Merge branch 'master' into r300g-glsl

Conflicts:
	src/gallium/drivers/r300/r300_tgsi_to_rc.c

Signed-off-by: Nicolai Hähnle <nhaehnle@gmail.com>

10 files changed, 131 insertions, 62 deletions

diff --git a/src/mesa/drivers/dri/intel/intel_batchbuffer.h b/src/mesa/drivers/dri/intel/intel_batchbuffer.h
index 51579df09e..9a619fbd5c 100644
--- a/src/mesa/drivers/dri/intel/intel_batchbuffer.h
+++ b/src/mesa/drivers/dri/intel/intel_batchbuffer.h
@@ -157,7 +157,7 @@ intel_batchbuffer_require_space(struct intel_batchbuffer *batch,
 #define OUT_BATCH(d) intel_batchbuffer_emit_dword(intel->batch, d)
 
 #define OUT_RELOC(buf, read_domains, write_domain, delta) do {		\
-   assert((delta) >= 0);						\
+   assert((unsigned) (delta) <= buf->size);				\
    intel_batchbuffer_emit_reloc(intel->batch, buf,			\
 				read_domains, write_domain, delta);	\
 } while (0)
diff --git a/src/mesa/drivers/dri/intel/intel_context.c b/src/mesa/drivers/dri/intel/intel_context.c
index d49d95768d..934f7aa187 100644
--- a/src/mesa/drivers/dri/intel/intel_context.c
+++ b/src/mesa/drivers/dri/intel/intel_context.c
@@ -984,41 +984,35 @@ intelMakeCurrent(__DRIcontextPrivate * driContextPriv,
 
       _mesa_make_current(&intel->ctx, &intel_fb->Base, readFb);
 
-      /* The drawbuffer won't always be updated by _mesa_make_current: 
-       */
-      if (intel->ctx.DrawBuffer == &intel_fb->Base) {
-
-	 if (intel->driReadDrawable != driReadPriv)
-	    intel->driReadDrawable = driReadPriv;
-
-	 if (intel->driDrawable != driDrawPriv) {
-	    if (driDrawPriv->swap_interval == (unsigned)-1) {
-	       int i;
-
-	       driDrawPriv->vblFlags = (intel->intelScreen->irq_active != 0)
-		  ? driGetDefaultVBlankFlags(&intel->optionCache)
-		 : VBLANK_FLAG_NO_IRQ;
-
-	       /* Prevent error printf if one crtc is disabled, this will
-		* be properly calculated in intelWindowMoved() next.
-		*/
-		driDrawPriv->vblFlags = intelFixupVblank(intel, driDrawPriv);
-
-	       (*psp->systemTime->getUST) (&intel_fb->swap_ust);
-	       driDrawableInitVBlank(driDrawPriv);
-	       intel_fb->vbl_waited = driDrawPriv->vblSeq;
-
-	       for (i = 0; i < 2; i++) {
-		  if (intel_fb->color_rb[i])
-		     intel_fb->color_rb[i]->vbl_pending = driDrawPriv->vblSeq;
-	       }
-	    }
-	    intel->driDrawable = driDrawPriv;
-	    intelWindowMoved(intel);
-	 }
+      intel->driReadDrawable = driReadPriv;
+
+      if (intel->driDrawable != driDrawPriv) {
+         if (driDrawPriv->swap_interval == (unsigned)-1) {
+            int i;
+
+            driDrawPriv->vblFlags = (intel->intelScreen->irq_active != 0)
+               ? driGetDefaultVBlankFlags(&intel->optionCache)
+               : VBLANK_FLAG_NO_IRQ;
 
-	 intel_draw_buffer(&intel->ctx, &intel_fb->Base);
+            /* Prevent error printf if one crtc is disabled, this will
+             * be properly calculated in intelWindowMoved() next.
+             */
+            driDrawPriv->vblFlags = intelFixupVblank(intel, driDrawPriv);
+
+            (*psp->systemTime->getUST) (&intel_fb->swap_ust);
+            driDrawableInitVBlank(driDrawPriv);
+            intel_fb->vbl_waited = driDrawPriv->vblSeq;
+
+            for (i = 0; i < 2; i++) {
+               if (intel_fb->color_rb[i])
+                  intel_fb->color_rb[i]->vbl_pending = driDrawPriv->vblSeq;
+            }
+         }
+         intel->driDrawable = driDrawPriv;
+         intelWindowMoved(intel);
       }
+
+      intel_draw_buffer(&intel->ctx, &intel_fb->Base);
    }
    else {
       _mesa_make_current(NULL, NULL, NULL);
diff --git a/src/mesa/drivers/dri/intel/intel_pixel_read.c b/src/mesa/drivers/dri/intel/intel_pixel_read.c
index bc67f6242a..4707500180 100644
--- a/src/mesa/drivers/dri/intel/intel_pixel_read.c
+++ b/src/mesa/drivers/dri/intel/intel_pixel_read.c
@@ -216,9 +216,8 @@ do_blit_readpixels(GLcontext * ctx,
       rowLength = -rowLength;
    }
 
-   /* XXX 64-bit cast? */
-   dst_offset = (GLuint) _mesa_image_address(2, pack, pixels, width, height,
-                                             format, type, 0, 0, 0);
+   dst_offset = (GLintptr) _mesa_image_address(2, pack, pixels, width, height,
+					       format, type, 0, 0, 0);
 
 
    /* Although the blits go on the command buffer, need to do this and
@@ -227,14 +226,14 @@ do_blit_readpixels(GLcontext * ctx,
    intelFlush(&intel->ctx);
    LOCK_HARDWARE(intel);
 
-   if (intel->driDrawable->numClipRects) {
+   if (intel->driReadDrawable->numClipRects) {
       GLboolean all = (width * height * src->cpp == dst->Base.Size &&
                        x == 0 && dst_offset == 0);
 
       dri_bo *dst_buffer = intel_bufferobj_buffer(intel, dst,
 						  all ? INTEL_WRITE_FULL :
 						  INTEL_WRITE_PART);
-      __DRIdrawablePrivate *dPriv = intel->driDrawable;
+      __DRIdrawablePrivate *dPriv = intel->driReadDrawable;
       int nbox = dPriv->numClipRects;
       drm_clip_rect_t *box = dPriv->pClipRects;
       drm_clip_rect_t rect;
diff --git a/src/mesa/drivers/dri/intel/intel_tex_copy.c b/src/mesa/drivers/dri/intel/intel_tex_copy.c
index 8f467e6514..f3e312420d 100644
--- a/src/mesa/drivers/dri/intel/intel_tex_copy.c
+++ b/src/mesa/drivers/dri/intel/intel_tex_copy.c
@@ -135,7 +135,7 @@ do_copy_texsubimage(struct intel_context *intel,
 
       if (ctx->ReadBuffer->Name == 0) {
 	 /* reading from a window, adjust x, y */
-	 __DRIdrawablePrivate *dPriv = intel->driDrawable;
+	 const __DRIdrawablePrivate *dPriv = intel->driReadDrawable;
 	 y = dPriv->y + (dPriv->h - (y + height));
 	 x += dPriv->x;
 
diff --git a/src/mesa/drivers/dri/r200/r200_state.c b/src/mesa/drivers/dri/r200/r200_state.c
index 76852e315c..d28e96d9d9 100644
--- a/src/mesa/drivers/dri/r200/r200_state.c
+++ b/src/mesa/drivers/dri/r200/r200_state.c
@@ -1578,13 +1578,6 @@ static void r200ClearStencil( GLcontext *ctx, GLint s )
  * Window position and viewport transformation
  */
 
-/*
- * To correctly position primitives:
- */
-#define SUBPIXEL_X 0.125
-#define SUBPIXEL_Y 0.125
-
-
 /**
  * Called when window size or position changes or viewport or depth range
  * state is changed.  We update the hardware viewport state here.
@@ -1609,9 +1602,9 @@ void r200UpdateWindow( GLcontext *ctx )
    }
 
    float_ui32_type sx = { v[MAT_SX] };
-   float_ui32_type tx = { v[MAT_TX] + xoffset + SUBPIXEL_X };
+   float_ui32_type tx = { v[MAT_TX] + xoffset };
    float_ui32_type sy = { v[MAT_SY] * y_scale };
-   float_ui32_type ty = { (v[MAT_TY] * y_scale) + y_bias + SUBPIXEL_Y };
+   float_ui32_type ty = { (v[MAT_TY] * y_scale) + y_bias };
    float_ui32_type sz = { v[MAT_SZ] * depthScale };
    float_ui32_type tz = { v[MAT_TZ] * depthScale };
 
@@ -1680,8 +1673,8 @@ void r200UpdateViewportOffset( GLcontext *ctx )
    float_ui32_type tx;
    float_ui32_type ty;
 
-   tx.f = v[MAT_TX] + xoffset + SUBPIXEL_X;
-   ty.f = (- v[MAT_TY]) + yoffset + SUBPIXEL_Y;
+   tx.f = v[MAT_TX] + xoffset;
+   ty.f = (- v[MAT_TY]) + yoffset;
 
    if ( rmesa->hw.vpt.cmd[VPT_SE_VPORT_XOFFSET] != tx.ui32 ||
 	rmesa->hw.vpt.cmd[VPT_SE_VPORT_YOFFSET] != ty.ui32 )
diff --git a/src/mesa/drivers/dri/r300/compiler/SConscript b/src/mesa/drivers/dri/r300/compiler/SConscript
new file mode 100755
index 0000000000..48fd65fb71
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/compiler/SConscript
@@ -0,0 +1,30 @@
+Import('*')
+
+env = env.Clone()
+env.Append(CPPPATH = '#/include')
+env.Append(CPPPATH = '#/src/mesa')
+
+# temporary fix
+env['CFLAGS'] = str(env['CFLAGS']).replace('-Werror=declaration-after-statement', '')
+
+r300compiler = env.ConvenienceLibrary(
+    target = 'r300compiler',
+    source = [
+        'radeon_code.c',
+        'radeon_compiler.c',
+        'radeon_nqssadce.c',
+        'radeon_program.c',
+        'radeon_program_alu.c',
+        'radeon_program_pair.c',
+        'r3xx_fragprog.c',
+        'r300_fragprog.c',
+        'r300_fragprog_swizzle.c',
+        'r300_fragprog_emit.c',
+        'r500_fragprog.c',
+        'r500_fragprog_emit.c',
+        'r3xx_vertprog.c',
+        'r3xx_vertprog_dump.c',
+        'memory_pool.c',
+    ])
+
+Return('r300compiler')
diff --git a/src/mesa/drivers/dri/r300/r300_context.c b/src/mesa/drivers/dri/r300/r300_context.c
index 6fcf209af6..2c2b16aa98 100644
--- a/src/mesa/drivers/dri/r300/r300_context.c
+++ b/src/mesa/drivers/dri/r300/r300_context.c
@@ -376,11 +376,21 @@ static void r300InitConstValues(GLcontext *ctx, radeonScreenPtr screen)
 	if (screen->chip_family >= CHIP_FAMILY_RV515) {
 		ctx->Const.FragmentProgram.MaxNativeTemps = R500_PFS_NUM_TEMP_REGS;
 		ctx->Const.FragmentProgram.MaxNativeAttribs = 11;	/* copy i915... */
-		ctx->Const.FragmentProgram.MaxNativeParameters = R500_PFS_NUM_CONST_REGS;
-		ctx->Const.FragmentProgram.MaxNativeAluInstructions = R500_PFS_MAX_INST;
-		ctx->Const.FragmentProgram.MaxNativeTexInstructions = R500_PFS_MAX_INST;
-		ctx->Const.FragmentProgram.MaxNativeInstructions = R500_PFS_MAX_INST;
-		ctx->Const.FragmentProgram.MaxNativeTexIndirections = R500_PFS_MAX_INST;
+
+		/* The hardware limits are higher than this,
+		 * but the non-KMS DRM interface artificially limits us
+		 * to this many instructions.
+		 *
+		 * We could of course work around it in the KMS path,
+		 * but it would be a mess, so it seems wiser
+		 * to leave it as is. Going forward, the Gallium driver
+		 * will not be subject to these limitations.
+		 */
+		ctx->Const.FragmentProgram.MaxNativeParameters = 255;
+		ctx->Const.FragmentProgram.MaxNativeAluInstructions = 255;
+		ctx->Const.FragmentProgram.MaxNativeTexInstructions = 255;
+		ctx->Const.FragmentProgram.MaxNativeInstructions = 255;
+		ctx->Const.FragmentProgram.MaxNativeTexIndirections = 255;
 		ctx->Const.FragmentProgram.MaxNativeAddressRegs = 0;
 	} else {
 		ctx->Const.FragmentProgram.MaxNativeTemps = R300_PFS_NUM_TEMP_REGS;
diff --git a/src/mesa/drivers/dri/r300/r300_fragprog_common.c b/src/mesa/drivers/dri/r300/r300_fragprog_common.c
index 39618b78dc..267ee81a7a 100644
--- a/src/mesa/drivers/dri/r300/r300_fragprog_common.c
+++ b/src/mesa/drivers/dri/r300/r300_fragprog_common.c
@@ -240,6 +240,19 @@ static void translate_fragment_program(GLcontext *ctx, struct r300_fragment_prog
 	rewriteFog(&compiler, fp);
 
 	r3xx_compile_fragment_program(&compiler);
+
+	if (compiler.is_r500) {
+		/* We need to support the non-KMS DRM interface, which
+		 * artificially limits the number of instructions and
+		 * constants which are available to us.
+		 *
+		 * See also the comment in r300_context.c where we
+		 * set the MAX_NATIVE_xxx values.
+		 */
+		if (fp->code.code.r500.inst_end >= 255 || fp->code.constants.Count > 255)
+			rc_error(&compiler.Base, "Program is too big (upgrade to r300g to avoid this limitation).\n");
+	}
+
 	fp->error = compiler.Base.Error;
 
 	fp->InputsRead = compiler.Base.Program.InputsRead;
diff --git a/src/mesa/drivers/dri/r600/r700_render.c b/src/mesa/drivers/dri/r600/r700_render.c
index 0aef0b7ea1..4f39d9f1bd 100644
--- a/src/mesa/drivers/dri/r600/r700_render.c
+++ b/src/mesa/drivers/dri/r600/r700_render.c
@@ -960,7 +960,6 @@ static GLboolean r700TryDrawPrims(GLcontext *ctx,
 
     _tnl_UpdateFixedFunctionProgram(ctx);
     r700SetVertexFormat(ctx, arrays, max_index + 1);
-    r700SetupIndexBuffer(ctx, ib);
     /* shaders need to be updated before buffers are validated */
     r700UpdateShaders2(ctx);
     if (!r600ValidateBuffers(ctx))
@@ -981,6 +980,7 @@ static GLboolean r700TryDrawPrims(GLcontext *ctx,
     GLuint emit_end = r700PredictRenderSize(ctx, nr_prims)
                     + context->radeon.cmdbuf.cs->cdw;
 
+    r700SetupIndexBuffer(ctx, ib);
     r700SetupStreams2(ctx, arrays, max_index + 1);
 
     radeonEmitState(radeon);
diff --git a/src/mesa/drivers/dri/r600/r700_vertprog.c b/src/mesa/drivers/dri/r600/r700_vertprog.c
index e7a209be9d..d12c39c9f7 100644
--- a/src/mesa/drivers/dri/r600/r700_vertprog.c
+++ b/src/mesa/drivers/dri/r600/r700_vertprog.c
@@ -319,8 +319,10 @@ struct r700_vertex_program* r700TranslateVertexShader(GLcontext *ctx,
                 _mesa_insert_mvp_code(ctx, vp->mesa_program);
         }
 
-	for(i=0; i<VERT_ATTRIB_MAX; i++)
+	if( 1 == nVer )
 	{
+	    for(i=0; i<VERT_ATTRIB_MAX; i++)
+	    {
 		unBit = 1 << i;
 		if(vp->mesa_program->Base.InputsRead & unBit) /* ctx->Array.ArrayObj->xxxxxxx */
 		{
@@ -328,7 +330,17 @@ struct r700_vertex_program* r700TranslateVertexShader(GLcontext *ctx,
 			vp->aos_desc[i].stride = vb->AttribPtr[i]->size * sizeof(GL_FLOAT);/* when emit array, data is packed. vb->AttribPtr[i]->stride;*/
 			vp->aos_desc[i].type   = GL_FLOAT;
 		}
+	    }
 	}
+	else
+	{
+		for(i=0; i<context->nNumActiveAos; i++)
+		{
+			vp->aos_desc[i].size   = context->stream_desc[i].size;
+			vp->aos_desc[i].stride = context->stream_desc[i].stride;
+			vp->aos_desc[i].type   = context->stream_desc[i].type;
+		}
+	}	
 	
 	if (context->radeon.radeonScreen->chip_family < CHIP_FAMILY_RV770)
 	{
@@ -388,17 +400,35 @@ void r700SelectVertexShader(GLcontext *ctx, GLint nVersion)
     
     for (vp = vpc->progs; vp; vp = vp->next)
     {
+	if (vp->uiVersion != nVersion ) 
+	    continue;
 	match = GL_TRUE;	
-	for(i=0; i<VERT_ATTRIB_MAX; i++)
+	if ( 1 == nVersion ) 
 	{
+	    for(i=0; i<VERT_ATTRIB_MAX; i++)
+	    {
 		unBit = 1 << i;
 		if(InputsRead & unBit)
 		{
-			if (vp->aos_desc[i].size != vb->AttribPtr[i]->size)
-				match = GL_FALSE;
-				break;
+		    if (vp->aos_desc[i].size != vb->AttribPtr[i]->size)
+		    {
+			match = GL_FALSE;
+			break;
+		    }
 		}
+	    }
 	}
+	else
+	{
+	    for(i=0; i<context->nNumActiveAos; i++)
+	    {
+		if (vp->aos_desc[i].size != context->stream_desc[i].size)
+		{
+		    match = GL_FALSE;
+		    break;
+		}
+	    }
+	}	
 	if (match) 
 	{
 		context->selected_vp = vp;