X-Git-Url: https://git.libre-soc.org/?a=blobdiff_plain;ds=sidebyside;f=src%2Fmesa%2Fstate_tracker%2Fst_glsl_to_tgsi.cpp;h=f3535c5810d7849a386d88bed26a3344784ef5bc;hb=04b7e65814cd2174185109d3c55c86eb4134f09b;hp=643a9bbb09d1af0b2f43df39973cf639ec9f9774;hpb=ecfb404e8d4fcd35524d1c4b3421e24980fe3976;p=mesa.git

diff --git a/src/mesa/state_tracker/st_glsl_to_tgsi.cpp b/src/mesa/state_tracker/st_glsl_to_tgsi.cpp
index 643a9bbb09d..f3535c5810d 100644
--- a/src/mesa/state_tracker/st_glsl_to_tgsi.cpp
+++ b/src/mesa/state_tracker/st_glsl_to_tgsi.cpp
@@ -34,7 +34,6 @@
 #include "main/compiler.h"
 #include "ir.h"
 #include "ir_visitor.h"
-#include "ir_print_visitor.h"
 #include "ir_expression_flattening.h"
 #include "glsl_types.h"
 #include "glsl_parser_extras.h"
@@ -44,11 +43,11 @@
 
 #include "main/mtypes.h"
 #include "main/shaderobj.h"
+#include "main/uniforms.h"
 #include "program/hash_table.h"
 
 extern "C" {
 #include "main/shaderapi.h"
-#include "main/uniforms.h"
 #include "program/prog_instruction.h"
 #include "program/prog_optimize.h"
 #include "program/prog_print.h"
@@ -71,9 +70,7 @@ extern "C" {
 }
 
 #define PROGRAM_IMMEDIATE PROGRAM_FILE_MAX
-#define PROGRAM_ANY_CONST ((1 << PROGRAM_LOCAL_PARAM) |  \
-                           (1 << PROGRAM_ENV_PARAM) |    \
-                           (1 << PROGRAM_STATE_VAR) |    \
+#define PROGRAM_ANY_CONST ((1 << PROGRAM_STATE_VAR) |    \
                            (1 << PROGRAM_CONSTANT) |     \
                            (1 << PROGRAM_UNIFORM))
 
@@ -85,8 +82,12 @@ extern "C" {
  */
 #define MAX_TEMPS         4096
 
-/* will be 4 for GLSL 4.00 */
-#define MAX_GLSL_TEXTURE_OFFSET 1
+/**
+ * Maximum number of arrays
+ */
+#define MAX_ARRAYS        256
+
+#define MAX_GLSL_TEXTURE_OFFSET 4
 
 class st_src_reg;
 class st_dst_reg;
@@ -110,6 +111,8 @@ public:
       this->index2D = 0;
       this->type = type ? type->base_type : GLSL_TYPE_ERROR;
       this->reladdr = NULL;
+      this->reladdr2 = NULL;
+      this->has_index2 = false;
    }
 
    st_src_reg(gl_register_file file, int index, int type)
@@ -121,6 +124,8 @@ public:
       this->swizzle = SWIZZLE_XYZW;
       this->negate = 0;
       this->reladdr = NULL;
+      this->reladdr2 = NULL;
+      this->has_index2 = false;
    }
 
    st_src_reg(gl_register_file file, int index, int type, int index2D)
@@ -132,6 +137,8 @@ public:
       this->swizzle = SWIZZLE_XYZW;
       this->negate = 0;
       this->reladdr = NULL;
+      this->reladdr2 = NULL;
+      this->has_index2 = false;
    }
 
    st_src_reg()
@@ -143,22 +150,36 @@ public:
       this->swizzle = 0;
       this->negate = 0;
       this->reladdr = NULL;
+      this->reladdr2 = NULL;
+      this->has_index2 = false;
    }
 
    explicit st_src_reg(st_dst_reg reg);
 
    gl_register_file file; /**< PROGRAM_* from Mesa */
-   int index; /**< temporary index, VERT_ATTRIB_*, FRAG_ATTRIB_*, etc. */
+   int index; /**< temporary index, VERT_ATTRIB_*, VARYING_SLOT_*, etc. */
    int index2D;
    GLuint swizzle; /**< SWIZZLE_XYZWONEZERO swizzles from Mesa. */
    int negate; /**< NEGATE_XYZW mask from mesa */
    int type; /** GLSL_TYPE_* from GLSL IR (enum glsl_base_type) */
    /** Register index should be offset by the integer in this reg. */
    st_src_reg *reladdr;
+   st_src_reg *reladdr2;
+   bool has_index2;
 };
 
 class st_dst_reg {
 public:
+   st_dst_reg(gl_register_file file, int writemask, int type, int index)
+   {
+      this->file = file;
+      this->index = index;
+      this->writemask = writemask;
+      this->cond_mask = COND_TR;
+      this->reladdr = NULL;
+      this->type = type;
+   }
+
    st_dst_reg(gl_register_file file, int writemask, int type)
    {
       this->file = file;
@@ -182,7 +203,7 @@ public:
    explicit st_dst_reg(st_src_reg reg);
 
    gl_register_file file; /**< PROGRAM_* from Mesa */
-   int index; /**< temporary index, VERT_ATTRIB_*, FRAG_ATTRIB_*, etc. */
+   int index; /**< temporary index, VERT_ATTRIB_*, VARYING_SLOT_*, etc. */
    int writemask; /**< Bitfield of WRITEMASK_[XYZW] */
    GLuint cond_mask:4;
    int type; /** GLSL_TYPE_* from GLSL IR (enum glsl_base_type) */
@@ -199,6 +220,8 @@ st_src_reg::st_src_reg(st_dst_reg reg)
    this->negate = 0;
    this->reladdr = reg.reladdr;
    this->index2D = 0;
+   this->reladdr2 = NULL;
+   this->has_index2 = false;
 }
 
 st_dst_reg::st_dst_reg(st_src_reg reg)
@@ -213,21 +236,11 @@ st_dst_reg::st_dst_reg(st_src_reg reg)
 
 class glsl_to_tgsi_instruction : public exec_node {
 public:
-   /* Callers of this ralloc-based new need not call delete. It's
-    * easier to just ralloc_free 'ctx' (or any of its ancestors). */
-   static void* operator new(size_t size, void *ctx)
-   {
-      void *node;
-
-      node = rzalloc_size(ctx, size);
-      assert(node != NULL);
-
-      return node;
-   }
+   DECLARE_RALLOC_CXX_OPERATORS(glsl_to_tgsi_instruction)
 
    unsigned op;
    st_dst_reg dst;
-   st_src_reg src[3];
+   st_src_reg src[4];
    /** Pointer to the ir source this tree came from for debugging */
    ir_instruction *ir;
    GLboolean cond_update;
@@ -235,7 +248,8 @@ public:
    int sampler; /**< sampler index */
    int tex_target; /**< One of TEXTURE_*_INDEX */
    GLboolean tex_shadow;
-   struct tgsi_texture_offset tex_offsets[MAX_GLSL_TEXTURE_OFFSET];
+
+   st_src_reg tex_offsets[MAX_GLSL_TEXTURE_OFFSET];
    unsigned tex_offset_num_offset;
    int dead_mask; /**< Used in dead code elimination */
 
@@ -315,13 +329,16 @@ public:
 
    int next_temp;
 
+   unsigned array_sizes[MAX_ARRAYS];
+   unsigned next_array;
+
    int num_address_regs;
    int samplers_used;
-   bool indirect_addr_temps;
    bool indirect_addr_consts;
    
    int glsl_version;
    bool native_integers;
+   bool have_sqrt;
 
    variable_storage *find_variable_storage(ir_variable *var);
 
@@ -362,6 +379,8 @@ public:
    virtual void visit(ir_discard *);
    virtual void visit(ir_texture *);
    virtual void visit(ir_if *);
+   virtual void visit(ir_emit_vertex *);
+   virtual void visit(ir_end_primitive *);
    /*@}*/
 
    st_src_reg result;
@@ -391,7 +410,12 @@ public:
    glsl_to_tgsi_instruction *emit(ir_instruction *ir, unsigned op,
         		        st_dst_reg dst,
         		        st_src_reg src0, st_src_reg src1, st_src_reg src2);
-   
+
+   glsl_to_tgsi_instruction *emit(ir_instruction *ir, unsigned op,
+                                  st_dst_reg dst,
+                                  st_src_reg src0, st_src_reg src1,
+                                  st_src_reg src2, st_src_reg src3);
+
    unsigned get_opcode(ir_instruction *ir, unsigned op,
                     st_dst_reg dst,
                     st_src_reg src0, st_src_reg src1);
@@ -411,8 +435,6 @@ public:
    void emit_scalar(ir_instruction *ir, unsigned op,
         	    st_dst_reg dst, st_src_reg src0, st_src_reg src1);
 
-   void try_emit_float_set(ir_instruction *ir, unsigned op, st_dst_reg dst);
-
    void emit_arl(ir_instruction *ir, st_dst_reg dst, st_src_reg src0);
 
    void emit_scs(ir_instruction *ir, unsigned op,
@@ -437,8 +459,7 @@ public:
    int get_last_temp_write(int index);
 
    void copy_propagate(void);
-   void eliminate_dead_code(void);
-   int eliminate_dead_code_advanced(void);
+   int eliminate_dead_code(void);
    void merge_registers(void);
    void renumber_registers(void);
 
@@ -452,7 +473,8 @@ static st_src_reg undef_src = st_src_reg(PROGRAM_UNDEFINED, 0, GLSL_TYPE_ERROR);
 
 static st_dst_reg undef_dst = st_dst_reg(PROGRAM_UNDEFINED, SWIZZLE_NOOP, GLSL_TYPE_ERROR);
 
-static st_dst_reg address_reg = st_dst_reg(PROGRAM_ADDRESS, WRITEMASK_X, GLSL_TYPE_FLOAT);
+static st_dst_reg address_reg = st_dst_reg(PROGRAM_ADDRESS, WRITEMASK_X, GLSL_TYPE_FLOAT, 0);
+static st_dst_reg address_reg2 = st_dst_reg(PROGRAM_ADDRESS, WRITEMASK_X, GLSL_TYPE_FLOAT, 1);
 
 static void
 fail_link(struct gl_shader_program *prog, const char *fmt, ...) PRINTFLIKE(2, 3);
@@ -505,8 +527,9 @@ num_inst_src_regs(unsigned opcode)
 
 glsl_to_tgsi_instruction *
 glsl_to_tgsi_visitor::emit(ir_instruction *ir, unsigned op,
-        		 st_dst_reg dst,
-        		 st_src_reg src0, st_src_reg src1, st_src_reg src2)
+                           st_dst_reg dst,
+                           st_src_reg src0, st_src_reg src1,
+                           st_src_reg src2, st_src_reg src3)
 {
    glsl_to_tgsi_instruction *inst = new(mem_ctx) glsl_to_tgsi_instruction();
    int num_reladdr = 0, i;
@@ -518,10 +541,12 @@ glsl_to_tgsi_visitor::emit(ir_instruction *ir, unsigned op,
     * sources into temps.
     */
    num_reladdr += dst.reladdr != NULL;
-   num_reladdr += src0.reladdr != NULL;
-   num_reladdr += src1.reladdr != NULL;
-   num_reladdr += src2.reladdr != NULL;
+   num_reladdr += src0.reladdr != NULL || src0.reladdr2 != NULL;
+   num_reladdr += src1.reladdr != NULL || src1.reladdr2 != NULL;
+   num_reladdr += src2.reladdr != NULL || src2.reladdr2 != NULL;
+   num_reladdr += src3.reladdr != NULL || src3.reladdr2 != NULL;
 
+   reladdr_to_temp(ir, &src3, &num_reladdr);
    reladdr_to_temp(ir, &src2, &num_reladdr);
    reladdr_to_temp(ir, &src1, &num_reladdr);
    reladdr_to_temp(ir, &src0, &num_reladdr);
@@ -537,22 +562,15 @@ glsl_to_tgsi_visitor::emit(ir_instruction *ir, unsigned op,
    inst->src[0] = src0;
    inst->src[1] = src1;
    inst->src[2] = src2;
+   inst->src[3] = src3;
    inst->ir = ir;
    inst->dead_mask = 0;
 
    inst->function = NULL;
    
-   if (op == TGSI_OPCODE_ARL || op == TGSI_OPCODE_UARL)
-      this->num_address_regs = 1;
-   
    /* Update indirect addressing status used by TGSI */
    if (dst.reladdr) {
       switch(dst.file) {
-      case PROGRAM_TEMPORARY:
-         this->indirect_addr_temps = true;
-         break;
-      case PROGRAM_LOCAL_PARAM:
-      case PROGRAM_ENV_PARAM:
       case PROGRAM_STATE_VAR:
       case PROGRAM_CONSTANT:
       case PROGRAM_UNIFORM:
@@ -566,14 +584,9 @@ glsl_to_tgsi_visitor::emit(ir_instruction *ir, unsigned op,
       }
    }
    else {
-      for (i=0; i<3; i++) {
+      for (i=0; i<4; i++) {
          if(inst->src[i].reladdr) {
             switch(inst->src[i].file) {
-            case PROGRAM_TEMPORARY:
-               this->indirect_addr_temps = true;
-               break;
-            case PROGRAM_LOCAL_PARAM:
-            case PROGRAM_ENV_PARAM:
             case PROGRAM_STATE_VAR:
             case PROGRAM_CONSTANT:
             case PROGRAM_UNIFORM:
@@ -591,18 +604,22 @@ glsl_to_tgsi_visitor::emit(ir_instruction *ir, unsigned op,
 
    this->instructions.push_tail(inst);
 
-   if (native_integers)
-      try_emit_float_set(ir, op, dst);
-
    return inst;
 }
 
+glsl_to_tgsi_instruction *
+glsl_to_tgsi_visitor::emit(ir_instruction *ir, unsigned op,
+                           st_dst_reg dst, st_src_reg src0,
+                           st_src_reg src1, st_src_reg src2)
+{
+   return emit(ir, op, dst, src0, src1, src2, undef_src);
+}
 
 glsl_to_tgsi_instruction *
 glsl_to_tgsi_visitor::emit(ir_instruction *ir, unsigned op,
         		 st_dst_reg dst, st_src_reg src0, st_src_reg src1)
 {
-   return emit(ir, op, dst, src0, src1, undef_src);
+   return emit(ir, op, dst, src0, src1, undef_src, undef_src);
 }
 
 glsl_to_tgsi_instruction *
@@ -610,32 +627,13 @@ glsl_to_tgsi_visitor::emit(ir_instruction *ir, unsigned op,
         		 st_dst_reg dst, st_src_reg src0)
 {
    assert(dst.writemask != 0);
-   return emit(ir, op, dst, src0, undef_src, undef_src);
+   return emit(ir, op, dst, src0, undef_src, undef_src, undef_src);
 }
 
 glsl_to_tgsi_instruction *
 glsl_to_tgsi_visitor::emit(ir_instruction *ir, unsigned op)
 {
-   return emit(ir, op, undef_dst, undef_src, undef_src, undef_src);
-}
-
- /**
- * Emits the code to convert the result of float SET instructions to integers.
- */
-void
-glsl_to_tgsi_visitor::try_emit_float_set(ir_instruction *ir, unsigned op,
-        		 st_dst_reg dst)
-{
-   if ((op == TGSI_OPCODE_SEQ ||
-        op == TGSI_OPCODE_SNE ||
-        op == TGSI_OPCODE_SGE ||
-        op == TGSI_OPCODE_SLT))
-   {
-      st_src_reg src = st_src_reg(dst);
-      src.negate = ~src.negate;
-      dst.type = GLSL_TYPE_FLOAT;
-      emit(ir, TGSI_OPCODE_F2I, dst, src);
-   }
+   return emit(ir, op, undef_dst, undef_src, undef_src, undef_src, undef_src);
 }
 
 /**
@@ -648,7 +646,10 @@ glsl_to_tgsi_visitor::get_opcode(ir_instruction *ir, unsigned op,
         		 st_src_reg src0, st_src_reg src1)
 {
    int type = GLSL_TYPE_FLOAT;
-   
+
+   if (op == TGSI_OPCODE_MOV)
+       return op;
+
    assert(src0.type != GLSL_TYPE_ARRAY);
    assert(src0.type != GLSL_TYPE_STRUCT);
    assert(src1.type != GLSL_TYPE_ARRAY);
@@ -661,14 +662,30 @@ glsl_to_tgsi_visitor::get_opcode(ir_instruction *ir, unsigned op,
 
 #define case4(c, f, i, u) \
    case TGSI_OPCODE_##c: \
-      if (type == GLSL_TYPE_INT) op = TGSI_OPCODE_##i; \
-      else if (type == GLSL_TYPE_UINT) op = TGSI_OPCODE_##u; \
-      else op = TGSI_OPCODE_##f; \
+      if (type == GLSL_TYPE_INT) \
+         op = TGSI_OPCODE_##i; \
+      else if (type == GLSL_TYPE_UINT) \
+         op = TGSI_OPCODE_##u; \
+      else \
+         op = TGSI_OPCODE_##f; \
       break;
+
 #define case3(f, i, u)  case4(f, f, i, u)
 #define case2fi(f, i)   case4(f, f, i, i)
 #define case2iu(i, u)   case4(i, LAST, i, u)
-   
+
+#define casecomp(c, f, i, u) \
+   case TGSI_OPCODE_##c: \
+      if (type == GLSL_TYPE_INT) \
+         op = TGSI_OPCODE_##i; \
+      else if (type == GLSL_TYPE_UINT) \
+         op = TGSI_OPCODE_##u; \
+      else if (native_integers) \
+         op = TGSI_OPCODE_##f; \
+      else \
+         op = TGSI_OPCODE_##c; \
+      break;
+
    switch(op) {
       case2fi(ADD, UADD);
       case2fi(MUL, UMUL);
@@ -677,17 +694,20 @@ glsl_to_tgsi_visitor::get_opcode(ir_instruction *ir, unsigned op,
       case3(MAX, IMAX, UMAX);
       case3(MIN, IMIN, UMIN);
       case2iu(MOD, UMOD);
-      
-      case2fi(SEQ, USEQ);
-      case2fi(SNE, USNE);
-      case3(SGE, ISGE, USGE);
-      case3(SLT, ISLT, USLT);
-      
+
+      casecomp(SEQ, FSEQ, USEQ, USEQ);
+      casecomp(SNE, FSNE, USNE, USNE);
+      casecomp(SGE, FSGE, ISGE, USGE);
+      casecomp(SLT, FSLT, ISLT, USLT);
+
       case2iu(ISHR, USHR);
 
       case2fi(SSG, ISSG);
       case3(ABS, IABS, IABS);
-      
+
+      case2iu(IBFE, UBFE);
+      case2iu(IMSB, UMSB);
+      case2iu(IMUL_HI, UMUL_HI);
       default: break;
    }
    
@@ -780,6 +800,10 @@ glsl_to_tgsi_visitor::emit_arl(ir_instruction *ir,
    if (src0.type == GLSL_TYPE_INT || src0.type == GLSL_TYPE_UINT)
       op = TGSI_OPCODE_UARL;
 
+   assert(dst.file == PROGRAM_ADDRESS);
+   if (dst.index >= this->num_address_regs)
+      this->num_address_regs = dst.index + 1;
+
    emit(NULL, op, dst, src0);
 }
 
@@ -894,8 +918,8 @@ glsl_to_tgsi_visitor::add_constant(gl_register_file file,
       /* Search immediate storage to see if we already have an identical
        * immediate that we can use instead of adding a duplicate entry.
        */
-      foreach_iter(exec_list_iterator, iter, this->immediates) {
-         entry = (immediate_storage *)iter.get();
+      foreach_list(node, &this->immediates) {
+         entry = (immediate_storage *) node;
          
          if (entry->size == size &&
              entry->type == datatype &&
@@ -980,10 +1004,13 @@ type_size(const struct glsl_type *type)
       }
       return size;
    case GLSL_TYPE_SAMPLER:
+   case GLSL_TYPE_IMAGE:
       /* Samplers take up one slot in UNIFORMS[], but they're baked in
        * at link time.
        */
       return 1;
+   case GLSL_TYPE_ATOMIC_UINT:
+   case GLSL_TYPE_INTERFACE:
    case GLSL_TYPE_VOID:
    case GLSL_TYPE_ERROR:
       assert(!"Invalid type in type_size");
@@ -1003,17 +1030,28 @@ glsl_to_tgsi_visitor::get_temp(const glsl_type *type)
    st_src_reg src;
 
    src.type = native_integers ? type->base_type : GLSL_TYPE_FLOAT;
-   src.file = PROGRAM_TEMPORARY;
-   src.index = next_temp;
    src.reladdr = NULL;
-   next_temp += type_size(type);
+   src.negate = 0;
+
+   if (!options->EmitNoIndirectTemp &&
+       (type->is_array() || type->is_matrix())) {
+
+      src.file = PROGRAM_ARRAY;
+      src.index = next_array << 16 | 0x8000;
+      array_sizes[next_array] = type_size(type);
+      ++next_array;
+
+   } else {
+      src.file = PROGRAM_TEMPORARY;
+      src.index = next_temp;
+      next_temp += type_size(type);
+   }
 
    if (type->is_array() || type->is_record()) {
       src.swizzle = SWIZZLE_NOOP;
    } else {
       src.swizzle = swizzle_for_size(type->vector_elements);
    }
-   src.negate = 0;
 
    return src;
 }
@@ -1024,8 +1062,8 @@ glsl_to_tgsi_visitor::find_variable_storage(ir_variable *var)
    
    variable_storage *entry;
 
-   foreach_iter(exec_list_iterator, iter, this->variables) {
-      entry = (variable_storage *)iter.get();
+   foreach_list(node, &this->variables) {
+      entry = (variable_storage *) node;
 
       if (entry->var == var)
          return entry;
@@ -1040,11 +1078,11 @@ glsl_to_tgsi_visitor::visit(ir_variable *ir)
    if (strcmp(ir->name, "gl_FragCoord") == 0) {
       struct gl_fragment_program *fp = (struct gl_fragment_program *)this->prog;
 
-      fp->OriginUpperLeft = ir->origin_upper_left;
-      fp->PixelCenterInteger = ir->pixel_center_integer;
+      fp->OriginUpperLeft = ir->data.origin_upper_left;
+      fp->PixelCenterInteger = ir->data.pixel_center_integer;
    }
 
-   if (ir->mode == ir_var_uniform && strncmp(ir->name, "gl_", 3) == 0) {
+   if (ir->data.mode == ir_var_uniform && strncmp(ir->name, "gl_", 3) == 0) {
       unsigned int i;
       const ir_state_slot *const slots = ir->state_slots;
       assert(ir->state_slots != NULL);
@@ -1076,13 +1114,11 @@ glsl_to_tgsi_visitor::visit(ir_variable *ir)
           */
          assert((int) ir->num_state_slots == type_size(ir->type));
 
-         storage = new(mem_ctx) variable_storage(ir, PROGRAM_TEMPORARY,
-        					 this->next_temp);
-         this->variables.push_tail(storage);
-         this->next_temp += type_size(ir->type);
+         dst = st_dst_reg(get_temp(ir->type));
 
-         dst = st_dst_reg(st_src_reg(PROGRAM_TEMPORARY, storage->index,
-               native_integers ? ir->type->base_type : GLSL_TYPE_FLOAT));
+         storage = new(mem_ctx) variable_storage(ir, dst.file, dst.index);
+
+         this->variables.push_tail(storage);
       }
 
 
@@ -1123,53 +1159,10 @@ glsl_to_tgsi_visitor::visit(ir_variable *ir)
 void
 glsl_to_tgsi_visitor::visit(ir_loop *ir)
 {
-   ir_dereference_variable *counter = NULL;
-
-   if (ir->counter != NULL)
-      counter = new(ir) ir_dereference_variable(ir->counter);
-
-   if (ir->from != NULL) {
-      assert(ir->counter != NULL);
-
-      ir_assignment *a = new(ir) ir_assignment(counter, ir->from, NULL);
-
-      a->accept(this);
-      delete a;
-   }
-
    emit(NULL, TGSI_OPCODE_BGNLOOP);
 
-   if (ir->to) {
-      ir_expression *e =
-         new(ir) ir_expression(ir->cmp, glsl_type::bool_type,
-        		       counter, ir->to);
-      ir_if *if_stmt =  new(ir) ir_if(e);
-
-      ir_loop_jump *brk = new(ir) ir_loop_jump(ir_loop_jump::jump_break);
-
-      if_stmt->then_instructions.push_tail(brk);
-
-      if_stmt->accept(this);
-
-      delete if_stmt;
-      delete e;
-      delete brk;
-   }
-
    visit_exec_list(&ir->body_instructions, this);
 
-   if (ir->increment) {
-      ir_expression *e =
-         new(ir) ir_expression(ir_binop_add, counter->type,
-        		       counter, ir->increment);
-
-      ir_assignment *a = new(ir) ir_assignment(counter, e, NULL);
-
-      a->accept(this);
-      delete a;
-      delete e;
-   }
-
    emit(NULL, TGSI_OPCODE_ENDLOOP);
 }
 
@@ -1204,12 +1197,12 @@ glsl_to_tgsi_visitor::visit(ir_function *ir)
       const ir_function_signature *sig;
       exec_list empty;
 
-      sig = ir->matching_signature(&empty);
+      sig = ir->matching_signature(NULL, &empty);
 
       assert(sig);
 
-      foreach_iter(exec_list_iterator, iter, sig->body) {
-         ir_instruction *ir = (ir_instruction *)iter.get();
+      foreach_list(node, &sig->body) {
+         ir_instruction *ir = (ir_instruction *) node;
 
          ir->accept(this);
       }
@@ -1285,11 +1278,12 @@ glsl_to_tgsi_visitor::try_emit_mad_for_and_not(ir_expression *ir, int try_operan
 bool
 glsl_to_tgsi_visitor::try_emit_sat(ir_expression *ir)
 {
-   /* Saturates were only introduced to vertex programs in
-    * NV_vertex_program3, so don't give them to drivers in the VP.
+   /* Emit saturates in the vertex shader only if SM 3.0 is supported.
     */
-   if (this->prog->Target == GL_VERTEX_PROGRAM_ARB)
+   if (this->prog->Target == GL_VERTEX_PROGRAM_ARB &&
+       !st_context(this->ctx)->has_shader_model3) {
       return false;
+   }
 
    ir_rvalue *sat_src = ir->as_rvalue_to_saturate();
    if (!sat_src)
@@ -1332,10 +1326,11 @@ void
 glsl_to_tgsi_visitor::reladdr_to_temp(ir_instruction *ir,
         			    st_src_reg *reg, int *num_reladdr)
 {
-   if (!reg->reladdr)
+   if (!reg->reladdr && !reg->reladdr2)
       return;
 
-   emit_arl(ir, address_reg, *reg->reladdr);
+   if (reg->reladdr) emit_arl(ir, address_reg, *reg->reladdr);
+   if (reg->reladdr2) emit_arl(ir, address_reg2, *reg->reladdr2);
 
    if (*num_reladdr != 1) {
       st_src_reg temp = get_temp(glsl_type::vec4_type);
@@ -1383,9 +1378,9 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
       this->result.file = PROGRAM_UNDEFINED;
       ir->operands[operand]->accept(this);
       if (this->result.file == PROGRAM_UNDEFINED) {
-         ir_print_visitor v;
          printf("Failed to get tree for expression operand:\n");
-         ir->operands[operand]->accept(&v);
+         ir->operands[operand]->print();
+         printf("\n");
          exit(1);
       }
       op[operand] = this->result;
@@ -1675,30 +1670,82 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
    case ir_unop_any: {
       assert(ir->operands[0]->type->is_vector());
 
-      /* After the dot-product, the value will be an integer on the
-       * range [0,4].  Zero stays zero, and positive values become 1.0.
-       */
-      glsl_to_tgsi_instruction *const dp =
-         emit_dp(ir, result_dst, op[0], op[0],
-                 ir->operands[0]->type->vector_elements);
-      if (this->prog->Target == GL_FRAGMENT_PROGRAM_ARB &&
-          result_dst.type == GLSL_TYPE_FLOAT) {
-	      /* The clamping to [0,1] can be done for free in the fragment
-	       * shader with a saturate.
-	       */
-	      dp->saturate = true;
-      } else if (result_dst.type == GLSL_TYPE_FLOAT) {
-	      /* Negating the result of the dot-product gives values on the range
-	       * [-4, 0].  Zero stays zero, and negative values become 1.0.  This
-	       * is achieved using SLT.
-	       */
-	      st_src_reg slt_src = result_src;
-	      slt_src.negate = ~slt_src.negate;
-	      emit(ir, TGSI_OPCODE_SLT, result_dst, slt_src, st_src_reg_for_float(0.0));
-      }
-      else {
-         /* Use SNE 0 if integers are being used as boolean values. */
-         emit(ir, TGSI_OPCODE_SNE, result_dst, result_src, st_src_reg_for_int(0));
+      if (native_integers) {
+         int dst_swizzle = 0, op0_swizzle, i;
+         st_src_reg accum = op[0];
+
+         op0_swizzle = op[0].swizzle;
+         accum.swizzle = MAKE_SWIZZLE4(GET_SWZ(op0_swizzle, 0),
+                                       GET_SWZ(op0_swizzle, 0),
+                                       GET_SWZ(op0_swizzle, 0),
+                                       GET_SWZ(op0_swizzle, 0));
+         for (i = 0; i < 4; i++) {
+            if (result_dst.writemask & (1 << i)) {
+               dst_swizzle = MAKE_SWIZZLE4(i, i, i, i);
+               break;
+            }
+         }
+         assert(i != 4);
+         assert(ir->operands[0]->type->is_boolean());
+
+         /* OR all the components together, since they should be either 0 or ~0
+          */
+         switch (ir->operands[0]->type->vector_elements) {
+         case 4:
+            op[0].swizzle = MAKE_SWIZZLE4(GET_SWZ(op0_swizzle, 3),
+                                          GET_SWZ(op0_swizzle, 3),
+                                          GET_SWZ(op0_swizzle, 3),
+                                          GET_SWZ(op0_swizzle, 3));
+            emit(ir, TGSI_OPCODE_OR, result_dst, accum, op[0]);
+            accum = st_src_reg(result_dst);
+            accum.swizzle = dst_swizzle;
+            /* fallthrough */
+         case 3:
+            op[0].swizzle = MAKE_SWIZZLE4(GET_SWZ(op0_swizzle, 2),
+                                          GET_SWZ(op0_swizzle, 2),
+                                          GET_SWZ(op0_swizzle, 2),
+                                          GET_SWZ(op0_swizzle, 2));
+            emit(ir, TGSI_OPCODE_OR, result_dst, accum, op[0]);
+            accum = st_src_reg(result_dst);
+            accum.swizzle = dst_swizzle;
+            /* fallthrough */
+         case 2:
+            op[0].swizzle = MAKE_SWIZZLE4(GET_SWZ(op0_swizzle, 1),
+                                          GET_SWZ(op0_swizzle, 1),
+                                          GET_SWZ(op0_swizzle, 1),
+                                          GET_SWZ(op0_swizzle, 1));
+            emit(ir, TGSI_OPCODE_OR, result_dst, accum, op[0]);
+            break;
+         default:
+            assert(!"Unexpected vector size");
+            break;
+         }
+      } else {
+         /* After the dot-product, the value will be an integer on the
+          * range [0,4].  Zero stays zero, and positive values become 1.0.
+          */
+         glsl_to_tgsi_instruction *const dp =
+            emit_dp(ir, result_dst, op[0], op[0],
+                    ir->operands[0]->type->vector_elements);
+         if (this->prog->Target == GL_FRAGMENT_PROGRAM_ARB &&
+             result_dst.type == GLSL_TYPE_FLOAT) {
+            /* The clamping to [0,1] can be done for free in the fragment
+             * shader with a saturate.
+             */
+            dp->saturate = true;
+         } else if (result_dst.type == GLSL_TYPE_FLOAT) {
+            /* Negating the result of the dot-product gives values on the range
+             * [-4, 0].  Zero stays zero, and negative values become 1.0.  This
+             * is achieved using SLT.
+             */
+            st_src_reg slt_src = result_src;
+            slt_src.negate = ~slt_src.negate;
+            emit(ir, TGSI_OPCODE_SLT, result_dst, slt_src, st_src_reg_for_float(0.0));
+         }
+         else {
+            /* Use SNE 0 if integers are being used as boolean values. */
+            emit(ir, TGSI_OPCODE_SNE, result_dst, result_src, st_src_reg_for_int(0));
+         }
       }
       break;
    }
@@ -1760,13 +1807,18 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
       break;
 
    case ir_unop_sqrt:
-      /* sqrt(x) = x * rsq(x). */
-      emit_scalar(ir, TGSI_OPCODE_RSQ, result_dst, op[0]);
-      emit(ir, TGSI_OPCODE_MUL, result_dst, result_src, op[0]);
-      /* For incoming channels <= 0, set the result to 0. */
-      op[0].negate = ~op[0].negate;
-      emit(ir, TGSI_OPCODE_CMP, result_dst,
-        		  op[0], result_src, st_src_reg_for_float(0.0));
+      if (have_sqrt) {
+         emit_scalar(ir, TGSI_OPCODE_SQRT, result_dst, op[0]);
+      }
+      else {
+         /* sqrt(x) = x * rsq(x). */
+         emit_scalar(ir, TGSI_OPCODE_RSQ, result_dst, op[0]);
+         emit(ir, TGSI_OPCODE_MUL, result_dst, result_src, op[0]);
+         /* For incoming channels <= 0, set the result to 0. */
+         op[0].negate = ~op[0].negate;
+         emit(ir, TGSI_OPCODE_CMP, result_dst,
+              op[0], result_src, st_src_reg_for_float(0.0));
+      }
       break;
    case ir_unop_rsq:
       emit_scalar(ir, TGSI_OPCODE_RSQ, result_dst, op[0]);
@@ -1815,10 +1867,17 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
          emit(ir, TGSI_OPCODE_TRUNC, result_dst, op[0]);
       break;
    case ir_unop_bitcast_f2i:
+      result_src = op[0];
+      result_src.type = GLSL_TYPE_INT;
+      break;
    case ir_unop_bitcast_f2u:
+      result_src = op[0];
+      result_src.type = GLSL_TYPE_UINT;
+      break;
    case ir_unop_bitcast_i2f:
    case ir_unop_bitcast_u2f:
       result_src = op[0];
+      result_src.type = GLSL_TYPE_FLOAT;
       break;
    case ir_unop_f2b:
       emit(ir, TGSI_OPCODE_SNE, result_dst, op[0], st_src_reg_for_float(0.0));
@@ -1927,17 +1986,74 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
 
       if (ir->type->base_type == GLSL_TYPE_BOOL) {
          emit(ir, TGSI_OPCODE_USNE, result_dst, cbuf, st_src_reg_for_int(0));
-         result_src.negate = 1;
-         emit(ir, TGSI_OPCODE_UCMP, result_dst, result_src, st_src_reg_for_int(~0), st_src_reg_for_int(0));
       } else {
          emit(ir, TGSI_OPCODE_MOV, result_dst, cbuf);
       }
       break;
    }
+   case ir_triop_lrp:
+      /* note: we have to reorder the three args here */
+      emit(ir, TGSI_OPCODE_LRP, result_dst, op[2], op[1], op[0]);
+      break;
+   case ir_triop_csel:
+      if (this->ctx->Const.NativeIntegers)
+         emit(ir, TGSI_OPCODE_UCMP, result_dst, op[0], op[1], op[2]);
+      else {
+         op[0].negate = ~op[0].negate;
+         emit(ir, TGSI_OPCODE_CMP, result_dst, op[0], op[1], op[2]);
+      }
+      break;
+   case ir_triop_bitfield_extract:
+      emit(ir, TGSI_OPCODE_IBFE, result_dst, op[0], op[1], op[2]);
+      break;
+   case ir_quadop_bitfield_insert:
+      emit(ir, TGSI_OPCODE_BFI, result_dst, op[0], op[1], op[2], op[3]);
+      break;
+   case ir_unop_bitfield_reverse:
+      emit(ir, TGSI_OPCODE_BREV, result_dst, op[0]);
+      break;
+   case ir_unop_bit_count:
+      emit(ir, TGSI_OPCODE_POPC, result_dst, op[0]);
+      break;
+   case ir_unop_find_msb:
+      emit(ir, TGSI_OPCODE_IMSB, result_dst, op[0]);
+      break;
+   case ir_unop_find_lsb:
+      emit(ir, TGSI_OPCODE_LSB, result_dst, op[0]);
+      break;
+   case ir_binop_imul_high:
+      emit(ir, TGSI_OPCODE_IMUL_HI, result_dst, op[0], op[1]);
+      break;
+   case ir_triop_fma:
+      /* NOTE: Perhaps there should be a special opcode that enforces fused
+       * mul-add. Just use MAD for now.
+       */
+      emit(ir, TGSI_OPCODE_MAD, result_dst, op[0], op[1], op[2]);
+      break;
+   case ir_unop_pack_snorm_2x16:
+   case ir_unop_pack_unorm_2x16:
+   case ir_unop_pack_half_2x16:
+   case ir_unop_pack_snorm_4x8:
+   case ir_unop_pack_unorm_4x8:
+   case ir_unop_unpack_snorm_2x16:
+   case ir_unop_unpack_unorm_2x16:
+   case ir_unop_unpack_half_2x16:
+   case ir_unop_unpack_half_2x16_split_x:
+   case ir_unop_unpack_half_2x16_split_y:
+   case ir_unop_unpack_snorm_4x8:
+   case ir_unop_unpack_unorm_4x8:
+   case ir_binop_pack_half_2x16_split:
+   case ir_binop_bfm:
+   case ir_triop_bfi:
    case ir_quadop_vector:
-      /* This operation should have already been handled.
+   case ir_binop_vector_extract:
+   case ir_triop_vector_insert:
+   case ir_binop_ldexp:
+   case ir_binop_carry:
+   case ir_binop_borrow:
+      /* This operation is not supported, or should have already been handled.
        */
-      assert(!"Should not get here.");
+      assert(!"Invalid ir opcode in glsl_to_tgsi_visitor::visit()");
       break;
    }
 
@@ -1997,10 +2113,10 @@ glsl_to_tgsi_visitor::visit(ir_dereference_variable *ir)
    ir_variable *var = ir->var;
 
    if (!entry) {
-      switch (var->mode) {
+      switch (var->data.mode) {
       case ir_var_uniform:
          entry = new(mem_ctx) variable_storage(var, PROGRAM_UNIFORM,
-        				       var->location);
+        				       var->data.location);
          this->variables.push_tail(entry);
          break;
       case ir_var_shader_in:
@@ -2009,29 +2125,30 @@ glsl_to_tgsi_visitor::visit(ir_dereference_variable *ir)
           * generic attributes (glBindVertexLocation), and
           * user-defined varyings.
           */
-         assert(var->location != -1);
+         assert(var->data.location != -1);
          entry = new(mem_ctx) variable_storage(var,
                                                PROGRAM_INPUT,
-                                               var->location);
+                                               var->data.location);
          break;
       case ir_var_shader_out:
-         assert(var->location != -1);
+         assert(var->data.location != -1);
          entry = new(mem_ctx) variable_storage(var,
                                                PROGRAM_OUTPUT,
-                                               var->location + var->index);
+                                               var->data.location
+                                               + var->data.index);
          break;
       case ir_var_system_value:
          entry = new(mem_ctx) variable_storage(var,
                                                PROGRAM_SYSTEM_VALUE,
-                                               var->location);
+                                               var->data.location);
          break;
       case ir_var_auto:
       case ir_var_temporary:
-         entry = new(mem_ctx) variable_storage(var, PROGRAM_TEMPORARY,
-        				       this->next_temp);
+         st_src_reg src = get_temp(var->type);
+
+         entry = new(mem_ctx) variable_storage(var, src.file, src.index);
          this->variables.push_tail(entry);
 
-         next_temp += type_size(var->type);
          break;
       }
 
@@ -2052,14 +2169,26 @@ glsl_to_tgsi_visitor::visit(ir_dereference_array *ir)
    ir_constant *index;
    st_src_reg src;
    int element_size = type_size(ir->type);
+   bool is_2D_input;
 
    index = ir->array_index->constant_expression_value();
 
    ir->array->accept(this);
    src = this->result;
 
+   is_2D_input = this->prog->Target == GL_GEOMETRY_PROGRAM_NV &&
+                 src.file == PROGRAM_INPUT &&
+                 ir->array->ir_type != ir_type_dereference_array;
+
+   if (is_2D_input)
+      element_size = 1;
+
    if (index) {
-      src.index += index->value.i[0] * element_size;
+      if (is_2D_input) {
+         src.index2D = index->value.i[0];
+         src.has_index2 = true;
+      } else
+         src.index += index->value.i[0] * element_size;
    } else {
       /* Variable index array dereference.  It eats the "vec4" of the
        * base of the array and an index that offsets the TGSI register
@@ -2082,7 +2211,7 @@ glsl_to_tgsi_visitor::visit(ir_dereference_array *ir)
       /* If there was already a relative address register involved, add the
        * new and the old together to get the new offset.
        */
-      if (src.reladdr != NULL) {
+      if (!is_2D_input && src.reladdr != NULL) {
          st_src_reg accum_reg = get_temp(native_integers ?
                                 glsl_type::int_type : glsl_type::float_type);
 
@@ -2092,8 +2221,15 @@ glsl_to_tgsi_visitor::visit(ir_dereference_array *ir)
          index_reg = accum_reg;
       }
 
-      src.reladdr = ralloc(mem_ctx, st_src_reg);
-      memcpy(src.reladdr, &index_reg, sizeof(index_reg));
+      if (is_2D_input) {
+         src.reladdr2 = ralloc(mem_ctx, st_src_reg);
+         memcpy(src.reladdr2, &index_reg, sizeof(index_reg));
+         src.index2D = 0;
+         src.has_index2 = true;
+      } else {
+         src.reladdr = ralloc(mem_ctx, st_src_reg);
+         memcpy(src.reladdr, &index_reg, sizeof(index_reg));
+      }
    }
 
    /* If the type is smaller than a vec4, replicate the last channel out. */
@@ -2303,7 +2439,7 @@ glsl_to_tgsi_visitor::visit(ir_assignment *ir)
       assert(!ir->lhs->type->is_scalar() && !ir->lhs->type->is_vector());
       l.writemask = WRITEMASK_XYZW;
    } else if (ir->lhs->type->is_scalar() &&
-              ir->lhs->variable_referenced()->mode == ir_var_shader_out) {
+              ir->lhs->variable_referenced()->data.mode == ir_var_shader_out) {
       /* FINISHME: This hack makes writing to gl_FragDepth, which lives in the
        * FINISHME: W component of fragment shader output zero, work correctly.
        */
@@ -2353,8 +2489,8 @@ glsl_to_tgsi_visitor::visit(ir_assignment *ir)
          if (native_integers) {
             /* This is necessary because TGSI's CMP instruction expects the
              * condition to be a float, and we store booleans as integers.
-             * If TGSI had a UCMP instruction or similar, this extra
-             * instruction would not be necessary.
+             * TODO: really want to avoid i2f path and use UCMP. Requires
+             * changes to process_move_condition though too.
              */
             condition_temp = get_temp(glsl_type::vec4_type);
             condition.negate = 0;
@@ -2412,8 +2548,8 @@ glsl_to_tgsi_visitor::visit(ir_constant *ir)
       st_src_reg temp_base = get_temp(ir->type);
       st_dst_reg temp = st_dst_reg(temp_base);
 
-      foreach_iter(exec_list_iterator, iter, ir->components) {
-         ir_constant *field_value = (ir_constant *)iter.get();
+      foreach_list(node, &ir->components) {
+         ir_constant *field_value = (ir_constant *) node;
          int size = type_size(field_value->type);
 
          assert(size > 0);
@@ -2529,8 +2665,8 @@ glsl_to_tgsi_visitor::get_function_signature(ir_function_signature *sig)
 {
    function_entry *entry;
 
-   foreach_iter(exec_list_iterator, iter, this->function_signatures) {
-      entry = (function_entry *)iter.get();
+   foreach_list(node, &this->function_signatures) {
+      entry = (function_entry *) node;
 
       if (entry->sig == sig)
          return entry;
@@ -2542,18 +2678,17 @@ glsl_to_tgsi_visitor::get_function_signature(ir_function_signature *sig)
    entry->bgn_inst = NULL;
 
    /* Allocate storage for all the parameters. */
-   foreach_iter(exec_list_iterator, iter, sig->parameters) {
-      ir_variable *param = (ir_variable *)iter.get();
+   foreach_list(node, &sig->parameters) {
+      ir_variable *param = (ir_variable *) node;
       variable_storage *storage;
 
       storage = find_variable_storage(param);
       assert(!storage);
 
-      storage = new(mem_ctx) variable_storage(param, PROGRAM_TEMPORARY,
-        				      this->next_temp);
-      this->variables.push_tail(storage);
+      st_src_reg src = get_temp(param->type);
 
-      this->next_temp += type_size(param->type);
+      storage = new(mem_ctx) variable_storage(param, src.file, src.index);
+      this->variables.push_tail(storage);
    }
 
    if (!sig->return_type->is_void()) {
@@ -2575,13 +2710,13 @@ glsl_to_tgsi_visitor::visit(ir_call *ir)
    int i;
 
    /* Process in parameters. */
-   exec_list_iterator sig_iter = sig->parameters.iterator();
-   foreach_iter(exec_list_iterator, iter, *ir) {
-      ir_rvalue *param_rval = (ir_rvalue *)iter.get();
-      ir_variable *param = (ir_variable *)sig_iter.get();
+   foreach_two_lists(formal_node, &sig->parameters,
+                     actual_node, &ir->actual_parameters) {
+      ir_rvalue *param_rval = (ir_rvalue *) actual_node;
+      ir_variable *param = (ir_variable *) formal_node;
 
-      if (param->mode == ir_var_function_in ||
-          param->mode == ir_var_function_inout) {
+      if (param->data.mode == ir_var_function_in ||
+          param->data.mode == ir_var_function_inout) {
          variable_storage *storage = find_variable_storage(param);
          assert(storage);
 
@@ -2601,23 +2736,20 @@ glsl_to_tgsi_visitor::visit(ir_call *ir)
             r.index++;
          }
       }
-
-      sig_iter.next();
    }
-   assert(!sig_iter.has_next());
 
    /* Emit call instruction */
    call_inst = emit(ir, TGSI_OPCODE_CAL);
    call_inst->function = entry;
 
    /* Process out parameters. */
-   sig_iter = sig->parameters.iterator();
-   foreach_iter(exec_list_iterator, iter, *ir) {
-      ir_rvalue *param_rval = (ir_rvalue *)iter.get();
-      ir_variable *param = (ir_variable *)sig_iter.get();
+   foreach_two_lists(formal_node, &sig->parameters,
+                     actual_node, &ir->actual_parameters) {
+      ir_rvalue *param_rval = (ir_rvalue *) actual_node;
+      ir_variable *param = (ir_variable *) formal_node;
 
-      if (param->mode == ir_var_function_out ||
-          param->mode == ir_var_function_inout) {
+      if (param->data.mode == ir_var_function_out ||
+          param->data.mode == ir_var_function_inout) {
          variable_storage *storage = find_variable_storage(param);
          assert(storage);
 
@@ -2637,10 +2769,7 @@ glsl_to_tgsi_visitor::visit(ir_call *ir)
             r.index++;
          }
       }
-
-      sig_iter.next();
    }
-   assert(!sig_iter.has_next());
 
    /* Process return value. */
    this->result = entry->return_reg;
@@ -2649,12 +2778,13 @@ glsl_to_tgsi_visitor::visit(ir_call *ir)
 void
 glsl_to_tgsi_visitor::visit(ir_texture *ir)
 {
-   st_src_reg result_src, coord, cube_sc, lod_info, projector, dx, dy, offset;
+   st_src_reg result_src, coord, cube_sc, lod_info, projector, dx, dy, offset[MAX_GLSL_TEXTURE_OFFSET], sample_index, component;
    st_dst_reg result_dst, coord_dst, cube_sc_dst;
    glsl_to_tgsi_instruction *inst = NULL;
    unsigned opcode = TGSI_OPCODE_NOP;
    const glsl_type *sampler_type = ir->sampler->type;
    bool is_cube_array = false;
+   unsigned i;
 
    /* if we are a cube array sampler */
    if ((sampler_type->sampler_dimensionality == GLSL_SAMPLER_DIM_CUBE &&
@@ -2672,6 +2802,7 @@ glsl_to_tgsi_visitor::visit(ir_texture *ir)
        */
       coord = get_temp(glsl_type::vec4_type);
       coord_dst = st_dst_reg(coord);
+      coord_dst.writemask = (1 << ir->coordinate->type->vector_elements) - 1;
       emit(ir, TGSI_OPCODE_MOV, coord_dst, this->result);
    }
 
@@ -2689,16 +2820,28 @@ glsl_to_tgsi_visitor::visit(ir_texture *ir)
    switch (ir->op) {
    case ir_tex:
       opcode = (is_cube_array && ir->shadow_comparitor) ? TGSI_OPCODE_TEX2 : TGSI_OPCODE_TEX; 
+      if (ir->offset) {
+         ir->offset->accept(this);
+         offset[0] = this->result;
+      }
       break;
    case ir_txb:
       opcode = is_cube_array ? TGSI_OPCODE_TXB2 : TGSI_OPCODE_TXB;
       ir->lod_info.bias->accept(this);
       lod_info = this->result;
+      if (ir->offset) {
+         ir->offset->accept(this);
+         offset[0] = this->result;
+      }
       break;
    case ir_txl:
       opcode = is_cube_array ? TGSI_OPCODE_TXL2 : TGSI_OPCODE_TXL;
       ir->lod_info.lod->accept(this);
       lod_info = this->result;
+      if (ir->offset) {
+         ir->offset->accept(this);
+         offset[0] = this->result;
+      }
       break;
    case ir_txd:
       opcode = TGSI_OPCODE_TXD;
@@ -2706,6 +2849,10 @@ glsl_to_tgsi_visitor::visit(ir_texture *ir)
       dx = this->result;
       ir->lod_info.grad.dPdy->accept(this);
       dy = this->result;
+      if (ir->offset) {
+         ir->offset->accept(this);
+         offset[0] = this->result;
+      }
       break;
    case ir_txs:
       opcode = TGSI_OPCODE_TXQ;
@@ -2717,10 +2864,40 @@ glsl_to_tgsi_visitor::visit(ir_texture *ir)
       ir->lod_info.lod->accept(this);
       lod_info = this->result;
       if (ir->offset) {
-	 ir->offset->accept(this);
-	 offset = this->result;
+         ir->offset->accept(this);
+         offset[0] = this->result;
       }
       break;
+   case ir_txf_ms:
+      opcode = TGSI_OPCODE_TXF;
+      ir->lod_info.sample_index->accept(this);
+      sample_index = this->result;
+      break;
+   case ir_tg4:
+      opcode = TGSI_OPCODE_TG4;
+      ir->lod_info.component->accept(this);
+      component = this->result;
+      if (ir->offset) {
+         ir->offset->accept(this);
+         if (ir->offset->type->base_type == GLSL_TYPE_ARRAY) {
+            const glsl_type *elt_type = ir->offset->type->fields.array;
+            for (i = 0; i < ir->offset->type->length; i++) {
+               offset[i] = this->result;
+               offset[i].index += i * type_size(elt_type);
+               offset[i].type = elt_type->base_type;
+               offset[i].swizzle = swizzle_for_size(elt_type->vector_elements);
+            }
+         } else {
+            offset[0] = this->result;
+         }
+      }
+      break;
+   case ir_lod:
+      opcode = TGSI_OPCODE_LODQ;
+      break;
+   case ir_query_levels:
+      assert(!"Unexpected ir_query_levels opcode");
+      break;
    }
 
    if (ir->projector) {
@@ -2803,7 +2980,11 @@ glsl_to_tgsi_visitor::visit(ir_texture *ir)
       }
    }
 
-   if (opcode == TGSI_OPCODE_TXL || opcode == TGSI_OPCODE_TXB ||
+   if (ir->op == ir_txf_ms) {
+      coord_dst.writemask = WRITEMASK_W;
+      emit(ir, TGSI_OPCODE_MOV, coord_dst, sample_index);
+      coord_dst.writemask = WRITEMASK_XYZW;
+   } else if (opcode == TGSI_OPCODE_TXL || opcode == TGSI_OPCODE_TXB ||
        opcode == TGSI_OPCODE_TXF) {
       /* TGSI stores LOD or LOD bias in the last channel of the coords. */
       coord_dst.writemask = WRITEMASK_W;
@@ -2821,7 +3002,13 @@ glsl_to_tgsi_visitor::visit(ir_texture *ir)
       inst = emit(ir, opcode, result_dst, coord, lod_info);
    } else if (opcode == TGSI_OPCODE_TEX2) {
       inst = emit(ir, opcode, result_dst, coord, cube_sc);
-   } else 
+   } else if (opcode == TGSI_OPCODE_TG4) {
+      if (is_cube_array && ir->shadow_comparitor) {
+         inst = emit(ir, opcode, result_dst, coord, cube_sc);
+      } else {
+         inst = emit(ir, opcode, result_dst, coord, component);
+      }
+   } else
       inst = emit(ir, opcode, result_dst, coord);
 
    if (ir->shadow_comparitor)
@@ -2832,12 +3019,9 @@ glsl_to_tgsi_visitor::visit(ir_texture *ir)
         					   this->prog);
 
    if (ir->offset) {
-       inst->tex_offset_num_offset = 1;
-       inst->tex_offsets[0].Index = offset.index;
-       inst->tex_offsets[0].File = offset.file;
-       inst->tex_offsets[0].SwizzleX = GET_SWZ(offset.swizzle, 0);
-       inst->tex_offsets[0].SwizzleY = GET_SWZ(offset.swizzle, 1);
-       inst->tex_offsets[0].SwizzleZ = GET_SWZ(offset.swizzle, 2);
+      for (i = 0; i < MAX_GLSL_TEXTURE_OFFSET && offset[i].file != PROGRAM_UNDEFINED; i++)
+         inst->tex_offsets[i] = offset[i];
+      inst->tex_offset_num_offset = i;
    }
 
    switch (sampler_type->sampler_dimensionality) {
@@ -2865,6 +3049,10 @@ glsl_to_tgsi_visitor::visit(ir_texture *ir)
    case GLSL_SAMPLER_DIM_EXTERNAL:
       inst->tex_target = TEXTURE_EXTERNAL_INDEX;
       break;
+   case GLSL_SAMPLER_DIM_MS:
+      inst->tex_target = (sampler_type->sampler_array)
+         ? TEXTURE_2D_MULTISAMPLE_ARRAY_INDEX : TEXTURE_2D_MULTISAMPLE_INDEX;
+      break;
    default:
       assert(!"Should not get here.");
    }
@@ -2902,41 +3090,25 @@ glsl_to_tgsi_visitor::visit(ir_discard *ir)
    if (ir->condition) {
       ir->condition->accept(this);
       this->result.negate = ~this->result.negate;
-      emit(ir, TGSI_OPCODE_KIL, undef_dst, this->result);
+      emit(ir, TGSI_OPCODE_KILL_IF, undef_dst, this->result);
    } else {
-      emit(ir, TGSI_OPCODE_KILP);
+      /* unconditional kil */
+      emit(ir, TGSI_OPCODE_KILL);
    }
 }
 
 void
 glsl_to_tgsi_visitor::visit(ir_if *ir)
 {
-   glsl_to_tgsi_instruction *cond_inst, *if_inst;
-   glsl_to_tgsi_instruction *prev_inst;
-
-   prev_inst = (glsl_to_tgsi_instruction *)this->instructions.get_tail();
+   unsigned if_opcode;
+   glsl_to_tgsi_instruction *if_inst;
 
    ir->condition->accept(this);
    assert(this->result.file != PROGRAM_UNDEFINED);
 
-   if (this->options->EmitCondCodes) {
-      cond_inst = (glsl_to_tgsi_instruction *)this->instructions.get_tail();
-
-      /* See if we actually generated any instruction for generating
-       * the condition.  If not, then cook up a move to a temp so we
-       * have something to set cond_update on.
-       */
-      if (cond_inst == prev_inst) {
-         st_src_reg temp = get_temp(glsl_type::bool_type);
-         cond_inst = emit(ir->condition, TGSI_OPCODE_MOV, st_dst_reg(temp), result);
-      }
-      cond_inst->cond_update = GL_TRUE;
+   if_opcode = native_integers ? TGSI_OPCODE_UIF : TGSI_OPCODE_IF;
 
-      if_inst = emit(ir->condition, TGSI_OPCODE_IF);
-      if_inst->dst.cond_mask = COND_NE;
-   } else {
-      if_inst = emit(ir->condition, TGSI_OPCODE_IF, undef_dst, this->result);
-   }
+   if_inst = emit(ir->condition, if_opcode, undef_dst, this->result);
 
    this->instructions.push_tail(if_inst);
 
@@ -2950,16 +3122,31 @@ glsl_to_tgsi_visitor::visit(ir_if *ir)
    if_inst = emit(ir->condition, TGSI_OPCODE_ENDIF);
 }
 
+
+void
+glsl_to_tgsi_visitor::visit(ir_emit_vertex *ir)
+{
+   assert(this->prog->Target == GL_GEOMETRY_PROGRAM_NV);
+   emit(ir, TGSI_OPCODE_EMIT);
+}
+
+void
+glsl_to_tgsi_visitor::visit(ir_end_primitive *ir)
+{
+   assert(this->prog->Target == GL_GEOMETRY_PROGRAM_NV);
+   emit(ir, TGSI_OPCODE_ENDPRIM);
+}
+
 glsl_to_tgsi_visitor::glsl_to_tgsi_visitor()
 {
    result.file = PROGRAM_UNDEFINED;
    next_temp = 1;
+   next_array = 0;
    next_signature_id = 1;
    num_immediates = 0;
    current_function = NULL;
    num_address_regs = 0;
    samplers_used = 0;
-   indirect_addr_temps = false;
    indirect_addr_consts = false;
    glsl_version = 0;
    native_integers = false;
@@ -2990,8 +3177,8 @@ count_resources(glsl_to_tgsi_visitor *v, gl_program *prog)
 {
    v->samplers_used = 0;
 
-   foreach_iter(exec_list_iterator, iter, v->instructions) {
-      glsl_to_tgsi_instruction *inst = (glsl_to_tgsi_instruction *)iter.get();
+   foreach_list(node, &v->instructions) {
+      glsl_to_tgsi_instruction *inst = (glsl_to_tgsi_instruction *) node;
 
       if (is_tex_instruction(inst->op)) {
          v->samplers_used |= 1 << inst->sampler;
@@ -3038,7 +3225,7 @@ set_uniform_initializer(struct gl_context *ctx, void *mem_ctx,
         	"Couldn't find uniform for initializer %s\n", name);
       return;
    }
-   int loc = _mesa_uniform_merge_location_offset(index, offset);
+   int loc = _mesa_uniform_merge_location_offset(shader_program, index, offset);
 
    for (unsigned int i = 0; i < (type->is_array() ? type->length : 1); i++) {
       ir_constant *element;
@@ -3136,8 +3323,8 @@ glsl_to_tgsi_visitor::simplify_cmp(void)
    memset(tempWrites, 0, sizeof(unsigned) * MAX_TEMPS);
    memset(outputWrites, 0, sizeof(outputWrites));
 
-   foreach_iter(exec_list_iterator, iter, this->instructions) {
-      glsl_to_tgsi_instruction *inst = (glsl_to_tgsi_instruction *)iter.get();
+   foreach_list(node, &this->instructions) {
+      glsl_to_tgsi_instruction *inst = (glsl_to_tgsi_instruction *) node;
       unsigned prevWriteMask = 0;
 
       /* Give up if we encounter relative addressing or flow control. */
@@ -3159,7 +3346,8 @@ glsl_to_tgsi_visitor::simplify_cmp(void)
          assert(inst->dst.index < MAX_TEMPS);
          prevWriteMask = tempWrites[inst->dst.index];
          tempWrites[inst->dst.index] |= inst->dst.writemask;
-      }
+      } else
+         continue;
 
       /* For a CMP to be considered a conditional write, the destination
        * register and source register two must be the same. */
@@ -3181,8 +3369,8 @@ glsl_to_tgsi_visitor::simplify_cmp(void)
 void
 glsl_to_tgsi_visitor::rename_temp_register(int index, int new_index)
 {
-   foreach_iter(exec_list_iterator, iter, this->instructions) {
-      glsl_to_tgsi_instruction *inst = (glsl_to_tgsi_instruction *)iter.get();
+   foreach_list(node, &this->instructions) {
+      glsl_to_tgsi_instruction *inst = (glsl_to_tgsi_instruction *) node;
       unsigned j;
       
       for (j=0; j < num_inst_src_regs(inst->op); j++) {
@@ -3191,6 +3379,13 @@ glsl_to_tgsi_visitor::rename_temp_register(int index, int new_index)
             inst->src[j].index = new_index;
          }
       }
+
+      for (j=0; j < inst->tex_offset_num_offset; j++) {
+         if (inst->tex_offsets[j].file == PROGRAM_TEMPORARY && 
+             inst->tex_offsets[j].index == index) {
+            inst->tex_offsets[j].index = new_index;
+         }
+      }
       
       if (inst->dst.file == PROGRAM_TEMPORARY && inst->dst.index == index) {
          inst->dst.index = new_index;
@@ -3205,8 +3400,8 @@ glsl_to_tgsi_visitor::get_first_temp_read(int index)
    int loop_start = -1; /* index of the first active BGNLOOP (if any) */
    unsigned i = 0, j;
    
-   foreach_iter(exec_list_iterator, iter, this->instructions) {
-      glsl_to_tgsi_instruction *inst = (glsl_to_tgsi_instruction *)iter.get();
+   foreach_list(node, &this->instructions) {
+      glsl_to_tgsi_instruction *inst = (glsl_to_tgsi_instruction *) node;
       
       for (j=0; j < num_inst_src_regs(inst->op); j++) {
          if (inst->src[j].file == PROGRAM_TEMPORARY && 
@@ -3214,6 +3409,12 @@ glsl_to_tgsi_visitor::get_first_temp_read(int index)
             return (depth == 0) ? i : loop_start;
          }
       }
+      for (j=0; j < inst->tex_offset_num_offset; j++) {
+         if (inst->tex_offsets[j].file == PROGRAM_TEMPORARY && 
+             inst->tex_offsets[j].index == index) {
+            return (depth == 0) ? i : loop_start;
+         }
+      }
       
       if (inst->op == TGSI_OPCODE_BGNLOOP) {
          if(depth++ == 0)
@@ -3237,8 +3438,8 @@ glsl_to_tgsi_visitor::get_first_temp_write(int index)
    int loop_start = -1; /* index of the first active BGNLOOP (if any) */
    int i = 0;
    
-   foreach_iter(exec_list_iterator, iter, this->instructions) {
-      glsl_to_tgsi_instruction *inst = (glsl_to_tgsi_instruction *)iter.get();
+   foreach_list(node, &this->instructions) {
+      glsl_to_tgsi_instruction *inst = (glsl_to_tgsi_instruction *) node;
       
       if (inst->dst.file == PROGRAM_TEMPORARY && inst->dst.index == index) {
          return (depth == 0) ? i : loop_start;
@@ -3266,8 +3467,8 @@ glsl_to_tgsi_visitor::get_last_temp_read(int index)
    int last = -1; /* index of last instruction that reads the temporary */
    unsigned i = 0, j;
    
-   foreach_iter(exec_list_iterator, iter, this->instructions) {
-      glsl_to_tgsi_instruction *inst = (glsl_to_tgsi_instruction *)iter.get();
+   foreach_list(node, &this->instructions) {
+      glsl_to_tgsi_instruction *inst = (glsl_to_tgsi_instruction *) node;
       
       for (j=0; j < num_inst_src_regs(inst->op); j++) {
          if (inst->src[j].file == PROGRAM_TEMPORARY && 
@@ -3275,6 +3476,11 @@ glsl_to_tgsi_visitor::get_last_temp_read(int index)
             last = (depth == 0) ? i : -2;
          }
       }
+      for (j=0; j < inst->tex_offset_num_offset; j++) {
+          if (inst->tex_offsets[j].file == PROGRAM_TEMPORARY &&
+              inst->tex_offsets[j].index == index)
+              last = (depth == 0) ? i : -2;
+      }
       
       if (inst->op == TGSI_OPCODE_BGNLOOP)
          depth++;
@@ -3297,8 +3503,8 @@ glsl_to_tgsi_visitor::get_last_temp_write(int index)
    int last = -1; /* index of last instruction that writes to the temporary */
    int i = 0;
    
-   foreach_iter(exec_list_iterator, iter, this->instructions) {
-      glsl_to_tgsi_instruction *inst = (glsl_to_tgsi_instruction *)iter.get();
+   foreach_list(node, &this->instructions) {
+      glsl_to_tgsi_instruction *inst = (glsl_to_tgsi_instruction *) node;
       
       if (inst->dst.file == PROGRAM_TEMPORARY && inst->dst.index == index)
          last = (depth == 0) ? i : -2;
@@ -3346,8 +3552,8 @@ glsl_to_tgsi_visitor::copy_propagate(void)
    int *acp_level = rzalloc_array(mem_ctx, int, this->next_temp * 4);
    int level = 0;
 
-   foreach_iter(exec_list_iterator, iter, this->instructions) {
-      glsl_to_tgsi_instruction *inst = (glsl_to_tgsi_instruction *)iter.get();
+   foreach_list(node, &this->instructions) {
+      glsl_to_tgsi_instruction *inst = (glsl_to_tgsi_instruction *) node;
 
       assert(inst->dst.file != PROGRAM_TEMPORARY
              || inst->dst.index < this->next_temp);
@@ -3359,7 +3565,8 @@ glsl_to_tgsi_visitor::copy_propagate(void)
          int acp_base = inst->src[r].index * 4;
 
          if (inst->src[r].file != PROGRAM_TEMPORARY ||
-             inst->src[r].reladdr)
+             inst->src[r].reladdr ||
+             inst->src[r].reladdr2)
             continue;
 
          /* See if we can find entries in the ACP consisting of MOVs
@@ -3394,6 +3601,8 @@ glsl_to_tgsi_visitor::copy_propagate(void)
              */
             inst->src[r].file = first->src[0].file;
             inst->src[r].index = first->src[0].index;
+            inst->src[r].index2D = first->src[0].index2D;
+            inst->src[r].has_index2 = first->src[0].has_index2;
 
             int swizzle = 0;
             for (int i = 0; i < 4; i++) {
@@ -3414,6 +3623,7 @@ glsl_to_tgsi_visitor::copy_propagate(void)
          break;
 
       case TGSI_OPCODE_IF:
+      case TGSI_OPCODE_UIF:
          ++level;
          break;
 
@@ -3492,9 +3702,12 @@ glsl_to_tgsi_visitor::copy_propagate(void)
       /* If this is a copy, add it to the ACP. */
       if (inst->op == TGSI_OPCODE_MOV &&
           inst->dst.file == PROGRAM_TEMPORARY &&
+          !(inst->dst.file == inst->src[0].file &&
+             inst->dst.index == inst->src[0].index) &&
           !inst->dst.reladdr &&
           !inst->saturate &&
           !inst->src[0].reladdr &&
+          !inst->src[0].reladdr2 &&
           !inst->src[0].negate) {
          for (int i = 0; i < 4; i++) {
             if (inst->dst.writemask & (1 << i)) {
@@ -3510,7 +3723,8 @@ glsl_to_tgsi_visitor::copy_propagate(void)
 }
 
 /*
- * Tracks available PROGRAM_TEMPORARY registers for dead code elimination.
+ * On a basic block basis, tracks available PROGRAM_TEMPORARY registers for dead
+ * code elimination.
  *
  * The glsl_to_tgsi_visitor lazily produces code assuming that this pass
  * will occur.  As an example, a TXP production after copy propagation but 
@@ -3523,48 +3737,9 @@ glsl_to_tgsi_visitor::copy_propagate(void)
  * and after this pass:
  *
  * 0: TXP TEMP[2], INPUT[4].xyyw, texture[0], 2D;
- * 
- * FIXME: assumes that all functions are inlined (no support for BGNSUB/ENDSUB)
- * FIXME: doesn't eliminate all dead code inside of loops; it steps around them
- */
-void
-glsl_to_tgsi_visitor::eliminate_dead_code(void)
-{
-   int i;
-   
-   for (i=0; i < this->next_temp; i++) {
-      int last_read = get_last_temp_read(i);
-      int j = 0;
-      
-      foreach_iter(exec_list_iterator, iter, this->instructions) {
-         glsl_to_tgsi_instruction *inst = (glsl_to_tgsi_instruction *)iter.get();
-
-         if (inst->dst.file == PROGRAM_TEMPORARY && inst->dst.index == i &&
-             j > last_read)
-         {
-            iter.remove();
-            delete inst;
-         }
-         
-         j++;
-      }
-   }
-}
-
-/*
- * On a basic block basis, tracks available PROGRAM_TEMPORARY registers for dead
- * code elimination.  This is less primitive than eliminate_dead_code(), as it
- * is per-channel and can detect consecutive writes without a read between them
- * as dead code.  However, there is some dead code that can be eliminated by 
- * eliminate_dead_code() but not this function - for example, this function 
- * cannot eliminate an instruction writing to a register that is never read and
- * is the only instruction writing to that register.
- *
- * The glsl_to_tgsi_visitor lazily produces code assuming that this pass
- * will occur.
  */
 int
-glsl_to_tgsi_visitor::eliminate_dead_code_advanced(void)
+glsl_to_tgsi_visitor::eliminate_dead_code(void)
 {
    glsl_to_tgsi_instruction **writes = rzalloc_array(mem_ctx,
                                                      glsl_to_tgsi_instruction *,
@@ -3573,8 +3748,8 @@ glsl_to_tgsi_visitor::eliminate_dead_code_advanced(void)
    int level = 0;
    int removed = 0;
 
-   foreach_iter(exec_list_iterator, iter, this->instructions) {
-      glsl_to_tgsi_instruction *inst = (glsl_to_tgsi_instruction *)iter.get();
+   foreach_list(node, &this->instructions) {
+      glsl_to_tgsi_instruction *inst = (glsl_to_tgsi_instruction *) node;
 
       assert(inst->dst.file != PROGRAM_TEMPORARY
              || inst->dst.index < this->next_temp);
@@ -3616,6 +3791,7 @@ glsl_to_tgsi_visitor::eliminate_dead_code_advanced(void)
          break;
 
       case TGSI_OPCODE_IF:
+      case TGSI_OPCODE_UIF:
          ++level;
          /* fallthrough to default case to mark the condition as read */
       
@@ -3643,6 +3819,26 @@ glsl_to_tgsi_visitor::eliminate_dead_code_advanced(void)
                }
             }
          }
+         for (unsigned i = 0; i < inst->tex_offset_num_offset; i++) {
+            if (inst->tex_offsets[i].file == PROGRAM_TEMPORARY && inst->tex_offsets[i].reladdr){
+               /* Any temporary might be read, so no dead code elimination 
+                * across this instruction.
+                */
+               memset(writes, 0, sizeof(*writes) * this->next_temp * 4);
+            } else if (inst->tex_offsets[i].file == PROGRAM_TEMPORARY) {
+               /* Clear where it's used as src. */
+               int src_chans = 1 << GET_SWZ(inst->tex_offsets[i].swizzle, 0);
+               src_chans |= 1 << GET_SWZ(inst->tex_offsets[i].swizzle, 1);
+               src_chans |= 1 << GET_SWZ(inst->tex_offsets[i].swizzle, 2);
+               src_chans |= 1 << GET_SWZ(inst->tex_offsets[i].swizzle, 3);
+               
+               for (int c = 0; c < 4; c++) {
+              	   if (src_chans & (1 << c)) {
+              	      writes[4 * inst->tex_offsets[i].index + c] = NULL;
+              	   }
+               }
+            }
+         }
          break;
       }
 
@@ -3680,13 +3876,13 @@ glsl_to_tgsi_visitor::eliminate_dead_code_advanced(void)
    /* Now actually remove the instructions that are completely dead and update
     * the writemask of other instructions with dead channels.
     */
-   foreach_iter(exec_list_iterator, iter, this->instructions) {
-      glsl_to_tgsi_instruction *inst = (glsl_to_tgsi_instruction *)iter.get();
+   foreach_list_safe(node, &this->instructions) {
+      glsl_to_tgsi_instruction *inst = (glsl_to_tgsi_instruction *) node;
       
       if (!inst->dead_mask || !inst->dst.writemask)
          continue;
       else if ((inst->dst.writemask & ~inst->dead_mask) == 0) {
-         iter.remove();
+         inst->remove();
          delete inst;
          removed++;
       } else
@@ -3797,7 +3993,6 @@ get_pixel_transfer_visitor(struct st_fragment_program *fp,
    v->next_temp = original->next_temp;
    v->num_address_regs = original->num_address_regs;
    v->samplers_used = prog->SamplersUsed = original->samplers_used;
-   v->indirect_addr_temps = original->indirect_addr_temps;
    v->indirect_addr_consts = original->indirect_addr_consts;
    memcpy(&v->immediates, &original->immediates, sizeof(v->immediates));
    v->num_immediates = original->num_immediates;
@@ -3806,14 +4001,14 @@ get_pixel_transfer_visitor(struct st_fragment_program *fp,
     * Get initial pixel color from the texture.
     * TEX colorTemp, fragment.texcoord[0], texture[0], 2D;
     */
-   coord = st_src_reg(PROGRAM_INPUT, FRAG_ATTRIB_TEX0, glsl_type::vec2_type);
+   coord = st_src_reg(PROGRAM_INPUT, VARYING_SLOT_TEX0, glsl_type::vec2_type);
    src0 = v->get_temp(glsl_type::vec4_type);
    dst0 = st_dst_reg(src0);
    inst = v->emit(NULL, TGSI_OPCODE_TEX, dst0, coord);
    inst->sampler = 0;
    inst->tex_target = TEXTURE_2D_INDEX;
 
-   prog->InputsRead |= FRAG_BIT_TEX0;
+   prog->InputsRead |= VARYING_BIT_TEX0;
    prog->SamplersUsed |= (1 << 0); /* mark sampler 0 as used */
    v->samplers_used |= (1 << 0);
 
@@ -3868,8 +4063,8 @@ get_pixel_transfer_visitor(struct st_fragment_program *fp,
 
    /* Now copy the instructions from the original glsl_to_tgsi_visitor into the
     * new visitor. */
-   foreach_iter(exec_list_iterator, iter, original->instructions) {
-      glsl_to_tgsi_instruction *inst = (glsl_to_tgsi_instruction *)iter.get();
+   foreach_list(node, &original->instructions) {
+      glsl_to_tgsi_instruction *inst = (glsl_to_tgsi_instruction *) node;
       glsl_to_tgsi_instruction *newinst;
       st_src_reg src_regs[3];
 
@@ -3879,7 +4074,7 @@ get_pixel_transfer_visitor(struct st_fragment_program *fp,
       for (int i=0; i<3; i++) {
          src_regs[i] = inst->src[i];
          if (src_regs[i].file == PROGRAM_INPUT &&
-             src_regs[i].index == FRAG_ATTRIB_COL0)
+             src_regs[i].index == VARYING_SLOT_COL0)
          {
             src_regs[i].file = PROGRAM_TEMPORARY;
             src_regs[i].index = src0.index;
@@ -3928,20 +4123,19 @@ get_bitmap_visitor(struct st_fragment_program *fp,
    v->next_temp = original->next_temp;
    v->num_address_regs = original->num_address_regs;
    v->samplers_used = prog->SamplersUsed = original->samplers_used;
-   v->indirect_addr_temps = original->indirect_addr_temps;
    v->indirect_addr_consts = original->indirect_addr_consts;
    memcpy(&v->immediates, &original->immediates, sizeof(v->immediates));
    v->num_immediates = original->num_immediates;
 
    /* TEX tmp0, fragment.texcoord[0], texture[0], 2D; */
-   coord = st_src_reg(PROGRAM_INPUT, FRAG_ATTRIB_TEX0, glsl_type::vec2_type);
+   coord = st_src_reg(PROGRAM_INPUT, VARYING_SLOT_TEX0, glsl_type::vec2_type);
    src0 = v->get_temp(glsl_type::vec4_type);
    dst0 = st_dst_reg(src0);
    inst = v->emit(NULL, TGSI_OPCODE_TEX, dst0, coord);
    inst->sampler = samplerIndex;
    inst->tex_target = TEXTURE_2D_INDEX;
 
-   prog->InputsRead |= FRAG_BIT_TEX0;
+   prog->InputsRead |= VARYING_BIT_TEX0;
    prog->SamplersUsed |= (1 << samplerIndex); /* mark sampler as used */
    v->samplers_used |= (1 << samplerIndex);
 
@@ -3949,12 +4143,12 @@ get_bitmap_visitor(struct st_fragment_program *fp,
    src0.negate = NEGATE_XYZW;
    if (st->bitmap.tex_format == PIPE_FORMAT_L8_UNORM)
       src0.swizzle = SWIZZLE_XXXX;
-   inst = v->emit(NULL, TGSI_OPCODE_KIL, undef_dst, src0);
+   inst = v->emit(NULL, TGSI_OPCODE_KILL_IF, undef_dst, src0);
 
    /* Now copy the instructions from the original glsl_to_tgsi_visitor into the
     * new visitor. */
-   foreach_iter(exec_list_iterator, iter, original->instructions) {
-      glsl_to_tgsi_instruction *inst = (glsl_to_tgsi_instruction *)iter.get();
+   foreach_list(node, &original->instructions) {
+      glsl_to_tgsi_instruction *inst = (glsl_to_tgsi_instruction *) node;
       glsl_to_tgsi_instruction *newinst;
       st_src_reg src_regs[3];
 
@@ -3990,13 +4184,16 @@ struct st_translate {
    struct ureg_program *ureg;
 
    struct ureg_dst temps[MAX_TEMPS];
+   struct ureg_dst arrays[MAX_ARRAYS];
    struct ureg_src *constants;
    struct ureg_src *immediates;
    struct ureg_dst outputs[PIPE_MAX_SHADER_OUTPUTS];
    struct ureg_src inputs[PIPE_MAX_SHADER_INPUTS];
-   struct ureg_dst address[1];
+   struct ureg_dst address[2];
    struct ureg_src samplers[PIPE_MAX_SAMPLERS];
    struct ureg_src systemValues[SYSTEM_VALUE_MAX];
+   struct tgsi_texture_offset tex_offsets[MAX_GLSL_TEXTURE_OFFSET];
+   unsigned array_sizes[MAX_ARRAYS];
 
    const GLuint *inputMapping;
    const GLuint *outputMapping;
@@ -4026,7 +4223,11 @@ struct st_translate {
 static unsigned mesa_sysval_to_semantic[SYSTEM_VALUE_MAX] = {
    TGSI_SEMANTIC_FACE,
    TGSI_SEMANTIC_VERTEXID,
-   TGSI_SEMANTIC_INSTANCEID
+   TGSI_SEMANTIC_INSTANCEID,
+   TGSI_SEMANTIC_SAMPLEID,
+   TGSI_SEMANTIC_SAMPLEPOS,
+   TGSI_SEMANTIC_SAMPLEMASK,
+   TGSI_SEMANTIC_INVOCATIONID,
 };
 
 /**
@@ -4108,23 +4309,41 @@ dst_register(struct st_translate *t,
              gl_register_file file,
              GLuint index)
 {
+   unsigned array;
+
    switch(file) {
    case PROGRAM_UNDEFINED:
       return ureg_dst_undef();
 
    case PROGRAM_TEMPORARY:
+      assert(index >= 0);
+      assert(index < (int) Elements(t->temps));
+ 
       if (ureg_dst_is_undef(t->temps[index]))
          t->temps[index] = ureg_DECL_local_temporary(t->ureg);
 
       return t->temps[index];
 
+   case PROGRAM_ARRAY:
+      array = index >> 16;
+
+      assert(array >= 0);
+      assert(array < (int) Elements(t->arrays));
+
+      if (ureg_dst_is_undef(t->arrays[array]))
+         t->arrays[array] = ureg_DECL_array_temporary(
+            t->ureg, t->array_sizes[array], TRUE);
+
+      return ureg_dst_array_offset(t->arrays[array],
+                                   (int)(index & 0xFFFF) - 0x8000);
+
    case PROGRAM_OUTPUT:
       if (t->procType == TGSI_PROCESSOR_VERTEX)
-         assert(index < VERT_RESULT_MAX);
+         assert(index < VARYING_SLOT_MAX);
       else if (t->procType == TGSI_PROCESSOR_FRAGMENT)
          assert(index < FRAG_RESULT_MAX);
       else
-         assert(index < GEOM_RESULT_MAX);
+         assert(index < VARYING_SLOT_MAX);
 
       assert(t->outputMapping[index] < Elements(t->outputs));
 
@@ -4152,14 +4371,9 @@ src_register(struct st_translate *t,
       return ureg_src_undef();
 
    case PROGRAM_TEMPORARY:
-      assert(index >= 0);
-      assert(index < (int) Elements(t->temps));
-      if (ureg_dst_is_undef(t->temps[index]))
-         t->temps[index] = ureg_DECL_local_temporary(t->ureg);
-      return ureg_src(t->temps[index]);
+   case PROGRAM_ARRAY:
+      return ureg_src(dst_register(t, file, index));
 
-   case PROGRAM_ENV_PARAM:
-   case PROGRAM_LOCAL_PARAM:
    case PROGRAM_UNIFORM:
       assert(index >= 0);
       return t->constants[index];
@@ -4222,24 +4436,27 @@ translate_dst(struct st_translate *t,
       case TGSI_PROCESSOR_VERTEX:
          /* XXX if the geometry shader is present, this must be done there
           * instead of here. */
-         if (dst_reg->index == VERT_RESULT_COL0 ||
-             dst_reg->index == VERT_RESULT_COL1 ||
-             dst_reg->index == VERT_RESULT_BFC0 ||
-             dst_reg->index == VERT_RESULT_BFC1) {
+         if (dst_reg->index == VARYING_SLOT_COL0 ||
+             dst_reg->index == VARYING_SLOT_COL1 ||
+             dst_reg->index == VARYING_SLOT_BFC0 ||
+             dst_reg->index == VARYING_SLOT_BFC1) {
             dst = ureg_saturate(dst);
          }
          break;
 
       case TGSI_PROCESSOR_FRAGMENT:
-         if (dst_reg->index >= FRAG_RESULT_COLOR) {
+         if (dst_reg->index == FRAG_RESULT_COLOR ||
+             dst_reg->index >= FRAG_RESULT_DATA0) {
             dst = ureg_saturate(dst);
          }
          break;
       }
    }
 
-   if (dst_reg->reladdr != NULL)
+   if (dst_reg->reladdr != NULL) {
+      assert(dst_reg->file != PROGRAM_TEMPORARY);
       dst = ureg_dst_indirect(dst, ureg_src(t->address[0]));
+   }
 
    return dst;
 }
@@ -4252,6 +4469,15 @@ translate_src(struct st_translate *t, const st_src_reg *src_reg)
 {
    struct ureg_src src = src_register(t, src_reg->file, src_reg->index, src_reg->index2D);
 
+   if (t->procType == TGSI_PROCESSOR_GEOMETRY && src_reg->has_index2) {
+      src = src_register(t, src_reg->file, src_reg->index, src_reg->index2D);
+      if (src_reg->reladdr2)
+         src = ureg_src_dimension_indirect(src, ureg_src(t->address[1]),
+                                           src_reg->index2D);
+      else
+         src = ureg_src_dimension(src, src_reg->index2D);
+   }
+
    src = ureg_swizzle(src,
                       GET_SWZ(src_reg->swizzle, 0) & 0x3,
                       GET_SWZ(src_reg->swizzle, 1) & 0x3,
@@ -4262,26 +4488,8 @@ translate_src(struct st_translate *t, const st_src_reg *src_reg)
       src = ureg_negate(src);
 
    if (src_reg->reladdr != NULL) {
-      /* Normally ureg_src_indirect() would be used here, but a stupid compiler 
-       * bug in g++ makes ureg_src_indirect (an inline C function) erroneously 
-       * set the bit for src.Negate.  So we have to do the operation manually
-       * here to work around the compiler's problems. */
-      /*src = ureg_src_indirect(src, ureg_src(t->address[0]));*/
-      struct ureg_src addr = ureg_src(t->address[0]);
-      src.Indirect = 1;
-      src.IndirectFile = addr.File;
-      src.IndirectIndex = addr.Index;
-      src.IndirectSwizzle = addr.SwizzleX;
-      
-      if (src_reg->file != PROGRAM_INPUT &&
-          src_reg->file != PROGRAM_OUTPUT) {
-         /* If src_reg->index was negative, it was set to zero in
-          * src_register().  Reassign it now.  But don't do this
-          * for input/output regs since they get remapped while
-          * const buffers don't.
-          */
-         src.Index = src_reg->index;
-      }
+      assert(src_reg->file != PROGRAM_TEMPORARY);
+      src = ureg_src_indirect(src, ureg_src(t->address[0]));
    }
 
    return src;
@@ -4289,22 +4497,50 @@ translate_src(struct st_translate *t, const st_src_reg *src_reg)
 
 static struct tgsi_texture_offset
 translate_tex_offset(struct st_translate *t,
-                     const struct tgsi_texture_offset *in_offset)
+                     const st_src_reg *in_offset, int idx)
 {
    struct tgsi_texture_offset offset;
    struct ureg_src imm_src;
+   struct ureg_dst dst;
+   int array;
 
-   assert(in_offset->File == PROGRAM_IMMEDIATE);
-   imm_src = t->immediates[in_offset->Index];
-
-   offset.File = imm_src.File;
-   offset.Index = imm_src.Index;
-   offset.SwizzleX = imm_src.SwizzleX;
-   offset.SwizzleY = imm_src.SwizzleY;
-   offset.SwizzleZ = imm_src.SwizzleZ;
-   offset.File = TGSI_FILE_IMMEDIATE;
-   offset.Padding = 0;
-
+   switch (in_offset->file) {
+   case PROGRAM_IMMEDIATE:
+      imm_src = t->immediates[in_offset->index];
+
+      offset.File = imm_src.File;
+      offset.Index = imm_src.Index;
+      offset.SwizzleX = imm_src.SwizzleX;
+      offset.SwizzleY = imm_src.SwizzleY;
+      offset.SwizzleZ = imm_src.SwizzleZ;
+      offset.Padding = 0;
+      break;
+   case PROGRAM_TEMPORARY:
+      imm_src = ureg_src(t->temps[in_offset->index]);
+      offset.File = imm_src.File;
+      offset.Index = imm_src.Index;
+      offset.SwizzleX = GET_SWZ(in_offset->swizzle, 0);
+      offset.SwizzleY = GET_SWZ(in_offset->swizzle, 1);
+      offset.SwizzleZ = GET_SWZ(in_offset->swizzle, 2);
+      offset.Padding = 0;
+      break;
+   case PROGRAM_ARRAY:
+      array = in_offset->index >> 16;
+
+      assert(array >= 0);
+      assert(array < (int) Elements(t->arrays));
+
+      dst = t->arrays[array];
+      offset.File = dst.File;
+      offset.Index = dst.Index + (in_offset->index & 0xFFFF) - 0x8000;
+      offset.SwizzleX = GET_SWZ(in_offset->swizzle, 0);
+      offset.SwizzleY = GET_SWZ(in_offset->swizzle, 1);
+      offset.SwizzleZ = GET_SWZ(in_offset->swizzle, 2);
+      offset.Padding = 0;
+      break;
+   default:
+      break;
+   }
    return offset;
 }
 
@@ -4341,6 +4577,7 @@ compile_tgsi_instruction(struct st_translate *t,
    case TGSI_OPCODE_ELSE:
    case TGSI_OPCODE_ENDLOOP:
    case TGSI_OPCODE_IF:
+   case TGSI_OPCODE_UIF:
       assert(num_dst == 0);
       ureg_label_insn(ureg,
                       inst->op,
@@ -4359,9 +4596,11 @@ compile_tgsi_instruction(struct st_translate *t,
    case TGSI_OPCODE_TEX2:
    case TGSI_OPCODE_TXB2:
    case TGSI_OPCODE_TXL2:
+   case TGSI_OPCODE_TG4:
+   case TGSI_OPCODE_LODQ:
       src[num_src++] = t->samplers[inst->sampler];
       for (i = 0; i < inst->tex_offset_num_offset; i++) {
-         texoffsets[i] = translate_tex_offset(t, &inst->tex_offsets[i]);
+         texoffsets[i] = translate_tex_offset(t, &inst->tex_offsets[i], i);
       }
       tex_target = st_translate_texture_target(inst->tex_target, inst->tex_shadow);
 
@@ -4417,7 +4656,7 @@ emit_wpos_adjustment( struct st_translate *t,
 
    struct ureg_src wpostrans = ureg_DECL_constant( ureg, wposTransConst );
    struct ureg_dst wpos_temp = ureg_DECL_temporary( ureg );
-   struct ureg_src wpos_input = t->inputs[t->inputMapping[FRAG_ATTRIB_WPOS]];
+   struct ureg_src wpos_input = t->inputs[t->inputMapping[VARYING_SLOT_POS]];
 
    /* First, apply the coordinate shift: */
    if (adjX || adjY[0] || adjY[1]) {
@@ -4468,7 +4707,7 @@ emit_wpos_adjustment( struct st_translate *t,
 
    /* Use wpos_temp as position input from here on:
     */
-   t->inputs[t->inputMapping[FRAG_ATTRIB_WPOS]] = ureg_src(wpos_temp);
+   t->inputs[t->inputMapping[VARYING_SLOT_POS]] = ureg_src(wpos_temp);
 }
 
 
@@ -4586,21 +4825,21 @@ emit_face_var(struct st_translate *t)
 {
    struct ureg_program *ureg = t->ureg;
    struct ureg_dst face_temp = ureg_DECL_temporary(ureg);
-   struct ureg_src face_input = t->inputs[t->inputMapping[FRAG_ATTRIB_FACE]];
+   struct ureg_src face_input = t->inputs[t->inputMapping[VARYING_SLOT_FACE]];
 
    /* MOV_SAT face_temp, input[face] */
    face_temp = ureg_saturate(face_temp);
    ureg_MOV(ureg, face_temp, face_input);
 
    /* Use face_temp as face input from here on: */
-   t->inputs[t->inputMapping[FRAG_ATTRIB_FACE]] = ureg_src(face_temp);
+   t->inputs[t->inputMapping[VARYING_SLOT_FACE]] = ureg_src(face_temp);
 }
 
 static void
 emit_edgeflags(struct st_translate *t)
 {
    struct ureg_program *ureg = t->ureg;
-   struct ureg_dst edge_dst = t->outputs[t->outputMapping[VERT_RESULT_EDGE]];
+   struct ureg_dst edge_dst = t->outputs[t->outputMapping[VARYING_SLOT_EDGE]];
    struct ureg_src edge_src = t->inputs[t->inputMapping[VERT_ATTRIB_EDGEFLAG]];
 
    ureg_MOV(ureg, edge_dst, edge_src);
@@ -4686,14 +4925,14 @@ st_translate_program(
                                                         is_centroid[i]);
       }
 
-      if (proginfo->InputsRead & FRAG_BIT_WPOS) {
+      if (proginfo->InputsRead & VARYING_BIT_POS) {
          /* Must do this after setting up t->inputs, and before
           * emitting constant references, below:
           */
           emit_wpos(st_context(ctx), t, proginfo, ureg);
       }
 
-      if (proginfo->InputsRead & FRAG_BIT_FACE)
+      if (proginfo->InputsRead & VARYING_BIT_FACE)
          emit_face_var(t);
 
       /*
@@ -4718,6 +4957,15 @@ st_translate_program(
                                              TGSI_SEMANTIC_COLOR,
                                              outputSemanticIndex[i]);
             break;
+         case TGSI_SEMANTIC_SAMPLEMASK:
+            t->outputs[i] = ureg_DECL_output(ureg,
+                                             TGSI_SEMANTIC_SAMPLEMASK,
+                                             outputSemanticIndex[i]);
+            /* TODO: If we ever support more than 32 samples, this will have
+             * to become an array.
+             */
+            t->outputs[i] = ureg_writemask(t->outputs[i], TGSI_WRITEMASK_X);
+            break;
          default:
             assert(!"fragment shader outputs must be POSITION/STENCIL/COLOR");
             ret = PIPE_ERROR_BAD_INPUT;
@@ -4750,6 +4998,13 @@ st_translate_program(
          t->outputs[i] = ureg_DECL_output(ureg,
                                           outputSemanticName[i],
                                           outputSemanticIndex[i]);
+         if (outputSemanticName[i] == TGSI_SEMANTIC_FOG) {
+            /* force register to contain a fog coordinate in the form (F, 0, 0, 1). */
+            ureg_MOV(ureg,
+                     ureg_writemask(t->outputs[i], TGSI_WRITEMASK_YZW),
+                     ureg_imm4f(ureg, 0.0f, 0.0f, 0.0f, 1.0f));
+            t->outputs[i] = ureg_writemask(t->outputs[i], TGSI_WRITEMASK_X);
+	 }
       }
       if (passthrough_edgeflags)
          emit_edgeflags(t);
@@ -4758,8 +5013,10 @@ st_translate_program(
    /* Declare address register.
     */
    if (program->num_address_regs > 0) {
-      assert(program->num_address_regs == 1);
+      assert(program->num_address_regs <= 2);
       t->address[0] = ureg_DECL_address(ureg);
+      if (program->num_address_regs == 2)
+         t->address[1] = ureg_DECL_address(ureg);
    }
 
    /* Declare misc input registers
@@ -4796,16 +5053,9 @@ st_translate_program(
       }
    }
 
-   if (program->indirect_addr_temps) {
-      /* If temps are accessed with indirect addressing, declare temporaries
-       * in sequential order.  Else, we declare them on demand elsewhere.
-       * (Note: the number of temporaries is equal to program->next_temp)
-       */
-      for (i = 0; i < (unsigned)program->next_temp; i++) {
-         /* XXX use TGSI_FILE_TEMPORARY_ARRAY when it's supported by ureg */
-         t->temps[i] = ureg_DECL_local_temporary(t->ureg);
-      }
-   }
+   /* Copy over array sizes
+    */
+   memcpy(t->array_sizes, program->array_sizes, sizeof(unsigned) * program->next_array);
 
    /* Emit constants and uniforms.  TGSI uses a single index space for these, 
     * so we put all the translated regs in t->constants.
@@ -4820,8 +5070,6 @@ st_translate_program(
 
       for (i = 0; i < proginfo->Parameters->NumParameters; i++) {
          switch (proginfo->Parameters->Parameters[i].Type) {
-         case PROGRAM_ENV_PARAM:
-         case PROGRAM_LOCAL_PARAM:
          case PROGRAM_STATE_VAR:
          case PROGRAM_UNIFORM:
             t->constants[i] = ureg_DECL_constant(ureg, i);
@@ -4865,15 +5113,15 @@ st_translate_program(
       goto out;
    }
    i = 0;
-   foreach_iter(exec_list_iterator, iter, program->immediates) {
-      immediate_storage *imm = (immediate_storage *)iter.get();
+   foreach_list(node, &program->immediates) {
+      immediate_storage *imm = (immediate_storage *) node;
       assert(i < program->num_immediates);
       t->immediates[i++] = emit_immediate(t, imm->values, imm->type, imm->size);
    }
    assert(i == program->num_immediates);
 
    /* texture samplers */
-   for (i = 0; i < ctx->Const.MaxTextureImageUnits; i++) {
+   for (i = 0; i < ctx->Const.Program[MESA_SHADER_FRAGMENT].MaxTextureImageUnits; i++) {
       if (program->samplers_used & (1 << i)) {
          t->samplers[i] = ureg_DECL_sampler(ureg, i);
       }
@@ -4881,10 +5129,9 @@ st_translate_program(
 
    /* Emit each instruction in turn:
     */
-   foreach_iter(exec_list_iterator, iter, program->instructions) {
+   foreach_list(n, &program->instructions) {
       set_insn_start(t, ureg_get_instruction_number(ureg));
-      compile_tgsi_instruction(t, (glsl_to_tgsi_instruction *)iter.get(),
-                               clamp_color);
+      compile_tgsi_instruction(t, (glsl_to_tgsi_instruction *) n, clamp_color);
    }
 
    /* Fix up all emitted labels:
@@ -4899,7 +5146,7 @@ st_translate_program(
        * prog->ParameterValues to get reallocated (e.g., anything that adds a
        * program constant) has to happen before creating this linkage.
        */
-      for (unsigned i = 0; i < MESA_SHADER_TYPES; i++) {
+      for (unsigned i = 0; i < MESA_SHADER_STAGES; i++) {
          if (program->shader_program->_LinkedShaders[i] == NULL)
             continue;
 
@@ -4926,6 +5173,26 @@ out:
 }
 /* ----------------------------- End TGSI code ------------------------------ */
 
+
+static unsigned
+shader_stage_to_ptarget(gl_shader_stage stage)
+{
+   switch (stage) {
+   case MESA_SHADER_VERTEX:
+      return PIPE_SHADER_VERTEX;
+   case MESA_SHADER_FRAGMENT:
+      return PIPE_SHADER_FRAGMENT;
+   case MESA_SHADER_GEOMETRY:
+      return PIPE_SHADER_GEOMETRY;
+   case MESA_SHADER_COMPUTE:
+      return PIPE_SHADER_COMPUTE;
+   }
+
+   assert(!"should not be reached");
+   return PIPE_SHADER_VERTEX;
+}
+
+
 /**
  * Convert a shader's GLSL IR into a Mesa gl_program, although without 
  * generating Mesa IR.
@@ -4937,29 +5204,12 @@ get_mesa_program(struct gl_context *ctx,
 {
    glsl_to_tgsi_visitor* v;
    struct gl_program *prog;
-   GLenum target;
-   const char *target_string;
+   GLenum target = _mesa_shader_stage_to_program(shader->Stage);
    bool progress;
    struct gl_shader_compiler_options *options =
-         &ctx->ShaderCompilerOptions[_mesa_shader_type_to_index(shader->Type)];
-
-   switch (shader->Type) {
-   case GL_VERTEX_SHADER:
-      target = GL_VERTEX_PROGRAM_ARB;
-      target_string = "vertex";
-      break;
-   case GL_FRAGMENT_SHADER:
-      target = GL_FRAGMENT_PROGRAM_ARB;
-      target_string = "fragment";
-      break;
-   case GL_GEOMETRY_SHADER:
-      target = GL_GEOMETRY_PROGRAM_NV;
-      target_string = "geometry";
-      break;
-   default:
-      assert(!"should not be reached");
-      return NULL;
-   }
+         &ctx->ShaderCompilerOptions[_mesa_shader_enum_to_shader_stage(shader->Type)];
+   struct pipe_screen *pscreen = ctx->st->pipe->screen;
+   unsigned ptarget = shader_stage_to_ptarget(shader->Stage);
 
    validate_ir_tree(shader->ir);
 
@@ -4975,6 +5225,9 @@ get_mesa_program(struct gl_context *ctx,
    v->glsl_version = ctx->Const.GLSLVersion;
    v->native_integers = ctx->Const.NativeIntegers;
 
+   v->have_sqrt = pscreen->get_shader_param(pscreen, ptarget,
+                                            PIPE_SHADER_CAP_TGSI_SQRT_SUPPORTED);
+
    _mesa_generate_parameters_list_for_uniforms(shader_program, shader,
 					       prog->Parameters);
 
@@ -4988,8 +5241,8 @@ get_mesa_program(struct gl_context *ctx,
    do {
       progress = GL_FALSE;
 
-      foreach_iter(exec_list_iterator, iter, v->function_signatures) {
-         function_entry *entry = (function_entry *)iter.get();
+      foreach_list(node, &v->function_signatures) {
+         function_entry *entry = (function_entry *) node;
 
          if (!entry->bgn_inst) {
             v->current_function = entry;
@@ -5030,27 +5283,20 @@ get_mesa_program(struct gl_context *ctx,
    /* Perform optimizations on the instructions in the glsl_to_tgsi_visitor. */
    v->simplify_cmp();
    v->copy_propagate();
-   while (v->eliminate_dead_code_advanced());
+   while (v->eliminate_dead_code());
 
-   /* FIXME: These passes to optimize temporary registers don't work when there
-    * is indirect addressing of the temporary register space.  We need proper 
-    * array support so that we don't have to give up these passes in every 
-    * shader that uses arrays.
-    */
-   if (!v->indirect_addr_temps) {
-      v->eliminate_dead_code();
-      v->merge_registers();
-      v->renumber_registers();
-   }
+   v->merge_registers();
+   v->renumber_registers();
    
    /* Write the END instruction. */
    v->emit(NULL, TGSI_OPCODE_END);
 
-   if (ctx->Shader.Flags & GLSL_DUMP) {
+   if (ctx->_Shader->Flags & GLSL_DUMP) {
       printf("\n");
-      printf("GLSL IR for linked %s program %d:\n", target_string,
+      printf("GLSL IR for linked %s program %d:\n",
+             _mesa_shader_stage_to_string(shader->Stage),
              shader_program->Name);
-      _mesa_print_ir(shader->ir, NULL);
+      _mesa_print_ir(stdout, shader->ir, NULL);
       printf("\n");
       printf("\n");
       fflush(stdout);
@@ -5059,7 +5305,7 @@ get_mesa_program(struct gl_context *ctx,
    prog->Instructions = NULL;
    prog->NumInstructions = 0;
 
-   do_set_program_inouts(shader->ir, prog, shader->Type == GL_FRAGMENT_SHADER);
+   do_set_program_inouts(shader->ir, prog, shader->Stage);
    count_resources(v, prog);
 
    _mesa_reference_program(ctx, &shader->Program, prog);
@@ -5089,6 +5335,10 @@ get_mesa_program(struct gl_context *ctx,
    case GL_GEOMETRY_SHADER:
       stgp = (struct st_geometry_program *)prog;
       stgp->glsl_to_tgsi = v;
+      stgp->Base.InputType = shader_program->Geom.InputType;
+      stgp->Base.OutputType = shader_program->Geom.OutputType;
+      stgp->Base.VerticesOut = shader_program->Geom.VerticesOut;
+      stgp->Base.Invocations = shader_program->Geom.Invocations;
       break;
    default:
       assert(!"should not be reached");
@@ -5109,6 +5359,7 @@ st_new_shader(struct gl_context *ctx, GLuint name, GLuint type)
    shader = rzalloc(NULL, struct gl_shader);
    if (shader) {
       shader->Type = type;
+      shader->Stage = _mesa_shader_enum_to_shader_stage(type);
       shader->Name = name;
       _mesa_init_shader(ctx, shader);
    }
@@ -5136,70 +5387,85 @@ st_new_shader_program(struct gl_context *ctx, GLuint name)
 GLboolean
 st_link_shader(struct gl_context *ctx, struct gl_shader_program *prog)
 {
+   struct pipe_screen *pscreen = ctx->st->pipe->screen;
    assert(prog->LinkStatus);
 
-   for (unsigned i = 0; i < MESA_SHADER_TYPES; i++) {
+   for (unsigned i = 0; i < MESA_SHADER_STAGES; i++) {
       if (prog->_LinkedShaders[i] == NULL)
          continue;
 
       bool progress;
       exec_list *ir = prog->_LinkedShaders[i]->ir;
       const struct gl_shader_compiler_options *options =
-            &ctx->ShaderCompilerOptions[_mesa_shader_type_to_index(prog->_LinkedShaders[i]->Type)];
+            &ctx->ShaderCompilerOptions[_mesa_shader_enum_to_shader_stage(prog->_LinkedShaders[i]->Type)];
 
-      do {
-         unsigned what_to_lower = MOD_TO_FRACT | DIV_TO_MUL_RCP |
-            EXP_TO_EXP2 | LOG_TO_LOG2;
-         if (options->EmitNoPow)
-            what_to_lower |= POW_TO_EXP2;
-         if (!ctx->Const.NativeIntegers)
-            what_to_lower |= INT_DIV_TO_MUL_RCP;
+      /* If there are forms of indirect addressing that the driver
+       * cannot handle, perform the lowering pass.
+       */
+      if (options->EmitNoIndirectInput || options->EmitNoIndirectOutput ||
+          options->EmitNoIndirectTemp || options->EmitNoIndirectUniform) {
+         lower_variable_index_to_cond_assign(ir,
+                                             options->EmitNoIndirectInput,
+                                             options->EmitNoIndirectOutput,
+                                             options->EmitNoIndirectTemp,
+                                             options->EmitNoIndirectUniform);
+      }
 
-         progress = false;
+      if (ctx->Extensions.ARB_shading_language_packing) {
+         unsigned lower_inst = LOWER_PACK_SNORM_2x16 |
+                               LOWER_UNPACK_SNORM_2x16 |
+                               LOWER_PACK_UNORM_2x16 |
+                               LOWER_UNPACK_UNORM_2x16 |
+                               LOWER_PACK_SNORM_4x8 |
+                               LOWER_UNPACK_SNORM_4x8 |
+                               LOWER_UNPACK_UNORM_4x8 |
+                               LOWER_PACK_UNORM_4x8 |
+                               LOWER_PACK_HALF_2x16 |
+                               LOWER_UNPACK_HALF_2x16;
+
+         lower_packing_builtins(ir, lower_inst);
+      }
 
-         /* Lowering */
-         do_mat_op_to_vec(ir);
-         lower_instructions(ir, what_to_lower);
+      if (!pscreen->get_param(pscreen, PIPE_CAP_TEXTURE_GATHER_OFFSETS))
+         lower_offset_arrays(ir);
+      do_mat_op_to_vec(ir);
+      lower_instructions(ir,
+                         MOD_TO_FRACT |
+                         DIV_TO_MUL_RCP |
+                         EXP_TO_EXP2 |
+                         LOG_TO_LOG2 |
+                         LDEXP_TO_ARITH |
+                         CARRY_TO_ARITH |
+                         BORROW_TO_ARITH |
+                         (options->EmitNoPow ? POW_TO_EXP2 : 0) |
+                         (!ctx->Const.NativeIntegers ? INT_DIV_TO_MUL_RCP : 0));
+
+      lower_ubo_reference(prog->_LinkedShaders[i], ir);
+      do_vec_index_to_cond_assign(ir);
+      lower_vector_insert(ir, true);
+      lower_quadop_vector(ir, false);
+      lower_noise(ir);
+      if (options->MaxIfDepth == 0) {
+         lower_discard(ir);
+      }
 
-         lower_ubo_reference(prog->_LinkedShaders[i], ir);
+      do {
+         progress = false;
 
          progress = do_lower_jumps(ir, true, true, options->EmitNoMainReturn, options->EmitNoCont, options->EmitNoLoops) || progress;
 
-         progress = do_common_optimization(ir, true, true,
-					   options->MaxUnrollIterations)
+         progress = do_common_optimization(ir, true, true, options,
+                                           ctx->Const.NativeIntegers)
 	   || progress;
 
-         progress = lower_quadop_vector(ir, false) || progress;
-
-         if (options->MaxIfDepth == 0)
-            progress = lower_discard(ir) || progress;
-
          progress = lower_if_to_cond_assign(ir, options->MaxIfDepth) || progress;
 
-         if (options->EmitNoNoise)
-            progress = lower_noise(ir) || progress;
-
-         /* If there are forms of indirect addressing that the driver
-          * cannot handle, perform the lowering pass.
-          */
-         if (options->EmitNoIndirectInput || options->EmitNoIndirectOutput
-             || options->EmitNoIndirectTemp || options->EmitNoIndirectUniform)
-           progress =
-             lower_variable_index_to_cond_assign(ir,
-        					 options->EmitNoIndirectInput,
-        					 options->EmitNoIndirectOutput,
-        					 options->EmitNoIndirectTemp,
-        					 options->EmitNoIndirectUniform)
-             || progress;
-
-         progress = do_vec_index_to_cond_assign(ir) || progress;
-
       } while (progress);
 
       validate_ir_tree(ir);
    }
 
-   for (unsigned i = 0; i < MESA_SHADER_TYPES; i++) {
+   for (unsigned i = 0; i < MESA_SHADER_STAGES; i++) {
       struct gl_program *linked_prog;
 
       if (prog->_LinkedShaders[i] == NULL)
@@ -5208,15 +5474,11 @@ st_link_shader(struct gl_context *ctx, struct gl_shader_program *prog)
       linked_prog = get_mesa_program(ctx, prog, prog->_LinkedShaders[i]);
 
       if (linked_prog) {
-	 static const GLenum targets[] = {
-	    GL_VERTEX_PROGRAM_ARB,
-	    GL_FRAGMENT_PROGRAM_ARB,
-	    GL_GEOMETRY_PROGRAM_NV
-	 };
-
 	 _mesa_reference_program(ctx, &prog->_LinkedShaders[i]->Program,
 				 linked_prog);
-         if (!ctx->Driver.ProgramStringNotify(ctx, targets[i], linked_prog)) {
+         if (!ctx->Driver.ProgramStringNotify(ctx,
+                                              _mesa_shader_stage_to_program(i),
+                                              linked_prog)) {
 	    _mesa_reference_program(ctx, &prog->_LinkedShaders[i]->Program,
 				    NULL);
             _mesa_reference_program(ctx, &linked_prog, NULL);