src/gallium/drivers/r300/compiler/r500_fragprog_emit.c

/*
 * Copyright (C) 2005 Ben Skeggs.
 *
 * Copyright 2008 Corbin Simpson <MostAwesomeDude@gmail.com>
 * Adaptation and modification for ATI/AMD Radeon R500 GPU chipsets.
 *
 * All Rights Reserved.
 *
 * Permission is hereby granted, free of charge, to any person obtaining
 * a copy of this software and associated documentation files (the
 * "Software"), to deal in the Software without restriction, including
 * without limitation the rights to use, copy, modify, merge, publish,
 * distribute, sublicense, and/or sell copies of the Software, and to
 * permit persons to whom the Software is furnished to do so, subject to
 * the following conditions:
 *
 * The above copyright notice and this permission notice (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 COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS 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
 *
 * \author Ben Skeggs <darktama@iinet.net.au>
 *
 * \author Jerome Glisse <j.glisse@gmail.com>
 *
 * \author Corbin Simpson <MostAwesomeDude@gmail.com>
 *
 */

#include "r500_fragprog.h"

#include "../r300_reg.h"

#include "radeon_program_pair.h"

#define PROG_CODE \
	struct r500_fragment_program_code *code = &c->code->code.r500

#define error(fmt, args...) do {			\
		rc_error(&c->Base, "%s::%s(): " fmt "\n",	\
			__FILE__, __FUNCTION__, ##args);	\
	} while(0)


struct branch_info {
	int If;
	int Else;
	int Endif;
};

struct r500_loop_info {
	int BgnLoop;

	int BranchDepth;
	int * Brks;
	int BrkCount;
	int BrkReserved;

	int * Conts;
	int ContCount;
	int ContReserved;
};

struct emit_state {
	struct radeon_compiler * C;
	struct r500_fragment_program_code * Code;

	struct branch_info * Branches;
	unsigned int CurrentBranchDepth;
	unsigned int BranchesReserved;

	struct r500_loop_info * Loops;
	unsigned int CurrentLoopDepth;
	unsigned int LoopsReserved;

	unsigned int MaxBranchDepth;

};

static unsigned int translate_rgb_op(struct r300_fragment_program_compiler *c, rc_opcode opcode)
{
	switch(opcode) {
	case RC_OPCODE_CMP: return R500_ALU_RGBA_OP_CMP;
	case RC_OPCODE_CND: return R500_ALU_RGBA_OP_CND;
	case RC_OPCODE_DDX: return R500_ALU_RGBA_OP_MDH;
	case RC_OPCODE_DDY: return R500_ALU_RGBA_OP_MDV;
	case RC_OPCODE_DP3: return R500_ALU_RGBA_OP_DP3;
	case RC_OPCODE_DP4: return R500_ALU_RGBA_OP_DP4;
	case RC_OPCODE_FRC: return R500_ALU_RGBA_OP_FRC;
	default:
		error("translate_rgb_op: unknown opcode %s\n", rc_get_opcode_info(opcode)->Name);
		/* fall through */
	case RC_OPCODE_NOP:
		/* fall through */
	case RC_OPCODE_MAD: return R500_ALU_RGBA_OP_MAD;
	case RC_OPCODE_MAX: return R500_ALU_RGBA_OP_MAX;
	case RC_OPCODE_MIN: return R500_ALU_RGBA_OP_MIN;
	case RC_OPCODE_REPL_ALPHA: return R500_ALU_RGBA_OP_SOP;
	}
}

static unsigned int translate_alpha_op(struct r300_fragment_program_compiler *c, rc_opcode opcode)
{
	switch(opcode) {
	case RC_OPCODE_CMP: return R500_ALPHA_OP_CMP;
	case RC_OPCODE_CND: return R500_ALPHA_OP_CND;
	case RC_OPCODE_COS: return R500_ALPHA_OP_COS;
	case RC_OPCODE_DDX: return R500_ALPHA_OP_MDH;
	case RC_OPCODE_DDY: return R500_ALPHA_OP_MDV;
	case RC_OPCODE_DP3: return R500_ALPHA_OP_DP;
	case RC_OPCODE_DP4: return R500_ALPHA_OP_DP;
	case RC_OPCODE_EX2: return R500_ALPHA_OP_EX2;
	case RC_OPCODE_FRC: return R500_ALPHA_OP_FRC;
	case RC_OPCODE_LG2: return R500_ALPHA_OP_LN2;
	default:
		error("translate_alpha_op: unknown opcode %s\n", rc_get_opcode_info(opcode)->Name);
		/* fall through */
	case RC_OPCODE_NOP:
		/* fall through */
	case RC_OPCODE_MAD: return R500_ALPHA_OP_MAD;
	case RC_OPCODE_MAX: return R500_ALPHA_OP_MAX;
	case RC_OPCODE_MIN: return R500_ALPHA_OP_MIN;
	case RC_OPCODE_RCP: return R500_ALPHA_OP_RCP;
	case RC_OPCODE_RSQ: return R500_ALPHA_OP_RSQ;
	case RC_OPCODE_SIN: return R500_ALPHA_OP_SIN;
	}
}

static unsigned int fix_hw_swizzle(unsigned int swz)
{
    switch (swz) {
        case RC_SWIZZLE_ZERO:
        case RC_SWIZZLE_UNUSED:
            swz = 4;
            break;
        case RC_SWIZZLE_HALF:
            swz = 5;
            break;
        case RC_SWIZZLE_ONE:
            swz = 6;
            break;
    }

	return swz;
}

static unsigned int translate_arg_rgb(struct rc_pair_instruction *inst, int arg)
{
	unsigned int t = inst->RGB.Arg[arg].Source;
	int comp;
	t |= inst->RGB.Arg[arg].Negate << 11;
	t |= inst->RGB.Arg[arg].Abs << 12;

	for(comp = 0; comp < 3; ++comp)
		t |= fix_hw_swizzle(GET_SWZ(inst->RGB.Arg[arg].Swizzle, comp)) << (3*comp + 2);

	return t;
}

static unsigned int translate_arg_alpha(struct rc_pair_instruction *inst, int i)
{
	unsigned int t = inst->Alpha.Arg[i].Source;
	t |= fix_hw_swizzle(GET_SWZ(inst->Alpha.Arg[i].Swizzle, 0)) << 2;
	t |= inst->Alpha.Arg[i].Negate << 5;
	t |= inst->Alpha.Arg[i].Abs << 6;
	return t;
}

static uint32_t translate_alu_result_op(struct r300_fragment_program_compiler * c, rc_compare_func func)
{
	switch(func) {
	case RC_COMPARE_FUNC_EQUAL: return R500_INST_ALU_RESULT_OP_EQ;
	case RC_COMPARE_FUNC_LESS: return R500_INST_ALU_RESULT_OP_LT;
	case RC_COMPARE_FUNC_GEQUAL: return R500_INST_ALU_RESULT_OP_GE;
	case RC_COMPARE_FUNC_NOTEQUAL: return R500_INST_ALU_RESULT_OP_NE;
	default:
		rc_error(&c->Base, "%s: unsupported compare func %i\n", __FUNCTION__, func);
		return 0;
	}
}

static void use_temporary(struct r500_fragment_program_code* code, unsigned int index)
{
	if (index > code->max_temp_idx)
		code->max_temp_idx = index;
}

static unsigned int use_source(struct r500_fragment_program_code* code, struct rc_pair_instruction_source src)
{
	/* From docs:
	 *   Note that inline constants set the MSB of ADDR0 and clear ADDR0_CONST.
	 * MSB = 1 << 7 */
	if (!src.Used)
		return 1 << 7;

	if (src.File == RC_FILE_CONSTANT) {
		return src.Index | R500_RGB_ADDR0_CONST;
	} else if (src.File == RC_FILE_TEMPORARY || src.File == RC_FILE_INPUT) {
		use_temporary(code, src.Index);
		return src.Index;
	} else if (src.File == RC_FILE_INLINE) {
		return src.Index | (1 << 7);
	}

	return 0;
}

/**
 * NOP the specified instruction if it is not a texture lookup.
 */
static void alu_nop(struct r300_fragment_program_compiler *c, int ip)
{
	PROG_CODE;

	if ((code->inst[ip].inst0 & 0x3) != R500_INST_TYPE_TEX) {
		code->inst[ip].inst0 |= R500_INST_NOP;
	}
}

/**
 * Emit a paired ALU instruction.
 */
static void emit_paired(struct r300_fragment_program_compiler *c, struct rc_pair_instruction *inst)
{
	int ip;
	PROG_CODE;

	if (code->inst_end >= c->Base.max_alu_insts-1) {
		error("emit_alu: Too many instructions");
		return;
	}

	ip = ++code->inst_end;

	/* Quirk: MDH/MDV (DDX/DDY) need a NOP on previous non-TEX instructions. */
	if (inst->RGB.Opcode == RC_OPCODE_DDX || inst->Alpha.Opcode == RC_OPCODE_DDX ||
		inst->RGB.Opcode == RC_OPCODE_DDY || inst->Alpha.Opcode == RC_OPCODE_DDY) {
		if (ip > 0) {
			alu_nop(c, ip - 1);
		}
	}

	code->inst[ip].inst5 = translate_rgb_op(c, inst->RGB.Opcode);
	code->inst[ip].inst4 = translate_alpha_op(c, inst->Alpha.Opcode);

	if (inst->RGB.OutputWriteMask || inst->Alpha.OutputWriteMask || inst->Alpha.DepthWriteMask) {
		code->inst[ip].inst0 = R500_INST_TYPE_OUT;
		if (inst->WriteALUResult) {
			error("Cannot write output and ALU result at the same time");
			return;
		}
	} else {
		code->inst[ip].inst0 = R500_INST_TYPE_ALU;
	}
	code->inst[ip].inst0 |= (inst->SemWait << R500_INST_TEX_SEM_WAIT_SHIFT);

	code->inst[ip].inst0 |= (inst->RGB.WriteMask << 11);
	code->inst[ip].inst0 |= inst->Alpha.WriteMask ? 1 << 14 : 0;
	code->inst[ip].inst0 |= (inst->RGB.OutputWriteMask << 15) | (inst->Alpha.OutputWriteMask << 18);
	if (inst->Nop) {
		code->inst[ip].inst0 |= R500_INST_NOP;
	}
	if (inst->Alpha.DepthWriteMask) {
		code->inst[ip].inst4 |= R500_ALPHA_W_OMASK;
		c->code->writes_depth = 1;
	}

	code->inst[ip].inst4 |= R500_ALPHA_ADDRD(inst->Alpha.DestIndex);
	code->inst[ip].inst5 |= R500_ALU_RGBA_ADDRD(inst->RGB.DestIndex);
	use_temporary(code, inst->Alpha.DestIndex);
	use_temporary(code, inst->RGB.DestIndex);

	if (inst->RGB.Saturate)
		code->inst[ip].inst0 |= R500_INST_RGB_CLAMP;
	if (inst->Alpha.Saturate)
		code->inst[ip].inst0 |= R500_INST_ALPHA_CLAMP;

	/* Set the presubtract operation. */
	switch(inst->RGB.Src[RC_PAIR_PRESUB_SRC].Index) {
		case RC_PRESUB_BIAS:
			code->inst[ip].inst1 |= R500_RGB_SRCP_OP_1_MINUS_2RGB0;
			break;
		case RC_PRESUB_SUB:
			code->inst[ip].inst1 |= R500_RGB_SRCP_OP_RGB1_MINUS_RGB0;
			break;
		case RC_PRESUB_ADD:
			code->inst[ip].inst1 |= R500_RGB_SRCP_OP_RGB1_PLUS_RGB0;
			break;
		case RC_PRESUB_INV:
			code->inst[ip].inst1 |= R500_RGB_SRCP_OP_1_MINUS_RGB0;
			break;
		default:
			break;
	}
	switch(inst->Alpha.Src[RC_PAIR_PRESUB_SRC].Index) {
		case RC_PRESUB_BIAS:
			code->inst[ip].inst2 |= R500_ALPHA_SRCP_OP_1_MINUS_2A0;
			break;
		case RC_PRESUB_SUB:
			code->inst[ip].inst2 |= R500_ALPHA_SRCP_OP_A1_MINUS_A0;
			break;
		case RC_PRESUB_ADD:
			code->inst[ip].inst2 |= R500_ALPHA_SRCP_OP_A1_PLUS_A0;
			break;
		case RC_PRESUB_INV:
			code->inst[ip].inst2 |= R500_ALPHA_SRCP_OP_1_MINUS_A0;
			break;
		default:
			break;
	}

	/* Set the output modifier */
	code->inst[ip].inst3 |= inst->RGB.Omod << R500_ALU_RGB_OMOD_SHIFT;
	code->inst[ip].inst4 |= inst->Alpha.Omod << R500_ALPHA_OMOD_SHIFT;

	code->inst[ip].inst1 |= R500_RGB_ADDR0(use_source(code, inst->RGB.Src[0]));
	code->inst[ip].inst1 |= R500_RGB_ADDR1(use_source(code, inst->RGB.Src[1]));
	code->inst[ip].inst1 |= R500_RGB_ADDR2(use_source(code, inst->RGB.Src[2]));

	code->inst[ip].inst2 |= R500_ALPHA_ADDR0(use_source(code, inst->Alpha.Src[0]));
	code->inst[ip].inst2 |= R500_ALPHA_ADDR1(use_source(code, inst->Alpha.Src[1]));
	code->inst[ip].inst2 |= R500_ALPHA_ADDR2(use_source(code, inst->Alpha.Src[2]));

	code->inst[ip].inst3 |= translate_arg_rgb(inst, 0) << R500_ALU_RGB_SEL_A_SHIFT;
	code->inst[ip].inst3 |= translate_arg_rgb(inst, 1) << R500_ALU_RGB_SEL_B_SHIFT;
	code->inst[ip].inst5 |= translate_arg_rgb(inst, 2) << R500_ALU_RGBA_SEL_C_SHIFT;

	code->inst[ip].inst4 |= translate_arg_alpha(inst, 0) << R500_ALPHA_SEL_A_SHIFT;
	code->inst[ip].inst4 |= translate_arg_alpha(inst, 1) << R500_ALPHA_SEL_B_SHIFT;
	code->inst[ip].inst5 |= translate_arg_alpha(inst, 2) << R500_ALU_RGBA_ALPHA_SEL_C_SHIFT;

	code->inst[ip].inst3 |= R500_ALU_RGB_TARGET(inst->RGB.Target);
	code->inst[ip].inst4 |= R500_ALPHA_TARGET(inst->Alpha.Target);

	if (inst->WriteALUResult) {
		code->inst[ip].inst3 |= R500_ALU_RGB_WMASK;

		if (inst->WriteALUResult == RC_ALURESULT_X)
			code->inst[ip].inst0 |= R500_INST_ALU_RESULT_SEL_RED;
		else
			code->inst[ip].inst0 |= R500_INST_ALU_RESULT_SEL_ALPHA;

		code->inst[ip].inst0 |= translate_alu_result_op(c, inst->ALUResultCompare);
	}
}

static unsigned int translate_strq_swizzle(unsigned int swizzle)
{
	unsigned int swiz = 0;
	int i;
	for (i = 0; i < 4; i++)
		swiz |= (GET_SWZ(swizzle, i) & 0x3) << i*2;
	return swiz;
}

/**
 * Emit a single TEX instruction
 */
static int emit_tex(struct r300_fragment_program_compiler *c, struct rc_sub_instruction *inst)
{
	int ip;
	PROG_CODE;

	if (code->inst_end >= c->Base.max_alu_insts-1) {
		error("emit_tex: Too many instructions");
		return 0;
	}

	ip = ++code->inst_end;

	code->inst[ip].inst0 = R500_INST_TYPE_TEX
		| (inst->DstReg.WriteMask << 11)
		| (inst->TexSemWait << R500_INST_TEX_SEM_WAIT_SHIFT);
	code->inst[ip].inst1 = R500_TEX_ID(inst->TexSrcUnit)
		| (inst->TexSemAcquire << R500_TEX_SEM_ACQUIRE_SHIFT);

	if (inst->TexSrcTarget == RC_TEXTURE_RECT)
		code->inst[ip].inst1 |= R500_TEX_UNSCALED;

	switch (inst->Opcode) {
	case RC_OPCODE_KIL:
		code->inst[ip].inst1 |= R500_TEX_INST_TEXKILL;
		break;
	case RC_OPCODE_TEX:
		code->inst[ip].inst1 |= R500_TEX_INST_LD;
		break;
	case RC_OPCODE_TXB:
		code->inst[ip].inst1 |= R500_TEX_INST_LODBIAS;
		break;
	case RC_OPCODE_TXP:
		code->inst[ip].inst1 |= R500_TEX_INST_PROJ;
		break;
	case RC_OPCODE_TXD:
		code->inst[ip].inst1 |= R500_TEX_INST_DXDY;
		break;
	case RC_OPCODE_TXL:
		code->inst[ip].inst1 |= R500_TEX_INST_LOD;
		break;
	default:
		error("emit_tex can't handle opcode %s\n", rc_get_opcode_info(inst->Opcode)->Name);
	}

	use_temporary(code, inst->SrcReg[0].Index);
	if (inst->Opcode != RC_OPCODE_KIL)
		use_temporary(code, inst->DstReg.Index);

	code->inst[ip].inst2 = R500_TEX_SRC_ADDR(inst->SrcReg[0].Index)
		| (translate_strq_swizzle(inst->SrcReg[0].Swizzle) << 8)
		| R500_TEX_DST_ADDR(inst->DstReg.Index)
		| (GET_SWZ(inst->TexSwizzle, 0) << 24)
		| (GET_SWZ(inst->TexSwizzle, 1) << 26)
		| (GET_SWZ(inst->TexSwizzle, 2) << 28)
		| (GET_SWZ(inst->TexSwizzle, 3) << 30)
		;

	if (inst->Opcode == RC_OPCODE_TXD) {
		use_temporary(code, inst->SrcReg[1].Index);
		use_temporary(code, inst->SrcReg[2].Index);

		/* DX and DY parameters are specified in a separate register. */
		code->inst[ip].inst3 =
			R500_DX_ADDR(inst->SrcReg[1].Index) |
			(translate_strq_swizzle(inst->SrcReg[1].Swizzle) << 8) |
			R500_DY_ADDR(inst->SrcReg[2].Index) |
			(translate_strq_swizzle(inst->SrcReg[2].Swizzle) << 24);
	}

	return 1;
}

static void emit_flowcontrol(struct emit_state * s, struct rc_instruction * inst)
{
	unsigned int newip;

	if (s->Code->inst_end >= s->C->max_alu_insts-1) {
		rc_error(s->C, "emit_tex: Too many instructions");
		return;
	}

	newip = ++s->Code->inst_end;

	/* Currently all loops use the same integer constant to intialize
	 * the loop variables. */
	if(!s->Code->int_constants[0]) {
		s->Code->int_constants[0] = R500_FC_INT_CONST_KR(0xff);
		s->Code->int_constant_count = 1;
	}
	s->Code->inst[newip].inst0 = R500_INST_TYPE_FC | R500_INST_ALU_WAIT;

	switch(inst->U.I.Opcode){
	struct branch_info * branch;
	struct r500_loop_info * loop;
	case RC_OPCODE_BGNLOOP:
		memory_pool_array_reserve(&s->C->Pool, struct r500_loop_info,
			s->Loops, s->CurrentLoopDepth, s->LoopsReserved, 1);

		loop = &s->Loops[s->CurrentLoopDepth++];
		memset(loop, 0, sizeof(struct r500_loop_info));
		loop->BranchDepth = s->CurrentBranchDepth;
		loop->BgnLoop = newip;

		s->Code->inst[newip].inst2 = R500_FC_OP_LOOP
			| R500_FC_JUMP_FUNC(0x00)
			| R500_FC_IGNORE_UNCOVERED
			;
		break;
	case RC_OPCODE_BRK:
		loop = &s->Loops[s->CurrentLoopDepth - 1];
		memory_pool_array_reserve(&s->C->Pool, int, loop->Brks,
					loop->BrkCount, loop->BrkReserved, 1);

		loop->Brks[loop->BrkCount++] = newip;
		s->Code->inst[newip].inst2 = R500_FC_OP_BREAKLOOP
			| R500_FC_JUMP_FUNC(0xff)
			| R500_FC_B_OP1_DECR
			| R500_FC_B_POP_CNT(
				s->CurrentBranchDepth - loop->BranchDepth)
			| R500_FC_IGNORE_UNCOVERED
			;
		break;

	case RC_OPCODE_CONT:
		loop = &s->Loops[s->CurrentLoopDepth - 1];
		memory_pool_array_reserve(&s->C->Pool, int, loop->Conts,
					loop->ContCount, loop->ContReserved, 1);
		loop->Conts[loop->ContCount++] = newip;
		s->Code->inst[newip].inst2 = R500_FC_OP_CONTINUE
			| R500_FC_JUMP_FUNC(0xff)
			| R500_FC_B_OP1_DECR
			| R500_FC_B_POP_CNT(
				s->CurrentBranchDepth -	loop->BranchDepth)
			| R500_FC_IGNORE_UNCOVERED
			;
		break;

	case RC_OPCODE_ENDLOOP:
	{
		loop = &s->Loops[s->CurrentLoopDepth - 1];
		/* Emit ENDLOOP */
		s->Code->inst[newip].inst2 = R500_FC_OP_ENDLOOP
			| R500_FC_JUMP_FUNC(0xff)
			| R500_FC_JUMP_ANY
			| R500_FC_IGNORE_UNCOVERED
			;
		/* The constant integer at index 0 is used by all loops. */
		s->Code->inst[newip].inst3 = R500_FC_INT_ADDR(0)
			| R500_FC_JUMP_ADDR(loop->BgnLoop + 1)
			;

		/* Set jump address and int constant for BGNLOOP */
		s->Code->inst[loop->BgnLoop].inst3 = R500_FC_INT_ADDR(0)
			| R500_FC_JUMP_ADDR(newip)
			;

		/* Set jump address for the BRK instructions. */
		while(loop->BrkCount--) {
			s->Code->inst[loop->Brks[loop->BrkCount]].inst3 =
						R500_FC_JUMP_ADDR(newip + 1);
		}

		/* Set jump address for CONT instructions. */
		while(loop->ContCount--) {
			s->Code->inst[loop->Conts[loop->ContCount]].inst3 =
						R500_FC_JUMP_ADDR(newip);
		}
		s->CurrentLoopDepth--;
		break;
	}
	case RC_OPCODE_IF:
		if ( s->CurrentBranchDepth >= R500_PFS_MAX_BRANCH_DEPTH_FULL) {
			rc_error(s->C, "Branch depth exceeds hardware limit");
			return;
		}
		memory_pool_array_reserve(&s->C->Pool, struct branch_info,
				s->Branches, s->CurrentBranchDepth, s->BranchesReserved, 1);

		branch = &s->Branches[s->CurrentBranchDepth++];
		branch->If = newip;
		branch->Else = -1;
		branch->Endif = -1;

		if (s->CurrentBranchDepth > s->MaxBranchDepth)
			s->MaxBranchDepth = s->CurrentBranchDepth;

		/* actual instruction is filled in at ENDIF time */
		break;
	
	case RC_OPCODE_ELSE:
		if (!s->CurrentBranchDepth) {
			rc_error(s->C, "%s: got ELSE outside a branch", __FUNCTION__);
			return;
		}

		branch = &s->Branches[s->CurrentBranchDepth - 1];
		branch->Else = newip;

		/* actual instruction is filled in at ENDIF time */
		break;

	case RC_OPCODE_ENDIF:
		if (!s->CurrentBranchDepth) {
			rc_error(s->C, "%s: got ELSE outside a branch", __FUNCTION__);
			return;
		}

		branch = &s->Branches[s->CurrentBranchDepth - 1];
		branch->Endif = newip;

		s->Code->inst[branch->Endif].inst2 = R500_FC_OP_JUMP
			| R500_FC_A_OP_NONE /* no address stack */
			| R500_FC_JUMP_ANY /* docs says set this, but I don't understand why */
			| R500_FC_B_OP0_DECR /* decrement branch counter if stay */
			| R500_FC_B_OP1_NONE /* no branch counter if stay */
			| R500_FC_B_POP_CNT(1)
			;
		s->Code->inst[branch->Endif].inst3 = R500_FC_JUMP_ADDR(branch->Endif + 1);
		s->Code->inst[branch->If].inst2 = R500_FC_OP_JUMP
			| R500_FC_A_OP_NONE /* no address stack */
			| R500_FC_JUMP_FUNC(0x0f) /* jump if ALU result is false */
			| R500_FC_B_OP0_INCR /* increment branch counter if stay */
			| R500_FC_IGNORE_UNCOVERED
		;

		if (branch->Else >= 0) {
			/* increment branch counter also if jump */
			s->Code->inst[branch->If].inst2 |= R500_FC_B_OP1_INCR;
			s->Code->inst[branch->If].inst3 = R500_FC_JUMP_ADDR(branch->Else + 1);

			s->Code->inst[branch->Else].inst2 = R500_FC_OP_JUMP
				| R500_FC_A_OP_NONE /* no address stack */
				| R500_FC_B_ELSE /* all active pixels want to jump */
				| R500_FC_B_OP0_NONE /* no counter op if stay */
				| R500_FC_B_OP1_DECR /* decrement branch counter if jump */
				| R500_FC_B_POP_CNT(1)
			;
			s->Code->inst[branch->Else].inst3 = R500_FC_JUMP_ADDR(branch->Endif + 1);
		} else {
			/* don't touch branch counter on jump */
			s->Code->inst[branch->If].inst2 |= R500_FC_B_OP1_NONE;
			s->Code->inst[branch->If].inst3 = R500_FC_JUMP_ADDR(branch->Endif + 1);
		}


		s->CurrentBranchDepth--;
		break;
	default:
		rc_error(s->C, "%s: unknown opcode %s\n", __FUNCTION__, rc_get_opcode_info(inst->U.I.Opcode)->Name);
	}
}

void r500BuildFragmentProgramHwCode(struct radeon_compiler *c, void *user)
{
	struct r300_fragment_program_compiler *compiler = (struct r300_fragment_program_compiler*)c;
	struct emit_state s;
	struct r500_fragment_program_code *code = &compiler->code->code.r500;

	memset(&s, 0, sizeof(s));
	s.C = &compiler->Base;
	s.Code = code;

	memset(code, 0, sizeof(*code));
	code->max_temp_idx = 1;
	code->inst_end = -1;

	for(struct rc_instruction * inst = compiler->Base.Program.Instructions.Next;
	    inst != &compiler->Base.Program.Instructions && !compiler->Base.Error;
	    inst = inst->Next) {
		if (inst->Type == RC_INSTRUCTION_NORMAL) {
			const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode);

			if (opcode->IsFlowControl) {
				emit_flowcontrol(&s, inst);
			} else if (inst->U.I.Opcode == RC_OPCODE_BEGIN_TEX) {
				continue;
			} else {
				emit_tex(compiler, &inst->U.I);
			}
		} else {
			emit_paired(compiler, &inst->U.P);
		}
	}

	if (code->max_temp_idx >= compiler->Base.max_temp_regs)
		rc_error(&compiler->Base, "Too many hardware temporaries used");

	if (compiler->Base.Error)
		return;

	if (code->inst_end == -1 ||
	    (code->inst[code->inst_end].inst0 & R500_INST_TYPE_MASK) != R500_INST_TYPE_OUT) {
		int ip;

		/* This may happen when dead-code elimination is disabled or
		 * when most of the fragment program logic is leading to a KIL */
		if (code->inst_end >= compiler->Base.max_alu_insts-1) {
			rc_error(&compiler->Base, "Introducing fake OUT: Too many instructions");
			return;
		}

		ip = ++code->inst_end;
		code->inst[ip].inst0 = R500_INST_TYPE_OUT | R500_INST_TEX_SEM_WAIT;
	}

	/* Make sure TEX_SEM_WAIT is set on the last instruction */
	code->inst[code->inst_end].inst0 |= R500_INST_TEX_SEM_WAIT;

	/* Enable full flow control mode if we are using loops or have if
	 * statements nested at least four deep. */
	if (s.MaxBranchDepth >= 4 || s.LoopsReserved > 0) {
		if (code->max_temp_idx < 1)
			code->max_temp_idx = 1;

		code->us_fc_ctrl |= R500_FC_FULL_FC_EN;
	}
}