Major rework of tnl module

New array_cache module
Support 8 texture units in core mesa (now support 8 everywhere)
Rework core mesa statechange operations to avoid flushing on many
noop statechanges.

1 files changed, 311 insertions, 0 deletions

diff --git a/src/mesa/tnl/t_vb_vertex.c b/src/mesa/tnl/t_vb_vertex.c
new file mode 100644
index 0000000000..7667d423a8
--- /dev/null
+++ b/src/mesa/tnl/t_vb_vertex.c
@@ -0,0 +1,311 @@
+/* $Id: t_vb_vertex.c,v 1.1 2000/12/26 05:09:33 keithw Exp $ */
+
+/*
+ * Mesa 3-D graphics library
+ * Version:  3.5
+ *
+ * Copyright (C) 1999-2000  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.
+ *
+ * Author:
+ *    Keith Whitwell <keithw@valinux.com>
+ */
+
+
+#include "glheader.h"
+#include "colormac.h"
+#include "context.h"
+#include "macros.h"
+#include "mem.h"
+#include "mmath.h"
+#include "mtypes.h"
+
+#include "math/m_xform.h"
+
+#include "t_context.h"
+#include "t_pipeline.h"
+
+
+
+struct vertex_stage_data {
+   GLvector4f eye;
+   GLvector4f clip;
+   GLvector4f proj;	
+   GLubyte *clipmask;
+   GLubyte ormask;
+   GLubyte andmask;
+
+
+   /* Need these because it's difficult to replay the sideeffects
+    * analytically.
+    */
+   GLvector4f *save_eyeptr;
+   GLvector4f *save_clipptr;
+   GLvector4f *save_projptr;
+};
+
+#define VERTEX_STAGE_DATA(stage) ((struct vertex_stage_data *)stage->private)
+
+
+
+
+/* This function implements cliptesting for user-defined clip planes.
+ * The clipping of primitives to these planes is implemented in
+ * t_render_clip.h.  
+ */
+#define USER_CLIPTEST(NAME, SZ)					\
+static void NAME( GLcontext *ctx,				\
+		  GLvector4f *clip,				\
+		  GLubyte *clipmask,				\
+		  GLubyte *clipormask,				\
+		  GLubyte *clipandmask )			\
+{								\
+   GLuint p;							\
+								\
+   for (p = 0; p < ctx->Const.MaxClipPlanes; p++)		\
+      if (ctx->Transform.ClipEnabled[p]) {			\
+	 GLuint nr, i;						\
+	 const GLfloat a = ctx->Transform._ClipUserPlane[p][0];	\
+	 const GLfloat b = ctx->Transform._ClipUserPlane[p][1];	\
+	 const GLfloat c = ctx->Transform._ClipUserPlane[p][2];	\
+	 const GLfloat d = ctx->Transform._ClipUserPlane[p][3];	\
+         GLfloat *coord = (GLfloat *)clip->data;		\
+         GLuint stride = clip->stride;				\
+         GLuint count = clip->count;				\
+								\
+	 for (nr = 0, i = 0 ; i < count ; i++) {		\
+	    GLfloat dp = coord[0] * a + coord[1] * b;		\
+	    if (SZ > 2) dp += coord[2] * c;			\
+	    if (SZ > 3) dp += coord[3] * d; else dp += d;	\
+								\
+	    if (dp < 0) {					\
+	       nr++;						\
+	       clipmask[i] |= CLIP_USER_BIT;			\
+	    }							\
+								\
+	    STRIDE_F(coord, stride);				\
+	 }							\
+								\
+	 if (nr > 0) {						\
+	    *clipormask |= CLIP_USER_BIT;			\
+	    if (nr == count) {					\
+	       *clipandmask |= CLIP_USER_BIT;			\
+	       return;						\
+	    }							\
+	 }							\
+      }								\
+}
+
+
+USER_CLIPTEST(userclip2, 2)		
+USER_CLIPTEST(userclip3, 3)		
+USER_CLIPTEST(userclip4, 4)		
+
+static void (*(usercliptab[5]))( GLcontext *,
+				 GLvector4f *, GLubyte *,
+				 GLubyte *, GLubyte * ) = 
+{
+   0,
+   0,
+   userclip2,
+   userclip3,
+   userclip4
+};
+
+
+
+static GLboolean run_vertex_stage( GLcontext *ctx, 
+				   struct gl_pipeline_stage *stage )
+{
+   struct vertex_stage_data *store = (struct vertex_stage_data *)stage->private;
+   struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb; 
+   
+   if (stage->changed_inputs) 
+   {
+/*        VB->ObjPtr->size = 4; */
+
+      if (ctx->_NeedEyeCoords) {
+	 /* Seperate modelview and project transformations:
+	  */
+	 if (ctx->ModelView.type == MATRIX_IDENTITY)
+	    VB->EyePtr = VB->ObjPtr;
+	 else
+	    VB->EyePtr = TransformRaw( &store->eye, &ctx->ModelView, 
+				       VB->ObjPtr);
+
+	 if (ctx->ProjectionMatrix.type == MATRIX_IDENTITY)
+	    VB->ClipPtr = VB->EyePtr;
+	 else
+	    VB->ClipPtr = TransformRaw( &store->clip, &ctx->ProjectionMatrix, 
+					VB->EyePtr );
+      } 
+      else 
+      {
+	 /* Combined modelviewproject transform:
+	  */
+	 if (ctx->_ModelProjectMatrix.type == MATRIX_IDENTITY)
+	    VB->ClipPtr = VB->ObjPtr;
+	 else
+	    VB->ClipPtr = TransformRaw( &store->clip, &ctx->_ModelProjectMatrix,
+					VB->ObjPtr );
+      }
+
+      /* Cliptest and perspective divide.  Clip functions must clear
+       * the clipmask.  
+       */
+      store->ormask = 0;
+      store->andmask = CLIP_ALL_BITS;
+
+      VB->ProjectedClipPtr = 
+	 gl_clip_tab[VB->ClipPtr->size]( VB->ClipPtr,
+					 &store->proj,
+					 store->clipmask,
+					 &store->ormask,
+					 &store->andmask );
+
+      if (store->andmask) 
+	 return GL_FALSE;
+
+
+      /* Test userclip planes.  This contributes to VB->ClipMask, so
+       * is essentially required to be in this stage.
+       */
+      if (ctx->Transform._AnyClip) {
+	 usercliptab[VB->ClipPtr->size]( ctx,
+					 VB->ClipPtr,
+					 store->clipmask,
+					 &store->ormask,
+					 &store->andmask );
+
+	 if (store->andmask)
+	    return GL_FALSE;
+      }
+      
+      VB->ClipOrMask = store->ormask;
+      VB->ClipMask = store->clipmask;
+
+      if (VB->ClipPtr == VB->ObjPtr && (VB->importable_data & VERT_OBJ))
+	 VB->importable_data |= VERT_CLIP;
+
+      /* Drivers expect this to be size 4...
+       */
+      if (VB->ProjectedClipPtr->size < 4) {
+	 ASSERT(VB->ProjectedClipPtr == VB->ClipPtr);
+	 if (VB->ProjectedClipPtr->flags & VEC_NOT_WRITEABLE) {
+	    ASSERT(VB->ProjectedClipPtr == VB->ObjPtr);
+	    VB->import_data( ctx, VERT_OBJ, VEC_NOT_WRITEABLE );
+	    VB->ProjectedClipPtr = VB->ClipPtr = VB->ObjPtr;
+	 }
+	 if (VB->ClipPtr->size == 2) 
+	    gl_vector4f_clean_elem( VB->ClipPtr, VB->Count, 2 );
+	 gl_vector4f_clean_elem( VB->ClipPtr, VB->Count, 3 );
+	 VB->ClipPtr->size = 4;
+      }
+
+      store->save_eyeptr = VB->EyePtr;
+      store->save_clipptr = VB->ClipPtr;
+      store->save_projptr = VB->ProjectedClipPtr;
+   } 
+   else {
+      /* Replay the sideeffects.  
+       */
+      VB->EyePtr = store->save_eyeptr;
+      VB->ClipPtr = store->save_clipptr;
+      VB->ProjectedClipPtr = store->save_projptr;
+      VB->ClipMask = store->clipmask;
+      VB->ClipOrMask = store->ormask;
+      if (VB->ClipPtr == VB->ObjPtr && (VB->importable_data & VERT_OBJ))
+	 VB->importable_data |= VERT_CLIP;
+      if (store->andmask)
+	 return GL_FALSE;
+   }
+
+   return GL_TRUE;
+}
+
+
+static void check_vertex( GLcontext *ctx, struct gl_pipeline_stage *stage )
+{
+   (void) ctx; 
+   (void) stage;
+}
+
+static GLboolean init_vertex_stage( GLcontext *ctx, 
+				    struct gl_pipeline_stage *stage )
+{
+   struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb; 
+   struct vertex_stage_data *store;
+   GLuint size = VB->Size;
+
+   stage->private = CALLOC(sizeof(*store));
+   store = VERTEX_STAGE_DATA(stage);
+   if (!store)
+      return GL_FALSE;
+   
+   gl_vector4f_alloc( &store->eye, 0, size, 32 );
+   gl_vector4f_alloc( &store->clip, 0, size, 32 );
+   gl_vector4f_alloc( &store->proj, 0, size, 32 );  
+
+   store->clipmask = (GLubyte *) ALIGN_MALLOC(sizeof(GLubyte)*size, 32 );
+
+   if (!store->clipmask ||
+       !store->eye.data || 
+       !store->clip.data || 
+       !store->proj.data) 
+      return GL_FALSE;
+
+   /* Now run the stage.
+    */
+   stage->run = run_vertex_stage;
+   return stage->run( ctx, stage );
+}
+
+static void dtr( struct gl_pipeline_stage *stage )
+{
+   struct vertex_stage_data *store = VERTEX_STAGE_DATA(stage);
+
+   if (store) {
+      gl_vector4f_free( &store->eye );
+      gl_vector4f_free( &store->clip );
+      gl_vector4f_free( &store->proj );
+      ALIGN_FREE( store->clipmask );
+      FREE(store);
+      stage->private = 0;
+      stage->run = init_vertex_stage;
+   }
+}
+
+
+const struct gl_pipeline_stage _tnl_vertex_transform_stage = 
+{ 
+   "modelview/project/cliptest/divide",
+   0,				/* re-check -- always on */
+   _NEW_MODELVIEW|
+   _NEW_PROJECTION|
+   _NEW_TRANSFORM,		/* re-run */
+   GL_TRUE,			/* active */
+   VERT_OBJ,		/* inputs */
+   VERT_EYE|VERT_CLIP,		/* outputs */
+   0, 0,			/* changed_inputs, private */
+   dtr,				/* destructor */
+   check_vertex,		/* check */
+   init_vertex_stage		/* run -- initially set to init */
+};
+
+