diff options
Diffstat (limited to 'src/mesa')
40 files changed, 1027 insertions, 430 deletions
diff --git a/src/mesa/drivers/dri/common/spantmp2.h b/src/mesa/drivers/dri/common/spantmp2.h index f2868cb58a..89c815722f 100644 --- a/src/mesa/drivers/dri/common/spantmp2.h +++ b/src/mesa/drivers/dri/common/spantmp2.h @@ -82,6 +82,71 @@ rgba[3] = 0xff; \ } while (0) +#elif (SPANTMP_PIXEL_FMT == GL_BGRA) && (SPANTMP_PIXEL_TYPE == GL_UNSIGNED_SHORT_4_4_4_4_REV) + +/** + ** GL_BGRA, GL_UNSIGNED_SHORT_4_4_4_4_REV + **/ + +#ifndef GET_VALUE +#ifndef GET_PTR +#define GET_PTR(_x, _y) (buf + (_x) * 2 + (_y) * pitch) +#endif + +#define GET_VALUE(_x, _y) *(volatile GLushort *)(GET_PTR(_x, _y)) +#define PUT_VALUE(_x, _y, _v) *(volatile GLushort *)(GET_PTR(_x, _y)) = (_v) +#endif /* GET_VALUE */ + +#define INIT_MONO_PIXEL(p, color) \ + p = PACK_COLOR_4444(color[3], color[0], color[1], color[2]) + +#define WRITE_RGBA( _x, _y, r, g, b, a ) \ + PUT_VALUE(_x, _y, PACK_COLOR_4444(a, r, g, b)) \ + +#define WRITE_PIXEL( _x, _y, p ) PUT_VALUE(_x, _y, p) + +#define READ_RGBA( rgba, _x, _y ) \ + do { \ + GLushort p = GET_VALUE(_x, _y); \ + rgba[0] = ((p >> 8) & 0xf) * 0x11; \ + rgba[1] = ((p >> 4) & 0xf) * 0x11; \ + rgba[2] = ((p >> 0) & 0xf) * 0x11; \ + rgba[3] = ((p >> 12) & 0xf) * 0x11; \ + } while (0) + + +#elif (SPANTMP_PIXEL_FMT == GL_BGRA) && (SPANTMP_PIXEL_TYPE == GL_UNSIGNED_SHORT_1_5_5_5_REV) + +/** + ** GL_BGRA, GL_UNSIGNED_SHORT_1_5_5_5_REV + **/ + +#ifndef GET_VALUE +#ifndef GET_PTR +#define GET_PTR(_x, _y) (buf + (_x) * 2 + (_y) * pitch) +#endif + +#define GET_VALUE(_x, _y) *(volatile GLushort *)(GET_PTR(_x, _y)) +#define PUT_VALUE(_x, _y, _v) *(volatile GLushort *)(GET_PTR(_x, _y)) = (_v) +#endif /* GET_VALUE */ + +#define INIT_MONO_PIXEL(p, color) \ + p = PACK_COLOR_1555(color[3], color[0], color[1], color[2]) + +#define WRITE_RGBA( _x, _y, r, g, b, a ) \ + PUT_VALUE(_x, _y, PACK_COLOR_1555(a, r, g, b)) \ + +#define WRITE_PIXEL( _x, _y, p ) PUT_VALUE(_x, _y, p) + +#define READ_RGBA( rgba, _x, _y ) \ + do { \ + GLushort p = GET_VALUE(_x, _y); \ + rgba[0] = ((p >> 7) & 0xf8) * 255 / 0xf8; \ + rgba[1] = ((p >> 2) & 0xf8) * 255 / 0xf8; \ + rgba[2] = ((p << 3) & 0xf8) * 255 / 0xf8; \ + rgba[3] = ((p >> 15) & 0x1) * 0xff; \ + } while (0) + #elif (SPANTMP_PIXEL_FMT == GL_BGRA) && (SPANTMP_PIXEL_TYPE == GL_UNSIGNED_INT_8_8_8_8_REV) /** diff --git a/src/mesa/drivers/dri/i915/Makefile b/src/mesa/drivers/dri/i915/Makefile index 954a7e2af1..9f4bd1699f 100644 --- a/src/mesa/drivers/dri/i915/Makefile +++ b/src/mesa/drivers/dri/i915/Makefile @@ -11,7 +11,6 @@ DRIVER_SOURCES = \ i830_metaops.c \ i830_state.c \ i830_texblend.c \ - i830_tex.c \ i830_texstate.c \ i830_vtbl.c \ intel_render.c \ @@ -36,7 +35,6 @@ DRIVER_SOURCES = \ intel_buffers.c \ intel_blit.c \ intel_swapbuffers.c \ - i915_tex.c \ i915_tex_layout.c \ i915_texstate.c \ i915_context.c \ diff --git a/src/mesa/drivers/dri/i915/i830_context.c b/src/mesa/drivers/dri/i915/i830_context.c index 9c540cb2bb..10b9bf371c 100644 --- a/src/mesa/drivers/dri/i915/i830_context.c +++ b/src/mesa/drivers/dri/i915/i830_context.c @@ -47,7 +47,6 @@ i830InitDriverFunctions(struct dd_function_table *functions) { intelInitDriverFunctions(functions); i830InitStateFuncs(functions); - i830InitTextureFuncs(functions); } extern const struct tnl_pipeline_stage *intel_pipeline[]; diff --git a/src/mesa/drivers/dri/i915/i830_tex.c b/src/mesa/drivers/dri/i915/i830_tex.c deleted file mode 100644 index 34ac42a78e..0000000000 --- a/src/mesa/drivers/dri/i915/i830_tex.c +++ /dev/null @@ -1,100 +0,0 @@ -/************************************************************************** - * - * Copyright 2003 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 "main/glheader.h" -#include "main/mtypes.h" -#include "main/imports.h" -#include "main/simple_list.h" -#include "main/enums.h" -#include "main/image.h" -#include "main/mm.h" -#include "main/texstore.h" -#include "main/texformat.h" -#include "swrast/swrast.h" - -#include "texmem.h" - -#include "i830_context.h" -#include "i830_reg.h" - - - -static void -i830TexEnv(GLcontext * ctx, GLenum target, - GLenum pname, const GLfloat * param) -{ - - switch (pname) { - case GL_TEXTURE_ENV_COLOR: - case GL_TEXTURE_ENV_MODE: - case GL_COMBINE_RGB: - case GL_COMBINE_ALPHA: - case GL_SOURCE0_RGB: - case GL_SOURCE1_RGB: - case GL_SOURCE2_RGB: - case GL_SOURCE0_ALPHA: - case GL_SOURCE1_ALPHA: - case GL_SOURCE2_ALPHA: - case GL_OPERAND0_RGB: - case GL_OPERAND1_RGB: - case GL_OPERAND2_RGB: - case GL_OPERAND0_ALPHA: - case GL_OPERAND1_ALPHA: - case GL_OPERAND2_ALPHA: - case GL_RGB_SCALE: - case GL_ALPHA_SCALE: - break; - - case GL_TEXTURE_LOD_BIAS:{ - struct i830_context *i830 = i830_context(ctx); - GLuint unit = ctx->Texture.CurrentUnit; - int b = (int) ((*param) * 16.0); - if (b > 63) - b = 63; - if (b < -64) - b = -64; - I830_STATECHANGE(i830, I830_UPLOAD_TEX(unit)); - i830->lodbias_tm0s3[unit] = - ((b << TM0S3_LOD_BIAS_SHIFT) & TM0S3_LOD_BIAS_MASK); - break; - } - - default: - break; - } -} - - - - -void -i830InitTextureFuncs(struct dd_function_table *functions) -{ -/* - functions->TexEnv = i830TexEnv; -*/ -} diff --git a/src/mesa/drivers/dri/i915/i830_vtbl.c b/src/mesa/drivers/dri/i915/i830_vtbl.c index 1a94921078..3bf02de61f 100644 --- a/src/mesa/drivers/dri/i915/i830_vtbl.c +++ b/src/mesa/drivers/dri/i915/i830_vtbl.c @@ -26,12 +26,14 @@ **************************************************************************/ #include "glapi/glapi.h" +#include "main/texformat.h" #include "i830_context.h" #include "i830_reg.h" #include "intel_batchbuffer.h" #include "intel_regions.h" #include "intel_tris.h" +#include "intel_fbo.h" #include "tnl/t_context.h" #include "tnl/t_vertex.h" @@ -614,6 +616,8 @@ i830_state_draw_region(struct intel_context *intel, { struct i830_context *i830 = i830_context(&intel->ctx); GLcontext *ctx = &intel->ctx; + struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[0]; + struct intel_renderbuffer *irb = intel_renderbuffer(rb); GLuint value; ASSERT(state == &i830->state || state == &i830->meta); @@ -651,13 +655,27 @@ i830_state_draw_region(struct intel_context *intel, */ value = (DSTORG_HORT_BIAS(0x8) | /* .5 */ DSTORG_VERT_BIAS(0x8) | DEPTH_IS_Z); /* .5 */ - - if (color_region && color_region->cpp == 4) { - value |= DV_PF_8888; - } - else { - value |= DV_PF_565; + + if (irb != NULL) { + switch (irb->texformat->MesaFormat) { + case MESA_FORMAT_ARGB8888: + value |= DV_PF_8888; + break; + case MESA_FORMAT_RGB565: + value |= DV_PF_565; + break; + case MESA_FORMAT_ARGB1555: + value |= DV_PF_1555; + break; + case MESA_FORMAT_ARGB4444: + value |= DV_PF_4444; + break; + default: + _mesa_problem(ctx, "Bad renderbuffer format: %d\n", + irb->texformat->MesaFormat); + } } + if (depth_region && depth_region->cpp == 4) { value |= DEPTH_FRMT_24_FIXED_8_OTHER; } diff --git a/src/mesa/drivers/dri/i915/i915_context.c b/src/mesa/drivers/dri/i915/i915_context.c index 7549029a1b..fdd2cf6109 100644 --- a/src/mesa/drivers/dri/i915/i915_context.c +++ b/src/mesa/drivers/dri/i915/i915_context.c @@ -83,7 +83,6 @@ i915InitDriverFunctions(struct dd_function_table *functions) { intelInitDriverFunctions(functions); i915InitStateFunctions(functions); - i915InitTextureFuncs(functions); i915InitFragProgFuncs(functions); functions->UpdateState = i915InvalidateState; } diff --git a/src/mesa/drivers/dri/i915/i915_fragprog.c b/src/mesa/drivers/dri/i915/i915_fragprog.c index a5158de945..2db10c60e9 100644 --- a/src/mesa/drivers/dri/i915/i915_fragprog.c +++ b/src/mesa/drivers/dri/i915/i915_fragprog.c @@ -323,7 +323,8 @@ upload_program(struct i915_fragment_program *p) p->ctx->FragmentProgram._Current; const struct prog_instruction *inst = program->Base.Instructions; -/* _mesa_debug_fp_inst(program->Base.NumInstructions, inst); */ + if (INTEL_DEBUG & DEBUG_WM) + _mesa_print_program(&program->Base); /* Is this a parse-failed program? Ensure a valid program is * loaded, as the flagging of an error isn't sufficient to stop @@ -1049,9 +1050,6 @@ i915ProgramStringNotify(GLcontext * ctx, _mesa_append_fog_code(ctx, &p->FragProg); p->FragProg.FogOption = GL_NONE; } - - if (INTEL_DEBUG & DEBUG_STATE) - _mesa_print_program(prog); } _tnl_program_string(ctx, target, prog); diff --git a/src/mesa/drivers/dri/i915/i915_tex.c b/src/mesa/drivers/dri/i915/i915_tex.c deleted file mode 100644 index e38d8fe79d..0000000000 --- a/src/mesa/drivers/dri/i915/i915_tex.c +++ /dev/null @@ -1,78 +0,0 @@ -/************************************************************************** - * - * Copyright 2003 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 "main/glheader.h" -#include "main/mtypes.h" -#include "main/imports.h" -#include "main/simple_list.h" -#include "main/enums.h" -#include "main/image.h" -#include "main/mm.h" -#include "main/texstore.h" -#include "main/texformat.h" -#include "swrast/swrast.h" - -#include "texmem.h" - -#include "i915_context.h" -#include "i915_reg.h" - - - -static void -i915TexEnv(GLcontext * ctx, GLenum target, - GLenum pname, const GLfloat * param) -{ - struct i915_context *i915 = I915_CONTEXT(ctx); - - switch (pname) { - case GL_TEXTURE_LOD_BIAS:{ - GLuint unit = ctx->Texture.CurrentUnit; - GLint b = (int) ((*param) * 16.0); - if (b > 255) - b = 255; - if (b < -256) - b = -256; - I915_STATECHANGE(i915, I915_UPLOAD_TEX(unit)); - i915->lodbias_ss2[unit] = - ((b << SS2_LOD_BIAS_SHIFT) & SS2_LOD_BIAS_MASK); - break; - } - - default: - break; - } -} - - -void -i915InitTextureFuncs(struct dd_function_table *functions) -{ -/* - functions->TexEnv = i915TexEnv; -*/ -} diff --git a/src/mesa/drivers/dri/i915/i915_vtbl.c b/src/mesa/drivers/dri/i915/i915_vtbl.c index 3f6d282d34..115004616f 100644 --- a/src/mesa/drivers/dri/i915/i915_vtbl.c +++ b/src/mesa/drivers/dri/i915/i915_vtbl.c @@ -32,6 +32,7 @@ #include "main/imports.h" #include "main/macros.h" #include "main/colormac.h" +#include "main/texformat.h" #include "tnl/t_context.h" #include "tnl/t_vertex.h" @@ -40,6 +41,7 @@ #include "intel_tex.h" #include "intel_regions.h" #include "intel_tris.h" +#include "intel_fbo.h" #include "i915_reg.h" #include "i915_context.h" @@ -542,6 +544,8 @@ i915_state_draw_region(struct intel_context *intel, { struct i915_context *i915 = i915_context(&intel->ctx); GLcontext *ctx = &intel->ctx; + struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[0]; + struct intel_renderbuffer *irb = intel_renderbuffer(rb); GLuint value; ASSERT(state == &i915->state || state == &i915->meta); @@ -580,12 +584,26 @@ i915_state_draw_region(struct intel_context *intel, value = (DSTORG_HORT_BIAS(0x8) | /* .5 */ DSTORG_VERT_BIAS(0x8) | /* .5 */ LOD_PRECLAMP_OGL | TEX_DEFAULT_COLOR_OGL); - if (color_region && color_region->cpp == 4) { - value |= DV_PF_8888; - } - else { - value |= (DITHER_FULL_ALWAYS | DV_PF_565); + if (irb != NULL) { + switch (irb->texformat->MesaFormat) { + case MESA_FORMAT_ARGB8888: + value |= DV_PF_8888; + break; + case MESA_FORMAT_RGB565: + value |= DV_PF_565 | DITHER_FULL_ALWAYS; + break; + case MESA_FORMAT_ARGB1555: + value |= DV_PF_1555 | DITHER_FULL_ALWAYS; + break; + case MESA_FORMAT_ARGB4444: + value |= DV_PF_4444 | DITHER_FULL_ALWAYS; + break; + default: + _mesa_problem(ctx, "Bad renderbuffer format: %d\n", + irb->texformat->MesaFormat); + } } + if (depth_region && depth_region->cpp == 4) { value |= DEPTH_FRMT_24_FIXED_8_OTHER; } diff --git a/src/mesa/drivers/dri/i965/brw_curbe.c b/src/mesa/drivers/dri/i965/brw_curbe.c index dfab14aa74..fe1de821f0 100644 --- a/src/mesa/drivers/dri/i965/brw_curbe.c +++ b/src/mesa/drivers/dri/i965/brw_curbe.c @@ -353,6 +353,14 @@ update_constant_buffer(struct brw_context *brw, map = const_buffer->virtual; memcpy(map, params->ParameterValues, size); dri_bo_unmap(const_buffer); + + if (0) { + int i; + for (i = 0; i < params->NumParameters; i++) { + float *p = params->ParameterValues[i]; + printf("%d: %f %f %f %f\n", i, p[0], p[1], p[2], p[3]); + } + } } } @@ -363,6 +371,10 @@ update_vertex_constant_buffer(struct brw_context *brw) { struct brw_vertex_program *vp = (struct brw_vertex_program *) brw->vertex_program; + if (0) { + printf("update VS constants in buffer %p\n", vp->const_buffer); + printf("program %u\n", vp->program.Base.Id); + } update_constant_buffer(brw, vp->program.Base.Parameters, vp->const_buffer); } diff --git a/src/mesa/drivers/dri/i965/brw_eu.h b/src/mesa/drivers/dri/i965/brw_eu.h index 66f8eb840c..896e67dbfe 100644 --- a/src/mesa/drivers/dri/i965/brw_eu.h +++ b/src/mesa/drivers/dri/i965/brw_eu.h @@ -868,6 +868,7 @@ void brw_dp_READ_4( struct brw_compile *p, void brw_dp_READ_4_vs( struct brw_compile *p, struct brw_reg dest, GLboolean relAddr, + struct brw_reg addrReg, GLuint location, GLuint bind_table_index ); diff --git a/src/mesa/drivers/dri/i965/brw_eu_emit.c b/src/mesa/drivers/dri/i965/brw_eu_emit.c index c731a93a8d..df2141660c 100644 --- a/src/mesa/drivers/dri/i965/brw_eu_emit.c +++ b/src/mesa/drivers/dri/i965/brw_eu_emit.c @@ -1003,15 +1003,18 @@ void brw_dp_READ_4( struct brw_compile *p, /** - * Read float[4] constant from VS constant buffer. + * Read float[4] constant(s) from VS constant buffer. + * For relative addressing, two float[4] constants will be read into 'dest'. + * Otherwise, one float[4] constant will be read into the lower half of 'dest'. */ void brw_dp_READ_4_vs(struct brw_compile *p, struct brw_reg dest, GLboolean relAddr, + struct brw_reg addrReg, GLuint location, GLuint bind_table_index) { - const GLuint msg_reg_nr = 1; + GLuint msg_reg_nr = 1; /* printf("vs const read msg, location %u, msg_reg_nr %d\n", @@ -1034,7 +1037,12 @@ void brw_dp_READ_4_vs(struct brw_compile *p, b = brw_message_reg(msg_reg_nr); b = retype(b, BRW_REGISTER_TYPE_UD); /*b = get_element_ud(b, 2);*/ - brw_MOV(p, b, brw_imm_ud(location)); + if (relAddr) { + brw_ADD(p, b, addrReg, brw_imm_ud(location)); + } + else { + brw_MOV(p, b, brw_imm_ud(location)); + } brw_pop_insn_state(p); } @@ -1053,13 +1061,56 @@ void brw_dp_READ_4_vs(struct brw_compile *p, brw_set_dp_read_message(insn, bind_table_index, - 0, /* msg_control (0 means 1 Oword) */ + 0, /* msg_control (0 means 1 Oword, lower half) */ BRW_DATAPORT_READ_MESSAGE_OWORD_BLOCK_READ, /* msg_type */ 0, /* source cache = data cache */ 1, /* msg_length */ 1, /* response_length (1 Oword) */ 0); /* eot */ } + + if (relAddr) { + /* second read to get second constant */ + msg_reg_nr++; + { + /* Setup MRF[1] with location/offset into const buffer */ + struct brw_reg b; + + brw_push_insn_state(p); + brw_set_compression_control(p, BRW_COMPRESSION_NONE); + brw_set_mask_control(p, BRW_MASK_DISABLE); + brw_set_predicate_control(p, BRW_PREDICATE_NONE); + + b = brw_message_reg(msg_reg_nr); + b = retype(b, BRW_REGISTER_TYPE_UD); + addrReg = suboffset(addrReg, 1); /* upper half of addrReg */ + brw_ADD(p, b, addrReg, brw_imm_ud(location)); + + brw_pop_insn_state(p); + } + + { + struct brw_instruction *insn = next_insn(p, BRW_OPCODE_SEND); + + insn->header.predicate_control = BRW_PREDICATE_NONE; + insn->header.compression_control = BRW_COMPRESSION_NONE; + insn->header.destreg__conditonalmod = msg_reg_nr; + insn->header.mask_control = BRW_MASK_DISABLE; + /*insn->header.access_mode = BRW_ALIGN_16;*/ + + brw_set_dest(insn, dest); + brw_set_src0(insn, brw_null_reg()); + + brw_set_dp_read_message(insn, + bind_table_index, + 1, /* msg_control (1 means 1 Oword, upper half) */ + BRW_DATAPORT_READ_MESSAGE_OWORD_BLOCK_READ, /* msg_type */ + 0, /* source cache = data cache */ + 1, /* msg_length */ + 1, /* response_length (1 Oword) */ + 0); /* eot */ + } + } } diff --git a/src/mesa/drivers/dri/i965/brw_state_dump.c b/src/mesa/drivers/dri/i965/brw_state_dump.c index 5d332d010c..a713262269 100644 --- a/src/mesa/drivers/dri/i965/brw_state_dump.c +++ b/src/mesa/drivers/dri/i965/brw_state_dump.c @@ -84,6 +84,19 @@ get_965_surfacetype(unsigned int surfacetype) } } +static const char * +get_965_surface_format(unsigned int surface_format) +{ + switch (surface_format) { + case 0x000: return "r32g32b32a32_float"; + case 0x0c1: return "b8g8r8a8_unorm"; + case 0x100: return "b5g6r5_unorm"; + case 0x102: return "b5g5r5a1_unorm"; + case 0x104: return "b4g4r4a4_unorm"; + default: return "unknown"; + } +} + static void dump_wm_surface_state(struct brw_context *brw) { int i; @@ -95,7 +108,7 @@ static void dump_wm_surface_state(struct brw_context *brw) char name[20]; if (surf_bo == NULL) { - fprintf(stderr, "WM SS%d: NULL\n", i); + fprintf(stderr, " WM SS%d: NULL\n", i); continue; } dri_bo_map(surf_bo, GL_FALSE); @@ -103,8 +116,9 @@ static void dump_wm_surface_state(struct brw_context *brw) surf = (struct brw_surface_state *)(surf_bo->virtual); sprintf(name, "WM SS%d", i); - state_out(name, surf, surfoff, 0, "%s\n", - get_965_surfacetype(surf->ss0.surface_type)); + state_out(name, surf, surfoff, 0, "%s %s\n", + get_965_surfacetype(surf->ss0.surface_type), + get_965_surface_format(surf->ss0.surface_format)); state_out(name, surf, surfoff, 1, "offset\n"); state_out(name, surf, surfoff, 2, "%dx%d size, %d mips\n", surf->ss2.width + 1, surf->ss2.height + 1, surf->ss2.mip_count); diff --git a/src/mesa/drivers/dri/i965/brw_vs_emit.c b/src/mesa/drivers/dri/i965/brw_vs_emit.c index 42f6a99142..af0f501621 100644 --- a/src/mesa/drivers/dri/i965/brw_vs_emit.c +++ b/src/mesa/drivers/dri/i965/brw_vs_emit.c @@ -170,6 +170,14 @@ static void brw_vs_alloc_regs( struct brw_vs_compile *c ) reg++; } + if (c->use_const_buffer) { + for (i = 0; i < 3; i++) { + c->current_const[i].index = -1; + c->current_const[i].reg = brw_vec8_grf(reg, 0); + reg++; + } + } + for (i = 0; i < 128; i++) { if (c->output_regs[i].used_in_src) { c->output_regs[i].reg = brw_vec8_grf(reg, 0); @@ -194,13 +202,6 @@ static void brw_vs_alloc_regs( struct brw_vs_compile *c ) c->prog_data.urb_entry_size = (c->nr_outputs + 2 + 3) / 4; c->prog_data.total_grf = reg; - if (c->use_const_buffer) { - for (i = 0; i < 3; i++) { - c->current_const[i].index = -1; - c->current_const[i].reg = get_tmp(c); - } - } - if (INTEL_DEBUG & DEBUG_VS) { _mesa_printf("%s NumAddrRegs %d\n", __FUNCTION__, c->vp->program.Base.NumAddressRegs); _mesa_printf("%s NumTemps %d\n", __FUNCTION__, c->vp->program.Base.NumTemporaries); @@ -655,6 +656,8 @@ static void emit_lit_noalias( struct brw_vs_compile *c, } brw_ENDIF(p, if_insn); + + release_tmp(c, tmp); } static void emit_lrp_noalias(struct brw_vs_compile *c, @@ -704,7 +707,9 @@ get_constant(struct brw_vs_compile *c, struct brw_compile *p = &c->func; struct brw_reg const_reg; - if (c->current_const[argIndex].index != src->Index) { + assert(argIndex < 3); + + if (c->current_const[argIndex].index != src->Index || src->RelAddr) { c->current_const[argIndex].index = src->Index; @@ -717,15 +722,18 @@ get_constant(struct brw_vs_compile *c, brw_dp_READ_4_vs(p, c->current_const[argIndex].reg, /* writeback dest */ src->RelAddr, /* relative indexing? */ + c->regs[PROGRAM_ADDRESS][0], /* address register */ 16 * src->Index, /* byte offset */ SURF_INDEX_VERT_CONST_BUFFER /* binding table index */ ); } - /* replicate lower four floats into upper four floats (to get XYZWXYZW) */ const_reg = c->current_const[argIndex].reg; - const_reg = stride(const_reg, 0, 4, 0); - const_reg.subnr = 0; + if (!src->RelAddr) { + /* replicate lower four floats into upper half (to get XYZWXYZW) */ + const_reg = stride(const_reg, 0, 4, 0); + const_reg.subnr = 0; + } return const_reg; } @@ -767,6 +775,42 @@ static struct brw_reg get_reg( struct brw_vs_compile *c, /** + * Indirect addressing: get reg[[arg] + offset]. + */ +static struct brw_reg deref( struct brw_vs_compile *c, + struct brw_reg arg, + GLint offset) +{ + struct brw_compile *p = &c->func; + struct brw_reg tmp = vec4(get_tmp(c)); + struct brw_reg addr_reg = c->regs[PROGRAM_ADDRESS][0]; + struct brw_reg vp_address = retype(vec1(addr_reg), BRW_REGISTER_TYPE_UW); + GLuint byte_offset = arg.nr * 32 + arg.subnr + offset * 16; + struct brw_reg indirect = brw_vec4_indirect(0,0); + + { + brw_push_insn_state(p); + brw_set_access_mode(p, BRW_ALIGN_1); + + /* This is pretty clunky - load the address register twice and + * fetch each 4-dword value in turn. There must be a way to do + * this in a single pass, but I couldn't get it to work. + */ + brw_ADD(p, brw_address_reg(0), vp_address, brw_imm_d(byte_offset)); + brw_MOV(p, tmp, indirect); + + brw_ADD(p, brw_address_reg(0), suboffset(vp_address, 8), brw_imm_d(byte_offset)); + brw_MOV(p, suboffset(tmp, 4), indirect); + + brw_pop_insn_state(p); + } + + /* NOTE: tmp not released */ + return vec8(tmp); +} + + +/** * Get brw reg corresponding to the instruction's [argIndex] src reg. * TODO: relative addressing! */ @@ -777,19 +821,29 @@ get_src_reg( struct brw_vs_compile *c, { const GLuint file = inst->SrcReg[argIndex].File; const GLint index = inst->SrcReg[argIndex].Index; + const GLboolean relAddr = inst->SrcReg[argIndex].RelAddr; switch (file) { case PROGRAM_TEMPORARY: case PROGRAM_INPUT: case PROGRAM_OUTPUT: - assert(c->regs[file][index].nr != 0); - return c->regs[file][index]; + if (relAddr) { + return deref(c, c->regs[file][0], index); + } + else { + assert(c->regs[file][index].nr != 0); + return c->regs[file][index]; + } + case PROGRAM_STATE_VAR: case PROGRAM_CONSTANT: case PROGRAM_UNIFORM: if (c->use_const_buffer) { return get_constant(c, inst, argIndex); } + else if (relAddr) { + return deref(c, c->regs[PROGRAM_STATE_VAR][0], index); + } else { assert(c->regs[PROGRAM_STATE_VAR][index].nr != 0); return c->regs[PROGRAM_STATE_VAR][index]; @@ -812,41 +866,6 @@ get_src_reg( struct brw_vs_compile *c, } -/** - * Indirect addressing: get reg[[arg] + offset]. - */ -static struct brw_reg deref( struct brw_vs_compile *c, - struct brw_reg arg, - GLint offset) -{ - struct brw_compile *p = &c->func; - struct brw_reg tmp = vec4(get_tmp(c)); - struct brw_reg addr_reg = c->regs[PROGRAM_ADDRESS][0]; - struct brw_reg vp_address = retype(vec1(addr_reg), BRW_REGISTER_TYPE_UW); - GLuint byte_offset = arg.nr * 32 + arg.subnr + offset * 16; - struct brw_reg indirect = brw_vec4_indirect(0,0); - - { - brw_push_insn_state(p); - brw_set_access_mode(p, BRW_ALIGN_1); - - /* This is pretty clunky - load the address register twice and - * fetch each 4-dword value in turn. There must be a way to do - * this in a single pass, but I couldn't get it to work. - */ - brw_ADD(p, brw_address_reg(0), vp_address, brw_imm_d(byte_offset)); - brw_MOV(p, tmp, indirect); - - brw_ADD(p, brw_address_reg(0), suboffset(vp_address, 8), brw_imm_d(byte_offset)); - brw_MOV(p, suboffset(tmp, 4), indirect); - - brw_pop_insn_state(p); - } - - return vec8(tmp); -} - - static void emit_arl( struct brw_vs_compile *c, struct brw_reg dst, struct brw_reg arg0 ) @@ -858,8 +877,8 @@ static void emit_arl( struct brw_vs_compile *c, if (need_tmp) tmp = get_tmp(c); - brw_RNDD(p, tmp, arg0); - brw_MUL(p, dst, tmp, brw_imm_d(16)); + brw_RNDD(p, tmp, arg0); /* tmp = round(arg0) */ + brw_MUL(p, dst, tmp, brw_imm_d(16)); /* dst = tmp * 16 */ if (need_tmp) release_tmp(c, tmp); @@ -882,13 +901,7 @@ static struct brw_reg get_arg( struct brw_vs_compile *c, if (src->File == PROGRAM_UNDEFINED) return brw_null_reg(); - if (src->RelAddr) { - /* XXX fix */ - reg = deref(c, c->regs[PROGRAM_STATE_VAR][0], src->Index); - } - else { - reg = get_src_reg(c, inst, argIndex); - } + reg = get_src_reg(c, inst, argIndex); /* Convert 3-bit swizzle to 2-bit. */ @@ -919,6 +932,10 @@ static struct brw_reg get_dst( struct brw_vs_compile *c, assert(c->regs[dst.File][dst.Index].nr != 0); reg = c->regs[dst.File][dst.Index]; break; + case PROGRAM_ADDRESS: + assert(dst.Index == 0); + reg = c->regs[dst.File][dst.Index]; + break; case PROGRAM_UNDEFINED: /* we may hit this for OPCODE_END, OPCODE_KIL, etc */ reg = brw_null_reg(); @@ -979,10 +996,7 @@ static void emit_swz( struct brw_vs_compile *c, if (src_mask) { struct brw_reg arg0; - if (src.RelAddr) - arg0 = deref(c, c->regs[PROGRAM_STATE_VAR][0], src.Index); - else - arg0 = get_src_reg(c, inst, argIndex); + arg0 = get_src_reg(c, inst, argIndex); arg0 = brw_swizzle(arg0, src_swz[0], src_swz[1], @@ -1178,6 +1192,11 @@ void brw_vs_emit(struct brw_vs_compile *c ) struct brw_reg args[3], dst; GLuint i; +#if 0 + printf("%d: ", insn); + _mesa_print_instruction(inst); +#endif + /* Get argument regs. SWZ is special and does this itself. */ if (inst->Opcode != OPCODE_SWZ) diff --git a/src/mesa/drivers/dri/i965/brw_wm_glsl.c b/src/mesa/drivers/dri/i965/brw_wm_glsl.c index 385efd2dd3..2d13088ddb 100644 --- a/src/mesa/drivers/dri/i965/brw_wm_glsl.c +++ b/src/mesa/drivers/dri/i965/brw_wm_glsl.c @@ -346,10 +346,10 @@ get_src_reg_const(struct brw_wm_compile *c, const_reg = brw_abs(const_reg); #if 0 - printf(" form const[%d] for arg %d, comp %d, reg %d\n", + printf(" form const[%d].%d for arg %d, reg %d\n", c->current_const[srcRegIndex].index, - srcRegIndex, component, + srcRegIndex, const_reg.nr); #endif @@ -407,7 +407,7 @@ static struct brw_reg get_src_reg_imm(struct brw_wm_compile *c, if (src->Abs) value = FABSF(value); #if 0 - printf(" form imm reg %f\n", value); + printf(" form immed value %f for chan %d\n", value, channel); #endif return brw_imm_f(value); } @@ -527,7 +527,9 @@ static void emit_mov( struct brw_wm_compile *c, if (mask & (1<<i)) { struct brw_reg src, dst; dst = get_dst_reg(c, inst, i); - src = get_src_reg_imm(c, inst, 0, i); + /* XXX some moves from immediate value don't work reliably!!! */ + /*src = get_src_reg_imm(c, inst, 0, i);*/ + src = get_src_reg(c, inst, 0, i); brw_MOV(p, dst, src); } } diff --git a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c index ce5dbb334b..0dc377be65 100644 --- a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c +++ b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c @@ -38,7 +38,7 @@ #include "intel_mipmap_tree.h" #include "intel_batchbuffer.h" #include "intel_tex.h" - +#include "intel_fbo.h" #include "brw_context.h" #include "brw_state.h" @@ -505,15 +505,18 @@ brw_update_vs_constant_surface( GLcontext *ctx, * usable for further buffers when doing ARB_draw_buffer support. */ static void -brw_update_region_surface(struct brw_context *brw, struct intel_region *region, - unsigned int unit, GLboolean cached) +brw_update_renderbuffer_surface(struct brw_context *brw, + struct gl_renderbuffer *rb, + unsigned int unit, GLboolean cached) { GLcontext *ctx = &brw->intel.ctx; dri_bo *region_bo = NULL; + struct intel_renderbuffer *irb = intel_renderbuffer(rb); + struct intel_region *region = irb ? irb->region : NULL; struct { unsigned int surface_type; unsigned int surface_format; - unsigned int width, height, cpp; + unsigned int width, height, pitch, cpp; GLubyte color_mask[4]; GLboolean color_blend; uint32_t tiling; @@ -525,13 +528,27 @@ brw_update_region_surface(struct brw_context *brw, struct intel_region *region, region_bo = region->buffer; key.surface_type = BRW_SURFACE_2D; - if (region->cpp == 4) + switch (irb->texformat->MesaFormat) { + case MESA_FORMAT_ARGB8888: key.surface_format = BRW_SURFACEFORMAT_B8G8R8A8_UNORM; - else + break; + case MESA_FORMAT_RGB565: key.surface_format = BRW_SURFACEFORMAT_B5G6R5_UNORM; + break; + case MESA_FORMAT_ARGB1555: + key.surface_format = BRW_SURFACEFORMAT_B5G5R5A1_UNORM; + break; + case MESA_FORMAT_ARGB4444: + key.surface_format = BRW_SURFACEFORMAT_B4G4R4A4_UNORM; + break; + default: + _mesa_problem(ctx, "Bad renderbuffer format: %d\n", + irb->texformat->MesaFormat); + } key.tiling = region->tiling; - key.width = region->pitch; /* XXX: not really! */ - key.height = region->height; + key.width = rb->Width; + key.height = rb->Height; + key.pitch = region->pitch; key.cpp = region->cpp; } else { key.surface_type = BRW_SURFACE_NULL; @@ -567,7 +584,7 @@ brw_update_region_surface(struct brw_context *brw, struct intel_region *region, surf.ss2.width = key.width - 1; surf.ss2.height = key.height - 1; brw_set_surface_tiling(&surf, key.tiling); - surf.ss3.pitch = (key.width * key.cpp) - 1; + surf.ss3.pitch = (key.pitch * key.cpp) - 1; /* _NEW_COLOR */ surf.ss0.color_blend = key.color_blend; @@ -655,14 +672,17 @@ static void prepare_wm_surfaces(struct brw_context *brw ) GLuint i; int old_nr_surfaces; + /* _NEW_BUFFERS */ /* Update surfaces for drawing buffers */ - if (brw->state.nr_color_regions > 1) { - for (i = 0; i < brw->state.nr_color_regions; i++) { - brw_update_region_surface(brw, brw->state.color_regions[i], i, - GL_FALSE); + if (ctx->DrawBuffer->_NumColorDrawBuffers >= 1) { + for (i = 0; i < ctx->DrawBuffer->_NumColorDrawBuffers; i++) { + brw_update_renderbuffer_surface(brw, + ctx->DrawBuffer->_ColorDrawBuffers[i], + i, + GL_FALSE); } } else { - brw_update_region_surface(brw, brw->state.color_regions[0], 0, GL_TRUE); + brw_update_renderbuffer_surface(brw, NULL, 0, GL_TRUE); } old_nr_surfaces = brw->wm.nr_surfaces; diff --git a/src/mesa/drivers/dri/intel/intel_blit.c b/src/mesa/drivers/dri/intel/intel_blit.c index 4ae9b118a3..4919828131 100644 --- a/src/mesa/drivers/dri/intel/intel_blit.c +++ b/src/mesa/drivers/dri/intel/intel_blit.c @@ -32,6 +32,8 @@ #include "main/mtypes.h" #include "main/context.h" #include "main/enums.h" +#include "main/texformat.h" +#include "main/colormac.h" #include "intel_blit.h" #include "intel_buffers.h" @@ -484,10 +486,9 @@ intelClearWithBlit(GLcontext *ctx, GLbitfield mask) const GLbitfield bufBit = 1 << buf; if ((clearMask & bufBit) && !(bufBit & skipBuffers)) { /* OK, clear this renderbuffer */ - struct intel_region *irb_region = - intel_get_rb_region(fb, buf); + struct intel_renderbuffer *irb = intel_get_renderbuffer(fb, buf); dri_bo *write_buffer = - intel_region_buffer(intel, irb_region, + intel_region_buffer(intel, irb->region, all ? INTEL_WRITE_FULL : INTEL_WRITE_PART); @@ -495,15 +496,13 @@ intelClearWithBlit(GLcontext *ctx, GLbitfield mask) GLint pitch, cpp; GLuint BR13, CMD; - ASSERT(irb_region); - - pitch = irb_region->pitch; - cpp = irb_region->cpp; + pitch = irb->region->pitch; + cpp = irb->region->cpp; DBG("%s dst:buf(%p)/%d+%d %d,%d sz:%dx%d\n", __FUNCTION__, - irb_region->buffer, (pitch * cpp), - irb_region->draw_offset, + irb->region->buffer, (pitch * cpp), + irb->region->draw_offset, b.x1, b.y1, b.x2 - b.x1, b.y2 - b.y1); BR13 = 0xf0 << 16; @@ -529,7 +528,7 @@ intelClearWithBlit(GLcontext *ctx, GLbitfield mask) } #ifndef I915 - if (irb_region->tiling != I915_TILING_NONE) { + if (irb->region->tiling != I915_TILING_NONE) { CMD |= XY_DST_TILED; pitch /= 4; } @@ -540,9 +539,36 @@ intelClearWithBlit(GLcontext *ctx, GLbitfield mask) clearVal = clear_depth; } else { - clearVal = (cpp == 4) - ? intel->ClearColor8888 : intel->ClearColor565; - } + uint8_t clear[4]; + GLclampf *color = ctx->Color.ClearColor; + + CLAMPED_FLOAT_TO_UBYTE(clear[0], color[0]); + CLAMPED_FLOAT_TO_UBYTE(clear[1], color[1]); + CLAMPED_FLOAT_TO_UBYTE(clear[2], color[2]); + CLAMPED_FLOAT_TO_UBYTE(clear[3], color[3]); + + switch (irb->texformat->MesaFormat) { + case MESA_FORMAT_ARGB8888: + clearVal = intel->ClearColor8888; + break; + case MESA_FORMAT_RGB565: + clearVal = intel->ClearColor565; + break; + case MESA_FORMAT_ARGB4444: + clearVal = PACK_COLOR_4444(clear[3], clear[0], + clear[1], clear[2]); + break; + case MESA_FORMAT_ARGB1555: + clearVal = PACK_COLOR_1555(clear[3], clear[0], + clear[1], clear[2]); + break; + default: + _mesa_problem(ctx, "Unexpected renderbuffer format: %d\n", + irb->texformat->MesaFormat); + clearVal = 0; + } + } + /* _mesa_debug(ctx, "hardware blit clear buf %d rb id %d\n", buf, irb->Base.Name); @@ -558,7 +584,7 @@ intelClearWithBlit(GLcontext *ctx, GLbitfield mask) OUT_BATCH((b.y2 << 16) | b.x2); OUT_RELOC(write_buffer, I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER, - irb_region->draw_offset); + irb->region->draw_offset); OUT_BATCH(clearVal); ADVANCE_BATCH(); clearMask &= ~bufBit; /* turn off bit, for faster loop exit */ diff --git a/src/mesa/drivers/dri/intel/intel_context.h b/src/mesa/drivers/dri/intel/intel_context.h index bd3810549a..d798225ddd 100644 --- a/src/mesa/drivers/dri/intel/intel_context.h +++ b/src/mesa/drivers/dri/intel/intel_context.h @@ -48,6 +48,8 @@ #define DV_PF_555 (1<<8) #define DV_PF_565 (2<<8) #define DV_PF_8888 (3<<8) +#define DV_PF_4444 (8<<8) +#define DV_PF_1555 (9<<8) struct intel_region; struct intel_context; @@ -337,6 +339,7 @@ extern char *__progname; #define ARRAY_SIZE(x) (sizeof(x) / sizeof(x[0])) #define ALIGN(value, alignment) ((value + alignment - 1) & ~(alignment - 1)) +#define IS_POWER_OF_TWO(val) (((val) & (val - 1)) == 0) #define INTEL_FIREVERTICES(intel) \ do { \ diff --git a/src/mesa/drivers/dri/intel/intel_decode.c b/src/mesa/drivers/dri/intel/intel_decode.c index f04638206d..a9dfe281cb 100644 --- a/src/mesa/drivers/dri/intel/intel_decode.c +++ b/src/mesa/drivers/dri/intel/intel_decode.c @@ -800,6 +800,7 @@ static int decode_3d_1d(uint32_t *data, int count, uint32_t hw_offset, int *failures, int i830) { unsigned int len, i, c, opcode, word, map, sampler, instr; + char *format; struct { uint32_t opcode; @@ -1001,6 +1002,35 @@ decode_3d_1d(uint32_t *data, int count, uint32_t hw_offset, int *failures, int i (*failures)++; } return len; + case 0x85: + len = (data[0] & 0x0000000f) + 2; + + if (len != 2) + fprintf(out, "Bad count in 3DSTATE_DEST_BUFFER_VARIABLES\n"); + if (count < 2) + BUFFER_FAIL(count, len, "3DSTATE_DEST_BUFFER_VARIABLES"); + + instr_out(data, hw_offset, 0, + "3DSTATE_DEST_BUFFER_VARIABLES\n"); + + switch ((data[1] >> 8) & 0xf) { + case 0x0: format = "g8"; break; + case 0x1: format = "x1r5g5b5"; break; + case 0x2: format = "r5g6b5"; break; + case 0x3: format = "a8r8g8b8"; break; + case 0x4: format = "ycrcb_swapy"; break; + case 0x5: format = "ycrcb_normal"; break; + case 0x6: format = "ycrcb_swapuv"; break; + case 0x7: format = "ycrcb_swapuvy"; break; + case 0x8: format = "a4r4g4b4"; break; + case 0x9: format = "a1r5g5b5"; break; + case 0xa: format = "a2r10g10b10"; break; + default: format = "BAD"; break; + } + instr_out(data, hw_offset, 1, "%s format, early Z %sabled\n", + format, + (data[1] & (1 << 31)) ? "en" : "dis"); + return len; } for (opcode = 0; opcode < sizeof(opcodes_3d_1d) / sizeof(opcodes_3d_1d[0]); diff --git a/src/mesa/drivers/dri/intel/intel_fbo.c b/src/mesa/drivers/dri/intel/intel_fbo.c index a401f730ba..52647ddf8b 100644 --- a/src/mesa/drivers/dri/intel/intel_fbo.c +++ b/src/mesa/drivers/dri/intel/intel_fbo.c @@ -119,6 +119,7 @@ intel_alloc_renderbuffer_storage(GLcontext * ctx, struct gl_renderbuffer *rb, rb->RedBits = 5; rb->GreenBits = 6; rb->BlueBits = 5; + irb->texformat = &_mesa_texformat_rgb565; cpp = 2; break; case GL_RGB: @@ -132,6 +133,7 @@ intel_alloc_renderbuffer_storage(GLcontext * ctx, struct gl_renderbuffer *rb, rb->GreenBits = 8; rb->BlueBits = 8; rb->AlphaBits = 0; + irb->texformat = &_mesa_texformat_argb8888; /* XXX: Need xrgb8888 */ cpp = 4; break; case GL_RGBA: @@ -148,6 +150,7 @@ intel_alloc_renderbuffer_storage(GLcontext * ctx, struct gl_renderbuffer *rb, rb->GreenBits = 8; rb->BlueBits = 8; rb->AlphaBits = 8; + irb->texformat = &_mesa_texformat_argb8888; cpp = 4; break; case GL_STENCIL_INDEX: @@ -160,12 +163,14 @@ intel_alloc_renderbuffer_storage(GLcontext * ctx, struct gl_renderbuffer *rb, rb->DataType = GL_UNSIGNED_INT_24_8_EXT; rb->StencilBits = 8; cpp = 4; + irb->texformat = &_mesa_texformat_s8_z24; break; case GL_DEPTH_COMPONENT16: rb->_ActualFormat = GL_DEPTH_COMPONENT16; rb->DataType = GL_UNSIGNED_SHORT; rb->DepthBits = 16; cpp = 2; + irb->texformat = &_mesa_texformat_z16; break; case GL_DEPTH_COMPONENT: case GL_DEPTH_COMPONENT24: @@ -174,6 +179,7 @@ intel_alloc_renderbuffer_storage(GLcontext * ctx, struct gl_renderbuffer *rb, rb->DataType = GL_UNSIGNED_INT_24_8_EXT; rb->DepthBits = 24; cpp = 4; + irb->texformat = &_mesa_texformat_s8_z24; break; case GL_DEPTH_STENCIL_EXT: case GL_DEPTH24_STENCIL8_EXT: @@ -182,6 +188,7 @@ intel_alloc_renderbuffer_storage(GLcontext * ctx, struct gl_renderbuffer *rb, rb->DepthBits = 24; rb->StencilBits = 8; cpp = 4; + irb->texformat = &_mesa_texformat_s8_z24; break; default: _mesa_problem(ctx, @@ -322,6 +329,7 @@ intel_create_renderbuffer(GLenum intFormat) irb->Base.GreenBits = 6; irb->Base.BlueBits = 5; irb->Base.DataType = GL_UNSIGNED_BYTE; + irb->texformat = &_mesa_texformat_rgb565; break; case GL_RGB8: irb->Base._ActualFormat = GL_RGB8; @@ -331,6 +339,7 @@ intel_create_renderbuffer(GLenum intFormat) irb->Base.BlueBits = 8; irb->Base.AlphaBits = 0; irb->Base.DataType = GL_UNSIGNED_BYTE; + irb->texformat = &_mesa_texformat_argb8888; /* XXX: Need xrgb8888 */ break; case GL_RGBA8: irb->Base._ActualFormat = GL_RGBA8; @@ -340,24 +349,28 @@ intel_create_renderbuffer(GLenum intFormat) irb->Base.BlueBits = 8; irb->Base.AlphaBits = 8; irb->Base.DataType = GL_UNSIGNED_BYTE; + irb->texformat = &_mesa_texformat_argb8888; break; case GL_STENCIL_INDEX8_EXT: irb->Base._ActualFormat = GL_STENCIL_INDEX8_EXT; irb->Base._BaseFormat = GL_STENCIL_INDEX; irb->Base.StencilBits = 8; irb->Base.DataType = GL_UNSIGNED_BYTE; + irb->texformat = &_mesa_texformat_s8_z24; break; case GL_DEPTH_COMPONENT16: irb->Base._ActualFormat = GL_DEPTH_COMPONENT16; irb->Base._BaseFormat = GL_DEPTH_COMPONENT; irb->Base.DepthBits = 16; irb->Base.DataType = GL_UNSIGNED_SHORT; + irb->texformat = &_mesa_texformat_z16; break; case GL_DEPTH_COMPONENT24: irb->Base._ActualFormat = GL_DEPTH24_STENCIL8_EXT; irb->Base._BaseFormat = GL_DEPTH_COMPONENT; irb->Base.DepthBits = 24; irb->Base.DataType = GL_UNSIGNED_INT; + irb->texformat = &_mesa_texformat_s8_z24; break; case GL_DEPTH24_STENCIL8_EXT: irb->Base._ActualFormat = GL_DEPTH24_STENCIL8_EXT; @@ -365,6 +378,7 @@ intel_create_renderbuffer(GLenum intFormat) irb->Base.DepthBits = 24; irb->Base.StencilBits = 8; irb->Base.DataType = GL_UNSIGNED_INT_24_8_EXT; + irb->texformat = &_mesa_texformat_s8_z24; break; default: _mesa_problem(NULL, @@ -449,6 +463,8 @@ static GLboolean intel_update_wrapper(GLcontext *ctx, struct intel_renderbuffer *irb, struct gl_texture_image *texImage) { + irb->texformat = texImage->TexFormat; + if (texImage->TexFormat == &_mesa_texformat_argb8888) { irb->Base._ActualFormat = GL_RGBA8; irb->Base._BaseFormat = GL_RGBA; @@ -458,9 +474,21 @@ intel_update_wrapper(GLcontext *ctx, struct intel_renderbuffer *irb, else if (texImage->TexFormat == &_mesa_texformat_rgb565) { irb->Base._ActualFormat = GL_RGB5; irb->Base._BaseFormat = GL_RGB; - irb->Base.DataType = GL_UNSIGNED_SHORT; + irb->Base.DataType = GL_UNSIGNED_BYTE; DBG("Render to RGB5 texture OK\n"); } + else if (texImage->TexFormat == &_mesa_texformat_argb1555) { + irb->Base._ActualFormat = GL_RGB5_A1; + irb->Base._BaseFormat = GL_RGBA; + irb->Base.DataType = GL_UNSIGNED_BYTE; + DBG("Render to ARGB1555 texture OK\n"); + } + else if (texImage->TexFormat == &_mesa_texformat_argb4444) { + irb->Base._ActualFormat = GL_RGBA4; + irb->Base._BaseFormat = GL_RGBA; + irb->Base.DataType = GL_UNSIGNED_BYTE; + DBG("Render to ARGB4444 texture OK\n"); + } else if (texImage->TexFormat == &_mesa_texformat_z16) { irb->Base._ActualFormat = GL_DEPTH_COMPONENT16; irb->Base._BaseFormat = GL_DEPTH_COMPONENT; @@ -631,11 +659,11 @@ intel_finish_render_texture(GLcontext * ctx, static void intel_validate_framebuffer(GLcontext *ctx, struct gl_framebuffer *fb) { - struct intel_context *intel = intel_context(ctx); const struct intel_renderbuffer *depthRb = intel_get_renderbuffer(fb, BUFFER_DEPTH); const struct intel_renderbuffer *stencilRb = intel_get_renderbuffer(fb, BUFFER_STENCIL); + int i; if (stencilRb && stencilRb != depthRb) { /* we only support combined depth/stencil buffers, not separate @@ -644,32 +672,21 @@ intel_validate_framebuffer(GLcontext *ctx, struct gl_framebuffer *fb) fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT; } - /* check that texture color buffers are a format we can render into */ - { - const struct gl_texture_format *supportedFormat; - GLuint i; - - /* The texture format we can render into seems to depend on the - * screen depth. There currently seems to be a problem when - * rendering into a rgb565 texture when the screen is abgr8888. - */ + for (i = 0; i < ctx->Const.MaxDrawBuffers; i++) { + struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[i]; + struct intel_renderbuffer *irb = intel_renderbuffer(rb); - if (intel->ctx.Visual.rgbBits >= 24) - supportedFormat = &_mesa_texformat_argb8888; - else - supportedFormat = &_mesa_texformat_rgb565; + if (rb == NULL) + continue; - for (i = 0; i < ctx->Const.MaxDrawBuffers; i++) { - const struct gl_texture_object *texObj = - fb->Attachment[BUFFER_COLOR0 + i].Texture; - if (texObj) { - const struct gl_texture_image *texImg = - texObj->Image[0][texObj->BaseLevel]; - if (texImg && texImg->TexFormat != supportedFormat) { - fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT; - break; - } - } + switch (irb->texformat->MesaFormat) { + case MESA_FORMAT_ARGB8888: + case MESA_FORMAT_RGB565: + case MESA_FORMAT_ARGB1555: + case MESA_FORMAT_ARGB4444: + break; + default: + fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT; } } } diff --git a/src/mesa/drivers/dri/intel/intel_fbo.h b/src/mesa/drivers/dri/intel/intel_fbo.h index 7226ee026f..f0665af482 100644 --- a/src/mesa/drivers/dri/intel/intel_fbo.h +++ b/src/mesa/drivers/dri/intel/intel_fbo.h @@ -61,6 +61,8 @@ struct intel_renderbuffer struct gl_renderbuffer Base; struct intel_region *region; + const struct gl_texture_format *texformat; + GLuint vbl_pending; /**< vblank sequence number of pending flip */ uint8_t *span_cache; diff --git a/src/mesa/drivers/dri/intel/intel_span.c b/src/mesa/drivers/dri/intel/intel_span.c index c3a873f1ab..34b78ebc1a 100644 --- a/src/mesa/drivers/dri/intel/intel_span.c +++ b/src/mesa/drivers/dri/intel/intel_span.c @@ -29,6 +29,7 @@ #include "main/macros.h" #include "main/mtypes.h" #include "main/colormac.h" +#include "main/texformat.h" #include "intel_buffers.h" #include "intel_fbo.h" @@ -313,6 +314,22 @@ static uint32_t y_tile_swizzle(struct intel_renderbuffer *irb, #define INTEL_TAG(x) x##_RGB565 #include "intel_spantmp.h" +/* a4r4g4b4 color span and pixel functions */ +#define INTEL_PIXEL_FMT GL_BGRA +#define INTEL_PIXEL_TYPE GL_UNSIGNED_SHORT_4_4_4_4_REV +#define INTEL_READ_VALUE(offset) pread_16(irb, offset) +#define INTEL_WRITE_VALUE(offset, v) pwrite_16(irb, offset, v) +#define INTEL_TAG(x) x##_ARGB4444 +#include "intel_spantmp.h" + +/* a1r5g5b5 color span and pixel functions */ +#define INTEL_PIXEL_FMT GL_BGRA +#define INTEL_PIXEL_TYPE GL_UNSIGNED_SHORT_1_5_5_5_REV +#define INTEL_READ_VALUE(offset) pread_16(irb, offset) +#define INTEL_WRITE_VALUE(offset, v) pwrite_16(irb, offset, v) +#define INTEL_TAG(x) x##_ARGB1555 +#include "intel_spantmp.h" + /* a8r8g8b8 color span and pixel functions */ #define INTEL_PIXEL_FMT GL_BGRA #define INTEL_PIXEL_TYPE GL_UNSIGNED_INT_8_8_8_8_REV @@ -561,8 +578,8 @@ intel_set_span_functions(struct intel_context *intel, else tiling = I915_TILING_NONE; - if (rb->_ActualFormat == GL_RGB5) { - /* 565 RGB */ + switch (irb->texformat->MesaFormat) { + case MESA_FORMAT_RGB565: switch (tiling) { case I915_TILING_NONE: default: @@ -575,38 +592,67 @@ intel_set_span_functions(struct intel_context *intel, intel_YTile_InitPointers_RGB565(rb); break; } - } - else if (rb->_ActualFormat == GL_RGB8) { - /* 8888 RGBx */ + break; + case MESA_FORMAT_ARGB4444: switch (tiling) { case I915_TILING_NONE: default: - intelInitPointers_xRGB8888(rb); + intelInitPointers_ARGB4444(rb); break; case I915_TILING_X: - intel_XTile_InitPointers_xRGB8888(rb); + intel_XTile_InitPointers_ARGB4444(rb); break; case I915_TILING_Y: - intel_YTile_InitPointers_xRGB8888(rb); + intel_YTile_InitPointers_ARGB4444(rb); break; } - } - else if (rb->_ActualFormat == GL_RGBA8) { - /* 8888 RGBA */ + break; + case MESA_FORMAT_ARGB1555: switch (tiling) { case I915_TILING_NONE: default: - intelInitPointers_ARGB8888(rb); + intelInitPointers_ARGB1555(rb); break; case I915_TILING_X: - intel_XTile_InitPointers_ARGB8888(rb); + intel_XTile_InitPointers_ARGB1555(rb); break; case I915_TILING_Y: - intel_YTile_InitPointers_ARGB8888(rb); + intel_YTile_InitPointers_ARGB1555(rb); break; } - } - else if (rb->_ActualFormat == GL_DEPTH_COMPONENT16) { + break; + case MESA_FORMAT_ARGB8888: + if (rb->AlphaBits == 0) { /* XXX: Need xRGB8888 Mesa format */ + /* 8888 RGBx */ + switch (tiling) { + case I915_TILING_NONE: + default: + intelInitPointers_xRGB8888(rb); + break; + case I915_TILING_X: + intel_XTile_InitPointers_xRGB8888(rb); + break; + case I915_TILING_Y: + intel_YTile_InitPointers_xRGB8888(rb); + break; + } + } else { + /* 8888 RGBA */ + switch (tiling) { + case I915_TILING_NONE: + default: + intelInitPointers_ARGB8888(rb); + break; + case I915_TILING_X: + intel_XTile_InitPointers_ARGB8888(rb); + break; + case I915_TILING_Y: + intel_YTile_InitPointers_ARGB8888(rb); + break; + } + } + break; + case MESA_FORMAT_Z16: switch (tiling) { case I915_TILING_NONE: default: @@ -619,51 +665,57 @@ intel_set_span_functions(struct intel_context *intel, intel_YTile_InitDepthPointers_z16(rb); break; } - } - else if (rb->_ActualFormat == GL_DEPTH_COMPONENT24) { - switch (tiling) { - case I915_TILING_NONE: - default: - intelInitDepthPointers_z24(rb); - break; - case I915_TILING_X: - intel_XTile_InitDepthPointers_z24(rb); - break; - case I915_TILING_Y: - intel_YTile_InitDepthPointers_z24(rb); - break; - } - } - else if (rb->_ActualFormat == GL_DEPTH24_STENCIL8_EXT) { - switch (tiling) { - case I915_TILING_NONE: - default: - intelInitDepthPointers_z24_s8(rb); - break; - case I915_TILING_X: - intel_XTile_InitDepthPointers_z24_s8(rb); - break; - case I915_TILING_Y: - intel_YTile_InitDepthPointers_z24_s8(rb); - break; - } - } - else if (rb->_ActualFormat == GL_STENCIL_INDEX8_EXT) { - switch (tiling) { - case I915_TILING_NONE: - default: - intelInitStencilPointers_z24_s8(rb); - break; - case I915_TILING_X: - intel_XTile_InitStencilPointers_z24_s8(rb); - break; - case I915_TILING_Y: - intel_YTile_InitStencilPointers_z24_s8(rb); - break; + break; + case MESA_FORMAT_S8_Z24: + /* There are a few different ways SW asks us to access the S8Z24 data: + * Z24 depth-only depth reads + * S8Z24 depth reads + * S8Z24 stencil reads. + */ + if (rb->_ActualFormat == GL_DEPTH_COMPONENT24) { + switch (tiling) { + case I915_TILING_NONE: + default: + intelInitDepthPointers_z24(rb); + break; + case I915_TILING_X: + intel_XTile_InitDepthPointers_z24(rb); + break; + case I915_TILING_Y: + intel_YTile_InitDepthPointers_z24(rb); + break; + } + } else if (rb->_ActualFormat == GL_DEPTH24_STENCIL8_EXT) { + switch (tiling) { + case I915_TILING_NONE: + default: + intelInitDepthPointers_z24_s8(rb); + break; + case I915_TILING_X: + intel_XTile_InitDepthPointers_z24_s8(rb); + break; + case I915_TILING_Y: + intel_YTile_InitDepthPointers_z24_s8(rb); + break; + } + } else if (rb->_ActualFormat == GL_STENCIL_INDEX8_EXT) { + switch (tiling) { + case I915_TILING_NONE: + default: + intelInitStencilPointers_z24_s8(rb); + break; + case I915_TILING_X: + intel_XTile_InitStencilPointers_z24_s8(rb); + break; + case I915_TILING_Y: + intel_YTile_InitStencilPointers_z24_s8(rb); + break; + } } - } - else { + break; + default: _mesa_problem(NULL, - "Unexpected _ActualFormat in intelSetSpanFunctions"); + "Unexpected MesaFormat in intelSetSpanFunctions"); + break; } } diff --git a/src/mesa/drivers/dri/intel/intel_tex_image.c b/src/mesa/drivers/dri/intel/intel_tex_image.c index 71561cf85c..1f192dafbe 100644 --- a/src/mesa/drivers/dri/intel/intel_tex_image.c +++ b/src/mesa/drivers/dri/intel/intel_tex_image.c @@ -315,8 +315,8 @@ intelTexImage(GLcontext * ctx, GLint postConvWidth = width; GLint postConvHeight = height; GLint texelBytes, sizeInBytes; - GLuint dstRowStride, srcRowStride = texImage->RowStride; - + GLuint dstRowStride = 0, srcRowStride = texImage->RowStride; + GLboolean needs_map; DBG("%s target %s level %d %dx%dx%d border %d\n", __FUNCTION__, _mesa_lookup_enum_by_nr(target), level, width, height, depth, border); @@ -482,8 +482,15 @@ intelTexImage(GLcontext * ctx, LOCK_HARDWARE(intel); + /* Two cases where we need a mapping of the miptree: when the user supplied + * data is mapped as well (non-PBO, memcpy upload) or when we're going to do + * (software) mipmap generation. + */ + needs_map = (pixels != NULL) || (level == texObj->BaseLevel && + texObj->GenerateMipmap); + if (intelImage->mt) { - if (pixels) + if (needs_map) texImage->Data = intel_miptree_image_map(intel, intelImage->mt, intelImage->face, @@ -509,8 +516,9 @@ intelTexImage(GLcontext * ctx, } DBG("Upload image %dx%dx%d row_len %d " - "pitch %d\n", - width, height, depth, width * texelBytes, dstRowStride); + "pitch %d pixels %d compressed %d\n", + width, height, depth, width * texelBytes, dstRowStride, + pixels ? 1 : 0, compressed); /* Copy data. Would like to know when it's ok for us to eg. use * the blitter to copy. Or, use the hardware to do the format @@ -523,7 +531,7 @@ intelTexImage(GLcontext * ctx, _mesa_copy_rect(texImage->Data, dst->cpp, dst->pitch, 0, 0, intelImage->mt->level[level].width, - intelImage->mt->level[level].height/4, + (intelImage->mt->level[level].height+3)/4, pixels, srcRowStride, 0, 0); @@ -549,7 +557,7 @@ intelTexImage(GLcontext * ctx, _mesa_unmap_teximage_pbo(ctx, unpack); if (intelImage->mt) { - if (pixels) + if (needs_map) intel_miptree_image_unmap(intel, intelImage->mt); texImage->Data = NULL; } diff --git a/src/mesa/main/attrib.c b/src/mesa/main/attrib.c index d5d0a552db..e43fa96dd3 100644 --- a/src/mesa/main/attrib.c +++ b/src/mesa/main/attrib.c @@ -1265,7 +1265,7 @@ _mesa_PopAttrib(void) /* restore clip planes */ for (i = 0; i < MAX_CLIP_PLANES; i++) { - const GLuint mask = 1 << 1; + const GLuint mask = 1 << i; const GLfloat *eyePlane = xform->EyeUserPlane[i]; COPY_4V(ctx->Transform.EyeUserPlane[i], eyePlane); if (xform->ClipPlanesEnabled & mask) { diff --git a/src/mesa/main/state.c b/src/mesa/main/state.c index cc37d63636..4f9088dd22 100644 --- a/src/mesa/main/state.c +++ b/src/mesa/main/state.c @@ -470,7 +470,8 @@ _mesa_update_state_locked( GLcontext *ctx ) /* Determine which state flags effect vertex/fragment program state */ if (ctx->FragmentProgram._MaintainTexEnvProgram) { - prog_flags |= (_NEW_TEXTURE | _NEW_FOG | _DD_NEW_SEPARATE_SPECULAR); + prog_flags |= (_NEW_TEXTURE | _NEW_FOG | _DD_NEW_SEPARATE_SPECULAR | + _NEW_ARRAY); } if (ctx->VertexProgram._MaintainTnlProgram) { prog_flags |= (_NEW_ARRAY | _NEW_TEXTURE | _NEW_TEXTURE_MATRIX | diff --git a/src/mesa/main/texcompress_s3tc.c b/src/mesa/main/texcompress_s3tc.c index d17e18da6b..a1c0f18f36 100644 --- a/src/mesa/main/texcompress_s3tc.c +++ b/src/mesa/main/texcompress_s3tc.c @@ -205,7 +205,7 @@ texstore_rgb_dxt1(TEXSTORE_PARAMS) dst, dstRowStride); } else { - _mesa_warning(ctx, "external dxt library not available"); + _mesa_warning(ctx, "external dxt library not available: texstore_rgb_dxt1"); } if (tempImage) @@ -267,7 +267,7 @@ texstore_rgba_dxt1(TEXSTORE_PARAMS) dst, dstRowStride); } else { - _mesa_warning(ctx, "external dxt library not available"); + _mesa_warning(ctx, "external dxt library not available: texstore_rgba_dxt1"); } if (tempImage) @@ -328,7 +328,7 @@ texstore_rgba_dxt3(TEXSTORE_PARAMS) dst, dstRowStride); } else { - _mesa_warning(ctx, "external dxt library not available"); + _mesa_warning(ctx, "external dxt library not available: texstore_rgba_dxt3"); } if (tempImage) @@ -389,7 +389,7 @@ texstore_rgba_dxt5(TEXSTORE_PARAMS) dst, dstRowStride); } else { - _mesa_warning(ctx, "external dxt library not available"); + _mesa_warning(ctx, "external dxt library not available: texstore_rgba_dxt5"); } if (tempImage) @@ -410,7 +410,7 @@ fetch_texel_2d_rgb_dxt1( const struct gl_texture_image *texImage, (GLubyte *)(texImage)->Data, i, j, texel); } else - _mesa_debug(NULL, "attempted to decode s3tc texture without library available\n"); + _mesa_debug(NULL, "attempted to decode s3tc texture without library available: fetch_texel_2d_rgb_dxt1"); } @@ -438,7 +438,7 @@ fetch_texel_2d_rgba_dxt1( const struct gl_texture_image *texImage, (GLubyte *)(texImage)->Data, i, j, texel); } else - _mesa_debug(NULL, "attempted to decode s3tc texture without library available\n"); + _mesa_debug(NULL, "attempted to decode s3tc texture without library available: fetch_texel_2d_rgba_dxt1\n"); } @@ -467,7 +467,7 @@ fetch_texel_2d_rgba_dxt3( const struct gl_texture_image *texImage, i, j, texel); } else - _mesa_debug(NULL, "attempted to decode s3tc texture without library available\n"); + _mesa_debug(NULL, "attempted to decode s3tc texture without library available: fetch_texel_2d_rgba_dxt3\n"); } @@ -495,7 +495,7 @@ fetch_texel_2d_rgba_dxt5( const struct gl_texture_image *texImage, i, j, texel); } else - _mesa_debug(NULL, "attempted to decode s3tc texture without library available\n"); + _mesa_debug(NULL, "attempted to decode s3tc texture without library available: fetch_texel_2d_rgba_dxt5\n"); } diff --git a/src/mesa/state_tracker/st_cb_accum.c b/src/mesa/state_tracker/st_cb_accum.c index 3f9a825a15..1510a1e236 100644 --- a/src/mesa/state_tracker/st_cb_accum.c +++ b/src/mesa/state_tracker/st_cb_accum.c @@ -40,6 +40,7 @@ #include "st_draw.h" #include "st_public.h" #include "st_format.h" +#include "st_texture.h" #include "pipe/p_context.h" #include "pipe/p_defines.h" #include "pipe/p_inlines.h" @@ -118,6 +119,9 @@ st_clear_accum_buffer(GLcontext *ctx, struct gl_renderbuffer *rb) const GLint height = ctx->DrawBuffer->_Ymax - ypos; GLubyte *map; + st_teximage_flush_before_map(ctx->st, acc_strb->texture, 0, 0, + PIPE_TRANSFER_WRITE); + acc_pt = screen->get_tex_transfer(screen, acc_strb->texture, 0, 0, 0, PIPE_TRANSFER_WRITE, xpos, ypos, width, height); @@ -163,6 +167,9 @@ accum_mad(GLcontext *ctx, GLfloat scale, GLfloat bias, struct pipe_transfer *acc_pt; GLubyte *map; + st_teximage_flush_before_map(ctx->st, acc_strb->texture, 0, 0, + PIPE_TRANSFER_READ_WRITE); + acc_pt = screen->get_tex_transfer(screen, acc_strb->texture, 0, 0, 0, PIPE_TRANSFER_READ_WRITE, xpos, ypos, width, height); @@ -192,20 +199,27 @@ accum_mad(GLcontext *ctx, GLfloat scale, GLfloat bias, static void -accum_accum(struct pipe_context *pipe, GLfloat value, +accum_accum(struct st_context *st, GLfloat value, GLint xpos, GLint ypos, GLint width, GLint height, struct st_renderbuffer *acc_strb, struct st_renderbuffer *color_strb) { + struct pipe_context *pipe = st->pipe; struct pipe_screen *screen = pipe->screen; struct pipe_transfer *acc_trans, *color_trans; GLfloat *colorBuf, *accBuf; GLint i; + st_teximage_flush_before_map(st, acc_strb->texture, 0, 0, + PIPE_TRANSFER_READ); + acc_trans = screen->get_tex_transfer(screen, acc_strb->texture, 0, 0, 0, PIPE_TRANSFER_READ, xpos, ypos, width, height); + st_teximage_flush_before_map(st, color_strb->texture, 0, 0, + PIPE_TRANSFER_READ); + color_trans = screen->get_tex_transfer(screen, color_strb->texture, 0, 0, 0, PIPE_TRANSFER_READ, xpos, ypos, width, height); @@ -235,20 +249,27 @@ accum_accum(struct pipe_context *pipe, GLfloat value, static void -accum_load(struct pipe_context *pipe, GLfloat value, +accum_load(struct st_context *st, GLfloat value, GLint xpos, GLint ypos, GLint width, GLint height, struct st_renderbuffer *acc_strb, struct st_renderbuffer *color_strb) { + struct pipe_context *pipe = st->pipe; struct pipe_screen *screen = pipe->screen; struct pipe_transfer *acc_trans, *color_trans; GLfloat *buf; GLint i; + st_teximage_flush_before_map(st, acc_strb->texture, 0, 0, + PIPE_TRANSFER_WRITE); + acc_trans = screen->get_tex_transfer(screen, acc_strb->texture, 0, 0, 0, PIPE_TRANSFER_WRITE, xpos, ypos, width, height); + st_teximage_flush_before_map(st, color_strb->texture, 0, 0, + PIPE_TRANSFER_READ); + color_trans = screen->get_tex_transfer(screen, color_strb->texture, 0, 0, 0, PIPE_TRANSFER_READ, xpos, ypos, width, height); @@ -284,10 +305,16 @@ accum_return(GLcontext *ctx, GLfloat value, abuf = (GLfloat *) _mesa_malloc(width * height * 4 * sizeof(GLfloat)); + st_teximage_flush_before_map(ctx->st, acc_strb->texture, 0, 0, + PIPE_TRANSFER_READ); + acc_trans = screen->get_tex_transfer(screen, acc_strb->texture, 0, 0, 0, PIPE_TRANSFER_READ, xpos, ypos, width, height); + st_teximage_flush_before_map(ctx->st, color_strb->texture, 0, 0, + PIPE_TRANSFER_READ_WRITE); + color_trans = screen->get_tex_transfer(screen, color_strb->texture, 0, 0, 0, PIPE_TRANSFER_READ_WRITE, xpos, ypos, width, height); @@ -325,7 +352,6 @@ static void st_Accum(GLcontext *ctx, GLenum op, GLfloat value) { struct st_context *st = ctx->st; - struct pipe_context *pipe = st->pipe; struct st_renderbuffer *acc_strb = st_renderbuffer(ctx->DrawBuffer->Attachment[BUFFER_ACCUM].Renderbuffer); struct st_renderbuffer *color_strb @@ -352,11 +378,11 @@ st_Accum(GLcontext *ctx, GLenum op, GLfloat value) break; case GL_ACCUM: if (value != 0.0F) { - accum_accum(pipe, value, xpos, ypos, width, height, acc_strb, color_strb); + accum_accum(st, value, xpos, ypos, width, height, acc_strb, color_strb); } break; case GL_LOAD: - accum_load(pipe, value, xpos, ypos, width, height, acc_strb, color_strb); + accum_load(st, value, xpos, ypos, width, height, acc_strb, color_strb); break; case GL_RETURN: accum_return(ctx, value, xpos, ypos, width, height, acc_strb, color_strb); diff --git a/src/mesa/state_tracker/st_cb_bufferobjects.c b/src/mesa/state_tracker/st_cb_bufferobjects.c index 3651e4ae7d..fdb800fbd0 100644 --- a/src/mesa/state_tracker/st_cb_bufferobjects.c +++ b/src/mesa/state_tracker/st_cb_bufferobjects.c @@ -32,6 +32,7 @@ #include "st_context.h" #include "st_cb_bufferobjects.h" +#include "st_public.h" #include "pipe/p_context.h" #include "pipe/p_defines.h" @@ -103,6 +104,9 @@ st_bufferobj_subdata(GLcontext *ctx, if (offset >= st_obj->size || size > (st_obj->size - offset)) return; + if (pipe->is_buffer_referenced(pipe, st_obj->buffer)) + st_flush(st_context(ctx), PIPE_FLUSH_RENDER_CACHE, NULL); + pipe_buffer_write(pipe->screen, st_obj->buffer, offset, size, data); } @@ -123,6 +127,10 @@ st_bufferobj_get_subdata(GLcontext *ctx, if (offset >= st_obj->size || size > (st_obj->size - offset)) return; + if (pipe->is_buffer_referenced(pipe, st_obj->buffer) & + PIPE_REFERENCED_FOR_WRITE) + st_flush(st_context(ctx), PIPE_FLUSH_RENDER_CACHE, NULL); + pipe_buffer_read(pipe->screen, st_obj->buffer, offset, size, data); } @@ -171,7 +179,7 @@ st_bufferobj_data(GLcontext *ctx, st_obj->size = size; if (data) - st_bufferobj_subdata(ctx, target, 0, size, data, obj); + pipe_buffer_write(pipe->screen, st_obj->buffer, 0, size, data); } @@ -185,6 +193,7 @@ st_bufferobj_map(GLcontext *ctx, GLenum target, GLenum access, struct pipe_context *pipe = st_context(ctx)->pipe; struct st_buffer_object *st_obj = st_buffer_object(obj); GLuint flags; + unsigned referenced; switch (access) { case GL_WRITE_ONLY: @@ -200,6 +209,11 @@ st_bufferobj_map(GLcontext *ctx, GLenum target, GLenum access, break; } + referenced = pipe->is_buffer_referenced(pipe, st_obj->buffer); + if (referenced && ((referenced & PIPE_REFERENCED_FOR_WRITE) || + (flags & PIPE_BUFFER_USAGE_CPU_WRITE))) + st_flush(st_context(ctx), PIPE_FLUSH_RENDER_CACHE, NULL); + obj->Pointer = pipe_buffer_map(pipe->screen, st_obj->buffer, flags); if(obj->Pointer) { obj->Offset = 0; diff --git a/src/mesa/state_tracker/st_cb_drawpixels.c b/src/mesa/state_tracker/st_cb_drawpixels.c index 0a4430501f..c67b026413 100644 --- a/src/mesa/state_tracker/st_cb_drawpixels.c +++ b/src/mesa/state_tracker/st_cb_drawpixels.c @@ -631,8 +631,6 @@ draw_stencil_pixels(GLcontext *ctx, GLint x, GLint y, GLint skipPixels; ubyte *stmap; - pipe->flush(pipe, PIPE_FLUSH_RENDER_CACHE, NULL); - strb = st_renderbuffer(ctx->DrawBuffer-> Attachment[BUFFER_STENCIL].Renderbuffer); @@ -640,6 +638,9 @@ draw_stencil_pixels(GLcontext *ctx, GLint x, GLint y, y = ctx->DrawBuffer->Height - y - height; } + st_teximage_flush_before_map(ctx->st, strb->texture, 0, 0, + PIPE_TRANSFER_WRITE); + pt = screen->get_tex_transfer(screen, strb->texture, 0, 0, 0, PIPE_TRANSFER_WRITE, x, y, width, height); @@ -825,6 +826,9 @@ copy_stencil_pixels(GLcontext *ctx, GLint srcx, GLint srcy, GL_STENCIL_INDEX, GL_UNSIGNED_BYTE, &ctx->DefaultPacking, buffer); + st_teximage_flush_before_map(ctx->st, rbDraw->texture, 0, 0, + PIPE_TRANSFER_WRITE); + ptDraw = screen->get_tex_transfer(screen, rbDraw->texture, 0, 0, 0, PIPE_TRANSFER_WRITE, dstx, dsty, width, height); diff --git a/src/mesa/state_tracker/st_cb_readpixels.c b/src/mesa/state_tracker/st_cb_readpixels.c index 9ce5f3fe84..519ad6660f 100644 --- a/src/mesa/state_tracker/st_cb_readpixels.c +++ b/src/mesa/state_tracker/st_cb_readpixels.c @@ -42,13 +42,14 @@ #include "pipe/p_defines.h" #include "pipe/p_inlines.h" #include "util/u_tile.h" + #include "st_context.h" #include "st_cb_bitmap.h" #include "st_cb_readpixels.h" #include "st_cb_fbo.h" #include "st_format.h" #include "st_public.h" - +#include "st_texture.h" /** * Special case for reading stencil buffer. @@ -73,6 +74,10 @@ st_read_stencil_pixels(GLcontext *ctx, GLint x, GLint y, } /* Create a read transfer from the renderbuffer's texture */ + + st_teximage_flush_before_map(ctx->st, strb->texture, 0, 0, + PIPE_TRANSFER_READ); + pt = screen->get_tex_transfer(screen, strb->texture, 0, 0, 0, PIPE_TRANSFER_READ, x, y, width, height); @@ -240,6 +245,9 @@ st_fast_readpixels(GLcontext *ctx, struct st_renderbuffer *strb, y = strb->texture->height[0] - y - height; } + st_teximage_flush_before_map(ctx->st, strb->texture, 0, 0, + PIPE_TRANSFER_READ); + trans = screen->get_tex_transfer(screen, strb->texture, 0, 0, 0, PIPE_TRANSFER_READ, x, y, width, height); if (!trans) { diff --git a/src/mesa/state_tracker/st_cb_texture.c b/src/mesa/state_tracker/st_cb_texture.c index 1f14b3705d..98dc6ec74d 100644 --- a/src/mesa/state_tracker/st_cb_texture.c +++ b/src/mesa/state_tracker/st_cb_texture.c @@ -26,6 +26,7 @@ **************************************************************************/ #include "main/mfeatures.h" +#include "main/bufferobj.h" #if FEATURE_convolve #include "main/convolve.h" #endif @@ -55,6 +56,7 @@ #include "pipe/p_inlines.h" #include "util/u_tile.h" #include "util/u_blit.h" +#include "util/u_surface.h" #define DBG if (0) printf @@ -110,6 +112,25 @@ compressed_num_bytes(GLuint mesaFormat) } +static GLboolean +is_compressed_mesa_format(const struct gl_texture_format *format) +{ + switch (format->MesaFormat) { + case MESA_FORMAT_RGB_DXT1: + case MESA_FORMAT_RGBA_DXT1: + case MESA_FORMAT_RGBA_DXT3: + case MESA_FORMAT_RGBA_DXT5: + case MESA_FORMAT_SRGB_DXT1: + case MESA_FORMAT_SRGBA_DXT1: + case MESA_FORMAT_SRGBA_DXT3: + case MESA_FORMAT_SRGBA_DXT5: + return GL_TRUE; + default: + return GL_FALSE; + } +} + + /** called via ctx->Driver.NewTextureImage() */ static struct gl_texture_image * st_NewTextureImage(GLcontext * ctx) @@ -379,6 +400,110 @@ strip_texture_border(GLint border, /** + * Try to do texture compression via rendering. If the Gallium driver + * can render into a compressed surface this will allow us to do texture + * compression. + * \return GL_TRUE for success, GL_FALSE for failure + */ +static GLboolean +compress_with_blit(GLcontext * ctx, + GLenum target, GLint level, + GLint xoffset, GLint yoffset, GLint zoffset, + GLint width, GLint height, GLint depth, + GLenum format, GLenum type, const void *pixels, + const struct gl_pixelstore_attrib *unpack, + struct gl_texture_image *texImage) +{ + const GLuint dstImageOffsets[1] = {0}; + struct st_texture_image *stImage = st_texture_image(texImage); + struct pipe_screen *screen = ctx->st->pipe->screen; + const GLuint face = _mesa_tex_target_to_face(target); + const struct gl_texture_format *mesa_format; + struct pipe_texture templ; + struct pipe_texture *src_tex; + struct pipe_surface *dst_surface; + struct pipe_transfer *tex_xfer; + void *map; + + + if (!stImage->pt) { + /* XXX: Can this happen? Should we assert? */ + return GL_FALSE; + } + + /* get destination surface (in the compressed texture) */ + dst_surface = screen->get_tex_surface(screen, stImage->pt, + stImage->face, stImage->level, 0, + PIPE_BUFFER_USAGE_GPU_WRITE); + if (!dst_surface) { + /* can't render into this format (or other problem) */ + return GL_FALSE; + } + + /* Choose format for the temporary RGBA texture image. + */ + mesa_format = st_ChooseTextureFormat(ctx, GL_RGBA, format, type); + assert(mesa_format); + if (!mesa_format) + return GL_FALSE; + + /* Create the temporary source texture + */ + memset(&templ, 0, sizeof(templ)); + templ.target = PIPE_TEXTURE_2D; + templ.format = st_mesa_format_to_pipe_format(mesa_format->MesaFormat); + pf_get_block(templ.format, &templ.block); + templ.width[0] = width; + templ.height[0] = height; + templ.depth[0] = 1; + templ.last_level = 0; + templ.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER; + src_tex = screen->texture_create(screen, &templ); + + if (!src_tex) + return GL_FALSE; + + /* Put user's tex data into the temporary texture + */ + tex_xfer = screen->get_tex_transfer(screen, src_tex, + face, level, 0, + PIPE_TRANSFER_WRITE, + 0, 0, width, height); /* x, y, w, h */ + map = screen->transfer_map(screen, tex_xfer); + + mesa_format->StoreImage(ctx, 2, GL_RGBA, mesa_format, + map, /* dest ptr */ + 0, 0, 0, /* dest x/y/z offset */ + tex_xfer->stride, /* dest row stride (bytes) */ + dstImageOffsets, /* image offsets (for 3D only) */ + width, height, 1, /* size */ + format, type, /* source format/type */ + pixels, /* source data */ + unpack); /* source data packing */ + + screen->transfer_unmap(screen, tex_xfer); + screen->tex_transfer_destroy(tex_xfer); + + /* copy / compress image */ + util_blit_pixels_tex(ctx->st->blit, + src_tex, /* pipe_texture (src) */ + 0, 0, /* src x0, y0 */ + width, height, /* src x1, y1 */ + dst_surface, /* pipe_surface (dst) */ + xoffset, yoffset, /* dst x0, y0 */ + xoffset + width, /* dst x1 */ + yoffset + height, /* dst y1 */ + 0.0, /* z */ + PIPE_TEX_MIPFILTER_NEAREST); + + pipe_surface_reference(&dst_surface, NULL); + pipe_texture_reference(&src_tex, NULL); + + return GL_TRUE; +} + + +/** * Do glTexImage1/2/3D(). */ static void @@ -392,8 +517,9 @@ st_TexImage(GLcontext * ctx, const struct gl_pixelstore_attrib *unpack, struct gl_texture_object *texObj, struct gl_texture_image *texImage, - GLsizei imageSize, GLboolean compressed) + GLsizei imageSize, GLboolean compressed_src) { + struct pipe_screen *screen = ctx->st->pipe->screen; struct st_texture_object *stObj = st_texture_object(texObj); struct st_texture_image *stImage = st_texture_image(texImage); GLint postConvWidth, postConvHeight; @@ -522,7 +648,7 @@ st_TexImage(GLcontext * ctx, * the expectation that the texture will be set up but nothing * more will be done. This is where those calls return: */ - if (compressed) { + if (compressed_src) { pixels = _mesa_validate_pbo_compressed_teximage(ctx, imageSize, pixels, unpack, "glCompressedTexImage"); @@ -535,6 +661,21 @@ st_TexImage(GLcontext * ctx, if (!pixels) return; + /* See if we can do texture compression with a blit/render. + */ + if (!compressed_src && + !ctx->Mesa_DXTn && + is_compressed_mesa_format(texImage->TexFormat) && + screen->is_format_supported(screen, + stImage->pt->format, + stImage->pt->target, + PIPE_TEXTURE_USAGE_RENDER_TARGET, 0)) { + if (compress_with_blit(ctx, target, level, 0, 0, 0, width, height, depth, + format, type, pixels, unpack, texImage)) { + return; + } + } + if (stImage->pt) { texImage->Data = st_texture_image_map(ctx->st, stImage, 0, PIPE_TRANSFER_WRITE, 0, 0, @@ -570,7 +711,7 @@ st_TexImage(GLcontext * ctx, * the blitter to copy. Or, use the hardware to do the format * conversion and copy: */ - if (compressed) { + if (compressed_src) { memcpy(texImage->Data, pixels, imageSize); } else { @@ -607,7 +748,7 @@ st_TexImage(GLcontext * ctx, _mesa_unmap_teximage_pbo(ctx, unpack); - if (stImage->pt) { + if (stImage->pt && texImage->Data) { st_texture_image_unmap(ctx->st, stImage); texImage->Data = NULL; } @@ -678,6 +819,90 @@ st_CompressedTexImage2D(GLcontext *ctx, GLenum target, GLint level, } + +/** + * glGetTexImage() helper: decompress a compressed texture by rendering + * a textured quad. Store the results in the user's buffer. + */ +static void +decompress_with_blit(GLcontext * ctx, GLenum target, GLint level, + GLenum format, GLenum type, GLvoid *pixels, + struct gl_texture_object *texObj, + struct gl_texture_image *texImage) +{ + struct pipe_screen *screen = ctx->st->pipe->screen; + struct st_texture_image *stImage = st_texture_image(texImage); + const GLuint width = texImage->Width; + const GLuint height = texImage->Height; + struct pipe_surface *dst_surface; + struct pipe_texture *dst_texture; + struct pipe_transfer *tex_xfer; + + /* create temp / dest surface */ + if (!util_create_rgba_surface(screen, width, height, + &dst_texture, &dst_surface)) { + _mesa_problem(ctx, "util_create_rgba_surface() failed " + "in decompress_with_blit()"); + return; + } + + /* blit/render/decompress */ + util_blit_pixels_tex(ctx->st->blit, + stImage->pt, /* pipe_texture (src) */ + 0, 0, /* src x0, y0 */ + width, height, /* src x1, y1 */ + dst_surface, /* pipe_surface (dst) */ + 0, 0, /* dst x0, y0 */ + width, height, /* dst x1, y1 */ + 0.0, /* z */ + PIPE_TEX_MIPFILTER_NEAREST); + + /* map the dst_surface so we can read from it */ + tex_xfer = screen->get_tex_transfer(screen, dst_texture, 0, 0, 0, + PIPE_TRANSFER_READ, + 0, 0, width, height); + + pixels = _mesa_map_readpix_pbo(ctx, &ctx->Pack, pixels); + + /* copy/pack data into user buffer */ + if (st_equal_formats(stImage->pt->format, format, type)) { + /* memcpy */ + const uint bytesPerRow = width * pf_get_size(stImage->pt->format); + ubyte *map = screen->transfer_map(screen, tex_xfer); + GLuint row; + for (row = 0; row < height; row++) { + GLvoid *dest = _mesa_image_address2d(&ctx->Pack, pixels, width, + height, format, type, row, 0); + memcpy(dest, map, bytesPerRow); + map += tex_xfer->stride; + } + screen->transfer_unmap(screen, tex_xfer); + } + else { + /* format translation via floats */ + GLuint row; + for (row = 0; row < height; row++) { + const GLbitfield transferOps = 0x0; /* bypassed for glGetTexImage() */ + GLfloat rgba[4 * MAX_WIDTH]; + GLvoid *dest = _mesa_image_address2d(&ctx->Pack, pixels, width, + height, format, type, row, 0); + + /* get float[4] rgba row from surface */ + pipe_get_tile_rgba(tex_xfer, 0, row, width, 1, rgba); + + _mesa_pack_rgba_span_float(ctx, width, (GLfloat (*)[4]) rgba, format, + type, dest, &ctx->Pack, transferOps); + } + } + + _mesa_unmap_readpix_pbo(ctx, &ctx->Pack); + + /* destroy the temp / dest surface */ + util_destroy_rgba_surface(dst_texture, dst_surface); +} + + + /** * Need to map texture image into memory before copying image data, * then unmap it. @@ -686,7 +911,7 @@ static void st_get_tex_image(GLcontext * ctx, GLenum target, GLint level, GLenum format, GLenum type, GLvoid * pixels, struct gl_texture_object *texObj, - struct gl_texture_image *texImage, GLboolean compressed) + struct gl_texture_image *texImage, GLboolean compressed_dst) { struct st_texture_image *stImage = st_texture_image(texImage); const GLuint dstImageStride = @@ -695,11 +920,27 @@ st_get_tex_image(GLcontext * ctx, GLenum target, GLint level, GLuint depth, i; GLubyte *dest; + if (stImage->pt && + pf_is_compressed(stImage->pt->format) && + !compressed_dst) { + /* Need to decompress the texture. + * We'll do this by rendering a textured quad. + * Note that we only expect RGBA formats (no Z/depth formats). + */ + decompress_with_blit(ctx, target, level, format, type, pixels, + texObj, texImage); + return; + } + /* Map */ if (stImage->pt) { /* Image is stored in hardware format in a buffer managed by the * kernel. Need to explicitly map and unmap it. */ + + st_teximage_flush_before_map(ctx->st, stImage->pt, 0, level, + PIPE_TRANSFER_READ); + texImage->Data = st_texture_image_map(ctx->st, stImage, 0, PIPE_TRANSFER_READ, 0, 0, stImage->base.Width, @@ -724,7 +965,7 @@ st_get_tex_image(GLcontext * ctx, GLenum target, GLint level, dest = (GLubyte *) pixels; for (i = 0; i < depth; i++) { - if (compressed) { + if (compressed_dst) { _mesa_get_compressed_teximage(ctx, target, level, dest, texObj, texImage); } @@ -787,6 +1028,7 @@ st_TexSubimage(GLcontext *ctx, GLint dims, GLenum target, GLint level, struct gl_texture_object *texObj, struct gl_texture_image *texImage) { + struct pipe_screen *screen = ctx->st->pipe->screen; struct st_texture_image *stImage = st_texture_image(texImage); GLuint dstRowStride; const GLuint srcImageStride = @@ -804,10 +1046,28 @@ st_TexSubimage(GLcontext *ctx, GLint dims, GLenum target, GLint level, if (!pixels) return; + /* See if we can do texture compression with a blit/render. + */ + if (!ctx->Mesa_DXTn && + is_compressed_mesa_format(texImage->TexFormat) && + screen->is_format_supported(screen, + stImage->pt->format, + stImage->pt->target, + PIPE_TEXTURE_USAGE_RENDER_TARGET, 0)) { + if (compress_with_blit(ctx, target, level, + xoffset, yoffset, zoffset, + width, height, depth, + format, type, pixels, packing, texImage)) { + return; + } + } + /* Map buffer if necessary. Need to lock to prevent other contexts * from uploading the buffer under us. */ if (stImage->pt) { + st_teximage_flush_before_map(ctx->st, stImage->pt, 0, level, + PIPE_TRANSFER_WRITE); texImage->Data = st_texture_image_map(ctx->st, stImage, zoffset, PIPE_TRANSFER_WRITE, xoffset, yoffset, @@ -932,6 +1192,9 @@ fallback_copy_texsubimage(GLcontext *ctx, GLenum target, GLint level, srcY = strb->Base.Height - srcY - height; } + st_teximage_flush_before_map(ctx->st, strb->texture, 0, 0, + PIPE_TRANSFER_READ); + src_trans = screen->get_tex_transfer( screen, strb->texture, 0, 0, 0, @@ -939,6 +1202,9 @@ fallback_copy_texsubimage(GLcontext *ctx, GLenum target, GLint level, srcX, srcY, width, height); + st_teximage_flush_before_map(ctx->st, stImage->pt, 0, 0, + PIPE_TRANSFER_WRITE); + texDest = st_texture_image_map(ctx->st, stImage, 0, PIPE_TRANSFER_WRITE, destX, destY, width, height); @@ -1318,6 +1584,11 @@ copy_image_data_to_texture(struct st_context *st, /* More straightforward upload. */ + + st_teximage_flush_before_map(st, stObj->pt, stImage->face, dstLevel, + PIPE_TRANSFER_WRITE); + + st_texture_image_data(st->pipe, stObj->pt, stImage->face, diff --git a/src/mesa/state_tracker/st_format.c b/src/mesa/state_tracker/st_format.c index 9e2d60c926..d507e3e58d 100644 --- a/src/mesa/state_tracker/st_format.c +++ b/src/mesa/state_tracker/st_format.c @@ -716,3 +716,23 @@ st_ChooseTextureFormat(GLcontext *ctx, GLint internalFormat, return translate_gallium_format_to_mesa_format(pFormat); } + + +/** + * Test if a gallium format is equivalent to a GL format/type. + */ +GLboolean +st_equal_formats(enum pipe_format pFormat, GLenum format, GLenum type) +{ + switch (pFormat) { + case PIPE_FORMAT_R8G8B8A8_UNORM: + return format == GL_RGBA && type == GL_UNSIGNED_BYTE; + case PIPE_FORMAT_B8G8R8A8_UNORM: + return format == GL_BGRA && type == GL_UNSIGNED_BYTE; + case PIPE_FORMAT_R5G6B5_UNORM: + return format == GL_RGB && type == GL_UNSIGNED_SHORT_5_6_5; + /* XXX more combos... */ + default: + return GL_FALSE; + } +} diff --git a/src/mesa/state_tracker/st_format.h b/src/mesa/state_tracker/st_format.h index 3f5ac3201b..7bbbe2d570 100644 --- a/src/mesa/state_tracker/st_format.h +++ b/src/mesa/state_tracker/st_format.h @@ -76,4 +76,8 @@ st_ChooseTextureFormat(GLcontext * ctx, GLint internalFormat, GLenum format, GLenum type); +extern GLboolean +st_equal_formats(enum pipe_format pFormat, GLenum format, GLenum type); + + #endif /* ST_CB_TEXIMAGE_H */ diff --git a/src/mesa/state_tracker/st_gen_mipmap.c b/src/mesa/state_tracker/st_gen_mipmap.c index 9cc2176d5e..6e9aa5245e 100644 --- a/src/mesa/state_tracker/st_gen_mipmap.c +++ b/src/mesa/state_tracker/st_gen_mipmap.c @@ -123,10 +123,17 @@ fallback_generate_mipmap(GLcontext *ctx, GLenum target, const ubyte *srcData; ubyte *dstData; + st_teximage_flush_before_map(ctx->st, pt, face, srcLevel, + PIPE_TRANSFER_READ); + srcTrans = screen->get_tex_transfer(screen, pt, face, srcLevel, zslice, PIPE_TRANSFER_READ, 0, 0, pt->width[srcLevel], pt->height[srcLevel]); + + st_teximage_flush_before_map(ctx->st, pt, face, dstLevel, + PIPE_TRANSFER_WRITE); + dstTrans = screen->get_tex_transfer(screen, pt, face, dstLevel, zslice, PIPE_TRANSFER_WRITE, 0, 0, pt->width[dstLevel], diff --git a/src/mesa/state_tracker/st_program.c b/src/mesa/state_tracker/st_program.c index 6348e83d8a..2795570cf1 100644 --- a/src/mesa/state_tracker/st_program.c +++ b/src/mesa/state_tracker/st_program.c @@ -484,14 +484,14 @@ st_translate_fragment_program(struct st_context *st, /* handled above */ assert(0); break; - case FRAG_RESULT_COLOR: + default: + assert(attr == FRAG_RESULT_COLOR || + (FRAG_RESULT_DATA0 <= attr && attr < FRAG_RESULT_MAX)); fs_output_semantic_name[fs_num_outputs] = TGSI_SEMANTIC_COLOR; fs_output_semantic_index[fs_num_outputs] = numColors; outputMapping[attr] = fs_num_outputs; numColors++; break; - default: - assert(0); } output_flags[fs_num_outputs] = stfp->Base.Base.OutputFlags[attr]; diff --git a/src/mesa/state_tracker/st_texture.c b/src/mesa/state_tracker/st_texture.c index 19eb7e2f69..fcbaeb6989 100644 --- a/src/mesa/state_tracker/st_texture.c +++ b/src/mesa/state_tracker/st_texture.c @@ -188,8 +188,10 @@ st_texture_image_map(struct st_context *st, struct st_texture_image *stImage, GLuint zoffset, enum pipe_transfer_usage usage, GLuint x, GLuint y, GLuint w, GLuint h) { - struct pipe_screen *screen = st->pipe->screen; + struct pipe_context *pipe = st->pipe; + struct pipe_screen *screen = pipe->screen; struct pipe_texture *pt = stImage->pt; + DBG("%s \n", __FUNCTION__); stImage->transfer = screen->get_tex_transfer(screen, pt, stImage->face, @@ -265,6 +267,7 @@ st_texture_image_data(struct pipe_context *pipe, struct pipe_transfer *dst_transfer; DBG("%s\n", __FUNCTION__); + for (i = 0; i < depth; i++) { dst_transfer = screen->get_tex_transfer(screen, dst, face, level, i, PIPE_TRANSFER_WRITE, 0, 0, @@ -481,3 +484,20 @@ st_release_teximage(struct st_framebuffer *stfb, uint surfIndex, return 1; } + +void +st_teximage_flush_before_map(struct st_context *st, + struct pipe_texture *pt, + unsigned int face, + unsigned int level, + enum pipe_transfer_usage usage) +{ + struct pipe_context *pipe = st->pipe; + unsigned referenced = + pipe->is_texture_referenced(pipe, pt, face, level); + + if (referenced && ((referenced & PIPE_REFERENCED_FOR_WRITE) || + usage == PIPE_TRANSFER_WRITE || + usage == PIPE_TRANSFER_READ_WRITE)) + st_flush(st, PIPE_FLUSH_RENDER_CACHE, NULL); +} diff --git a/src/mesa/state_tracker/st_texture.h b/src/mesa/state_tracker/st_texture.h index 60c000115e..a392e3d57c 100644 --- a/src/mesa/state_tracker/st_texture.h +++ b/src/mesa/state_tracker/st_texture.h @@ -170,5 +170,10 @@ st_texture_image_copy(struct pipe_context *pipe, struct pipe_texture *src, GLuint face); - +extern void +st_teximage_flush_before_map(struct st_context *st, + struct pipe_texture *pt, + unsigned int face, + unsigned int level, + enum pipe_transfer_usage usage); #endif diff --git a/src/mesa/vbo/vbo_save.h b/src/mesa/vbo/vbo_save.h index 9558f83883..86bbd24f7b 100644 --- a/src/mesa/vbo/vbo_save.h +++ b/src/mesa/vbo/vbo_save.h @@ -64,6 +64,13 @@ struct vbo_save_vertex_list { GLubyte attrsz[VBO_ATTRIB_MAX]; GLuint vertex_size; + /* Copy of the final vertex from node->vertex_store->bufferobj. + * Keep this in regular (non-VBO) memory to avoid repeated + * map/unmap of the VBO when updating GL current data. + */ + GLfloat *current_data; + GLuint current_size; + GLuint buffer_offset; GLuint count; GLuint wrap_count; /* number of copied vertices at start */ diff --git a/src/mesa/vbo/vbo_save_api.c b/src/mesa/vbo/vbo_save_api.c index 52b6f1884e..868226075a 100644 --- a/src/mesa/vbo/vbo_save_api.c +++ b/src/mesa/vbo/vbo_save_api.c @@ -289,6 +289,31 @@ static void _save_compile_vertex_list( GLcontext *ctx ) node->vertex_store->refcount++; node->prim_store->refcount++; + + node->current_size = node->vertex_size - node->attrsz[0]; + node->current_data = NULL; + + if (node->current_size) { + /* If the malloc fails, we just pull the data out of the VBO + * later instead. + */ + node->current_data = MALLOC( node->current_size * sizeof(GLfloat) ); + if (node->current_data) { + const char *buffer = (const char *)save->vertex_store->buffer; + unsigned attr_offset = node->attrsz[0] * sizeof(GLfloat); + unsigned vertex_offset = 0; + + if (node->count) + vertex_offset = (node->count-1) * node->vertex_size * sizeof(GLfloat); + + memcpy( node->current_data, + buffer + node->buffer_offset + vertex_offset + attr_offset, + node->current_size * sizeof(GLfloat) ); + } + } + + + assert(node->attrsz[VBO_ATTRIB_POS] != 0 || node->count == 0); diff --git a/src/mesa/vbo/vbo_save_draw.c b/src/mesa/vbo/vbo_save_draw.c index f59e1036d0..5110648c28 100644 --- a/src/mesa/vbo/vbo_save_draw.c +++ b/src/mesa/vbo/vbo_save_draw.c @@ -46,20 +46,31 @@ static void _playback_copy_to_current( GLcontext *ctx, const struct vbo_save_vertex_list *node ) { struct vbo_context *vbo = vbo_context(ctx); - GLfloat vertex[VBO_ATTRIB_MAX * 4], *data = vertex; + GLfloat vertex[VBO_ATTRIB_MAX * 4]; + GLfloat *data; GLuint i, offset; - if (node->count) - offset = (node->buffer_offset + - (node->count-1) * node->vertex_size * sizeof(GLfloat)); - else - offset = node->buffer_offset; + if (node->current_size == 0) + return; - ctx->Driver.GetBufferSubData( ctx, 0, offset, - node->vertex_size * sizeof(GLfloat), - data, node->vertex_store->bufferobj ); + if (node->current_data) { + data = node->current_data; + } + else { + data = vertex; + + if (node->count) + offset = (node->buffer_offset + + (node->count-1) * node->vertex_size * sizeof(GLfloat)); + else + offset = node->buffer_offset; - data += node->attrsz[0]; /* skip vertex position */ + ctx->Driver.GetBufferSubData( ctx, 0, offset, + node->vertex_size * sizeof(GLfloat), + data, node->vertex_store->bufferobj ); + + data += node->attrsz[0]; /* skip vertex position */ + } for (i = VBO_ATTRIB_POS+1 ; i < VBO_ATTRIB_MAX ; i++) { if (node->attrsz[i]) { |