Merge commit 'origin/gallium-0.1' into gallium-0.2

Conflicts:

	src/gallium/auxiliary/rtasm/rtasm_execmem.c
	src/mesa/shader/slang/slang_emit.c
	src/mesa/shader/slang/slang_log.c
	src/mesa/state_tracker/st_atom_framebuffer.c

21 files changed, 299 insertions, 38 deletions

diff --git a/src/gallium/auxiliary/draw/draw_pt.c b/src/gallium/auxiliary/draw/draw_pt.c
index 87ec6ae20c..b98c0a0ecf 100644
--- a/src/gallium/auxiliary/draw/draw_pt.c
+++ b/src/gallium/auxiliary/draw/draw_pt.c
@@ -33,6 +33,8 @@
 #include "draw/draw_context.h"
 #include "draw/draw_private.h"
 #include "draw/draw_pt.h"
+#include "draw/draw_vs.h"
+#include "tgsi/tgsi_dump.h"
 
 static unsigned trim( unsigned count, unsigned first, unsigned incr )
 {
@@ -195,6 +197,28 @@ draw_arrays(struct draw_context *draw, unsigned prim,
       draw->reduced_prim = reduced_prim;
    }
 
+#if 0
+   {
+      int i;
+      debug_printf("draw_arrays(prim=%u start=%u count=%u):\n",
+                   prim, start, count);
+      tgsi_dump(draw->vs.vertex_shader->state.tokens, 0);
+      debug_printf("Elements:\n");
+      for (i = 0; i < draw->pt.nr_vertex_elements; i++) {
+         debug_printf("  format=%s comps=%u\n",
+                      pf_name(draw->pt.vertex_element[i].src_format),
+                      draw->pt.vertex_element[i].nr_components);
+      }
+      debug_printf("Buffers:\n");
+      for (i = 0; i < draw->pt.nr_vertex_buffers; i++) {
+         debug_printf("  pitch=%u offset=%u ptr=%p\n",
+                      draw->pt.vertex_buffer[i].pitch,
+                      draw->pt.vertex_buffer[i].buffer_offset,
+                      draw->pt.user.vbuffer[i]);
+      }
+   }
+#endif
+
    /* drawing done here: */
    draw_pt_arrays(draw, prim, start, count);
 }
diff --git a/src/gallium/auxiliary/draw/draw_vs_aos.c b/src/gallium/auxiliary/draw/draw_vs_aos.c
index a6880685db..6141ba9cbf 100644
--- a/src/gallium/auxiliary/draw/draw_vs_aos.c
+++ b/src/gallium/auxiliary/draw/draw_vs_aos.c
@@ -2190,7 +2190,8 @@ static struct draw_vs_varient *varient_aos_sse( struct draw_vertex_shader *vs,
    if (!vaos->buffer)
       goto fail;
 
-   debug_printf("nr_vb: %d const: %x\n", vaos->nr_vb, vaos->base.key.const_vbuffers);
+   if (0)
+      debug_printf("nr_vb: %d const: %x\n", vaos->nr_vb, vaos->base.key.const_vbuffers);
 
 #if 0
    tgsi_dump(vs->state.tokens, 0);
diff --git a/src/gallium/auxiliary/draw/draw_vs_exec.c b/src/gallium/auxiliary/draw/draw_vs_exec.c
index 13d4fcfdbf..80c3606657 100644
--- a/src/gallium/auxiliary/draw/draw_vs_exec.c
+++ b/src/gallium/auxiliary/draw/draw_vs_exec.c
@@ -123,6 +123,12 @@ vs_exec_run_linear( struct draw_vertex_shader *shader,
 	 input = (const float (*)[4])((const char *)input + input_stride);
       } 
 
+      tgsi_set_exec_mask(machine,
+                         1,
+                         max_vertices > 1,
+                         max_vertices > 2,
+                         max_vertices > 3);
+
       /* run interpreter */
       tgsi_exec_machine_run( machine );
 
diff --git a/src/gallium/auxiliary/draw/draw_vs_sse.c b/src/gallium/auxiliary/draw/draw_vs_sse.c
index b11ae31662..0e2036f12a 100644
--- a/src/gallium/auxiliary/draw/draw_vs_sse.c
+++ b/src/gallium/auxiliary/draw/draw_vs_sse.c
@@ -99,9 +99,23 @@ vs_sse_run_linear( struct draw_vertex_shader *base,
    struct tgsi_exec_machine *machine = shader->machine;
    unsigned int i;
 
+   /* By default, execute all channels.  XXX move this inside the loop
+    * below when we support shader conditionals/loops.
+    */
+   tgsi_set_exec_mask(machine, 1, 1, 1, 1);
+
    for (i = 0; i < count; i += MAX_TGSI_VERTICES) {
       unsigned int max_vertices = MIN2(MAX_TGSI_VERTICES, count - i);
 
+      if (max_vertices < 4) {
+         /* disable the unused execution channels */
+         tgsi_set_exec_mask(machine,
+                            1,
+                            max_vertices > 1,
+                            max_vertices > 2,
+                            0);
+      }
+
       /* run compiled shader
        */
       shader->func(machine->Inputs,
diff --git a/src/gallium/auxiliary/tgsi/tgsi_exec.c b/src/gallium/auxiliary/tgsi/tgsi_exec.c
index 1a5294eabc..e32779123f 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_exec.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_exec.c
@@ -958,6 +958,10 @@ fetch_src_file_channel(
       switch( file ) {
       case TGSI_FILE_CONSTANT:
          assert(mach->Consts);
+         assert(index->i[0] >= 0);
+         assert(index->i[1] >= 0);
+         assert(index->i[2] >= 0);
+         assert(index->i[3] >= 0);
          chan->f[0] = mach->Consts[index->i[0]][swizzle];
          chan->f[1] = mach->Consts[index->i[1]][swizzle];
          chan->f[2] = mach->Consts[index->i[2]][swizzle];
diff --git a/src/gallium/auxiliary/tgsi/tgsi_exec.h b/src/gallium/auxiliary/tgsi/tgsi_exec.h
index c4e649e69c..fc40a25e09 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_exec.h
+++ b/src/gallium/auxiliary/tgsi/tgsi_exec.h
@@ -165,6 +165,10 @@ struct tgsi_exec_labels
 #define TGSI_EXEC_TEMP_HALF_I       (TGSI_EXEC_NUM_TEMPS + 3)
 #define TGSI_EXEC_TEMP_HALF_C       1
 
+/* execution mask, each value is either 0 or ~0 */
+#define TGSI_EXEC_MASK_I            (TGSI_EXEC_NUM_TEMPS + 3)
+#define TGSI_EXEC_MASK_C            2
+
 #define TGSI_EXEC_TEMP_R0           (TGSI_EXEC_NUM_TEMPS + 4)
 
 #define TGSI_EXEC_TEMP_ADDR         (TGSI_EXEC_NUM_TEMPS + 5)
@@ -265,6 +269,27 @@ void
 tgsi_exec_machine_free_data(struct tgsi_exec_machine *mach);
 
 
+static INLINE void
+tgsi_set_kill_mask(struct tgsi_exec_machine *mach, unsigned mask)
+{
+   mach->Temps[TGSI_EXEC_TEMP_KILMASK_I].xyzw[TGSI_EXEC_TEMP_KILMASK_C].u[0] =
+      mask;
+}
+
+
+/** Set execution mask values prior to executing the shader */
+static INLINE void
+tgsi_set_exec_mask(struct tgsi_exec_machine *mach,
+                   boolean ch0, boolean ch1, boolean ch2, boolean ch3)
+{
+   int *mask = mach->Temps[TGSI_EXEC_MASK_I].xyzw[TGSI_EXEC_MASK_C].i;
+   mask[0] = ch0 ? ~0 : 0;
+   mask[1] = ch1 ? ~0 : 0;
+   mask[2] = ch2 ? ~0 : 0;
+   mask[3] = ch3 ? ~0 : 0;
+}
+
+
 #if defined __cplusplus
 } /* extern "C" */
 #endif
diff --git a/src/gallium/auxiliary/tgsi/tgsi_sse2.c b/src/gallium/auxiliary/tgsi/tgsi_sse2.c
index 47e52c8424..3df0c5db3f 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_sse2.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_sse2.c
@@ -72,6 +72,9 @@
 
 #define TEMP_R0   TGSI_EXEC_TEMP_R0
 #define TEMP_ADDR TGSI_EXEC_TEMP_ADDR
+#define TEMP_EXEC_MASK_I TGSI_EXEC_MASK_I
+#define TEMP_EXEC_MASK_C TGSI_EXEC_MASK_C
+
 
 /**
  * X86 utility functions.
@@ -233,6 +236,9 @@ emit_const(
    int indirectIndex )
 {
    if (indirect) {
+      /* 'vec' is the offset from the address register's value.
+       * We're loading CONST[ADDR+vec] into an xmm register.
+       */
       struct x86_reg r0 = get_input_base();
       struct x86_reg r1 = get_output_base();
       uint i;
@@ -243,18 +249,40 @@ emit_const(
       x86_push( func, r0 );
       x86_push( func, r1 );
 
+      /*
+       * Loop over the four pixels or vertices in the quad.
+       * Get the value of the address (offset) register for pixel/vertex[i],
+       * add it to the src offset and index into the constant buffer.
+       * Note that we're working on SOA data.
+       * If any of the pixel/vertex execution channels are unused their
+       * values will be garbage.  It's very important that we don't use
+       * those garbage values as indexes into the constant buffer since
+       * that'll cause segfaults.
+       * The solution is to bitwise-AND the offset with the execution mask
+       * register whose values are either 0 or ~0.
+       * The caller must setup the execution mask register to indicate
+       * which channels are valid/alive before running the shader.
+       * The execution mask will also figure into loops and conditionals
+       * someday.
+       */
       for (i = 0; i < QUAD_SIZE; i++) {
-         x86_lea( func, r0, get_const( vec, chan ) );
+         /* r1 = address register[i] */
          x86_mov( func, r1, x86_make_disp( get_temp( TEMP_ADDR, CHAN_X ), i * 4 ) );
+         /* r0 = execution mask[i] */
+         x86_mov( func, r0, x86_make_disp( get_temp( TEMP_EXEC_MASK_I, TEMP_EXEC_MASK_C ), i * 4 ) );
+         /* r1 = r1 & r0 */
+         x86_and( func, r1, r0 );
+         /* r0 = 'vec', the offset */
+         x86_lea( func, r0, get_const( vec, chan ) );
 
-         /* Quick hack to multiply by 16 -- need to add SHL to rtasm.
+         /* Quick hack to multiply r1 by 16 -- need to add SHL to rtasm.
           */
          x86_add( func, r1, r1 );
          x86_add( func, r1, r1 );
          x86_add( func, r1, r1 );
          x86_add( func, r1, r1 );
 
-         x86_add( func, r0, r1 );
+         x86_add( func, r0, r1 );  /* r0 = r0 + r1 */
          x86_mov( func, r1, x86_deref( r0 ) );
          x86_mov( func, x86_make_disp( get_temp( TEMP_R0, CHAN_X ), i * 4 ), r1 );
       }
@@ -268,6 +296,7 @@ emit_const(
          get_temp( TEMP_R0, CHAN_X ) );
    }
    else {
+      /* 'vec' is the index into the src register file, such as TEMP[vec] */
       assert( vec >= 0 );
 
       sse_movss(
diff --git a/src/gallium/auxiliary/util/u_mm.c b/src/gallium/auxiliary/util/u_mm.c
index 592ace00fc..45ce257b5e 100644
--- a/src/gallium/auxiliary/util/u_mm.c
+++ b/src/gallium/auxiliary/util/u_mm.c
@@ -172,6 +172,10 @@ u_mmAllocMem(struct mem_block *heap, int size, int align2, int startSearch)
    int startofs = 0;
    int endofs;
 
+   assert(size >= 0);
+   assert(align2 >= 0);
+   assert(align2 <= 12); /* sanity check, 2^12 (4KB) enough? */
+
    if (!heap || align2 < 0 || size <= 0)
       return NULL;
 
diff --git a/src/gallium/drivers/softpipe/sp_fs_sse.c b/src/gallium/drivers/softpipe/sp_fs_sse.c
index 0111469405..8aa597f633 100644
--- a/src/gallium/drivers/softpipe/sp_fs_sse.c
+++ b/src/gallium/drivers/softpipe/sp_fs_sse.c
@@ -92,7 +92,8 @@ fs_sse_run( const struct sp_fragment_shader *base,
 		       machine->Temps);
 
    /* init kill mask */
-   machine->Temps[TGSI_EXEC_TEMP_KILMASK_I].xyzw[TGSI_EXEC_TEMP_KILMASK_C].u[0] = 0x0;
+   tgsi_set_kill_mask(machine, 0x0);
+   tgsi_set_exec_mask(machine, 1, 1, 1, 1);
 
    shader->func( machine->Inputs,
 		 machine->Outputs,
diff --git a/src/gallium/include/pipe/p_debug.h b/src/gallium/include/pipe/p_debug.h
index cb6196aa9f..3b00fb9aa8 100644
--- a/src/gallium/include/pipe/p_debug.h
+++ b/src/gallium/include/pipe/p_debug.h
@@ -49,7 +49,7 @@ extern "C" {
 #endif
 
 
-#ifdef DBG
+#if defined(DBG) || defined(DEBUG)
 #ifndef DEBUG
 #define DEBUG 1
 #endif
diff --git a/src/mesa/main/texenvprogram.c b/src/mesa/main/texenvprogram.c
index 2c7df7b8d9..237ddaa683 100644
--- a/src/mesa/main/texenvprogram.c
+++ b/src/mesa/main/texenvprogram.c
@@ -233,6 +233,12 @@ static GLbitfield get_fp_input_mask( GLcontext *ctx )
       /* Fixed function logic */
       GLbitfield varying_inputs = ctx->varying_vp_inputs;
 
+      /* These get generated in the setup routine regardless of the
+       * vertex program:
+       */
+      if (ctx->Point.PointSprite)
+         varying_inputs |= FRAG_BITS_TEX_ANY;
+
       /* First look at what values may be computed by the generated
        * vertex program:
        */
@@ -260,6 +266,12 @@ static GLbitfield get_fp_input_mask( GLcontext *ctx )
       /* calculate from vp->outputs */
       GLbitfield vp_outputs = ctx->VertexProgram._Current->Base.OutputsWritten;
 
+      /* These get generated in the setup routine regardless of the
+       * vertex program:
+       */
+      if (ctx->Point.PointSprite)
+         vp_outputs |= FRAG_BITS_TEX_ANY;
+
       if (vp_outputs & (1 << VERT_RESULT_COL0)) fp_inputs |= FRAG_BIT_COL0;
       if (vp_outputs & (1 << VERT_RESULT_COL1)) fp_inputs |= FRAG_BIT_COL1;
 
diff --git a/src/mesa/shader/prog_uniform.c b/src/mesa/shader/prog_uniform.c
index 25d7c0997b..11f2e3e561 100644
--- a/src/mesa/shader/prog_uniform.c
+++ b/src/mesa/shader/prog_uniform.c
@@ -87,6 +87,7 @@ _mesa_append_uniform(struct gl_uniform_list *list,
       list->Uniforms[oldNum].Name = _mesa_strdup(name);
       list->Uniforms[oldNum].VertPos = -1;
       list->Uniforms[oldNum].FragPos = -1;
+      list->Uniforms[oldNum].Initialized = GL_FALSE;
       index = oldNum;
       list->NumUniforms++;
    }
diff --git a/src/mesa/shader/prog_uniform.h b/src/mesa/shader/prog_uniform.h
index 735de28705..deea732991 100644
--- a/src/mesa/shader/prog_uniform.h
+++ b/src/mesa/shader/prog_uniform.h
@@ -50,6 +50,7 @@ struct gl_uniform
    const char *Name;        /**< Null-terminated string */
    GLint VertPos;
    GLint FragPos;
+   GLboolean Initialized;   /**< For debug.  Has this uniform been set? */
 #if 0
    GLenum DataType;         /**< GL_FLOAT, GL_FLOAT_VEC2, etc */
    GLuint Size;             /**< Number of components (1..4) */
diff --git a/src/mesa/shader/shader_api.c b/src/mesa/shader/shader_api.c
index 266ecc4ef2..7f020fe5d3 100644
--- a/src/mesa/shader/shader_api.c
+++ b/src/mesa/shader/shader_api.c
@@ -1512,10 +1512,12 @@ set_program_uniform(GLcontext *ctx, struct gl_program *program,
                     GLenum type, GLsizei count, GLint elems,
                     const void *values)
 {
+   struct gl_program_parameter *param =
+      &program->Parameters->Parameters[index];
+
    assert(offset >= 0);
 
-   if (!compatible_types(type,
-                         program->Parameters->Parameters[index].DataType)) {
+   if (!compatible_types(type, param->DataType)) {
       _mesa_error(ctx, GL_INVALID_OPERATION, "glUniform(type mismatch)");
       return;
    }
@@ -1525,7 +1527,7 @@ set_program_uniform(GLcontext *ctx, struct gl_program *program,
       return;
    }
 
-   if (program->Parameters->Parameters[index].Type == PROGRAM_SAMPLER) {
+   if (param->Type == PROGRAM_SAMPLER) {
       /* This controls which texture unit which is used by a sampler */
       GLuint texUnit, sampler;
 
@@ -1556,9 +1558,9 @@ set_program_uniform(GLcontext *ctx, struct gl_program *program,
    else {
       /* ordinary uniform variable */
       GLsizei k, i;
-      GLint slots = (program->Parameters->Parameters[index].Size + 3) / 4;
+      GLint slots = (param->Size + 3) / 4;
 
-      if (count * elems > (GLint) program->Parameters->Parameters[index].Size) {
+      if (count * elems > (GLint) param->Size) {
          _mesa_error(ctx, GL_INVALID_OPERATION, "glUniform(count too large)");
          return;
       }
@@ -1567,7 +1569,8 @@ set_program_uniform(GLcontext *ctx, struct gl_program *program,
          count = slots;
 
       for (k = 0; k < count; k++) {
-         GLfloat *uniformVal = program->Parameters->ParameterValues[index + offset + k];
+         GLfloat *uniformVal =
+            program->Parameters->ParameterValues[index + offset + k];
          if (is_integer_type(type)) {
             const GLint *iValues = ((const GLint *) values) + k * elems;
             for (i = 0; i < elems; i++) {
@@ -1582,7 +1585,7 @@ set_program_uniform(GLcontext *ctx, struct gl_program *program,
          }
 
          /* if the uniform is bool-valued, convert to 1.0 or 0.0 */
-         if (is_boolean_type(program->Parameters->Parameters[index].DataType)) {
+         if (is_boolean_type(param->DataType)) {
             for (i = 0; i < elems; i++) {
                uniformVal[i] = uniformVal[i] ? 1.0f : 0.0f;
             }
@@ -1666,6 +1669,8 @@ _mesa_uniform(GLcontext *ctx, GLint location, GLsizei count,
                              index, offset, type, count, elems, values);
       }
    }
+
+   shProg->Uniforms->Uniforms[location].Initialized = GL_TRUE;
 }
 
 
@@ -1776,6 +1781,8 @@ _mesa_uniform_matrix(GLcontext *ctx, GLint cols, GLint rows,
                                     count, rows, cols, transpose, values);
       }
    }
+
+   shProg->Uniforms->Uniforms[location].Initialized = GL_TRUE;
 }
 
 
diff --git a/src/mesa/shader/slang/slang_codegen.c b/src/mesa/shader/slang/slang_codegen.c
index e7b2bad8c2..d83e3b01e6 100644
--- a/src/mesa/shader/slang/slang_codegen.c
+++ b/src/mesa/shader/slang/slang_codegen.c
@@ -3253,7 +3253,7 @@ _slang_gen_array_element(slang_assemble_ctx * A, slang_operation *oper)
       index = _slang_gen_operation(A, &oper->children[1]);
       if (array && index) {
          /* bounds check */
-         GLint constIndex = 0;
+         GLint constIndex = -1;
          if (index->Opcode == IR_FLOAT) {
             constIndex = (int) index->Value[0];
             if (constIndex < 0 || constIndex >= arrayLen) {
diff --git a/src/mesa/shader/slang/slang_emit.c b/src/mesa/shader/slang/slang_emit.c
index 6ab06864b9..672ec4bd60 100644
--- a/src/mesa/shader/slang/slang_emit.c
+++ b/src/mesa/shader/slang/slang_emit.c
@@ -330,6 +330,17 @@ constant_to_src_reg(struct prog_src_register *src, GLfloat val,
 }
 
 
+static void
+address_to_dst_reg(struct prog_dst_register *dst, GLuint index)
+{
+   assert(index == 0); /* only one address reg at this time */
+   dst->File = PROGRAM_ADDRESS;
+   dst->Index = index;
+   dst->WriteMask = WRITEMASK_X;
+}
+
+
+
 /**
  * Add new instruction at end of given program.
  * \param prog  the program to append instruction onto
@@ -614,6 +625,7 @@ emit_arith(slang_emit_info *emitInfo, slang_ir_node *n)
 
    /* result storage */
    alloc_node_storage(emitInfo, n, -1);
+
    assert(n->Store->Index >= 0);
    if (n->Store->Size == 2)
       n->Writemask = WRITEMASK_XY;
@@ -1546,6 +1558,60 @@ emit_swizzle(slang_emit_info *emitInfo, slang_ir_node *n)
 
 
 /**
+ * Move a block registers from src to dst (or move a single register).
+ * \param size  size of block, in floats (<=4 means one register)
+ */
+static struct prog_instruction *
+move_block(slang_emit_info *emitInfo,
+           GLuint size, GLboolean relAddr,
+           const slang_ir_storage *dst,
+           const slang_ir_storage *src)
+{
+   struct prog_instruction *inst;
+
+   if (size > 4) {
+      /* move matrix/struct etc (block of registers) */
+      slang_ir_storage dstStore = *dst;
+      slang_ir_storage srcStore = *src;
+      //GLint size = srcStore.Size;
+      /*ASSERT(n->Children[0]->Writemask == WRITEMASK_XYZW);
+      ASSERT(n->Children[1]->Store->Swizzle == SWIZZLE_NOOP);
+      */
+      dstStore.Size = 4;
+      srcStore.Size = 4;
+      while (size >= 4) {
+         inst = new_instruction(emitInfo, OPCODE_MOV);
+         inst->Comment = _mesa_strdup("IR_COPY block");
+         storage_to_dst_reg(&inst->DstReg, &dstStore, WRITEMASK_XYZW);
+         storage_to_src_reg(&inst->SrcReg[0], &srcStore);
+         inst->SrcReg[0].RelAddr = relAddr;
+         srcStore.Index++;
+         dstStore.Index++;
+         size -= 4;
+      }
+   }
+   else {
+      /* single register move */
+      GLuint writemask;
+      if (size == 1) {
+         GLuint comp = GET_SWZ(src->Swizzle, 0);
+         assert(comp < 4);
+         writemask = WRITEMASK_X << comp;
+      }
+      else {
+         writemask = WRITEMASK_XYZW;
+      }
+      inst = new_instruction(emitInfo, OPCODE_MOV);
+      storage_to_dst_reg(&inst->DstReg, dst, writemask);
+      storage_to_src_reg(&inst->SrcReg[0], src);
+      inst->SrcReg[0].RelAddr = relAddr;
+   }
+   return inst;
+}
+
+
+
+/**
  * Dereference array element.  Just resolve storage for the array
  * element represented by this node.
  */
@@ -1587,24 +1653,47 @@ emit_array_element(slang_emit_info *emitInfo, slang_ir_node *n)
    else {
       /* Variable array index */
       struct prog_instruction *inst;
-      slang_ir_storage dstStore = *n->Store;
 
       /* do codegen for array index expression */
       emit(emitInfo, n->Children[1]);
 
-      inst = new_instruction(emitInfo, OPCODE_ARL);
+      /* allocate temp storage for the array element */
+      assert(n->Store->Index < 0);
+      n->Store->File = PROGRAM_TEMPORARY;
+      n->Store->Parent = NULL;
+      alloc_node_storage(emitInfo, n, -1);
+
+      if (n->Store->Size > 4) {
+         /* need to multiply the index by the element size */
+         GLint elemSize = (n->Store->Size + 3) / 4;
+         slang_ir_storage indexTemp;
 
-      if (dstStore.Size > 4)
-         dstStore.Size = 4; /* only emit one instruction */
+         /* allocate 1 float indexTemp */
+         alloc_local_temp(emitInfo, &indexTemp, 1);
+
+         /* MUL temp, index, elemSize */
+         inst = new_instruction(emitInfo, OPCODE_MUL);
+         storage_to_dst_reg(&inst->DstReg, &indexTemp, WRITEMASK_X);
+         storage_to_src_reg(&inst->SrcReg[0], n->Children[1]->Store);
+         constant_to_src_reg(&inst->SrcReg[1], elemSize, emitInfo);
 
-      storage_to_dst_reg(&inst->DstReg, &dstStore, n->Writemask);
-      storage_to_src_reg(&inst->SrcReg[0], n->Children[1]->Store);
+         /* load ADDR[0].X = temp */
+         inst = new_instruction(emitInfo, OPCODE_ARL);
+         storage_to_src_reg(&inst->SrcReg[0], &indexTemp);
+         address_to_dst_reg(&inst->DstReg, 0);
 
-      inst->DstReg.File = PROGRAM_ADDRESS;
-      inst->DstReg.Index = 0; /* always address register [0] */
-      inst->Comment = _mesa_strdup("ARL ADDR");
+         _slang_free_temp(emitInfo->vt, &indexTemp);
+      }
+      else {
+         /* simply load address reg w/ array index */
+         inst = new_instruction(emitInfo, OPCODE_ARL);
+         storage_to_src_reg(&inst->SrcReg[0], n->Children[1]->Store);
+         address_to_dst_reg(&inst->DstReg, 0);
+      }
 
-      n->Store->RelAddr = GL_TRUE;
+      /* copy from array element to temp storage */
+      move_block(emitInfo, n->Store->Size, GL_TRUE,
+                 n->Store, n->Children[0]->Store);
    }
 
    /* if array element size is one, make sure we only access X */
diff --git a/src/mesa/shader/slang/slang_log.c b/src/mesa/shader/slang/slang_log.c
index dc838c72ad..25f696f67e 100644
--- a/src/mesa/shader/slang/slang_log.c
+++ b/src/mesa/shader/slang/slang_log.c
@@ -89,7 +89,7 @@ slang_info_log_message(slang_info_log * log, const char *prefix,
    slang_string_concat(log->text, "\n");
 
    if (MESA_VERBOSE & VERBOSE_GLSL) {
-      _mesa_printf("Mesa: GLSL %s\n", log->text);
+      _mesa_printf("Mesa: GLSL %s", log->text);
    }
 
    return 1;
diff --git a/src/mesa/state_tracker/st_atom_framebuffer.c b/src/mesa/state_tracker/st_atom_framebuffer.c
index b4f42c6f93..ca1a719a9a 100644
--- a/src/mesa/state_tracker/st_atom_framebuffer.c
+++ b/src/mesa/state_tracker/st_atom_framebuffer.c
@@ -116,10 +116,11 @@ update_framebuffer_state( struct st_context *st )
          /* rendering to a GL texture, may have to update surface */
          update_renderbuffer_surface(st, strb);
       }
-      
-      assert(strb->surface);
-      framebuffer->cbufs[framebuffer->num_cbufs] = strb->surface;
-      framebuffer->num_cbufs++;
+
+      if (strb->surface) {
+         framebuffer->cbufs[framebuffer->num_cbufs] = strb->surface;
+         framebuffer->num_cbufs++;
+      }
    }
 
    strb = st_renderbuffer(fb->Attachment[BUFFER_DEPTH].Renderbuffer);
@@ -130,7 +131,6 @@ update_framebuffer_state( struct st_context *st )
          update_renderbuffer_surface(st, strb);
       }
 
-      assert(strb->surface);
       framebuffer->zsbuf = strb->surface;
    }
    else {
diff --git a/src/mesa/state_tracker/st_cb_clear.c b/src/mesa/state_tracker/st_cb_clear.c
index ee282e8e20..1412a3761f 100644
--- a/src/mesa/state_tracker/st_cb_clear.c
+++ b/src/mesa/state_tracker/st_cb_clear.c
@@ -406,13 +406,17 @@ check_clear_stencil_with_quad(GLcontext *ctx, struct gl_renderbuffer *rb)
 static void
 clear_color_buffer(GLcontext *ctx, struct gl_renderbuffer *rb)
 {
+   struct st_renderbuffer *strb = st_renderbuffer(rb);
+
+   if (!strb->surface)
+      return;
+
    if (check_clear_color_with_quad( ctx, rb )) {
       /* masking or scissoring */
       clear_with_quad(ctx, GL_TRUE, GL_FALSE, GL_FALSE);
    }
    else {
       /* clear whole buffer w/out masking */
-      struct st_renderbuffer *strb = st_renderbuffer(rb);
       uint clearValue;
       /* NOTE: we always pass the clear color as PIPE_FORMAT_A8R8G8B8_UNORM
        * at this time!
@@ -426,13 +430,16 @@ clear_color_buffer(GLcontext *ctx, struct gl_renderbuffer *rb)
 static void
 clear_depth_buffer(GLcontext *ctx, struct gl_renderbuffer *rb)
 {
+   struct st_renderbuffer *strb = st_renderbuffer(rb);
+
+   if (!strb->surface)
+      return;
+
    if (check_clear_depth_with_quad(ctx, rb)) {
       /* scissoring or we have a combined depth/stencil buffer */
       clear_with_quad(ctx, GL_FALSE, GL_TRUE, GL_FALSE);
    }
    else {
-      struct st_renderbuffer *strb = st_renderbuffer(rb);
-
       /* simple clear of whole buffer */
       uint clearValue = util_pack_z(strb->surface->format, ctx->Depth.Clear);
       ctx->st->pipe->clear(ctx->st->pipe, strb->surface, clearValue);
@@ -443,13 +450,16 @@ clear_depth_buffer(GLcontext *ctx, struct gl_renderbuffer *rb)
 static void
 clear_stencil_buffer(GLcontext *ctx, struct gl_renderbuffer *rb)
 {
+   struct st_renderbuffer *strb = st_renderbuffer(rb);
+
+   if (!strb->surface)
+      return;
+
    if (check_clear_stencil_with_quad(ctx, rb)) {
       /* masking or scissoring or combined depth/stencil buffer */
       clear_with_quad(ctx, GL_FALSE, GL_FALSE, GL_TRUE);
    }
    else {
-      struct st_renderbuffer *strb = st_renderbuffer(rb);
-
       /* simple clear of whole buffer */
       GLuint clearValue = ctx->Stencil.Clear;
 
@@ -469,14 +479,16 @@ clear_stencil_buffer(GLcontext *ctx, struct gl_renderbuffer *rb)
 static void
 clear_depth_stencil_buffer(GLcontext *ctx, struct gl_renderbuffer *rb)
 {
+   struct st_renderbuffer *strb = st_renderbuffer(rb);
+
+   if (!strb->surface)
+      return;
 
    if (check_clear_depth_stencil_with_quad(ctx, rb)) {
       /* masking or scissoring */
       clear_with_quad(ctx, GL_FALSE, GL_TRUE, GL_TRUE);
    }
    else {
-      struct st_renderbuffer *strb = st_renderbuffer(rb);
-
       /* clear whole buffer w/out masking */
       GLuint clearValue = util_pack_z(strb->surface->format, ctx->Depth.Clear);
 
diff --git a/src/mesa/state_tracker/st_draw.c b/src/mesa/state_tracker/st_draw.c
index 61949a9388..7cf06da43c 100644
--- a/src/mesa/state_tracker/st_draw.c
+++ b/src/mesa/state_tracker/st_draw.c
@@ -33,6 +33,7 @@
 #include "main/imports.h"
 #include "main/image.h"
 #include "main/macros.h"
+#include "shader/prog_uniform.h"
 
 #include "vbo/vbo.h"
 
@@ -483,6 +484,28 @@ setup_non_interleaved_attribs(GLcontext *ctx,
 
 
 
+/**
+ * Prior to drawing, check that any uniforms referenced by the
+ * current shader have been set.  If a uniform has not been set,
+ * issue a warning.
+ */
+static void
+check_uniforms(GLcontext *ctx)
+{
+   const struct gl_shader_program *shProg = ctx->Shader.CurrentProgram;
+   if (shProg && shProg->LinkStatus) {
+      GLuint i;
+      for (i = 0; i < shProg->Uniforms->NumUniforms; i++) {
+         const struct gl_uniform *u = &shProg->Uniforms->Uniforms[i];
+         if (!u->Initialized) {
+            _mesa_warning(ctx,
+                          "Using shader with uninitialized uniform: %s",
+                          u->Name);
+         }
+      }
+   }
+}
+
 
 /**
  * This function gets plugged into the VBO module and is called when
@@ -516,6 +539,10 @@ st_draw_vbo(GLcontext *ctx,
    vp = ctx->st->vp;
    vs = &ctx->st->vp->state;
 
+   if (MESA_VERBOSE & VERBOSE_GLSL) {
+      check_uniforms(ctx);
+   }
+
    /*
     * Setup the vbuffer[] and velements[] arrays.
     */
@@ -557,6 +584,9 @@ st_draw_vbo(GLcontext *ctx,
    pipe->set_vertex_buffers(pipe, num_vbuffers, vbuffer);
    pipe->set_vertex_elements(pipe, num_velements, velements);
 
+   if (num_vbuffers == 0 || num_velements == 0)
+      return;
+
    /* do actual drawing */
    if (ib) {
       /* indexed primitive */
diff --git a/src/mesa/state_tracker/st_framebuffer.c b/src/mesa/state_tracker/st_framebuffer.c
index fe9900d92c..6fd731d209 100644
--- a/src/mesa/state_tracker/st_framebuffer.c
+++ b/src/mesa/state_tracker/st_framebuffer.c
@@ -291,7 +291,8 @@ st_notify_swapbuffers_complete(struct st_framebuffer *stfb)
       for (i = 0; i < BUFFER_COUNT; i++) {
 	 if (stfb->Base.Attachment[i].Renderbuffer) {
 	    strb = st_renderbuffer(stfb->Base.Attachment[i].Renderbuffer);
-	    strb->surface->status = PIPE_SURFACE_STATUS_UNDEFINED;
+            if (strb->surface)
+               strb->surface->status = PIPE_SURFACE_STATUS_UNDEFINED;
 	 }
       }
    }