { /* dirty */ + (_NEW_POLYGONSTIPPLE), /* mesa */ + 0, /* st */ }, - .update = update_stipple + update_stipple /* update */ }; diff --git a/src/mesa/state_tracker/st_atom_texture.c b/src/mesa/state_tracker/st_atom_texture.c index 01c07dc26d..767654f3d0 100644 --- a/src/mesa/state_tracker/st_atom_texture.c +++ b/src/mesa/state_tracker/st_atom_texture.c @@ -38,6 +38,23 @@ #include "st_cb_texture.h" #include "pipe/p_context.h" #include "pipe/p_inlines.h" +#include "cso_cache/cso_context.h" +#include "util/u_simple_shaders.h" + + +static void * +get_passthrough_fs(struct st_context *st) +{ + struct pipe_shader_state shader; + + if (!st->passthrough_fs) { + st->passthrough_fs = + util_make_fragment_passthrough_shader(st->pipe, &shader); + free((void *) shader.tokens); + } + + return st->passthrough_fs; +} /** @@ -49,6 +66,7 @@ update_textures(struct st_context *st) { struct gl_fragment_program *fprog = st->ctx->FragmentProgram._Current; GLuint su; + GLboolean missing_textures = GL_FALSE; st->state.num_textures = 0; @@ -67,10 +85,13 @@ update_textures(struct st_context *st) retval = st_finalize_texture(st->ctx, st->pipe, texObj, &flush); if (!retval) { /* out of mem */ + missing_textures = GL_TRUE; continue; } st->state.num_textures = su + 1; + + stObj->teximage_realloc = TRUE; } pt = st_get_stobj_texture(stObj); @@ -79,21 +100,23 @@ update_textures(struct st_context *st) pipe_texture_reference(&st->state.sampler_texture[su], pt); } - st->pipe->set_sampler_textures(st->pipe, st->state.num_textures, - st->state.sampler_texture); + cso_set_sampler_textures(st->cso_context, + st->state.num_textures, + st->state.sampler_texture); + + if (missing_textures) { + /* use a pass-through frag shader that uses no textures */ + void *fs = get_passthrough_fs(st); + cso_set_fragment_shader_handle(st->cso_context, fs); + } } const struct st_tracked_state st_update_texture = { - .name = "st_update_texture", - .dirty = { - .mesa = _NEW_TEXTURE, - .st = ST_NEW_FRAGMENT_PROGRAM, + "st_update_texture", /* name */ + { /* dirty */ + _NEW_TEXTURE, /* mesa */ + ST_NEW_FRAGMENT_PROGRAM, /* st */ }, - .update = update_textures + update_textures /* update */ }; - - - - - diff --git a/src/mesa/state_tracker/st_atom_viewport.c b/src/mesa/state_tracker/st_atom_viewport.c index eb3f62cfbe..4b51521470 100644 --- a/src/mesa/state_tracker/st_atom_viewport.c +++ b/src/mesa/state_tracker/st_atom_viewport.c @@ -82,10 +82,10 @@ update_viewport( struct st_context *st ) const struct st_tracked_state st_update_viewport = { - .name = "st_update_viewport", - .dirty = { - .mesa = _NEW_BUFFERS | _NEW_VIEWPORT, - .st = 0, + "st_update_viewport", /* name */ + { /* dirty */ + _NEW_BUFFERS | _NEW_VIEWPORT, /* mesa */ + 0, /* st */ }, - .update = update_viewport + update_viewport /* update */ }; diff --git a/src/mesa/state_tracker/st_cache.c b/src/mesa/state_tracker/st_cache.c deleted file mode 100644 index 7ee4fadc37..0000000000 --- a/src/mesa/state_tracker/st_cache.c +++ /dev/null @@ -1,181 +0,0 @@ -/************************************************************************** - * - * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - - /* - * Authors: - * Zack Rusin <zack@tungstengraphics.com> - */ - -#include "st_cache.h" - -#include "st_context.h" - -#include "pipe/p_state.h" - -#include "cso_cache/cso_cache.h" -#include "cso_cache/cso_hash.h" - - -/* Those function will either find the state of the given template - * in the cache or they will create a new state from the given - * template, insert it in the cache and return it. - */ - -/* - * If the driver returns 0 from the create method then they will assign - * the data member of the cso to be the template itself. - */ - -const struct cso_blend * st_cached_blend_state(struct st_context *st, - const struct pipe_blend_state *templ) -{ - unsigned hash_key = cso_construct_key((void*)templ, sizeof(struct pipe_blend_state)); - struct cso_hash_iter iter = cso_find_state_template(st->cache, - hash_key, CSO_BLEND, - (void*)templ); - if (cso_hash_iter_is_null(iter)) { - struct cso_blend *cso = malloc(sizeof(struct cso_blend)); - memcpy(&cso->state, templ, sizeof(struct pipe_blend_state)); - cso->data = st->pipe->create_blend_state(st->pipe, &cso->state); - if (!cso->data) - cso->data = &cso->state; - cso->delete_state = (cso_state_callback)st->pipe->delete_blend_state; - cso->context = st->pipe; - iter = cso_insert_state(st->cache, hash_key, CSO_BLEND, cso); - } - return ((struct cso_blend *)cso_hash_iter_data(iter)); -} - -const struct cso_sampler * -st_cached_sampler_state(struct st_context *st, - const struct pipe_sampler_state *templ) -{ - unsigned hash_key = cso_construct_key((void*)templ, sizeof(struct pipe_sampler_state)); - struct cso_hash_iter iter = cso_find_state_template(st->cache, - hash_key, CSO_SAMPLER, - (void*)templ); - if (cso_hash_iter_is_null(iter)) { - struct cso_sampler *cso = malloc(sizeof(struct cso_sampler)); - memcpy(&cso->state, templ, sizeof(struct pipe_sampler_state)); - cso->data = st->pipe->create_sampler_state(st->pipe, &cso->state); - if (!cso->data) - cso->data = &cso->state; - cso->delete_state = (cso_state_callback)st->pipe->delete_sampler_state; - cso->context = st->pipe; - iter = cso_insert_state(st->cache, hash_key, CSO_SAMPLER, cso); - } - return (struct cso_sampler*)(cso_hash_iter_data(iter)); -} - -const struct cso_depth_stencil_alpha * -st_cached_depth_stencil_alpha_state(struct st_context *st, - const struct pipe_depth_stencil_alpha_state *templ) -{ - unsigned hash_key = cso_construct_key((void*)templ, - sizeof(struct pipe_depth_stencil_alpha_state)); - struct cso_hash_iter iter = cso_find_state_template(st->cache, - hash_key, - CSO_DEPTH_STENCIL_ALPHA, - (void*)templ); - if (cso_hash_iter_is_null(iter)) { - struct cso_depth_stencil_alpha *cso = malloc(sizeof(struct cso_depth_stencil_alpha)); - memcpy(&cso->state, templ, sizeof(struct pipe_depth_stencil_alpha_state)); - cso->data = st->pipe->create_depth_stencil_alpha_state(st->pipe, &cso->state); - if (!cso->data) - cso->data = &cso->state; - cso->delete_state = (cso_state_callback)st->pipe->delete_depth_stencil_alpha_state; - cso->context = st->pipe; - iter = cso_insert_state(st->cache, hash_key, CSO_DEPTH_STENCIL_ALPHA, cso); - } - return (struct cso_depth_stencil_alpha*)(cso_hash_iter_data(iter)); -} - -const struct cso_rasterizer* st_cached_rasterizer_state( - struct st_context *st, - const struct pipe_rasterizer_state *templ) -{ - unsigned hash_key = cso_construct_key((void*)templ, - sizeof(struct pipe_rasterizer_state)); - struct cso_hash_iter iter = cso_find_state_template(st->cache, - hash_key, CSO_RASTERIZER, - (void*)templ); - if (cso_hash_iter_is_null(iter)) { - struct cso_rasterizer *cso = malloc(sizeof(struct cso_rasterizer)); - memcpy(&cso->state, templ, sizeof(struct pipe_rasterizer_state)); - cso->data = st->pipe->create_rasterizer_state(st->pipe, &cso->state); - if (!cso->data) - cso->data = &cso->state; - cso->delete_state = (cso_state_callback)st->pipe->delete_rasterizer_state; - cso->context = st->pipe; - iter = cso_insert_state(st->cache, hash_key, CSO_RASTERIZER, cso); - } - return (struct cso_rasterizer*)(cso_hash_iter_data(iter)); -} - -const struct cso_fragment_shader * -st_cached_fs_state(struct st_context *st, - const struct pipe_shader_state *templ) -{ - unsigned hash_key = cso_construct_key((void*)templ, - sizeof(struct pipe_shader_state)); - struct cso_hash_iter iter = cso_find_state_template(st->cache, - hash_key, CSO_FRAGMENT_SHADER, - (void*)templ); - if (cso_hash_iter_is_null(iter)) { - struct cso_fragment_shader *cso = malloc(sizeof(struct cso_fragment_shader)); - memcpy(&cso->state, templ, sizeof(struct pipe_shader_state)); - cso->data = st->pipe->create_fs_state(st->pipe, &cso->state); - if (!cso->data) - cso->data = &cso->state; - cso->delete_state = (cso_state_callback)st->pipe->delete_fs_state; - cso->context = st->pipe; - iter = cso_insert_state(st->cache, hash_key, CSO_FRAGMENT_SHADER, cso); - } - return (struct cso_fragment_shader*)(cso_hash_iter_data(iter)); -} - -const struct cso_vertex_shader * -st_cached_vs_state(struct st_context *st, - const struct pipe_shader_state *templ) -{ - unsigned hash_key = cso_construct_key((void*)templ, - sizeof(struct pipe_shader_state)); - struct cso_hash_iter iter = cso_find_state_template(st->cache, - hash_key, CSO_VERTEX_SHADER, - (void*)templ); - if (cso_hash_iter_is_null(iter)) { - struct cso_vertex_shader *cso = malloc(sizeof(struct cso_vertex_shader)); - memcpy(&cso->state, templ, sizeof(struct pipe_shader_state)); - cso->data = st->pipe->create_vs_state(st->pipe, &cso->state); - if (!cso->data) - cso->data = &cso->state; - cso->delete_state = (cso_state_callback)st->pipe->delete_vs_state; - cso->context = st->pipe; - iter = cso_insert_state(st->cache, hash_key, CSO_VERTEX_SHADER, cso); - } - return (struct cso_vertex_shader*)(cso_hash_iter_data(iter)); -} diff --git a/src/mesa/state_tracker/st_cb_accum.c b/src/mesa/state_tracker/st_cb_accum.c index a623d0bcc0..1636bed91a 100644 --- a/src/mesa/state_tracker/st_cb_accum.c +++ b/src/mesa/state_tracker/st_cb_accum.c @@ -56,54 +56,125 @@ */ +/** + * Wrapper for pipe_get_tile_rgba(). Do format/cpp override to make the + * tile util function think the surface is 16bit/channel, even if it's not. + * See also: st_renderbuffer_alloc_storage() + */ +static void +acc_get_tile_rgba(struct pipe_context *pipe, struct pipe_surface *acc_ps, + uint x, uint y, uint w, uint h, float *p) +{ + const enum pipe_format f = acc_ps->format; + const int cpp = acc_ps->cpp; + + acc_ps->format = DEFAULT_ACCUM_PIPE_FORMAT; + acc_ps->cpp = 8; + + pipe_get_tile_rgba(pipe, acc_ps, x, y, w, h, p); + + acc_ps->format = f; + acc_ps->cpp = cpp; +} + + +/** + * Wrapper for pipe_put_tile_rgba(). Do format/cpp override to make the + * tile util function think the surface is 16bit/channel, even if it's not. + * See also: st_renderbuffer_alloc_storage() + */ +static void +acc_put_tile_rgba(struct pipe_context *pipe, struct pipe_surface *acc_ps, + uint x, uint y, uint w, uint h, const float *p) +{ + enum pipe_format f = acc_ps->format; + const int cpp = acc_ps->cpp; + + acc_ps->format = DEFAULT_ACCUM_PIPE_FORMAT; + acc_ps->cpp = 8; + + pipe_put_tile_rgba(pipe, acc_ps, x, y, w, h, p); + + acc_ps->format = f; + acc_ps->cpp = cpp; +} + + + void st_clear_accum_buffer(GLcontext *ctx, struct gl_renderbuffer *rb) { - struct pipe_context *pipe = ctx->st->pipe; struct st_renderbuffer *acc_strb = st_renderbuffer(rb); struct pipe_surface *acc_ps = acc_strb->surface; const GLint xpos = ctx->DrawBuffer->_Xmin; const GLint ypos = ctx->DrawBuffer->_Ymin; const GLint width = ctx->DrawBuffer->_Xmax - xpos; const GLint height = ctx->DrawBuffer->_Ymax - ypos; - const GLfloat r = ctx->Accum.ClearColor[0]; - const GLfloat g = ctx->Accum.ClearColor[1]; - const GLfloat b = ctx->Accum.ClearColor[2]; - const GLfloat a = ctx->Accum.ClearColor[3]; - GLfloat *accBuf = (GLfloat *) malloc(width * height * 4 * sizeof(GLfloat)); - int i; - - for (i = 0; i < width * height; i++) { - accBuf[i*4+0] = r; - accBuf[i*4+1] = g; - accBuf[i*4+2] = b; - accBuf[i*4+3] = a; + GLvoid *map; + + map = pipe_surface_map(acc_ps); + + /* note acc_strb->format might not equal acc_ps->format */ + switch (acc_strb->format) { + case PIPE_FORMAT_R16G16B16A16_SNORM: + { + GLshort r = FLOAT_TO_SHORT(ctx->Accum.ClearColor[0]); + GLshort g = FLOAT_TO_SHORT(ctx->Accum.ClearColor[1]); + GLshort b = FLOAT_TO_SHORT(ctx->Accum.ClearColor[2]); + GLshort a = FLOAT_TO_SHORT(ctx->Accum.ClearColor[3]); + int i, j; + for (i = 0; i < height; i++) { + GLshort *dst = ((GLshort *) map + + ((ypos + i) * acc_ps->pitch + xpos) * 4); + for (j = 0; j < width; j++) { + dst[0] = r; + dst[1] = g; + dst[2] = b; + dst[3] = a; + dst += 4; + } + } + } + break; + default: + _mesa_problem(ctx, "unexpected format in st_clear_accum_buffer()"); } - pipe_put_tile_rgba(pipe, acc_ps, xpos, ypos, width, height, accBuf); + pipe_surface_unmap(acc_ps); } /** For ADD/MULT */ static void -accum_mad(struct pipe_context *pipe, GLfloat scale, GLfloat bias, +accum_mad(GLcontext *ctx, GLfloat scale, GLfloat bias, GLint xpos, GLint ypos, GLint width, GLint height, - struct pipe_surface *acc_ps) + struct st_renderbuffer *acc_strb) { - GLfloat *accBuf; - GLint i; - - accBuf = (GLfloat *) malloc(width * height * 4 * sizeof(GLfloat)); - - pipe_get_tile_rgba(pipe, acc_ps, xpos, ypos, width, height, accBuf); - - for (i = 0; i < 4 * width * height; i++) { - accBuf[i] = accBuf[i] * scale + bias; + struct pipe_surface *acc_ps = acc_strb->surface; + GLvoid *map; + + map = pipe_surface_map(acc_ps); + + /* note acc_strb->format might not equal acc_ps->format */ + switch (acc_strb->format) { + case PIPE_FORMAT_R16G16B16A16_SNORM: + { + int i, j; + for (i = 0; i < height; i++) { + GLshort *acc = ((GLshort *) map + + ((ypos + i) * acc_ps->pitch + xpos) * 4); + for (j = 0; j < width * 4; j++) { + float val = SHORT_TO_FLOAT(acc[j]) * scale + bias; + acc[j] = FLOAT_TO_SHORT(val); + } + } + } + break; + default: + _mesa_problem(NULL, "unexpected format in st_clear_accum_buffer()"); } - pipe_put_tile_rgba(pipe, acc_ps, xpos, ypos, width, height, accBuf); - - free(accBuf); + pipe_surface_unmap(acc_ps); } @@ -120,13 +191,13 @@ accum_accum(struct pipe_context *pipe, GLfloat value, accBuf = (GLfloat *) malloc(width * height * 4 * sizeof(GLfloat)); pipe_get_tile_rgba(pipe, color_ps, xpos, ypos, width, height, colorBuf); - pipe_get_tile_rgba(pipe, acc_ps, xpos, ypos, width, height, accBuf); + acc_get_tile_rgba(pipe, acc_ps, xpos, ypos, width, height, accBuf); for (i = 0; i < 4 * width * height; i++) { accBuf[i] = accBuf[i] + colorBuf[i] * value; } - pipe_put_tile_rgba(pipe, acc_ps, xpos, ypos, width, height, accBuf); + acc_put_tile_rgba(pipe, acc_ps, xpos, ypos, width, height, accBuf); free(colorBuf); free(accBuf); @@ -150,7 +221,7 @@ accum_load(struct pipe_context *pipe, GLfloat value, buf[i] = buf[i] * value; } - pipe_put_tile_rgba(pipe, acc_ps, xpos, ypos, width, height, buf); + acc_put_tile_rgba(pipe, acc_ps, xpos, ypos, width, height, buf); free(buf); } @@ -169,7 +240,7 @@ accum_return(GLcontext *ctx, GLfloat value, abuf = (GLfloat *) malloc(width * height * 4 * sizeof(GLfloat)); - pipe_get_tile_rgba(pipe, acc_ps, xpos, ypos, width, height, abuf); + acc_get_tile_rgba(pipe, acc_ps, xpos, ypos, width, height, abuf); if (!colormask[0] || !colormask[1] || !colormask[2] || !colormask[3]) { cbuf = (GLfloat *) malloc(width * height * 4 * sizeof(GLfloat)); @@ -219,12 +290,12 @@ st_Accum(GLcontext *ctx, GLenum op, GLfloat value) switch (op) { case GL_ADD: if (value != 0.0F) { - accum_mad(pipe, 1.0, value, xpos, ypos, width, height, acc_ps); + accum_mad(ctx, 1.0, value, xpos, ypos, width, height, acc_strb); } break; case GL_MULT: if (value != 1.0F) { - accum_mad(pipe, value, 0.0, xpos, ypos, width, height, acc_ps); + accum_mad(ctx, value, 0.0, xpos, ypos, width, height, acc_strb); } break; case GL_ACCUM: diff --git a/src/mesa/state_tracker/st_cb_bitmap.c b/src/mesa/state_tracker/st_cb_bitmap.c index 59f3c385df..e95ff5e2e0 100644 --- a/src/mesa/state_tracker/st_cb_bitmap.c +++ b/src/mesa/state_tracker/st_cb_bitmap.c @@ -50,7 +50,6 @@ #include "pipe/p_context.h" #include "pipe/p_defines.h" #include "pipe/p_inlines.h" -#include "pipe/p_winsys.h" #include "util/p_tile.h" #include "util/u_draw_quad.h" #include "util/u_simple_shaders.h" @@ -90,10 +89,14 @@ struct bitmap_cache GLint xpos, ypos; /** Bounds of region used in window coords */ GLint xmin, ymin, xmax, ymax; + struct pipe_texture *texture; + struct pipe_surface *surf; + GLboolean empty; + /** An I8 texture image: */ - GLubyte buffer[BITMAP_CACHE_HEIGHT][BITMAP_CACHE_WIDTH]; + ubyte *buffer; }; @@ -105,7 +108,7 @@ struct bitmap_cache * This program will be combined with the user's fragment program. */ static struct st_fragment_program * -make_bitmap_fragment_program(GLcontext *ctx) +make_bitmap_fragment_program(GLcontext *ctx, GLuint samplerIndex) { struct st_fragment_program *stfp; struct gl_program *p; @@ -130,7 +133,7 @@ make_bitmap_fragment_program(GLcontext *ctx) p->Instructions[ic].DstReg.Index = 0; p->Instructions[ic].SrcReg[0].File = PROGRAM_INPUT; p->Instructions[ic].SrcReg[0].Index = FRAG_ATTRIB_TEX0; - p->Instructions[ic].TexSrcUnit = 0; + p->Instructions[ic].TexSrcUnit = samplerIndex; p->Instructions[ic].TexSrcTarget = TEXTURE_2D_INDEX; ic++; @@ -148,7 +151,7 @@ make_bitmap_fragment_program(GLcontext *ctx) p->InputsRead = FRAG_BIT_TEX0; p->OutputsWritten = 0x0; - p->SamplersUsed = 0x1; /* sampler 0 (bit 0) is used */ + p->SamplersUsed = (1 << samplerIndex); stfp = (struct st_fragment_program *) p; stfp->Base.UsesKill = GL_TRUE; @@ -158,6 +161,19 @@ make_bitmap_fragment_program(GLcontext *ctx) } +static int +find_free_bit(uint bitfield) +{ + int i; + for (i = 0; i < 32; i++) { + if ((bitfield & (1 << i)) == 0) { + return i; + } + } + return -1; +} + + /** * Combine basic bitmap fragment program with the user-defined program. */ @@ -165,28 +181,30 @@ static struct st_fragment_program * combined_bitmap_fragment_program(GLcontext *ctx) { struct st_context *st = ctx->st; - struct st_fragment_program *stfp; + struct st_fragment_program *stfp = st->fp; - if (!st->bitmap.program) { - /* create the basic bitmap fragment program */ - st->bitmap.program = make_bitmap_fragment_program(ctx); - } - - if (st->bitmap.user_prog_sn == st->fp->serialNo) { - /* re-use */ - stfp = st->bitmap.combined_prog; - } - else { - /* Concatenate the bitmap program with the current user-defined program. + if (!stfp->bitmap_program) { + /* + * Generate new program which is the user-defined program prefixed + * with the bitmap sampler/kill instructions. */ - stfp = (struct st_fragment_program *) + struct st_fragment_program *bitmap_prog; + uint sampler; + + sampler = find_free_bit(st->fp->Base.Base.SamplersUsed); + bitmap_prog = make_bitmap_fragment_program(ctx, sampler); + + stfp->bitmap_program = (struct st_fragment_program *) _mesa_combine_programs(ctx, - &st->bitmap.program->Base.Base, - &st->fp->Base.Base); + &bitmap_prog->Base.Base, &stfp->Base.Base); + stfp->bitmap_program->bitmap_sampler = sampler; + + /* done with this after combining */ + st_reference_fragprog(st, &bitmap_prog, NULL); #if 0 { - struct gl_program *p = &stfp->Base.Base; + struct gl_program *p = &stfp->bitmap_program->Base.Base; printf("Combined bitmap program:\n"); _mesa_print_program(p); printf("InputsRead: 0x%x\n", p->InputsRead); @@ -196,11 +214,7 @@ combined_bitmap_fragment_program(GLcontext *ctx) #endif /* translate to TGSI tokens */ - st_translate_fragment_program(st, stfp, NULL); - - /* save new program, update serial numbers */ - st->bitmap.user_prog_sn = st->fp->serialNo; - st->bitmap.combined_prog = stfp; + st_translate_fragment_program(st, stfp->bitmap_program, NULL); } /* Ideally we'd have updated the pipe constants during the normal @@ -208,81 +222,42 @@ combined_bitmap_fragment_program(GLcontext *ctx) */ st_upload_constants(st, stfp->Base.Base.Parameters, PIPE_SHADER_FRAGMENT); - return stfp; + return stfp->bitmap_program; } /** - * Create a texture which represents a bitmap image. + * Copy user-provide bitmap bits into texture buffer, expanding + * bits into texels. + * "On" bits will set texels to 0xff. + * "Off" bits will not modify texels. + * Note that the image is actually going to be upside down in + * the texture. We deal with that with texcoords. */ -static struct pipe_texture * -make_bitmap_texture(GLcontext *ctx, GLsizei width, GLsizei height, - const struct gl_pixelstore_attrib *unpack, - const GLubyte *bitmap) +static void +unpack_bitmap(struct st_context *st, + GLint px, GLint py, GLsizei width, GLsizei height, + const struct gl_pixelstore_attrib *unpack, + const GLubyte *bitmap, + ubyte *destBuffer, uint destStride) { - struct pipe_context *pipe = ctx->st->pipe; - struct pipe_screen *screen = pipe->screen; - struct pipe_surface *surface; - uint format = 0, cpp, comp; - ubyte *dest; - struct pipe_texture *pt; - int row, col; - - /* find a texture format we know */ - if (screen->is_format_supported( screen, PIPE_FORMAT_U_I8, PIPE_TEXTURE )) { - format = PIPE_FORMAT_U_I8; - cpp = 1; - comp = 0; - } - else if (screen->is_format_supported( screen, PIPE_FORMAT_A8R8G8B8_UNORM, PIPE_TEXTURE )) { - format = PIPE_FORMAT_A8R8G8B8_UNORM; - cpp = 4; - comp = 3; /* alpha channel */ /*XXX little-endian dependency */ - } - else { - /* XXX support more formats */ - assert( 0 ); - } - - /* PBO source... */ - bitmap = _mesa_map_bitmap_pbo(ctx, unpack, bitmap); - if (!bitmap) { - return NULL; - } - - /** - * Create texture to hold bitmap pattern. - */ - pt = st_texture_create(ctx->st, PIPE_TEXTURE_2D, format, 0, width, height, - 1, 0); - if (!pt) { - _mesa_unmap_bitmap_pbo(ctx, unpack); - return NULL; - } + GLint row, col; - surface = screen->get_tex_surface(screen, pt, 0, 0, 0); - - /* map texture surface */ - dest = pipe_surface_map(surface); - - /* Put image into texture surface. - * Note that the image is actually going to be upside down in - * the texture. We deal with that with texcoords. - */ +#define SET_PIXEL(COL, ROW) \ + destBuffer[(py + (ROW)) * destStride + px + (COL)] = 0x0; for (row = 0; row < height; row++) { const GLubyte *src = (const GLubyte *) _mesa_image_address2d(unpack, bitmap, width, height, GL_COLOR_INDEX, GL_BITMAP, row, 0); - ubyte *destRow = dest + row * surface->pitch * cpp; if (unpack->LsbFirst) { /* Lsb first */ GLubyte mask = 1U << (unpack->SkipPixels & 0x7); for (col = 0; col < width; col++) { - /* set texel to 255 if bit is set */ - destRow[comp] = (*src & mask) ? 0x0 : 0xff; - destRow += cpp; + if (*src & mask) { + SET_PIXEL(col, row); + } if (mask == 128U) { src++; @@ -302,9 +277,9 @@ make_bitmap_texture(GLcontext *ctx, GLsizei width, GLsizei height, GLubyte mask = 128U >> (unpack->SkipPixels & 0x7); for (col = 0; col < width; col++) { - /* set texel to 255 if bit is set */ - destRow[comp] =(*src & mask) ? 0x0 : 0xff; - destRow += cpp; + if (*src & mask) { + SET_PIXEL(col, row); + } if (mask == 1U) { src++; @@ -322,6 +297,50 @@ make_bitmap_texture(GLcontext *ctx, GLsizei width, GLsizei height, } /* row */ +#undef SET_PIXEL +} + + +/** + * Create a texture which represents a bitmap image. + */ +static struct pipe_texture * +make_bitmap_texture(GLcontext *ctx, GLsizei width, GLsizei height, + const struct gl_pixelstore_attrib *unpack, + const GLubyte *bitmap) +{ + struct pipe_context *pipe = ctx->st->pipe; + struct pipe_screen *screen = pipe->screen; + struct pipe_surface *surface; + ubyte *dest; + struct pipe_texture *pt; + + /* PBO source... */ + bitmap = _mesa_map_bitmap_pbo(ctx, unpack, bitmap); + if (!bitmap) { + return NULL; + } + + /** + * Create texture to hold bitmap pattern. + */ + pt = st_texture_create(ctx->st, PIPE_TEXTURE_2D, ctx->st->bitmap.tex_format, + 0, width, height, 1, 0); + if (!pt) { + _mesa_unmap_bitmap_pbo(ctx, unpack); + return NULL; + } + + surface = screen->get_tex_surface(screen, pt, 0, 0, 0); + + /* map texture surface */ + dest = pipe_surface_map(surface); + + /* Put image into texture surface */ + memset(dest, 0xff, height * surface->pitch); + unpack_bitmap(ctx->st, 0, 0, width, height, unpack, bitmap, + dest, surface->pitch); + _mesa_unmap_bitmap_pbo(ctx, unpack); /* Release surface */ @@ -329,8 +348,6 @@ make_bitmap_texture(GLcontext *ctx, GLsizei width, GLsizei height, pipe_surface_reference(&surface, NULL); pipe->texture_update(pipe, pt, 0, 0x1); - pt->format = format; - return pt; } @@ -348,11 +365,8 @@ setup_bitmap_vertex_data(struct st_context *st, const GLfloat x1 = x + width; const GLfloat y0 = y; const GLfloat y1 = y + height; - const GLfloat bias = st->bitmap_texcoord_bias; - const GLfloat xBias = bias / (x1-x0); - const GLfloat yBias = bias / (y1-y0); - const GLfloat sLeft = 0.0 + xBias, sRight = 1.0 + xBias; - const GLfloat tTop = yBias, tBot = 1.0 - tTop - yBias; + const GLfloat sLeft = 0.0F, sRight = 1.0F; + const GLfloat tTop = 0.0, tBot = 1.0 - tTop; const GLfloat clip_x0 = x0 / fb_width * 2.0 - 1.0; const GLfloat clip_y0 = y0 / fb_height * 2.0 - 1.0; const GLfloat clip_x1 = x1 / fb_width * 2.0 - 1.0; @@ -361,9 +375,8 @@ setup_bitmap_vertex_data(struct st_context *st, void *buf; if (!st->bitmap.vbuf) { - st->bitmap.vbuf = pipe->winsys->buffer_create(pipe->winsys, 32, - PIPE_BUFFER_USAGE_VERTEX, - sizeof(st->bitmap.vertices)); + st->bitmap.vbuf = pipe_buffer_create(pipe, 32, PIPE_BUFFER_USAGE_VERTEX, + sizeof(st->bitmap.vertices)); } /* Positions are in clip coords since we need to do clipping in case @@ -402,10 +415,9 @@ setup_bitmap_vertex_data(struct st_context *st, } /* put vertex data into vbuf */ - buf = pipe->winsys->buffer_map(pipe->winsys, st->bitmap.vbuf, - PIPE_BUFFER_USAGE_CPU_WRITE); + buf = pipe_buffer_map(pipe, st->bitmap.vbuf, PIPE_BUFFER_USAGE_CPU_WRITE); memcpy(buf, st->bitmap.vertices, sizeof(st->bitmap.vertices)); - pipe->winsys->buffer_unmap(pipe->winsys, st->bitmap.vbuf); + pipe_buffer_unmap(pipe, st->bitmap.vbuf); } @@ -438,21 +450,38 @@ draw_bitmap_quad(GLcontext *ctx, GLint x, GLint y, GLfloat z, cso_save_samplers(cso); cso_save_sampler_textures(cso); cso_save_viewport(cso); + cso_save_fragment_shader(cso); + cso_save_vertex_shader(cso); /* rasterizer state: just scissor */ st->bitmap.rasterizer.scissor = ctx->Scissor.Enabled; cso_set_rasterizer(cso, &st->bitmap.rasterizer); /* fragment shader state: TEX lookup program */ - pipe->bind_fs_state(pipe, stfp->driver_shader); + cso_set_fragment_shader_handle(cso, stfp->driver_shader); /* vertex shader state: position + texcoord pass-through */ - pipe->bind_vs_state(pipe, st->bitmap.vs); + cso_set_vertex_shader_handle(cso, st->bitmap.vs); - /* sampler / texture state */ - cso_single_sampler(cso, 0, &st->bitmap.sampler); - cso_single_sampler_done(cso); - pipe->set_sampler_textures(pipe, 1, &pt); + /* user samplers, plus our bitmap sampler */ + { + struct pipe_sampler_state *samplers[PIPE_MAX_SAMPLERS]; + uint num = MAX2(stfp->bitmap_sampler + 1, st->state.num_samplers); + uint i; + for (i = 0; i < st->state.num_samplers; i++) { + samplers[i] = &st->state.samplers[i]; + } + samplers[stfp->bitmap_sampler] = &st->bitmap.sampler; + cso_set_samplers(cso, num, (const struct pipe_sampler_state **) samplers); } + + /* user textures, plus the bitmap texture */ + { + struct pipe_texture *textures[PIPE_MAX_SAMPLERS]; + uint num = MAX2(stfp->bitmap_sampler + 1, st->state.num_textures); + memcpy(textures, st->state.sampler_texture, sizeof(textures)); + textures[stfp->bitmap_sampler] = pt; + cso_set_sampler_textures(cso, num, textures); + } /* viewport state: viewport matching window dims */ { @@ -488,55 +517,42 @@ draw_bitmap_quad(GLcontext *ctx, GLint x, GLint y, GLfloat z, cso_restore_samplers(cso); cso_restore_sampler_textures(cso); cso_restore_viewport(cso); - /* shaders don't go through cso yet */ - pipe->bind_fs_state(pipe, st->fp->driver_shader); - pipe->bind_vs_state(pipe, st->vp->driver_shader); + cso_restore_fragment_shader(cso); + cso_restore_vertex_shader(cso); } static void reset_cache(struct st_context *st) { - memset(st->bitmap.cache->buffer, 0xff, sizeof(st->bitmap.cache->buffer)); - st->bitmap.cache->empty = GL_TRUE; - - st->bitmap.cache->xmin = 1000000; - st->bitmap.cache->xmax = -1000000; - st->bitmap.cache->ymin = 1000000; - st->bitmap.cache->ymax = -1000000; -} - - -static void -init_bitmap_cache(struct st_context *st) -{ struct pipe_context *pipe = st->pipe; struct pipe_screen *screen = pipe->screen; - enum pipe_format format; + struct bitmap_cache *cache = st->bitmap.cache; - st->bitmap.cache = CALLOC_STRUCT(bitmap_cache); - if (!st->bitmap.cache) - return; + //memset(cache->buffer, 0xff, sizeof(cache->buffer)); + cache->empty = GL_TRUE; - /* find a usable texture format */ - if (screen->is_format_supported(screen, PIPE_FORMAT_U_I8, PIPE_TEXTURE)) { - format = PIPE_FORMAT_U_I8; - } - else { - /* XXX support more formats */ - assert(0); - } + cache->xmin = 1000000; + cache->xmax = -1000000; + cache->ymin = 1000000; + cache->ymax = -1000000; - st->bitmap.cache->texture - = st_texture_create(st, PIPE_TEXTURE_2D, format, 0, - BITMAP_CACHE_WIDTH, BITMAP_CACHE_HEIGHT, 1, 0); - if (!st->bitmap.cache->texture) { - FREE(st->bitmap.cache); - st->bitmap.cache = NULL; - return; - } + assert(!cache->texture); - reset_cache(st); + /* allocate a new texture */ + cache->texture = st_texture_create(st, PIPE_TEXTURE_2D, + st->bitmap.tex_format, 0, + BITMAP_CACHE_WIDTH, BITMAP_CACHE_HEIGHT, + 1, 0); + + /* Map the texture surface. + * Subsequent glBitmap calls will write into the texture image. + */ + cache->surf = screen->get_tex_surface(screen, cache->texture, 0, 0, 0); + cache->buffer = pipe_surface_map(cache->surf); + + /* init image to all 0xff */ + memset(cache->buffer, 0xff, BITMAP_CACHE_WIDTH * BITMAP_CACHE_HEIGHT); } @@ -547,36 +563,37 @@ void st_flush_bitmap_cache(struct st_context *st) { if (!st->bitmap.cache->empty) { - struct bitmap_cache *cache = st->bitmap.cache; - struct pipe_context *pipe = st->pipe; - struct pipe_screen *screen = pipe->screen; - struct pipe_surface *surf; - void *dest; + if (st->ctx->DrawBuffer) { + struct bitmap_cache *cache = st->bitmap.cache; + struct pipe_context *pipe = st->pipe; + + assert(cache->xmin <= cache->xmax); + /* + printf("flush size %d x %d at %d, %d\n", + cache->xmax - cache->xmin, + cache->ymax - cache->ymin, + cache->xpos, cache->ypos); + */ + + /* The texture surface has been mapped until now. + * So unmap and release the texture surface before drawing. + */ + pipe_surface_unmap(cache->surf); + pipe_surface_reference(&cache->surf, NULL); - assert(cache->xmin <= cache->xmax); - /* - printf("flush size %d x %d at %d, %d\n", - cache->xmax - cache->xmin, - cache->ymax - cache->ymin, - cache->xpos, cache->ypos); - */ - - /* update the texture map image */ - surf = screen->get_tex_surface(screen, cache->texture, 0, 0, 0); - dest = pipe_surface_map(surf); - memcpy(dest, cache->buffer, sizeof(cache->buffer)); - pipe_surface_unmap(surf); - pipe_surface_reference(&surf, NULL); - - pipe->texture_update(pipe, cache->texture, 0, 0x1); - - draw_bitmap_quad(st->ctx, - cache->xpos, - cache->ypos, - st->ctx->Current.RasterPos[2], - BITMAP_CACHE_WIDTH, BITMAP_CACHE_HEIGHT, - cache->texture); + /* XXX is this needed? */ + pipe->texture_update(pipe, cache->texture, 0, 0x1); + + draw_bitmap_quad(st->ctx, + cache->xpos, + cache->ypos, + st->ctx->Current.RasterPos[2], + BITMAP_CACHE_WIDTH, BITMAP_CACHE_HEIGHT, + cache->texture); + /* release/free the texture */ + pipe_texture_reference(&cache->texture, NULL); + } reset_cache(st); } } @@ -593,7 +610,6 @@ accum_bitmap(struct st_context *st, const GLubyte *bitmap ) { struct bitmap_cache *cache = st->bitmap.cache; - int row, col; int px = -999, py; if (width > BITMAP_CACHE_WIDTH || @@ -632,60 +648,8 @@ accum_bitmap(struct st_context *st, if (y + height > cache->ymax) cache->ymax = y + height; - /* XXX try to combine this code with code in make_bitmap_texture() */ -#define SET_PIXEL(COL, ROW) \ - cache->buffer[py + (ROW)][px + (COL)] = 0x0; - - for (row = 0; row < height; row++) { - const GLubyte *src = (const GLubyte *) _mesa_image_address2d(unpack, - bitmap, width, height, GL_COLOR_INDEX, GL_BITMAP, row, 0); - - if (unpack->LsbFirst) { - /* Lsb first */ - GLubyte mask = 1U << (unpack->SkipPixels & 0x7); - for (col = 0; col < width; col++) { - - if (*src & mask) { - SET_PIXEL(col, row); - } - - if (mask == 128U) { - src++; - mask = 1U; - } - else { - mask = mask << 1; - } - } - - /* get ready for next row */ - if (mask != 1) - src++; - } - else { - /* Msb first */ - GLubyte mask = 128U >> (unpack->SkipPixels & 0x7); - for (col = 0; col < width; col++) { - - if (*src & mask) { - SET_PIXEL(col, row); - } - - if (mask == 1U) { - src++; - mask = 128U; - } - else { - mask = mask >> 1; - } - } - - /* get ready for next row */ - if (mask != 128) - src++; - } - - } /* row */ + unpack_bitmap(st, px, py, width, height, unpack, bitmap, + cache->buffer, BITMAP_CACHE_WIDTH); return GL_TRUE; /* accumulated */ } @@ -724,6 +688,7 @@ st_Bitmap(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, assert(pt->target == PIPE_TEXTURE_2D); draw_bitmap_quad(ctx, x, y, ctx->Current.RasterPos[2], width, height, pt); + /* release/free the texture */ pipe_texture_reference(&pt, NULL); } } @@ -742,6 +707,8 @@ void st_init_bitmap(struct st_context *st) { struct pipe_sampler_state *sampler = &st->bitmap.sampler; + struct pipe_context *pipe = st->pipe; + struct pipe_screen *screen = pipe->screen; /* init sampler state once */ memset(sampler, 0, sizeof(*sampler)); @@ -758,7 +725,19 @@ st_init_bitmap(struct st_context *st) st->bitmap.rasterizer.gl_rasterization_rules = 1; st->bitmap.rasterizer.bypass_vs = 1; - init_bitmap_cache(st); + /* find a usable texture format */ + if (screen->is_format_supported(screen, PIPE_FORMAT_I8_UNORM, PIPE_TEXTURE)) { + st->bitmap.tex_format = PIPE_FORMAT_I8_UNORM; + } + else { + /* XXX support more formats */ + assert(0); + } + + /* alloc bitmap cache object */ + st->bitmap.cache = CALLOC_STRUCT(bitmap_cache); + + reset_cache(st); } @@ -768,21 +747,13 @@ st_destroy_bitmap(struct st_context *st) { struct pipe_context *pipe = st->pipe; - if (st->bitmap.combined_prog) { - st_delete_program(st->ctx, &st->bitmap.combined_prog->Base.Base); - } - - if (st->bitmap.program) { - st_delete_program(st->ctx, &st->bitmap.program->Base.Base); - } - if (st->bitmap.vs) { - pipe->delete_vs_state(pipe, st->bitmap.vs); + cso_delete_vertex_shader(st->cso_context, st->bitmap.vs); st->bitmap.vs = NULL; } if (st->bitmap.vbuf) { - pipe->winsys->buffer_destroy(pipe->winsys, st->bitmap.vbuf); + pipe_buffer_destroy(pipe, st->bitmap.vbuf); st->bitmap.vbuf = NULL; } diff --git a/src/mesa/state_tracker/st_cb_blit.c b/src/mesa/state_tracker/st_cb_blit.c index 63211d8b66..0533d9460f 100644 --- a/src/mesa/state_tracker/st_cb_blit.c +++ b/src/mesa/state_tracker/st_cb_blit.c @@ -71,7 +71,6 @@ st_BlitFramebuffer(GLcontext *ctx, GLbitfield mask, GLenum filter) { struct st_context *st = ctx->st; - struct pipe_context *pipe = st->pipe; const uint pFilter = ((filter == GL_NEAREST) ? PIPE_TEX_MIPFILTER_NEAREST @@ -100,10 +99,6 @@ st_BlitFramebuffer(GLcontext *ctx, 0.0, pFilter); } - - /* shaders don't go through CSO yet */ - pipe->bind_fs_state(pipe, st->fp->driver_shader); - pipe->bind_vs_state(pipe, st->vp->driver_shader); } diff --git a/src/mesa/state_tracker/st_cb_bufferobjects.c b/src/mesa/state_tracker/st_cb_bufferobjects.c index fa1254ff7c..af79aefa96 100644 --- a/src/mesa/state_tracker/st_cb_bufferobjects.c +++ b/src/mesa/state_tracker/st_cb_bufferobjects.c @@ -35,7 +35,6 @@ #include "pipe/p_context.h" #include "pipe/p_defines.h" -#include "pipe/p_winsys.h" #include "pipe/p_inlines.h" @@ -79,7 +78,7 @@ st_bufferobj_free(GLcontext *ctx, struct gl_buffer_object *obj) struct st_buffer_object *st_obj = st_buffer_object(obj); if (st_obj->buffer) - pipe_buffer_reference(pipe->winsys, &st_obj->buffer, NULL); + pipe_reference_buffer(pipe, &st_obj->buffer, NULL); free(st_obj); } @@ -106,10 +105,9 @@ st_bufferobj_subdata(GLcontext *ctx, if (offset >= st_obj->size || size > (st_obj->size - offset)) return; - map = pipe->winsys->buffer_map(pipe->winsys, st_obj->buffer, - PIPE_BUFFER_USAGE_CPU_WRITE); + map = pipe_buffer_map(pipe, st_obj->buffer, PIPE_BUFFER_USAGE_CPU_WRITE); memcpy(map + offset, data, size); - pipe->winsys->buffer_unmap(pipe->winsys, st_obj->buffer); + pipe_buffer_unmap(pipe, st_obj->buffer); } @@ -130,10 +128,9 @@ st_bufferobj_get_subdata(GLcontext *ctx, if (offset >= st_obj->size || size > (st_obj->size - offset)) return; - map = pipe->winsys->buffer_map(pipe->winsys, st_obj->buffer, - PIPE_BUFFER_USAGE_CPU_READ); + map = pipe_buffer_map(pipe, st_obj->buffer, PIPE_BUFFER_USAGE_CPU_READ); memcpy(data, map + offset, size); - pipe->winsys->buffer_unmap(pipe->winsys, st_obj->buffer); + pipe_buffer_unmap(pipe, st_obj->buffer); } @@ -174,10 +171,9 @@ st_bufferobj_data(GLcontext *ctx, buffer_usage = 0; } - pipe_buffer_reference( pipe->winsys, &st_obj->buffer, NULL ); + pipe_reference_buffer( pipe, &st_obj->buffer, NULL ); - st_obj->buffer = pipe->winsys->buffer_create( pipe->winsys, 32, buffer_usage, - size ); + st_obj->buffer = pipe_buffer_create( pipe, 32, buffer_usage, size ); st_obj->size = size; @@ -211,7 +207,7 @@ st_bufferobj_map(GLcontext *ctx, GLenum target, GLenum access, break; } - obj->Pointer = pipe->winsys->buffer_map(pipe->winsys, st_obj->buffer, flags); + obj->Pointer = pipe_buffer_map(pipe, st_obj->buffer, flags); return obj->Pointer; } @@ -225,7 +221,7 @@ st_bufferobj_unmap(GLcontext *ctx, GLenum target, struct gl_buffer_object *obj) struct pipe_context *pipe = st_context(ctx)->pipe; struct st_buffer_object *st_obj = st_buffer_object(obj); - pipe->winsys->buffer_unmap(pipe->winsys, st_obj->buffer); + pipe_buffer_unmap(pipe, st_obj->buffer); obj->Pointer = NULL; return GL_TRUE; } diff --git a/src/mesa/state_tracker/st_cb_clear.c b/src/mesa/state_tracker/st_cb_clear.c index fa9f986f11..cdfcdcee72 100644 --- a/src/mesa/state_tracker/st_cb_clear.c +++ b/src/mesa/state_tracker/st_cb_clear.c @@ -34,8 +34,8 @@ #include "main/glheader.h" #include "main/macros.h" #include "shader/prog_instruction.h" -#include "st_atom.h" #include "st_context.h" +#include "st_atom.h" #include "st_cb_accum.h" #include "st_cb_clear.h" #include "st_cb_fbo.h" @@ -45,9 +45,9 @@ #include "st_mesa_to_tgsi.h" #include "pipe/p_context.h" +#include "pipe/p_inlines.h" #include "pipe/p_state.h" #include "pipe/p_defines.h" -#include "pipe/p_winsys.h" #include "util/u_pack_color.h" #include "util/u_simple_shaders.h" #include "util/u_draw_quad.h" @@ -97,16 +97,26 @@ st_destroy_clear(struct st_context *st) { struct pipe_context *pipe = st->pipe; + if (st->clear.vert_shader.tokens) { + FREE((void *) st->clear.vert_shader.tokens); + st->clear.vert_shader.tokens = NULL; + } + + if (st->clear.frag_shader.tokens) { + FREE((void *) st->clear.frag_shader.tokens); + st->clear.frag_shader.tokens = NULL; + } + if (st->clear.fs) { - pipe->delete_fs_state(pipe, st->clear.fs); + cso_delete_fragment_shader(st->cso_context, st->clear.fs); st->clear.fs = NULL; } if (st->clear.vs) { - pipe->delete_vs_state(pipe, st->clear.vs); + cso_delete_vertex_shader(st->cso_context, st->clear.vs); st->clear.vs = NULL; } if (st->clear.vbuf) { - pipe->winsys->buffer_destroy(pipe->winsys, st->clear.vbuf); + pipe_buffer_destroy(pipe, st->clear.vbuf); st->clear.vbuf = NULL; } } @@ -142,9 +152,8 @@ draw_quad(GLcontext *ctx, void *buf; if (!st->clear.vbuf) { - st->clear.vbuf = pipe->winsys->buffer_create(pipe->winsys, 32, - PIPE_BUFFER_USAGE_VERTEX, - sizeof(st->clear.vertices)); + st->clear.vbuf = pipe_buffer_create(pipe, 32, PIPE_BUFFER_USAGE_VERTEX, + sizeof(st->clear.vertices)); } /* positions */ @@ -171,10 +180,9 @@ draw_quad(GLcontext *ctx, } /* put vertex data into vbuf */ - buf = pipe->winsys->buffer_map(pipe->winsys, st->clear.vbuf, - PIPE_BUFFER_USAGE_CPU_WRITE); + buf = pipe_buffer_map(pipe, st->clear.vbuf, PIPE_BUFFER_USAGE_CPU_WRITE); memcpy(buf, st->clear.vertices, sizeof(st->clear.vertices)); - pipe->winsys->buffer_unmap(pipe->winsys, st->clear.vbuf); + pipe_buffer_unmap(pipe, st->clear.vbuf); /* draw */ util_draw_vertex_buffer(pipe, st->clear.vbuf, @@ -195,7 +203,6 @@ clear_with_quad(GLcontext *ctx, GLboolean color, GLboolean depth, GLboolean stencil) { struct st_context *st = ctx->st; - struct pipe_context *pipe = st->pipe; const GLfloat x0 = ctx->DrawBuffer->_Xmin; const GLfloat x1 = ctx->DrawBuffer->_Xmax; GLfloat y0, y1; @@ -222,6 +229,8 @@ clear_with_quad(GLcontext *ctx, cso_save_depth_stencil_alpha(st->cso_context); cso_save_rasterizer(st->cso_context); cso_save_viewport(st->cso_context); + cso_save_fragment_shader(st->cso_context); + cso_save_vertex_shader(st->cso_context); /* blend state: RGBA masking */ { @@ -273,8 +282,8 @@ clear_with_quad(GLcontext *ctx, cso_set_rasterizer(st->cso_context, &st->clear.raster); cso_set_viewport(st->cso_context, &st->clear.viewport); - pipe->bind_fs_state(pipe, st->clear.fs); - pipe->bind_vs_state(pipe, st->clear.vs); + cso_set_fragment_shader_handle(st->cso_context, st->clear.fs); + cso_set_vertex_shader_handle(st->cso_context, st->clear.vs); /* draw quad matching scissor rect (XXX verify coord round-off) */ draw_quad(ctx, x0, y0, x1, y1, ctx->Depth.Clear, ctx->Color.ClearColor); @@ -284,9 +293,8 @@ clear_with_quad(GLcontext *ctx, cso_restore_depth_stencil_alpha(st->cso_context); cso_restore_rasterizer(st->cso_context); cso_restore_viewport(st->cso_context); - /* these don't go through cso yet */ - pipe->bind_fs_state(pipe, st->fp->driver_shader); - pipe->bind_vs_state(pipe, st->vp->driver_shader); + cso_restore_fragment_shader(st->cso_context); + cso_restore_vertex_shader(st->cso_context); } @@ -406,7 +414,7 @@ clear_color_buffer(GLcontext *ctx, struct gl_renderbuffer *rb) /* clear whole buffer w/out masking */ struct st_renderbuffer *strb = st_renderbuffer(rb); uint clearValue; - util_pack_color(ctx->Color.ClearColor, strb->surface->format, &clearValue); + util_pack_color(ctx->Color.ClearColor, PIPE_FORMAT_A8R8G8B8_UNORM, &clearValue); ctx->st->pipe->clear(ctx->st->pipe, strb->surface, clearValue); } } diff --git a/src/mesa/state_tracker/st_cb_drawpixels.c b/src/mesa/state_tracker/st_cb_drawpixels.c index c57e05312a..fbbe8d2d64 100644 --- a/src/mesa/state_tracker/st_cb_drawpixels.c +++ b/src/mesa/state_tracker/st_cb_drawpixels.c @@ -55,7 +55,6 @@ #include "pipe/p_context.h" #include "pipe/p_defines.h" #include "pipe/p_inlines.h" -#include "pipe/p_winsys.h" #include "util/p_tile.h" #include "util/u_draw_quad.h" #include "shader/prog_instruction.h" @@ -165,14 +164,16 @@ static struct st_fragment_program * make_fragment_shader_z(struct st_context *st) { GLcontext *ctx = st->ctx; - /* only make programs once and re-use */ - static struct st_fragment_program *stfp = NULL; struct gl_program *p; GLuint ic = 0; - if (stfp) - return stfp; + if (st->drawpix.z_shader) { + return st->drawpix.z_shader; + } + /* + * Create shader now + */ p = ctx->Driver.NewProgram(ctx, GL_FRAGMENT_PROGRAM_ARB, 0); if (!p) return NULL; @@ -214,10 +215,10 @@ make_fragment_shader_z(struct st_context *st) p->OutputsWritten = (1 << FRAG_RESULT_COLR) | (1 << FRAG_RESULT_DEPR); p->SamplersUsed = 0x1; /* sampler 0 (bit 0) is used */ - stfp = (struct st_fragment_program *) p; - st_translate_fragment_program(st, stfp, NULL); + st->drawpix.z_shader = (struct st_fragment_program *) p; + st_translate_fragment_program(st, st->drawpix.z_shader, NULL); - return stfp; + return st->drawpix.z_shader; } @@ -229,16 +230,17 @@ make_fragment_shader_z(struct st_context *st) static struct st_vertex_program * st_make_passthrough_vertex_shader(struct st_context *st, GLboolean passColor) { - /* only make programs once and re-use */ - static struct st_vertex_program *progs[2] = { NULL, NULL }; GLcontext *ctx = st->ctx; struct st_vertex_program *stvp; struct gl_program *p; GLuint ic = 0; - if (progs[passColor]) - return progs[passColor]; + if (st->drawpix.vert_shaders[passColor]) + return st->drawpix.vert_shaders[passColor]; + /* + * Create shader now + */ p = ctx->Driver.NewProgram(ctx, GL_VERTEX_PROGRAM_ARB, 0); if (!p) return NULL; @@ -294,7 +296,7 @@ st_make_passthrough_vertex_shader(struct st_context *st, GLboolean passColor) stvp = (struct st_vertex_program *) p; st_translate_vertex_program(st, stvp, NULL); - progs[passColor] = stvp; + st->drawpix.vert_shaders[passColor] = stvp; return stvp; } @@ -349,7 +351,7 @@ make_texture(struct st_context *st, pt = st_texture_create(st, PIPE_TEXTURE_2D, pipeFormat, 0, width, height, 1, 0); if (!pt) { - _mesa_unmap_drapix_pbo(ctx, unpack); + _mesa_unmap_drawpix_pbo(ctx, unpack); return NULL; } @@ -395,111 +397,107 @@ make_texture(struct st_context *st, ctx->_ImageTransferState = imageTransferStateSave; } - _mesa_unmap_drapix_pbo(ctx, unpack); + _mesa_unmap_drawpix_pbo(ctx, unpack); return pt; } /** - * Draw textured quad. + * Draw quad with texcoords and optional color. * Coords are window coords with y=0=bottom. + * \param color may be null + * \param invertTex if true, flip texcoords vertically */ static void draw_quad(GLcontext *ctx, GLfloat x0, GLfloat y0, GLfloat z, - GLfloat x1, GLfloat y1, GLboolean invertTex) -{ - GLfloat verts[4][2][4]; /* four verts, two attribs, XYZW */ - GLuint i; - GLfloat sLeft = 0.0, sRight = 1.0; - GLfloat tTop = invertTex, tBot = 1.0 - tTop; - - /* upper-left */ - verts[0][0][0] = x0; /* attr[0].x */ - verts[0][0][1] = y0; /* attr[0].y */ - verts[0][1][0] = sLeft; /* attr[1].s */ - verts[0][1][1] = tTop; /* attr[1].t */ - - /* upper-right */ - verts[1][0][0] = x1; - verts[1][0][1] = y0; - verts[1][1][0] = sRight; - verts[1][1][1] = tTop; - - /* lower-right */ - verts[2][0][0] = x1; - verts[2][0][1] = y1; - verts[2][1][0] = sRight; - verts[2][1][1] = tBot; - - /* lower-left */ - verts[3][0][0] = x0; - verts[3][0][1] = y1; - verts[3][1][0] = sLeft; - verts[3][1][1] = tBot; - - /* same for all verts: */ - for (i = 0; i < 4; i++) { - verts[i][0][2] = z; /*Z*/ - verts[i][0][3] = 1.0; /*W*/ - verts[i][1][2] = 0.0; /*R*/ - verts[i][1][3] = 1.0; /*Q*/ - } - - st_draw_vertices(ctx, PIPE_PRIM_QUADS, 4, (float *) verts, 2, GL_FALSE); -} - - -static void -draw_quad_colored(GLcontext *ctx, GLfloat x0, GLfloat y0, GLfloat z, - GLfloat x1, GLfloat y1, const GLfloat *color, - GLboolean invertTex) + GLfloat x1, GLfloat y1, const GLfloat *color, + GLboolean invertTex) { - GLfloat bias = ctx->st->bitmap_texcoord_bias; + struct st_context *st = ctx->st; + struct pipe_context *pipe = ctx->st->pipe; GLfloat verts[4][3][4]; /* four verts, three attribs, XYZW */ - GLuint i; - GLfloat xBias = bias / (x1-x0); - GLfloat yBias = bias / (y1-y0); - GLfloat sLeft = 0.0 + xBias, sRight = 1.0 + xBias; - GLfloat tTop = invertTex - yBias, tBot = 1.0 - tTop - yBias; - - /* upper-left */ - verts[0][0][0] = x0; /* attr[0].x */ - verts[0][0][1] = y0; /* attr[0].y */ - verts[0][2][0] = sLeft; /* attr[2].s */ - verts[0][2][1] = tTop; /* attr[2].t */ - - /* upper-right */ - verts[1][0][0] = x1; - verts[1][0][1] = y0; - verts[1][2][0] = sRight; - verts[1][2][1] = tTop; - - /* lower-right */ - verts[2][0][0] = x1; - verts[2][0][1] = y1; - verts[2][2][0] = sRight; - verts[2][2][1] = tBot; - - /* lower-left */ - verts[3][0][0] = x0; - verts[3][0][1] = y1; - verts[3][2][0] = sLeft; - verts[3][2][1] = tBot; - - /* same for all verts: */ - for (i = 0; i < 4; i++) { - verts[i][0][2] = z; /*Z*/ - verts[i][0][3] = 1.0; /*W*/ - verts[i][1][0] = color[0]; - verts[i][1][1] = color[1]; - verts[i][1][2] = color[2]; - verts[i][1][3] = color[3]; - verts[i][2][2] = 0.0; /*R*/ - verts[i][2][3] = 1.0; /*Q*/ + + /* setup vertex data */ + { + const struct gl_framebuffer *fb = st->ctx->DrawBuffer; + const GLfloat fb_width = fb->Width; + const GLfloat fb_height = fb->Height; + const GLfloat clip_x0 = x0 / fb_width * 2.0 - 1.0; + const GLfloat clip_y0 = y0 / fb_height * 2.0 - 1.0; + const GLfloat clip_x1 = x1 / fb_width * 2.0 - 1.0; + const GLfloat clip_y1 = y1 / fb_height * 2.0 - 1.0; + const GLfloat sLeft = 0.0F, sRight = 1.0F; + const GLfloat tTop = invertTex, tBot = 1.0 - tTop; + GLuint tex, i; + + /* upper-left */ + verts[0][0][0] = clip_x0; /* v[0].attr[0].x */ + verts[0][0][1] = clip_y0; /* v[0].attr[0].y */ + + /* upper-right */ + verts[1][0][0] = clip_x1; + verts[1][0][1] = clip_y0; + + /* lower-right */ + verts[2][0][0] = clip_x1; + verts[2][0][1] = clip_y1; + + /* lower-left */ + verts[3][0][0] = clip_x0; + verts[3][0][1] = clip_y1; + + tex = color ? 2 : 1; + verts[0][tex][0] = sLeft; /* v[0].attr[tex].s */ + verts[0][tex][1] = tTop; /* v[0].attr[tex].t */ + verts[1][tex][0] = sRight; + verts[1][tex][1] = tTop; + verts[2][tex][0] = sRight; + verts[2][tex][1] = tBot; + verts[3][tex][0] = sLeft; + verts[3][tex][1] = tBot; + + /* same for all verts: */ + if (color) { + for (i = 0; i < 4; i++) { + verts[i][0][2] = z; /*Z*/ + verts[i][0][3] = 1.0; /*W*/ + verts[i][1][0] = color[0]; + verts[i][1][1] = color[1]; + verts[i][1][2] = color[2]; + verts[i][1][3] = color[3]; + verts[i][2][2] = 0.0; /*R*/ + verts[i][2][3] = 1.0; /*Q*/ + } + } + else { + for (i = 0; i < 4; i++) { + verts[i][0][2] = z; /*Z*/ + verts[i][0][3] = 1.0; /*W*/ + verts[i][1][2] = 0.0; /*R*/ + verts[i][1][3] = 1.0; /*Q*/ + } + } } - st_draw_vertices(ctx, PIPE_PRIM_QUADS, 4, (float *) verts, 3, GL_FALSE); + { + struct pipe_buffer *buf; + ubyte *map; + + /* allocate/load buffer object with vertex data */ + buf = pipe_buffer_create(pipe,32, PIPE_BUFFER_USAGE_VERTEX, + sizeof(verts)); + map = pipe_buffer_map(pipe, buf, PIPE_BUFFER_USAGE_CPU_WRITE); + memcpy(map, verts, sizeof(verts)); + pipe_buffer_unmap(pipe, buf); + + util_draw_vertex_buffer(pipe, buf, + PIPE_PRIM_QUADS, + 4, /* verts */ + 3); /* attribs/vert */ + + pipe_buffer_destroy(pipe, buf); + } } @@ -532,6 +530,8 @@ draw_textured_quad(GLcontext *ctx, GLint x, GLint y, GLfloat z, cso_save_viewport(cso); cso_save_samplers(cso); cso_save_sampler_textures(cso); + cso_save_fragment_shader(cso); + cso_save_vertex_shader(cso); /* rasterizer state: just scissor */ { @@ -543,10 +543,10 @@ draw_textured_quad(GLcontext *ctx, GLint x, GLint y, GLfloat z, } /* fragment shader state: TEX lookup program */ - pipe->bind_fs_state(pipe, stfp->driver_shader); + cso_set_fragment_shader_handle(cso, stfp->driver_shader); /* vertex shader state: position + texcoord pass-through */ - pipe->bind_vs_state(pipe, stvp->driver_shader); + cso_set_vertex_shader_handle(cso, stvp->driver_shader); /* texture sampling state: */ @@ -562,6 +562,9 @@ draw_textured_quad(GLcontext *ctx, GLint x, GLint y, GLfloat z, sampler.normalized_coords = 1; cso_single_sampler(cso, 0, &sampler); + if (st->pixel_xfer.pixelmap_enabled) { + cso_single_sampler(cso, 1, &sampler); + } cso_single_sampler_done(cso); } @@ -582,7 +585,15 @@ draw_textured_quad(GLcontext *ctx, GLint x, GLint y, GLfloat z, } /* texture state: */ - pipe->set_sampler_textures(pipe, 1, &pt); + if (st->pixel_xfer.pixelmap_enabled) { + struct pipe_texture *textures[2]; + textures[0] = pt; + textures[1] = st->pixel_xfer.pixelmap_texture; + pipe->set_sampler_textures(pipe, 2, textures); + } + else { + pipe->set_sampler_textures(pipe, 1, &pt); + } /* Compute window coords (y=0=bottom) with pixel zoom. * Recall that these coords are transformed by the current @@ -592,22 +603,15 @@ draw_textured_quad(GLcontext *ctx, GLint x, GLint y, GLfloat z, x1 = x + width * ctx->Pixel.ZoomX; y0 = y; y1 = y + height * ctx->Pixel.ZoomY; - - /* draw textured quad */ - if (color) - draw_quad_colored(ctx, x0, y0, z, x1, y1, color, invertTex); - else - draw_quad(ctx, x0, y0, z, x1, y1, invertTex); + draw_quad(ctx, x0, y0, z, x1, y1, color, invertTex); /* restore state */ cso_restore_rasterizer(cso); cso_restore_viewport(cso); cso_restore_samplers(cso); cso_restore_sampler_textures(cso); - - /* shaders don't go through cso yet */ - pipe->bind_fs_state(pipe, st->fp->driver_shader); - pipe->bind_vs_state(pipe, st->vp->driver_shader); + cso_restore_fragment_shader(cso); + cso_restore_vertex_shader(cso); } @@ -769,7 +773,7 @@ draw_stencil_pixels(GLcontext *ctx, GLint x, GLint y, } switch (ps->format) { - case PIPE_FORMAT_U_S8: + case PIPE_FORMAT_S8_UNORM: { ubyte *dest = stmap + spanY * ps->pitch + spanX; memcpy(dest, values, spanWidth); @@ -867,11 +871,9 @@ copy_stencil_pixels(GLcontext *ctx, GLint srcx, GLint srcy, GLsizei width, GLsizei height, GLint dstx, GLint dsty) { - struct st_renderbuffer *rbRead = st_renderbuffer(ctx->ReadBuffer->_StencilBuffer); struct st_renderbuffer *rbDraw = st_renderbuffer(ctx->DrawBuffer->_StencilBuffer); - struct pipe_surface *psRead = rbRead->surface; struct pipe_surface *psDraw = rbDraw->surface; - ubyte *readMap, *drawMap; + ubyte *drawMap; ubyte *buffer; int i; @@ -881,14 +883,13 @@ copy_stencil_pixels(GLcontext *ctx, GLint srcx, GLint srcy, return; } - /* map the stencil buffers */ - readMap = pipe_surface_map(psRead); - drawMap = pipe_surface_map(psDraw); - /* this will do stencil pixel transfer ops */ st_read_stencil_pixels(ctx, srcx, srcy, width, height, GL_UNSIGNED_BYTE, &ctx->DefaultPacking, buffer); + /* map the stencil buffer */ + drawMap = pipe_surface_map(psDraw); + /* draw */ /* XXX PixelZoom not handled yet */ for (i = 0; i < height; i++) { @@ -916,7 +917,7 @@ copy_stencil_pixels(GLcontext *ctx, GLint srcx, GLint srcy, } } break; - case PIPE_FORMAT_U_S8: + case PIPE_FORMAT_S8_UNORM: memcpy(dst, src, width); break; default: @@ -926,8 +927,7 @@ copy_stencil_pixels(GLcontext *ctx, GLint srcx, GLint srcy, free(buffer); - /* unmap the stencil buffers */ - pipe_surface_unmap(psRead); + /* unmap the stencil buffer */ pipe_surface_unmap(psDraw); } @@ -961,7 +961,7 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy, } if (type == GL_COLOR) { - rbRead = st_renderbuffer(ctx->ReadBuffer->_ColorReadBuffer); + rbRead = st_get_color_read_renderbuffer(ctx); color = NULL; stfp = combined_drawpix_fragment_program(ctx); stvp = st_make_passthrough_vertex_shader(ctx->st, GL_FALSE); @@ -983,22 +983,8 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy, else { /* srcFormat can't be used as a texture format */ if (type == GL_DEPTH) { - static const enum pipe_format zFormats[] = { - PIPE_FORMAT_Z16_UNORM, - PIPE_FORMAT_Z32_UNORM, - PIPE_FORMAT_S8Z24_UNORM, - PIPE_FORMAT_Z24S8_UNORM - }; - uint i; - texFormat = 0; - for (i = 0; i < Elements(zFormats); i++) { - if (screen->is_format_supported(screen, zFormats[i], - PIPE_TEXTURE)) { - texFormat = zFormats[i]; - break; - } - } - assert(texFormat); /* XXX no depth texture formats??? */ + texFormat = st_choose_format(pipe, GL_DEPTH_COMPONENT, PIPE_TEXTURE); + assert(texFormat != PIPE_FORMAT_NONE); /* XXX no depth texture formats??? */ } else { /* todo */ @@ -1017,11 +1003,7 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy, srcy = ctx->DrawBuffer->Height - srcy - height; } - /* For some drivers (like Xlib) it's not possible to treat the - * front/back color buffers as surfaces (they're XImages and Pixmaps). - * So, this var tells us if we can use surface_copy here... - */ - if (st->haveFramebufferSurfaces && srcFormat == texFormat) { + if (srcFormat == texFormat) { /* copy source framebuffer surface into mipmap/texture */ pipe->surface_copy(pipe, FALSE, @@ -1063,3 +1045,15 @@ void st_init_drawpixels_functions(struct dd_function_table *functions) functions->DrawPixels = st_DrawPixels; functions->CopyPixels = st_CopyPixels; } + + +void +st_destroy_drawpix(struct st_context *st) +{ + st_reference_fragprog(st, &st->drawpix.z_shader, NULL); + st_reference_fragprog(st, &st->pixel_xfer.combined_prog, NULL); + st_reference_vertprog(st, &st->drawpix.vert_shaders[0], NULL); + st_reference_vertprog(st, &st->drawpix.vert_shaders[1], NULL); +} + + diff --git a/src/mesa/state_tracker/st_cb_drawpixels.h b/src/mesa/state_tracker/st_cb_drawpixels.h index 71ba487020..26fe864d18 100644 --- a/src/mesa/state_tracker/st_cb_drawpixels.h +++ b/src/mesa/state_tracker/st_cb_drawpixels.h @@ -32,5 +32,8 @@ extern void st_init_drawpixels_functions(struct dd_function_table *functions); +extern void +st_destroy_drawpix(struct st_context *st); + #endif /* ST_CB_DRAWPIXELS_H */ diff --git a/src/mesa/state_tracker/st_cb_fbo.c b/src/mesa/state_tracker/st_cb_fbo.c index ec7788923a..fc8a5ea7f6 100644 --- a/src/mesa/state_tracker/st_cb_fbo.c +++ b/src/mesa/state_tracker/st_cb_fbo.c @@ -80,6 +80,8 @@ init_renderbuffer_bits(struct st_renderbuffer *strb, /** * gl_renderbuffer::AllocStorage() + * This is called to allocate the original drawing surface, and + * during window resize. */ static GLboolean st_renderbuffer_alloc_storage(GLcontext * ctx, struct gl_renderbuffer *rb, @@ -90,8 +92,10 @@ st_renderbuffer_alloc_storage(GLcontext * ctx, struct gl_renderbuffer *rb, struct st_renderbuffer *strb = st_renderbuffer(rb); enum pipe_format pipeFormat; GLbitfield flags = 0x0; /* XXX needed? */ + int ret; if (!strb->surface) { + /* first time surface creation */ strb->surface = pipe->winsys->surface_alloc(pipe->winsys); assert(strb->surface); assert(strb->surface->refcount); @@ -99,10 +103,10 @@ st_renderbuffer_alloc_storage(GLcontext * ctx, struct gl_renderbuffer *rb, if (!strb->surface) return GL_FALSE; } - - if (strb->surface->buffer) - pipe_buffer_reference(pipe->winsys, &strb->surface->buffer, - NULL); + else if (strb->surface->buffer) { + /* release/discard the old surface buffer */ + pipe_reference_buffer(pipe, &strb->surface->buffer, NULL); + } /* Determine surface format here */ if (strb->format != PIPE_FORMAT_NONE) { @@ -116,20 +120,44 @@ st_renderbuffer_alloc_storage(GLcontext * ctx, struct gl_renderbuffer *rb, init_renderbuffer_bits(strb, pipeFormat); - pipe->winsys->surface_alloc_storage(pipe->winsys, - strb->surface, - width, - height, - pipeFormat, - flags); - if (!strb->surface->buffer) - return GL_FALSE; /* out of memory, try s/w buffer? */ + ret = pipe->winsys->surface_alloc_storage(pipe->winsys, + strb->surface, + width, + height, + pipeFormat, + flags); + if (ret || !strb->surface->buffer) { + if (pipeFormat == DEFAULT_ACCUM_PIPE_FORMAT) { + /* Accum buffer. Try a different surface format. Since accum + * buffers are s/w only for now, the surface pixel format doesn't + * really matter, only that the buffer is large enough. + */ + int sz, mult; + enum pipe_format accum_format; + + /* allocate a buffer of (typically) double height to get 64bpp */ + accum_format = st_choose_renderbuffer_format(pipe, GL_RGBA); + sz = pf_get_size(accum_format); + mult = pf_get_size(DEFAULT_ACCUM_PIPE_FORMAT) / sz; + + ret = pipe->winsys->surface_alloc_storage(pipe->winsys, + strb->surface, + width, height * mult, + accum_format, flags); + if (ret) + return GL_FALSE; /* we've _really_ failed */ + + } + else { + return GL_FALSE; /* out of memory, try s/w buffer? */ + } + } ASSERT(strb->surface->buffer); ASSERT(strb->surface->format); ASSERT(strb->surface->cpp); ASSERT(strb->surface->width == width); - ASSERT(strb->surface->height == height); + /*ASSERT(strb->surface->height == height);*/ ASSERT(strb->surface->pitch); strb->Base.Width = width; @@ -224,6 +252,8 @@ st_new_renderbuffer_fb(enum pipe_format format) switch (format) { case PIPE_FORMAT_A8R8G8B8_UNORM: case PIPE_FORMAT_B8G8R8A8_UNORM: + case PIPE_FORMAT_X8R8G8B8_UNORM: + case PIPE_FORMAT_B8G8R8X8_UNORM: case PIPE_FORMAT_A1R5G5B5_UNORM: case PIPE_FORMAT_A4R4G4B4_UNORM: case PIPE_FORMAT_R5G6B5_UNORM: @@ -240,6 +270,8 @@ st_new_renderbuffer_fb(enum pipe_format format) break; case PIPE_FORMAT_S8Z24_UNORM: case PIPE_FORMAT_Z24S8_UNORM: + case PIPE_FORMAT_X8Z24_UNORM: + case PIPE_FORMAT_Z24X8_UNORM: strb->Base.InternalFormat = GL_DEPTH24_STENCIL8_EXT; strb->Base._BaseFormat = GL_DEPTH_STENCIL_EXT; break; @@ -247,7 +279,7 @@ st_new_renderbuffer_fb(enum pipe_format format) strb->Base.InternalFormat = GL_STENCIL_INDEX8_EXT; strb->Base._BaseFormat = GL_STENCIL_INDEX; break; - case PIPE_FORMAT_R16G16B16A16_SNORM: + case DEFAULT_ACCUM_PIPE_FORMAT: /*PIPE_FORMAT_R16G16B16A16_SNORM*/ strb->Base.InternalFormat = GL_RGBA16; strb->Base._BaseFormat = GL_RGBA; break; @@ -338,6 +370,8 @@ st_render_texture(GLcontext *ctx, att->TextureLevel, att->Zoffset); assert(strb->surface); + assert(screen->is_format_supported(screen, strb->surface->format, PIPE_TEXTURE)); + assert(screen->is_format_supported(screen, strb->surface->format, PIPE_SURFACE)); init_renderbuffer_bits(strb, pt->format); @@ -368,6 +402,10 @@ st_finish_render_texture(GLcontext *ctx, ctx->st->pipe->flush(ctx->st->pipe, PIPE_FLUSH_RENDER_CACHE, NULL); + ctx->st->pipe->texture_update(ctx->st->pipe, + st_get_texobj_texture(att->Texture), + att->CubeMapFace, 1 << att->TextureLevel); + /* printf("FINISH RENDER TO TEXTURE surf=%p\n", strb->surface); */ diff --git a/src/mesa/state_tracker/st_cb_fbo.h b/src/mesa/state_tracker/st_cb_fbo.h index 21e531d1d0..c1aa14f9b2 100644 --- a/src/mesa/state_tracker/st_cb_fbo.h +++ b/src/mesa/state_tracker/st_cb_fbo.h @@ -30,6 +30,9 @@ #define ST_CB_FBO_H +#define DEFAULT_ACCUM_PIPE_FORMAT PIPE_FORMAT_R16G16B16A16_SNORM + + /** * Derived renderbuffer class. Just need to add a pointer to the diff --git a/src/mesa/state_tracker/st_cb_feedback.c b/src/mesa/state_tracker/st_cb_feedback.c index 5315294c07..1b50792bd1 100644 --- a/src/mesa/state_tracker/st_cb_feedback.c +++ b/src/mesa/state_tracker/st_cb_feedback.c @@ -56,7 +56,7 @@ #include "cso_cache/cso_cache.h" #include "draw/draw_context.h" -#include "draw/draw_private.h" +#include "draw/draw_pipe.h" /** @@ -238,6 +238,12 @@ select_reset_stipple_counter( struct draw_stage *stage ) /* no-op */ } +static void +select_destroy( struct draw_stage *stage ) +{ + /* no-op */ +} + /** * Create GL selection mode drawing stage. @@ -254,6 +260,7 @@ draw_glselect_stage(GLcontext *ctx, struct draw_context *draw) fs->stage.tri = select_tri; fs->stage.flush = select_flush; fs->stage.reset_stipple_counter = select_reset_stipple_counter; + fs->stage.destroy = select_destroy; fs->ctx = ctx; return &fs->stage; diff --git a/src/mesa/state_tracker/st_cb_flush.c b/src/mesa/state_tracker/st_cb_flush.c index c7efa40e38..5e866b0d41 100644 --- a/src/mesa/state_tracker/st_cb_flush.c +++ b/src/mesa/state_tracker/st_cb_flush.c @@ -44,76 +44,49 @@ #include "pipe/p_winsys.h" -void st_flush( struct st_context *st, uint pipeFlushFlags, - struct pipe_fence_handle **fence ) +static INLINE GLboolean +is_front_buffer_dirty(struct st_context *st) { - FLUSH_VERTICES(st->ctx, 0); - - st_flush_bitmap_cache(st); - - st->pipe->flush( st->pipe, pipeFlushFlags, fence ); + return st->frontbuffer_status == FRONT_STATUS_DIRTY; } -static void st_gl_flush( struct st_context *st, uint pipeFlushFlags, - struct pipe_fence_handle **fence ) +/** + * Tell the winsys to display the front color buffer on-screen. + */ +static void +display_front_buffer(struct st_context *st) { GLframebuffer *fb = st->ctx->DrawBuffer; + struct st_renderbuffer *strb + = st_renderbuffer(fb->Attachment[BUFFER_FRONT_LEFT].Renderbuffer); + struct pipe_surface *front_surf = strb->surface; - st_flush_bitmap_cache(st); - - FLUSH_VERTICES(st->ctx, 0); - - if (!fb) - return; - - /* XXX: temporary hack. This flag should only be set if we do any - * rendering to the front buffer. - * - * Further more, the scissor rectangle could be tracked to - * construct a dirty region of the front buffer, to avoid - * situations where it must be copied repeatedly. - * - * In the extreme case, some kind of timer could be set up to allow - * coalescing of multiple flushes to the frontbuffer, which can be - * quite a performance drain if there are a sufficient number of - * them. + /* Hook for copying "fake" frontbuffer if necessary: */ - st->flags.frontbuffer_dirty - = (fb->_ColorDrawBufferMask[0] & BUFFER_BIT_FRONT_LEFT); - - if (st->flags.frontbuffer_dirty) { - struct st_renderbuffer *strb - = st_renderbuffer(fb->Attachment[BUFFER_FRONT_LEFT].Renderbuffer); - struct pipe_surface *front_surf = strb->surface; - - /* If we aren't rendering to the frontbuffer, this is a noop. - * This should be uncontroversial for glFlush, though people may - * feel more strongly about glFinish. - * - * Additionally, need to make sure that the frontbuffer_dirty - * flag really gets set on frontbuffer rendering. - */ - st->pipe->flush( st->pipe, pipeFlushFlags, fence ); - - /* Hook for copying "fake" frontbuffer if necessary: - */ - st->pipe->winsys->flush_frontbuffer( st->pipe->winsys, front_surf, - st->pipe->priv ); - st->flags.frontbuffer_dirty = 0; - } + st->pipe->winsys->flush_frontbuffer( st->pipe->winsys, front_surf, + st->pipe->priv ); + + /* + st->frontbuffer_status = FRONT_STATUS_UNDEFINED; + */ } -/** - * Called via ctx->Driver.Flush() - */ -static void st_glFlush(GLcontext *ctx) +void st_flush( struct st_context *st, uint pipeFlushFlags, + struct pipe_fence_handle **fence ) { - st_gl_flush(ctx->st, PIPE_FLUSH_RENDER_CACHE, NULL); + FLUSH_VERTICES(st->ctx, 0); + + st_flush_bitmap_cache(st); + + st->pipe->flush( st->pipe, pipeFlushFlags, fence ); } +/** + * Flush, and wait for completion. + */ void st_finish( struct st_context *st ) { struct pipe_fence_handle *fence = NULL; @@ -125,12 +98,36 @@ void st_finish( struct st_context *st ) } + +/** + * Called via ctx->Driver.Flush() + */ +static void st_glFlush(GLcontext *ctx) +{ + struct st_context *st = ctx->st; + + if (is_front_buffer_dirty(st)) { + st_finish(st); + display_front_buffer(st); + } + else { + st_flush(st, PIPE_FLUSH_RENDER_CACHE, NULL); + } +} + + /** * Called via ctx->Driver.Finish() */ static void st_glFinish(GLcontext *ctx) { - st_finish( ctx->st ); + struct st_context *st = ctx->st; + + st_finish(st); + + if (is_front_buffer_dirty(st)) { + display_front_buffer(st); + } } diff --git a/src/mesa/state_tracker/st_cb_program.c b/src/mesa/state_tracker/st_cb_program.c index 003ec0d2b7..5f753afdd3 100644 --- a/src/mesa/state_tracker/st_cb_program.c +++ b/src/mesa/state_tracker/st_cb_program.c @@ -39,7 +39,7 @@ #include "shader/programopt.h" #include "shader/shader_api.h" -#include "cso_cache/cso_cache.h" +#include "cso_cache/cso_context.h" #include "draw/draw_context.h" #include "st_context.h" @@ -89,8 +89,12 @@ static void st_use_program( GLcontext *ctx, +/** + * Called via ctx->Driver.NewProgram() to allocate a new vertex or + * fragment program. + */ static struct gl_program *st_new_program( GLcontext *ctx, - GLenum target, + GLenum target, GLuint id ) { switch (target) { @@ -118,7 +122,8 @@ static struct gl_program *st_new_program( GLcontext *ctx, } default: - return _mesa_new_program(ctx, target, id); + assert(0); + return NULL; } } @@ -127,7 +132,6 @@ void st_delete_program(GLcontext *ctx, struct gl_program *prog) { struct st_context *st = st_context(ctx); - struct pipe_context *pipe = st->pipe; switch( prog->Target ) { case GL_VERTEX_PROGRAM_ARB: @@ -135,10 +139,16 @@ st_delete_program(GLcontext *ctx, struct gl_program *prog) struct st_vertex_program *stvp = (struct st_vertex_program *) prog; if (stvp->driver_shader) { - pipe->delete_vs_state(pipe, stvp->driver_shader); + cso_delete_vertex_shader(st->cso_context, stvp->driver_shader); stvp->driver_shader = NULL; } + if (stvp->draw_shader) { + /* this would only have been allocated for the RasterPos path */ + draw_delete_vertex_shader(st->draw, stvp->draw_shader); + stvp->draw_shader = NULL; + } + if (stvp->state.tokens) { FREE((void *) stvp->state.tokens); stvp->state.tokens = NULL; @@ -150,7 +160,7 @@ st_delete_program(GLcontext *ctx, struct gl_program *prog) struct st_fragment_program *stfp = (struct st_fragment_program *) prog; if (stfp->driver_shader) { - pipe->delete_fs_state(pipe, stfp->driver_shader); + cso_delete_fragment_shader(st->cso_context, stfp->driver_shader); stfp->driver_shader = NULL; } @@ -159,6 +169,12 @@ st_delete_program(GLcontext *ctx, struct gl_program *prog) stfp->state.tokens = NULL; } + if (stfp->bitmap_program) { + struct gl_program *prg = &stfp->bitmap_program->Base.Base; + _mesa_reference_program(ctx, &prg, NULL); + stfp->bitmap_program = NULL; + } + st_free_translated_vertex_programs(st, stfp->vertex_programs); } break; @@ -183,7 +199,6 @@ static void st_program_string_notify( GLcontext *ctx, struct gl_program *prog ) { struct st_context *st = st_context(ctx); - struct pipe_context *pipe = st->pipe; if (target == GL_FRAGMENT_PROGRAM_ARB) { struct st_fragment_program *stfp = (struct st_fragment_program *) prog; @@ -191,7 +206,7 @@ static void st_program_string_notify( GLcontext *ctx, stfp->serialNo++; if (stfp->driver_shader) { - pipe->delete_fs_state(pipe, stfp->driver_shader); + cso_delete_fragment_shader(st->cso_context, stfp->driver_shader); stfp->driver_shader = NULL; } @@ -211,11 +226,12 @@ static void st_program_string_notify( GLcontext *ctx, stvp->serialNo++; if (stvp->driver_shader) { - pipe->delete_vs_state(pipe, stvp->driver_shader); + cso_delete_vertex_shader(st->cso_context, stvp->driver_shader); stvp->driver_shader = NULL; } if (stvp->draw_shader) { + /* this would only have been allocated for the RasterPos path */ draw_delete_vertex_shader(st->draw, stvp->draw_shader); stvp->draw_shader = NULL; } diff --git a/src/mesa/state_tracker/st_cb_queryobj.c b/src/mesa/state_tracker/st_cb_queryobj.c index c1d0d086b4..21c2c7dd9f 100644 --- a/src/mesa/state_tracker/st_cb_queryobj.c +++ b/src/mesa/state_tracker/st_cb_queryobj.c @@ -90,11 +90,7 @@ st_DeleteQuery(GLcontext *ctx, struct gl_query_object *q) FREE(stq); } -/** - * Do glReadPixels by getting rows from the framebuffer surface with - * get_tile(). Convert to requested format/type with Mesa image routines. - * Image transfer ops are done in software too. - */ + static void st_BeginQuery(GLcontext *ctx, struct gl_query_object *q) { diff --git a/src/mesa/state_tracker/st_cb_rasterpos.c b/src/mesa/state_tracker/st_cb_rasterpos.c index 2ed228778e..3cb7b68bea 100644 --- a/src/mesa/state_tracker/st_cb_rasterpos.c +++ b/src/mesa/state_tracker/st_cb_rasterpos.c @@ -46,7 +46,7 @@ #include "st_cb_rasterpos.h" #include "st_draw.h" #include "draw/draw_context.h" -#include "draw/draw_private.h" +#include "draw/draw_pipe.h" #include "shader/prog_instruction.h" #include "vbo/vbo.h" diff --git a/src/mesa/state_tracker/st_cb_readpixels.c b/src/mesa/state_tracker/st_cb_readpixels.c index 2bcc8c99fb..745755566e 100644 --- a/src/mesa/state_tracker/st_cb_readpixels.c +++ b/src/mesa/state_tracker/st_cb_readpixels.c @@ -87,7 +87,7 @@ st_read_stencil_pixels(GLcontext *ctx, GLint x, GLint y, /* get stencil values */ switch (ps->format) { - case PIPE_FORMAT_U_S8: + case PIPE_FORMAT_S8_UNORM: { const ubyte *src = stmap + srcY * ps->pitch + x; memcpy(values, src, width); @@ -129,6 +129,34 @@ st_read_stencil_pixels(GLcontext *ctx, GLint x, GLint y, /** + * Return renderbuffer to use for reading color pixels for glRead/CopyPixel + * commands. + * Special care is needed for the front buffer. + */ +struct st_renderbuffer * +st_get_color_read_renderbuffer(GLcontext *ctx) +{ + struct gl_framebuffer *fb = ctx->ReadBuffer; + struct st_renderbuffer *strb = + st_renderbuffer(fb->_ColorReadBuffer); + struct st_renderbuffer *front = + st_renderbuffer(fb->Attachment[BUFFER_FRONT_LEFT].Renderbuffer); + + if (strb == front + && ctx->st->frontbuffer_status == FRONT_STATUS_COPY_OF_BACK) { + /* reading from front color buffer, which is a logical copy of the + * back color buffer. + */ + struct st_renderbuffer *back = + st_renderbuffer(fb->Attachment[BUFFER_BACK_LEFT].Renderbuffer); + strb = back; + } + + return strb; +} + + +/** * Do glReadPixels by getting rows from the framebuffer surface with * get_tile(). Convert to requested format/type with Mesa image routines. * Image transfer ops are done in software too. @@ -173,12 +201,13 @@ st_readpixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, strb = st_renderbuffer(ctx->ReadBuffer->_DepthBuffer); } else { - strb = st_renderbuffer(ctx->ReadBuffer->_ColorReadBuffer); + /* Read color buffer */ + strb = st_get_color_read_renderbuffer(ctx); } + if (!strb) return; - if (format == GL_RGBA && type == GL_FLOAT) { /* write tile(row) directly into user's buffer */ df = (GLfloat *) _mesa_image_address2d(&clippedPacking, dest, width, @@ -211,7 +240,8 @@ st_readpixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, const GLint dstStride = _mesa_image_row_stride(&clippedPacking, width, format, type); - if (strb->surface->format == PIPE_FORMAT_S8Z24_UNORM) { + if (strb->surface->format == PIPE_FORMAT_S8Z24_UNORM || + strb->surface->format == PIPE_FORMAT_X8Z24_UNORM) { if (format == GL_DEPTH_COMPONENT) { for (i = 0; i < height; i++) { GLuint ztemp[MAX_WIDTH], j; @@ -240,7 +270,7 @@ st_readpixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, } else if (strb->surface->format == PIPE_FORMAT_Z16_UNORM) { for (i = 0; i < height; i++) { - GLshort ztemp[MAX_WIDTH], j; + GLushort ztemp[MAX_WIDTH], j; GLfloat zfloat[MAX_WIDTH]; const double scale = 1.0 / 0xffff; pipe_get_tile_raw(pipe, strb->surface, x, y, width, 1, ztemp, 0); diff --git a/src/mesa/state_tracker/st_cb_readpixels.h b/src/mesa/state_tracker/st_cb_readpixels.h index 79acdad88e..9e151be51f 100644 --- a/src/mesa/state_tracker/st_cb_readpixels.h +++ b/src/mesa/state_tracker/st_cb_readpixels.h @@ -29,6 +29,9 @@ #ifndef ST_CB_READPIXELS_H #define ST_CB_READPIXELS_H +extern struct st_renderbuffer * +st_get_color_read_renderbuffer(GLcontext *ctx); + extern void st_read_stencil_pixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, GLenum type, diff --git a/src/mesa/state_tracker/st_cb_strings.c b/src/mesa/state_tracker/st_cb_strings.c index 247519ab3d..066124f8f3 100644 --- a/src/mesa/state_tracker/st_cb_strings.c +++ b/src/mesa/state_tracker/st_cb_strings.c @@ -53,21 +53,21 @@ st_get_string(GLcontext * ctx, GLenum name) const char *vendor = screen->get_vendor( screen ); const char *tungsten = "Tungsten Graphics, Inc."; - /* Tungsten developed the state_tracker module (and much of - * Mesa), but the driver itself may come from elsewhere. The - * additional string allows "and XyzCorp" to reflect this. + /* Tungsten Graphics, Inc. developed the state_tracker module + * (and much of Mesa), but the driver itself may come from elsewhere. + * The additional string allows "and XyzCorp" to reflect this. */ if (vendor && strcmp(vendor, tungsten) != 0) - snprintf(st->vendor, sizeof(st->vendor), + util_snprintf(st->vendor, sizeof(st->vendor), "%s and %s", tungsten, vendor); else - snprintf(st->vendor, sizeof(st->vendor), "%s", tungsten); + util_snprintf(st->vendor, sizeof(st->vendor), "%s", tungsten); return (GLubyte *) st->vendor; } case GL_RENDERER: - snprintf(st->renderer, sizeof(st->renderer), "Gallium %s, %s on %s", + util_snprintf(st->renderer, sizeof(st->renderer), "Gallium %s, %s on %s", ST_VERSION_STRING, screen->get_name( screen ), screen->winsys->get_name( screen->winsys )); diff --git a/src/mesa/state_tracker/st_cb_texture.c b/src/mesa/state_tracker/st_cb_texture.c index 43b43ec14c..cc17bebc6b 100644 --- a/src/mesa/state_tracker/st_cb_texture.c +++ b/src/mesa/state_tracker/st_cb_texture.c @@ -31,6 +31,7 @@ #include "main/image.h" #include "main/macros.h" #include "main/mipmap.h" +#include "main/pixel.h" #include "main/texcompress.h" #include "main/texformat.h" #include "main/teximage.h" @@ -41,6 +42,7 @@ #include "state_tracker/st_cb_fbo.h" #include "state_tracker/st_cb_texture.h" #include "state_tracker/st_format.h" +#include "state_tracker/st_public.h" #include "state_tracker/st_texture.h" #include "state_tracker/st_gen_mipmap.h" @@ -48,6 +50,7 @@ #include "pipe/p_defines.h" #include "pipe/p_inlines.h" #include "util/p_tile.h" +#include "util/u_blit.h" #define DBG if (0) printf @@ -148,7 +151,7 @@ st_DeleteTextureObject(GLcontext *ctx, { struct st_texture_object *stObj = st_texture_object(texObj); if (stObj->pt) - pipe_texture_release(&stObj->pt); + pipe_texture_reference(&stObj->pt, NULL); _mesa_delete_texture_object(ctx, texObj); } @@ -162,7 +165,7 @@ st_FreeTextureImageData(GLcontext * ctx, struct gl_texture_image *texImage) DBG("%s\n", __FUNCTION__); if (stImage->pt) { - pipe_texture_release(&stImage->pt); + pipe_texture_reference(&stImage->pt, NULL); } if (texImage->Data) { @@ -491,6 +494,10 @@ st_TexImage(GLcontext * ctx, strip_texture_border(border, &width, &height, &depth, unpack, &unpackNB); unpack = &unpackNB; + texImage->Width = width; + texImage->Height = height; + texImage->Depth = depth; + texImage->Border = 0; border = 0; } @@ -537,7 +544,7 @@ st_TexImage(GLcontext * ctx, * Release any old malloced memory. */ if (stImage->pt) { - pipe_texture_release(&stImage->pt); + pipe_texture_reference(&stImage->pt, NULL); assert(!texImage->Data); } else if (texImage->Data) { @@ -549,15 +556,17 @@ st_TexImage(GLcontext * ctx, * waiting on any outstanding fences. */ if (stObj->pt && - /*stObj->pt->first_level == level &&*/ - stObj->pt->last_level == level && - stObj->pt->target != PIPE_TEXTURE_CUBE && - !st_texture_match_image(stObj->pt, &stImage->base, - stImage->face, stImage->level)) { + (stObj->teximage_realloc || + (/*stObj->pt->first_level == level &&*/ + stObj->pt->last_level == level && + stObj->pt->target != PIPE_TEXTURE_CUBE && + !st_texture_match_image(stObj->pt, &stImage->base, + stImage->face, stImage->level)))) { DBG("release it\n"); - pipe_texture_release(&stObj->pt); + pipe_texture_reference(&stObj->pt, NULL); assert(!stObj->pt); + stObj->teximage_realloc = FALSE; } if (!stObj->pt) { @@ -894,6 +903,11 @@ st_TexSubimage(GLcontext * ctx, dstRowStride = stImage->surface->pitch * stImage->surface->cpp; } + if (!texImage->Data) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexSubImage"); + return; + } + src = (const GLubyte *) pixels; for (i = 0; i++ < depth;) { @@ -905,10 +919,11 @@ st_TexSubimage(GLcontext * ctx, texImage->ImageOffsets, width, height, 1, format, type, src, packing)) { - _mesa_error(ctx, GL_OUT_OF_MEMORY, "st_TexSubImage"); + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexSubImage"); } if (stImage->pt && i < depth) { + /* map next slice of 3D texture */ st_texture_image_unmap(stImage); texImage->Data = st_texture_image_map(ctx->st, stImage, zoffset + i); src += srcImageStride; @@ -1030,9 +1045,10 @@ fallback_copy_texsubimage(GLcontext *ctx, const uint face = texture_face(target); struct pipe_texture *pt = stImage->pt; struct pipe_surface *src_surf, *dest_surf; - GLfloat *data; GLint row, yStep; + st_flush(ctx->st, PIPE_FLUSH_RENDER_CACHE, NULL); + /* determine bottom-to-top vs. top-to-bottom order */ if (st_fb_orientation(ctx->ReadBuffer) == Y_0_TOP) { destY = height - 1 - destY; @@ -1046,25 +1062,38 @@ fallback_copy_texsubimage(GLcontext *ctx, dest_surf = screen->get_tex_surface(screen, pt, face, level, destZ); - /* buffer for one row */ - data = (GLfloat *) malloc(width * 4 * sizeof(GLfloat)); - - /* do copy row by row */ - for (row = 0; row < height; row++) { - pipe_get_tile_rgba(pipe, src_surf, srcX, srcY + row, width, 1, data); + assert(width <= MAX_WIDTH); - /* XXX we're ignoring convolution for now */ - if (ctx->_ImageTransferState) { - _mesa_apply_rgba_transfer_ops(ctx, - ctx->_ImageTransferState & ~IMAGE_CONVOLUTION_BIT, - width, (GLfloat (*)[4])data); + /* + * To avoid a large temp memory allocation, do copy row by row. + */ + if (baseFormat == GL_DEPTH_COMPONENT) { + const GLboolean scaleOrBias = (ctx->Pixel.DepthScale != 1.0F || + ctx->Pixel.DepthBias != 0.0F); + + for (row = 0; row < height; row++, srcY++, destY += yStep) { + uint data[MAX_WIDTH]; + pipe_get_tile_z(pipe, src_surf, srcX, srcY, width, 1, data); + if (scaleOrBias) { + _mesa_scale_and_bias_depth_uint(ctx, width, data); + } + pipe_put_tile_z(pipe, dest_surf, destX, destY, width, 1, data); + } + } + else { + /* RGBA format */ + for (row = 0; row < height; row++, srcY++, destY += yStep) { + float data[4 * MAX_WIDTH]; + pipe_get_tile_rgba(pipe, src_surf, srcX, srcY, width, 1, data); + /* XXX we're ignoring convolution for now */ + if (ctx->_ImageTransferState) { + _mesa_apply_rgba_transfer_ops(ctx, + ctx->_ImageTransferState & ~IMAGE_CONVOLUTION_BIT, + width, (GLfloat (*)[4]) data); + } + pipe_put_tile_rgba(pipe, dest_surf, destX, destY, width, 1, data); } - - pipe_put_tile_rgba(pipe, dest_surf, destX, destY, width, 1, data); - destY += yStep; } - - free(data); } @@ -1101,6 +1130,7 @@ do_copy_texsubimage(GLcontext *ctx, struct pipe_surface *dest_surface; uint dest_format, src_format; uint do_flip = FALSE; + GLboolean use_fallback = GL_TRUE; (void) texImage; @@ -1131,11 +1161,9 @@ do_copy_texsubimage(GLcontext *ctx, dest_surface = screen->get_tex_surface(screen, stImage->pt, stImage->face, stImage->level, destZ); - if (src_format == dest_format && - ctx->_ImageTransferState == 0x0 && + if (ctx->_ImageTransferState == 0x0 && strb->surface->buffer && - dest_surface->buffer && - strb->surface->cpp == stImage->pt->cpp) { + dest_surface->buffer) { /* do blit-style copy */ /* XXX may need to invert image depending on window @@ -1159,19 +1187,36 @@ do_copy_texsubimage(GLcontext *ctx, GL_COPY); /* ? */ #else - pipe->surface_copy(pipe, - do_flip, - /* dest */ - dest_surface, - destX, destY, - /* src */ - strb->surface, - srcX, srcY, - /* size */ - width, height); + if (src_format == dest_format) { + pipe->surface_copy(pipe, + do_flip, + /* dest */ + dest_surface, + destX, destY, + /* src */ + strb->surface, + srcX, srcY, + /* size */ + width, height); + use_fallback = GL_FALSE; + } + else if (screen->is_format_supported(screen, strb->surface->format, + PIPE_TEXTURE) && + screen->is_format_supported(screen, dest_surface->format, + PIPE_SURFACE)) { + util_blit_pixels(ctx->st->blit, + strb->surface, + srcX, do_flip ? srcY + height : srcY, + srcX + width, do_flip ? srcY : srcY + height, + dest_surface, + destX, destY, destX + width, destY + height, + 0.0, PIPE_TEX_MIPFILTER_NEAREST); + use_fallback = GL_FALSE; + } #endif } - else { + + if (use_fallback) { fallback_copy_texsubimage(ctx, target, level, strb, stImage, baseFormat, destX, destY, destZ, @@ -1299,8 +1344,6 @@ static void calculate_first_last_level(struct st_texture_object *stObj) { struct gl_texture_object *tObj = &stObj->base; - const struct gl_texture_image *const baseImage = - tObj->Image[0][tObj->BaseLevel]; /* These must be signed values. MinLod and MaxLod can be negative numbers, * and having firstLevel and lastLevel as signed prevents the need for @@ -1323,7 +1366,7 @@ calculate_first_last_level(struct st_texture_object *stObj) } else { firstLevel = 0; - lastLevel = MIN2(tObj->MaxLevel - tObj->BaseLevel, baseImage->MaxLog2); + lastLevel = MIN2(tObj->MaxLevel, tObj->Image[0][tObj->BaseLevel]->WidthLog2); } break; case GL_TEXTURE_RECTANGLE_NV: @@ -1353,9 +1396,9 @@ copy_image_data_to_texture(struct st_context *st, stImage->face ); - pipe_texture_release(&stImage->pt); + pipe_texture_reference(&stImage->pt, NULL); } - else { + else if (stImage->base.Data) { assert(stImage->base.Data != NULL); /* More straightforward upload. @@ -1363,7 +1406,7 @@ copy_image_data_to_texture(struct st_context *st, st_texture_image_data(st->pipe, stObj->pt, stImage->face, - stImage->level, + dstLevel, stImage->base.Data, stImage->base.RowStride, stImage->base.RowStride * @@ -1405,17 +1448,6 @@ st_finalize_texture(GLcontext *ctx, calculate_first_last_level(stObj); firstImage = st_texture_image(stObj->base.Image[0][stObj->base.BaseLevel]); -#if 0 - /* Fallback case: - */ - if (firstImage->base.Border) { - if (stObj->pt) { - pipe_texture_release(&stObj->pt); - } - return GL_FALSE; - } -#endif - /* If both firstImage and stObj point to a texture which can contain * all active images, favour firstImage. Note that because of the * completeness requirement, we know that the image dimensions @@ -1425,9 +1457,6 @@ st_finalize_texture(GLcontext *ctx, firstImage->pt != stObj->pt && firstImage->pt->last_level >= stObj->lastLevel) { - if (stObj->pt) - pipe_texture_release(&stObj->pt); - pipe_texture_reference(&stObj->pt, firstImage->pt); } @@ -1439,24 +1468,25 @@ st_finalize_texture(GLcontext *ctx, cpp = firstImage->base.TexFormat->TexelBytes; } - /* Check texture can hold all active levels. Check texture matches - * target, imageFormat, etc. + /* If we already have a gallium texture, check that it matches the texture + * object's format, target, size, num_levels, etc. */ - if (stObj->pt && - (stObj->pt->target != gl_target_to_pipe(stObj->base.Target) || - stObj->pt->format != - st_mesa_format_to_pipe_format(firstImage->base.TexFormat->MesaFormat) || - stObj->pt->last_level != stObj->lastLevel || - stObj->pt->width[0] != firstImage->base.Width2 || - stObj->pt->height[0] != firstImage->base.Height2 || - stObj->pt->depth[0] != firstImage->base.Depth2 || - stObj->pt->cpp != cpp || - stObj->pt->compressed != firstImage->base.IsCompressed)) { - pipe_texture_release(&stObj->pt); + if (stObj->pt) { + const enum pipe_format fmt = + st_mesa_format_to_pipe_format(firstImage->base.TexFormat->MesaFormat); + if (stObj->pt->target != gl_target_to_pipe(stObj->base.Target) || + stObj->pt->format != fmt || + stObj->pt->last_level < stObj->lastLevel || + stObj->pt->width[0] != firstImage->base.Width2 || + stObj->pt->height[0] != firstImage->base.Height2 || + stObj->pt->depth[0] != firstImage->base.Depth2 || + stObj->pt->cpp != cpp || + stObj->pt->compressed != firstImage->base.IsCompressed) { + pipe_texture_release(&stObj->pt); + } } - - /* May need to create a new texture: + /* May need to create a new gallium texture: */ if (!stObj->pt) { stObj->pt = st_texture_create(ctx->st, diff --git a/src/mesa/state_tracker/st_context.c b/src/mesa/state_tracker/st_context.c index 7511c28074..4081722bc7 100644 --- a/src/mesa/state_tracker/st_context.c +++ b/src/mesa/state_tracker/st_context.c @@ -55,7 +55,6 @@ #include "st_gen_mipmap.h" #include "st_program.h" #include "pipe/p_context.h" -#include "pipe/p_winsys.h" #include "pipe/p_inlines.h" #include "draw/draw_context.h" #include "cso_cache/cso_cache.h" @@ -128,8 +127,6 @@ st_create_context_priv( GLcontext *ctx, struct pipe_context *pipe ) st->ctx->VertexProgram._MaintainTnlProgram = GL_TRUE; - st->haveFramebufferSurfaces = GL_TRUE; - st->pixel_xfer.cache = _mesa_new_program_cache(); /* GL limits and extensions */ @@ -159,7 +156,6 @@ struct st_context *st_create_context(struct pipe_context *pipe, static void st_destroy_context_priv( struct st_context *st ) { - struct pipe_winsys *ws = st->pipe->winsys; uint i; draw_destroy(st->draw); @@ -169,29 +165,46 @@ static void st_destroy_context_priv( struct st_context *st ) st_destroy_bitmap(st); st_destroy_blit(st); st_destroy_clear(st); + st_destroy_drawpix(st); _vbo_DestroyContext(st->ctx); - cso_destroy_context(st->cso_context); - - _mesa_delete_program_cache(st->ctx, st->pixel_xfer.cache); + for (i = 0; i < Elements(st->state.sampler_texture); i++) { + pipe_texture_reference(&st->state.sampler_texture[i], NULL); + } for (i = 0; i < Elements(st->state.constants); i++) { if (st->state.constants[i].buffer) { - pipe_buffer_reference(ws, &st->state.constants[i].buffer, NULL); + pipe_reference_buffer(st->pipe, &st->state.constants[i].buffer, NULL); } } - st->pipe->destroy( st->pipe ); free( st ); } void st_destroy_context( struct st_context *st ) { + struct pipe_context *pipe = st->pipe; + struct cso_context *cso = st->cso_context; GLcontext *ctx = st->ctx; + + /* need to unbind and destroy CSO objects before anything else */ + cso_release_all(st->cso_context); + + st_reference_fragprog(st, &st->fp, NULL); + st_reference_vertprog(st, &st->vp, NULL); + + _mesa_delete_program_cache(st->ctx, st->pixel_xfer.cache); + _mesa_free_context_data(ctx); + st_destroy_context_priv(st); + + cso_destroy_context(cso); + + pipe->destroy( pipe ); + free(ctx); } diff --git a/src/mesa/state_tracker/st_context.h b/src/mesa/state_tracker/st_context.h index bcebbd4943..46c16e45e6 100644 --- a/src/mesa/state_tracker/st_context.h +++ b/src/mesa/state_tracker/st_context.h @@ -45,6 +45,11 @@ struct blit_state; struct bitmap_cache; +#define FRONT_STATUS_UNDEFINED 0 +#define FRONT_STATUS_DIRTY 1 +#define FRONT_STATUS_COPY_OF_BACK 2 + + #define ST_NEW_MESA 0x1 /* Mesa state has changed */ #define ST_NEW_FRAGMENT_PROGRAM 0x2 #define ST_NEW_VERTEX_PROGRAM 0x4 @@ -104,19 +109,11 @@ struct st_context struct gl_fragment_program *fragment_program; } cb; - struct { - GLuint frontbuffer_dirty:1; - } flags; + GLuint frontbuffer_status; /**< one of FRONT_STATUS_ */ char vendor[100]; char renderer[100]; - /** Can we access the front/back color buffers as pipe_surfaces? - * We can't with the Xlib driver... - * This is a hack that should be fixed someday. - */ - GLboolean haveFramebufferSurfaces; - /* State to be validated: */ struct st_tracked_state **atoms; @@ -126,8 +123,6 @@ struct st_context GLfloat polygon_offset_scale; /* ?? */ - GLfloat bitmap_texcoord_bias; - /** Mapping from VERT_RESULT_x to post-transformed vertex slot */ const GLuint *vertex_result_to_slot; @@ -141,22 +136,28 @@ struct st_context GLuint user_prog_sn; /**< user fragment program serial no. */ struct st_fragment_program *combined_prog; GLuint combined_prog_sn; + struct pipe_texture *pixelmap_texture; + boolean pixelmap_enabled; /**< use the pixelmap texture? */ } pixel_xfer; /** for glBitmap */ struct { - struct st_fragment_program *program; /**< bitmap tex/kil program */ - GLuint user_prog_sn; /**< user fragment program serial no. */ - struct st_fragment_program *combined_prog; struct pipe_rasterizer_state rasterizer; struct pipe_sampler_state sampler; struct pipe_shader_state vert_shader; + enum pipe_format tex_format; void *vs; float vertices[4][3][4]; /**< vertex pos + color + texcoord */ struct pipe_buffer *vbuf; struct bitmap_cache *cache; } bitmap; + /** for glDraw/CopyPixels */ + struct { + struct st_fragment_program *z_shader; + struct st_vertex_program *vert_shaders[2]; + } drawpix; + /** for glClear */ struct { struct pipe_shader_state vert_shader; @@ -169,6 +170,8 @@ struct st_context struct pipe_buffer *vbuf; } clear; + void *passthrough_fs; /**< simple pass-through frag shader */ + struct gen_mipmap_state *gen_mipmap; struct blit_state *blit; diff --git a/src/mesa/state_tracker/st_draw.c b/src/mesa/state_tracker/st_draw.c index f0f62246dd..a3bffbfc95 100644 --- a/src/mesa/state_tracker/st_draw.c +++ b/src/mesa/state_tracker/st_draw.c @@ -35,15 +35,14 @@ #include "vbo/vbo.h" -#include "st_atom.h" #include "st_context.h" +#include "st_atom.h" #include "st_cb_bufferobjects.h" #include "st_draw.h" #include "st_program.h" #include "pipe/p_context.h" #include "pipe/p_defines.h" -#include "pipe/p_winsys.h" #include "pipe/p_inlines.h" #include "draw/draw_private.h" @@ -190,6 +189,60 @@ pipe_vertex_format(GLenum type, GLuint size, GLboolean normalized) } +/* + * If edge flags are needed, setup an bitvector of flags and call + * pipe->set_edgeflags(). + * XXX memleak: need to free the returned pointer at some point + */ +static void * +setup_edgeflags(GLcontext *ctx, GLenum primMode, GLint start, GLint count, + const struct gl_client_array *array) +{ + struct pipe_context *pipe = ctx->st->pipe; + + if ((primMode == GL_TRIANGLES || + primMode == GL_QUADS || + primMode == GL_POLYGON) && + (ctx->Polygon.FrontMode != GL_FILL || + ctx->Polygon.BackMode != GL_FILL)) { + /* need edge flags */ + GLuint i; + unsigned *vec; + struct st_buffer_object *stobj = st_buffer_object(array->BufferObj); + ubyte *map; + + if (!stobj) + return NULL; + + vec = (unsigned *) calloc(sizeof(unsigned), (count + 31) / 32); + if (!vec) + return NULL; + + map = pipe_buffer_map(pipe, stobj->buffer, PIPE_BUFFER_USAGE_CPU_READ); + map = ADD_POINTERS(map, array->Ptr); + + for (i = 0; i < count; i++) { + if (*((float *) map)) + vec[i/32] |= 1 << (i % 32); + + map += array->StrideB; + } + + pipe_buffer_unmap(pipe, stobj->buffer); + + pipe->set_edgeflags(pipe, vec); + + return vec; + } + else { + /* edge flags not needed */ + pipe->set_edgeflags(pipe, NULL); + return NULL; + } +} + + + /** * This function gets plugged into the VBO module and is called when * we have something to render. @@ -205,7 +258,6 @@ st_draw_vbo(GLcontext *ctx, GLuint max_index) { struct pipe_context *pipe = ctx->st->pipe; - struct pipe_winsys *winsys = pipe->winsys; const struct st_vertex_program *vp; const struct pipe_shader_state *vs; struct pipe_vertex_buffer vbuffer[PIPE_MAX_SHADER_INPUTS]; @@ -237,7 +289,7 @@ st_draw_vbo(GLcontext *ctx, assert(stobj->buffer); vbuffer[attr].buffer = NULL; - pipe_buffer_reference(winsys, &vbuffer[attr].buffer, stobj->buffer); + pipe_reference_buffer(pipe, &vbuffer[attr].buffer, stobj->buffer); vbuffer[attr].buffer_offset = (unsigned) arrays[0]->Ptr;/* in bytes */ velements[attr].src_offset = arrays[mesaAttr]->Ptr - arrays[0]->Ptr; assert(velements[attr].src_offset <= 2048); /* 11-bit field */ @@ -255,9 +307,8 @@ st_draw_vbo(GLcontext *ctx, /* wrap user data */ vbuffer[attr].buffer - = winsys->user_buffer_create(winsys, - (void *) arrays[mesaAttr]->Ptr, - bytes); + = pipe_user_buffer_create(pipe, (void *) arrays[mesaAttr]->Ptr, + bytes); vbuffer[attr].buffer_offset = 0; velements[attr].src_offset = 0; } @@ -277,7 +328,6 @@ st_draw_vbo(GLcontext *ctx, pipe->set_vertex_buffers(pipe, vp->num_inputs, vbuffer); pipe->set_vertex_elements(pipe, vp->num_inputs, velements); - /* do actual drawing */ if (ib) { /* indexed primitive */ @@ -304,37 +354,44 @@ st_draw_vbo(GLcontext *ctx, if (bufobj && bufobj->Name) { /* elements/indexes are in a real VBO */ struct st_buffer_object *stobj = st_buffer_object(bufobj); - pipe_buffer_reference(winsys, &indexBuf, stobj->buffer); + pipe_reference_buffer(pipe, &indexBuf, stobj->buffer); indexOffset = (unsigned) ib->ptr / indexSize; } else { /* element/indicies are in user space memory */ - indexBuf = winsys->user_buffer_create(winsys, - (void *) ib->ptr, - ib->count * indexSize); + indexBuf = pipe_user_buffer_create(pipe, (void *) ib->ptr, + ib->count * indexSize); indexOffset = 0; } /* draw */ for (i = 0; i < nr_prims; i++) { + setup_edgeflags(ctx, prims[i].mode, + prims[i].start + indexOffset, prims[i].count, + arrays[VERT_ATTRIB_EDGEFLAG]); + pipe->draw_elements(pipe, indexBuf, indexSize, prims[i].mode, prims[i].start + indexOffset, prims[i].count); } - pipe_buffer_reference(winsys, &indexBuf, NULL); + pipe_reference_buffer(pipe, &indexBuf, NULL); } else { /* non-indexed */ GLuint i; for (i = 0; i < nr_prims; i++) { + setup_edgeflags(ctx, prims[i].mode, + prims[i].start, prims[i].count, + arrays[VERT_ATTRIB_EDGEFLAG]); + pipe->draw_arrays(pipe, prims[i].mode, prims[i].start, prims[i].count); } } /* unreference buffers (frees wrapped user-space buffer objects) */ for (attr = 0; attr < vp->num_inputs; attr++) { - pipe_buffer_reference(winsys, &vbuffer[attr].buffer, NULL); + pipe_reference_buffer(pipe, &vbuffer[attr].buffer, NULL); assert(!vbuffer[attr].buffer); } pipe->set_vertex_buffers(pipe, vp->num_inputs, vbuffer); @@ -343,76 +400,6 @@ st_draw_vbo(GLcontext *ctx, /** - * Utility function for drawing simple primitives (such as quads for - * glClear and glDrawPixels). Coordinates are in screen space. - * \param mode one of PIPE_PRIM_x - * \param numVertex number of vertices - * \param verts vertex data (all attributes are float[4]) - * \param numAttribs number of attributes per vertex - */ -void -st_draw_vertices(GLcontext *ctx, unsigned prim, - unsigned numVertex, float *verts, - unsigned numAttribs, - GLboolean inClipCoords) -{ - const float width = ctx->DrawBuffer->Width; - const float height = ctx->DrawBuffer->Height; - const unsigned vertex_bytes = numVertex * numAttribs * 4 * sizeof(float); - struct pipe_context *pipe = ctx->st->pipe; - struct pipe_buffer *vbuf; - struct pipe_vertex_buffer vbuffer; - struct pipe_vertex_element velements[PIPE_MAX_ATTRIBS]; - unsigned i; - - assert(numAttribs > 0); - - if (!inClipCoords) { - /* convert to clip coords */ - for (i = 0; i < numVertex; i++) { - float x = verts[i * numAttribs * 4 + 0]; - float y = verts[i * numAttribs * 4 + 1]; - x = x / width * 2.0 - 1.0; - y = y / height * 2.0 - 1.0; - verts[i * numAttribs * 4 + 0] = x; - verts[i * numAttribs * 4 + 1] = y; - } - } - - /* XXX create one-time */ - vbuf = pipe->winsys->buffer_create(pipe->winsys, 32, - PIPE_BUFFER_USAGE_VERTEX, vertex_bytes); - assert(vbuf); - - memcpy(pipe->winsys->buffer_map(pipe->winsys, vbuf, - PIPE_BUFFER_USAGE_CPU_WRITE), - verts, vertex_bytes); - pipe->winsys->buffer_unmap(pipe->winsys, vbuf); - - /* tell pipe about the vertex buffer */ - vbuffer.buffer = vbuf; - vbuffer.pitch = numAttribs * 4 * sizeof(float); /* vertex size */ - vbuffer.buffer_offset = 0; - pipe->set_vertex_buffers(pipe, 1, &vbuffer); - - /* tell pipe about the vertex attributes */ - for (i = 0; i < numAttribs; i++) { - velements[i].src_offset = i * 4 * sizeof(GLfloat); - velements[i].vertex_buffer_index = 0; - velements[i].src_format = PIPE_FORMAT_R32G32B32A32_FLOAT; - velements[i].nr_components = 4; - } - pipe->set_vertex_elements(pipe, numAttribs, velements); - - /* draw */ - pipe->draw_arrays(pipe, prim, 0, numVertex); - - /* XXX: do one-time */ - pipe_buffer_reference(pipe->winsys, &vbuf, NULL); -} - - -/** * Set the (private) draw module's post-transformed vertex format when in * GL_SELECT or GL_FEEDBACK mode or for glRasterPos. */ @@ -466,7 +453,6 @@ st_feedback_draw_vbo(GLcontext *ctx, struct st_context *st = ctx->st; struct pipe_context *pipe = st->pipe; struct draw_context *draw = st->draw; - struct pipe_winsys *winsys = pipe->winsys; const struct st_vertex_program *vp; const struct pipe_shader_state *vs; struct pipe_buffer *index_buffer_handle = 0; @@ -517,7 +503,7 @@ st_feedback_draw_vbo(GLcontext *ctx, assert(stobj->buffer); vbuffers[attr].buffer = NULL; - pipe_buffer_reference(winsys, &vbuffers[attr].buffer, stobj->buffer); + pipe_reference_buffer(pipe, &vbuffers[attr].buffer, stobj->buffer); vbuffers[attr].buffer_offset = (unsigned) arrays[0]->Ptr;/* in bytes */ velements[attr].src_offset = arrays[mesaAttr]->Ptr - arrays[0]->Ptr; } @@ -529,9 +515,8 @@ st_feedback_draw_vbo(GLcontext *ctx, /* wrap user data */ vbuffers[attr].buffer - = winsys->user_buffer_create(winsys, - (void *) arrays[mesaAttr]->Ptr, - bytes); + = pipe_user_buffer_create(pipe, (void *) arrays[mesaAttr]->Ptr, + bytes); vbuffers[attr].buffer_offset = 0; velements[attr].src_offset = 0; } @@ -552,9 +537,8 @@ st_feedback_draw_vbo(GLcontext *ctx, #endif /* map the attrib buffer */ - map = pipe->winsys->buffer_map(pipe->winsys, - vbuffers[attr].buffer, - PIPE_BUFFER_USAGE_CPU_READ); + map = pipe_buffer_map(pipe, vbuffers[attr].buffer, + PIPE_BUFFER_USAGE_CPU_READ); draw_set_mapped_vertex_buffer(draw, attr, map); } @@ -565,9 +549,10 @@ st_feedback_draw_vbo(GLcontext *ctx, unsigned indexSize; struct gl_buffer_object *bufobj = ib->obj; struct st_buffer_object *stobj = st_buffer_object(bufobj); - index_buffer_handle = stobj->buffer; void *map; + index_buffer_handle = stobj->buffer; + switch (ib->type) { case GL_UNSIGNED_INT: indexSize = 4; @@ -580,9 +565,8 @@ st_feedback_draw_vbo(GLcontext *ctx, return; } - map = pipe->winsys->buffer_map(pipe->winsys, - index_buffer_handle, - PIPE_BUFFER_USAGE_CPU_READ); + map = pipe_buffer_map(pipe, index_buffer_handle, + PIPE_BUFFER_USAGE_CPU_READ); draw_set_mapped_element_buffer(draw, indexSize, map); } else { @@ -592,7 +576,7 @@ st_feedback_draw_vbo(GLcontext *ctx, /* map constant buffers */ - mapped_constants = winsys->buffer_map(winsys, + mapped_constants = pipe_buffer_map(pipe, st->state.constants[PIPE_SHADER_VERTEX].buffer, PIPE_BUFFER_USAGE_CPU_READ); draw_set_mapped_constant_buffer(st->draw, mapped_constants); @@ -605,21 +589,20 @@ st_feedback_draw_vbo(GLcontext *ctx, /* unmap constant buffers */ - winsys->buffer_unmap(winsys, st->state.constants[PIPE_SHADER_VERTEX].buffer); + pipe_buffer_unmap(pipe, st->state.constants[PIPE_SHADER_VERTEX].buffer); /* * unmap vertex/index buffers */ for (i = 0; i < PIPE_MAX_ATTRIBS; i++) { - if (draw->vertex_buffer[i].buffer) { - pipe->winsys->buffer_unmap(pipe->winsys, - draw->vertex_buffer[i].buffer); - pipe_buffer_reference(winsys, &draw->vertex_buffer[i].buffer, NULL); + if (draw->pt.vertex_buffer[i].buffer) { + pipe_buffer_unmap(pipe, draw->pt.vertex_buffer[i].buffer); + pipe_reference_buffer(pipe, &draw->pt.vertex_buffer[i].buffer, NULL); draw_set_mapped_vertex_buffer(draw, i, NULL); } } if (ib) { - pipe->winsys->buffer_unmap(pipe->winsys, index_buffer_handle); + pipe_buffer_unmap(pipe, index_buffer_handle); draw_set_mapped_element_buffer(draw, 0, NULL); } } diff --git a/src/mesa/state_tracker/st_draw.h b/src/mesa/state_tracker/st_draw.h index 171bde57e5..c81f2b25da 100644 --- a/src/mesa/state_tracker/st_draw.h +++ b/src/mesa/state_tracker/st_draw.h @@ -59,11 +59,4 @@ st_feedback_draw_vbo(GLcontext *ctx, GLuint min_index, GLuint max_index); -void -st_draw_vertices(GLcontext *ctx, unsigned prim, - unsigned numVertex, float *verts, - unsigned numAttribs, - GLboolean inClipCoords); - - #endif diff --git a/src/mesa/state_tracker/st_extensions.c b/src/mesa/state_tracker/st_extensions.c index 47a50d40ca..6f94ba39ae 100644 --- a/src/mesa/state_tracker/st_extensions.c +++ b/src/mesa/state_tracker/st_extensions.c @@ -38,17 +38,17 @@ #include "st_extensions.h" -static int min(int a, int b) +static int _min(int a, int b) { return (a < b) ? a : b; } -static int max(int a, int b) +static int _max(int a, int b) { return (a > b) ? a : b; } -static int clamp(int a, int min, int max) +static int _clamp(int a, int min, int max) { if (a < min) return min; @@ -69,48 +69,49 @@ void st_init_limits(struct st_context *st) struct gl_constants *c = &st->ctx->Const; c->MaxTextureLevels - = min(screen->get_param(screen, PIPE_CAP_MAX_TEXTURE_2D_LEVELS), + = _min(screen->get_param(screen, PIPE_CAP_MAX_TEXTURE_2D_LEVELS), MAX_TEXTURE_LEVELS); c->Max3DTextureLevels - = min(screen->get_param(screen, PIPE_CAP_MAX_TEXTURE_3D_LEVELS), + = _min(screen->get_param(screen, PIPE_CAP_MAX_TEXTURE_3D_LEVELS), MAX_3D_TEXTURE_LEVELS); c->MaxCubeTextureLevels - = min(screen->get_param(screen, PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS), + = _min(screen->get_param(screen, PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS), MAX_CUBE_TEXTURE_LEVELS); c->MaxTextureRectSize - = min(1 << (c->MaxTextureLevels - 1), MAX_TEXTURE_RECT_SIZE); + = _min(1 << (c->MaxTextureLevels - 1), MAX_TEXTURE_RECT_SIZE); c->MaxTextureUnits = c->MaxTextureImageUnits = c->MaxTextureCoordUnits - = min(screen->get_param(screen, PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS), + = _min(screen->get_param(screen, PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS), MAX_TEXTURE_IMAGE_UNITS); c->MaxDrawBuffers - = clamp(screen->get_param(screen, PIPE_CAP_MAX_RENDER_TARGETS), + = _clamp(screen->get_param(screen, PIPE_CAP_MAX_RENDER_TARGETS), 1, MAX_DRAW_BUFFERS); c->MaxLineWidth - = max(1.0, screen->get_paramf(screen, PIPE_CAP_MAX_LINE_WIDTH)); + = _max(1.0, screen->get_paramf(screen, PIPE_CAP_MAX_LINE_WIDTH)); c->MaxLineWidthAA - = max(1.0, screen->get_paramf(screen, PIPE_CAP_MAX_LINE_WIDTH_AA)); + = _max(1.0, screen->get_paramf(screen, PIPE_CAP_MAX_LINE_WIDTH_AA)); c->MaxPointSize - = max(1.0, screen->get_paramf(screen, PIPE_CAP_MAX_POINT_WIDTH)); + = _max(1.0, screen->get_paramf(screen, PIPE_CAP_MAX_POINT_WIDTH)); c->MaxPointSizeAA - = max(1.0, screen->get_paramf(screen, PIPE_CAP_MAX_POINT_WIDTH_AA)); + = _max(1.0, screen->get_paramf(screen, PIPE_CAP_MAX_POINT_WIDTH_AA)); c->MaxTextureMaxAnisotropy - = max(2.0, screen->get_paramf(screen, PIPE_CAP_MAX_TEXTURE_ANISOTROPY)); + = _max(2.0, screen->get_paramf(screen, PIPE_CAP_MAX_TEXTURE_ANISOTROPY)); c->MaxTextureLodBias = screen->get_paramf(screen, PIPE_CAP_MAX_TEXTURE_LOD_BIAS); - st->bitmap_texcoord_bias - = screen->get_paramf(screen, PIPE_CAP_BITMAP_TEXCOORD_BIAS); + c->MaxDrawBuffers + = CLAMP(screen->get_param(screen, PIPE_CAP_MAX_RENDER_TARGETS), + 1, MAX_DRAW_BUFFERS); } @@ -160,6 +161,7 @@ void st_init_extensions(struct st_context *st) ctx->Extensions.NV_blend_square = GL_TRUE; ctx->Extensions.NV_texgen_reflection = GL_TRUE; + ctx->Extensions.SGI_color_matrix = GL_TRUE; ctx->Extensions.SGIS_generate_mipmap = GL_TRUE; /* XXX temp */ /* @@ -215,11 +217,12 @@ void st_init_extensions(struct st_context *st) ctx->Extensions.EXT_texture_sRGB = GL_TRUE; } +#if 01 if (screen->is_format_supported(screen, PIPE_FORMAT_DXT5_RGBA, PIPE_TEXTURE)) { ctx->Extensions.EXT_texture_compression_s3tc = GL_TRUE; } - +#endif if (screen->is_format_supported(screen, PIPE_FORMAT_YCBCR, PIPE_TEXTURE) || screen->is_format_supported(screen, PIPE_FORMAT_YCBCR_REV, PIPE_TEXTURE)) { ctx->Extensions.MESA_ycbcr_texture = GL_TRUE; diff --git a/src/mesa/state_tracker/st_format.c b/src/mesa/state_tracker/st_format.c index 20defd5e16..17a3cfd5a4 100644 --- a/src/mesa/state_tracker/st_format.c +++ b/src/mesa/state_tracker/st_format.c @@ -261,13 +261,13 @@ st_mesa_format_to_pipe_format(GLuint mesaFormat) case MESA_FORMAT_RGB565: return PIPE_FORMAT_R5G6B5_UNORM; case MESA_FORMAT_AL88: - return PIPE_FORMAT_U_A8_L8; + return PIPE_FORMAT_A8L8_UNORM; case MESA_FORMAT_A8: - return PIPE_FORMAT_U_A8; + return PIPE_FORMAT_A8_UNORM; case MESA_FORMAT_L8: - return PIPE_FORMAT_U_L8; + return PIPE_FORMAT_L8_UNORM; case MESA_FORMAT_I8: - return PIPE_FORMAT_U_I8; + return PIPE_FORMAT_I8_UNORM; case MESA_FORMAT_Z16: return PIPE_FORMAT_Z16_UNORM; case MESA_FORMAT_Z32: @@ -293,7 +293,7 @@ st_mesa_format_to_pipe_format(GLuint mesaFormat) /** * Find an RGBA format supported by the context/winsys. */ -static GLuint +static enum pipe_format default_rgba_format(struct pipe_screen *screen, uint type) { static const enum pipe_format colorFormats[] = { @@ -315,20 +315,23 @@ default_rgba_format(struct pipe_screen *screen, uint type) /** * Search list of formats for first RGBA format with >8 bits/channel. */ -static GLuint +static enum pipe_format default_deep_rgba_format(struct pipe_screen *screen, uint type) { if (screen->is_format_supported(screen, PIPE_FORMAT_R16G16B16A16_SNORM, type)) { return PIPE_FORMAT_R16G16B16A16_SNORM; } - return PIPE_FORMAT_NONE; + if (type == PIPE_TEXTURE) + return default_rgba_format(screen, type); + else + return PIPE_FORMAT_NONE; } /** * Find an Z format supported by the context/winsys. */ -static GLuint +static enum pipe_format default_depth_format(struct pipe_screen *screen, uint type) { static const enum pipe_format zFormats[] = { @@ -351,8 +354,9 @@ default_depth_format(struct pipe_screen *screen, uint type) * Given an OpenGL internalFormat value for a texture or surface, return * the best matching PIPE_FORMAT_x, or PIPE_FORMAT_NONE if there's no match. */ -static enum pipe_format -choose_format(struct pipe_context *pipe, GLint internalFormat, uint surfType) +enum pipe_format +st_choose_format(struct pipe_context *pipe, GLint internalFormat, + uint surfType) { struct pipe_screen *screen = pipe->screen; @@ -405,8 +409,8 @@ choose_format(struct pipe_context *pipe, GLint internalFormat, uint surfType) case GL_ALPHA12: case GL_ALPHA16: case GL_COMPRESSED_ALPHA: - if (screen->is_format_supported( screen, PIPE_FORMAT_U_A8, surfType )) - return PIPE_FORMAT_U_A8; + if (screen->is_format_supported( screen, PIPE_FORMAT_A8_UNORM, surfType )) + return PIPE_FORMAT_A8_UNORM; return default_rgba_format( screen, surfType ); case 1: @@ -416,8 +420,8 @@ choose_format(struct pipe_context *pipe, GLint internalFormat, uint surfType) case GL_LUMINANCE12: case GL_LUMINANCE16: case GL_COMPRESSED_LUMINANCE: - if (screen->is_format_supported( screen, PIPE_FORMAT_U_L8, surfType )) - return PIPE_FORMAT_U_L8; + if (screen->is_format_supported( screen, PIPE_FORMAT_L8_UNORM, surfType )) + return PIPE_FORMAT_L8_UNORM; return default_rgba_format( screen, surfType ); case 2: @@ -429,8 +433,8 @@ choose_format(struct pipe_context *pipe, GLint internalFormat, uint surfType) case GL_LUMINANCE12_ALPHA12: case GL_LUMINANCE16_ALPHA16: case GL_COMPRESSED_LUMINANCE_ALPHA: - if (screen->is_format_supported( screen, PIPE_FORMAT_U_A8_L8, surfType )) - return PIPE_FORMAT_U_A8_L8; + if (screen->is_format_supported( screen, PIPE_FORMAT_A8L8_UNORM, surfType )) + return PIPE_FORMAT_A8L8_UNORM; return default_rgba_format( screen, surfType ); case GL_INTENSITY: @@ -439,8 +443,8 @@ choose_format(struct pipe_context *pipe, GLint internalFormat, uint surfType) case GL_INTENSITY12: case GL_INTENSITY16: case GL_COMPRESSED_INTENSITY: - if (screen->is_format_supported( screen, PIPE_FORMAT_U_I8, surfType )) - return PIPE_FORMAT_U_I8; + if (screen->is_format_supported( screen, PIPE_FORMAT_I8_UNORM, surfType )) + return PIPE_FORMAT_I8_UNORM; return default_rgba_format( screen, surfType ); case GL_YCBCR_MESA: @@ -499,8 +503,8 @@ choose_format(struct pipe_context *pipe, GLint internalFormat, uint surfType) case GL_STENCIL_INDEX4_EXT: case GL_STENCIL_INDEX8_EXT: case GL_STENCIL_INDEX16_EXT: - if (screen->is_format_supported( screen, PIPE_FORMAT_U_S8, surfType )) - return PIPE_FORMAT_U_S8; + if (screen->is_format_supported( screen, PIPE_FORMAT_S8_UNORM, surfType )) + return PIPE_FORMAT_S8_UNORM; if (screen->is_format_supported( screen, PIPE_FORMAT_S8Z24_UNORM, surfType )) return PIPE_FORMAT_S8Z24_UNORM; if (screen->is_format_supported( screen, PIPE_FORMAT_Z24S8_UNORM, surfType )) @@ -527,7 +531,7 @@ choose_format(struct pipe_context *pipe, GLint internalFormat, uint surfType) enum pipe_format st_choose_renderbuffer_format(struct pipe_context *pipe, GLint internalFormat) { - return choose_format(pipe, internalFormat, PIPE_SURFACE); + return st_choose_format(pipe, internalFormat, PIPE_SURFACE); } @@ -543,13 +547,13 @@ translate_gallium_format_to_mesa_format(enum pipe_format format) return &_mesa_texformat_argb4444; case PIPE_FORMAT_R5G6B5_UNORM: return &_mesa_texformat_rgb565; - case PIPE_FORMAT_U_A8_L8: + case PIPE_FORMAT_A8L8_UNORM: return &_mesa_texformat_al88; - case PIPE_FORMAT_U_A8: + case PIPE_FORMAT_A8_UNORM: return &_mesa_texformat_a8; - case PIPE_FORMAT_U_L8: + case PIPE_FORMAT_L8_UNORM: return &_mesa_texformat_l8; - case PIPE_FORMAT_U_I8: + case PIPE_FORMAT_I8_UNORM: return &_mesa_texformat_i8; case PIPE_FORMAT_Z16_UNORM: return &_mesa_texformat_z16; @@ -591,7 +595,7 @@ st_ChooseTextureFormat(GLcontext *ctx, GLint internalFormat, (void) format; (void) type; - pFormat = choose_format(ctx->st->pipe, internalFormat, PIPE_TEXTURE); + pFormat = st_choose_format(ctx->st->pipe, internalFormat, PIPE_TEXTURE); if (pFormat == PIPE_FORMAT_NONE) return NULL; diff --git a/src/mesa/state_tracker/st_format.h b/src/mesa/state_tracker/st_format.h index c9a11de504..ff0fd042db 100644 --- a/src/mesa/state_tracker/st_format.h +++ b/src/mesa/state_tracker/st_format.h @@ -64,6 +64,10 @@ st_mesa_format_to_pipe_format(GLuint mesaFormat); extern enum pipe_format +st_choose_format(struct pipe_context *pipe, GLint internalFormat, + uint surfType); + +extern enum pipe_format st_choose_renderbuffer_format(struct pipe_context *pipe, GLint internalFormat); diff --git a/src/mesa/state_tracker/st_framebuffer.c b/src/mesa/state_tracker/st_framebuffer.c index 075e9d1bd6..47466c97d8 100644 --- a/src/mesa/state_tracker/st_framebuffer.c +++ b/src/mesa/state_tracker/st_framebuffer.c @@ -107,7 +107,7 @@ st_create_framebuffer( const __GLcontextModes *visual, if (visual->accumRedBits > 0) { /* 16-bit/channel accum */ struct gl_renderbuffer *accumRb - = st_new_renderbuffer_fb(PIPE_FORMAT_R16G16B16A16_SNORM); + = st_new_renderbuffer_fb(DEFAULT_ACCUM_PIPE_FORMAT); _mesa_add_renderbuffer(&stfb->Base, BUFFER_ACCUM, accumRb); } @@ -188,6 +188,7 @@ st_notify_swapbuffers(struct st_framebuffer *stfb) st_flush( ctx->st, PIPE_FLUSH_RENDER_CACHE | PIPE_FLUSH_SWAPBUFFERS, NULL ); + ctx->st->frontbuffer_status = FRONT_STATUS_COPY_OF_BACK; } } diff --git a/src/mesa/state_tracker/st_gen_mipmap.c b/src/mesa/state_tracker/st_gen_mipmap.c index a931911227..047743689d 100644 --- a/src/mesa/state_tracker/st_gen_mipmap.c +++ b/src/mesa/state_tracker/st_gen_mipmap.c @@ -36,7 +36,6 @@ #include "pipe/p_context.h" #include "pipe/p_defines.h" #include "pipe/p_inlines.h" -#include "pipe/p_winsys.h" #include "util/u_gen_mipmap.h" #include "cso_cache/cso_cache.h" @@ -95,12 +94,6 @@ st_render_mipmap(struct st_context *st, util_gen_mipmap(st->gen_mipmap, pt, face, baseLevel, lastLevel, PIPE_TEX_FILTER_LINEAR); - /* shaders don't go through CSO yet */ - if (st->fp) - pipe->bind_fs_state(pipe, st->fp->driver_shader); - if (st->vp) - pipe->bind_vs_state(pipe, st->vp->driver_shader); - return TRUE; } @@ -111,7 +104,6 @@ fallback_generate_mipmap(GLcontext *ctx, GLenum target, { struct pipe_context *pipe = ctx->st->pipe; struct pipe_screen *screen = pipe->screen; - struct pipe_winsys *ws = pipe->winsys; struct pipe_texture *pt = st_get_texobj_texture(texObj); const uint baseLevel = texObj->BaseLevel; const uint lastLevel = pt->last_level; @@ -134,11 +126,11 @@ fallback_generate_mipmap(GLcontext *ctx, GLenum target, srcSurf = screen->get_tex_surface(screen, pt, face, srcLevel, zslice); dstSurf = screen->get_tex_surface(screen, pt, face, dstLevel, zslice); - srcData = (ubyte *) ws->buffer_map(ws, srcSurf->buffer, - PIPE_BUFFER_USAGE_CPU_READ) + srcData = (ubyte *) pipe_buffer_map(pipe, srcSurf->buffer, + PIPE_BUFFER_USAGE_CPU_READ) + srcSurf->offset; - dstData = (ubyte *) ws->buffer_map(ws, dstSurf->buffer, - PIPE_BUFFER_USAGE_CPU_WRITE) + dstData = (ubyte *) pipe_buffer_map(pipe, dstSurf->buffer, + PIPE_BUFFER_USAGE_CPU_WRITE) + dstSurf->offset; _mesa_generate_mipmap_level(target, datatype, comps, @@ -150,8 +142,8 @@ fallback_generate_mipmap(GLcontext *ctx, GLenum target, dstSurf->pitch * dstSurf->cpp, /* stride in bytes */ dstData); - ws->buffer_unmap(ws, srcSurf->buffer); - ws->buffer_unmap(ws, dstSurf->buffer); + pipe_buffer_unmap(pipe, srcSurf->buffer); + pipe_buffer_unmap(pipe, dstSurf->buffer); pipe_surface_reference(&srcSurf, NULL); pipe_surface_reference(&dstSurf, NULL); @@ -205,6 +197,6 @@ st_generate_mipmap(GLcontext *ctx, GLenum target, dstImage->TexFormat = srcImage->TexFormat; stImage = (struct st_texture_image *) dstImage; - stImage->pt = pt; + pipe_texture_reference(&stImage->pt, pt); } } diff --git a/src/mesa/state_tracker/st_mesa_to_tgsi.c b/src/mesa/state_tracker/st_mesa_to_tgsi.c index 578fd2ecb0..d2be450bdb 100644 --- a/src/mesa/state_tracker/st_mesa_to_tgsi.c +++ b/src/mesa/state_tracker/st_mesa_to_tgsi.c @@ -767,12 +767,14 @@ tgsi_translate_mesa_program( switch (outputSemanticName[i]) { case TGSI_SEMANTIC_POSITION: fulldecl = make_output_decl(i, - TGSI_SEMANTIC_POSITION, 0, /* Z / Depth */ + TGSI_SEMANTIC_POSITION, /* Z / Depth */ + outputSemanticIndex[i], TGSI_WRITEMASK_Z ); break; case TGSI_SEMANTIC_COLOR: fulldecl = make_output_decl(i, - TGSI_SEMANTIC_COLOR, 0, + TGSI_SEMANTIC_COLOR, + outputSemanticIndex[i], TGSI_WRITEMASK_XYZW ); break; default: @@ -803,7 +805,7 @@ tgsi_translate_mesa_program( { GLboolean tempsUsed[MAX_PROGRAM_TEMPS + 1]; GLboolean inside_range = GL_FALSE; - GLuint start_range; + GLuint start_range = 0; find_temporaries(program, tempsUsed); tempsUsed[MAX_PROGRAM_TEMPS] = GL_FALSE; diff --git a/src/mesa/state_tracker/st_program.c b/src/mesa/state_tracker/st_program.c index 4d1ef1bf3d..d450c30694 100644 --- a/src/mesa/state_tracker/st_program.c +++ b/src/mesa/state_tracker/st_program.c @@ -270,7 +270,7 @@ st_translate_vertex_program(struct st_context *st, stvp->state.tokens = NULL; } if (stvp->driver_shader) { - pipe->delete_vs_state(pipe, stvp->driver_shader); + cso_delete_vertex_shader(st->cso_context, stvp->driver_shader); stvp->driver_shader = NULL; } @@ -315,7 +315,7 @@ st_translate_vertex_program(struct st_context *st, * \param tokensOut destination for TGSI tokens * \return pointer to cached pipe_shader object. */ -const struct cso_fragment_shader * +void st_translate_fragment_program(struct st_context *st, struct st_fragment_program *stfp, const GLuint inputMapping[]) @@ -325,7 +325,6 @@ st_translate_fragment_program(struct st_context *st, GLuint outputMapping[FRAG_RESULT_MAX]; GLuint defaultInputMapping[FRAG_ATTRIB_MAX]; struct pipe_shader_state fs; - const struct cso_fragment_shader *cso; GLuint interpMode[16]; /* XXX size? */ GLuint attr; const GLbitfield inputsRead = stfp->Base.Base.InputsRead; @@ -475,7 +474,5 @@ st_translate_fragment_program(struct st_context *st, if (TGSI_DEBUG) tgsi_dump( fs.tokens, 0/*TGSI_DUMP_VERBOSE*/ ); - - return cso; } diff --git a/src/mesa/state_tracker/st_program.h b/src/mesa/state_tracker/st_program.h index 63d6590540..bf07a50789 100644 --- a/src/mesa/state_tracker/st_program.h +++ b/src/mesa/state_tracker/st_program.h @@ -34,7 +34,8 @@ #ifndef ST_PROGRAM_H #define ST_PROGRAM_H -#include "mtypes.h" +#include "main/mtypes.h" +#include "shader/program.h" #include "pipe/p_shader_tokens.h" @@ -61,7 +62,7 @@ struct st_fragment_program GLuint input_map[PIPE_MAX_SHADER_INPUTS]; struct pipe_shader_state state; - struct pipe_shader_state *driver_shader; + void *driver_shader; GLuint param_state; @@ -69,6 +70,10 @@ struct st_fragment_program * outputs match this fragment program's inputs. */ struct translated_vertex_program *vertex_programs; + + /** Program prefixed with glBitmap prologue */ + struct st_fragment_program *bitmap_program; + uint bitmap_sampler; }; @@ -88,7 +93,7 @@ struct st_vertex_program GLuint num_inputs; struct pipe_shader_state state; - struct pipe_shader_state *driver_shader; + void *driver_shader; /** For using our private draw module (glRasterPos) */ struct draw_vertex_shader *draw_shader; @@ -97,21 +102,42 @@ struct st_vertex_program }; -static inline struct st_fragment_program * +static INLINE struct st_fragment_program * st_fragment_program( struct gl_fragment_program *fp ) { return (struct st_fragment_program *)fp; } -static inline struct st_vertex_program * +static INLINE struct st_vertex_program * st_vertex_program( struct gl_vertex_program *vp ) { return (struct st_vertex_program *)vp; } -extern const struct cso_fragment_shader * +static INLINE void +st_reference_vertprog(struct st_context *st, + struct st_vertex_program **ptr, + struct st_vertex_program *prog) +{ + _mesa_reference_program(st->ctx, + (struct gl_program **) ptr, + (struct gl_program *) prog); +} + +static INLINE void +st_reference_fragprog(struct st_context *st, + struct st_fragment_program **ptr, + struct st_fragment_program *prog) +{ + _mesa_reference_program(st->ctx, + (struct gl_program **) ptr, + (struct gl_program *) prog); +} + + +extern void st_translate_fragment_program(struct st_context *st, struct st_fragment_program *fp, const GLuint inputMapping[]); diff --git a/src/mesa/state_tracker/st_texobj.c b/src/mesa/state_tracker/st_texobj.c deleted file mode 100644 index eb5bdb2d08..0000000000 --- a/src/mesa/state_tracker/st_texobj.c +++ /dev/null @@ -1,102 +0,0 @@ -/************************************************************************** - * - * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - -/* - * Authors: - * Brian Paul - */ - -#include "imports.h" -#include "texformat.h" - -#include "st_context.h" -#include "st_texobj.h" -#include "pipe/p_defines.h" - - -/** - * Create a pipe_texture_object from a Mesa texture object. - * Eventually, gl_texture_object may be derived from this... - */ -struct pipe_texture_object * -create_texture_object(struct gl_texture_object *texObj) -{ - struct pipe_texture_object *pto; - const struct gl_texture_image *texImage; - - pto = calloc(1, sizeof(*pto)); - if (!pto) - return NULL; - - /* XXX: Member not defined. Comment-out to get it compile. */ - /*assert(texObj->Complete);*/ - - switch (texObj->Target) { - case GL_TEXTURE_1D: - pto->type = PIPE_TEXTURE_1D; - break; - case GL_TEXTURE_2D: - pto->type = PIPE_TEXTURE_2D; - break; - case GL_TEXTURE_3D: - pto->type = PIPE_TEXTURE_3D; - break; - case GL_TEXTURE_CUBE_MAP: - pto->type = PIPE_TEXTURE_CUBE; - break; - default: - assert(0); - return NULL; - } - - texImage = texObj->Image[0][texObj->BaseLevel]; - assert(texImage); - - switch (texImage->TexFormat->MesaFormat) { - case MESA_FORMAT_RGBA8888: - pto->format = PIPE_FORMAT_U_R8_G8_B8_A8; - break; - case MESA_FORMAT_RGB565: - pto->format = PIPE_FORMAT_U_R5_G6_B5; - break; - - /* XXX fill in more formats */ - - default: - assert(0); - return NULL; - } - - pto->width = texImage->Width; - pto->height = texImage->Height; - pto->depth = texImage->Depth; - - /* XXX verify this */ - pto->mipmapped = texObj->Image[0][texObj->BaseLevel + 1] != NULL; - - return pto; -} diff --git a/src/mesa/state_tracker/st_texture.c b/src/mesa/state_tracker/st_texture.c index 8e3235cc99..f68bef1207 100644 --- a/src/mesa/state_tracker/st_texture.c +++ b/src/mesa/state_tracker/st_texture.c @@ -38,7 +38,6 @@ #include "pipe/p_inlines.h" #include "pipe/p_util.h" #include "pipe/p_inlines.h" -#include "pipe/p_winsys.h" #define DBG if(0) printf @@ -290,8 +289,7 @@ st_texture_image_copy(struct pipe_context *pipe, GLuint i; /* XXX this is a hack */ - if (dst->compressed) - height /= 4; + const GLuint copyHeight = dst->compressed ? height / 4 : height; for (i = 0; i < depth; i++) { GLuint srcLevel; @@ -315,7 +313,7 @@ st_texture_image_copy(struct pipe_context *pipe, 0, 0, /* destX, Y */ src_surface, 0, 0, /* srcX, Y */ - width, height); + width, copyHeight); pipe_surface_reference(&dst_surface, NULL); pipe_surface_reference(&src_surface, NULL); diff --git a/src/mesa/state_tracker/st_texture.h b/src/mesa/state_tracker/st_texture.h index 7abccb3a69..b05762113c 100644 --- a/src/mesa/state_tracker/st_texture.h +++ b/src/mesa/state_tracker/st_texture.h @@ -68,9 +68,7 @@ struct st_texture_object */ struct pipe_texture *pt; - GLboolean imageOverride; - GLint depthOverride; - GLuint pitchOverride; + GLboolean teximage_realloc; }; diff --git a/src/mesa/swrast/s_drawpix.c b/src/mesa/swrast/s_drawpix.c index fb04d9f746..cbf6617058 100644 --- a/src/mesa/swrast/s_drawpix.c +++ b/src/mesa/swrast/s_drawpix.c @@ -877,7 +877,7 @@ end: RENDER_FINISH(swrast,ctx); - _mesa_unmap_drapix_pbo(ctx, unpack); + _mesa_unmap_drawpix_pbo(ctx, unpack); } diff --git a/src/mesa/swrast/s_fragprog.c b/src/mesa/swrast/s_fragprog.c index 6ee8bfd0a5..8eeb40f7c5 100644 --- a/src/mesa/swrast/s_fragprog.c +++ b/src/mesa/swrast/s_fragprog.c @@ -1,8 +1,8 @@ /* * Mesa 3-D graphics library - * Version: 6.5.2 + * Version: 7.0.3 * - * Copyright (C) 1999-2006 Brian Paul All Rights Reserved. + * Copyright (C) 1999-2007 Brian Paul All Rights Reserved. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the "Software"), @@ -33,18 +33,21 @@ /** - * Fetch a texel. + * Fetch a texel with given lod. + * Called via machine->FetchTexelLod() */ static void -fetch_texel( GLcontext *ctx, const GLfloat texcoord[4], GLfloat lambda, - GLuint unit, GLfloat color[4] ) +fetch_texel_lod( GLcontext *ctx, const GLfloat texcoord[4], GLfloat lambda, + GLuint unit, GLfloat color[4] ) { GLchan rgba[4]; SWcontext *swrast = SWRAST_CONTEXT(ctx); + const struct gl_texture_object *texObj = ctx->Texture.Unit[unit]._Current; + + lambda = CLAMP(lambda, texObj->MinLod, texObj->MaxLod); /* XXX use a float-valued TextureSample routine here!!! */ - swrast->TextureSample[unit](ctx, ctx->Texture.Unit[unit]._Current, - 1, (const GLfloat (*)[4]) texcoord, + swrast->TextureSample[unit](ctx, texObj, 1, (const GLfloat (*)[4]) texcoord, &lambda, &rgba); color[0] = CHAN_TO_FLOAT(rgba[0]); color[1] = CHAN_TO_FLOAT(rgba[1]); @@ -56,11 +59,12 @@ fetch_texel( GLcontext *ctx, const GLfloat texcoord[4], GLfloat lambda, /** * Fetch a texel with the given partial derivatives to compute a level * of detail in the mipmap. + * Called via machine->FetchTexelDeriv() */ static void fetch_texel_deriv( GLcontext *ctx, const GLfloat texcoord[4], const GLfloat texdx[4], const GLfloat texdy[4], - GLuint unit, GLfloat color[4] ) + GLfloat lodBias, GLuint unit, GLfloat color[4] ) { SWcontext *swrast = SWRAST_CONTEXT(ctx); const struct gl_texture_object *texObj = ctx->Texture.Unit[unit]._Current; @@ -69,15 +73,17 @@ fetch_texel_deriv( GLcontext *ctx, const GLfloat texcoord[4], const GLfloat texH = (GLfloat) texImg->HeightScale; GLchan rgba[4]; - GLfloat lambda = _swrast_compute_lambda(texdx[0], texdy[0], /* ds/dx, ds/dy */ - texdx[1], texdy[1], /* dt/dx, dt/dy */ - texdx[3], texdy[2], /* dq/dx, dq/dy */ - texW, texH, - texcoord[0], texcoord[1], texcoord[3], - 1.0F / texcoord[3]); + GLfloat lambda + = _swrast_compute_lambda(texdx[0], texdy[0], /* ds/dx, ds/dy */ + texdx[1], texdy[1], /* dt/dx, dt/dy */ + texdx[3], texdy[2], /* dq/dx, dq/dy */ + texW, texH, + texcoord[0], texcoord[1], texcoord[3], + 1.0F / texcoord[3]) + lodBias; + + lambda = CLAMP(lambda, texObj->MinLod, texObj->MaxLod); - swrast->TextureSample[unit](ctx, ctx->Texture.Unit[unit]._Current, - 1, (const GLfloat (*)[4]) texcoord, + swrast->TextureSample[unit](ctx, texObj, 1, (const GLfloat (*)[4]) texcoord, &lambda, &rgba); color[0] = CHAN_TO_FLOAT(rgba[0]); color[1] = CHAN_TO_FLOAT(rgba[1]); @@ -132,7 +138,7 @@ init_machine(GLcontext *ctx, struct gl_program_machine *machine, /* init call stack */ machine->StackDepth = 0; - machine->FetchTexelLod = fetch_texel; + machine->FetchTexelLod = fetch_texel_lod; machine->FetchTexelDeriv = fetch_texel_deriv; } diff --git a/src/mesa/vbo/vbo_exec.h b/src/mesa/vbo/vbo_exec.h index b7e8c9fe79..ddbcbe1181 100644 --- a/src/mesa/vbo/vbo_exec.h +++ b/src/mesa/vbo/vbo_exec.h @@ -162,4 +162,7 @@ void vbo_exec_do_EvalCoord2f( struct vbo_exec_context *exec, void vbo_exec_do_EvalCoord1f( struct vbo_exec_context *exec, GLfloat u); +extern GLboolean +vbo_validate_shaders(GLcontext *ctx); + #endif diff --git a/src/mesa/vbo/vbo_exec_api.c b/src/mesa/vbo/vbo_exec_api.c index b7f4d8a307..98580170e3 100644 --- a/src/mesa/vbo/vbo_exec_api.c +++ b/src/mesa/vbo/vbo_exec_api.c @@ -477,6 +477,23 @@ static void GLAPIENTRY vbo_exec_EvalPoint2( GLint i, GLint j ) } +/** + * Check if programs/shaders are enabled and valid at glBegin time. + */ +GLboolean +vbo_validate_shaders(GLcontext *ctx) +{ + if ((ctx->VertexProgram.Enabled && !ctx->VertexProgram._Enabled) || + (ctx->FragmentProgram.Enabled && !ctx->FragmentProgram._Enabled)) { + return GL_FALSE; + } + if (ctx->Shader.CurrentProgram && !ctx->Shader.CurrentProgram->LinkStatus) { + return GL_FALSE; + } + return GL_TRUE; +} + + /* Build a list of primitives on the fly. Keep * ctx->Driver.CurrentExecPrimitive uptodate as well. */ @@ -491,18 +508,16 @@ static void GLAPIENTRY vbo_exec_Begin( GLenum mode ) if (ctx->NewState) { _mesa_update_state( ctx ); - /* XXX also need to check if shader enabled, but invalid */ - if ((ctx->VertexProgram.Enabled && !ctx->VertexProgram._Enabled) || - (ctx->FragmentProgram.Enabled && !ctx->FragmentProgram._Enabled)) { - _mesa_error(ctx, GL_INVALID_OPERATION, - "glBegin (invalid vertex/fragment program)"); - return; - } - CALL_Begin(ctx->Exec, (mode)); return; } + if (!vbo_validate_shaders(ctx)) { + _mesa_error(ctx, GL_INVALID_OPERATION, + "glBegin (invalid vertex/fragment program)"); + return; + } + /* Heuristic: attempt to isolate attributes occuring outside * begin/end pairs. */ diff --git a/src/mesa/vbo/vbo_exec_array.c b/src/mesa/vbo/vbo_exec_array.c index 77f3cf1455..a52521db64 100644 --- a/src/mesa/vbo/vbo_exec_array.c +++ b/src/mesa/vbo/vbo_exec_array.c @@ -245,6 +245,11 @@ vbo_exec_DrawArrays(GLenum mode, GLint start, GLsizei count) if (ctx->NewState) _mesa_update_state( ctx ); + if (!vbo_validate_shaders(ctx)) { + _mesa_error(ctx, GL_INVALID_OPERATION, "glDrawArrays(bad shader)"); + return; + } + bind_arrays( ctx ); prim[0].begin = 1; @@ -280,6 +285,11 @@ vbo_exec_DrawRangeElements(GLenum mode, if (ctx->NewState) _mesa_update_state( ctx ); + if (!vbo_validate_shaders(ctx)) { + _mesa_error(ctx, GL_INVALID_OPERATION, "glDrawRangeElements(bad shader)"); + return; + } + bind_arrays( ctx ); ib.count = count; @@ -340,6 +350,11 @@ vbo_exec_DrawElements(GLenum mode, GLsizei count, GLenum type, const GLvoid *ind if (!_mesa_validate_DrawElements( ctx, mode, count, type, indices )) return; + if (!vbo_validate_shaders(ctx)) { + _mesa_error(ctx, GL_INVALID_OPERATION, "glDrawElements(bad shader)"); + return; + } + if (ctx->Array.ElementArrayBufferObj->Name) { const GLvoid *map = ctx->Driver.MapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER_ARB, diff --git a/src/mesa/vbo/vbo_exec_draw.c b/src/mesa/vbo/vbo_exec_draw.c index 23a3658386..6bb6849a6a 100644 --- a/src/mesa/vbo/vbo_exec_draw.c +++ b/src/mesa/vbo/vbo_exec_draw.c @@ -198,7 +198,7 @@ static void vbo_exec_bind_arrays( GLcontext *ctx ) arrays[attr].BufferObj = exec->vtx.bufferobj; arrays[attr]._MaxElement = count; /* ??? */ - data += exec->vtx.attrsz[attr] * sizeof(GLfloat); + data += exec->vtx.attrsz[src] * sizeof(GLfloat); } } } diff --git a/src/mesa/vbo/vbo_save_draw.c b/src/mesa/vbo/vbo_save_draw.c index 3c6f0fccd9..bf5c6d4eef 100644 --- a/src/mesa/vbo/vbo_save_draw.c +++ b/src/mesa/vbo/vbo_save_draw.c @@ -131,10 +131,12 @@ static void vbo_bind_vertex_list( GLcontext *ctx, break; } - for (attr = 0; attr < VBO_ATTRIB_MAX; attr++) { - if (node->attrsz[attr]) { + for (attr = 0; attr < VERT_ATTRIB_MAX; attr++) { + GLuint src = map[attr]; + + if (node->attrsz[src]) { arrays[attr].Ptr = (const GLubyte *)data; - arrays[attr].Size = node->attrsz[attr]; + arrays[attr].Size = node->attrsz[src]; arrays[attr].StrideB = node->vertex_size * sizeof(GLfloat); arrays[attr].Stride = node->vertex_size * sizeof(GLfloat); arrays[attr].Type = GL_FLOAT; @@ -144,7 +146,7 @@ static void vbo_bind_vertex_list( GLcontext *ctx, assert(arrays[attr].BufferObj->Name); - data += node->attrsz[attr] * sizeof(GLfloat); + data += node->attrsz[src] * sizeof(GLfloat); } } } diff --git a/src/mesa/x86-64/glapi_x86-64.S b/src/mesa/x86-64/glapi_x86-64.S index a239a6749d..cb34061b36 100644 --- a/src/mesa/x86-64/glapi_x86-64.S +++ b/src/mesa/x86-64/glapi_x86-64.S @@ -25268,192 +25268,12 @@ GL_PREFIX(LoadProgramNV): .size GL_PREFIX(LoadProgramNV), .-GL_PREFIX(LoadProgramNV) .p2align 4,,15 - .globl GL_PREFIX(ProgramParameter4dNV) - .type GL_PREFIX(ProgramParameter4dNV), @function -GL_PREFIX(ProgramParameter4dNV): -#if defined(GLX_USE_TLS) - call _x86_64_get_dispatch@PLT - movq 5384(%rax), %r11 - jmp *%r11 -#elif defined(PTHREADS) - subq $56, %rsp - movq %rdi, (%rsp) - movq %rsi, 8(%rsp) - movq %xmm0, 16(%rsp) - movq %xmm1, 24(%rsp) - movq %xmm2, 32(%rsp) - movq %xmm3, 40(%rsp) - call _x86_64_get_dispatch@PLT - movq 40(%rsp), %xmm3 - movq 32(%rsp), %xmm2 - movq 24(%rsp), %xmm1 - movq 16(%rsp), %xmm0 - movq 8(%rsp), %rsi - movq (%rsp), %rdi - addq $56, %rsp - movq 5384(%rax), %r11 - jmp *%r11 -#else - movq _glapi_Dispatch(%rip), %rax - testq %rax, %rax - je 1f - movq 5384(%rax), %r11 - jmp *%r11 -1: - subq $56, %rsp - movq %rdi, (%rsp) - movq %rsi, 8(%rsp) - movq %xmm0, 16(%rsp) - movq %xmm1, 24(%rsp) - movq %xmm2, 32(%rsp) - movq %xmm3, 40(%rsp) - call _glapi_get_dispatch - movq 40(%rsp), %xmm3 - movq 32(%rsp), %xmm2 - movq 24(%rsp), %xmm1 - movq 16(%rsp), %xmm0 - movq 8(%rsp), %rsi - movq (%rsp), %rdi - addq $56, %rsp - movq 5384(%rax), %r11 - jmp *%r11 -#endif /* defined(GLX_USE_TLS) */ - .size GL_PREFIX(ProgramParameter4dNV), .-GL_PREFIX(ProgramParameter4dNV) - - .p2align 4,,15 - .globl GL_PREFIX(ProgramParameter4dvNV) - .type GL_PREFIX(ProgramParameter4dvNV), @function -GL_PREFIX(ProgramParameter4dvNV): -#if defined(GLX_USE_TLS) - call _x86_64_get_dispatch@PLT - movq 5392(%rax), %r11 - jmp *%r11 -#elif defined(PTHREADS) - pushq %rdi - pushq %rsi - pushq %rdx - call _x86_64_get_dispatch@PLT - popq %rdx - popq %rsi - popq %rdi - movq 5392(%rax), %r11 - jmp *%r11 -#else - movq _glapi_Dispatch(%rip), %rax - testq %rax, %rax - je 1f - movq 5392(%rax), %r11 - jmp *%r11 -1: - pushq %rdi - pushq %rsi - pushq %rdx - call _glapi_get_dispatch - popq %rdx - popq %rsi - popq %rdi - movq 5392(%rax), %r11 - jmp *%r11 -#endif /* defined(GLX_USE_TLS) */ - .size GL_PREFIX(ProgramParameter4dvNV), .-GL_PREFIX(ProgramParameter4dvNV) - - .p2align 4,,15 - .globl GL_PREFIX(ProgramParameter4fNV) - .type GL_PREFIX(ProgramParameter4fNV), @function -GL_PREFIX(ProgramParameter4fNV): -#if defined(GLX_USE_TLS) - call _x86_64_get_dispatch@PLT - movq 5400(%rax), %r11 - jmp *%r11 -#elif defined(PTHREADS) - subq $56, %rsp - movq %rdi, (%rsp) - movq %rsi, 8(%rsp) - movq %xmm0, 16(%rsp) - movq %xmm1, 24(%rsp) - movq %xmm2, 32(%rsp) - movq %xmm3, 40(%rsp) - call _x86_64_get_dispatch@PLT - movq 40(%rsp), %xmm3 - movq 32(%rsp), %xmm2 - movq 24(%rsp), %xmm1 - movq 16(%rsp), %xmm0 - movq 8(%rsp), %rsi - movq (%rsp), %rdi - addq $56, %rsp - movq 5400(%rax), %r11 - jmp *%r11 -#else - movq _glapi_Dispatch(%rip), %rax - testq %rax, %rax - je 1f - movq 5400(%rax), %r11 - jmp *%r11 -1: - subq $56, %rsp - movq %rdi, (%rsp) - movq %rsi, 8(%rsp) - movq %xmm0, 16(%rsp) - movq %xmm1, 24(%rsp) - movq %xmm2, 32(%rsp) - movq %xmm3, 40(%rsp) - call _glapi_get_dispatch - movq 40(%rsp), %xmm3 - movq 32(%rsp), %xmm2 - movq 24(%rsp), %xmm1 - movq 16(%rsp), %xmm0 - movq 8(%rsp), %rsi - movq (%rsp), %rdi - addq $56, %rsp - movq 5400(%rax), %r11 - jmp *%r11 -#endif /* defined(GLX_USE_TLS) */ - .size GL_PREFIX(ProgramParameter4fNV), .-GL_PREFIX(ProgramParameter4fNV) - - .p2align 4,,15 - .globl GL_PREFIX(ProgramParameter4fvNV) - .type GL_PREFIX(ProgramParameter4fvNV), @function -GL_PREFIX(ProgramParameter4fvNV): -#if defined(GLX_USE_TLS) - call _x86_64_get_dispatch@PLT - movq 5408(%rax), %r11 - jmp *%r11 -#elif defined(PTHREADS) - pushq %rdi - pushq %rsi - pushq %rdx - call _x86_64_get_dispatch@PLT - popq %rdx - popq %rsi - popq %rdi - movq 5408(%rax), %r11 - jmp *%r11 -#else - movq _glapi_Dispatch(%rip), %rax - testq %rax, %rax - je 1f - movq 5408(%rax), %r11 - jmp *%r11 -1: - pushq %rdi - pushq %rsi - pushq %rdx - call _glapi_get_dispatch - popq %rdx - popq %rsi - popq %rdi - movq 5408(%rax), %r11 - jmp *%r11 -#endif /* defined(GLX_USE_TLS) */ - .size GL_PREFIX(ProgramParameter4fvNV), .-GL_PREFIX(ProgramParameter4fvNV) - - .p2align 4,,15 .globl GL_PREFIX(ProgramParameters4dvNV) .type GL_PREFIX(ProgramParameters4dvNV), @function GL_PREFIX(ProgramParameters4dvNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5416(%rax), %r11 + movq 5384(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -25467,13 +25287,13 @@ GL_PREFIX(ProgramParameters4dvNV): popq %rdx popq %rsi popq %rdi - movq 5416(%rax), %r11 + movq 5384(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5416(%rax), %r11 + movq 5384(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -25487,7 +25307,7 @@ GL_PREFIX(ProgramParameters4dvNV): popq %rdx popq %rsi popq %rdi - movq 5416(%rax), %r11 + movq 5384(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(ProgramParameters4dvNV), .-GL_PREFIX(ProgramParameters4dvNV) @@ -25498,7 +25318,7 @@ GL_PREFIX(ProgramParameters4dvNV): GL_PREFIX(ProgramParameters4fvNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5424(%rax), %r11 + movq 5392(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -25512,13 +25332,13 @@ GL_PREFIX(ProgramParameters4fvNV): popq %rdx popq %rsi popq %rdi - movq 5424(%rax), %r11 + movq 5392(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5424(%rax), %r11 + movq 5392(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -25532,7 +25352,7 @@ GL_PREFIX(ProgramParameters4fvNV): popq %rdx popq %rsi popq %rdi - movq 5424(%rax), %r11 + movq 5392(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(ProgramParameters4fvNV), .-GL_PREFIX(ProgramParameters4fvNV) @@ -25543,7 +25363,7 @@ GL_PREFIX(ProgramParameters4fvNV): GL_PREFIX(RequestResidentProgramsNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5432(%rax), %r11 + movq 5400(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -25553,13 +25373,13 @@ GL_PREFIX(RequestResidentProgramsNV): popq %rbp popq %rsi popq %rdi - movq 5432(%rax), %r11 + movq 5400(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5432(%rax), %r11 + movq 5400(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -25569,7 +25389,7 @@ GL_PREFIX(RequestResidentProgramsNV): popq %rbp popq %rsi popq %rdi - movq 5432(%rax), %r11 + movq 5400(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(RequestResidentProgramsNV), .-GL_PREFIX(RequestResidentProgramsNV) @@ -25580,7 +25400,7 @@ GL_PREFIX(RequestResidentProgramsNV): GL_PREFIX(TrackMatrixNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5440(%rax), %r11 + movq 5408(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -25594,13 +25414,13 @@ GL_PREFIX(TrackMatrixNV): popq %rdx popq %rsi popq %rdi - movq 5440(%rax), %r11 + movq 5408(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5440(%rax), %r11 + movq 5408(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -25614,7 +25434,7 @@ GL_PREFIX(TrackMatrixNV): popq %rdx popq %rsi popq %rdi - movq 5440(%rax), %r11 + movq 5408(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(TrackMatrixNV), .-GL_PREFIX(TrackMatrixNV) @@ -25625,7 +25445,7 @@ GL_PREFIX(TrackMatrixNV): GL_PREFIX(VertexAttrib1dNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5448(%rax), %r11 + movq 5416(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) subq $24, %rsp @@ -25635,13 +25455,13 @@ GL_PREFIX(VertexAttrib1dNV): movq 8(%rsp), %xmm0 movq (%rsp), %rdi addq $24, %rsp - movq 5448(%rax), %r11 + movq 5416(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5448(%rax), %r11 + movq 5416(%rax), %r11 jmp *%r11 1: subq $24, %rsp @@ -25651,7 +25471,7 @@ GL_PREFIX(VertexAttrib1dNV): movq 8(%rsp), %xmm0 movq (%rsp), %rdi addq $24, %rsp - movq 5448(%rax), %r11 + movq 5416(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttrib1dNV), .-GL_PREFIX(VertexAttrib1dNV) @@ -25662,7 +25482,7 @@ GL_PREFIX(VertexAttrib1dNV): GL_PREFIX(VertexAttrib1dvNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5456(%rax), %r11 + movq 5424(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -25672,13 +25492,13 @@ GL_PREFIX(VertexAttrib1dvNV): popq %rbp popq %rsi popq %rdi - movq 5456(%rax), %r11 + movq 5424(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5456(%rax), %r11 + movq 5424(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -25688,7 +25508,7 @@ GL_PREFIX(VertexAttrib1dvNV): popq %rbp popq %rsi popq %rdi - movq 5456(%rax), %r11 + movq 5424(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttrib1dvNV), .-GL_PREFIX(VertexAttrib1dvNV) @@ -25699,7 +25519,7 @@ GL_PREFIX(VertexAttrib1dvNV): GL_PREFIX(VertexAttrib1fNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5464(%rax), %r11 + movq 5432(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) subq $24, %rsp @@ -25709,13 +25529,13 @@ GL_PREFIX(VertexAttrib1fNV): movq 8(%rsp), %xmm0 movq (%rsp), %rdi addq $24, %rsp - movq 5464(%rax), %r11 + movq 5432(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5464(%rax), %r11 + movq 5432(%rax), %r11 jmp *%r11 1: subq $24, %rsp @@ -25725,7 +25545,7 @@ GL_PREFIX(VertexAttrib1fNV): movq 8(%rsp), %xmm0 movq (%rsp), %rdi addq $24, %rsp - movq 5464(%rax), %r11 + movq 5432(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttrib1fNV), .-GL_PREFIX(VertexAttrib1fNV) @@ -25736,7 +25556,7 @@ GL_PREFIX(VertexAttrib1fNV): GL_PREFIX(VertexAttrib1fvNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5472(%rax), %r11 + movq 5440(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -25746,13 +25566,13 @@ GL_PREFIX(VertexAttrib1fvNV): popq %rbp popq %rsi popq %rdi - movq 5472(%rax), %r11 + movq 5440(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5472(%rax), %r11 + movq 5440(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -25762,7 +25582,7 @@ GL_PREFIX(VertexAttrib1fvNV): popq %rbp popq %rsi popq %rdi - movq 5472(%rax), %r11 + movq 5440(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttrib1fvNV), .-GL_PREFIX(VertexAttrib1fvNV) @@ -25773,7 +25593,7 @@ GL_PREFIX(VertexAttrib1fvNV): GL_PREFIX(VertexAttrib1sNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5480(%rax), %r11 + movq 5448(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -25783,13 +25603,13 @@ GL_PREFIX(VertexAttrib1sNV): popq %rbp popq %rsi popq %rdi - movq 5480(%rax), %r11 + movq 5448(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5480(%rax), %r11 + movq 5448(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -25799,7 +25619,7 @@ GL_PREFIX(VertexAttrib1sNV): popq %rbp popq %rsi popq %rdi - movq 5480(%rax), %r11 + movq 5448(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttrib1sNV), .-GL_PREFIX(VertexAttrib1sNV) @@ -25810,7 +25630,7 @@ GL_PREFIX(VertexAttrib1sNV): GL_PREFIX(VertexAttrib1svNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5488(%rax), %r11 + movq 5456(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -25820,13 +25640,13 @@ GL_PREFIX(VertexAttrib1svNV): popq %rbp popq %rsi popq %rdi - movq 5488(%rax), %r11 + movq 5456(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5488(%rax), %r11 + movq 5456(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -25836,7 +25656,7 @@ GL_PREFIX(VertexAttrib1svNV): popq %rbp popq %rsi popq %rdi - movq 5488(%rax), %r11 + movq 5456(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttrib1svNV), .-GL_PREFIX(VertexAttrib1svNV) @@ -25847,7 +25667,7 @@ GL_PREFIX(VertexAttrib1svNV): GL_PREFIX(VertexAttrib2dNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5496(%rax), %r11 + movq 5464(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) subq $24, %rsp @@ -25859,13 +25679,13 @@ GL_PREFIX(VertexAttrib2dNV): movq 8(%rsp), %xmm0 movq (%rsp), %rdi addq $24, %rsp - movq 5496(%rax), %r11 + movq 5464(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5496(%rax), %r11 + movq 5464(%rax), %r11 jmp *%r11 1: subq $24, %rsp @@ -25877,7 +25697,7 @@ GL_PREFIX(VertexAttrib2dNV): movq 8(%rsp), %xmm0 movq (%rsp), %rdi addq $24, %rsp - movq 5496(%rax), %r11 + movq 5464(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttrib2dNV), .-GL_PREFIX(VertexAttrib2dNV) @@ -25888,7 +25708,7 @@ GL_PREFIX(VertexAttrib2dNV): GL_PREFIX(VertexAttrib2dvNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5504(%rax), %r11 + movq 5472(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -25898,13 +25718,13 @@ GL_PREFIX(VertexAttrib2dvNV): popq %rbp popq %rsi popq %rdi - movq 5504(%rax), %r11 + movq 5472(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5504(%rax), %r11 + movq 5472(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -25914,7 +25734,7 @@ GL_PREFIX(VertexAttrib2dvNV): popq %rbp popq %rsi popq %rdi - movq 5504(%rax), %r11 + movq 5472(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttrib2dvNV), .-GL_PREFIX(VertexAttrib2dvNV) @@ -25925,7 +25745,7 @@ GL_PREFIX(VertexAttrib2dvNV): GL_PREFIX(VertexAttrib2fNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5512(%rax), %r11 + movq 5480(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) subq $24, %rsp @@ -25937,13 +25757,13 @@ GL_PREFIX(VertexAttrib2fNV): movq 8(%rsp), %xmm0 movq (%rsp), %rdi addq $24, %rsp - movq 5512(%rax), %r11 + movq 5480(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5512(%rax), %r11 + movq 5480(%rax), %r11 jmp *%r11 1: subq $24, %rsp @@ -25955,7 +25775,7 @@ GL_PREFIX(VertexAttrib2fNV): movq 8(%rsp), %xmm0 movq (%rsp), %rdi addq $24, %rsp - movq 5512(%rax), %r11 + movq 5480(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttrib2fNV), .-GL_PREFIX(VertexAttrib2fNV) @@ -25966,7 +25786,7 @@ GL_PREFIX(VertexAttrib2fNV): GL_PREFIX(VertexAttrib2fvNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5520(%rax), %r11 + movq 5488(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -25976,13 +25796,13 @@ GL_PREFIX(VertexAttrib2fvNV): popq %rbp popq %rsi popq %rdi - movq 5520(%rax), %r11 + movq 5488(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5520(%rax), %r11 + movq 5488(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -25992,7 +25812,7 @@ GL_PREFIX(VertexAttrib2fvNV): popq %rbp popq %rsi popq %rdi - movq 5520(%rax), %r11 + movq 5488(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttrib2fvNV), .-GL_PREFIX(VertexAttrib2fvNV) @@ -26003,7 +25823,7 @@ GL_PREFIX(VertexAttrib2fvNV): GL_PREFIX(VertexAttrib2sNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5528(%rax), %r11 + movq 5496(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -26013,13 +25833,13 @@ GL_PREFIX(VertexAttrib2sNV): popq %rdx popq %rsi popq %rdi - movq 5528(%rax), %r11 + movq 5496(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5528(%rax), %r11 + movq 5496(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -26029,7 +25849,7 @@ GL_PREFIX(VertexAttrib2sNV): popq %rdx popq %rsi popq %rdi - movq 5528(%rax), %r11 + movq 5496(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttrib2sNV), .-GL_PREFIX(VertexAttrib2sNV) @@ -26040,7 +25860,7 @@ GL_PREFIX(VertexAttrib2sNV): GL_PREFIX(VertexAttrib2svNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5536(%rax), %r11 + movq 5504(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -26050,13 +25870,13 @@ GL_PREFIX(VertexAttrib2svNV): popq %rbp popq %rsi popq %rdi - movq 5536(%rax), %r11 + movq 5504(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5536(%rax), %r11 + movq 5504(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -26066,7 +25886,7 @@ GL_PREFIX(VertexAttrib2svNV): popq %rbp popq %rsi popq %rdi - movq 5536(%rax), %r11 + movq 5504(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttrib2svNV), .-GL_PREFIX(VertexAttrib2svNV) @@ -26077,7 +25897,7 @@ GL_PREFIX(VertexAttrib2svNV): GL_PREFIX(VertexAttrib3dNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5544(%rax), %r11 + movq 5512(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) subq $40, %rsp @@ -26091,13 +25911,13 @@ GL_PREFIX(VertexAttrib3dNV): movq 8(%rsp), %xmm0 movq (%rsp), %rdi addq $40, %rsp - movq 5544(%rax), %r11 + movq 5512(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5544(%rax), %r11 + movq 5512(%rax), %r11 jmp *%r11 1: subq $40, %rsp @@ -26111,7 +25931,7 @@ GL_PREFIX(VertexAttrib3dNV): movq 8(%rsp), %xmm0 movq (%rsp), %rdi addq $40, %rsp - movq 5544(%rax), %r11 + movq 5512(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttrib3dNV), .-GL_PREFIX(VertexAttrib3dNV) @@ -26122,7 +25942,7 @@ GL_PREFIX(VertexAttrib3dNV): GL_PREFIX(VertexAttrib3dvNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5552(%rax), %r11 + movq 5520(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -26132,13 +25952,13 @@ GL_PREFIX(VertexAttrib3dvNV): popq %rbp popq %rsi popq %rdi - movq 5552(%rax), %r11 + movq 5520(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5552(%rax), %r11 + movq 5520(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -26148,7 +25968,7 @@ GL_PREFIX(VertexAttrib3dvNV): popq %rbp popq %rsi popq %rdi - movq 5552(%rax), %r11 + movq 5520(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttrib3dvNV), .-GL_PREFIX(VertexAttrib3dvNV) @@ -26159,7 +25979,7 @@ GL_PREFIX(VertexAttrib3dvNV): GL_PREFIX(VertexAttrib3fNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5560(%rax), %r11 + movq 5528(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) subq $40, %rsp @@ -26173,13 +25993,13 @@ GL_PREFIX(VertexAttrib3fNV): movq 8(%rsp), %xmm0 movq (%rsp), %rdi addq $40, %rsp - movq 5560(%rax), %r11 + movq 5528(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5560(%rax), %r11 + movq 5528(%rax), %r11 jmp *%r11 1: subq $40, %rsp @@ -26193,7 +26013,7 @@ GL_PREFIX(VertexAttrib3fNV): movq 8(%rsp), %xmm0 movq (%rsp), %rdi addq $40, %rsp - movq 5560(%rax), %r11 + movq 5528(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttrib3fNV), .-GL_PREFIX(VertexAttrib3fNV) @@ -26204,7 +26024,7 @@ GL_PREFIX(VertexAttrib3fNV): GL_PREFIX(VertexAttrib3fvNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5568(%rax), %r11 + movq 5536(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -26214,13 +26034,13 @@ GL_PREFIX(VertexAttrib3fvNV): popq %rbp popq %rsi popq %rdi - movq 5568(%rax), %r11 + movq 5536(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5568(%rax), %r11 + movq 5536(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -26230,7 +26050,7 @@ GL_PREFIX(VertexAttrib3fvNV): popq %rbp popq %rsi popq %rdi - movq 5568(%rax), %r11 + movq 5536(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttrib3fvNV), .-GL_PREFIX(VertexAttrib3fvNV) @@ -26241,7 +26061,7 @@ GL_PREFIX(VertexAttrib3fvNV): GL_PREFIX(VertexAttrib3sNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5576(%rax), %r11 + movq 5544(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -26255,13 +26075,13 @@ GL_PREFIX(VertexAttrib3sNV): popq %rdx popq %rsi popq %rdi - movq 5576(%rax), %r11 + movq 5544(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5576(%rax), %r11 + movq 5544(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -26275,7 +26095,7 @@ GL_PREFIX(VertexAttrib3sNV): popq %rdx popq %rsi popq %rdi - movq 5576(%rax), %r11 + movq 5544(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttrib3sNV), .-GL_PREFIX(VertexAttrib3sNV) @@ -26286,7 +26106,7 @@ GL_PREFIX(VertexAttrib3sNV): GL_PREFIX(VertexAttrib3svNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5584(%rax), %r11 + movq 5552(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -26296,13 +26116,13 @@ GL_PREFIX(VertexAttrib3svNV): popq %rbp popq %rsi popq %rdi - movq 5584(%rax), %r11 + movq 5552(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5584(%rax), %r11 + movq 5552(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -26312,7 +26132,7 @@ GL_PREFIX(VertexAttrib3svNV): popq %rbp popq %rsi popq %rdi - movq 5584(%rax), %r11 + movq 5552(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttrib3svNV), .-GL_PREFIX(VertexAttrib3svNV) @@ -26323,7 +26143,7 @@ GL_PREFIX(VertexAttrib3svNV): GL_PREFIX(VertexAttrib4dNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5592(%rax), %r11 + movq 5560(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) subq $40, %rsp @@ -26339,13 +26159,13 @@ GL_PREFIX(VertexAttrib4dNV): movq 8(%rsp), %xmm0 movq (%rsp), %rdi addq $40, %rsp - movq 5592(%rax), %r11 + movq 5560(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5592(%rax), %r11 + movq 5560(%rax), %r11 jmp *%r11 1: subq $40, %rsp @@ -26361,7 +26181,7 @@ GL_PREFIX(VertexAttrib4dNV): movq 8(%rsp), %xmm0 movq (%rsp), %rdi addq $40, %rsp - movq 5592(%rax), %r11 + movq 5560(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttrib4dNV), .-GL_PREFIX(VertexAttrib4dNV) @@ -26372,7 +26192,7 @@ GL_PREFIX(VertexAttrib4dNV): GL_PREFIX(VertexAttrib4dvNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5600(%rax), %r11 + movq 5568(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -26382,13 +26202,13 @@ GL_PREFIX(VertexAttrib4dvNV): popq %rbp popq %rsi popq %rdi - movq 5600(%rax), %r11 + movq 5568(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5600(%rax), %r11 + movq 5568(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -26398,7 +26218,7 @@ GL_PREFIX(VertexAttrib4dvNV): popq %rbp popq %rsi popq %rdi - movq 5600(%rax), %r11 + movq 5568(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttrib4dvNV), .-GL_PREFIX(VertexAttrib4dvNV) @@ -26409,7 +26229,7 @@ GL_PREFIX(VertexAttrib4dvNV): GL_PREFIX(VertexAttrib4fNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5608(%rax), %r11 + movq 5576(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) subq $40, %rsp @@ -26425,13 +26245,13 @@ GL_PREFIX(VertexAttrib4fNV): movq 8(%rsp), %xmm0 movq (%rsp), %rdi addq $40, %rsp - movq 5608(%rax), %r11 + movq 5576(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5608(%rax), %r11 + movq 5576(%rax), %r11 jmp *%r11 1: subq $40, %rsp @@ -26447,7 +26267,7 @@ GL_PREFIX(VertexAttrib4fNV): movq 8(%rsp), %xmm0 movq (%rsp), %rdi addq $40, %rsp - movq 5608(%rax), %r11 + movq 5576(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttrib4fNV), .-GL_PREFIX(VertexAttrib4fNV) @@ -26458,7 +26278,7 @@ GL_PREFIX(VertexAttrib4fNV): GL_PREFIX(VertexAttrib4fvNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5616(%rax), %r11 + movq 5584(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -26468,13 +26288,13 @@ GL_PREFIX(VertexAttrib4fvNV): popq %rbp popq %rsi popq %rdi - movq 5616(%rax), %r11 + movq 5584(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5616(%rax), %r11 + movq 5584(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -26484,7 +26304,7 @@ GL_PREFIX(VertexAttrib4fvNV): popq %rbp popq %rsi popq %rdi - movq 5616(%rax), %r11 + movq 5584(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttrib4fvNV), .-GL_PREFIX(VertexAttrib4fvNV) @@ -26495,7 +26315,7 @@ GL_PREFIX(VertexAttrib4fvNV): GL_PREFIX(VertexAttrib4sNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5624(%rax), %r11 + movq 5592(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -26509,13 +26329,13 @@ GL_PREFIX(VertexAttrib4sNV): popq %rdx popq %rsi popq %rdi - movq 5624(%rax), %r11 + movq 5592(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5624(%rax), %r11 + movq 5592(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -26529,7 +26349,7 @@ GL_PREFIX(VertexAttrib4sNV): popq %rdx popq %rsi popq %rdi - movq 5624(%rax), %r11 + movq 5592(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttrib4sNV), .-GL_PREFIX(VertexAttrib4sNV) @@ -26540,7 +26360,7 @@ GL_PREFIX(VertexAttrib4sNV): GL_PREFIX(VertexAttrib4svNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5632(%rax), %r11 + movq 5600(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -26550,13 +26370,13 @@ GL_PREFIX(VertexAttrib4svNV): popq %rbp popq %rsi popq %rdi - movq 5632(%rax), %r11 + movq 5600(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5632(%rax), %r11 + movq 5600(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -26566,7 +26386,7 @@ GL_PREFIX(VertexAttrib4svNV): popq %rbp popq %rsi popq %rdi - movq 5632(%rax), %r11 + movq 5600(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttrib4svNV), .-GL_PREFIX(VertexAttrib4svNV) @@ -26577,7 +26397,7 @@ GL_PREFIX(VertexAttrib4svNV): GL_PREFIX(VertexAttrib4ubNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5640(%rax), %r11 + movq 5608(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -26591,13 +26411,13 @@ GL_PREFIX(VertexAttrib4ubNV): popq %rdx popq %rsi popq %rdi - movq 5640(%rax), %r11 + movq 5608(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5640(%rax), %r11 + movq 5608(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -26611,7 +26431,7 @@ GL_PREFIX(VertexAttrib4ubNV): popq %rdx popq %rsi popq %rdi - movq 5640(%rax), %r11 + movq 5608(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttrib4ubNV), .-GL_PREFIX(VertexAttrib4ubNV) @@ -26622,7 +26442,7 @@ GL_PREFIX(VertexAttrib4ubNV): GL_PREFIX(VertexAttrib4ubvNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5648(%rax), %r11 + movq 5616(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -26632,13 +26452,13 @@ GL_PREFIX(VertexAttrib4ubvNV): popq %rbp popq %rsi popq %rdi - movq 5648(%rax), %r11 + movq 5616(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5648(%rax), %r11 + movq 5616(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -26648,7 +26468,7 @@ GL_PREFIX(VertexAttrib4ubvNV): popq %rbp popq %rsi popq %rdi - movq 5648(%rax), %r11 + movq 5616(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttrib4ubvNV), .-GL_PREFIX(VertexAttrib4ubvNV) @@ -26659,7 +26479,7 @@ GL_PREFIX(VertexAttrib4ubvNV): GL_PREFIX(VertexAttribPointerNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5656(%rax), %r11 + movq 5624(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -26673,13 +26493,13 @@ GL_PREFIX(VertexAttribPointerNV): popq %rdx popq %rsi popq %rdi - movq 5656(%rax), %r11 + movq 5624(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5656(%rax), %r11 + movq 5624(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -26693,7 +26513,7 @@ GL_PREFIX(VertexAttribPointerNV): popq %rdx popq %rsi popq %rdi - movq 5656(%rax), %r11 + movq 5624(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttribPointerNV), .-GL_PREFIX(VertexAttribPointerNV) @@ -26704,7 +26524,7 @@ GL_PREFIX(VertexAttribPointerNV): GL_PREFIX(VertexAttribs1dvNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5664(%rax), %r11 + movq 5632(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -26714,13 +26534,13 @@ GL_PREFIX(VertexAttribs1dvNV): popq %rdx popq %rsi popq %rdi - movq 5664(%rax), %r11 + movq 5632(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5664(%rax), %r11 + movq 5632(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -26730,7 +26550,7 @@ GL_PREFIX(VertexAttribs1dvNV): popq %rdx popq %rsi popq %rdi - movq 5664(%rax), %r11 + movq 5632(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttribs1dvNV), .-GL_PREFIX(VertexAttribs1dvNV) @@ -26741,7 +26561,7 @@ GL_PREFIX(VertexAttribs1dvNV): GL_PREFIX(VertexAttribs1fvNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5672(%rax), %r11 + movq 5640(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -26751,13 +26571,13 @@ GL_PREFIX(VertexAttribs1fvNV): popq %rdx popq %rsi popq %rdi - movq 5672(%rax), %r11 + movq 5640(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5672(%rax), %r11 + movq 5640(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -26767,7 +26587,7 @@ GL_PREFIX(VertexAttribs1fvNV): popq %rdx popq %rsi popq %rdi - movq 5672(%rax), %r11 + movq 5640(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttribs1fvNV), .-GL_PREFIX(VertexAttribs1fvNV) @@ -26778,7 +26598,7 @@ GL_PREFIX(VertexAttribs1fvNV): GL_PREFIX(VertexAttribs1svNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5680(%rax), %r11 + movq 5648(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -26788,13 +26608,13 @@ GL_PREFIX(VertexAttribs1svNV): popq %rdx popq %rsi popq %rdi - movq 5680(%rax), %r11 + movq 5648(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5680(%rax), %r11 + movq 5648(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -26804,7 +26624,7 @@ GL_PREFIX(VertexAttribs1svNV): popq %rdx popq %rsi popq %rdi - movq 5680(%rax), %r11 + movq 5648(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttribs1svNV), .-GL_PREFIX(VertexAttribs1svNV) @@ -26815,7 +26635,7 @@ GL_PREFIX(VertexAttribs1svNV): GL_PREFIX(VertexAttribs2dvNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5688(%rax), %r11 + movq 5656(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -26825,13 +26645,13 @@ GL_PREFIX(VertexAttribs2dvNV): popq %rdx popq %rsi popq %rdi - movq 5688(%rax), %r11 + movq 5656(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5688(%rax), %r11 + movq 5656(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -26841,7 +26661,7 @@ GL_PREFIX(VertexAttribs2dvNV): popq %rdx popq %rsi popq %rdi - movq 5688(%rax), %r11 + movq 5656(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttribs2dvNV), .-GL_PREFIX(VertexAttribs2dvNV) @@ -26852,7 +26672,7 @@ GL_PREFIX(VertexAttribs2dvNV): GL_PREFIX(VertexAttribs2fvNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5696(%rax), %r11 + movq 5664(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -26862,13 +26682,13 @@ GL_PREFIX(VertexAttribs2fvNV): popq %rdx popq %rsi popq %rdi - movq 5696(%rax), %r11 + movq 5664(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5696(%rax), %r11 + movq 5664(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -26878,7 +26698,7 @@ GL_PREFIX(VertexAttribs2fvNV): popq %rdx popq %rsi popq %rdi - movq 5696(%rax), %r11 + movq 5664(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttribs2fvNV), .-GL_PREFIX(VertexAttribs2fvNV) @@ -26889,7 +26709,7 @@ GL_PREFIX(VertexAttribs2fvNV): GL_PREFIX(VertexAttribs2svNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5704(%rax), %r11 + movq 5672(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -26899,13 +26719,13 @@ GL_PREFIX(VertexAttribs2svNV): popq %rdx popq %rsi popq %rdi - movq 5704(%rax), %r11 + movq 5672(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5704(%rax), %r11 + movq 5672(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -26915,7 +26735,7 @@ GL_PREFIX(VertexAttribs2svNV): popq %rdx popq %rsi popq %rdi - movq 5704(%rax), %r11 + movq 5672(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttribs2svNV), .-GL_PREFIX(VertexAttribs2svNV) @@ -26926,7 +26746,7 @@ GL_PREFIX(VertexAttribs2svNV): GL_PREFIX(VertexAttribs3dvNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5712(%rax), %r11 + movq 5680(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -26936,13 +26756,13 @@ GL_PREFIX(VertexAttribs3dvNV): popq %rdx popq %rsi popq %rdi - movq 5712(%rax), %r11 + movq 5680(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5712(%rax), %r11 + movq 5680(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -26952,7 +26772,7 @@ GL_PREFIX(VertexAttribs3dvNV): popq %rdx popq %rsi popq %rdi - movq 5712(%rax), %r11 + movq 5680(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttribs3dvNV), .-GL_PREFIX(VertexAttribs3dvNV) @@ -26963,7 +26783,7 @@ GL_PREFIX(VertexAttribs3dvNV): GL_PREFIX(VertexAttribs3fvNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5720(%rax), %r11 + movq 5688(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -26973,13 +26793,13 @@ GL_PREFIX(VertexAttribs3fvNV): popq %rdx popq %rsi popq %rdi - movq 5720(%rax), %r11 + movq 5688(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5720(%rax), %r11 + movq 5688(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -26989,7 +26809,7 @@ GL_PREFIX(VertexAttribs3fvNV): popq %rdx popq %rsi popq %rdi - movq 5720(%rax), %r11 + movq 5688(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttribs3fvNV), .-GL_PREFIX(VertexAttribs3fvNV) @@ -27000,7 +26820,7 @@ GL_PREFIX(VertexAttribs3fvNV): GL_PREFIX(VertexAttribs3svNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5728(%rax), %r11 + movq 5696(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -27010,13 +26830,13 @@ GL_PREFIX(VertexAttribs3svNV): popq %rdx popq %rsi popq %rdi - movq 5728(%rax), %r11 + movq 5696(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5728(%rax), %r11 + movq 5696(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -27026,7 +26846,7 @@ GL_PREFIX(VertexAttribs3svNV): popq %rdx popq %rsi popq %rdi - movq 5728(%rax), %r11 + movq 5696(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttribs3svNV), .-GL_PREFIX(VertexAttribs3svNV) @@ -27037,7 +26857,7 @@ GL_PREFIX(VertexAttribs3svNV): GL_PREFIX(VertexAttribs4dvNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5736(%rax), %r11 + movq 5704(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -27047,13 +26867,13 @@ GL_PREFIX(VertexAttribs4dvNV): popq %rdx popq %rsi popq %rdi - movq 5736(%rax), %r11 + movq 5704(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5736(%rax), %r11 + movq 5704(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -27063,7 +26883,7 @@ GL_PREFIX(VertexAttribs4dvNV): popq %rdx popq %rsi popq %rdi - movq 5736(%rax), %r11 + movq 5704(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttribs4dvNV), .-GL_PREFIX(VertexAttribs4dvNV) @@ -27074,7 +26894,7 @@ GL_PREFIX(VertexAttribs4dvNV): GL_PREFIX(VertexAttribs4fvNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5744(%rax), %r11 + movq 5712(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -27084,13 +26904,13 @@ GL_PREFIX(VertexAttribs4fvNV): popq %rdx popq %rsi popq %rdi - movq 5744(%rax), %r11 + movq 5712(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5744(%rax), %r11 + movq 5712(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -27100,7 +26920,7 @@ GL_PREFIX(VertexAttribs4fvNV): popq %rdx popq %rsi popq %rdi - movq 5744(%rax), %r11 + movq 5712(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttribs4fvNV), .-GL_PREFIX(VertexAttribs4fvNV) @@ -27111,7 +26931,7 @@ GL_PREFIX(VertexAttribs4fvNV): GL_PREFIX(VertexAttribs4svNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5752(%rax), %r11 + movq 5720(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -27121,13 +26941,13 @@ GL_PREFIX(VertexAttribs4svNV): popq %rdx popq %rsi popq %rdi - movq 5752(%rax), %r11 + movq 5720(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5752(%rax), %r11 + movq 5720(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -27137,7 +26957,7 @@ GL_PREFIX(VertexAttribs4svNV): popq %rdx popq %rsi popq %rdi - movq 5752(%rax), %r11 + movq 5720(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttribs4svNV), .-GL_PREFIX(VertexAttribs4svNV) @@ -27148,7 +26968,7 @@ GL_PREFIX(VertexAttribs4svNV): GL_PREFIX(VertexAttribs4ubvNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5760(%rax), %r11 + movq 5728(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -27158,13 +26978,13 @@ GL_PREFIX(VertexAttribs4ubvNV): popq %rdx popq %rsi popq %rdi - movq 5760(%rax), %r11 + movq 5728(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5760(%rax), %r11 + movq 5728(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -27174,7 +26994,7 @@ GL_PREFIX(VertexAttribs4ubvNV): popq %rdx popq %rsi popq %rdi - movq 5760(%rax), %r11 + movq 5728(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(VertexAttribs4ubvNV), .-GL_PREFIX(VertexAttribs4ubvNV) @@ -27185,7 +27005,7 @@ GL_PREFIX(VertexAttribs4ubvNV): GL_PREFIX(AlphaFragmentOp1ATI): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5768(%rax), %r11 + movq 5736(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -27203,13 +27023,13 @@ GL_PREFIX(AlphaFragmentOp1ATI): popq %rdx popq %rsi popq %rdi - movq 5768(%rax), %r11 + movq 5736(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5768(%rax), %r11 + movq 5736(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -27227,7 +27047,7 @@ GL_PREFIX(AlphaFragmentOp1ATI): popq %rdx popq %rsi popq %rdi - movq 5768(%rax), %r11 + movq 5736(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(AlphaFragmentOp1ATI), .-GL_PREFIX(AlphaFragmentOp1ATI) @@ -27238,7 +27058,7 @@ GL_PREFIX(AlphaFragmentOp1ATI): GL_PREFIX(AlphaFragmentOp2ATI): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5776(%rax), %r11 + movq 5744(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -27256,13 +27076,13 @@ GL_PREFIX(AlphaFragmentOp2ATI): popq %rdx popq %rsi popq %rdi - movq 5776(%rax), %r11 + movq 5744(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5776(%rax), %r11 + movq 5744(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -27280,7 +27100,7 @@ GL_PREFIX(AlphaFragmentOp2ATI): popq %rdx popq %rsi popq %rdi - movq 5776(%rax), %r11 + movq 5744(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(AlphaFragmentOp2ATI), .-GL_PREFIX(AlphaFragmentOp2ATI) @@ -27291,7 +27111,7 @@ GL_PREFIX(AlphaFragmentOp2ATI): GL_PREFIX(AlphaFragmentOp3ATI): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5784(%rax), %r11 + movq 5752(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -27309,13 +27129,13 @@ GL_PREFIX(AlphaFragmentOp3ATI): popq %rdx popq %rsi popq %rdi - movq 5784(%rax), %r11 + movq 5752(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5784(%rax), %r11 + movq 5752(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -27333,7 +27153,7 @@ GL_PREFIX(AlphaFragmentOp3ATI): popq %rdx popq %rsi popq %rdi - movq 5784(%rax), %r11 + movq 5752(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(AlphaFragmentOp3ATI), .-GL_PREFIX(AlphaFragmentOp3ATI) @@ -27344,25 +27164,25 @@ GL_PREFIX(AlphaFragmentOp3ATI): GL_PREFIX(BeginFragmentShaderATI): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5792(%rax), %r11 + movq 5760(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rbp call _x86_64_get_dispatch@PLT popq %rbp - movq 5792(%rax), %r11 + movq 5760(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5792(%rax), %r11 + movq 5760(%rax), %r11 jmp *%r11 1: pushq %rbp call _glapi_get_dispatch popq %rbp - movq 5792(%rax), %r11 + movq 5760(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(BeginFragmentShaderATI), .-GL_PREFIX(BeginFragmentShaderATI) @@ -27373,25 +27193,25 @@ GL_PREFIX(BeginFragmentShaderATI): GL_PREFIX(BindFragmentShaderATI): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5800(%rax), %r11 + movq 5768(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi call _x86_64_get_dispatch@PLT popq %rdi - movq 5800(%rax), %r11 + movq 5768(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5800(%rax), %r11 + movq 5768(%rax), %r11 jmp *%r11 1: pushq %rdi call _glapi_get_dispatch popq %rdi - movq 5800(%rax), %r11 + movq 5768(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(BindFragmentShaderATI), .-GL_PREFIX(BindFragmentShaderATI) @@ -27402,7 +27222,7 @@ GL_PREFIX(BindFragmentShaderATI): GL_PREFIX(ColorFragmentOp1ATI): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5808(%rax), %r11 + movq 5776(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -27420,13 +27240,13 @@ GL_PREFIX(ColorFragmentOp1ATI): popq %rdx popq %rsi popq %rdi - movq 5808(%rax), %r11 + movq 5776(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5808(%rax), %r11 + movq 5776(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -27444,7 +27264,7 @@ GL_PREFIX(ColorFragmentOp1ATI): popq %rdx popq %rsi popq %rdi - movq 5808(%rax), %r11 + movq 5776(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(ColorFragmentOp1ATI), .-GL_PREFIX(ColorFragmentOp1ATI) @@ -27455,7 +27275,7 @@ GL_PREFIX(ColorFragmentOp1ATI): GL_PREFIX(ColorFragmentOp2ATI): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5816(%rax), %r11 + movq 5784(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -27473,13 +27293,13 @@ GL_PREFIX(ColorFragmentOp2ATI): popq %rdx popq %rsi popq %rdi - movq 5816(%rax), %r11 + movq 5784(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5816(%rax), %r11 + movq 5784(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -27497,7 +27317,7 @@ GL_PREFIX(ColorFragmentOp2ATI): popq %rdx popq %rsi popq %rdi - movq 5816(%rax), %r11 + movq 5784(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(ColorFragmentOp2ATI), .-GL_PREFIX(ColorFragmentOp2ATI) @@ -27508,7 +27328,7 @@ GL_PREFIX(ColorFragmentOp2ATI): GL_PREFIX(ColorFragmentOp3ATI): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5824(%rax), %r11 + movq 5792(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -27526,13 +27346,13 @@ GL_PREFIX(ColorFragmentOp3ATI): popq %rdx popq %rsi popq %rdi - movq 5824(%rax), %r11 + movq 5792(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5824(%rax), %r11 + movq 5792(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -27550,7 +27370,7 @@ GL_PREFIX(ColorFragmentOp3ATI): popq %rdx popq %rsi popq %rdi - movq 5824(%rax), %r11 + movq 5792(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(ColorFragmentOp3ATI), .-GL_PREFIX(ColorFragmentOp3ATI) @@ -27561,25 +27381,25 @@ GL_PREFIX(ColorFragmentOp3ATI): GL_PREFIX(DeleteFragmentShaderATI): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5832(%rax), %r11 + movq 5800(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi call _x86_64_get_dispatch@PLT popq %rdi - movq 5832(%rax), %r11 + movq 5800(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5832(%rax), %r11 + movq 5800(%rax), %r11 jmp *%r11 1: pushq %rdi call _glapi_get_dispatch popq %rdi - movq 5832(%rax), %r11 + movq 5800(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(DeleteFragmentShaderATI), .-GL_PREFIX(DeleteFragmentShaderATI) @@ -27590,25 +27410,25 @@ GL_PREFIX(DeleteFragmentShaderATI): GL_PREFIX(EndFragmentShaderATI): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5840(%rax), %r11 + movq 5808(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rbp call _x86_64_get_dispatch@PLT popq %rbp - movq 5840(%rax), %r11 + movq 5808(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5840(%rax), %r11 + movq 5808(%rax), %r11 jmp *%r11 1: pushq %rbp call _glapi_get_dispatch popq %rbp - movq 5840(%rax), %r11 + movq 5808(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(EndFragmentShaderATI), .-GL_PREFIX(EndFragmentShaderATI) @@ -27619,25 +27439,25 @@ GL_PREFIX(EndFragmentShaderATI): GL_PREFIX(GenFragmentShadersATI): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5848(%rax), %r11 + movq 5816(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi call _x86_64_get_dispatch@PLT popq %rdi - movq 5848(%rax), %r11 + movq 5816(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5848(%rax), %r11 + movq 5816(%rax), %r11 jmp *%r11 1: pushq %rdi call _glapi_get_dispatch popq %rdi - movq 5848(%rax), %r11 + movq 5816(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(GenFragmentShadersATI), .-GL_PREFIX(GenFragmentShadersATI) @@ -27648,7 +27468,7 @@ GL_PREFIX(GenFragmentShadersATI): GL_PREFIX(PassTexCoordATI): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5856(%rax), %r11 + movq 5824(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -27658,13 +27478,13 @@ GL_PREFIX(PassTexCoordATI): popq %rdx popq %rsi popq %rdi - movq 5856(%rax), %r11 + movq 5824(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5856(%rax), %r11 + movq 5824(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -27674,7 +27494,7 @@ GL_PREFIX(PassTexCoordATI): popq %rdx popq %rsi popq %rdi - movq 5856(%rax), %r11 + movq 5824(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(PassTexCoordATI), .-GL_PREFIX(PassTexCoordATI) @@ -27685,7 +27505,7 @@ GL_PREFIX(PassTexCoordATI): GL_PREFIX(SampleMapATI): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5864(%rax), %r11 + movq 5832(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -27695,13 +27515,13 @@ GL_PREFIX(SampleMapATI): popq %rdx popq %rsi popq %rdi - movq 5864(%rax), %r11 + movq 5832(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5864(%rax), %r11 + movq 5832(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -27711,7 +27531,7 @@ GL_PREFIX(SampleMapATI): popq %rdx popq %rsi popq %rdi - movq 5864(%rax), %r11 + movq 5832(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(SampleMapATI), .-GL_PREFIX(SampleMapATI) @@ -27722,7 +27542,7 @@ GL_PREFIX(SampleMapATI): GL_PREFIX(SetFragmentShaderConstantATI): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5872(%rax), %r11 + movq 5840(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -27732,13 +27552,13 @@ GL_PREFIX(SetFragmentShaderConstantATI): popq %rbp popq %rsi popq %rdi - movq 5872(%rax), %r11 + movq 5840(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5872(%rax), %r11 + movq 5840(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -27748,7 +27568,7 @@ GL_PREFIX(SetFragmentShaderConstantATI): popq %rbp popq %rsi popq %rdi - movq 5872(%rax), %r11 + movq 5840(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(SetFragmentShaderConstantATI), .-GL_PREFIX(SetFragmentShaderConstantATI) @@ -27759,7 +27579,7 @@ GL_PREFIX(SetFragmentShaderConstantATI): GL_PREFIX(PointParameteriNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5880(%rax), %r11 + movq 5848(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -27769,13 +27589,13 @@ GL_PREFIX(PointParameteriNV): popq %rbp popq %rsi popq %rdi - movq 5880(%rax), %r11 + movq 5848(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5880(%rax), %r11 + movq 5848(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -27785,7 +27605,7 @@ GL_PREFIX(PointParameteriNV): popq %rbp popq %rsi popq %rdi - movq 5880(%rax), %r11 + movq 5848(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(PointParameteriNV), .-GL_PREFIX(PointParameteriNV) @@ -27796,7 +27616,7 @@ GL_PREFIX(PointParameteriNV): GL_PREFIX(PointParameterivNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5888(%rax), %r11 + movq 5856(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -27806,13 +27626,13 @@ GL_PREFIX(PointParameterivNV): popq %rbp popq %rsi popq %rdi - movq 5888(%rax), %r11 + movq 5856(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5888(%rax), %r11 + movq 5856(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -27822,79 +27642,79 @@ GL_PREFIX(PointParameterivNV): popq %rbp popq %rsi popq %rdi - movq 5888(%rax), %r11 + movq 5856(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(PointParameterivNV), .-GL_PREFIX(PointParameterivNV) .p2align 4,,15 - .globl GL_PREFIX(_dispatch_stub_737) - .type GL_PREFIX(_dispatch_stub_737), @function - HIDDEN(GL_PREFIX(_dispatch_stub_737)) -GL_PREFIX(_dispatch_stub_737): + .globl GL_PREFIX(_dispatch_stub_733) + .type GL_PREFIX(_dispatch_stub_733), @function + HIDDEN(GL_PREFIX(_dispatch_stub_733)) +GL_PREFIX(_dispatch_stub_733): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5896(%rax), %r11 + movq 5864(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi call _x86_64_get_dispatch@PLT popq %rdi - movq 5896(%rax), %r11 + movq 5864(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5896(%rax), %r11 + movq 5864(%rax), %r11 jmp *%r11 1: pushq %rdi call _glapi_get_dispatch popq %rdi - movq 5896(%rax), %r11 + movq 5864(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ - .size GL_PREFIX(_dispatch_stub_737), .-GL_PREFIX(_dispatch_stub_737) + .size GL_PREFIX(_dispatch_stub_733), .-GL_PREFIX(_dispatch_stub_733) .p2align 4,,15 - .globl GL_PREFIX(_dispatch_stub_738) - .type GL_PREFIX(_dispatch_stub_738), @function - HIDDEN(GL_PREFIX(_dispatch_stub_738)) -GL_PREFIX(_dispatch_stub_738): + .globl GL_PREFIX(_dispatch_stub_734) + .type GL_PREFIX(_dispatch_stub_734), @function + HIDDEN(GL_PREFIX(_dispatch_stub_734)) +GL_PREFIX(_dispatch_stub_734): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5904(%rax), %r11 + movq 5872(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi call _x86_64_get_dispatch@PLT popq %rdi - movq 5904(%rax), %r11 + movq 5872(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5904(%rax), %r11 + movq 5872(%rax), %r11 jmp *%r11 1: pushq %rdi call _glapi_get_dispatch popq %rdi - movq 5904(%rax), %r11 + movq 5872(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ - .size GL_PREFIX(_dispatch_stub_738), .-GL_PREFIX(_dispatch_stub_738) + .size GL_PREFIX(_dispatch_stub_734), .-GL_PREFIX(_dispatch_stub_734) .p2align 4,,15 - .globl GL_PREFIX(_dispatch_stub_739) - .type GL_PREFIX(_dispatch_stub_739), @function - HIDDEN(GL_PREFIX(_dispatch_stub_739)) -GL_PREFIX(_dispatch_stub_739): + .globl GL_PREFIX(_dispatch_stub_735) + .type GL_PREFIX(_dispatch_stub_735), @function + HIDDEN(GL_PREFIX(_dispatch_stub_735)) +GL_PREFIX(_dispatch_stub_735): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5912(%rax), %r11 + movq 5880(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -27904,13 +27724,13 @@ GL_PREFIX(_dispatch_stub_739): popq %rbp popq %rsi popq %rdi - movq 5912(%rax), %r11 + movq 5880(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5912(%rax), %r11 + movq 5880(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -27920,19 +27740,19 @@ GL_PREFIX(_dispatch_stub_739): popq %rbp popq %rsi popq %rdi - movq 5912(%rax), %r11 + movq 5880(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ - .size GL_PREFIX(_dispatch_stub_739), .-GL_PREFIX(_dispatch_stub_739) + .size GL_PREFIX(_dispatch_stub_735), .-GL_PREFIX(_dispatch_stub_735) .p2align 4,,15 - .globl GL_PREFIX(_dispatch_stub_740) - .type GL_PREFIX(_dispatch_stub_740), @function - HIDDEN(GL_PREFIX(_dispatch_stub_740)) -GL_PREFIX(_dispatch_stub_740): + .globl GL_PREFIX(_dispatch_stub_736) + .type GL_PREFIX(_dispatch_stub_736), @function + HIDDEN(GL_PREFIX(_dispatch_stub_736)) +GL_PREFIX(_dispatch_stub_736): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5920(%rax), %r11 + movq 5888(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -27942,13 +27762,13 @@ GL_PREFIX(_dispatch_stub_740): popq %rbp popq %rsi popq %rdi - movq 5920(%rax), %r11 + movq 5888(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5920(%rax), %r11 + movq 5888(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -27958,40 +27778,40 @@ GL_PREFIX(_dispatch_stub_740): popq %rbp popq %rsi popq %rdi - movq 5920(%rax), %r11 + movq 5888(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ - .size GL_PREFIX(_dispatch_stub_740), .-GL_PREFIX(_dispatch_stub_740) + .size GL_PREFIX(_dispatch_stub_736), .-GL_PREFIX(_dispatch_stub_736) .p2align 4,,15 - .globl GL_PREFIX(_dispatch_stub_741) - .type GL_PREFIX(_dispatch_stub_741), @function - HIDDEN(GL_PREFIX(_dispatch_stub_741)) -GL_PREFIX(_dispatch_stub_741): + .globl GL_PREFIX(_dispatch_stub_737) + .type GL_PREFIX(_dispatch_stub_737), @function + HIDDEN(GL_PREFIX(_dispatch_stub_737)) +GL_PREFIX(_dispatch_stub_737): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5928(%rax), %r11 + movq 5896(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi call _x86_64_get_dispatch@PLT popq %rdi - movq 5928(%rax), %r11 + movq 5896(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5928(%rax), %r11 + movq 5896(%rax), %r11 jmp *%r11 1: pushq %rdi call _glapi_get_dispatch popq %rdi - movq 5928(%rax), %r11 + movq 5896(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ - .size GL_PREFIX(_dispatch_stub_741), .-GL_PREFIX(_dispatch_stub_741) + .size GL_PREFIX(_dispatch_stub_737), .-GL_PREFIX(_dispatch_stub_737) .p2align 4,,15 .globl GL_PREFIX(GetProgramNamedParameterdvNV) @@ -27999,7 +27819,7 @@ GL_PREFIX(_dispatch_stub_741): GL_PREFIX(GetProgramNamedParameterdvNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5936(%rax), %r11 + movq 5904(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -28013,13 +27833,13 @@ GL_PREFIX(GetProgramNamedParameterdvNV): popq %rdx popq %rsi popq %rdi - movq 5936(%rax), %r11 + movq 5904(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5936(%rax), %r11 + movq 5904(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -28033,7 +27853,7 @@ GL_PREFIX(GetProgramNamedParameterdvNV): popq %rdx popq %rsi popq %rdi - movq 5936(%rax), %r11 + movq 5904(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(GetProgramNamedParameterdvNV), .-GL_PREFIX(GetProgramNamedParameterdvNV) @@ -28044,7 +27864,7 @@ GL_PREFIX(GetProgramNamedParameterdvNV): GL_PREFIX(GetProgramNamedParameterfvNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5944(%rax), %r11 + movq 5912(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -28058,13 +27878,13 @@ GL_PREFIX(GetProgramNamedParameterfvNV): popq %rdx popq %rsi popq %rdi - movq 5944(%rax), %r11 + movq 5912(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5944(%rax), %r11 + movq 5912(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -28078,7 +27898,7 @@ GL_PREFIX(GetProgramNamedParameterfvNV): popq %rdx popq %rsi popq %rdi - movq 5944(%rax), %r11 + movq 5912(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(GetProgramNamedParameterfvNV), .-GL_PREFIX(GetProgramNamedParameterfvNV) @@ -28089,7 +27909,7 @@ GL_PREFIX(GetProgramNamedParameterfvNV): GL_PREFIX(ProgramNamedParameter4dNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5952(%rax), %r11 + movq 5920(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) subq $56, %rsp @@ -28109,13 +27929,13 @@ GL_PREFIX(ProgramNamedParameter4dNV): movq 8(%rsp), %rsi movq (%rsp), %rdi addq $56, %rsp - movq 5952(%rax), %r11 + movq 5920(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5952(%rax), %r11 + movq 5920(%rax), %r11 jmp *%r11 1: subq $56, %rsp @@ -28135,7 +27955,7 @@ GL_PREFIX(ProgramNamedParameter4dNV): movq 8(%rsp), %rsi movq (%rsp), %rdi addq $56, %rsp - movq 5952(%rax), %r11 + movq 5920(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(ProgramNamedParameter4dNV), .-GL_PREFIX(ProgramNamedParameter4dNV) @@ -28146,7 +27966,7 @@ GL_PREFIX(ProgramNamedParameter4dNV): GL_PREFIX(ProgramNamedParameter4dvNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5960(%rax), %r11 + movq 5928(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -28160,13 +27980,13 @@ GL_PREFIX(ProgramNamedParameter4dvNV): popq %rdx popq %rsi popq %rdi - movq 5960(%rax), %r11 + movq 5928(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5960(%rax), %r11 + movq 5928(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -28180,7 +28000,7 @@ GL_PREFIX(ProgramNamedParameter4dvNV): popq %rdx popq %rsi popq %rdi - movq 5960(%rax), %r11 + movq 5928(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(ProgramNamedParameter4dvNV), .-GL_PREFIX(ProgramNamedParameter4dvNV) @@ -28191,7 +28011,7 @@ GL_PREFIX(ProgramNamedParameter4dvNV): GL_PREFIX(ProgramNamedParameter4fNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5968(%rax), %r11 + movq 5936(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) subq $56, %rsp @@ -28211,13 +28031,13 @@ GL_PREFIX(ProgramNamedParameter4fNV): movq 8(%rsp), %rsi movq (%rsp), %rdi addq $56, %rsp - movq 5968(%rax), %r11 + movq 5936(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5968(%rax), %r11 + movq 5936(%rax), %r11 jmp *%r11 1: subq $56, %rsp @@ -28237,7 +28057,7 @@ GL_PREFIX(ProgramNamedParameter4fNV): movq 8(%rsp), %rsi movq (%rsp), %rdi addq $56, %rsp - movq 5968(%rax), %r11 + movq 5936(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(ProgramNamedParameter4fNV), .-GL_PREFIX(ProgramNamedParameter4fNV) @@ -28248,7 +28068,7 @@ GL_PREFIX(ProgramNamedParameter4fNV): GL_PREFIX(ProgramNamedParameter4fvNV): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5976(%rax), %r11 + movq 5944(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -28262,13 +28082,13 @@ GL_PREFIX(ProgramNamedParameter4fvNV): popq %rdx popq %rsi popq %rdi - movq 5976(%rax), %r11 + movq 5944(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5976(%rax), %r11 + movq 5944(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -28282,19 +28102,19 @@ GL_PREFIX(ProgramNamedParameter4fvNV): popq %rdx popq %rsi popq %rdi - movq 5976(%rax), %r11 + movq 5944(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(ProgramNamedParameter4fvNV), .-GL_PREFIX(ProgramNamedParameter4fvNV) .p2align 4,,15 - .globl GL_PREFIX(_dispatch_stub_748) - .type GL_PREFIX(_dispatch_stub_748), @function - HIDDEN(GL_PREFIX(_dispatch_stub_748)) -GL_PREFIX(_dispatch_stub_748): + .globl GL_PREFIX(_dispatch_stub_744) + .type GL_PREFIX(_dispatch_stub_744), @function + HIDDEN(GL_PREFIX(_dispatch_stub_744)) +GL_PREFIX(_dispatch_stub_744): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5984(%rax), %r11 + movq 5952(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -28304,13 +28124,13 @@ GL_PREFIX(_dispatch_stub_748): popq %rbp popq %rsi popq %rdi - movq 5984(%rax), %r11 + movq 5952(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5984(%rax), %r11 + movq 5952(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -28320,19 +28140,19 @@ GL_PREFIX(_dispatch_stub_748): popq %rbp popq %rsi popq %rdi - movq 5984(%rax), %r11 + movq 5952(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ - .size GL_PREFIX(_dispatch_stub_748), .-GL_PREFIX(_dispatch_stub_748) + .size GL_PREFIX(_dispatch_stub_744), .-GL_PREFIX(_dispatch_stub_744) .p2align 4,,15 - .globl GL_PREFIX(_dispatch_stub_749) - .type GL_PREFIX(_dispatch_stub_749), @function - HIDDEN(GL_PREFIX(_dispatch_stub_749)) -GL_PREFIX(_dispatch_stub_749): + .globl GL_PREFIX(_dispatch_stub_745) + .type GL_PREFIX(_dispatch_stub_745), @function + HIDDEN(GL_PREFIX(_dispatch_stub_745)) +GL_PREFIX(_dispatch_stub_745): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 5992(%rax), %r11 + movq 5960(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -28342,13 +28162,13 @@ GL_PREFIX(_dispatch_stub_749): popq %rbp popq %rsi popq %rdi - movq 5992(%rax), %r11 + movq 5960(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 5992(%rax), %r11 + movq 5960(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -28358,10 +28178,10 @@ GL_PREFIX(_dispatch_stub_749): popq %rbp popq %rsi popq %rdi - movq 5992(%rax), %r11 + movq 5960(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ - .size GL_PREFIX(_dispatch_stub_749), .-GL_PREFIX(_dispatch_stub_749) + .size GL_PREFIX(_dispatch_stub_745), .-GL_PREFIX(_dispatch_stub_745) .p2align 4,,15 .globl GL_PREFIX(BindFramebufferEXT) @@ -28369,7 +28189,7 @@ GL_PREFIX(_dispatch_stub_749): GL_PREFIX(BindFramebufferEXT): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 6000(%rax), %r11 + movq 5968(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -28379,13 +28199,13 @@ GL_PREFIX(BindFramebufferEXT): popq %rbp popq %rsi popq %rdi - movq 6000(%rax), %r11 + movq 5968(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 6000(%rax), %r11 + movq 5968(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -28395,7 +28215,7 @@ GL_PREFIX(BindFramebufferEXT): popq %rbp popq %rsi popq %rdi - movq 6000(%rax), %r11 + movq 5968(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(BindFramebufferEXT), .-GL_PREFIX(BindFramebufferEXT) @@ -28406,7 +28226,7 @@ GL_PREFIX(BindFramebufferEXT): GL_PREFIX(BindRenderbufferEXT): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 6008(%rax), %r11 + movq 5976(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -28416,13 +28236,13 @@ GL_PREFIX(BindRenderbufferEXT): popq %rbp popq %rsi popq %rdi - movq 6008(%rax), %r11 + movq 5976(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 6008(%rax), %r11 + movq 5976(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -28432,7 +28252,7 @@ GL_PREFIX(BindRenderbufferEXT): popq %rbp popq %rsi popq %rdi - movq 6008(%rax), %r11 + movq 5976(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(BindRenderbufferEXT), .-GL_PREFIX(BindRenderbufferEXT) @@ -28443,25 +28263,25 @@ GL_PREFIX(BindRenderbufferEXT): GL_PREFIX(CheckFramebufferStatusEXT): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 6016(%rax), %r11 + movq 5984(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi call _x86_64_get_dispatch@PLT popq %rdi - movq 6016(%rax), %r11 + movq 5984(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 6016(%rax), %r11 + movq 5984(%rax), %r11 jmp *%r11 1: pushq %rdi call _glapi_get_dispatch popq %rdi - movq 6016(%rax), %r11 + movq 5984(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(CheckFramebufferStatusEXT), .-GL_PREFIX(CheckFramebufferStatusEXT) @@ -28472,7 +28292,7 @@ GL_PREFIX(CheckFramebufferStatusEXT): GL_PREFIX(DeleteFramebuffersEXT): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 6024(%rax), %r11 + movq 5992(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -28482,13 +28302,13 @@ GL_PREFIX(DeleteFramebuffersEXT): popq %rbp popq %rsi popq %rdi - movq 6024(%rax), %r11 + movq 5992(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 6024(%rax), %r11 + movq 5992(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -28498,7 +28318,7 @@ GL_PREFIX(DeleteFramebuffersEXT): popq %rbp popq %rsi popq %rdi - movq 6024(%rax), %r11 + movq 5992(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(DeleteFramebuffersEXT), .-GL_PREFIX(DeleteFramebuffersEXT) @@ -28509,7 +28329,7 @@ GL_PREFIX(DeleteFramebuffersEXT): GL_PREFIX(DeleteRenderbuffersEXT): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 6032(%rax), %r11 + movq 6000(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -28519,13 +28339,13 @@ GL_PREFIX(DeleteRenderbuffersEXT): popq %rbp popq %rsi popq %rdi - movq 6032(%rax), %r11 + movq 6000(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 6032(%rax), %r11 + movq 6000(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -28535,7 +28355,7 @@ GL_PREFIX(DeleteRenderbuffersEXT): popq %rbp popq %rsi popq %rdi - movq 6032(%rax), %r11 + movq 6000(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(DeleteRenderbuffersEXT), .-GL_PREFIX(DeleteRenderbuffersEXT) @@ -28546,7 +28366,7 @@ GL_PREFIX(DeleteRenderbuffersEXT): GL_PREFIX(FramebufferRenderbufferEXT): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 6040(%rax), %r11 + movq 6008(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -28560,13 +28380,13 @@ GL_PREFIX(FramebufferRenderbufferEXT): popq %rdx popq %rsi popq %rdi - movq 6040(%rax), %r11 + movq 6008(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 6040(%rax), %r11 + movq 6008(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -28580,7 +28400,7 @@ GL_PREFIX(FramebufferRenderbufferEXT): popq %rdx popq %rsi popq %rdi - movq 6040(%rax), %r11 + movq 6008(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(FramebufferRenderbufferEXT), .-GL_PREFIX(FramebufferRenderbufferEXT) @@ -28591,7 +28411,7 @@ GL_PREFIX(FramebufferRenderbufferEXT): GL_PREFIX(FramebufferTexture1DEXT): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 6048(%rax), %r11 + movq 6016(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -28605,13 +28425,13 @@ GL_PREFIX(FramebufferTexture1DEXT): popq %rdx popq %rsi popq %rdi - movq 6048(%rax), %r11 + movq 6016(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 6048(%rax), %r11 + movq 6016(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -28625,7 +28445,7 @@ GL_PREFIX(FramebufferTexture1DEXT): popq %rdx popq %rsi popq %rdi - movq 6048(%rax), %r11 + movq 6016(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(FramebufferTexture1DEXT), .-GL_PREFIX(FramebufferTexture1DEXT) @@ -28636,7 +28456,7 @@ GL_PREFIX(FramebufferTexture1DEXT): GL_PREFIX(FramebufferTexture2DEXT): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 6056(%rax), %r11 + movq 6024(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -28650,13 +28470,13 @@ GL_PREFIX(FramebufferTexture2DEXT): popq %rdx popq %rsi popq %rdi - movq 6056(%rax), %r11 + movq 6024(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 6056(%rax), %r11 + movq 6024(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -28670,7 +28490,7 @@ GL_PREFIX(FramebufferTexture2DEXT): popq %rdx popq %rsi popq %rdi - movq 6056(%rax), %r11 + movq 6024(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(FramebufferTexture2DEXT), .-GL_PREFIX(FramebufferTexture2DEXT) @@ -28681,7 +28501,7 @@ GL_PREFIX(FramebufferTexture2DEXT): GL_PREFIX(FramebufferTexture3DEXT): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 6064(%rax), %r11 + movq 6032(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -28699,13 +28519,13 @@ GL_PREFIX(FramebufferTexture3DEXT): popq %rdx popq %rsi popq %rdi - movq 6064(%rax), %r11 + movq 6032(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 6064(%rax), %r11 + movq 6032(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -28723,7 +28543,7 @@ GL_PREFIX(FramebufferTexture3DEXT): popq %rdx popq %rsi popq %rdi - movq 6064(%rax), %r11 + movq 6032(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(FramebufferTexture3DEXT), .-GL_PREFIX(FramebufferTexture3DEXT) @@ -28734,7 +28554,7 @@ GL_PREFIX(FramebufferTexture3DEXT): GL_PREFIX(GenFramebuffersEXT): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 6072(%rax), %r11 + movq 6040(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -28744,13 +28564,13 @@ GL_PREFIX(GenFramebuffersEXT): popq %rbp popq %rsi popq %rdi - movq 6072(%rax), %r11 + movq 6040(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 6072(%rax), %r11 + movq 6040(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -28760,7 +28580,7 @@ GL_PREFIX(GenFramebuffersEXT): popq %rbp popq %rsi popq %rdi - movq 6072(%rax), %r11 + movq 6040(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(GenFramebuffersEXT), .-GL_PREFIX(GenFramebuffersEXT) @@ -28771,7 +28591,7 @@ GL_PREFIX(GenFramebuffersEXT): GL_PREFIX(GenRenderbuffersEXT): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 6080(%rax), %r11 + movq 6048(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -28781,13 +28601,13 @@ GL_PREFIX(GenRenderbuffersEXT): popq %rbp popq %rsi popq %rdi - movq 6080(%rax), %r11 + movq 6048(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 6080(%rax), %r11 + movq 6048(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -28797,7 +28617,7 @@ GL_PREFIX(GenRenderbuffersEXT): popq %rbp popq %rsi popq %rdi - movq 6080(%rax), %r11 + movq 6048(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(GenRenderbuffersEXT), .-GL_PREFIX(GenRenderbuffersEXT) @@ -28808,25 +28628,25 @@ GL_PREFIX(GenRenderbuffersEXT): GL_PREFIX(GenerateMipmapEXT): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 6088(%rax), %r11 + movq 6056(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi call _x86_64_get_dispatch@PLT popq %rdi - movq 6088(%rax), %r11 + movq 6056(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 6088(%rax), %r11 + movq 6056(%rax), %r11 jmp *%r11 1: pushq %rdi call _glapi_get_dispatch popq %rdi - movq 6088(%rax), %r11 + movq 6056(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(GenerateMipmapEXT), .-GL_PREFIX(GenerateMipmapEXT) @@ -28837,7 +28657,7 @@ GL_PREFIX(GenerateMipmapEXT): GL_PREFIX(GetFramebufferAttachmentParameterivEXT): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 6096(%rax), %r11 + movq 6064(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -28851,13 +28671,13 @@ GL_PREFIX(GetFramebufferAttachmentParameterivEXT): popq %rdx popq %rsi popq %rdi - movq 6096(%rax), %r11 + movq 6064(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 6096(%rax), %r11 + movq 6064(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -28871,7 +28691,7 @@ GL_PREFIX(GetFramebufferAttachmentParameterivEXT): popq %rdx popq %rsi popq %rdi - movq 6096(%rax), %r11 + movq 6064(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(GetFramebufferAttachmentParameterivEXT), .-GL_PREFIX(GetFramebufferAttachmentParameterivEXT) @@ -28882,7 +28702,7 @@ GL_PREFIX(GetFramebufferAttachmentParameterivEXT): GL_PREFIX(GetRenderbufferParameterivEXT): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 6104(%rax), %r11 + movq 6072(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -28892,13 +28712,13 @@ GL_PREFIX(GetRenderbufferParameterivEXT): popq %rdx popq %rsi popq %rdi - movq 6104(%rax), %r11 + movq 6072(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 6104(%rax), %r11 + movq 6072(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -28908,7 +28728,7 @@ GL_PREFIX(GetRenderbufferParameterivEXT): popq %rdx popq %rsi popq %rdi - movq 6104(%rax), %r11 + movq 6072(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(GetRenderbufferParameterivEXT), .-GL_PREFIX(GetRenderbufferParameterivEXT) @@ -28919,25 +28739,25 @@ GL_PREFIX(GetRenderbufferParameterivEXT): GL_PREFIX(IsFramebufferEXT): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 6112(%rax), %r11 + movq 6080(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi call _x86_64_get_dispatch@PLT popq %rdi - movq 6112(%rax), %r11 + movq 6080(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 6112(%rax), %r11 + movq 6080(%rax), %r11 jmp *%r11 1: pushq %rdi call _glapi_get_dispatch popq %rdi - movq 6112(%rax), %r11 + movq 6080(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(IsFramebufferEXT), .-GL_PREFIX(IsFramebufferEXT) @@ -28948,25 +28768,25 @@ GL_PREFIX(IsFramebufferEXT): GL_PREFIX(IsRenderbufferEXT): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 6120(%rax), %r11 + movq 6088(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi call _x86_64_get_dispatch@PLT popq %rdi - movq 6120(%rax), %r11 + movq 6088(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 6120(%rax), %r11 + movq 6088(%rax), %r11 jmp *%r11 1: pushq %rdi call _glapi_get_dispatch popq %rdi - movq 6120(%rax), %r11 + movq 6088(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(IsRenderbufferEXT), .-GL_PREFIX(IsRenderbufferEXT) @@ -28977,7 +28797,7 @@ GL_PREFIX(IsRenderbufferEXT): GL_PREFIX(RenderbufferStorageEXT): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 6128(%rax), %r11 + movq 6096(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -28991,13 +28811,13 @@ GL_PREFIX(RenderbufferStorageEXT): popq %rdx popq %rsi popq %rdi - movq 6128(%rax), %r11 + movq 6096(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 6128(%rax), %r11 + movq 6096(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -29011,19 +28831,19 @@ GL_PREFIX(RenderbufferStorageEXT): popq %rdx popq %rsi popq %rdi - movq 6128(%rax), %r11 + movq 6096(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(RenderbufferStorageEXT), .-GL_PREFIX(RenderbufferStorageEXT) .p2align 4,,15 - .globl GL_PREFIX(_dispatch_stub_767) - .type GL_PREFIX(_dispatch_stub_767), @function - HIDDEN(GL_PREFIX(_dispatch_stub_767)) -GL_PREFIX(_dispatch_stub_767): + .globl GL_PREFIX(_dispatch_stub_763) + .type GL_PREFIX(_dispatch_stub_763), @function + HIDDEN(GL_PREFIX(_dispatch_stub_763)) +GL_PREFIX(_dispatch_stub_763): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 6136(%rax), %r11 + movq 6104(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -29041,13 +28861,13 @@ GL_PREFIX(_dispatch_stub_767): popq %rdx popq %rsi popq %rdi - movq 6136(%rax), %r11 + movq 6104(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 6136(%rax), %r11 + movq 6104(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -29065,10 +28885,10 @@ GL_PREFIX(_dispatch_stub_767): popq %rdx popq %rsi popq %rdi - movq 6136(%rax), %r11 + movq 6104(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ - .size GL_PREFIX(_dispatch_stub_767), .-GL_PREFIX(_dispatch_stub_767) + .size GL_PREFIX(_dispatch_stub_763), .-GL_PREFIX(_dispatch_stub_763) .p2align 4,,15 .globl GL_PREFIX(FramebufferTextureLayerEXT) @@ -29076,7 +28896,7 @@ GL_PREFIX(_dispatch_stub_767): GL_PREFIX(FramebufferTextureLayerEXT): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 6144(%rax), %r11 + movq 6112(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -29090,13 +28910,13 @@ GL_PREFIX(FramebufferTextureLayerEXT): popq %rdx popq %rsi popq %rdi - movq 6144(%rax), %r11 + movq 6112(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 6144(%rax), %r11 + movq 6112(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -29110,19 +28930,19 @@ GL_PREFIX(FramebufferTextureLayerEXT): popq %rdx popq %rsi popq %rdi - movq 6144(%rax), %r11 + movq 6112(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ .size GL_PREFIX(FramebufferTextureLayerEXT), .-GL_PREFIX(FramebufferTextureLayerEXT) .p2align 4,,15 - .globl GL_PREFIX(_dispatch_stub_769) - .type GL_PREFIX(_dispatch_stub_769), @function - HIDDEN(GL_PREFIX(_dispatch_stub_769)) -GL_PREFIX(_dispatch_stub_769): + .globl GL_PREFIX(_dispatch_stub_765) + .type GL_PREFIX(_dispatch_stub_765), @function + HIDDEN(GL_PREFIX(_dispatch_stub_765)) +GL_PREFIX(_dispatch_stub_765): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 6152(%rax), %r11 + movq 6120(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -29136,13 +28956,13 @@ GL_PREFIX(_dispatch_stub_769): popq %rdx popq %rsi popq %rdi - movq 6152(%rax), %r11 + movq 6120(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 6152(%rax), %r11 + movq 6120(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -29156,19 +28976,19 @@ GL_PREFIX(_dispatch_stub_769): popq %rdx popq %rsi popq %rdi - movq 6152(%rax), %r11 + movq 6120(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ - .size GL_PREFIX(_dispatch_stub_769), .-GL_PREFIX(_dispatch_stub_769) + .size GL_PREFIX(_dispatch_stub_765), .-GL_PREFIX(_dispatch_stub_765) .p2align 4,,15 - .globl GL_PREFIX(_dispatch_stub_770) - .type GL_PREFIX(_dispatch_stub_770), @function - HIDDEN(GL_PREFIX(_dispatch_stub_770)) -GL_PREFIX(_dispatch_stub_770): + .globl GL_PREFIX(_dispatch_stub_766) + .type GL_PREFIX(_dispatch_stub_766), @function + HIDDEN(GL_PREFIX(_dispatch_stub_766)) +GL_PREFIX(_dispatch_stub_766): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 6160(%rax), %r11 + movq 6128(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -29182,13 +29002,13 @@ GL_PREFIX(_dispatch_stub_770): popq %rdx popq %rsi popq %rdi - movq 6160(%rax), %r11 + movq 6128(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 6160(%rax), %r11 + movq 6128(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -29202,19 +29022,19 @@ GL_PREFIX(_dispatch_stub_770): popq %rdx popq %rsi popq %rdi - movq 6160(%rax), %r11 + movq 6128(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ - .size GL_PREFIX(_dispatch_stub_770), .-GL_PREFIX(_dispatch_stub_770) + .size GL_PREFIX(_dispatch_stub_766), .-GL_PREFIX(_dispatch_stub_766) .p2align 4,,15 - .globl GL_PREFIX(_dispatch_stub_771) - .type GL_PREFIX(_dispatch_stub_771), @function - HIDDEN(GL_PREFIX(_dispatch_stub_771)) -GL_PREFIX(_dispatch_stub_771): + .globl GL_PREFIX(_dispatch_stub_767) + .type GL_PREFIX(_dispatch_stub_767), @function + HIDDEN(GL_PREFIX(_dispatch_stub_767)) +GL_PREFIX(_dispatch_stub_767): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 6168(%rax), %r11 + movq 6136(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -29228,13 +29048,13 @@ GL_PREFIX(_dispatch_stub_771): popq %rdx popq %rsi popq %rdi - movq 6168(%rax), %r11 + movq 6136(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 6168(%rax), %r11 + movq 6136(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -29248,19 +29068,19 @@ GL_PREFIX(_dispatch_stub_771): popq %rdx popq %rsi popq %rdi - movq 6168(%rax), %r11 + movq 6136(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ - .size GL_PREFIX(_dispatch_stub_771), .-GL_PREFIX(_dispatch_stub_771) + .size GL_PREFIX(_dispatch_stub_767), .-GL_PREFIX(_dispatch_stub_767) .p2align 4,,15 - .globl GL_PREFIX(_dispatch_stub_772) - .type GL_PREFIX(_dispatch_stub_772), @function - HIDDEN(GL_PREFIX(_dispatch_stub_772)) -GL_PREFIX(_dispatch_stub_772): + .globl GL_PREFIX(_dispatch_stub_768) + .type GL_PREFIX(_dispatch_stub_768), @function + HIDDEN(GL_PREFIX(_dispatch_stub_768)) +GL_PREFIX(_dispatch_stub_768): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 6176(%rax), %r11 + movq 6144(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -29270,13 +29090,13 @@ GL_PREFIX(_dispatch_stub_772): popq %rdx popq %rsi popq %rdi - movq 6176(%rax), %r11 + movq 6144(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 6176(%rax), %r11 + movq 6144(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -29286,19 +29106,19 @@ GL_PREFIX(_dispatch_stub_772): popq %rdx popq %rsi popq %rdi - movq 6176(%rax), %r11 + movq 6144(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ - .size GL_PREFIX(_dispatch_stub_772), .-GL_PREFIX(_dispatch_stub_772) + .size GL_PREFIX(_dispatch_stub_768), .-GL_PREFIX(_dispatch_stub_768) .p2align 4,,15 - .globl GL_PREFIX(_dispatch_stub_773) - .type GL_PREFIX(_dispatch_stub_773), @function - HIDDEN(GL_PREFIX(_dispatch_stub_773)) -GL_PREFIX(_dispatch_stub_773): + .globl GL_PREFIX(_dispatch_stub_769) + .type GL_PREFIX(_dispatch_stub_769), @function + HIDDEN(GL_PREFIX(_dispatch_stub_769)) +GL_PREFIX(_dispatch_stub_769): #if defined(GLX_USE_TLS) call _x86_64_get_dispatch@PLT - movq 6184(%rax), %r11 + movq 6152(%rax), %r11 jmp *%r11 #elif defined(PTHREADS) pushq %rdi @@ -29308,13 +29128,13 @@ GL_PREFIX(_dispatch_stub_773): popq %rdx popq %rsi popq %rdi - movq 6184(%rax), %r11 + movq 6152(%rax), %r11 jmp *%r11 #else movq _glapi_Dispatch(%rip), %rax testq %rax, %rax je 1f - movq 6184(%rax), %r11 + movq 6152(%rax), %r11 jmp *%r11 1: pushq %rdi @@ -29324,10 +29144,10 @@ GL_PREFIX(_dispatch_stub_773): popq %rdx popq %rsi popq %rdi - movq 6184(%rax), %r11 + movq 6152(%rax), %r11 jmp *%r11 #endif /* defined(GLX_USE_TLS) */ - .size GL_PREFIX(_dispatch_stub_773), .-GL_PREFIX(_dispatch_stub_773) + .size GL_PREFIX(_dispatch_stub_769), .-GL_PREFIX(_dispatch_stub_769) .globl GL_PREFIX(ArrayElementEXT) ; .set GL_PREFIX(ArrayElementEXT), GL_PREFIX(ArrayElement) .globl GL_PREFIX(BindTextureEXT) ; .set GL_PREFIX(BindTextureEXT), GL_PREFIX(BindTexture) @@ -29419,6 +29239,10 @@ GL_PREFIX(_dispatch_stub_773): .globl GL_PREFIX(GetVertexAttribdv) ; .set GL_PREFIX(GetVertexAttribdv), GL_PREFIX(GetVertexAttribdvARB) .globl GL_PREFIX(GetVertexAttribfv) ; .set GL_PREFIX(GetVertexAttribfv), GL_PREFIX(GetVertexAttribfvARB) .globl GL_PREFIX(GetVertexAttribiv) ; .set GL_PREFIX(GetVertexAttribiv), GL_PREFIX(GetVertexAttribivARB) + .globl GL_PREFIX(ProgramParameter4dNV) ; .set GL_PREFIX(ProgramParameter4dNV), GL_PREFIX(ProgramEnvParameter4dARB) + .globl GL_PREFIX(ProgramParameter4dvNV) ; .set GL_PREFIX(ProgramParameter4dvNV), GL_PREFIX(ProgramEnvParameter4dvARB) + .globl GL_PREFIX(ProgramParameter4fNV) ; .set GL_PREFIX(ProgramParameter4fNV), GL_PREFIX(ProgramEnvParameter4fARB) + .globl GL_PREFIX(ProgramParameter4fvNV) ; .set GL_PREFIX(ProgramParameter4fvNV), GL_PREFIX(ProgramEnvParameter4fvARB) .globl GL_PREFIX(VertexAttrib1d) ; .set GL_PREFIX(VertexAttrib1d), GL_PREFIX(VertexAttrib1dARB) .globl GL_PREFIX(VertexAttrib1dv) ; .set GL_PREFIX(VertexAttrib1dv), GL_PREFIX(VertexAttrib1dvARB) .globl GL_PREFIX(VertexAttrib1f) ; .set GL_PREFIX(VertexAttrib1f), GL_PREFIX(VertexAttrib1fARB) @@ -29578,7 +29402,7 @@ GL_PREFIX(_dispatch_stub_773): .globl GL_PREFIX(IsProgramARB) ; .set GL_PREFIX(IsProgramARB), GL_PREFIX(IsProgramNV) .globl GL_PREFIX(PointParameteri) ; .set GL_PREFIX(PointParameteri), GL_PREFIX(PointParameteriNV) .globl GL_PREFIX(PointParameteriv) ; .set GL_PREFIX(PointParameteriv), GL_PREFIX(PointParameterivNV) - .globl GL_PREFIX(BlendEquationSeparate) ; .set GL_PREFIX(BlendEquationSeparate), GL_PREFIX(_dispatch_stub_749) + .globl GL_PREFIX(BlendEquationSeparate) ; .set GL_PREFIX(BlendEquationSeparate), GL_PREFIX(_dispatch_stub_745) #if defined(GLX_USE_TLS) && defined(__linux__) .section ".note.ABI-tag", "a" diff --git a/src/mesa/x86/glapi_x86.S b/src/mesa/x86/glapi_x86.S index 40ecc20753..b99c2b6520 100644 --- a/src/mesa/x86/glapi_x86.S +++ b/src/mesa/x86/glapi_x86.S @@ -844,10 +844,6 @@ GLNAME(gl_dispatch_functions_start): GL_STUB(GetVertexAttribivNV, _gloffset_GetVertexAttribivNV, GetVertexAttribivNV@12) GL_STUB(IsProgramNV, _gloffset_IsProgramNV, IsProgramNV@4) GL_STUB(LoadProgramNV, _gloffset_LoadProgramNV, LoadProgramNV@16) - GL_STUB(ProgramParameter4dNV, _gloffset_ProgramParameter4dNV, ProgramParameter4dNV@40) - GL_STUB(ProgramParameter4dvNV, _gloffset_ProgramParameter4dvNV, ProgramParameter4dvNV@12) - GL_STUB(ProgramParameter4fNV, _gloffset_ProgramParameter4fNV, ProgramParameter4fNV@24) - GL_STUB(ProgramParameter4fvNV, _gloffset_ProgramParameter4fvNV, ProgramParameter4fvNV@12) GL_STUB(ProgramParameters4dvNV, _gloffset_ProgramParameters4dvNV, ProgramParameters4dvNV@16) GL_STUB(ProgramParameters4fvNV, _gloffset_ProgramParameters4fvNV, ProgramParameters4fvNV@16) GL_STUB(RequestResidentProgramsNV, _gloffset_RequestResidentProgramsNV, RequestResidentProgramsNV@8) @@ -908,26 +904,26 @@ GLNAME(gl_dispatch_functions_start): GL_STUB(SetFragmentShaderConstantATI, _gloffset_SetFragmentShaderConstantATI, SetFragmentShaderConstantATI@8) GL_STUB(PointParameteriNV, _gloffset_PointParameteriNV, PointParameteriNV@8) GL_STUB(PointParameterivNV, _gloffset_PointParameterivNV, PointParameterivNV@8) - GL_STUB(_dispatch_stub_737, _gloffset_ActiveStencilFaceEXT, _dispatch_stub_737@4) + GL_STUB(_dispatch_stub_733, _gloffset_ActiveStencilFaceEXT, _dispatch_stub_733@4) + HIDDEN(GL_PREFIX(_dispatch_stub_733, _dispatch_stub_733@4)) + GL_STUB(_dispatch_stub_734, _gloffset_BindVertexArrayAPPLE, _dispatch_stub_734@4) + HIDDEN(GL_PREFIX(_dispatch_stub_734, _dispatch_stub_734@4)) + GL_STUB(_dispatch_stub_735, _gloffset_DeleteVertexArraysAPPLE, _dispatch_stub_735@8) + HIDDEN(GL_PREFIX(_dispatch_stub_735, _dispatch_stub_735@8)) + GL_STUB(_dispatch_stub_736, _gloffset_GenVertexArraysAPPLE, _dispatch_stub_736@8) + HIDDEN(GL_PREFIX(_dispatch_stub_736, _dispatch_stub_736@8)) + GL_STUB(_dispatch_stub_737, _gloffset_IsVertexArrayAPPLE, _dispatch_stub_737@4) HIDDEN(GL_PREFIX(_dispatch_stub_737, _dispatch_stub_737@4)) - GL_STUB(_dispatch_stub_738, _gloffset_BindVertexArrayAPPLE, _dispatch_stub_738@4) - HIDDEN(GL_PREFIX(_dispatch_stub_738, _dispatch_stub_738@4)) - GL_STUB(_dispatch_stub_739, _gloffset_DeleteVertexArraysAPPLE, _dispatch_stub_739@8) - HIDDEN(GL_PREFIX(_dispatch_stub_739, _dispatch_stub_739@8)) - GL_STUB(_dispatch_stub_740, _gloffset_GenVertexArraysAPPLE, _dispatch_stub_740@8) - HIDDEN(GL_PREFIX(_dispatch_stub_740, _dispatch_stub_740@8)) - GL_STUB(_dispatch_stub_741, _gloffset_IsVertexArrayAPPLE, _dispatch_stub_741@4) - HIDDEN(GL_PREFIX(_dispatch_stub_741, _dispatch_stub_741@4)) GL_STUB(GetProgramNamedParameterdvNV, _gloffset_GetProgramNamedParameterdvNV, GetProgramNamedParameterdvNV@16) GL_STUB(GetProgramNamedParameterfvNV, _gloffset_GetProgramNamedParameterfvNV, GetProgramNamedParameterfvNV@16) GL_STUB(ProgramNamedParameter4dNV, _gloffset_ProgramNamedParameter4dNV, ProgramNamedParameter4dNV@44) GL_STUB(ProgramNamedParameter4dvNV, _gloffset_ProgramNamedParameter4dvNV, ProgramNamedParameter4dvNV@16) GL_STUB(ProgramNamedParameter4fNV, _gloffset_ProgramNamedParameter4fNV, ProgramNamedParameter4fNV@28) GL_STUB(ProgramNamedParameter4fvNV, _gloffset_ProgramNamedParameter4fvNV, ProgramNamedParameter4fvNV@16) - GL_STUB(_dispatch_stub_748, _gloffset_DepthBoundsEXT, _dispatch_stub_748@16) - HIDDEN(GL_PREFIX(_dispatch_stub_748, _dispatch_stub_748@16)) - GL_STUB(_dispatch_stub_749, _gloffset_BlendEquationSeparateEXT, _dispatch_stub_749@8) - HIDDEN(GL_PREFIX(_dispatch_stub_749, _dispatch_stub_749@8)) + GL_STUB(_dispatch_stub_744, _gloffset_DepthBoundsEXT, _dispatch_stub_744@16) + HIDDEN(GL_PREFIX(_dispatch_stub_744, _dispatch_stub_744@16)) + GL_STUB(_dispatch_stub_745, _gloffset_BlendEquationSeparateEXT, _dispatch_stub_745@8) + HIDDEN(GL_PREFIX(_dispatch_stub_745, _dispatch_stub_745@8)) GL_STUB(BindFramebufferEXT, _gloffset_BindFramebufferEXT, BindFramebufferEXT@8) GL_STUB(BindRenderbufferEXT, _gloffset_BindRenderbufferEXT, BindRenderbufferEXT@8) GL_STUB(CheckFramebufferStatusEXT, _gloffset_CheckFramebufferStatusEXT, CheckFramebufferStatusEXT@4) @@ -945,19 +941,19 @@ GLNAME(gl_dispatch_functions_start): GL_STUB(IsFramebufferEXT, _gloffset_IsFramebufferEXT, IsFramebufferEXT@4) GL_STUB(IsRenderbufferEXT, _gloffset_IsRenderbufferEXT, IsRenderbufferEXT@4) GL_STUB(RenderbufferStorageEXT, _gloffset_RenderbufferStorageEXT, RenderbufferStorageEXT@16) - GL_STUB(_dispatch_stub_767, _gloffset_BlitFramebufferEXT, _dispatch_stub_767@40) - HIDDEN(GL_PREFIX(_dispatch_stub_767, _dispatch_stub_767@40)) + GL_STUB(_dispatch_stub_763, _gloffset_BlitFramebufferEXT, _dispatch_stub_763@40) + HIDDEN(GL_PREFIX(_dispatch_stub_763, _dispatch_stub_763@40)) GL_STUB(FramebufferTextureLayerEXT, _gloffset_FramebufferTextureLayerEXT, FramebufferTextureLayerEXT@20) - GL_STUB(_dispatch_stub_769, _gloffset_StencilFuncSeparateATI, _dispatch_stub_769@16) - HIDDEN(GL_PREFIX(_dispatch_stub_769, _dispatch_stub_769@16)) - GL_STUB(_dispatch_stub_770, _gloffset_ProgramEnvParameters4fvEXT, _dispatch_stub_770@16) - HIDDEN(GL_PREFIX(_dispatch_stub_770, _dispatch_stub_770@16)) - GL_STUB(_dispatch_stub_771, _gloffset_ProgramLocalParameters4fvEXT, _dispatch_stub_771@16) - HIDDEN(GL_PREFIX(_dispatch_stub_771, _dispatch_stub_771@16)) - GL_STUB(_dispatch_stub_772, _gloffset_GetQueryObjecti64vEXT, _dispatch_stub_772@12) - HIDDEN(GL_PREFIX(_dispatch_stub_772, _dispatch_stub_772@12)) - GL_STUB(_dispatch_stub_773, _gloffset_GetQueryObjectui64vEXT, _dispatch_stub_773@12) - HIDDEN(GL_PREFIX(_dispatch_stub_773, _dispatch_stub_773@12)) + GL_STUB(_dispatch_stub_765, _gloffset_StencilFuncSeparateATI, _dispatch_stub_765@16) + HIDDEN(GL_PREFIX(_dispatch_stub_765, _dispatch_stub_765@16)) + GL_STUB(_dispatch_stub_766, _gloffset_ProgramEnvParameters4fvEXT, _dispatch_stub_766@16) + HIDDEN(GL_PREFIX(_dispatch_stub_766, _dispatch_stub_766@16)) + GL_STUB(_dispatch_stub_767, _gloffset_ProgramLocalParameters4fvEXT, _dispatch_stub_767@16) + HIDDEN(GL_PREFIX(_dispatch_stub_767, _dispatch_stub_767@16)) + GL_STUB(_dispatch_stub_768, _gloffset_GetQueryObjecti64vEXT, _dispatch_stub_768@12) + HIDDEN(GL_PREFIX(_dispatch_stub_768, _dispatch_stub_768@12)) + GL_STUB(_dispatch_stub_769, _gloffset_GetQueryObjectui64vEXT, _dispatch_stub_769@12) + HIDDEN(GL_PREFIX(_dispatch_stub_769, _dispatch_stub_769@12)) GL_STUB_ALIAS(ArrayElementEXT, _gloffset_ArrayElement, ArrayElementEXT@4, ArrayElement, ArrayElement@4) GL_STUB_ALIAS(BindTextureEXT, _gloffset_BindTexture, BindTextureEXT@8, BindTexture, BindTexture@8) GL_STUB_ALIAS(DrawArraysEXT, _gloffset_DrawArrays, DrawArraysEXT@12, DrawArrays, DrawArrays@12) @@ -1107,6 +1103,10 @@ GLNAME(gl_dispatch_functions_start): GL_STUB_ALIAS(GetVertexAttribdv, _gloffset_GetVertexAttribdvARB, GetVertexAttribdv@12, GetVertexAttribdvARB, GetVertexAttribdvARB@12) GL_STUB_ALIAS(GetVertexAttribfv, _gloffset_GetVertexAttribfvARB, GetVertexAttribfv@12, GetVertexAttribfvARB, GetVertexAttribfvARB@12) GL_STUB_ALIAS(GetVertexAttribiv, _gloffset_GetVertexAttribivARB, GetVertexAttribiv@12, GetVertexAttribivARB, GetVertexAttribivARB@12) + GL_STUB_ALIAS(ProgramParameter4dNV, _gloffset_ProgramEnvParameter4dARB, ProgramParameter4dNV@40, ProgramEnvParameter4dARB, ProgramEnvParameter4dARB@40) + GL_STUB_ALIAS(ProgramParameter4dvNV, _gloffset_ProgramEnvParameter4dvARB, ProgramParameter4dvNV@12, ProgramEnvParameter4dvARB, ProgramEnvParameter4dvARB@12) + GL_STUB_ALIAS(ProgramParameter4fNV, _gloffset_ProgramEnvParameter4fARB, ProgramParameter4fNV@24, ProgramEnvParameter4fARB, ProgramEnvParameter4fARB@24) + GL_STUB_ALIAS(ProgramParameter4fvNV, _gloffset_ProgramEnvParameter4fvARB, ProgramParameter4fvNV@12, ProgramEnvParameter4fvARB, ProgramEnvParameter4fvARB@12) GL_STUB_ALIAS(VertexAttrib1d, _gloffset_VertexAttrib1dARB, VertexAttrib1d@12, VertexAttrib1dARB, VertexAttrib1dARB@12) GL_STUB_ALIAS(VertexAttrib1dv, _gloffset_VertexAttrib1dvARB, VertexAttrib1dv@8, VertexAttrib1dvARB, VertexAttrib1dvARB@8) GL_STUB_ALIAS(VertexAttrib1f, _gloffset_VertexAttrib1fARB, VertexAttrib1f@8, VertexAttrib1fARB, VertexAttrib1fARB@8) diff --git a/src/mesa/x86/rtasm/x86sse.c b/src/mesa/x86/rtasm/x86sse.c index 39c0e9b851..7a91364ed8 100644 --- a/src/mesa/x86/rtasm/x86sse.c +++ b/src/mesa/x86/rtasm/x86sse.c @@ -1,3 +1,4 @@ +#ifdef USE_X86_ASM #if defined(__i386__) || defined(__386__) #include "imports.h" @@ -290,7 +291,7 @@ void x86_call( struct x86_function *p, void (*label)()) void x86_call( struct x86_function *p, struct x86_reg reg) { emit_1ub(p, 0xff); - emit_modrm(p, reg, reg); + emit_modrm_noreg(p, 2, reg); } #endif @@ -1191,3 +1192,9 @@ void x86sse_dummy( void ) } #endif + +#else /* USE_X86_ASM */ + +int x86sse_c_dummy_var; /* silence warning */ + +#endif /* USE_X86_ASM */ @@ -89,105 +89,6 @@ def get_winddk_paths(env, version=None): exe_path = string.join(exe_paths, os.pathsep ) return (include_path, lib_path, exe_path) -def set_winddk_flags(env): - """Mimic WINDDK's builtin flags. - - See also: - - WINDDK's bin/makefile.new i386mk.inc for more info. - - buildchk_wxp_x86.log files, generated by the WINDDK's build - - http://alter.org.ua/docs/nt_kernel/vc8_proj/ - """ - - cppdefines = [ - ('_X86_', '1'), - ('i386', '1'), - 'STD_CALL', - ('CONDITION_HANDLING', '1'), - ('NT_INST', '0'), - ('WIN32', '100'), - ('_NT1X_', '100'), - ('WINNT', '1'), - ('_WIN32_WINNT', '0x0501'), # minimum required OS version - ('WINVER', '0x0501'), - ('_WIN32_IE', '0x0603'), - ('WIN32_LEAN_AND_MEAN', '1'), - ('DEVL', '1'), - ('__BUILDMACHINE__', 'WinDDK'), - ('FPO', '0'), - ] - if env.get('DEBUG', False): - cppdefines += [ - ('DBG', 1), - ] - env.Append(CPPDEFINES = cppdefines) - - # See also: - # - http://msdn2.microsoft.com/en-us/library/y0zzbyt4.aspx - # - cl /? - cflags = [ - '/Zl', # omit default library name in .OBJ - '/Zp8', # 8bytes struct member alignment - '/Gy', # separate functions for linker - '/Gm-', # disable minimal rebuild - '/W3', # warning level - '/WX', # treat warnings as errors - '/Gz', # __stdcall Calling convention - '/GX-', # disable C++ EH - '/GR-', # disable C++ RTTI - '/GF', # enable read-only string pooling - '/GS', # enable security checks - '/G6', # optimize for PPro, P-II, P-III - '/Ze', # enable extensions - #'/Gi-', # ??? - '/QIfdiv-', # disable Pentium FDIV fix - #'/hotpatch', # ??? - #'/Z7', #enable old-style debug info - ] - if env.get('debug', False): - cflags += [ - '/Od', # disable optimizations - '/Oi', # enable intrinsic functions - '/Oy-', # disable frame pointer omission - ] - else: - cflags += [ - '/Ox', # maximum optimizations - '/Oi', # enable intrinsic functions - '/Os', # favor code space - ] - env.Append(CFLAGS = cflags) - env.Append(CXXFLAGS = cflags) - - # See also: - # - http://msdn2.microsoft.com/en-us/library/y0zzbyt4.aspx - env.Append(LINKFLAGS = [ - '/merge:_PAGE=PAGE', - '/merge:_TEXT=.text', - '/section:INIT,d', - '/opt:ref', - '/opt:icf', - '/ignore:4198,4010,4037,4039,4065,4070,4078,4087,4089,4221', - '/incremental:no', - '/fullbuild', - '/release', - '/nodefaultlib', - '/wx', - '/debug', - '/debugtype:cv', - '/version:5.1', - '/osversion:5.1', - '/functionpadmin:5', - '/safeseh', - '/pdbcompress', - '/stack:0x40000,0x1000', - '/driver', - '/align:0x80', - '/subsystem:native,5.01', - '/base:0x10000', - - '/entry:DrvEnableDriver', - ]) - def validate_vars(env): """Validate the PCH and PCHSTOP construction variables.""" if env.has_key('PCH') and env['PCH']: @@ -314,8 +215,6 @@ def generate(env): SCons.Tool.mslink.generate(env) - set_winddk_flags(env) - if not env.has_key('ENV'): env['ENV'] = {} |