[mesa.git] / src / mesa / state_tracker / st_glsl_to_tgsi.cpp

/*
 * Copyright (C) 2005-2007  Brian Paul   All Rights Reserved.
 * Copyright (C) 2008  VMware, Inc.   All Rights Reserved.
 * Copyright © 2010 Intel Corporation
 * Copyright © 2011 Bryan Cain
 *
 * Permission is hereby granted, free of charge, to any person obtaining a
 * copy of this software and associated documentation files (the "Software"),
 * to deal in the Software without restriction, including without limitation
 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
 * and/or sell copies of the Software, and to permit persons to whom the
 * Software is furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice (including the next
 * paragraph) shall be included in all copies or substantial portions of the
 * Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
 * DEALINGS IN THE SOFTWARE.
 */

/**
 * \file glsl_to_tgsi.cpp
 *
 * Translate GLSL IR to TGSI.
 */

#include <stdio.h>
#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"
#include "../glsl/program.h"
#include "ir_optimization.h"
#include "ast.h"

#include "main/mtypes.h"
#include "main/shaderobj.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"
#include "program/program.h"
#include "program/prog_parameter.h"
#include "program/sampler.h"

#include "pipe/p_compiler.h"
#include "pipe/p_context.h"
#include "pipe/p_screen.h"
#include "pipe/p_shader_tokens.h"
#include "pipe/p_state.h"
#include "util/u_math.h"
#include "tgsi/tgsi_ureg.h"
#include "tgsi/tgsi_info.h"
#include "st_context.h"
#include "st_program.h"
#include "st_glsl_to_tgsi.h"
#include "st_mesa_to_tgsi.h"
}

#define PROGRAM_IMMEDIATE PROGRAM_FILE_MAX
#define PROGRAM_ANY_CONST ((1 << PROGRAM_LOCAL_PARAM) |  \
                           (1 << PROGRAM_ENV_PARAM) |    \
                           (1 << PROGRAM_STATE_VAR) |    \
                           (1 << PROGRAM_CONSTANT) |     \
                           (1 << PROGRAM_UNIFORM))

/**
 * Maximum number of temporary registers.
 *
 * It is too big for stack allocated arrays -- it will cause stack overflow on
 * Windows and likely Mac OS X.
 */
#define MAX_TEMPS         4096

/**
 * Maximum number of arrays
 */
#define MAX_ARRAYS        256

/* will be 4 for GLSL 4.00 */
#define MAX_GLSL_TEXTURE_OFFSET 1

class st_src_reg;
class st_dst_reg;

static int swizzle_for_size(int size);

/**
 * This struct is a corresponding struct to TGSI ureg_src.
 */
class st_src_reg {
public:
   st_src_reg(gl_register_file file, int index, const glsl_type *type)
   {
      this->file = file;
      this->index = index;
      if (type && (type->is_scalar() || type->is_vector() || type->is_matrix()))
         this->swizzle = swizzle_for_size(type->vector_elements);
      else
         this->swizzle = SWIZZLE_XYZW;
      this->negate = 0;
      this->index2D = 0;
      this->type = type ? type->base_type : GLSL_TYPE_ERROR;
      this->reladdr = NULL;
   }

   st_src_reg(gl_register_file file, int index, int type)
   {
      this->type = type;
      this->file = file;
      this->index = index;
      this->index2D = 0;
      this->swizzle = SWIZZLE_XYZW;
      this->negate = 0;
      this->reladdr = NULL;
   }

   st_src_reg(gl_register_file file, int index, int type, int index2D)
   {
      this->type = type;
      this->file = file;
      this->index = index;
      this->index2D = index2D;
      this->swizzle = SWIZZLE_XYZW;
      this->negate = 0;
      this->reladdr = NULL;
   }

   st_src_reg()
   {
      this->type = GLSL_TYPE_ERROR;
      this->file = PROGRAM_UNDEFINED;
      this->index = 0;
      this->index2D = 0;
      this->swizzle = 0;
      this->negate = 0;
      this->reladdr = NULL;
   }

   explicit st_src_reg(st_dst_reg reg);

   gl_register_file file; /**< PROGRAM_* from Mesa */
   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;
};

class st_dst_reg {
public:
   st_dst_reg(gl_register_file file, int writemask, int type)
   {
      this->file = file;
      this->index = 0;
      this->writemask = writemask;
      this->cond_mask = COND_TR;
      this->reladdr = NULL;
      this->type = type;
   }

   st_dst_reg()
   {
      this->type = GLSL_TYPE_ERROR;
      this->file = PROGRAM_UNDEFINED;
      this->index = 0;
      this->writemask = 0;
      this->cond_mask = COND_TR;
      this->reladdr = NULL;
   }

   explicit st_dst_reg(st_src_reg reg);

   gl_register_file file; /**< PROGRAM_* from Mesa */
   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) */
   /** Register index should be offset by the integer in this reg. */
   st_src_reg *reladdr;
};

st_src_reg::st_src_reg(st_dst_reg reg)
{
   this->type = reg.type;
   this->file = reg.file;
   this->index = reg.index;
   this->swizzle = SWIZZLE_XYZW;
   this->negate = 0;
   this->reladdr = reg.reladdr;
   this->index2D = 0;
}

st_dst_reg::st_dst_reg(st_src_reg reg)
{
   this->type = reg.type;
   this->file = reg.file;
   this->index = reg.index;
   this->writemask = WRITEMASK_XYZW;
   this->cond_mask = COND_TR;
   this->reladdr = reg.reladdr;
}

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;
   }

   unsigned op;
   st_dst_reg dst;
   st_src_reg src[3];
   /** Pointer to the ir source this tree came from for debugging */
   ir_instruction *ir;
   GLboolean cond_update;
   bool saturate;
   int sampler; /**< sampler index */
   int tex_target; /**< One of TEXTURE_*_INDEX */
   GLboolean tex_shadow;
   struct tgsi_texture_offset tex_offsets[MAX_GLSL_TEXTURE_OFFSET];
   unsigned tex_offset_num_offset;
   int dead_mask; /**< Used in dead code elimination */

   class function_entry *function; /* Set on TGSI_OPCODE_CAL or TGSI_OPCODE_BGNSUB */
};

class variable_storage : public exec_node {
public:
   variable_storage(ir_variable *var, gl_register_file file, int index)
      : file(file), index(index), var(var)
   {
      /* empty */
   }

   gl_register_file file;
   int index;
   ir_variable *var; /* variable that maps to this, if any */
};

class immediate_storage : public exec_node {
public:
   immediate_storage(gl_constant_value *values, int size, int type)
   {
      memcpy(this->values, values, size * sizeof(gl_constant_value));
      this->size = size;
      this->type = type;
   }
   
   gl_constant_value values[4];
   int size; /**< Number of components (1-4) */
   int type; /**< GL_FLOAT, GL_INT, GL_BOOL, or GL_UNSIGNED_INT */
};

class function_entry : public exec_node {
public:
   ir_function_signature *sig;

   /**
    * identifier of this function signature used by the program.
    *
    * At the point that TGSI instructions for function calls are
    * generated, we don't know the address of the first instruction of
    * the function body.  So we make the BranchTarget that is called a
    * small integer and rewrite them during set_branchtargets().
    */
   int sig_id;

   /**
    * Pointer to first instruction of the function body.
    *
    * Set during function body emits after main() is processed.
    */
   glsl_to_tgsi_instruction *bgn_inst;

   /**
    * Index of the first instruction of the function body in actual TGSI.
    *
    * Set after conversion from glsl_to_tgsi_instruction to TGSI.
    */
   int inst;

   /** Storage for the return value. */
   st_src_reg return_reg;
};

struct glsl_to_tgsi_visitor : public ir_visitor {
public:
   glsl_to_tgsi_visitor();
   ~glsl_to_tgsi_visitor();

   function_entry *current_function;

   struct gl_context *ctx;
   struct gl_program *prog;
   struct gl_shader_program *shader_program;
   struct gl_shader_compiler_options *options;

   int next_temp;

   unsigned array_sizes[MAX_ARRAYS];
   unsigned next_array;

   int num_address_regs;
   int samplers_used;
   bool indirect_addr_consts;
   
   int glsl_version;
   bool native_integers;
   bool have_sqrt;

   variable_storage *find_variable_storage(ir_variable *var);

   int add_constant(gl_register_file file, gl_constant_value values[4],
                    int size, int datatype, GLuint *swizzle_out);

   function_entry *get_function_signature(ir_function_signature *sig);

   st_src_reg get_temp(const glsl_type *type);
   void reladdr_to_temp(ir_instruction *ir, st_src_reg *reg, int *num_reladdr);

   st_src_reg st_src_reg_for_float(float val);
   st_src_reg st_src_reg_for_int(int val);
   st_src_reg st_src_reg_for_type(int type, int val);

   /**
    * \name Visit methods
    *
    * As typical for the visitor pattern, there must be one \c visit method for
    * each concrete subclass of \c ir_instruction.  Virtual base classes within
    * the hierarchy should not have \c visit methods.
    */
   /*@{*/
   virtual void visit(ir_variable *);
   virtual void visit(ir_loop *);
   virtual void visit(ir_loop_jump *);
   virtual void visit(ir_function_signature *);
   virtual void visit(ir_function *);
   virtual void visit(ir_expression *);
   virtual void visit(ir_swizzle *);
   virtual void visit(ir_dereference_variable  *);
   virtual void visit(ir_dereference_array *);
   virtual void visit(ir_dereference_record *);
   virtual void visit(ir_assignment *);
   virtual void visit(ir_constant *);
   virtual void visit(ir_call *);
   virtual void visit(ir_return *);
   virtual void visit(ir_discard *);
   virtual void visit(ir_texture *);
   virtual void visit(ir_if *);
   /*@}*/

   st_src_reg result;

   /** List of variable_storage */
   exec_list variables;

   /** List of immediate_storage */
   exec_list immediates;
   unsigned num_immediates;

   /** List of function_entry */
   exec_list function_signatures;
   int next_signature_id;

   /** List of glsl_to_tgsi_instruction */
   exec_list instructions;

   glsl_to_tgsi_instruction *emit(ir_instruction *ir, unsigned op);

   glsl_to_tgsi_instruction *emit(ir_instruction *ir, unsigned op,
                                st_dst_reg dst, st_src_reg src0);

   glsl_to_tgsi_instruction *emit(ir_instruction *ir, unsigned op,
                                st_dst_reg dst, st_src_reg src0, st_src_reg src1);

   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);
   
   unsigned get_opcode(ir_instruction *ir, unsigned op,
                    st_dst_reg dst,
                    st_src_reg src0, st_src_reg src1);

   /**
    * Emit the correct dot-product instruction for the type of arguments
    */
   glsl_to_tgsi_instruction *emit_dp(ir_instruction *ir,
                                     st_dst_reg dst,
                                     st_src_reg src0,
                                     st_src_reg src1,
                                     unsigned elements);

   void emit_scalar(ir_instruction *ir, unsigned op,
                    st_dst_reg dst, st_src_reg src0);

   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,
                 st_dst_reg dst, const st_src_reg &src);

   bool try_emit_mad(ir_expression *ir,
              int mul_operand);
   bool try_emit_mad_for_and_not(ir_expression *ir,
              int mul_operand);
   bool try_emit_sat(ir_expression *ir);

   void emit_swz(ir_expression *ir);

   bool process_move_condition(ir_rvalue *ir);

   void simplify_cmp(void);

   void rename_temp_register(int index, int new_index);
   int get_first_temp_read(int index);
   int get_first_temp_write(int index);
   int get_last_temp_read(int index);
   int get_last_temp_write(int index);

   void copy_propagate(void);
   void eliminate_dead_code(void);
   int eliminate_dead_code_advanced(void);
   void merge_registers(void);
   void renumber_registers(void);

   void emit_block_mov(ir_assignment *ir, const struct glsl_type *type,
                       st_dst_reg *l, st_src_reg *r);

   void *mem_ctx;
};

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 void
fail_link(struct gl_shader_program *prog, const char *fmt, ...) PRINTFLIKE(2, 3);

static void
fail_link(struct gl_shader_program *prog, const char *fmt, ...)
{
   va_list args;
   va_start(args, fmt);
   ralloc_vasprintf_append(&prog->InfoLog, fmt, args);
   va_end(args);

   prog->LinkStatus = GL_FALSE;
}

static int
swizzle_for_size(int size)
{
   int size_swizzles[4] = {
      MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X),
      MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y),
      MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_Z),
      MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_W),
   };

   assert((size >= 1) && (size <= 4));
   return size_swizzles[size - 1];
}

static bool
is_tex_instruction(unsigned opcode)
{
   const tgsi_opcode_info* info = tgsi_get_opcode_info(opcode);
   return info->is_tex;
}

static unsigned
num_inst_dst_regs(unsigned opcode)
{
   const tgsi_opcode_info* info = tgsi_get_opcode_info(opcode);
   return info->num_dst;
}

static unsigned
num_inst_src_regs(unsigned opcode)
{
   const tgsi_opcode_info* info = tgsi_get_opcode_info(opcode);
   return info->is_tex ? info->num_src - 1 : info->num_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, st_src_reg src2)
{
   glsl_to_tgsi_instruction *inst = new(mem_ctx) glsl_to_tgsi_instruction();
   int num_reladdr = 0, i;
   
   op = get_opcode(ir, op, dst, src0, src1);

   /* If we have to do relative addressing, we want to load the ARL
    * reg directly for one of the regs, and preload the other reladdr
    * sources into temps.
    */
   num_reladdr += dst.reladdr != NULL;
   num_reladdr += src0.reladdr != NULL;
   num_reladdr += src1.reladdr != NULL;
   num_reladdr += src2.reladdr != NULL;

   reladdr_to_temp(ir, &src2, &num_reladdr);
   reladdr_to_temp(ir, &src1, &num_reladdr);
   reladdr_to_temp(ir, &src0, &num_reladdr);

   if (dst.reladdr) {
      emit_arl(ir, address_reg, *dst.reladdr);
      num_reladdr--;
   }
   assert(num_reladdr == 0);

   inst->op = op;
   inst->dst = dst;
   inst->src[0] = src0;
   inst->src[1] = src1;
   inst->src[2] = src2;
   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_LOCAL_PARAM:
      case PROGRAM_ENV_PARAM:
      case PROGRAM_STATE_VAR:
      case PROGRAM_CONSTANT:
      case PROGRAM_UNIFORM:
         this->indirect_addr_consts = true;
         break;
      case PROGRAM_IMMEDIATE:
         assert(!"immediates should not have indirect addressing");
         break;
      default:
         break;
      }
   }
   else {
      for (i=0; i<3; i++) {
         if(inst->src[i].reladdr) {
            switch(inst->src[i].file) {
            case PROGRAM_LOCAL_PARAM:
            case PROGRAM_ENV_PARAM:
            case PROGRAM_STATE_VAR:
            case PROGRAM_CONSTANT:
            case PROGRAM_UNIFORM:
               this->indirect_addr_consts = true;
               break;
            case PROGRAM_IMMEDIATE:
               assert(!"immediates should not have indirect addressing");
               break;
            default:
               break;
            }
         }
      }
   }

   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)
{
   return emit(ir, op, dst, src0, src1, undef_src);
}

glsl_to_tgsi_instruction *
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);
}

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);
   }
}

/**
 * Determines whether to use an integer, unsigned integer, or float opcode 
 * based on the operands and input opcode, then emits the result.
 */
unsigned
glsl_to_tgsi_visitor::get_opcode(ir_instruction *ir, unsigned op,
                         st_dst_reg dst,
                         st_src_reg src0, st_src_reg src1)
{
   int type = GLSL_TYPE_FLOAT;
   
   assert(src0.type != GLSL_TYPE_ARRAY);
   assert(src0.type != GLSL_TYPE_STRUCT);
   assert(src1.type != GLSL_TYPE_ARRAY);
   assert(src1.type != GLSL_TYPE_STRUCT);

   if (src0.type == GLSL_TYPE_FLOAT || src1.type == GLSL_TYPE_FLOAT)
      type = GLSL_TYPE_FLOAT;
   else if (native_integers)
      type = src0.type == GLSL_TYPE_BOOL ? GLSL_TYPE_INT : src0.type;

#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; \
      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)
   
   switch(op) {
      case2fi(ADD, UADD);
      case2fi(MUL, UMUL);
      case2fi(MAD, UMAD);
      case3(DIV, IDIV, UDIV);
      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);
      
      case2iu(ISHR, USHR);

      case2fi(SSG, ISSG);
      case3(ABS, IABS, IABS);
      
      default: break;
   }
   
   assert(op != TGSI_OPCODE_LAST);
   return op;
}

glsl_to_tgsi_instruction *
glsl_to_tgsi_visitor::emit_dp(ir_instruction *ir,
                            st_dst_reg dst, st_src_reg src0, st_src_reg src1,
                            unsigned elements)
{
   static const unsigned dot_opcodes[] = {
      TGSI_OPCODE_DP2, TGSI_OPCODE_DP3, TGSI_OPCODE_DP4
   };

   return emit(ir, dot_opcodes[elements - 2], dst, src0, src1);
}

/**
 * Emits TGSI scalar opcodes to produce unique answers across channels.
 *
 * Some TGSI opcodes are scalar-only, like ARB_fp/vp.  The src X
 * channel determines the result across all channels.  So to do a vec4
 * of this operation, we want to emit a scalar per source channel used
 * to produce dest channels.
 */
void
glsl_to_tgsi_visitor::emit_scalar(ir_instruction *ir, unsigned op,
                                st_dst_reg dst,
                                st_src_reg orig_src0, st_src_reg orig_src1)
{
   int i, j;
   int done_mask = ~dst.writemask;

   /* TGSI RCP is a scalar operation splatting results to all channels,
    * like ARB_fp/vp.  So emit as many RCPs as necessary to cover our
    * dst channels.
    */
   for (i = 0; i < 4; i++) {
      GLuint this_mask = (1 << i);
      glsl_to_tgsi_instruction *inst;
      st_src_reg src0 = orig_src0;
      st_src_reg src1 = orig_src1;

      if (done_mask & this_mask)
         continue;

      GLuint src0_swiz = GET_SWZ(src0.swizzle, i);
      GLuint src1_swiz = GET_SWZ(src1.swizzle, i);
      for (j = i + 1; j < 4; j++) {
         /* If there is another enabled component in the destination that is
          * derived from the same inputs, generate its value on this pass as
          * well.
          */
         if (!(done_mask & (1 << j)) &&
             GET_SWZ(src0.swizzle, j) == src0_swiz &&
             GET_SWZ(src1.swizzle, j) == src1_swiz) {
            this_mask |= (1 << j);
         }
      }
      src0.swizzle = MAKE_SWIZZLE4(src0_swiz, src0_swiz,
                                   src0_swiz, src0_swiz);
      src1.swizzle = MAKE_SWIZZLE4(src1_swiz, src1_swiz,
                                  src1_swiz, src1_swiz);

      inst = emit(ir, op, dst, src0, src1);
      inst->dst.writemask = this_mask;
      done_mask |= this_mask;
   }
}

void
glsl_to_tgsi_visitor::emit_scalar(ir_instruction *ir, unsigned op,
                                st_dst_reg dst, st_src_reg src0)
{
   st_src_reg undef = undef_src;

   undef.swizzle = SWIZZLE_XXXX;

   emit_scalar(ir, op, dst, src0, undef);
}

void
glsl_to_tgsi_visitor::emit_arl(ir_instruction *ir,
                                st_dst_reg dst, st_src_reg src0)
{
   int op = TGSI_OPCODE_ARL;

   if (src0.type == GLSL_TYPE_INT || src0.type == GLSL_TYPE_UINT)
      op = TGSI_OPCODE_UARL;

   emit(NULL, op, dst, src0);
}

/**
 * Emit an TGSI_OPCODE_SCS instruction
 *
 * The \c SCS opcode functions a bit differently than the other TGSI opcodes.
 * Instead of splatting its result across all four components of the 
 * destination, it writes one value to the \c x component and another value to 
 * the \c y component.
 *
 * \param ir        IR instruction being processed
 * \param op        Either \c TGSI_OPCODE_SIN or \c TGSI_OPCODE_COS depending 
 *                  on which value is desired.
 * \param dst       Destination register
 * \param src       Source register
 */
void
glsl_to_tgsi_visitor::emit_scs(ir_instruction *ir, unsigned op,
                             st_dst_reg dst,
                             const st_src_reg &src)
{
   /* Vertex programs cannot use the SCS opcode.
    */
   if (this->prog->Target == GL_VERTEX_PROGRAM_ARB) {
      emit_scalar(ir, op, dst, src);
      return;
   }

   const unsigned component = (op == TGSI_OPCODE_SIN) ? 0 : 1;
   const unsigned scs_mask = (1U << component);
   int done_mask = ~dst.writemask;
   st_src_reg tmp;

   assert(op == TGSI_OPCODE_SIN || op == TGSI_OPCODE_COS);

   /* If there are compnents in the destination that differ from the component
    * that will be written by the SCS instrution, we'll need a temporary.
    */
   if (scs_mask != unsigned(dst.writemask)) {
      tmp = get_temp(glsl_type::vec4_type);
   }

   for (unsigned i = 0; i < 4; i++) {
      unsigned this_mask = (1U << i);
      st_src_reg src0 = src;

      if ((done_mask & this_mask) != 0)
         continue;

      /* The source swizzle specified which component of the source generates
       * sine / cosine for the current component in the destination.  The SCS
       * instruction requires that this value be swizzle to the X component.
       * Replace the current swizzle with a swizzle that puts the source in
       * the X component.
       */
      unsigned src0_swiz = GET_SWZ(src.swizzle, i);

      src0.swizzle = MAKE_SWIZZLE4(src0_swiz, src0_swiz,
                                   src0_swiz, src0_swiz);
      for (unsigned j = i + 1; j < 4; j++) {
         /* If there is another enabled component in the destination that is
          * derived from the same inputs, generate its value on this pass as
          * well.
          */
         if (!(done_mask & (1 << j)) &&
             GET_SWZ(src0.swizzle, j) == src0_swiz) {
            this_mask |= (1 << j);
         }
      }

      if (this_mask != scs_mask) {
         glsl_to_tgsi_instruction *inst;
         st_dst_reg tmp_dst = st_dst_reg(tmp);

         /* Emit the SCS instruction.
          */
         inst = emit(ir, TGSI_OPCODE_SCS, tmp_dst, src0);
         inst->dst.writemask = scs_mask;

         /* Move the result of the SCS instruction to the desired location in
          * the destination.
          */
         tmp.swizzle = MAKE_SWIZZLE4(component, component,
                                     component, component);
         inst = emit(ir, TGSI_OPCODE_SCS, dst, tmp);
         inst->dst.writemask = this_mask;
      } else {
         /* Emit the SCS instruction to write directly to the destination.
          */
         glsl_to_tgsi_instruction *inst = emit(ir, TGSI_OPCODE_SCS, dst, src0);
         inst->dst.writemask = scs_mask;
      }

      done_mask |= this_mask;
   }
}

int
glsl_to_tgsi_visitor::add_constant(gl_register_file file,
                             gl_constant_value values[4], int size, int datatype,
                             GLuint *swizzle_out)
{
   if (file == PROGRAM_CONSTANT) {
      return _mesa_add_typed_unnamed_constant(this->prog->Parameters, values,
                                              size, datatype, swizzle_out);
   } else {
      int index = 0;
      immediate_storage *entry;
      assert(file == PROGRAM_IMMEDIATE);

      /* 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();
         
         if (entry->size == size &&
             entry->type == datatype &&
             !memcmp(entry->values, values, size * sizeof(gl_constant_value))) {
             return index;
         }
         index++;
      }
      
      /* Add this immediate to the list. */
      entry = new(mem_ctx) immediate_storage(values, size, datatype);
      this->immediates.push_tail(entry);
      this->num_immediates++;
      return index;
   }
}

st_src_reg
glsl_to_tgsi_visitor::st_src_reg_for_float(float val)
{
   st_src_reg src(PROGRAM_IMMEDIATE, -1, GLSL_TYPE_FLOAT);
   union gl_constant_value uval;

   uval.f = val;
   src.index = add_constant(src.file, &uval, 1, GL_FLOAT, &src.swizzle);

   return src;
}

st_src_reg
glsl_to_tgsi_visitor::st_src_reg_for_int(int val)
{
   st_src_reg src(PROGRAM_IMMEDIATE, -1, GLSL_TYPE_INT);
   union gl_constant_value uval;
   
   assert(native_integers);

   uval.i = val;
   src.index = add_constant(src.file, &uval, 1, GL_INT, &src.swizzle);

   return src;
}

st_src_reg
glsl_to_tgsi_visitor::st_src_reg_for_type(int type, int val)
{
   if (native_integers)
      return type == GLSL_TYPE_FLOAT ? st_src_reg_for_float(val) : 
                                       st_src_reg_for_int(val);
   else
      return st_src_reg_for_float(val);
}

static int
type_size(const struct glsl_type *type)
{
   unsigned int i;
   int size;

   switch (type->base_type) {
   case GLSL_TYPE_UINT:
   case GLSL_TYPE_INT:
   case GLSL_TYPE_FLOAT:
   case GLSL_TYPE_BOOL:
      if (type->is_matrix()) {
         return type->matrix_columns;
      } else {
         /* Regardless of size of vector, it gets a vec4. This is bad
          * packing for things like floats, but otherwise arrays become a
          * mess.  Hopefully a later pass over the code can pack scalars
          * down if appropriate.
          */
         return 1;
      }
   case GLSL_TYPE_ARRAY:
      assert(type->length > 0);
      return type_size(type->fields.array) * type->length;
   case GLSL_TYPE_STRUCT:
      size = 0;
      for (i = 0; i < type->length; i++) {
         size += type_size(type->fields.structure[i].type);
      }
      return size;
   case GLSL_TYPE_SAMPLER:
      /* Samplers take up one slot in UNIFORMS[], but they're baked in
       * at link time.
       */
      return 1;
   case GLSL_TYPE_INTERFACE:
   case GLSL_TYPE_VOID:
   case GLSL_TYPE_ERROR:
      assert(!"Invalid type in type_size");
      break;
   }
   return 0;
}

/**
 * In the initial pass of codegen, we assign temporary numbers to
 * intermediate results.  (not SSA -- variable assignments will reuse
 * storage).
 */
st_src_reg
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.reladdr = NULL;
   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);
   }

   return src;
}

variable_storage *
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();

      if (entry->var == var)
         return entry;
   }

   return NULL;
}

void
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;
   }

   if (ir->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);

      /* Check if this statevar's setup in the STATE file exactly
       * matches how we'll want to reference it as a
       * struct/array/whatever.  If not, then we need to move it into
       * temporary storage and hope that it'll get copy-propagated
       * out.
       */
      for (i = 0; i < ir->num_state_slots; i++) {
         if (slots[i].swizzle != SWIZZLE_XYZW) {
            break;
         }
      }

      variable_storage *storage;
      st_dst_reg dst;
      if (i == ir->num_state_slots) {
         /* We'll set the index later. */
         storage = new(mem_ctx) variable_storage(ir, PROGRAM_STATE_VAR, -1);
         this->variables.push_tail(storage);

         dst = undef_dst;
      } else {
         /* The variable_storage constructor allocates slots based on the size
          * of the type.  However, this had better match the number of state
          * elements that we're going to copy into the new temporary.
          */
         assert((int) ir->num_state_slots == type_size(ir->type));

         dst = st_dst_reg(get_temp(ir->type));

         storage = new(mem_ctx) variable_storage(ir, dst.file, dst.index);

         this->variables.push_tail(storage);
      }


      for (unsigned int i = 0; i < ir->num_state_slots; i++) {
         int index = _mesa_add_state_reference(this->prog->Parameters,
                                               (gl_state_index *)slots[i].tokens);

         if (storage->file == PROGRAM_STATE_VAR) {
            if (storage->index == -1) {
               storage->index = index;
            } else {
               assert(index == storage->index + (int)i);
            }
         } else {
                /* We use GLSL_TYPE_FLOAT here regardless of the actual type of
                 * the data being moved since MOV does not care about the type of
                 * data it is moving, and we don't want to declare registers with
                 * array or struct types.
                 */
            st_src_reg src(PROGRAM_STATE_VAR, index, GLSL_TYPE_FLOAT);
            src.swizzle = slots[i].swizzle;
            emit(ir, TGSI_OPCODE_MOV, dst, src);
            /* even a float takes up a whole vec4 reg in a struct/array. */
            dst.index++;
         }
      }

      if (storage->file == PROGRAM_TEMPORARY &&
          dst.index != storage->index + (int) ir->num_state_slots) {
         fail_link(this->shader_program,
                   "failed to load builtin uniform `%s'  (%d/%d regs loaded)\n",
                   ir->name, dst.index - storage->index,
                   type_size(ir->type));
      }
   }
}

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);
}

void
glsl_to_tgsi_visitor::visit(ir_loop_jump *ir)
{
   switch (ir->mode) {
   case ir_loop_jump::jump_break:
      emit(NULL, TGSI_OPCODE_BRK);
      break;
   case ir_loop_jump::jump_continue:
      emit(NULL, TGSI_OPCODE_CONT);
      break;
   }
}


void
glsl_to_tgsi_visitor::visit(ir_function_signature *ir)
{
   assert(0);
   (void)ir;
}

void
glsl_to_tgsi_visitor::visit(ir_function *ir)
{
   /* Ignore function bodies other than main() -- we shouldn't see calls to
    * them since they should all be inlined before we get to glsl_to_tgsi.
    */
   if (strcmp(ir->name, "main") == 0) {
      const ir_function_signature *sig;
      exec_list empty;

      sig = ir->matching_signature(&empty);

      assert(sig);

      foreach_iter(exec_list_iterator, iter, sig->body) {
         ir_instruction *ir = (ir_instruction *)iter.get();

         ir->accept(this);
      }
   }
}

bool
glsl_to_tgsi_visitor::try_emit_mad(ir_expression *ir, int mul_operand)
{
   int nonmul_operand = 1 - mul_operand;
   st_src_reg a, b, c;
   st_dst_reg result_dst;

   ir_expression *expr = ir->operands[mul_operand]->as_expression();
   if (!expr || expr->operation != ir_binop_mul)
      return false;

   expr->operands[0]->accept(this);
   a = this->result;
   expr->operands[1]->accept(this);
   b = this->result;
   ir->operands[nonmul_operand]->accept(this);
   c = this->result;

   this->result = get_temp(ir->type);
   result_dst = st_dst_reg(this->result);
   result_dst.writemask = (1 << ir->type->vector_elements) - 1;
   emit(ir, TGSI_OPCODE_MAD, result_dst, a, b, c);

   return true;
}

/**
 * Emit MAD(a, -b, a) instead of AND(a, NOT(b))
 *
 * The logic values are 1.0 for true and 0.0 for false.  Logical-and is
 * implemented using multiplication, and logical-or is implemented using
 * addition.  Logical-not can be implemented as (true - x), or (1.0 - x).
 * As result, the logical expression (a & !b) can be rewritten as:
 *
 *     - a * !b
 *     - a * (1 - b)
 *     - (a * 1) - (a * b)
 *     - a + -(a * b)
 *     - a + (a * -b)
 *
 * This final expression can be implemented as a single MAD(a, -b, a)
 * instruction.
 */
bool
glsl_to_tgsi_visitor::try_emit_mad_for_and_not(ir_expression *ir, int try_operand)
{
   const int other_operand = 1 - try_operand;
   st_src_reg a, b;

   ir_expression *expr = ir->operands[try_operand]->as_expression();
   if (!expr || expr->operation != ir_unop_logic_not)
      return false;

   ir->operands[other_operand]->accept(this);
   a = this->result;
   expr->operands[0]->accept(this);
   b = this->result;

   b.negate = ~b.negate;

   this->result = get_temp(ir->type);
   emit(ir, TGSI_OPCODE_MAD, st_dst_reg(this->result), a, b, a);

   return true;
}

bool
glsl_to_tgsi_visitor::try_emit_sat(ir_expression *ir)
{
   /* Emit saturates in the vertex shader only if SM 3.0 is supported.
    */
   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)
      return false;

   sat_src->accept(this);
   st_src_reg src = this->result;

   /* If we generated an expression instruction into a temporary in
    * processing the saturate's operand, apply the saturate to that
    * instruction.  Otherwise, generate a MOV to do the saturate.
    *
    * Note that we have to be careful to only do this optimization if
    * the instruction in question was what generated src->result.  For
    * example, ir_dereference_array might generate a MUL instruction
    * to create the reladdr, and return us a src reg using that
    * reladdr.  That MUL result is not the value we're trying to
    * saturate.
    */
   ir_expression *sat_src_expr = sat_src->as_expression();
   if (sat_src_expr && (sat_src_expr->operation == ir_binop_mul ||
                        sat_src_expr->operation == ir_binop_add ||
                        sat_src_expr->operation == ir_binop_dot)) {
      glsl_to_tgsi_instruction *new_inst;
      new_inst = (glsl_to_tgsi_instruction *)this->instructions.get_tail();
      new_inst->saturate = true;
   } else {
      this->result = get_temp(ir->type);
      st_dst_reg result_dst = st_dst_reg(this->result);
      result_dst.writemask = (1 << ir->type->vector_elements) - 1;
      glsl_to_tgsi_instruction *inst;
      inst = emit(ir, TGSI_OPCODE_MOV, result_dst, src);
      inst->saturate = true;
   }

   return true;
}

void
glsl_to_tgsi_visitor::reladdr_to_temp(ir_instruction *ir,
                                    st_src_reg *reg, int *num_reladdr)
{
   if (!reg->reladdr)
      return;

   emit_arl(ir, address_reg, *reg->reladdr);

   if (*num_reladdr != 1) {
      st_src_reg temp = get_temp(glsl_type::vec4_type);

      emit(ir, TGSI_OPCODE_MOV, st_dst_reg(temp), *reg);
      *reg = temp;
   }

   (*num_reladdr)--;
}

void
glsl_to_tgsi_visitor::visit(ir_expression *ir)
{
   unsigned int operand;
   st_src_reg op[Elements(ir->operands)];
   st_src_reg result_src;
   st_dst_reg result_dst;

   /* Quick peephole: Emit MAD(a, b, c) instead of ADD(MUL(a, b), c)
    */
   if (ir->operation == ir_binop_add) {
      if (try_emit_mad(ir, 1))
         return;
      if (try_emit_mad(ir, 0))
         return;
   }

   /* Quick peephole: Emit OPCODE_MAD(-a, -b, a) instead of AND(a, NOT(b))
    */
   if (ir->operation == ir_binop_logic_and) {
      if (try_emit_mad_for_and_not(ir, 1))
         return;
      if (try_emit_mad_for_and_not(ir, 0))
         return;
   }

   if (try_emit_sat(ir))
      return;

   if (ir->operation == ir_quadop_vector)
      assert(!"ir_quadop_vector should have been lowered");

   for (operand = 0; operand < ir->get_num_operands(); operand++) {
      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);
         exit(1);
      }
      op[operand] = this->result;

      /* Matrix expression operands should have been broken down to vector
       * operations already.
       */
      assert(!ir->operands[operand]->type->is_matrix());
   }

   int vector_elements = ir->operands[0]->type->vector_elements;
   if (ir->operands[1]) {
      vector_elements = MAX2(vector_elements,
                             ir->operands[1]->type->vector_elements);
   }

   this->result.file = PROGRAM_UNDEFINED;

   /* Storage for our result.  Ideally for an assignment we'd be using
    * the actual storage for the result here, instead.
    */
   result_src = get_temp(ir->type);
   /* convenience for the emit functions below. */
   result_dst = st_dst_reg(result_src);
   /* Limit writes to the channels that will be used by result_src later.
    * This does limit this temp's use as a temporary for multi-instruction
    * sequences.
    */
   result_dst.writemask = (1 << ir->type->vector_elements) - 1;

   switch (ir->operation) {
   case ir_unop_logic_not:
      if (result_dst.type != GLSL_TYPE_FLOAT)
         emit(ir, TGSI_OPCODE_NOT, result_dst, op[0]);
      else {
         /* Previously 'SEQ dst, src, 0.0' was used for this.  However, many
          * older GPUs implement SEQ using multiple instructions (i915 uses two
          * SGE instructions and a MUL instruction).  Since our logic values are
          * 0.0 and 1.0, 1-x also implements !x.
          */
         op[0].negate = ~op[0].negate;
         emit(ir, TGSI_OPCODE_ADD, result_dst, op[0], st_src_reg_for_float(1.0));
      }
      break;
   case ir_unop_neg:
      if (result_dst.type == GLSL_TYPE_INT || result_dst.type == GLSL_TYPE_UINT)
         emit(ir, TGSI_OPCODE_INEG, result_dst, op[0]);
      else {
         op[0].negate = ~op[0].negate;
         result_src = op[0];
      }
      break;
   case ir_unop_abs:
      emit(ir, TGSI_OPCODE_ABS, result_dst, op[0]);
      break;
   case ir_unop_sign:
      emit(ir, TGSI_OPCODE_SSG, result_dst, op[0]);
      break;
   case ir_unop_rcp:
      emit_scalar(ir, TGSI_OPCODE_RCP, result_dst, op[0]);
      break;

   case ir_unop_exp2:
      emit_scalar(ir, TGSI_OPCODE_EX2, result_dst, op[0]);
      break;
   case ir_unop_exp:
   case ir_unop_log:
      assert(!"not reached: should be handled by ir_explog_to_explog2");
      break;
   case ir_unop_log2:
      emit_scalar(ir, TGSI_OPCODE_LG2, result_dst, op[0]);
      break;
   case ir_unop_sin:
      emit_scalar(ir, TGSI_OPCODE_SIN, result_dst, op[0]);
      break;
   case ir_unop_cos:
      emit_scalar(ir, TGSI_OPCODE_COS, result_dst, op[0]);
      break;
   case ir_unop_sin_reduced:
      emit_scs(ir, TGSI_OPCODE_SIN, result_dst, op[0]);
      break;
   case ir_unop_cos_reduced:
      emit_scs(ir, TGSI_OPCODE_COS, result_dst, op[0]);
      break;

   case ir_unop_dFdx:
      emit(ir, TGSI_OPCODE_DDX, result_dst, op[0]);
      break;
   case ir_unop_dFdy:
   {
      /* The X component contains 1 or -1 depending on whether the framebuffer
       * is a FBO or the window system buffer, respectively.
       * It is then multiplied with the source operand of DDY.
       */
      static const gl_state_index transform_y_state[STATE_LENGTH]
         = { STATE_INTERNAL, STATE_FB_WPOS_Y_TRANSFORM };

      unsigned transform_y_index =
         _mesa_add_state_reference(this->prog->Parameters,
                                   transform_y_state);

      st_src_reg transform_y = st_src_reg(PROGRAM_STATE_VAR,
                                          transform_y_index,
                                          glsl_type::vec4_type);
      transform_y.swizzle = SWIZZLE_XXXX;

      st_src_reg temp = get_temp(glsl_type::vec4_type);

      emit(ir, TGSI_OPCODE_MUL, st_dst_reg(temp), transform_y, op[0]);
      emit(ir, TGSI_OPCODE_DDY, result_dst, temp);
      break;
   }

   case ir_unop_noise: {
      /* At some point, a motivated person could add a better
       * implementation of noise.  Currently not even the nvidia
       * binary drivers do anything more than this.  In any case, the
       * place to do this is in the GL state tracker, not the poor
       * driver.
       */
      emit(ir, TGSI_OPCODE_MOV, result_dst, st_src_reg_for_float(0.5));
      break;
   }

   case ir_binop_add:
      emit(ir, TGSI_OPCODE_ADD, result_dst, op[0], op[1]);
      break;
   case ir_binop_sub:
      emit(ir, TGSI_OPCODE_SUB, result_dst, op[0], op[1]);
      break;

   case ir_binop_mul:
      emit(ir, TGSI_OPCODE_MUL, result_dst, op[0], op[1]);
      break;
   case ir_binop_div:
      if (result_dst.type == GLSL_TYPE_FLOAT)
         assert(!"not reached: should be handled by ir_div_to_mul_rcp");
      else
         emit(ir, TGSI_OPCODE_DIV, result_dst, op[0], op[1]);
      break;
   case ir_binop_mod:
      if (result_dst.type == GLSL_TYPE_FLOAT)
         assert(!"ir_binop_mod should have been converted to b * fract(a/b)");
      else
         emit(ir, TGSI_OPCODE_MOD, result_dst, op[0], op[1]);
      break;

   case ir_binop_less:
      emit(ir, TGSI_OPCODE_SLT, result_dst, op[0], op[1]);
      break;
   case ir_binop_greater:
      emit(ir, TGSI_OPCODE_SLT, result_dst, op[1], op[0]);
      break;
   case ir_binop_lequal:
      emit(ir, TGSI_OPCODE_SGE, result_dst, op[1], op[0]);
      break;
   case ir_binop_gequal:
      emit(ir, TGSI_OPCODE_SGE, result_dst, op[0], op[1]);
      break;
   case ir_binop_equal:
      emit(ir, TGSI_OPCODE_SEQ, result_dst, op[0], op[1]);
      break;
   case ir_binop_nequal:
      emit(ir, TGSI_OPCODE_SNE, result_dst, op[0], op[1]);
      break;
   case ir_binop_all_equal:
      /* "==" operator producing a scalar boolean. */
      if (ir->operands[0]->type->is_vector() ||
          ir->operands[1]->type->is_vector()) {
         st_src_reg temp = get_temp(native_integers ?
               glsl_type::get_instance(ir->operands[0]->type->base_type, 4, 1) :
               glsl_type::vec4_type);
         
         if (native_integers) {
            st_dst_reg temp_dst = st_dst_reg(temp);
            st_src_reg temp1 = st_src_reg(temp), temp2 = st_src_reg(temp);
            
            emit(ir, TGSI_OPCODE_SEQ, st_dst_reg(temp), op[0], op[1]);
            
            /* Emit 1-3 AND operations to combine the SEQ results. */
            switch (ir->operands[0]->type->vector_elements) {
            case 2:
               break;
            case 3:
               temp_dst.writemask = WRITEMASK_Y;
               temp1.swizzle = SWIZZLE_YYYY;
               temp2.swizzle = SWIZZLE_ZZZZ;
               emit(ir, TGSI_OPCODE_AND, temp_dst, temp1, temp2);
               break;
            case 4:
               temp_dst.writemask = WRITEMASK_X;
               temp1.swizzle = SWIZZLE_XXXX;
               temp2.swizzle = SWIZZLE_YYYY;
               emit(ir, TGSI_OPCODE_AND, temp_dst, temp1, temp2);
               temp_dst.writemask = WRITEMASK_Y;
               temp1.swizzle = SWIZZLE_ZZZZ;
               temp2.swizzle = SWIZZLE_WWWW;
               emit(ir, TGSI_OPCODE_AND, temp_dst, temp1, temp2);
            }
            
            temp1.swizzle = SWIZZLE_XXXX;
            temp2.swizzle = SWIZZLE_YYYY;
            emit(ir, TGSI_OPCODE_AND, result_dst, temp1, temp2);
         } else {
            emit(ir, TGSI_OPCODE_SNE, st_dst_reg(temp), op[0], op[1]);
            
            /* After the dot-product, the value will be an integer on the
             * range [0,4].  Zero becomes 1.0, and positive values become zero.
             */
            emit_dp(ir, result_dst, temp, temp, vector_elements);

            /* Negating the result of the dot-product gives values on the range
             * [-4, 0].  Zero becomes 1.0, and negative values become zero.
             * This is achieved using SGE.
             */
            st_src_reg sge_src = result_src;
            sge_src.negate = ~sge_src.negate;
            emit(ir, TGSI_OPCODE_SGE, result_dst, sge_src, st_src_reg_for_float(0.0));
         }
      } else {
         emit(ir, TGSI_OPCODE_SEQ, result_dst, op[0], op[1]);
      }
      break;
   case ir_binop_any_nequal:
      /* "!=" operator producing a scalar boolean. */
      if (ir->operands[0]->type->is_vector() ||
          ir->operands[1]->type->is_vector()) {
         st_src_reg temp = get_temp(native_integers ?
               glsl_type::get_instance(ir->operands[0]->type->base_type, 4, 1) :
               glsl_type::vec4_type);
         emit(ir, TGSI_OPCODE_SNE, st_dst_reg(temp), op[0], op[1]);

         if (native_integers) {
            st_dst_reg temp_dst = st_dst_reg(temp);
            st_src_reg temp1 = st_src_reg(temp), temp2 = st_src_reg(temp);
            
            /* Emit 1-3 OR operations to combine the SNE results. */
            switch (ir->operands[0]->type->vector_elements) {
            case 2:
               break;
            case 3:
               temp_dst.writemask = WRITEMASK_Y;
               temp1.swizzle = SWIZZLE_YYYY;
               temp2.swizzle = SWIZZLE_ZZZZ;
               emit(ir, TGSI_OPCODE_OR, temp_dst, temp1, temp2);
               break;
            case 4:
               temp_dst.writemask = WRITEMASK_X;
               temp1.swizzle = SWIZZLE_XXXX;
               temp2.swizzle = SWIZZLE_YYYY;
               emit(ir, TGSI_OPCODE_OR, temp_dst, temp1, temp2);
               temp_dst.writemask = WRITEMASK_Y;
               temp1.swizzle = SWIZZLE_ZZZZ;
               temp2.swizzle = SWIZZLE_WWWW;
               emit(ir, TGSI_OPCODE_OR, temp_dst, temp1, temp2);
            }
            
            temp1.swizzle = SWIZZLE_XXXX;
            temp2.swizzle = SWIZZLE_YYYY;
            emit(ir, TGSI_OPCODE_OR, result_dst, temp1, temp2);
         } 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, temp, temp, vector_elements);
            if (this->prog->Target == GL_FRAGMENT_PROGRAM_ARB) {
               /* The clamping to [0,1] can be done for free in the fragment
                * shader with a saturate.
                */
               dp->saturate = true;
            } else {
               /* Negating the result of the dot-product gives values on the range
                * [-4, 0].  Zero stays zero, and negative values become 1.0.  This
                * 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 {
         emit(ir, TGSI_OPCODE_SNE, result_dst, op[0], op[1]);
      }
      break;

   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));
      }
      break;
   }

   case ir_binop_logic_xor:
      if (native_integers)
         emit(ir, TGSI_OPCODE_XOR, result_dst, op[0], op[1]);
      else
         emit(ir, TGSI_OPCODE_SNE, result_dst, op[0], op[1]);
      break;

   case ir_binop_logic_or: {
      if (native_integers) {
         /* If integers are used as booleans, we can use an actual "or" 
          * instruction.
          */
         assert(native_integers);
         emit(ir, TGSI_OPCODE_OR, result_dst, op[0], op[1]);
      } else {
         /* After the addition, the value will be an integer on the
          * range [0,2].  Zero stays zero, and positive values become 1.0.
          */
         glsl_to_tgsi_instruction *add =
            emit(ir, TGSI_OPCODE_ADD, result_dst, op[0], op[1]);
         if (this->prog->Target == GL_FRAGMENT_PROGRAM_ARB) {
            /* The clamping to [0,1] can be done for free in the fragment
             * shader with a saturate if floats are being used as boolean values.
             */
            add->saturate = true;
         } else {
            /* Negating the result of the addition gives values on the range
             * [-2, 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));
         }
      }
      break;
   }

   case ir_binop_logic_and:
      /* If native integers are disabled, the bool args are stored as float 0.0
       * or 1.0, so "mul" gives us "and".  If they're enabled, just use the
       * actual AND opcode.
       */
      if (native_integers)
         emit(ir, TGSI_OPCODE_AND, result_dst, op[0], op[1]);
      else
         emit(ir, TGSI_OPCODE_MUL, result_dst, op[0], op[1]);
      break;

   case ir_binop_dot:
      assert(ir->operands[0]->type->is_vector());
      assert(ir->operands[0]->type == ir->operands[1]->type);
      emit_dp(ir, result_dst, op[0], op[1],
              ir->operands[0]->type->vector_elements);
      break;

   case ir_unop_sqrt:
      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]);
      break;
   case ir_unop_i2f:
      if (native_integers) {
         emit(ir, TGSI_OPCODE_I2F, result_dst, op[0]);
         break;
      }
      /* fallthrough to next case otherwise */
   case ir_unop_b2f:
      if (native_integers) {
         emit(ir, TGSI_OPCODE_AND, result_dst, op[0], st_src_reg_for_float(1.0));
         break;
      }
      /* fallthrough to next case otherwise */
   case ir_unop_i2u:
   case ir_unop_u2i:
      /* Converting between signed and unsigned integers is a no-op. */
      result_src = op[0];
      break;
   case ir_unop_b2i:
      if (native_integers) {
         /* Booleans are stored as integers using ~0 for true and 0 for false.
          * GLSL requires that int(bool) return 1 for true and 0 for false.
          * This conversion is done with AND, but it could be done with NEG.
          */
         emit(ir, TGSI_OPCODE_AND, result_dst, op[0], st_src_reg_for_int(1));
      } else {
         /* Booleans and integers are both stored as floats when native 
          * integers are disabled.
          */
         result_src = op[0];
      }
      break;
   case ir_unop_f2i:
      if (native_integers)
         emit(ir, TGSI_OPCODE_F2I, result_dst, op[0]);
      else
         emit(ir, TGSI_OPCODE_TRUNC, result_dst, op[0]);
      break;
   case ir_unop_f2u:
      if (native_integers)
         emit(ir, TGSI_OPCODE_F2U, result_dst, op[0]);
      else
         emit(ir, TGSI_OPCODE_TRUNC, result_dst, op[0]);
      break;
   case ir_unop_bitcast_f2i:
   case ir_unop_bitcast_f2u:
   case ir_unop_bitcast_i2f:
   case ir_unop_bitcast_u2f:
      result_src = op[0];
      break;
   case ir_unop_f2b:
      emit(ir, TGSI_OPCODE_SNE, result_dst, op[0], st_src_reg_for_float(0.0));
      break;
   case ir_unop_i2b:
      if (native_integers)
         emit(ir, TGSI_OPCODE_INEG, result_dst, op[0]);
      else
         emit(ir, TGSI_OPCODE_SNE, result_dst, op[0], st_src_reg_for_float(0.0));
      break;
   case ir_unop_trunc:
      emit(ir, TGSI_OPCODE_TRUNC, result_dst, op[0]);
      break;
   case ir_unop_ceil:
      emit(ir, TGSI_OPCODE_CEIL, result_dst, op[0]);
      break;
   case ir_unop_floor:
      emit(ir, TGSI_OPCODE_FLR, result_dst, op[0]);
      break;
   case ir_unop_round_even:
      emit(ir, TGSI_OPCODE_ROUND, result_dst, op[0]);
      break;
   case ir_unop_fract:
      emit(ir, TGSI_OPCODE_FRC, result_dst, op[0]);
      break;

   case ir_binop_min:
      emit(ir, TGSI_OPCODE_MIN, result_dst, op[0], op[1]);
      break;
   case ir_binop_max:
      emit(ir, TGSI_OPCODE_MAX, result_dst, op[0], op[1]);
      break;
   case ir_binop_pow:
      emit_scalar(ir, TGSI_OPCODE_POW, result_dst, op[0], op[1]);
      break;

   case ir_unop_bit_not:
      if (native_integers) {
         emit(ir, TGSI_OPCODE_NOT, result_dst, op[0]);
         break;
      }
   case ir_unop_u2f:
      if (native_integers) {
         emit(ir, TGSI_OPCODE_U2F, result_dst, op[0]);
         break;
      }
   case ir_binop_lshift:
      if (native_integers) {
         emit(ir, TGSI_OPCODE_SHL, result_dst, op[0], op[1]);
         break;
      }
   case ir_binop_rshift:
      if (native_integers) {
         emit(ir, TGSI_OPCODE_ISHR, result_dst, op[0], op[1]);
         break;
      }
   case ir_binop_bit_and:
      if (native_integers) {
         emit(ir, TGSI_OPCODE_AND, result_dst, op[0], op[1]);
         break;
      }
   case ir_binop_bit_xor:
      if (native_integers) {
         emit(ir, TGSI_OPCODE_XOR, result_dst, op[0], op[1]);
         break;
      }
   case ir_binop_bit_or:
      if (native_integers) {
         emit(ir, TGSI_OPCODE_OR, result_dst, op[0], op[1]);
         break;
      }

      assert(!"GLSL 1.30 features unsupported");
      break;

   case ir_binop_ubo_load: {
      ir_constant *uniform_block = ir->operands[0]->as_constant();
      ir_constant *const_offset_ir = ir->operands[1]->as_constant();
      unsigned const_offset = const_offset_ir ? const_offset_ir->value.u[0] : 0;
      st_src_reg index_reg = get_temp(glsl_type::uint_type);
      st_src_reg cbuf;

      cbuf.type = glsl_type::vec4_type->base_type;
      cbuf.file = PROGRAM_CONSTANT;
      cbuf.index = 0;
      cbuf.index2D = uniform_block->value.u[0] + 1;
      cbuf.reladdr = NULL;
      cbuf.negate = 0;
      
      assert(ir->type->is_vector() || ir->type->is_scalar());

      if (const_offset_ir) {
         index_reg = st_src_reg_for_int(const_offset / 16);
      } else {
         emit(ir, TGSI_OPCODE_USHR, st_dst_reg(index_reg), op[1], st_src_reg_for_int(4));
      }

      cbuf.swizzle = swizzle_for_size(ir->type->vector_elements);
      cbuf.swizzle += MAKE_SWIZZLE4(const_offset % 16 / 4,
                                    const_offset % 16 / 4,
                                    const_offset % 16 / 4,
                                    const_offset % 16 / 4);

      cbuf.reladdr = ralloc(mem_ctx, st_src_reg);
      memcpy(cbuf.reladdr, &index_reg, sizeof(index_reg));

      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_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_unop_bitfield_reverse:
   case ir_unop_bit_count:
   case ir_unop_find_msb:
   case ir_unop_find_lsb:
   case ir_triop_bitfield_extract:
   case ir_quadop_bitfield_insert:
   case ir_quadop_vector:
      /* This operation is not supported, or should have already been handled.
       */
      assert(!"Invalid ir opcode in glsl_to_tgsi_visitor::visit()");
      break;
   }

   this->result = result_src;
}


void
glsl_to_tgsi_visitor::visit(ir_swizzle *ir)
{
   st_src_reg src;
   int i;
   int swizzle[4];

   /* Note that this is only swizzles in expressions, not those on the left
    * hand side of an assignment, which do write masking.  See ir_assignment
    * for that.
    */

   ir->val->accept(this);
   src = this->result;
   assert(src.file != PROGRAM_UNDEFINED);

   for (i = 0; i < 4; i++) {
      if (i < ir->type->vector_elements) {
         switch (i) {
         case 0:
            swizzle[i] = GET_SWZ(src.swizzle, ir->mask.x);
            break;
         case 1:
            swizzle[i] = GET_SWZ(src.swizzle, ir->mask.y);
            break;
         case 2:
            swizzle[i] = GET_SWZ(src.swizzle, ir->mask.z);
            break;
         case 3:
            swizzle[i] = GET_SWZ(src.swizzle, ir->mask.w);
            break;
         }
      } else {
         /* If the type is smaller than a vec4, replicate the last
          * channel out.
          */
         swizzle[i] = swizzle[ir->type->vector_elements - 1];
      }
   }

   src.swizzle = MAKE_SWIZZLE4(swizzle[0], swizzle[1], swizzle[2], swizzle[3]);

   this->result = src;
}

void
glsl_to_tgsi_visitor::visit(ir_dereference_variable *ir)
{
   variable_storage *entry = find_variable_storage(ir->var);
   ir_variable *var = ir->var;

   if (!entry) {
      switch (var->mode) {
      case ir_var_uniform:
         entry = new(mem_ctx) variable_storage(var, PROGRAM_UNIFORM,
                                               var->location);
         this->variables.push_tail(entry);
         break;
      case ir_var_shader_in:
         /* The linker assigns locations for varyings and attributes,
          * including deprecated builtins (like gl_Color), user-assign
          * generic attributes (glBindVertexLocation), and
          * user-defined varyings.
          */
         assert(var->location != -1);
         entry = new(mem_ctx) variable_storage(var,
                                               PROGRAM_INPUT,
                                               var->location);
         break;
      case ir_var_shader_out:
         assert(var->location != -1);
         entry = new(mem_ctx) variable_storage(var,
                                               PROGRAM_OUTPUT,
                                               var->location + var->index);
         break;
      case ir_var_system_value:
         entry = new(mem_ctx) variable_storage(var,
                                               PROGRAM_SYSTEM_VALUE,
                                               var->location);
         break;
      case ir_var_auto:
      case ir_var_temporary:
         st_src_reg src = get_temp(var->type);

         entry = new(mem_ctx) variable_storage(var, src.file, src.index);
         this->variables.push_tail(entry);

         break;
      }

      if (!entry) {
         printf("Failed to make storage for %s\n", var->name);
         exit(1);
      }
   }

   this->result = st_src_reg(entry->file, entry->index, var->type);
   if (!native_integers)
      this->result.type = GLSL_TYPE_FLOAT;
}

void
glsl_to_tgsi_visitor::visit(ir_dereference_array *ir)
{
   ir_constant *index;
   st_src_reg src;
   int element_size = type_size(ir->type);

   index = ir->array_index->constant_expression_value();

   ir->array->accept(this);
   src = this->result;

   if (index) {
      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
       * index.
       */
      ir->array_index->accept(this);

      st_src_reg index_reg;

      if (element_size == 1) {
         index_reg = this->result;
      } else {
         index_reg = get_temp(native_integers ?
                              glsl_type::int_type : glsl_type::float_type);

         emit(ir, TGSI_OPCODE_MUL, st_dst_reg(index_reg),
              this->result, st_src_reg_for_type(index_reg.type, element_size));
      }

      /* 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) {
         st_src_reg accum_reg = get_temp(native_integers ?
                                glsl_type::int_type : glsl_type::float_type);

         emit(ir, TGSI_OPCODE_ADD, st_dst_reg(accum_reg),
              index_reg, *src.reladdr);

         index_reg = accum_reg;
      }

      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. */
   if (ir->type->is_scalar() || ir->type->is_vector())
      src.swizzle = swizzle_for_size(ir->type->vector_elements);
   else
      src.swizzle = SWIZZLE_NOOP;

   /* Change the register type to the element type of the array. */
   src.type = ir->type->base_type;

   this->result = src;
}

void
glsl_to_tgsi_visitor::visit(ir_dereference_record *ir)
{
   unsigned int i;
   const glsl_type *struct_type = ir->record->type;
   int offset = 0;

   ir->record->accept(this);

   for (i = 0; i < struct_type->length; i++) {
      if (strcmp(struct_type->fields.structure[i].name, ir->field) == 0)
         break;
      offset += type_size(struct_type->fields.structure[i].type);
   }

   /* If the type is smaller than a vec4, replicate the last channel out. */
   if (ir->type->is_scalar() || ir->type->is_vector())
      this->result.swizzle = swizzle_for_size(ir->type->vector_elements);
   else
      this->result.swizzle = SWIZZLE_NOOP;

   this->result.index += offset;
   this->result.type = ir->type->base_type;
}

/**
 * We want to be careful in assignment setup to hit the actual storage
 * instead of potentially using a temporary like we might with the
 * ir_dereference handler.
 */
static st_dst_reg
get_assignment_lhs(ir_dereference *ir, glsl_to_tgsi_visitor *v)
{
   /* The LHS must be a dereference.  If the LHS is a variable indexed array
    * access of a vector, it must be separated into a series conditional moves
    * before reaching this point (see ir_vec_index_to_cond_assign).
    */
   assert(ir->as_dereference());
   ir_dereference_array *deref_array = ir->as_dereference_array();
   if (deref_array) {
      assert(!deref_array->array->type->is_vector());
   }

   /* Use the rvalue deref handler for the most part.  We'll ignore
    * swizzles in it and write swizzles using writemask, though.
    */
   ir->accept(v);
   return st_dst_reg(v->result);
}

/**
 * Process the condition of a conditional assignment
 *
 * Examines the condition of a conditional assignment to generate the optimal
 * first operand of a \c CMP instruction.  If the condition is a relational
 * operator with 0 (e.g., \c ir_binop_less), the value being compared will be
 * used as the source for the \c CMP instruction.  Otherwise the comparison
 * is processed to a boolean result, and the boolean result is used as the
 * operand to the CMP instruction.
 */
bool
glsl_to_tgsi_visitor::process_move_condition(ir_rvalue *ir)
{
   ir_rvalue *src_ir = ir;
   bool negate = true;
   bool switch_order = false;

   ir_expression *const expr = ir->as_expression();
   if ((expr != NULL) && (expr->get_num_operands() == 2)) {
      bool zero_on_left = false;

      if (expr->operands[0]->is_zero()) {
         src_ir = expr->operands[1];
         zero_on_left = true;
      } else if (expr->operands[1]->is_zero()) {
         src_ir = expr->operands[0];
         zero_on_left = false;
      }

      /*      a is -  0  +            -  0  +
       * (a <  0)  T  F  F  ( a < 0)  T  F  F
       * (0 <  a)  F  F  T  (-a < 0)  F  F  T
       * (a <= 0)  T  T  F  (-a < 0)  F  F  T  (swap order of other operands)
       * (0 <= a)  F  T  T  ( a < 0)  T  F  F  (swap order of other operands)
       * (a >  0)  F  F  T  (-a < 0)  F  F  T
       * (0 >  a)  T  F  F  ( a < 0)  T  F  F
       * (a >= 0)  F  T  T  ( a < 0)  T  F  F  (swap order of other operands)
       * (0 >= a)  T  T  F  (-a < 0)  F  F  T  (swap order of other operands)
       *
       * Note that exchanging the order of 0 and 'a' in the comparison simply
       * means that the value of 'a' should be negated.
       */
      if (src_ir != ir) {
         switch (expr->operation) {
         case ir_binop_less:
            switch_order = false;
            negate = zero_on_left;
            break;

         case ir_binop_greater:
            switch_order = false;
            negate = !zero_on_left;
            break;

         case ir_binop_lequal:
            switch_order = true;
            negate = !zero_on_left;
            break;

         case ir_binop_gequal:
            switch_order = true;
            negate = zero_on_left;
            break;

         default:
            /* This isn't the right kind of comparison afterall, so make sure
             * the whole condition is visited.
             */
            src_ir = ir;
            break;
         }
      }
   }

   src_ir->accept(this);

   /* We use the TGSI_OPCODE_CMP (a < 0 ? b : c) for conditional moves, and the
    * condition we produced is 0.0 or 1.0.  By flipping the sign, we can
    * choose which value TGSI_OPCODE_CMP produces without an extra instruction
    * computing the condition.
    */
   if (negate)
      this->result.negate = ~this->result.negate;

   return switch_order;
}

void
glsl_to_tgsi_visitor::emit_block_mov(ir_assignment *ir, const struct glsl_type *type,
                                     st_dst_reg *l, st_src_reg *r)
{
   if (type->base_type == GLSL_TYPE_STRUCT) {
      for (unsigned int i = 0; i < type->length; i++) {
         emit_block_mov(ir, type->fields.structure[i].type, l, r);
      }
      return;
   }

   if (type->is_array()) {
      for (unsigned int i = 0; i < type->length; i++) {
         emit_block_mov(ir, type->fields.array, l, r);
      }
      return;
   }

   if (type->is_matrix()) {
      const struct glsl_type *vec_type;

      vec_type = glsl_type::get_instance(GLSL_TYPE_FLOAT,
                                         type->vector_elements, 1);

      for (int i = 0; i < type->matrix_columns; i++) {
         emit_block_mov(ir, vec_type, l, r);
      }
      return;
   }

   assert(type->is_scalar() || type->is_vector());

   r->type = type->base_type;
   emit(ir, TGSI_OPCODE_MOV, *l, *r);
   l->index++;
   r->index++;
}

void
glsl_to_tgsi_visitor::visit(ir_assignment *ir)
{
   st_dst_reg l;
   st_src_reg r;
   int i;

   ir->rhs->accept(this);
   r = this->result;

   l = get_assignment_lhs(ir->lhs, this);

   /* FINISHME: This should really set to the correct maximal writemask for each
    * FINISHME: component written (in the loops below).  This case can only
    * FINISHME: occur for matrices, arrays, and structures.
    */
   if (ir->write_mask == 0) {
      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) {
      /* FINISHME: This hack makes writing to gl_FragDepth, which lives in the
       * FINISHME: W component of fragment shader output zero, work correctly.
       */
      l.writemask = WRITEMASK_XYZW;
   } else {
      int swizzles[4];
      int first_enabled_chan = 0;
      int rhs_chan = 0;

      l.writemask = ir->write_mask;

      for (int i = 0; i < 4; i++) {
         if (l.writemask & (1 << i)) {
            first_enabled_chan = GET_SWZ(r.swizzle, i);
            break;
         }
      }

      /* Swizzle a small RHS vector into the channels being written.
       *
       * glsl ir treats write_mask as dictating how many channels are
       * present on the RHS while TGSI treats write_mask as just
       * showing which channels of the vec4 RHS get written.
       */
      for (int i = 0; i < 4; i++) {
         if (l.writemask & (1 << i))
            swizzles[i] = GET_SWZ(r.swizzle, rhs_chan++);
         else
            swizzles[i] = first_enabled_chan;
      }
      r.swizzle = MAKE_SWIZZLE4(swizzles[0], swizzles[1],
                                swizzles[2], swizzles[3]);
   }

   assert(l.file != PROGRAM_UNDEFINED);
   assert(r.file != PROGRAM_UNDEFINED);

   if (ir->condition) {
      const bool switch_order = this->process_move_condition(ir->condition);
      st_src_reg condition = this->result;

      for (i = 0; i < type_size(ir->lhs->type); i++) {
         st_src_reg l_src = st_src_reg(l);
         st_src_reg condition_temp = condition;
         l_src.swizzle = swizzle_for_size(ir->lhs->type->vector_elements);
         
         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.
             */
            condition_temp = get_temp(glsl_type::vec4_type);
            condition.negate = 0;
            emit(ir, TGSI_OPCODE_I2F, st_dst_reg(condition_temp), condition);
            condition_temp.swizzle = condition.swizzle;
         }
         
         if (switch_order) {
            emit(ir, TGSI_OPCODE_CMP, l, condition_temp, l_src, r);
         } else {
            emit(ir, TGSI_OPCODE_CMP, l, condition_temp, r, l_src);
         }

         l.index++;
         r.index++;
      }
   } else if (ir->rhs->as_expression() &&
              this->instructions.get_tail() &&
              ir->rhs == ((glsl_to_tgsi_instruction *)this->instructions.get_tail())->ir &&
              type_size(ir->lhs->type) == 1 &&
              l.writemask == ((glsl_to_tgsi_instruction *)this->instructions.get_tail())->dst.writemask) {
      /* To avoid emitting an extra MOV when assigning an expression to a 
       * variable, emit the last instruction of the expression again, but
       * replace the destination register with the target of the assignment.
       * Dead code elimination will remove the original instruction.
       */
      glsl_to_tgsi_instruction *inst, *new_inst;
      inst = (glsl_to_tgsi_instruction *)this->instructions.get_tail();
      new_inst = emit(ir, inst->op, l, inst->src[0], inst->src[1], inst->src[2]);
      new_inst->saturate = inst->saturate;
      inst->dead_mask = inst->dst.writemask;
   } else {
      emit_block_mov(ir, ir->rhs->type, &l, &r);
   }
}


void
glsl_to_tgsi_visitor::visit(ir_constant *ir)
{
   st_src_reg src;
   GLfloat stack_vals[4] = { 0 };
   gl_constant_value *values = (gl_constant_value *) stack_vals;
   GLenum gl_type = GL_NONE;
   unsigned int i;
   static int in_array = 0;
   gl_register_file file = in_array ? PROGRAM_CONSTANT : PROGRAM_IMMEDIATE;

   /* Unfortunately, 4 floats is all we can get into
    * _mesa_add_typed_unnamed_constant.  So, make a temp to store an
    * aggregate constant and move each constant value into it.  If we
    * get lucky, copy propagation will eliminate the extra moves.
    */
   if (ir->type->base_type == GLSL_TYPE_STRUCT) {
      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();
         int size = type_size(field_value->type);

         assert(size > 0);

         field_value->accept(this);
         src = this->result;

         for (i = 0; i < (unsigned int)size; i++) {
            emit(ir, TGSI_OPCODE_MOV, temp, src);

            src.index++;
            temp.index++;
         }
      }
      this->result = temp_base;
      return;
   }

   if (ir->type->is_array()) {
      st_src_reg temp_base = get_temp(ir->type);
      st_dst_reg temp = st_dst_reg(temp_base);
      int size = type_size(ir->type->fields.array);

      assert(size > 0);
      in_array++;

      for (i = 0; i < ir->type->length; i++) {
         ir->array_elements[i]->accept(this);
         src = this->result;
         for (int j = 0; j < size; j++) {
            emit(ir, TGSI_OPCODE_MOV, temp, src);

            src.index++;
            temp.index++;
         }
      }
      this->result = temp_base;
      in_array--;
      return;
   }

   if (ir->type->is_matrix()) {
      st_src_reg mat = get_temp(ir->type);
      st_dst_reg mat_column = st_dst_reg(mat);

      for (i = 0; i < ir->type->matrix_columns; i++) {
         assert(ir->type->base_type == GLSL_TYPE_FLOAT);
         values = (gl_constant_value *) &ir->value.f[i * ir->type->vector_elements];

         src = st_src_reg(file, -1, ir->type->base_type);
         src.index = add_constant(file,
                                  values,
                                  ir->type->vector_elements,
                                  GL_FLOAT,
                                  &src.swizzle);
         emit(ir, TGSI_OPCODE_MOV, mat_column, src);

         mat_column.index++;
      }

      this->result = mat;
      return;
   }

   switch (ir->type->base_type) {
   case GLSL_TYPE_FLOAT:
      gl_type = GL_FLOAT;
      for (i = 0; i < ir->type->vector_elements; i++) {
         values[i].f = ir->value.f[i];
      }
      break;
   case GLSL_TYPE_UINT:
      gl_type = native_integers ? GL_UNSIGNED_INT : GL_FLOAT;
      for (i = 0; i < ir->type->vector_elements; i++) {
         if (native_integers)
            values[i].u = ir->value.u[i];
         else
            values[i].f = ir->value.u[i];
      }
      break;
   case GLSL_TYPE_INT:
      gl_type = native_integers ? GL_INT : GL_FLOAT;
      for (i = 0; i < ir->type->vector_elements; i++) {
         if (native_integers)
            values[i].i = ir->value.i[i];
         else
            values[i].f = ir->value.i[i];
      }
      break;
   case GLSL_TYPE_BOOL:
      gl_type = native_integers ? GL_BOOL : GL_FLOAT;
      for (i = 0; i < ir->type->vector_elements; i++) {
         if (native_integers)
            values[i].u = ir->value.b[i] ? ~0 : 0;
         else
            values[i].f = ir->value.b[i];
      }
      break;
   default:
      assert(!"Non-float/uint/int/bool constant");
   }

   this->result = st_src_reg(file, -1, ir->type);
   this->result.index = add_constant(file,
                                     values,
                                     ir->type->vector_elements,
                                     gl_type,
                                     &this->result.swizzle);
}

function_entry *
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();

      if (entry->sig == sig)
         return entry;
   }

   entry = ralloc(mem_ctx, function_entry);
   entry->sig = sig;
   entry->sig_id = this->next_signature_id++;
   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();
      variable_storage *storage;

      storage = find_variable_storage(param);
      assert(!storage);

      st_src_reg src = get_temp(param->type);

      storage = new(mem_ctx) variable_storage(param, src.file, src.index);
      this->variables.push_tail(storage);
   }

   if (!sig->return_type->is_void()) {
      entry->return_reg = get_temp(sig->return_type);
   } else {
      entry->return_reg = undef_src;
   }

   this->function_signatures.push_tail(entry);
   return entry;
}

void
glsl_to_tgsi_visitor::visit(ir_call *ir)
{
   glsl_to_tgsi_instruction *call_inst;
   ir_function_signature *sig = ir->callee;
   function_entry *entry = get_function_signature(sig);
   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();

      if (param->mode == ir_var_function_in ||
          param->mode == ir_var_function_inout) {
         variable_storage *storage = find_variable_storage(param);
         assert(storage);

         param_rval->accept(this);
         st_src_reg r = this->result;

         st_dst_reg l;
         l.file = storage->file;
         l.index = storage->index;
         l.reladdr = NULL;
         l.writemask = WRITEMASK_XYZW;
         l.cond_mask = COND_TR;

         for (i = 0; i < type_size(param->type); i++) {
            emit(ir, TGSI_OPCODE_MOV, l, r);
            l.index++;
            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();

      if (param->mode == ir_var_function_out ||
          param->mode == ir_var_function_inout) {
         variable_storage *storage = find_variable_storage(param);
         assert(storage);

         st_src_reg r;
         r.file = storage->file;
         r.index = storage->index;
         r.reladdr = NULL;
         r.swizzle = SWIZZLE_NOOP;
         r.negate = 0;

         param_rval->accept(this);
         st_dst_reg l = st_dst_reg(this->result);

         for (i = 0; i < type_size(param->type); i++) {
            emit(ir, TGSI_OPCODE_MOV, l, r);
            l.index++;
            r.index++;
         }
      }

      sig_iter.next();
   }
   assert(!sig_iter.has_next());

   /* Process return value. */
   this->result = entry->return_reg;
}

void
glsl_to_tgsi_visitor::visit(ir_texture *ir)
{
   st_src_reg result_src, coord, cube_sc, lod_info, projector, dx, dy, offset, sample_index;
   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;

   /* if we are a cube array sampler */
   if ((sampler_type->sampler_dimensionality == GLSL_SAMPLER_DIM_CUBE &&
        sampler_type->sampler_array)) {
      is_cube_array = true;
   }

   if (ir->coordinate) {
      ir->coordinate->accept(this);

      /* Put our coords in a temp.  We'll need to modify them for shadow,
       * projection, or LOD, so the only case we'd use it as is is if
       * we're doing plain old texturing.  The optimization passes on
       * glsl_to_tgsi_visitor should handle cleaning up our mess in that case.
       */
      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);
   }

   if (ir->projector) {
      ir->projector->accept(this);
      projector = this->result;
   }

   /* Storage for our result.  Ideally for an assignment we'd be using
    * the actual storage for the result here, instead.
    */
   result_src = get_temp(ir->type);
   result_dst = st_dst_reg(result_src);

   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 = 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 = 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 = this->result;
      }
      break;
   case ir_txd:
      opcode = TGSI_OPCODE_TXD;
      ir->lod_info.grad.dPdx->accept(this);
      dx = this->result;
      ir->lod_info.grad.dPdy->accept(this);
      dy = this->result;
      if (ir->offset) {
         ir->offset->accept(this);
         offset = this->result;
      }
      break;
   case ir_txs:
      opcode = TGSI_OPCODE_TXQ;
      ir->lod_info.lod->accept(this);
      lod_info = this->result;
      break;
   case ir_txf:
      opcode = TGSI_OPCODE_TXF;
      ir->lod_info.lod->accept(this);
      lod_info = this->result;
      if (ir->offset) {
         ir->offset->accept(this);
         offset = 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_lod:
      assert(!"Unexpected ir_lod opcode");
      break;
   }

   if (ir->projector) {
      if (opcode == TGSI_OPCODE_TEX) {
         /* Slot the projector in as the last component of the coord. */
         coord_dst.writemask = WRITEMASK_W;
         emit(ir, TGSI_OPCODE_MOV, coord_dst, projector);
         coord_dst.writemask = WRITEMASK_XYZW;
         opcode = TGSI_OPCODE_TXP;
      } else {
         st_src_reg coord_w = coord;
         coord_w.swizzle = SWIZZLE_WWWW;

         /* For the other TEX opcodes there's no projective version
          * since the last slot is taken up by LOD info.  Do the
          * projective divide now.
          */
         coord_dst.writemask = WRITEMASK_W;
         emit(ir, TGSI_OPCODE_RCP, coord_dst, projector);

         /* In the case where we have to project the coordinates "by hand,"
          * the shadow comparator value must also be projected.
          */
         st_src_reg tmp_src = coord;
         if (ir->shadow_comparitor) {
            /* Slot the shadow value in as the second to last component of the
             * coord.
             */
            ir->shadow_comparitor->accept(this);

            tmp_src = get_temp(glsl_type::vec4_type);
            st_dst_reg tmp_dst = st_dst_reg(tmp_src);

            /* Projective division not allowed for array samplers. */
            assert(!sampler_type->sampler_array);

            tmp_dst.writemask = WRITEMASK_Z;
            emit(ir, TGSI_OPCODE_MOV, tmp_dst, this->result);

            tmp_dst.writemask = WRITEMASK_XY;
            emit(ir, TGSI_OPCODE_MOV, tmp_dst, coord);
         }

         coord_dst.writemask = WRITEMASK_XYZ;
         emit(ir, TGSI_OPCODE_MUL, coord_dst, tmp_src, coord_w);

         coord_dst.writemask = WRITEMASK_XYZW;
         coord.swizzle = SWIZZLE_XYZW;
      }
   }

   /* If projection is done and the opcode is not TGSI_OPCODE_TXP, then the shadow
    * comparator was put in the correct place (and projected) by the code,
    * above, that handles by-hand projection.
    */
   if (ir->shadow_comparitor && (!ir->projector || opcode == TGSI_OPCODE_TXP)) {
      /* Slot the shadow value in as the second to last component of the
       * coord.
       */
      ir->shadow_comparitor->accept(this);

      if (is_cube_array) {
         cube_sc = get_temp(glsl_type::float_type);
         cube_sc_dst = st_dst_reg(cube_sc);
         cube_sc_dst.writemask = WRITEMASK_X;
         emit(ir, TGSI_OPCODE_MOV, cube_sc_dst, this->result);
         cube_sc_dst.writemask = WRITEMASK_X;
      }
      else {
         if ((sampler_type->sampler_dimensionality == GLSL_SAMPLER_DIM_2D &&
              sampler_type->sampler_array) ||
             sampler_type->sampler_dimensionality == GLSL_SAMPLER_DIM_CUBE) {
            coord_dst.writemask = WRITEMASK_W;
         } else {
            coord_dst.writemask = WRITEMASK_Z;
         }
         
         emit(ir, TGSI_OPCODE_MOV, coord_dst, this->result);
         coord_dst.writemask = WRITEMASK_XYZW;
      }
   }

   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;
      emit(ir, TGSI_OPCODE_MOV, coord_dst, lod_info);
      coord_dst.writemask = WRITEMASK_XYZW;
   }

   if (opcode == TGSI_OPCODE_TXD)
      inst = emit(ir, opcode, result_dst, coord, dx, dy);
   else if (opcode == TGSI_OPCODE_TXQ)
      inst = emit(ir, opcode, result_dst, lod_info);
   else if (opcode == TGSI_OPCODE_TXF) {
      inst = emit(ir, opcode, result_dst, coord);
   } else if (opcode == TGSI_OPCODE_TXL2 || opcode == TGSI_OPCODE_TXB2) {
      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 
      inst = emit(ir, opcode, result_dst, coord);

   if (ir->shadow_comparitor)
      inst->tex_shadow = GL_TRUE;

   inst->sampler = _mesa_get_sampler_uniform_value(ir->sampler,
                                                   this->shader_program,
                                                   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);
   }

   switch (sampler_type->sampler_dimensionality) {
   case GLSL_SAMPLER_DIM_1D:
      inst->tex_target = (sampler_type->sampler_array)
         ? TEXTURE_1D_ARRAY_INDEX : TEXTURE_1D_INDEX;
      break;
   case GLSL_SAMPLER_DIM_2D:
      inst->tex_target = (sampler_type->sampler_array)
         ? TEXTURE_2D_ARRAY_INDEX : TEXTURE_2D_INDEX;
      break;
   case GLSL_SAMPLER_DIM_3D:
      inst->tex_target = TEXTURE_3D_INDEX;
      break;
   case GLSL_SAMPLER_DIM_CUBE:
      inst->tex_target = (sampler_type->sampler_array)
         ? TEXTURE_CUBE_ARRAY_INDEX : TEXTURE_CUBE_INDEX;
      break;
   case GLSL_SAMPLER_DIM_RECT:
      inst->tex_target = TEXTURE_RECT_INDEX;
      break;
   case GLSL_SAMPLER_DIM_BUF:
      inst->tex_target = TEXTURE_BUFFER_INDEX;
      break;
   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.");
   }

   this->result = result_src;
}

void
glsl_to_tgsi_visitor::visit(ir_return *ir)
{
   if (ir->get_value()) {
      st_dst_reg l;
      int i;

      assert(current_function);

      ir->get_value()->accept(this);
      st_src_reg r = this->result;

      l = st_dst_reg(current_function->return_reg);

      for (i = 0; i < type_size(current_function->sig->return_type); i++) {
         emit(ir, TGSI_OPCODE_MOV, l, r);
         l.index++;
         r.index++;
      }
   }

   emit(ir, TGSI_OPCODE_RET);
}

void
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);
   } else {
      emit(ir, TGSI_OPCODE_KILP);
   }
}

void
glsl_to_tgsi_visitor::visit(ir_if *ir)
{
   unsigned if_opcode;
   glsl_to_tgsi_instruction *if_inst;

   ir->condition->accept(this);
   assert(this->result.file != PROGRAM_UNDEFINED);

   if_opcode = native_integers ? TGSI_OPCODE_UIF : TGSI_OPCODE_IF;

   if_inst = emit(ir->condition, if_opcode, undef_dst, this->result);

   this->instructions.push_tail(if_inst);

   visit_exec_list(&ir->then_instructions, this);

   if (!ir->else_instructions.is_empty()) {
      emit(ir->condition, TGSI_OPCODE_ELSE);
      visit_exec_list(&ir->else_instructions, this);
   }

   if_inst = emit(ir->condition, TGSI_OPCODE_ENDIF);
}

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_consts = false;
   glsl_version = 0;
   native_integers = false;
   mem_ctx = ralloc_context(NULL);
   ctx = NULL;
   prog = NULL;
   shader_program = NULL;
   options = NULL;
}

glsl_to_tgsi_visitor::~glsl_to_tgsi_visitor()
{
   ralloc_free(mem_ctx);
}

extern "C" void free_glsl_to_tgsi_visitor(glsl_to_tgsi_visitor *v)
{
   delete v;
}


/**
 * Count resources used by the given gpu program (number of texture
 * samplers, etc).
 */
static void
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();

      if (is_tex_instruction(inst->op)) {
         v->samplers_used |= 1 << inst->sampler;

         if (inst->tex_shadow) {
            prog->ShadowSamplers |= 1 << inst->sampler;
         }
      }
   }
   
   prog->SamplersUsed = v->samplers_used;

   if (v->shader_program != NULL)
      _mesa_update_shader_textures_used(v->shader_program, prog);
}

static void
set_uniform_initializer(struct gl_context *ctx, void *mem_ctx,
                        struct gl_shader_program *shader_program,
                        const char *name, const glsl_type *type,
                        ir_constant *val)
{
   if (type->is_record()) {
      ir_constant *field_constant;

      field_constant = (ir_constant *)val->components.get_head();

      for (unsigned int i = 0; i < type->length; i++) {
         const glsl_type *field_type = type->fields.structure[i].type;
         const char *field_name = ralloc_asprintf(mem_ctx, "%s.%s", name,
                                            type->fields.structure[i].name);
         set_uniform_initializer(ctx, mem_ctx, shader_program, field_name,
                                 field_type, field_constant);
         field_constant = (ir_constant *)field_constant->next;
      }
      return;
   }

   unsigned offset;
   unsigned index = _mesa_get_uniform_location(ctx, shader_program, name,
                                               &offset);
   if (offset == GL_INVALID_INDEX) {
      fail_link(shader_program,
                "Couldn't find uniform for initializer %s\n", name);
      return;
   }
   int loc = _mesa_uniform_merge_location_offset(index, offset);

   for (unsigned int i = 0; i < (type->is_array() ? type->length : 1); i++) {
      ir_constant *element;
      const glsl_type *element_type;
      if (type->is_array()) {
         element = val->array_elements[i];
         element_type = type->fields.array;
      } else {
         element = val;
         element_type = type;
      }

      void *values;

      if (element_type->base_type == GLSL_TYPE_BOOL) {
         int *conv = ralloc_array(mem_ctx, int, element_type->components());
         for (unsigned int j = 0; j < element_type->components(); j++) {
            conv[j] = element->value.b[j];
         }
         values = (void *)conv;
         element_type = glsl_type::get_instance(GLSL_TYPE_INT,
                                                element_type->vector_elements,
                                                1);
      } else {
         values = &element->value;
      }

      if (element_type->is_matrix()) {
         _mesa_uniform_matrix(ctx, shader_program,
                              element_type->matrix_columns,
                              element_type->vector_elements,
                              loc, 1, GL_FALSE, (GLfloat *)values);
      } else {
         _mesa_uniform(ctx, shader_program, loc, element_type->matrix_columns,
                       values, element_type->gl_type);
      }

      loc++;
   }
}

/**
 * Returns the mask of channels (bitmask of WRITEMASK_X,Y,Z,W) which
 * are read from the given src in this instruction
 */
static int
get_src_arg_mask(st_dst_reg dst, st_src_reg src)
{
   int read_mask = 0, comp;

   /* Now, given the src swizzle and the written channels, find which
    * components are actually read
    */
   for (comp = 0; comp < 4; ++comp) {
      const unsigned coord = GET_SWZ(src.swizzle, comp);
      ASSERT(coord < 4);
      if (dst.writemask & (1 << comp) && coord <= SWIZZLE_W)
         read_mask |= 1 << coord;
   }

   return read_mask;
}

/**
 * This pass replaces CMP T0, T1 T2 T0 with MOV T0, T2 when the CMP
 * instruction is the first instruction to write to register T0.  There are
 * several lowering passes done in GLSL IR (e.g. branches and
 * relative addressing) that create a large number of conditional assignments
 * that ir_to_mesa converts to CMP instructions like the one mentioned above.
 *
 * Here is why this conversion is safe:
 * CMP T0, T1 T2 T0 can be expanded to:
 * if (T1 < 0.0)
 *      MOV T0, T2;
 * else
 *      MOV T0, T0;
 *
 * If (T1 < 0.0) evaluates to true then our replacement MOV T0, T2 is the same
 * as the original program.  If (T1 < 0.0) evaluates to false, executing
 * MOV T0, T0 will store a garbage value in T0 since T0 is uninitialized.
 * Therefore, it doesn't matter that we are replacing MOV T0, T0 with MOV T0, T2
 * because any instruction that was going to read from T0 after this was going
 * to read a garbage value anyway.
 */
void
glsl_to_tgsi_visitor::simplify_cmp(void)
{
   unsigned *tempWrites;
   unsigned outputWrites[MAX_PROGRAM_OUTPUTS];

   tempWrites = new unsigned[MAX_TEMPS];
   if (!tempWrites) {
      return;
   }
   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();
      unsigned prevWriteMask = 0;

      /* Give up if we encounter relative addressing or flow control. */
      if (inst->dst.reladdr ||
          tgsi_get_opcode_info(inst->op)->is_branch ||
          inst->op == TGSI_OPCODE_BGNSUB ||
          inst->op == TGSI_OPCODE_CONT ||
          inst->op == TGSI_OPCODE_END ||
          inst->op == TGSI_OPCODE_ENDSUB ||
          inst->op == TGSI_OPCODE_RET) {
         break;
      }

      if (inst->dst.file == PROGRAM_OUTPUT) {
         assert(inst->dst.index < MAX_PROGRAM_OUTPUTS);
         prevWriteMask = outputWrites[inst->dst.index];
         outputWrites[inst->dst.index] |= inst->dst.writemask;
      } else if (inst->dst.file == PROGRAM_TEMPORARY) {
         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. */
      if (inst->op == TGSI_OPCODE_CMP
          && !(inst->dst.writemask & prevWriteMask)
          && inst->src[2].file == inst->dst.file
          && inst->src[2].index == inst->dst.index
          && inst->dst.writemask == get_src_arg_mask(inst->dst, inst->src[2])) {

         inst->op = TGSI_OPCODE_MOV;
         inst->src[0] = inst->src[1];
      }
   }

   delete [] tempWrites;
}

/* Replaces all references to a temporary register index with another index. */
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();
      unsigned j;
      
      for (j=0; j < num_inst_src_regs(inst->op); j++) {
         if (inst->src[j].file == PROGRAM_TEMPORARY && 
             inst->src[j].index == index) {
            inst->src[j].index = new_index;
         }
      }
      
      if (inst->dst.file == PROGRAM_TEMPORARY && inst->dst.index == index) {
         inst->dst.index = new_index;
      }
   }
}

int
glsl_to_tgsi_visitor::get_first_temp_read(int index)
{
   int depth = 0; /* loop depth */
   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();
      
      for (j=0; j < num_inst_src_regs(inst->op); j++) {
         if (inst->src[j].file == PROGRAM_TEMPORARY && 
             inst->src[j].index == index) {
            return (depth == 0) ? i : loop_start;
         }
      }
      
      if (inst->op == TGSI_OPCODE_BGNLOOP) {
         if(depth++ == 0)
            loop_start = i;
      } else if (inst->op == TGSI_OPCODE_ENDLOOP) {
         if (--depth == 0)
            loop_start = -1;
      }
      assert(depth >= 0);
      
      i++;
   }
   
   return -1;
}

int
glsl_to_tgsi_visitor::get_first_temp_write(int index)
{
   int depth = 0; /* loop depth */
   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();
      
      if (inst->dst.file == PROGRAM_TEMPORARY && inst->dst.index == index) {
         return (depth == 0) ? i : loop_start;
      }
      
      if (inst->op == TGSI_OPCODE_BGNLOOP) {
         if(depth++ == 0)
            loop_start = i;
      } else if (inst->op == TGSI_OPCODE_ENDLOOP) {
         if (--depth == 0)
            loop_start = -1;
      }
      assert(depth >= 0);
      
      i++;
   }
   
   return -1;
}

int
glsl_to_tgsi_visitor::get_last_temp_read(int index)
{
   int depth = 0; /* loop depth */
   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();
      
      for (j=0; j < num_inst_src_regs(inst->op); j++) {
         if (inst->src[j].file == PROGRAM_TEMPORARY && 
             inst->src[j].index == index) {
            last = (depth == 0) ? i : -2;
         }
      }
      
      if (inst->op == TGSI_OPCODE_BGNLOOP)
         depth++;
      else if (inst->op == TGSI_OPCODE_ENDLOOP)
         if (--depth == 0 && last == -2)
            last = i;
      assert(depth >= 0);
      
      i++;
   }
   
   assert(last >= -1);
   return last;
}

int
glsl_to_tgsi_visitor::get_last_temp_write(int index)
{
   int depth = 0; /* loop depth */
   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();
      
      if (inst->dst.file == PROGRAM_TEMPORARY && inst->dst.index == index)
         last = (depth == 0) ? i : -2;
      
      if (inst->op == TGSI_OPCODE_BGNLOOP)
         depth++;
      else if (inst->op == TGSI_OPCODE_ENDLOOP)
         if (--depth == 0 && last == -2)
            last = i;
      assert(depth >= 0);
      
      i++;
   }
   
   assert(last >= -1);
   return last;
}

/*
 * On a basic block basis, tracks available PROGRAM_TEMPORARY register
 * channels for copy propagation and updates following instructions to
 * use the original versions.
 *
 * The glsl_to_tgsi_visitor lazily produces code assuming that this pass
 * will occur.  As an example, a TXP production before this pass:
 *
 * 0: MOV TEMP[1], INPUT[4].xyyy;
 * 1: MOV TEMP[1].w, INPUT[4].wwww;
 * 2: TXP TEMP[2], TEMP[1], texture[0], 2D;
 *
 * and after:
 *
 * 0: MOV TEMP[1], INPUT[4].xyyy;
 * 1: MOV TEMP[1].w, INPUT[4].wwww;
 * 2: TXP TEMP[2], INPUT[4].xyyw, texture[0], 2D;
 *
 * which allows for dead code elimination on TEMP[1]'s writes.
 */
void
glsl_to_tgsi_visitor::copy_propagate(void)
{
   glsl_to_tgsi_instruction **acp = rzalloc_array(mem_ctx,
                                                    glsl_to_tgsi_instruction *,
                                                    this->next_temp * 4);
   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();

      assert(inst->dst.file != PROGRAM_TEMPORARY
             || inst->dst.index < this->next_temp);

      /* First, do any copy propagation possible into the src regs. */
      for (int r = 0; r < 3; r++) {
         glsl_to_tgsi_instruction *first = NULL;
         bool good = true;
         int acp_base = inst->src[r].index * 4;

         if (inst->src[r].file != PROGRAM_TEMPORARY ||
             inst->src[r].reladdr)
            continue;

         /* See if we can find entries in the ACP consisting of MOVs
          * from the same src register for all the swizzled channels
          * of this src register reference.
          */
         for (int i = 0; i < 4; i++) {
            int src_chan = GET_SWZ(inst->src[r].swizzle, i);
            glsl_to_tgsi_instruction *copy_chan = acp[acp_base + src_chan];

            if (!copy_chan) {
               good = false;
               break;
            }

            assert(acp_level[acp_base + src_chan] <= level);

            if (!first) {
               first = copy_chan;
            } else {
               if (first->src[0].file != copy_chan->src[0].file ||
                   first->src[0].index != copy_chan->src[0].index) {
                  good = false;
                  break;
               }
            }
         }

         if (good) {
            /* We've now validated that we can copy-propagate to
             * replace this src register reference.  Do it.
             */
            inst->src[r].file = first->src[0].file;
            inst->src[r].index = first->src[0].index;

            int swizzle = 0;
            for (int i = 0; i < 4; i++) {
               int src_chan = GET_SWZ(inst->src[r].swizzle, i);
               glsl_to_tgsi_instruction *copy_inst = acp[acp_base + src_chan];
               swizzle |= (GET_SWZ(copy_inst->src[0].swizzle, src_chan) <<
                           (3 * i));
            }
            inst->src[r].swizzle = swizzle;
         }
      }

      switch (inst->op) {
      case TGSI_OPCODE_BGNLOOP:
      case TGSI_OPCODE_ENDLOOP:
         /* End of a basic block, clear the ACP entirely. */
         memset(acp, 0, sizeof(*acp) * this->next_temp * 4);
         break;

      case TGSI_OPCODE_IF:
      case TGSI_OPCODE_UIF:
         ++level;
         break;

      case TGSI_OPCODE_ENDIF:
      case TGSI_OPCODE_ELSE:
         /* Clear all channels written inside the block from the ACP, but
          * leaving those that were not touched.
          */
         for (int r = 0; r < this->next_temp; r++) {
            for (int c = 0; c < 4; c++) {
               if (!acp[4 * r + c])
                  continue;

               if (acp_level[4 * r + c] >= level)
                  acp[4 * r + c] = NULL;
            }
         }
         if (inst->op == TGSI_OPCODE_ENDIF)
            --level;
         break;

      default:
         /* Continuing the block, clear any written channels from
          * the ACP.
          */
         if (inst->dst.file == PROGRAM_TEMPORARY && inst->dst.reladdr) {
            /* Any temporary might be written, so no copy propagation
             * across this instruction.
             */
            memset(acp, 0, sizeof(*acp) * this->next_temp * 4);
         } else if (inst->dst.file == PROGRAM_OUTPUT &&
                    inst->dst.reladdr) {
            /* Any output might be written, so no copy propagation
             * from outputs across this instruction.
             */
            for (int r = 0; r < this->next_temp; r++) {
               for (int c = 0; c < 4; c++) {
                  if (!acp[4 * r + c])
                     continue;

                  if (acp[4 * r + c]->src[0].file == PROGRAM_OUTPUT)
                     acp[4 * r + c] = NULL;
               }
            }
         } else if (inst->dst.file == PROGRAM_TEMPORARY ||
                    inst->dst.file == PROGRAM_OUTPUT) {
            /* Clear where it's used as dst. */
            if (inst->dst.file == PROGRAM_TEMPORARY) {
               for (int c = 0; c < 4; c++) {
                  if (inst->dst.writemask & (1 << c)) {
                     acp[4 * inst->dst.index + c] = NULL;
                  }
               }
            }

            /* Clear where it's used as src. */
            for (int r = 0; r < this->next_temp; r++) {
               for (int c = 0; c < 4; c++) {
                  if (!acp[4 * r + c])
                     continue;

                  int src_chan = GET_SWZ(acp[4 * r + c]->src[0].swizzle, c);

                  if (acp[4 * r + c]->src[0].file == inst->dst.file &&
                      acp[4 * r + c]->src[0].index == inst->dst.index &&
                      inst->dst.writemask & (1 << src_chan))
                  {
                     acp[4 * r + c] = NULL;
                  }
               }
            }
         }
         break;
      }

      /* If this is a copy, add it to the ACP. */
      if (inst->op == TGSI_OPCODE_MOV &&
          inst->dst.file == PROGRAM_TEMPORARY &&
          !inst->dst.reladdr &&
          !inst->saturate &&
          !inst->src[0].reladdr &&
          !inst->src[0].negate) {
         for (int i = 0; i < 4; i++) {
            if (inst->dst.writemask & (1 << i)) {
               acp[4 * inst->dst.index + i] = inst;
               acp_level[4 * inst->dst.index + i] = level;
            }
         }
      }
   }

   ralloc_free(acp_level);
   ralloc_free(acp);
}

/*
 * 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 
 * before this pass:
 *
 * 0: MOV TEMP[1], INPUT[4].xyyy;
 * 1: MOV TEMP[1].w, INPUT[4].wwww;
 * 2: TXP TEMP[2], INPUT[4].xyyw, texture[0], 2D;
 *
 * 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_instruction **writes = rzalloc_array(mem_ctx,
                                                     glsl_to_tgsi_instruction *,
                                                     this->next_temp * 4);
   int *write_level = rzalloc_array(mem_ctx, int, this->next_temp * 4);
   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();

      assert(inst->dst.file != PROGRAM_TEMPORARY
             || inst->dst.index < this->next_temp);
      
      switch (inst->op) {
      case TGSI_OPCODE_BGNLOOP:
      case TGSI_OPCODE_ENDLOOP:
      case TGSI_OPCODE_CONT:
      case TGSI_OPCODE_BRK:
         /* End of a basic block, clear the write array entirely.
          *
          * This keeps us from killing dead code when the writes are
          * on either side of a loop, even when the register isn't touched
          * inside the loop.  However, glsl_to_tgsi_visitor doesn't seem to emit
          * dead code of this type, so it shouldn't make a difference as long as
          * the dead code elimination pass in the GLSL compiler does its job.
          */
         memset(writes, 0, sizeof(*writes) * this->next_temp * 4);
         break;

      case TGSI_OPCODE_ENDIF:
      case TGSI_OPCODE_ELSE:
         /* Promote the recorded level of all channels written inside the
          * preceding if or else block to the level above the if/else block.
          */
         for (int r = 0; r < this->next_temp; r++) {
            for (int c = 0; c < 4; c++) {
               if (!writes[4 * r + c])
                         continue;

               if (write_level[4 * r + c] == level)
                         write_level[4 * r + c] = level-1;
            }
         }

         if(inst->op == TGSI_OPCODE_ENDIF)
            --level;
         
         break;

      case TGSI_OPCODE_IF:
      case TGSI_OPCODE_UIF:
         ++level;
         /* fallthrough to default case to mark the condition as read */
      
      default:
         /* Continuing the block, clear any channels from the write array that
          * are read by this instruction.
          */
         for (unsigned i = 0; i < Elements(inst->src); i++) {
            if (inst->src[i].file == PROGRAM_TEMPORARY && inst->src[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->src[i].file == PROGRAM_TEMPORARY) {
               /* Clear where it's used as src. */
               int src_chans = 1 << GET_SWZ(inst->src[i].swizzle, 0);
               src_chans |= 1 << GET_SWZ(inst->src[i].swizzle, 1);
               src_chans |= 1 << GET_SWZ(inst->src[i].swizzle, 2);
               src_chans |= 1 << GET_SWZ(inst->src[i].swizzle, 3);
               
               for (int c = 0; c < 4; c++) {
                   if (src_chans & (1 << c)) {
                      writes[4 * inst->src[i].index + c] = NULL;
                   }
               }
            }
         }
         break;
      }

      /* If this instruction writes to a temporary, add it to the write array.
       * If there is already an instruction in the write array for one or more
       * of the channels, flag that channel write as dead.
       */
      if (inst->dst.file == PROGRAM_TEMPORARY &&
          !inst->dst.reladdr &&
          !inst->saturate) {
         for (int c = 0; c < 4; c++) {
            if (inst->dst.writemask & (1 << c)) {
               if (writes[4 * inst->dst.index + c]) {
                  if (write_level[4 * inst->dst.index + c] < level)
                     continue;
                  else
                     writes[4 * inst->dst.index + c]->dead_mask |= (1 << c);
               }
               writes[4 * inst->dst.index + c] = inst;
               write_level[4 * inst->dst.index + c] = level;
            }
         }
      }
   }

   /* Anything still in the write array at this point is dead code. */
   for (int r = 0; r < this->next_temp; r++) {
      for (int c = 0; c < 4; c++) {
         glsl_to_tgsi_instruction *inst = writes[4 * r + c];
         if (inst)
            inst->dead_mask |= (1 << c);
      }
   }

   /* 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();
      
      if (!inst->dead_mask || !inst->dst.writemask)
         continue;
      else if ((inst->dst.writemask & ~inst->dead_mask) == 0) {
         iter.remove();
         delete inst;
         removed++;
      } else
         inst->dst.writemask &= ~(inst->dead_mask);
   }

   ralloc_free(write_level);
   ralloc_free(writes);
   
   return removed;
}

/* Merges temporary registers together where possible to reduce the number of 
 * registers needed to run a program.
 * 
 * Produces optimal code only after copy propagation and dead code elimination 
 * have been run. */
void
glsl_to_tgsi_visitor::merge_registers(void)
{
   int *last_reads = rzalloc_array(mem_ctx, int, this->next_temp);
   int *first_writes = rzalloc_array(mem_ctx, int, this->next_temp);
   int i, j;
   
   /* Read the indices of the last read and first write to each temp register
    * into an array so that we don't have to traverse the instruction list as 
    * much. */
   for (i=0; i < this->next_temp; i++) {
      last_reads[i] = get_last_temp_read(i);
      first_writes[i] = get_first_temp_write(i);
   }
   
   /* Start looking for registers with non-overlapping usages that can be 
    * merged together. */
   for (i=0; i < this->next_temp; i++) {
      /* Don't touch unused registers. */
      if (last_reads[i] < 0 || first_writes[i] < 0) continue;
      
      for (j=0; j < this->next_temp; j++) {
         /* Don't touch unused registers. */
         if (last_reads[j] < 0 || first_writes[j] < 0) continue;
         
         /* We can merge the two registers if the first write to j is after or 
          * in the same instruction as the last read from i.  Note that the 
          * register at index i will always be used earlier or at the same time 
          * as the register at index j. */
         if (first_writes[i] <= first_writes[j] && 
             last_reads[i] <= first_writes[j])
         {
            rename_temp_register(j, i); /* Replace all references to j with i.*/
            
            /* Update the first_writes and last_reads arrays with the new 
             * values for the merged register index, and mark the newly unused 
             * register index as such. */
            last_reads[i] = last_reads[j];
            first_writes[j] = -1;
            last_reads[j] = -1;
         }
      }
   }
   
   ralloc_free(last_reads);
   ralloc_free(first_writes);
}

/* Reassign indices to temporary registers by reusing unused indices created 
 * by optimization passes. */
void
glsl_to_tgsi_visitor::renumber_registers(void)
{
   int i = 0;
   int new_index = 0;
   
   for (i=0; i < this->next_temp; i++) {
      if (get_first_temp_read(i) < 0) continue;
      if (i != new_index)
         rename_temp_register(i, new_index);
      new_index++;
   }
   
   this->next_temp = new_index;
}

/**
 * Returns a fragment program which implements the current pixel transfer ops.
 * Based on get_pixel_transfer_program in st_atom_pixeltransfer.c.
 */
extern "C" void
get_pixel_transfer_visitor(struct st_fragment_program *fp,
                           glsl_to_tgsi_visitor *original,
                           int scale_and_bias, int pixel_maps)
{
   glsl_to_tgsi_visitor *v = new glsl_to_tgsi_visitor();
   struct st_context *st = st_context(original->ctx);
   struct gl_program *prog = &fp->Base.Base;
   struct gl_program_parameter_list *params = _mesa_new_parameter_list();
   st_src_reg coord, src0;
   st_dst_reg dst0;
   glsl_to_tgsi_instruction *inst;

   /* Copy attributes of the glsl_to_tgsi_visitor in the original shader. */
   v->ctx = original->ctx;
   v->prog = prog;
   v->shader_program = NULL;
   v->glsl_version = original->glsl_version;
   v->native_integers = original->native_integers;
   v->options = original->options;
   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_consts = original->indirect_addr_consts;
   memcpy(&v->immediates, &original->immediates, sizeof(v->immediates));
   v->num_immediates = original->num_immediates;

   /*
    * Get initial pixel color from the texture.
    * TEX colorTemp, fragment.texcoord[0], texture[0], 2D;
    */
   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 |= VARYING_BIT_TEX0;
   prog->SamplersUsed |= (1 << 0); /* mark sampler 0 as used */
   v->samplers_used |= (1 << 0);

   if (scale_and_bias) {
      static const gl_state_index scale_state[STATE_LENGTH] =
         { STATE_INTERNAL, STATE_PT_SCALE,
           (gl_state_index) 0, (gl_state_index) 0, (gl_state_index) 0 };
      static const gl_state_index bias_state[STATE_LENGTH] =
         { STATE_INTERNAL, STATE_PT_BIAS,
           (gl_state_index) 0, (gl_state_index) 0, (gl_state_index) 0 };
      GLint scale_p, bias_p;
      st_src_reg scale, bias;

      scale_p = _mesa_add_state_reference(params, scale_state);
      bias_p = _mesa_add_state_reference(params, bias_state);

      /* MAD colorTemp, colorTemp, scale, bias; */
      scale = st_src_reg(PROGRAM_STATE_VAR, scale_p, GLSL_TYPE_FLOAT);
      bias = st_src_reg(PROGRAM_STATE_VAR, bias_p, GLSL_TYPE_FLOAT);
      inst = v->emit(NULL, TGSI_OPCODE_MAD, dst0, src0, scale, bias);
   }

   if (pixel_maps) {
      st_src_reg temp = v->get_temp(glsl_type::vec4_type);
      st_dst_reg temp_dst = st_dst_reg(temp);

      assert(st->pixel_xfer.pixelmap_texture);

      /* With a little effort, we can do four pixel map look-ups with
       * two TEX instructions:
       */

      /* TEX temp.rg, colorTemp.rgba, texture[1], 2D; */
      temp_dst.writemask = WRITEMASK_XY; /* write R,G */
      inst = v->emit(NULL, TGSI_OPCODE_TEX, temp_dst, src0);
      inst->sampler = 1;
      inst->tex_target = TEXTURE_2D_INDEX;

      /* TEX temp.ba, colorTemp.baba, texture[1], 2D; */
      src0.swizzle = MAKE_SWIZZLE4(SWIZZLE_Z, SWIZZLE_W, SWIZZLE_Z, SWIZZLE_W);
      temp_dst.writemask = WRITEMASK_ZW; /* write B,A */
      inst = v->emit(NULL, TGSI_OPCODE_TEX, temp_dst, src0);
      inst->sampler = 1;
      inst->tex_target = TEXTURE_2D_INDEX;

      prog->SamplersUsed |= (1 << 1); /* mark sampler 1 as used */
      v->samplers_used |= (1 << 1);

      /* MOV colorTemp, temp; */
      inst = v->emit(NULL, TGSI_OPCODE_MOV, dst0, temp);
   }

   /* 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();
      glsl_to_tgsi_instruction *newinst;
      st_src_reg src_regs[3];

      if (inst->dst.file == PROGRAM_OUTPUT)
         prog->OutputsWritten |= BITFIELD64_BIT(inst->dst.index);

      for (int i=0; i<3; i++) {
         src_regs[i] = inst->src[i];
         if (src_regs[i].file == PROGRAM_INPUT &&
             src_regs[i].index == VARYING_SLOT_COL0)
         {
            src_regs[i].file = PROGRAM_TEMPORARY;
            src_regs[i].index = src0.index;
         }
         else if (src_regs[i].file == PROGRAM_INPUT)
            prog->InputsRead |= BITFIELD64_BIT(src_regs[i].index);
      }

      newinst = v->emit(NULL, inst->op, inst->dst, src_regs[0], src_regs[1], src_regs[2]);
      newinst->tex_target = inst->tex_target;
   }

   /* Make modifications to fragment program info. */
   prog->Parameters = _mesa_combine_parameter_lists(params,
                                                    original->prog->Parameters);
   _mesa_free_parameter_list(params);
   count_resources(v, prog);
   fp->glsl_to_tgsi = v;
}

/**
 * Make fragment program for glBitmap:
 *   Sample the texture and kill the fragment if the bit is 0.
 * This program will be combined with the user's fragment program.
 *
 * Based on make_bitmap_fragment_program in st_cb_bitmap.c.
 */
extern "C" void
get_bitmap_visitor(struct st_fragment_program *fp,
                   glsl_to_tgsi_visitor *original, int samplerIndex)
{
   glsl_to_tgsi_visitor *v = new glsl_to_tgsi_visitor();
   struct st_context *st = st_context(original->ctx);
   struct gl_program *prog = &fp->Base.Base;
   st_src_reg coord, src0;
   st_dst_reg dst0;
   glsl_to_tgsi_instruction *inst;

   /* Copy attributes of the glsl_to_tgsi_visitor in the original shader. */
   v->ctx = original->ctx;
   v->prog = prog;
   v->shader_program = NULL;
   v->glsl_version = original->glsl_version;
   v->native_integers = original->native_integers;
   v->options = original->options;
   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_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, 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 |= VARYING_BIT_TEX0;
   prog->SamplersUsed |= (1 << samplerIndex); /* mark sampler as used */
   v->samplers_used |= (1 << samplerIndex);

   /* KIL if -tmp0 < 0 # texel=0 -> keep / texel=0 -> discard */
   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);

   /* 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();
      glsl_to_tgsi_instruction *newinst;
      st_src_reg src_regs[3];

      if (inst->dst.file == PROGRAM_OUTPUT)
         prog->OutputsWritten |= BITFIELD64_BIT(inst->dst.index);

      for (int i=0; i<3; i++) {
         src_regs[i] = inst->src[i];
         if (src_regs[i].file == PROGRAM_INPUT)
            prog->InputsRead |= BITFIELD64_BIT(src_regs[i].index);
      }

      newinst = v->emit(NULL, inst->op, inst->dst, src_regs[0], src_regs[1], src_regs[2]);
      newinst->tex_target = inst->tex_target;
   }

   /* Make modifications to fragment program info. */
   prog->Parameters = _mesa_clone_parameter_list(original->prog->Parameters);
   count_resources(v, prog);
   fp->glsl_to_tgsi = v;
}

/* ------------------------- TGSI conversion stuff -------------------------- */
struct label {
   unsigned branch_target;
   unsigned token;
};

/**
 * Intermediate state used during shader translation.
 */
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_src samplers[PIPE_MAX_SAMPLERS];
   struct ureg_src systemValues[SYSTEM_VALUE_MAX];

   unsigned array_sizes[MAX_ARRAYS];

   const GLuint *inputMapping;
   const GLuint *outputMapping;

   /* For every instruction that contains a label (eg CALL), keep
    * details so that we can go back afterwards and emit the correct
    * tgsi instruction number for each label.
    */
   struct label *labels;
   unsigned labels_size;
   unsigned labels_count;

   /* Keep a record of the tgsi instruction number that each mesa
    * instruction starts at, will be used to fix up labels after
    * translation.
    */
   unsigned *insn;
   unsigned insn_size;
   unsigned insn_count;

   unsigned procType;  /**< TGSI_PROCESSOR_VERTEX/FRAGMENT */

   boolean error;
};

/** Map Mesa's SYSTEM_VALUE_x to TGSI_SEMANTIC_x */
static unsigned mesa_sysval_to_semantic[SYSTEM_VALUE_MAX] = {
   TGSI_SEMANTIC_FACE,
   TGSI_SEMANTIC_VERTEXID,
   TGSI_SEMANTIC_INSTANCEID
};

/**
 * Make note of a branch to a label in the TGSI code.
 * After we've emitted all instructions, we'll go over the list
 * of labels built here and patch the TGSI code with the actual
 * location of each label.
 */
static unsigned *get_label(struct st_translate *t, unsigned branch_target)
{
   unsigned i;

   if (t->labels_count + 1 >= t->labels_size) {
      t->labels_size = 1 << (util_logbase2(t->labels_size) + 1);
      t->labels = (struct label *)realloc(t->labels, 
                                          t->labels_size * sizeof(struct label));
      if (t->labels == NULL) {
         static unsigned dummy;
         t->error = TRUE;
         return &dummy;
      }
   }

   i = t->labels_count++;
   t->labels[i].branch_target = branch_target;
   return &t->labels[i].token;
}

/**
 * Called prior to emitting the TGSI code for each instruction.
 * Allocate additional space for instructions if needed.
 * Update the insn[] array so the next glsl_to_tgsi_instruction points to
 * the next TGSI instruction.
 */
static void set_insn_start(struct st_translate *t, unsigned start)
{
   if (t->insn_count + 1 >= t->insn_size) {
      t->insn_size = 1 << (util_logbase2(t->insn_size) + 1);
      t->insn = (unsigned *)realloc(t->insn, t->insn_size * sizeof(t->insn[0]));
      if (t->insn == NULL) {
         t->error = TRUE;
         return;
      }
   }

   t->insn[t->insn_count++] = start;
}

/**
 * Map a glsl_to_tgsi constant/immediate to a TGSI immediate.
 */
static struct ureg_src
emit_immediate(struct st_translate *t,
               gl_constant_value values[4],
               int type, int size)
{
   struct ureg_program *ureg = t->ureg;

   switch(type)
   {
   case GL_FLOAT:
      return ureg_DECL_immediate(ureg, &values[0].f, size);
   case GL_INT:
      return ureg_DECL_immediate_int(ureg, &values[0].i, size);
   case GL_UNSIGNED_INT:
   case GL_BOOL:
      return ureg_DECL_immediate_uint(ureg, &values[0].u, size);
   default:
      assert(!"should not get here - type must be float, int, uint, or bool");
      return ureg_src_undef();
   }
}

/**
 * Map a glsl_to_tgsi dst register to a TGSI ureg_dst register.
 */
static struct ureg_dst
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 < VARYING_SLOT_MAX);
      else if (t->procType == TGSI_PROCESSOR_FRAGMENT)
         assert(index < FRAG_RESULT_MAX);
      else
         assert(index < VARYING_SLOT_MAX);

      assert(t->outputMapping[index] < Elements(t->outputs));

      return t->outputs[t->outputMapping[index]];

   case PROGRAM_ADDRESS:
      return t->address[index];

   default:
      assert(!"unknown dst register file");
      return ureg_dst_undef();
   }
}

/**
 * Map a glsl_to_tgsi src register to a TGSI ureg_src register.
 */
static struct ureg_src
src_register(struct st_translate *t,
             gl_register_file file,
             GLint index, GLint index2D)
{
   switch(file) {
   case PROGRAM_UNDEFINED:
      return ureg_src_undef();

   case PROGRAM_TEMPORARY:
   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];
   case PROGRAM_STATE_VAR:
   case PROGRAM_CONSTANT:       /* ie, immediate */
      if (index2D) {
         struct ureg_src src;
         src = ureg_src_register(TGSI_FILE_CONSTANT, 0);
         src.Dimension = 1;
         src.DimensionIndex = index2D;
         return src;
      } else if (index < 0)
         return ureg_DECL_constant(t->ureg, 0);
      else
         return t->constants[index];

   case PROGRAM_IMMEDIATE:
      return t->immediates[index];

   case PROGRAM_INPUT:
      assert(t->inputMapping[index] < Elements(t->inputs));
      return t->inputs[t->inputMapping[index]];

   case PROGRAM_OUTPUT:
      assert(t->outputMapping[index] < Elements(t->outputs));
      return ureg_src(t->outputs[t->outputMapping[index]]); /* not needed? */

   case PROGRAM_ADDRESS:
      return ureg_src(t->address[index]);

   case PROGRAM_SYSTEM_VALUE:
      assert(index < (int) Elements(t->systemValues));
      return t->systemValues[index];

   default:
      assert(!"unknown src register file");
      return ureg_src_undef();
   }
}

/**
 * Create a TGSI ureg_dst register from an st_dst_reg.
 */
static struct ureg_dst
translate_dst(struct st_translate *t,
              const st_dst_reg *dst_reg,
              bool saturate, bool clamp_color)
{
   struct ureg_dst dst = dst_register(t, 
                                      dst_reg->file,
                                      dst_reg->index);

   dst = ureg_writemask(dst, dst_reg->writemask);
   
   if (saturate)
      dst = ureg_saturate(dst);
   else if (clamp_color && dst_reg->file == PROGRAM_OUTPUT) {
      /* Clamp colors for ARB_color_buffer_float. */
      switch (t->procType) {
      case TGSI_PROCESSOR_VERTEX:
         /* XXX if the geometry shader is present, this must be done there
          * instead of here. */
         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) {
            dst = ureg_saturate(dst);
         }
         break;
      }
   }

   if (dst_reg->reladdr != NULL) {
      assert(dst_reg->file != PROGRAM_TEMPORARY);
      dst = ureg_dst_indirect(dst, ureg_src(t->address[0]));
   }

   return dst;
}

/**
 * Create a TGSI ureg_src register from an st_src_reg.
 */
static struct ureg_src
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);

   src = ureg_swizzle(src,
                      GET_SWZ(src_reg->swizzle, 0) & 0x3,
                      GET_SWZ(src_reg->swizzle, 1) & 0x3,
                      GET_SWZ(src_reg->swizzle, 2) & 0x3,
                      GET_SWZ(src_reg->swizzle, 3) & 0x3);

   if ((src_reg->negate & 0xf) == NEGATE_XYZW)
      src = ureg_negate(src);

   if (src_reg->reladdr != NULL) {
      assert(src_reg->file != PROGRAM_TEMPORARY);
      src = ureg_src_indirect(src, ureg_src(t->address[0]));
   }

   return src;
}

static struct tgsi_texture_offset
translate_tex_offset(struct st_translate *t,
                     const struct tgsi_texture_offset *in_offset)
{
   struct tgsi_texture_offset offset;
   struct ureg_src imm_src;

   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;

   return offset;
}

static void
compile_tgsi_instruction(struct st_translate *t,
                         const glsl_to_tgsi_instruction *inst,
                         bool clamp_dst_color_output)
{
   struct ureg_program *ureg = t->ureg;
   GLuint i;
   struct ureg_dst dst[1];
   struct ureg_src src[4];
   struct tgsi_texture_offset texoffsets[MAX_GLSL_TEXTURE_OFFSET];

   unsigned num_dst;
   unsigned num_src;
   unsigned tex_target;

   num_dst = num_inst_dst_regs(inst->op);
   num_src = num_inst_src_regs(inst->op);

   if (num_dst) 
      dst[0] = translate_dst(t, 
                             &inst->dst,
                             inst->saturate,
                             clamp_dst_color_output);

   for (i = 0; i < num_src; i++) 
      src[i] = translate_src(t, &inst->src[i]);

   switch(inst->op) {
   case TGSI_OPCODE_BGNLOOP:
   case TGSI_OPCODE_CAL:
   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,
                      src, num_src,
                      get_label(t, 
                                inst->op == TGSI_OPCODE_CAL ? inst->function->sig_id : 0));
      return;

   case TGSI_OPCODE_TEX:
   case TGSI_OPCODE_TXB:
   case TGSI_OPCODE_TXD:
   case TGSI_OPCODE_TXL:
   case TGSI_OPCODE_TXP:
   case TGSI_OPCODE_TXQ:
   case TGSI_OPCODE_TXF:
   case TGSI_OPCODE_TEX2:
   case TGSI_OPCODE_TXB2:
   case TGSI_OPCODE_TXL2:
      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]);
      }
      tex_target = st_translate_texture_target(inst->tex_target, inst->tex_shadow);

      ureg_tex_insn(ureg,
                    inst->op,
                    dst, num_dst,
                    tex_target,
                    texoffsets, inst->tex_offset_num_offset,
                    src, num_src);
      return;

   case TGSI_OPCODE_SCS:
      dst[0] = ureg_writemask(dst[0], TGSI_WRITEMASK_XY);
      ureg_insn(ureg, inst->op, dst, num_dst, src, num_src);
      break;

   default:
      ureg_insn(ureg,
                inst->op,
                dst, num_dst,
                src, num_src);
      break;
   }
}

/**
 * Emit the TGSI instructions for inverting and adjusting WPOS.
 * This code is unavoidable because it also depends on whether
 * a FBO is bound (STATE_FB_WPOS_Y_TRANSFORM).
 */
static void
emit_wpos_adjustment( struct st_translate *t,
                      const struct gl_program *program,
                      boolean invert,
                      GLfloat adjX, GLfloat adjY[2])
{
   struct ureg_program *ureg = t->ureg;

   /* Fragment program uses fragment position input.
    * Need to replace instances of INPUT[WPOS] with temp T
    * where T = INPUT[WPOS] by y is inverted.
    */
   static const gl_state_index wposTransformState[STATE_LENGTH]
      = { STATE_INTERNAL, STATE_FB_WPOS_Y_TRANSFORM, 
          (gl_state_index)0, (gl_state_index)0, (gl_state_index)0 };
   
   /* XXX: note we are modifying the incoming shader here!  Need to
    * do this before emitting the constant decls below, or this
    * will be missed:
    */
   unsigned wposTransConst = _mesa_add_state_reference(program->Parameters,
                                                       wposTransformState);

   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[VARYING_SLOT_POS]];

   /* First, apply the coordinate shift: */
   if (adjX || adjY[0] || adjY[1]) {
      if (adjY[0] != adjY[1]) {
         /* Adjust the y coordinate by adjY[1] or adjY[0] respectively
          * depending on whether inversion is actually going to be applied
          * or not, which is determined by testing against the inversion
          * state variable used below, which will be either +1 or -1.
          */
         struct ureg_dst adj_temp = ureg_DECL_local_temporary(ureg);

         ureg_CMP(ureg, adj_temp,
                  ureg_scalar(wpostrans, invert ? 2 : 0),
                  ureg_imm4f(ureg, adjX, adjY[0], 0.0f, 0.0f),
                  ureg_imm4f(ureg, adjX, adjY[1], 0.0f, 0.0f));
         ureg_ADD(ureg, wpos_temp, wpos_input, ureg_src(adj_temp));
      } else {
         ureg_ADD(ureg, wpos_temp, wpos_input,
                  ureg_imm4f(ureg, adjX, adjY[0], 0.0f, 0.0f));
      }
      wpos_input = ureg_src(wpos_temp);
   } else {
      /* MOV wpos_temp, input[wpos]
       */
      ureg_MOV( ureg, wpos_temp, wpos_input );
   }

   /* Now the conditional y flip: STATE_FB_WPOS_Y_TRANSFORM.xy/zw will be
    * inversion/identity, or the other way around if we're drawing to an FBO.
    */
   if (invert) {
      /* MAD wpos_temp.y, wpos_input, wpostrans.xxxx, wpostrans.yyyy
       */
      ureg_MAD( ureg,
                ureg_writemask(wpos_temp, TGSI_WRITEMASK_Y ),
                wpos_input,
                ureg_scalar(wpostrans, 0),
                ureg_scalar(wpostrans, 1));
   } else {
      /* MAD wpos_temp.y, wpos_input, wpostrans.zzzz, wpostrans.wwww
       */
      ureg_MAD( ureg,
                ureg_writemask(wpos_temp, TGSI_WRITEMASK_Y ),
                wpos_input,
                ureg_scalar(wpostrans, 2),
                ureg_scalar(wpostrans, 3));
   }

   /* Use wpos_temp as position input from here on:
    */
   t->inputs[t->inputMapping[VARYING_SLOT_POS]] = ureg_src(wpos_temp);
}


/**
 * Emit fragment position/ooordinate code.
 */
static void
emit_wpos(struct st_context *st,
          struct st_translate *t,
          const struct gl_program *program,
          struct ureg_program *ureg)
{
   const struct gl_fragment_program *fp =
      (const struct gl_fragment_program *) program;
   struct pipe_screen *pscreen = st->pipe->screen;
   GLfloat adjX = 0.0f;
   GLfloat adjY[2] = { 0.0f, 0.0f };
   boolean invert = FALSE;

   /* Query the pixel center conventions supported by the pipe driver and set
    * adjX, adjY to help out if it cannot handle the requested one internally.
    *
    * The bias of the y-coordinate depends on whether y-inversion takes place
    * (adjY[1]) or not (adjY[0]), which is in turn dependent on whether we are
    * drawing to an FBO (causes additional inversion), and whether the the pipe
    * driver origin and the requested origin differ (the latter condition is
    * stored in the 'invert' variable).
    *
    * For height = 100 (i = integer, h = half-integer, l = lower, u = upper):
    *
    * center shift only:
    * i -> h: +0.5
    * h -> i: -0.5
    *
    * inversion only:
    * l,i -> u,i: ( 0.0 + 1.0) * -1 + 100 = 99
    * l,h -> u,h: ( 0.5 + 0.0) * -1 + 100 = 99.5
    * u,i -> l,i: (99.0 + 1.0) * -1 + 100 = 0
    * u,h -> l,h: (99.5 + 0.0) * -1 + 100 = 0.5
    *
    * inversion and center shift:
    * l,i -> u,h: ( 0.0 + 0.5) * -1 + 100 = 99.5
    * l,h -> u,i: ( 0.5 + 0.5) * -1 + 100 = 99
    * u,i -> l,h: (99.0 + 0.5) * -1 + 100 = 0.5
    * u,h -> l,i: (99.5 + 0.5) * -1 + 100 = 0
    */
   if (fp->OriginUpperLeft) {
      /* Fragment shader wants origin in upper-left */
      if (pscreen->get_param(pscreen, PIPE_CAP_TGSI_FS_COORD_ORIGIN_UPPER_LEFT)) {
         /* the driver supports upper-left origin */
      }
      else if (pscreen->get_param(pscreen, PIPE_CAP_TGSI_FS_COORD_ORIGIN_LOWER_LEFT)) {
         /* the driver supports lower-left origin, need to invert Y */
         ureg_property_fs_coord_origin(ureg, TGSI_FS_COORD_ORIGIN_LOWER_LEFT);
         invert = TRUE;
      }
      else
         assert(0);
   }
   else {
      /* Fragment shader wants origin in lower-left */
      if (pscreen->get_param(pscreen, PIPE_CAP_TGSI_FS_COORD_ORIGIN_LOWER_LEFT))
         /* the driver supports lower-left origin */
         ureg_property_fs_coord_origin(ureg, TGSI_FS_COORD_ORIGIN_LOWER_LEFT);
      else if (pscreen->get_param(pscreen, PIPE_CAP_TGSI_FS_COORD_ORIGIN_UPPER_LEFT))
         /* the driver supports upper-left origin, need to invert Y */
         invert = TRUE;
      else
         assert(0);
   }
   
   if (fp->PixelCenterInteger) {
      /* Fragment shader wants pixel center integer */
      if (pscreen->get_param(pscreen, PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_INTEGER)) {
         /* the driver supports pixel center integer */
         adjY[1] = 1.0f;
         ureg_property_fs_coord_pixel_center(ureg, TGSI_FS_COORD_PIXEL_CENTER_INTEGER);
      }
      else if (pscreen->get_param(pscreen, PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_HALF_INTEGER)) {
         /* the driver supports pixel center half integer, need to bias X,Y */
         adjX = -0.5f;
         adjY[0] = -0.5f;
         adjY[1] = 0.5f;
      }
      else
         assert(0);
   }
   else {
      /* Fragment shader wants pixel center half integer */
      if (pscreen->get_param(pscreen, PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_HALF_INTEGER)) {
         /* the driver supports pixel center half integer */
      }
      else if (pscreen->get_param(pscreen, PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_INTEGER)) {
         /* the driver supports pixel center integer, need to bias X,Y */
         adjX = adjY[0] = adjY[1] = 0.5f;
         ureg_property_fs_coord_pixel_center(ureg, TGSI_FS_COORD_PIXEL_CENTER_INTEGER);
      }
      else
         assert(0);
   }

   /* we invert after adjustment so that we avoid the MOV to temporary,
    * and reuse the adjustment ADD instead */
   emit_wpos_adjustment(t, program, invert, adjX, adjY);
}

/**
 * OpenGL's fragment gl_FrontFace input is 1 for front-facing, 0 for back.
 * TGSI uses +1 for front, -1 for back.
 * This function converts the TGSI value to the GL value.  Simply clamping/
 * saturating the value to [0,1] does the job.
 */
static void
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[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[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[VARYING_SLOT_EDGE]];
   struct ureg_src edge_src = t->inputs[t->inputMapping[VERT_ATTRIB_EDGEFLAG]];

   ureg_MOV(ureg, edge_dst, edge_src);
}

/**
 * Translate intermediate IR (glsl_to_tgsi_instruction) to TGSI format.
 * \param program  the program to translate
 * \param numInputs  number of input registers used
 * \param inputMapping  maps Mesa fragment program inputs to TGSI generic
 *                      input indexes
 * \param inputSemanticName  the TGSI_SEMANTIC flag for each input
 * \param inputSemanticIndex  the semantic index (ex: which texcoord) for
 *                            each input
 * \param interpMode  the TGSI_INTERPOLATE_LINEAR/PERSP mode for each input
 * \param numOutputs  number of output registers used
 * \param outputMapping  maps Mesa fragment program outputs to TGSI
 *                       generic outputs
 * \param outputSemanticName  the TGSI_SEMANTIC flag for each output
 * \param outputSemanticIndex  the semantic index (ex: which texcoord) for
 *                             each output
 *
 * \return  PIPE_OK or PIPE_ERROR_OUT_OF_MEMORY
 */
extern "C" enum pipe_error
st_translate_program(
   struct gl_context *ctx,
   uint procType,
   struct ureg_program *ureg,
   glsl_to_tgsi_visitor *program,
   const struct gl_program *proginfo,
   GLuint numInputs,
   const GLuint inputMapping[],
   const ubyte inputSemanticName[],
   const ubyte inputSemanticIndex[],
   const GLuint interpMode[],
   const GLboolean is_centroid[],
   GLuint numOutputs,
   const GLuint outputMapping[],
   const ubyte outputSemanticName[],
   const ubyte outputSemanticIndex[],
   boolean passthrough_edgeflags,
   boolean clamp_color)
{
   struct st_translate *t;
   unsigned i;
   enum pipe_error ret = PIPE_OK;

   assert(numInputs <= Elements(t->inputs));
   assert(numOutputs <= Elements(t->outputs));

   t = CALLOC_STRUCT(st_translate);
   if (!t) {
      ret = PIPE_ERROR_OUT_OF_MEMORY;
      goto out;
   }

   memset(t, 0, sizeof *t);

   t->procType = procType;
   t->inputMapping = inputMapping;
   t->outputMapping = outputMapping;
   t->ureg = ureg;

   if (program->shader_program) {
      for (i = 0; i < program->shader_program->NumUserUniformStorage; i++) {
         struct gl_uniform_storage *const storage =
               &program->shader_program->UniformStorage[i];

         _mesa_uniform_detach_all_driver_storage(storage);
      }
   }

   /*
    * Declare input attributes.
    */
   if (procType == TGSI_PROCESSOR_FRAGMENT) {
      for (i = 0; i < numInputs; i++) {
         t->inputs[i] = ureg_DECL_fs_input_cyl_centroid(ureg,
                                                        inputSemanticName[i],
                                                        inputSemanticIndex[i],
                                                        interpMode[i], 0,
                                                        is_centroid[i]);
      }

      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 & VARYING_BIT_FACE)
         emit_face_var(t);

      /*
       * Declare output attributes.
       */
      for (i = 0; i < numOutputs; i++) {
         switch (outputSemanticName[i]) {
         case TGSI_SEMANTIC_POSITION:
            t->outputs[i] = ureg_DECL_output(ureg,
                                             TGSI_SEMANTIC_POSITION, /* Z/Depth */
                                             outputSemanticIndex[i]);
            t->outputs[i] = ureg_writemask(t->outputs[i], TGSI_WRITEMASK_Z);
            break;
         case TGSI_SEMANTIC_STENCIL:
            t->outputs[i] = ureg_DECL_output(ureg,
                                             TGSI_SEMANTIC_STENCIL, /* Stencil */
                                             outputSemanticIndex[i]);
            t->outputs[i] = ureg_writemask(t->outputs[i], TGSI_WRITEMASK_Y);
            break;
         case TGSI_SEMANTIC_COLOR:
            t->outputs[i] = ureg_DECL_output(ureg,
                                             TGSI_SEMANTIC_COLOR,
                                             outputSemanticIndex[i]);
            break;
         default:
            assert(!"fragment shader outputs must be POSITION/STENCIL/COLOR");
            ret = PIPE_ERROR_BAD_INPUT;
            goto out;
         }
      }
   }
   else if (procType == TGSI_PROCESSOR_GEOMETRY) {
      for (i = 0; i < numInputs; i++) {
         t->inputs[i] = ureg_DECL_gs_input(ureg,
                                           i,
                                           inputSemanticName[i],
                                           inputSemanticIndex[i]);
      }

      for (i = 0; i < numOutputs; i++) {
         t->outputs[i] = ureg_DECL_output(ureg,
                                          outputSemanticName[i],
                                          outputSemanticIndex[i]);
      }
   }
   else {
      assert(procType == TGSI_PROCESSOR_VERTEX);

      for (i = 0; i < numInputs; i++) {
         t->inputs[i] = ureg_DECL_vs_input(ureg, i);
      }

      for (i = 0; i < numOutputs; i++) {
         t->outputs[i] = ureg_DECL_output(ureg,
                                          outputSemanticName[i],
                                          outputSemanticIndex[i]);
      }
      if (passthrough_edgeflags)
         emit_edgeflags(t);
   }

   /* Declare address register.
    */
   if (program->num_address_regs > 0) {
      assert(program->num_address_regs == 1);
      t->address[0] = ureg_DECL_address(ureg);
   }

   /* Declare misc input registers
    */
   {
      GLbitfield sysInputs = proginfo->SystemValuesRead;
      unsigned numSys = 0;
      for (i = 0; sysInputs; i++) {
         if (sysInputs & (1 << i)) {
            unsigned semName = mesa_sysval_to_semantic[i];
            t->systemValues[i] = ureg_DECL_system_value(ureg, numSys, semName, 0);
            if (semName == TGSI_SEMANTIC_INSTANCEID ||
                semName == TGSI_SEMANTIC_VERTEXID) {
               /* From Gallium perspective, these system values are always
                * integer, and require native integer support.  However, if
                * native integer is supported on the vertex stage but not the
                * pixel stage (e.g, i915g + draw), Mesa will generate IR that
                * assumes these system values are floats. To resolve the
                * inconsistency, we insert a U2F.
                */
               struct st_context *st = st_context(ctx);
               struct pipe_screen *pscreen = st->pipe->screen;
               assert(procType == TGSI_PROCESSOR_VERTEX);
               assert(pscreen->get_shader_param(pscreen, PIPE_SHADER_VERTEX, PIPE_SHADER_CAP_INTEGERS));
               if (!ctx->Const.NativeIntegers) {
                  struct ureg_dst temp = ureg_DECL_local_temporary(t->ureg);
                  ureg_U2F( t->ureg, ureg_writemask(temp, TGSI_WRITEMASK_X), t->systemValues[i]);
                  t->systemValues[i] = ureg_scalar(ureg_src(temp), 0);
               }
            }
            numSys++;
            sysInputs &= ~(1 << i);
         }
      }
   }

   /* 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.
    */
   if (proginfo->Parameters) {
      t->constants = (struct ureg_src *)
         calloc(proginfo->Parameters->NumParameters, sizeof(t->constants[0]));
      if (t->constants == NULL) {
         ret = PIPE_ERROR_OUT_OF_MEMORY;
         goto out;
      }

      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);
            break;

         /* Emit immediates for PROGRAM_CONSTANT only when there's no indirect
          * addressing of the const buffer.
          * FIXME: Be smarter and recognize param arrays:
          * indirect addressing is only valid within the referenced
          * array.
          */
         case PROGRAM_CONSTANT:
            if (program->indirect_addr_consts)
               t->constants[i] = ureg_DECL_constant(ureg, i);
            else
               t->constants[i] = emit_immediate(t,
                                                proginfo->Parameters->ParameterValues[i],
                                                proginfo->Parameters->Parameters[i].DataType,
                                                4);
            break;
         default:
            break;
         }
      }
   }

   if (program->shader_program) {
      unsigned num_ubos = program->shader_program->NumUniformBlocks;

      for (i = 0; i < num_ubos; i++) {
         ureg_DECL_constant2D(t->ureg, 0, program->shader_program->UniformBlocks[i].UniformBufferSize / 4, i + 1);
      }
   }
   
   /* Emit immediate values.
    */
   t->immediates = (struct ureg_src *)
      calloc(program->num_immediates, sizeof(struct ureg_src));
   if (t->immediates == NULL) {
      ret = PIPE_ERROR_OUT_OF_MEMORY;
      goto out;
   }
   i = 0;
   foreach_iter(exec_list_iterator, iter, program->immediates) {
      immediate_storage *imm = (immediate_storage *)iter.get();
      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++) {
      if (program->samplers_used & (1 << i)) {
         t->samplers[i] = ureg_DECL_sampler(ureg, i);
      }
   }

   /* Emit each instruction in turn:
    */
   foreach_iter(exec_list_iterator, iter, program->instructions) {
      set_insn_start(t, ureg_get_instruction_number(ureg));
      compile_tgsi_instruction(t, (glsl_to_tgsi_instruction *)iter.get(),
                               clamp_color);
   }

   /* Fix up all emitted labels:
    */
   for (i = 0; i < t->labels_count; i++) {
      ureg_fixup_label(ureg, t->labels[i].token,
                       t->insn[t->labels[i].branch_target]);
   }

   if (program->shader_program) {
      /* This has to be done last.  Any operation the can cause
       * 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++) {
         if (program->shader_program->_LinkedShaders[i] == NULL)
            continue;

         _mesa_associate_uniform_storage(ctx, program->shader_program,
               program->shader_program->_LinkedShaders[i]->Program->Parameters);
      }
   }

out:
   if (t) {
      free(t->insn);
      free(t->labels);
      free(t->constants);
      free(t->immediates);

      if (t->error) {
         debug_printf("%s: translate error flag set\n", __FUNCTION__);
      }

      free(t);
   }

   return ret;
}
/* ----------------------------- End TGSI code ------------------------------ */

/**
 * Convert a shader's GLSL IR into a Mesa gl_program, although without 
 * generating Mesa IR.
 */
static struct gl_program *
get_mesa_program(struct gl_context *ctx,
                 struct gl_shader_program *shader_program,
                 struct gl_shader *shader)
{
   glsl_to_tgsi_visitor* v;
   struct gl_program *prog;
   GLenum target;
   const char *target_string;
   bool progress;
   struct gl_shader_compiler_options *options =
         &ctx->ShaderCompilerOptions[_mesa_shader_type_to_index(shader->Type)];
   struct pipe_screen *pscreen = ctx->st->pipe->screen;
   unsigned ptarget;

   switch (shader->Type) {
   case GL_VERTEX_SHADER:
      target = GL_VERTEX_PROGRAM_ARB;
      ptarget = PIPE_SHADER_VERTEX;
      target_string = "vertex";
      break;
   case GL_FRAGMENT_SHADER:
      target = GL_FRAGMENT_PROGRAM_ARB;
      ptarget = PIPE_SHADER_FRAGMENT;
      target_string = "fragment";
      break;
   case GL_GEOMETRY_SHADER:
      target = GL_GEOMETRY_PROGRAM_NV;
      ptarget = PIPE_SHADER_GEOMETRY;
      target_string = "geometry";
      break;
   default:
      assert(!"should not be reached");
      return NULL;
   }

   validate_ir_tree(shader->ir);

   prog = ctx->Driver.NewProgram(ctx, target, shader_program->Name);
   if (!prog)
      return NULL;
   prog->Parameters = _mesa_new_parameter_list();
   v = new glsl_to_tgsi_visitor();
   v->ctx = ctx;
   v->prog = prog;
   v->shader_program = shader_program;
   v->options = options;
   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);

   /* Remove reads from output registers. */
   lower_output_reads(shader->ir);

   /* Emit intermediate IR for main(). */
   visit_exec_list(shader->ir, v);

   /* Now emit bodies for any functions that were used. */
   do {
      progress = GL_FALSE;

      foreach_iter(exec_list_iterator, iter, v->function_signatures) {
         function_entry *entry = (function_entry *)iter.get();

         if (!entry->bgn_inst) {
            v->current_function = entry;

            entry->bgn_inst = v->emit(NULL, TGSI_OPCODE_BGNSUB);
            entry->bgn_inst->function = entry;

            visit_exec_list(&entry->sig->body, v);

            glsl_to_tgsi_instruction *last;
            last = (glsl_to_tgsi_instruction *)v->instructions.get_tail();
            if (last->op != TGSI_OPCODE_RET)
               v->emit(NULL, TGSI_OPCODE_RET);

            glsl_to_tgsi_instruction *end;
            end = v->emit(NULL, TGSI_OPCODE_ENDSUB);
            end->function = entry;

            progress = GL_TRUE;
         }
      }
   } while (progress);

#if 0
   /* Print out some information (for debugging purposes) used by the 
    * optimization passes. */
   for (i=0; i < v->next_temp; i++) {
      int fr = v->get_first_temp_read(i);
      int fw = v->get_first_temp_write(i);
      int lr = v->get_last_temp_read(i);
      int lw = v->get_last_temp_write(i);
      
      printf("Temp %d: FR=%3d FW=%3d LR=%3d LW=%3d\n", i, fr, fw, lr, lw);
      assert(fw <= fr);
   }
#endif

   /* Perform optimizations on the instructions in the glsl_to_tgsi_visitor. */
   v->simplify_cmp();
   v->copy_propagate();
   while (v->eliminate_dead_code_advanced());

   v->eliminate_dead_code();
   v->merge_registers();
   v->renumber_registers();
   
   /* Write the END instruction. */
   v->emit(NULL, TGSI_OPCODE_END);

   if (ctx->Shader.Flags & GLSL_DUMP) {
      printf("\n");
      printf("GLSL IR for linked %s program %d:\n", target_string,
             shader_program->Name);
      _mesa_print_ir(shader->ir, NULL);
      printf("\n");
      printf("\n");
      fflush(stdout);
   }

   prog->Instructions = NULL;
   prog->NumInstructions = 0;

   do_set_program_inouts(shader->ir, prog, shader->Type == GL_FRAGMENT_SHADER);
   count_resources(v, prog);

   _mesa_reference_program(ctx, &shader->Program, prog);
   
   /* This has to be done last.  Any operation the can cause
    * prog->ParameterValues to get reallocated (e.g., anything that adds a
    * program constant) has to happen before creating this linkage.
    */
   _mesa_associate_uniform_storage(ctx, shader_program, prog->Parameters);
   if (!shader_program->LinkStatus) {
      return NULL;
   }

   struct st_vertex_program *stvp;
   struct st_fragment_program *stfp;
   struct st_geometry_program *stgp;
   
   switch (shader->Type) {
   case GL_VERTEX_SHADER:
      stvp = (struct st_vertex_program *)prog;
      stvp->glsl_to_tgsi = v;
      break;
   case GL_FRAGMENT_SHADER:
      stfp = (struct st_fragment_program *)prog;
      stfp->glsl_to_tgsi = v;
      break;
   case GL_GEOMETRY_SHADER:
      stgp = (struct st_geometry_program *)prog;
      stgp->glsl_to_tgsi = v;
      break;
   default:
      assert(!"should not be reached");
      return NULL;
   }

   return prog;
}

extern "C" {

struct gl_shader *
st_new_shader(struct gl_context *ctx, GLuint name, GLuint type)
{
   struct gl_shader *shader;
   assert(type == GL_FRAGMENT_SHADER || type == GL_VERTEX_SHADER ||
          type == GL_GEOMETRY_SHADER_ARB);
   shader = rzalloc(NULL, struct gl_shader);
   if (shader) {
      shader->Type = type;
      shader->Name = name;
      _mesa_init_shader(ctx, shader);
   }
   return shader;
}

struct gl_shader_program *
st_new_shader_program(struct gl_context *ctx, GLuint name)
{
   struct gl_shader_program *shProg;
   shProg = rzalloc(NULL, struct gl_shader_program);
   if (shProg) {
      shProg->Name = name;
      _mesa_init_shader_program(ctx, shProg);
   }
   return shProg;
}

/**
 * Link a shader.
 * Called via ctx->Driver.LinkShader()
 * This actually involves converting GLSL IR into an intermediate TGSI-like IR 
 * with code lowering and other optimizations.
 */
GLboolean
st_link_shader(struct gl_context *ctx, struct gl_shader_program *prog)
{
   assert(prog->LinkStatus);

   for (unsigned i = 0; i < MESA_SHADER_TYPES; 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)];

      /* 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);
      }

      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);
      }

      do_mat_op_to_vec(ir);
      lower_instructions(ir,
                         MOD_TO_FRACT |
                         DIV_TO_MUL_RCP |
                         EXP_TO_EXP2 |
                         LOG_TO_LOG2 |
                         (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_quadop_vector(ir, false);
      lower_noise(ir);
      if (options->MaxIfDepth == 0) {
         lower_discard(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;

         progress = lower_if_to_cond_assign(ir, options->MaxIfDepth) || progress;

      } while (progress);

      validate_ir_tree(ir);
   }

   for (unsigned i = 0; i < MESA_SHADER_TYPES; i++) {
      struct gl_program *linked_prog;

      if (prog->_LinkedShaders[i] == NULL)
         continue;

      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)) {
            _mesa_reference_program(ctx, &prog->_LinkedShaders[i]->Program,
                                    NULL);
            _mesa_reference_program(ctx, &linked_prog, NULL);
            return GL_FALSE;
         }
      }

      _mesa_reference_program(ctx, &linked_prog, NULL);
   }

   return GL_TRUE;
}

void
st_translate_stream_output_info(glsl_to_tgsi_visitor *glsl_to_tgsi,
                                const GLuint outputMapping[],
                                struct pipe_stream_output_info *so)
{
   unsigned i;
   struct gl_transform_feedback_info *info =
      &glsl_to_tgsi->shader_program->LinkedTransformFeedback;

   for (i = 0; i < info->NumOutputs; i++) {
      so->output[i].register_index =
         outputMapping[info->Outputs[i].OutputRegister];
      so->output[i].start_component = info->Outputs[i].ComponentOffset;
      so->output[i].num_components = info->Outputs[i].NumComponents;
      so->output[i].output_buffer = info->Outputs[i].OutputBuffer;
      so->output[i].dst_offset = info->Outputs[i].DstOffset;
   }

   for (i = 0; i < PIPE_MAX_SO_BUFFERS; i++) {
      so->stride[i] = info->BufferStride[i];
   }
   so->num_outputs = info->NumOutputs;
}

} /* extern "C" */