diff options
author | Stéphane Marchesin <[email protected]> | 2011-08-26 17:37:25 -0700 |
---|---|---|
committer | Stéphane Marchesin <[email protected]> | 2011-08-26 17:37:25 -0700 |
commit | f8e6d19f3f40931be741b44d3edf210c38e13f0f (patch) | |
tree | e99e4c619901412ac6448534b0f57ce1c4295c6b /src/gallium/drivers/r300/compiler | |
parent | 974c49ed176de55aadb335a2956ef5dfec774a23 (diff) | |
parent | e3b0e3776646d0367206e4544229622eb22fe9f8 (diff) |
Merge branch 'master' of git://anongit.freedesktop.org/mesa/mesa
Diffstat (limited to 'src/gallium/drivers/r300/compiler')
60 files changed, 16933 insertions, 0 deletions
diff --git a/src/gallium/drivers/r300/compiler/memory_pool.c b/src/gallium/drivers/r300/compiler/memory_pool.c new file mode 100644 index 00000000000..ddcdddf9e3c --- /dev/null +++ b/src/gallium/drivers/r300/compiler/memory_pool.c @@ -0,0 +1,97 @@ +/* + * Copyright 2009 Nicolai Hähnle <[email protected]> + * + * 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 + * on the rights to use, copy, modify, merge, publish, distribute, sub + * license, 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 NON-INFRINGEMENT. IN NO EVENT SHALL + * THE AUTHOR(S) AND/OR THEIR 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. */ + +#include "memory_pool.h" + +#include <assert.h> +#include <stdlib.h> +#include <string.h> + + +#define POOL_LARGE_ALLOC 4096 +#define POOL_ALIGN 8 + + +struct memory_block { + struct memory_block * next; +}; + +void memory_pool_init(struct memory_pool * pool) +{ + memset(pool, 0, sizeof(struct memory_pool)); +} + + +void memory_pool_destroy(struct memory_pool * pool) +{ + while(pool->blocks) { + struct memory_block * block = pool->blocks; + pool->blocks = block->next; + free(block); + } +} + +static void refill_pool(struct memory_pool * pool) +{ + unsigned int blocksize = pool->total_allocated; + struct memory_block * newblock; + + if (!blocksize) + blocksize = 2*POOL_LARGE_ALLOC; + + newblock = (struct memory_block*)malloc(blocksize); + newblock->next = pool->blocks; + pool->blocks = newblock; + + pool->head = (unsigned char*)(newblock + 1); + pool->end = ((unsigned char*)newblock) + blocksize; + pool->total_allocated += blocksize; +} + + +void * memory_pool_malloc(struct memory_pool * pool, unsigned int bytes) +{ + if (bytes < POOL_LARGE_ALLOC) { + void * ptr; + + if (pool->head + bytes > pool->end) + refill_pool(pool); + + assert(pool->head + bytes <= pool->end); + + ptr = pool->head; + + pool->head += bytes; + pool->head = (unsigned char*)(((unsigned long)pool->head + POOL_ALIGN - 1) & ~(POOL_ALIGN - 1)); + + return ptr; + } else { + struct memory_block * block = (struct memory_block*)malloc(bytes + sizeof(struct memory_block)); + + block->next = pool->blocks; + pool->blocks = block; + + return (block + 1); + } +} + + diff --git a/src/gallium/drivers/r300/compiler/memory_pool.h b/src/gallium/drivers/r300/compiler/memory_pool.h new file mode 100644 index 00000000000..42344d0e3ba --- /dev/null +++ b/src/gallium/drivers/r300/compiler/memory_pool.h @@ -0,0 +1,80 @@ +/* + * Copyright 2009 Nicolai Hähnle <[email protected]> + * + * 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 + * on the rights to use, copy, modify, merge, publish, distribute, sub + * license, 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 NON-INFRINGEMENT. IN NO EVENT SHALL + * THE AUTHOR(S) AND/OR THEIR 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. */ + +#ifndef MEMORY_POOL_H +#define MEMORY_POOL_H + +struct memory_block; + +/** + * Provides a pool of memory that can quickly be allocated from, at the + * cost of being unable to explicitly free one of the allocated blocks. + * Instead, the entire pool can be freed at once. + * + * The idea is to allow one to quickly allocate a flexible amount of + * memory during operations like shader compilation while avoiding + * reference counting headaches. + */ +struct memory_pool { + unsigned char * head; + unsigned char * end; + unsigned int total_allocated; + struct memory_block * blocks; +}; + + +void memory_pool_init(struct memory_pool * pool); +void memory_pool_destroy(struct memory_pool * pool); +void * memory_pool_malloc(struct memory_pool * pool, unsigned int bytes); + + +/** + * Generic helper for growing an array that has separate size/count + * and reserved counters to accomodate up to num new element. + * + * type * Array; + * unsigned int Size; + * unsigned int Reserved; + * + * memory_pool_array_reserve(pool, type, Array, Size, Reserved, k); + * assert(Size + k < Reserved); + * + * \note Size is not changed by this macro. + * + * \warning Array, Size, Reserved have to be lvalues and may be evaluated + * several times. + */ +#define memory_pool_array_reserve(pool, type, array, size, reserved, num) do { \ + unsigned int _num = (num); \ + if ((size) + _num > (reserved)) { \ + unsigned int newreserve = (reserved) * 2; \ + type * newarray; \ + if (newreserve < _num) \ + newreserve = 4 * _num; /* arbitrary heuristic */ \ + newarray = memory_pool_malloc((pool), newreserve * sizeof(type)); \ + memcpy(newarray, (array), (size) * sizeof(type)); \ + (array) = newarray; \ + (reserved) = newreserve; \ + } \ +} while(0) + +#endif /* MEMORY_POOL_H */ diff --git a/src/gallium/drivers/r300/compiler/r300_fragprog.c b/src/gallium/drivers/r300/compiler/r300_fragprog.c new file mode 100644 index 00000000000..deba9ca834d --- /dev/null +++ b/src/gallium/drivers/r300/compiler/r300_fragprog.c @@ -0,0 +1,338 @@ +/* + * Copyright (C) 2005 Ben Skeggs. + * + * 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. + * + */ + +#include "r300_fragprog.h" + +#include <stdio.h> + +#include "../r300_reg.h" + +static void presub_string(char out[10], unsigned int inst) +{ + switch(inst & 0x600000){ + case R300_ALU_SRCP_1_MINUS_2_SRC0: + sprintf(out, "bias"); + break; + case R300_ALU_SRCP_SRC1_MINUS_SRC0: + sprintf(out, "sub"); + break; + case R300_ALU_SRCP_SRC1_PLUS_SRC0: + sprintf(out, "add"); + break; + case R300_ALU_SRCP_1_MINUS_SRC0: + sprintf(out, "inv "); + break; + } +} + +static int get_msb(unsigned int bit, unsigned int r400_ext_addr) +{ + return (r400_ext_addr & bit) ? 1 << 5 : 0; +} + +/* just some random things... */ +void r300FragmentProgramDump(struct radeon_compiler *c, void *user) +{ + struct r300_fragment_program_compiler *compiler = (struct r300_fragment_program_compiler*)c; + struct r300_fragment_program_code *code = &compiler->code->code.r300; + int n, i, j; + static int pc = 0; + + fprintf(stderr, "pc=%d*************************************\n", pc++); + + fprintf(stderr, "Hardware program\n"); + fprintf(stderr, "----------------\n"); + if (c->is_r400) { + fprintf(stderr, "code_offset_ext: %08x\n", code->r400_code_offset_ext); + } + + for (n = 0; n <= (code->config & 3); n++) { + uint32_t code_addr = code->code_addr[3 - (code->config & 3) + n]; + unsigned int alu_offset = ((code_addr & R300_ALU_START_MASK) >> R300_ALU_START_SHIFT) + + (((code->r400_code_offset_ext >> (24 - (n * 6))) & 0x7) << 6); + unsigned int alu_end = ((code_addr & R300_ALU_SIZE_MASK) >> R300_ALU_SIZE_SHIFT) + + (((code->r400_code_offset_ext >> (27 - (n * 6))) & 0x7) << 6); + int tex_offset = (code_addr & R300_TEX_START_MASK) >> R300_TEX_START_SHIFT; + int tex_end = (code_addr & R300_TEX_SIZE_MASK) >> R300_TEX_SIZE_SHIFT; + + fprintf(stderr, "NODE %d: alu_offset: %u, tex_offset: %d, " + "alu_end: %u, tex_end: %d (code_addr: %08x)\n", n, + alu_offset, tex_offset, alu_end, tex_end, code_addr); + + if (n > 0 || (code->config & R300_PFS_CNTL_FIRST_NODE_HAS_TEX)) { + fprintf(stderr, " TEX:\n"); + for (i = tex_offset; + i <= tex_offset + tex_end; + ++i) { + const char *instr; + + switch ((code->tex. + inst[i] >> R300_TEX_INST_SHIFT) & + 15) { + case R300_TEX_OP_LD: + instr = "TEX"; + break; + case R300_TEX_OP_KIL: + instr = "KIL"; + break; + case R300_TEX_OP_TXP: + instr = "TXP"; + break; + case R300_TEX_OP_TXB: + instr = "TXB"; + break; + default: + instr = "UNKNOWN"; + } + + fprintf(stderr, + " %s t%i, %c%i, texture[%i] (%08x)\n", + instr, + (code->tex. + inst[i] >> R300_DST_ADDR_SHIFT) & 31, + 't', + (code->tex. + inst[i] >> R300_SRC_ADDR_SHIFT) & 31, + (code->tex. + inst[i] & R300_TEX_ID_MASK) >> + R300_TEX_ID_SHIFT, + code->tex.inst[i]); + } + } + + for (i = alu_offset; + i <= alu_offset + alu_end; ++i) { + char srcc[4][10], dstc[20]; + char srca[4][10], dsta[20]; + char argc[3][20]; + char arga[3][20]; + char flags[5], tmp[10]; + + for (j = 0; j < 3; ++j) { + int regc = code->alu.inst[i].rgb_addr >> (j * 6); + int rega = code->alu.inst[i].alpha_addr >> (j * 6); + int msbc = get_msb(R400_ADDR_EXT_RGB_MSB_BIT(j), + code->alu.inst[i].r400_ext_addr); + int msba = get_msb(R400_ADDR_EXT_A_MSB_BIT(j), + code->alu.inst[i].r400_ext_addr); + + sprintf(srcc[j], "%c%i", + (regc & 32) ? 'c' : 't', (regc & 31) | msbc); + sprintf(srca[j], "%c%i", + (rega & 32) ? 'c' : 't', (rega & 31) | msba); + } + + dstc[0] = 0; + sprintf(flags, "%s%s%s", + (code->alu.inst[i]. + rgb_addr & R300_ALU_DSTC_REG_X) ? "x" : "", + (code->alu.inst[i]. + rgb_addr & R300_ALU_DSTC_REG_Y) ? "y" : "", + (code->alu.inst[i]. + rgb_addr & R300_ALU_DSTC_REG_Z) ? "z" : ""); + if (flags[0] != 0) { + unsigned int msb = get_msb( + R400_ADDRD_EXT_RGB_MSB_BIT, + code->alu.inst[i].r400_ext_addr); + + sprintf(dstc, "t%i.%s ", + ((code->alu.inst[i]. + rgb_addr >> R300_ALU_DSTC_SHIFT) + & 31) | msb, + flags); + } + sprintf(flags, "%s%s%s", + (code->alu.inst[i]. + rgb_addr & R300_ALU_DSTC_OUTPUT_X) ? "x" : "", + (code->alu.inst[i]. + rgb_addr & R300_ALU_DSTC_OUTPUT_Y) ? "y" : "", + (code->alu.inst[i]. + rgb_addr & R300_ALU_DSTC_OUTPUT_Z) ? "z" : ""); + if (flags[0] != 0) { + sprintf(tmp, "o%i.%s", + (code->alu.inst[i]. + rgb_addr >> 29) & 3, + flags); + strcat(dstc, tmp); + } + /* Presub */ + presub_string(srcc[3], code->alu.inst[i].rgb_inst); + presub_string(srca[3], code->alu.inst[i].alpha_inst); + + dsta[0] = 0; + if (code->alu.inst[i].alpha_addr & R300_ALU_DSTA_REG) { + unsigned int msb = get_msb( + R400_ADDRD_EXT_A_MSB_BIT, + code->alu.inst[i].r400_ext_addr); + sprintf(dsta, "t%i.w ", + ((code->alu.inst[i]. + alpha_addr >> R300_ALU_DSTA_SHIFT) & 31) + | msb); + } + if (code->alu.inst[i].alpha_addr & R300_ALU_DSTA_OUTPUT) { + sprintf(tmp, "o%i.w ", + (code->alu.inst[i]. + alpha_addr >> 25) & 3); + strcat(dsta, tmp); + } + if (code->alu.inst[i].alpha_addr & R300_ALU_DSTA_DEPTH) { + strcat(dsta, "Z"); + } + + fprintf(stderr, + "%3i: xyz: %3s %3s %3s %5s-> %-20s (%08x)\n" + " w: %3s %3s %3s %5s-> %-20s (%08x)\n", i, + srcc[0], srcc[1], srcc[2], srcc[3], dstc, + code->alu.inst[i].rgb_addr, srca[0], srca[1], + srca[2], srca[3], dsta, + code->alu.inst[i].alpha_addr); + + for (j = 0; j < 3; ++j) { + int regc = code->alu.inst[i].rgb_inst >> (j * 7); + int rega = code->alu.inst[i].alpha_inst >> (j * 7); + int d; + char buf[20]; + + d = regc & 31; + if (d < 12) { + switch (d % 4) { + case R300_ALU_ARGC_SRC0C_XYZ: + sprintf(buf, "%s.xyz", + srcc[d / 4]); + break; + case R300_ALU_ARGC_SRC0C_XXX: + sprintf(buf, "%s.xxx", + srcc[d / 4]); + break; + case R300_ALU_ARGC_SRC0C_YYY: + sprintf(buf, "%s.yyy", + srcc[d / 4]); + break; + case R300_ALU_ARGC_SRC0C_ZZZ: + sprintf(buf, "%s.zzz", + srcc[d / 4]); + break; + } + } else if (d < 15) { + sprintf(buf, "%s.www", srca[d - 12]); + } else if (d < 20 ) { + switch(d) { + case R300_ALU_ARGC_SRCP_XYZ: + sprintf(buf, "srcp.xyz"); + break; + case R300_ALU_ARGC_SRCP_XXX: + sprintf(buf, "srcp.xxx"); + break; + case R300_ALU_ARGC_SRCP_YYY: + sprintf(buf, "srcp.yyy"); + break; + case R300_ALU_ARGC_SRCP_ZZZ: + sprintf(buf, "srcp.zzz"); + break; + case R300_ALU_ARGC_SRCP_WWW: + sprintf(buf, "srcp.www"); + break; + } + } else if (d == 20) { + sprintf(buf, "0.0"); + } else if (d == 21) { + sprintf(buf, "1.0"); + } else if (d == 22) { + sprintf(buf, "0.5"); + } else if (d >= 23 && d < 32) { + d -= 23; + switch (d / 3) { + case 0: + sprintf(buf, "%s.yzx", + srcc[d % 3]); + break; + case 1: + sprintf(buf, "%s.zxy", + srcc[d % 3]); + break; + case 2: + sprintf(buf, "%s.Wzy", + srcc[d % 3]); + break; + } + } else { + sprintf(buf, "%i", d); + } + + sprintf(argc[j], "%s%s%s%s", + (regc & 32) ? "-" : "", + (regc & 64) ? "|" : "", + buf, (regc & 64) ? "|" : ""); + + d = rega & 31; + if (d < 9) { + sprintf(buf, "%s.%c", srcc[d / 3], + 'x' + (char)(d % 3)); + } else if (d < 12) { + sprintf(buf, "%s.w", srca[d - 9]); + } else if (d < 16) { + switch(d) { + case R300_ALU_ARGA_SRCP_X: + sprintf(buf, "srcp.x"); + break; + case R300_ALU_ARGA_SRCP_Y: + sprintf(buf, "srcp.y"); + break; + case R300_ALU_ARGA_SRCP_Z: + sprintf(buf, "srcp.z"); + break; + case R300_ALU_ARGA_SRCP_W: + sprintf(buf, "srcp.w"); + break; + } + } else if (d == 16) { + sprintf(buf, "0.0"); + } else if (d == 17) { + sprintf(buf, "1.0"); + } else if (d == 18) { + sprintf(buf, "0.5"); + } else { + sprintf(buf, "%i", d); + } + + sprintf(arga[j], "%s%s%s%s", + (rega & 32) ? "-" : "", + (rega & 64) ? "|" : "", + buf, (rega & 64) ? "|" : ""); + } + + fprintf(stderr, " xyz: %8s %8s %8s op: %08x %s\n" + " w: %8s %8s %8s op: %08x\n", + argc[0], argc[1], argc[2], + code->alu.inst[i].rgb_inst, + code->alu.inst[i].rgb_inst & R300_ALU_INSERT_NOP ? + "NOP" : "", + arga[0], arga[1],arga[2], + code->alu.inst[i].alpha_inst); + } + } +} diff --git a/src/gallium/drivers/r300/compiler/r300_fragprog.h b/src/gallium/drivers/r300/compiler/r300_fragprog.h new file mode 100644 index 00000000000..0c88bab2f33 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/r300_fragprog.h @@ -0,0 +1,44 @@ +/* + * Copyright (C) 2005 Ben Skeggs. + * + * 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. + * + */ + +/* + * Authors: + * Ben Skeggs <[email protected]> + * Jerome Glisse <[email protected]> + */ +#ifndef __R300_FRAGPROG_H_ +#define __R300_FRAGPROG_H_ + +#include "radeon_compiler.h" +#include "radeon_program.h" + + +extern void r300BuildFragmentProgramHwCode(struct radeon_compiler *c, void *user); + +extern void r300FragmentProgramDump(struct radeon_compiler *c, void *user); + +#endif diff --git a/src/gallium/drivers/r300/compiler/r300_fragprog_emit.c b/src/gallium/drivers/r300/compiler/r300_fragprog_emit.c new file mode 100644 index 00000000000..e6fd1fde62d --- /dev/null +++ b/src/gallium/drivers/r300/compiler/r300_fragprog_emit.c @@ -0,0 +1,536 @@ +/* + * Copyright (C) 2005 Ben Skeggs. + * + * 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 + * + * Emit the r300_fragment_program_code that can be understood by the hardware. + * Input is a pre-transformed radeon_program. + * + * \author Ben Skeggs <[email protected]> + * + * \author Jerome Glisse <[email protected]> + */ + +#include "r300_fragprog.h" + +#include "../r300_reg.h" + +#include "radeon_program_pair.h" +#include "r300_fragprog_swizzle.h" + + +struct r300_emit_state { + struct r300_fragment_program_compiler * compiler; + + unsigned current_node : 2; + unsigned node_first_tex : 8; + unsigned node_first_alu : 8; + uint32_t node_flags; +}; + +#define PROG_CODE \ + struct r300_fragment_program_compiler *c = emit->compiler; \ + struct r300_fragment_program_code *code = &c->code->code.r300 + +#define error(fmt, args...) do { \ + rc_error(&c->Base, "%s::%s(): " fmt "\n", \ + __FILE__, __FUNCTION__, ##args); \ + } while(0) + +static unsigned int get_msbs_alu(unsigned int bits) +{ + return (bits >> 6) & 0x7; +} + +/** + * @param lsbs The number of least significant bits + */ +static unsigned int get_msbs_tex(unsigned int bits, unsigned int lsbs) +{ + return (bits >> lsbs) & 0x15; +} + +#define R400_EXT_GET_MSBS(x, lsbs, mask) (((x) >> lsbs) & mask) + +/** + * Mark a temporary register as used. + */ +static void use_temporary(struct r300_fragment_program_code *code, unsigned int index) +{ + if (index > code->pixsize) + code->pixsize = index; +} + +static unsigned int use_source(struct r300_fragment_program_code* code, struct rc_pair_instruction_source src) +{ + if (!src.Used) + return 0; + + if (src.File == RC_FILE_CONSTANT) { + return src.Index | (1 << 5); + } else if (src.File == RC_FILE_TEMPORARY || src.File == RC_FILE_INPUT) { + use_temporary(code, src.Index); + return src.Index & 0x1f; + } + + return 0; +} + + +static unsigned int translate_rgb_opcode(struct r300_fragment_program_compiler * c, rc_opcode opcode) +{ + switch(opcode) { + case RC_OPCODE_CMP: return R300_ALU_OUTC_CMP; + case RC_OPCODE_CND: return R300_ALU_OUTC_CND; + case RC_OPCODE_DP3: return R300_ALU_OUTC_DP3; + case RC_OPCODE_DP4: return R300_ALU_OUTC_DP4; + case RC_OPCODE_FRC: return R300_ALU_OUTC_FRC; + default: + error("translate_rgb_opcode: Unknown opcode %s", rc_get_opcode_info(opcode)->Name); + /* fall through */ + case RC_OPCODE_NOP: + /* fall through */ + case RC_OPCODE_MAD: return R300_ALU_OUTC_MAD; + case RC_OPCODE_MAX: return R300_ALU_OUTC_MAX; + case RC_OPCODE_MIN: return R300_ALU_OUTC_MIN; + case RC_OPCODE_REPL_ALPHA: return R300_ALU_OUTC_REPL_ALPHA; + } +} + +static unsigned int translate_alpha_opcode(struct r300_fragment_program_compiler * c, rc_opcode opcode) +{ + switch(opcode) { + case RC_OPCODE_CMP: return R300_ALU_OUTA_CMP; + case RC_OPCODE_CND: return R300_ALU_OUTA_CND; + case RC_OPCODE_DP3: return R300_ALU_OUTA_DP4; + case RC_OPCODE_DP4: return R300_ALU_OUTA_DP4; + case RC_OPCODE_EX2: return R300_ALU_OUTA_EX2; + case RC_OPCODE_FRC: return R300_ALU_OUTA_FRC; + case RC_OPCODE_LG2: return R300_ALU_OUTA_LG2; + default: + error("translate_rgb_opcode: Unknown opcode %s", rc_get_opcode_info(opcode)->Name); + /* fall through */ + case RC_OPCODE_NOP: + /* fall through */ + case RC_OPCODE_MAD: return R300_ALU_OUTA_MAD; + case RC_OPCODE_MAX: return R300_ALU_OUTA_MAX; + case RC_OPCODE_MIN: return R300_ALU_OUTA_MIN; + case RC_OPCODE_RCP: return R300_ALU_OUTA_RCP; + case RC_OPCODE_RSQ: return R300_ALU_OUTA_RSQ; + } +} + +/** + * Emit one paired ALU instruction. + */ +static int emit_alu(struct r300_emit_state * emit, struct rc_pair_instruction* inst) +{ + int ip; + int j; + PROG_CODE; + + if (code->alu.length >= c->Base.max_alu_insts) { + error("Too many ALU instructions"); + return 0; + } + + ip = code->alu.length++; + + code->alu.inst[ip].rgb_inst = translate_rgb_opcode(c, inst->RGB.Opcode); + code->alu.inst[ip].alpha_inst = translate_alpha_opcode(c, inst->Alpha.Opcode); + + for(j = 0; j < 3; ++j) { + /* Set the RGB address */ + unsigned int src = use_source(code, inst->RGB.Src[j]); + unsigned int arg; + if (inst->RGB.Src[j].Index >= R300_PFS_NUM_TEMP_REGS) + code->alu.inst[ip].r400_ext_addr |= R400_ADDR_EXT_RGB_MSB_BIT(j); + + code->alu.inst[ip].rgb_addr |= src << (6*j); + + /* Set the Alpha address */ + src = use_source(code, inst->Alpha.Src[j]); + if (inst->Alpha.Src[j].Index >= R300_PFS_NUM_TEMP_REGS) + code->alu.inst[ip].r400_ext_addr |= R400_ADDR_EXT_A_MSB_BIT(j); + + code->alu.inst[ip].alpha_addr |= src << (6*j); + + arg = r300FPTranslateRGBSwizzle(inst->RGB.Arg[j].Source, inst->RGB.Arg[j].Swizzle); + arg |= inst->RGB.Arg[j].Abs << 6; + arg |= inst->RGB.Arg[j].Negate << 5; + code->alu.inst[ip].rgb_inst |= arg << (7*j); + + arg = r300FPTranslateAlphaSwizzle(inst->Alpha.Arg[j].Source, inst->Alpha.Arg[j].Swizzle); + arg |= inst->Alpha.Arg[j].Abs << 6; + arg |= inst->Alpha.Arg[j].Negate << 5; + code->alu.inst[ip].alpha_inst |= arg << (7*j); + } + + /* Presubtract */ + if (inst->RGB.Src[RC_PAIR_PRESUB_SRC].Used) { + switch(inst->RGB.Src[RC_PAIR_PRESUB_SRC].Index) { + case RC_PRESUB_BIAS: + code->alu.inst[ip].rgb_inst |= + R300_ALU_SRCP_1_MINUS_2_SRC0; + break; + case RC_PRESUB_ADD: + code->alu.inst[ip].rgb_inst |= + R300_ALU_SRCP_SRC1_PLUS_SRC0; + break; + case RC_PRESUB_SUB: + code->alu.inst[ip].rgb_inst |= + R300_ALU_SRCP_SRC1_MINUS_SRC0; + break; + case RC_PRESUB_INV: + code->alu.inst[ip].rgb_inst |= + R300_ALU_SRCP_1_MINUS_SRC0; + break; + default: + break; + } + } + + if (inst->Alpha.Src[RC_PAIR_PRESUB_SRC].Used) { + switch(inst->Alpha.Src[RC_PAIR_PRESUB_SRC].Index) { + case RC_PRESUB_BIAS: + code->alu.inst[ip].alpha_inst |= + R300_ALU_SRCP_1_MINUS_2_SRC0; + break; + case RC_PRESUB_ADD: + code->alu.inst[ip].alpha_inst |= + R300_ALU_SRCP_SRC1_PLUS_SRC0; + break; + case RC_PRESUB_SUB: + code->alu.inst[ip].alpha_inst |= + R300_ALU_SRCP_SRC1_MINUS_SRC0; + break; + case RC_PRESUB_INV: + code->alu.inst[ip].alpha_inst |= + R300_ALU_SRCP_1_MINUS_SRC0; + break; + default: + break; + } + } + + if (inst->RGB.Saturate) + code->alu.inst[ip].rgb_inst |= R300_ALU_OUTC_CLAMP; + if (inst->Alpha.Saturate) + code->alu.inst[ip].alpha_inst |= R300_ALU_OUTA_CLAMP; + + if (inst->RGB.WriteMask) { + use_temporary(code, inst->RGB.DestIndex); + if (inst->RGB.DestIndex >= R300_PFS_NUM_TEMP_REGS) + code->alu.inst[ip].r400_ext_addr |= R400_ADDRD_EXT_RGB_MSB_BIT; + code->alu.inst[ip].rgb_addr |= + ((inst->RGB.DestIndex & 0x1f) << R300_ALU_DSTC_SHIFT) | + (inst->RGB.WriteMask << R300_ALU_DSTC_REG_MASK_SHIFT); + } + if (inst->RGB.OutputWriteMask) { + code->alu.inst[ip].rgb_addr |= + (inst->RGB.OutputWriteMask << R300_ALU_DSTC_OUTPUT_MASK_SHIFT) | + R300_RGB_TARGET(inst->RGB.Target); + emit->node_flags |= R300_RGBA_OUT; + } + + if (inst->Alpha.WriteMask) { + use_temporary(code, inst->Alpha.DestIndex); + if (inst->Alpha.DestIndex >= R300_PFS_NUM_TEMP_REGS) + code->alu.inst[ip].r400_ext_addr |= R400_ADDRD_EXT_A_MSB_BIT; + code->alu.inst[ip].alpha_addr |= + ((inst->Alpha.DestIndex & 0x1f) << R300_ALU_DSTA_SHIFT) | + R300_ALU_DSTA_REG; + } + if (inst->Alpha.OutputWriteMask) { + code->alu.inst[ip].alpha_addr |= R300_ALU_DSTA_OUTPUT | + R300_ALPHA_TARGET(inst->Alpha.Target); + emit->node_flags |= R300_RGBA_OUT; + } + if (inst->Alpha.DepthWriteMask) { + code->alu.inst[ip].alpha_addr |= R300_ALU_DSTA_DEPTH; + emit->node_flags |= R300_W_OUT; + c->code->writes_depth = 1; + } + if (inst->Nop) + code->alu.inst[ip].rgb_inst |= R300_ALU_INSERT_NOP; + + return 1; +} + + +/** + * Finish the current node without advancing to the next one. + */ +static int finish_node(struct r300_emit_state * emit) +{ + struct r300_fragment_program_compiler * c = emit->compiler; + struct r300_fragment_program_code *code = &emit->compiler->code->code.r300; + unsigned alu_offset; + unsigned alu_end; + unsigned tex_offset; + unsigned tex_end; + + unsigned int alu_offset_msbs, alu_end_msbs; + + if (code->alu.length == emit->node_first_alu) { + /* Generate a single NOP for this node */ + struct rc_pair_instruction inst; + memset(&inst, 0, sizeof(inst)); + if (!emit_alu(emit, &inst)) + return 0; + } + + alu_offset = emit->node_first_alu; + alu_end = code->alu.length - alu_offset - 1; + tex_offset = emit->node_first_tex; + tex_end = code->tex.length - tex_offset - 1; + + if (code->tex.length == emit->node_first_tex) { + if (emit->current_node > 0) { + error("Node %i has no TEX instructions", emit->current_node); + return 0; + } + + tex_end = 0; + } else { + if (emit->current_node == 0) + code->config |= R300_PFS_CNTL_FIRST_NODE_HAS_TEX; + } + + /* Write the config register. + * Note: The order in which the words for each node are written + * is not correct here and needs to be fixed up once we're entirely + * done + * + * Also note that the register specification from AMD is slightly + * incorrect in its description of this register. */ + code->code_addr[emit->current_node] = + ((alu_offset << R300_ALU_START_SHIFT) + & R300_ALU_START_MASK) + | ((alu_end << R300_ALU_SIZE_SHIFT) + & R300_ALU_SIZE_MASK) + | ((tex_offset << R300_TEX_START_SHIFT) + & R300_TEX_START_MASK) + | ((tex_end << R300_TEX_SIZE_SHIFT) + & R300_TEX_SIZE_MASK) + | emit->node_flags + | (get_msbs_tex(tex_offset, 5) + << R400_TEX_START_MSB_SHIFT) + | (get_msbs_tex(tex_end, 5) + << R400_TEX_SIZE_MSB_SHIFT) + ; + + /* Write r400 extended instruction fields. These will be ignored on + * r300 cards. */ + alu_offset_msbs = get_msbs_alu(alu_offset); + alu_end_msbs = get_msbs_alu(alu_end); + switch(emit->current_node) { + case 0: + code->r400_code_offset_ext |= + alu_offset_msbs << R400_ALU_START3_MSB_SHIFT + | alu_end_msbs << R400_ALU_SIZE3_MSB_SHIFT; + break; + case 1: + code->r400_code_offset_ext |= + alu_offset_msbs << R400_ALU_START2_MSB_SHIFT + | alu_end_msbs << R400_ALU_SIZE2_MSB_SHIFT; + break; + case 2: + code->r400_code_offset_ext |= + alu_offset_msbs << R400_ALU_START1_MSB_SHIFT + | alu_end_msbs << R400_ALU_SIZE1_MSB_SHIFT; + break; + case 3: + code->r400_code_offset_ext |= + alu_offset_msbs << R400_ALU_START0_MSB_SHIFT + | alu_end_msbs << R400_ALU_SIZE0_MSB_SHIFT; + break; + } + return 1; +} + + +/** + * Begin a block of texture instructions. + * Create the necessary indirection. + */ +static int begin_tex(struct r300_emit_state * emit) +{ + PROG_CODE; + + if (code->alu.length == emit->node_first_alu && + code->tex.length == emit->node_first_tex) { + return 1; + } + + if (emit->current_node == 3) { + error("Too many texture indirections"); + return 0; + } + + if (!finish_node(emit)) + return 0; + + emit->current_node++; + emit->node_first_tex = code->tex.length; + emit->node_first_alu = code->alu.length; + emit->node_flags = 0; + return 1; +} + + +static int emit_tex(struct r300_emit_state * emit, struct rc_instruction * inst) +{ + unsigned int unit; + unsigned int dest; + unsigned int opcode; + PROG_CODE; + + if (code->tex.length >= emit->compiler->Base.max_tex_insts) { + error("Too many TEX instructions"); + return 0; + } + + unit = inst->U.I.TexSrcUnit; + dest = inst->U.I.DstReg.Index; + + switch(inst->U.I.Opcode) { + case RC_OPCODE_KIL: opcode = R300_TEX_OP_KIL; break; + case RC_OPCODE_TEX: opcode = R300_TEX_OP_LD; break; + case RC_OPCODE_TXB: opcode = R300_TEX_OP_TXB; break; + case RC_OPCODE_TXP: opcode = R300_TEX_OP_TXP; break; + default: + error("Unknown texture opcode %s", rc_get_opcode_info(inst->U.I.Opcode)->Name); + return 0; + } + + if (inst->U.I.Opcode == RC_OPCODE_KIL) { + unit = 0; + dest = 0; + } else { + use_temporary(code, dest); + } + + use_temporary(code, inst->U.I.SrcReg[0].Index); + + code->tex.inst[code->tex.length++] = + ((inst->U.I.SrcReg[0].Index << R300_SRC_ADDR_SHIFT) + & R300_SRC_ADDR_MASK) + | ((dest << R300_DST_ADDR_SHIFT) + & R300_DST_ADDR_MASK) + | (unit << R300_TEX_ID_SHIFT) + | (opcode << R300_TEX_INST_SHIFT) + | (inst->U.I.SrcReg[0].Index >= R300_PFS_NUM_TEMP_REGS ? + R400_SRC_ADDR_EXT_BIT : 0) + | (dest >= R300_PFS_NUM_TEMP_REGS ? + R400_DST_ADDR_EXT_BIT : 0) + ; + return 1; +} + + +/** + * Final compilation step: Turn the intermediate radeon_program into + * machine-readable instructions. + */ +void r300BuildFragmentProgramHwCode(struct radeon_compiler *c, void *user) +{ + struct r300_fragment_program_compiler *compiler = (struct r300_fragment_program_compiler*)c; + struct r300_emit_state emit; + struct r300_fragment_program_code *code = &compiler->code->code.r300; + unsigned int tex_end; + + memset(&emit, 0, sizeof(emit)); + emit.compiler = compiler; + + memset(code, 0, sizeof(struct r300_fragment_program_code)); + + 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) { + if (inst->U.I.Opcode == RC_OPCODE_BEGIN_TEX) { + begin_tex(&emit); + continue; + } + + emit_tex(&emit, inst); + } else { + emit_alu(&emit, &inst->U.P); + } + } + + if (code->pixsize >= compiler->Base.max_temp_regs) + rc_error(&compiler->Base, "Too many hardware temporaries used.\n"); + + if (compiler->Base.Error) + return; + + /* Finish the program */ + finish_node(&emit); + + code->config |= emit.current_node; /* FIRST_NODE_HAS_TEX set by finish_node */ + + /* Set r400 extended instruction fields. These values will be ignored + * on r300 cards. */ + code->r400_code_offset_ext |= + (get_msbs_alu(0) + << R400_ALU_OFFSET_MSB_SHIFT) + | (get_msbs_alu(code->alu.length - 1) + << R400_ALU_SIZE_MSB_SHIFT); + + tex_end = code->tex.length ? code->tex.length - 1 : 0; + code->code_offset = + ((0 << R300_PFS_CNTL_ALU_OFFSET_SHIFT) + & R300_PFS_CNTL_ALU_OFFSET_MASK) + | (((code->alu.length - 1) << R300_PFS_CNTL_ALU_END_SHIFT) + & R300_PFS_CNTL_ALU_END_MASK) + | ((0 << R300_PFS_CNTL_TEX_OFFSET_SHIFT) + & R300_PFS_CNTL_TEX_OFFSET_MASK) + | ((tex_end << R300_PFS_CNTL_TEX_END_SHIFT) + & R300_PFS_CNTL_TEX_END_MASK) + | (get_msbs_tex(0, 5) << R400_TEX_START_MSB_SHIFT) + | (get_msbs_tex(tex_end, 6) << R400_TEX_SIZE_MSB_SHIFT) + ; + + if (emit.current_node < 3) { + int shift = 3 - emit.current_node; + int i; + for(i = emit.current_node; i >= 0; --i) + code->code_addr[shift + i] = code->code_addr[i]; + for(i = 0; i < shift; ++i) + code->code_addr[i] = 0; + } + + if (code->pixsize >= R300_PFS_NUM_TEMP_REGS + || code->alu.length > R300_PFS_MAX_ALU_INST + || code->tex.length > R300_PFS_MAX_TEX_INST) { + + code->r390_mode = 1; + } +} diff --git a/src/gallium/drivers/r300/compiler/r300_fragprog_swizzle.c b/src/gallium/drivers/r300/compiler/r300_fragprog_swizzle.c new file mode 100644 index 00000000000..b7bca8c0cfa --- /dev/null +++ b/src/gallium/drivers/r300/compiler/r300_fragprog_swizzle.c @@ -0,0 +1,243 @@ +/* + * Copyright (C) 2008 Nicolai Haehnle. + * + * 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 + * Utilities to deal with the somewhat odd restriction on R300 fragment + * program swizzles. + */ + +#include "r300_fragprog_swizzle.h" + +#include <stdio.h> + +#include "../r300_reg.h" +#include "radeon_compiler.h" + +#define MAKE_SWZ3(x, y, z) (RC_MAKE_SWIZZLE(RC_SWIZZLE_##x, RC_SWIZZLE_##y, RC_SWIZZLE_##z, RC_SWIZZLE_ZERO)) + +struct swizzle_data { + unsigned int hash; /**< swizzle value this matches */ + unsigned int base; /**< base value for hw swizzle */ + unsigned int stride; /**< difference in base between arg0/1/2 */ + unsigned int srcp_stride; /**< difference in base between arg0/scrp */ +}; + +static const struct swizzle_data native_swizzles[] = { + {MAKE_SWZ3(X, Y, Z), R300_ALU_ARGC_SRC0C_XYZ, 4, 15}, + {MAKE_SWZ3(X, X, X), R300_ALU_ARGC_SRC0C_XXX, 4, 15}, + {MAKE_SWZ3(Y, Y, Y), R300_ALU_ARGC_SRC0C_YYY, 4, 15}, + {MAKE_SWZ3(Z, Z, Z), R300_ALU_ARGC_SRC0C_ZZZ, 4, 15}, + {MAKE_SWZ3(W, W, W), R300_ALU_ARGC_SRC0A, 1, 7}, + {MAKE_SWZ3(Y, Z, X), R300_ALU_ARGC_SRC0C_YZX, 1, 0}, + {MAKE_SWZ3(Z, X, Y), R300_ALU_ARGC_SRC0C_ZXY, 1, 0}, + {MAKE_SWZ3(W, Z, Y), R300_ALU_ARGC_SRC0CA_WZY, 1, 0}, + {MAKE_SWZ3(ONE, ONE, ONE), R300_ALU_ARGC_ONE, 0, 0}, + {MAKE_SWZ3(ZERO, ZERO, ZERO), R300_ALU_ARGC_ZERO, 0, 0}, + {MAKE_SWZ3(HALF, HALF, HALF), R300_ALU_ARGC_HALF, 0, 0} +}; + +static const int num_native_swizzles = sizeof(native_swizzles)/sizeof(native_swizzles[0]); + +/** + * Find a native RGB swizzle that matches the given swizzle. + * Returns 0 if none found. + */ +static const struct swizzle_data* lookup_native_swizzle(unsigned int swizzle) +{ + int i, comp; + + for(i = 0; i < num_native_swizzles; ++i) { + const struct swizzle_data* sd = &native_swizzles[i]; + for(comp = 0; comp < 3; ++comp) { + unsigned int swz = GET_SWZ(swizzle, comp); + if (swz == RC_SWIZZLE_UNUSED) + continue; + if (swz != GET_SWZ(sd->hash, comp)) + break; + } + if (comp == 3) + return sd; + } + + return 0; +} + +/** + * Determines if the given swizzle is valid for r300/r400. In most situations + * it is better to use r300_swizzle_is_native() which can be accesed via + * struct radeon_compiler *c; c->SwizzleCaps->IsNative(). + */ +int r300_swizzle_is_native_basic(unsigned int swizzle) +{ + if(lookup_native_swizzle(swizzle)) + return 1; + else + return 0; +} + +/** + * Check whether the given instruction supports the swizzle and negate + * combinations in the given source register. + */ +static int r300_swizzle_is_native(rc_opcode opcode, struct rc_src_register reg) +{ + const struct swizzle_data* sd; + unsigned int relevant; + int j; + + if (opcode == RC_OPCODE_KIL || + opcode == RC_OPCODE_TEX || + opcode == RC_OPCODE_TXB || + opcode == RC_OPCODE_TXP) { + if (reg.Abs || reg.Negate) + return 0; + + for(j = 0; j < 4; ++j) { + unsigned int swz = GET_SWZ(reg.Swizzle, j); + if (swz == RC_SWIZZLE_UNUSED) + continue; + if (swz != j) + return 0; + } + + return 1; + } + + relevant = 0; + + for(j = 0; j < 3; ++j) + if (GET_SWZ(reg.Swizzle, j) != RC_SWIZZLE_UNUSED) + relevant |= 1 << j; + + if ((reg.Negate & relevant) && ((reg.Negate & relevant) != relevant)) + return 0; + + sd = lookup_native_swizzle(reg.Swizzle); + if (!sd || (reg.File == RC_FILE_PRESUB && sd->srcp_stride == 0)) + return 0; + + return 1; +} + + +static void r300_swizzle_split( + struct rc_src_register src, unsigned int mask, + struct rc_swizzle_split * split) +{ + split->NumPhases = 0; + + while(mask) { + unsigned int best_matchcount = 0; + unsigned int best_matchmask = 0; + int i, comp; + + for(i = 0; i < num_native_swizzles; ++i) { + const struct swizzle_data *sd = &native_swizzles[i]; + unsigned int matchcount = 0; + unsigned int matchmask = 0; + for(comp = 0; comp < 3; ++comp) { + unsigned int swz; + if (!GET_BIT(mask, comp)) + continue; + swz = GET_SWZ(src.Swizzle, comp); + if (swz == RC_SWIZZLE_UNUSED) + continue; + if (swz == GET_SWZ(sd->hash, comp)) { + /* check if the negate bit of current component + * is the same for already matched components */ + if (matchmask && (!!(src.Negate & matchmask) != !!(src.Negate & (1 << comp)))) + continue; + + matchcount++; + matchmask |= 1 << comp; + } + } + if (matchcount > best_matchcount) { + best_matchcount = matchcount; + best_matchmask = matchmask; + if (matchmask == (mask & RC_MASK_XYZ)) + break; + } + } + + if (mask & RC_MASK_W) + best_matchmask |= RC_MASK_W; + + split->Phase[split->NumPhases++] = best_matchmask; + mask &= ~best_matchmask; + } +} + +struct rc_swizzle_caps r300_swizzle_caps = { + .IsNative = r300_swizzle_is_native, + .Split = r300_swizzle_split +}; + + +/** + * Translate an RGB (XYZ) swizzle into the hardware code for the given + * instruction source. + */ +unsigned int r300FPTranslateRGBSwizzle(unsigned int src, unsigned int swizzle) +{ + const struct swizzle_data* sd = lookup_native_swizzle(swizzle); + + if (!sd || (src == RC_PAIR_PRESUB_SRC && sd->srcp_stride == 0)) { + fprintf(stderr, "Not a native swizzle: %08x\n", swizzle); + return 0; + } + + if (src == RC_PAIR_PRESUB_SRC) { + return sd->base + sd->srcp_stride; + } else { + return sd->base + src*sd->stride; + } +} + + +/** + * Translate an Alpha (W) swizzle into the hardware code for the given + * instruction source. + */ +unsigned int r300FPTranslateAlphaSwizzle(unsigned int src, unsigned int swizzle) +{ + unsigned int swz = GET_SWZ(swizzle, 0); + if (src == RC_PAIR_PRESUB_SRC) { + return R300_ALU_ARGA_SRCP_X + swz; + } + if (swz < 3) + return swz + 3*src; + + switch(swz) { + case RC_SWIZZLE_W: return R300_ALU_ARGA_SRC0A + src; + case RC_SWIZZLE_ONE: return R300_ALU_ARGA_ONE; + case RC_SWIZZLE_ZERO: return R300_ALU_ARGA_ZERO; + case RC_SWIZZLE_HALF: return R300_ALU_ARGA_HALF; + default: return R300_ALU_ARGA_ONE; + } +} diff --git a/src/gallium/drivers/r300/compiler/r300_fragprog_swizzle.h b/src/gallium/drivers/r300/compiler/r300_fragprog_swizzle.h new file mode 100644 index 00000000000..f2635be140d --- /dev/null +++ b/src/gallium/drivers/r300/compiler/r300_fragprog_swizzle.h @@ -0,0 +1,39 @@ +/* + * Copyright (C) 2008 Nicolai Haehnle. + * + * 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. + * + */ + +#ifndef __R300_FRAGPROG_SWIZZLE_H_ +#define __R300_FRAGPROG_SWIZZLE_H_ + +#include "radeon_swizzle.h" + +extern struct rc_swizzle_caps r300_swizzle_caps; + +unsigned int r300FPTranslateRGBSwizzle(unsigned int src, unsigned int swizzle); +unsigned int r300FPTranslateAlphaSwizzle(unsigned int src, unsigned int swizzle); +int r300_swizzle_is_native_basic(unsigned int swizzle); + +#endif /* __R300_FRAGPROG_SWIZZLE_H_ */ diff --git a/src/gallium/drivers/r300/compiler/r3xx_fragprog.c b/src/gallium/drivers/r300/compiler/r3xx_fragprog.c new file mode 100644 index 00000000000..bb6c010e8e3 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/r3xx_fragprog.c @@ -0,0 +1,172 @@ +/* + * Copyright 2009 Nicolai Hähnle <[email protected]> + * + * 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 + * on the rights to use, copy, modify, merge, publish, distribute, sub + * license, 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 NON-INFRINGEMENT. IN NO EVENT SHALL + * THE AUTHOR(S) AND/OR THEIR 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. */ + +#include "radeon_compiler.h" + +#include <stdio.h> + +#include "radeon_compiler_util.h" +#include "radeon_dataflow.h" +#include "radeon_emulate_branches.h" +#include "radeon_emulate_loops.h" +#include "radeon_program_alu.h" +#include "radeon_program_tex.h" +#include "radeon_rename_regs.h" +#include "radeon_remove_constants.h" +#include "r300_fragprog.h" +#include "r300_fragprog_swizzle.h" +#include "r500_fragprog.h" + + +static void dataflow_outputs_mark_use(void * userdata, void * data, + void (*callback)(void *, unsigned int, unsigned int)) +{ + struct r300_fragment_program_compiler * c = userdata; + callback(data, c->OutputColor[0], RC_MASK_XYZW); + callback(data, c->OutputColor[1], RC_MASK_XYZW); + callback(data, c->OutputColor[2], RC_MASK_XYZW); + callback(data, c->OutputColor[3], RC_MASK_XYZW); + callback(data, c->OutputDepth, RC_MASK_W); +} + +static void rc_rewrite_depth_out(struct radeon_compiler *cc, void *user) +{ + struct r300_fragment_program_compiler *c = (struct r300_fragment_program_compiler*)cc; + struct rc_instruction *rci; + + for (rci = c->Base.Program.Instructions.Next; rci != &c->Base.Program.Instructions; rci = rci->Next) { + struct rc_sub_instruction * inst = &rci->U.I; + unsigned i; + const struct rc_opcode_info *info = rc_get_opcode_info(inst->Opcode); + + if (inst->DstReg.File != RC_FILE_OUTPUT || inst->DstReg.Index != c->OutputDepth) + continue; + + if (inst->DstReg.WriteMask & RC_MASK_Z) { + inst->DstReg.WriteMask = RC_MASK_W; + } else { + inst->DstReg.WriteMask = 0; + continue; + } + + if (!info->IsComponentwise) { + continue; + } + + for (i = 0; i < info->NumSrcRegs; i++) { + inst->SrcReg[i] = lmul_swizzle(RC_SWIZZLE_ZZZZ, inst->SrcReg[i]); + } + } +} + +static int radeon_saturate_output( + struct radeon_compiler * c, + struct rc_instruction * inst, + void* data) +{ + const struct rc_opcode_info *info = rc_get_opcode_info(inst->U.I.Opcode); + + if (!info->HasDstReg || inst->U.I.DstReg.File != RC_FILE_OUTPUT) + return 0; + + inst->U.I.SaturateMode = RC_SATURATE_ZERO_ONE; + return 1; +} + +void r3xx_compile_fragment_program(struct r300_fragment_program_compiler* c) +{ + int is_r500 = c->Base.is_r500; + int opt = !c->Base.disable_optimizations; + int sat_out = c->state.frag_clamp; + + /* Lists of instruction transformations. */ + struct radeon_program_transformation saturate_output[] = { + { &radeon_saturate_output, c }, + { 0, 0 } + }; + + struct radeon_program_transformation rewrite_tex[] = { + { &radeonTransformTEX, c }, + { 0, 0 } + }; + + struct radeon_program_transformation rewrite_if[] = { + { &r500_transform_IF, 0 }, + {0, 0} + }; + + struct radeon_program_transformation native_rewrite_r500[] = { + { &radeonTransformALU, 0 }, + { &radeonTransformDeriv, 0 }, + { &radeonTransformTrigScale, 0 }, + { 0, 0 } + }; + + struct radeon_program_transformation native_rewrite_r300[] = { + { &radeonTransformALU, 0 }, + { &r300_transform_trig_simple, 0 }, + { 0, 0 } + }; + + /* List of compiler passes. */ + struct radeon_compiler_pass fs_list[] = { + /* NAME DUMP PREDICATE FUNCTION PARAM */ + {"rewrite depth out", 1, 1, rc_rewrite_depth_out, NULL}, + /* This transformation needs to be done before any of the IF + * instructions are modified. */ + {"transform KILP", 1, 1, rc_transform_KILP, NULL}, + {"unroll loops", 1, is_r500, rc_unroll_loops, NULL}, + {"transform loops", 1, !is_r500, rc_transform_loops, NULL}, + {"emulate branches", 1, !is_r500, rc_emulate_branches, NULL}, + {"saturate output writes", 1, sat_out, rc_local_transform, saturate_output}, + {"transform TEX", 1, 1, rc_local_transform, rewrite_tex}, + {"transform IF", 1, is_r500, rc_local_transform, rewrite_if}, + {"native rewrite", 1, is_r500, rc_local_transform, native_rewrite_r500}, + {"native rewrite", 1, !is_r500, rc_local_transform, native_rewrite_r300}, + {"deadcode", 1, opt, rc_dataflow_deadcode, dataflow_outputs_mark_use}, + {"emulate loops", 1, !is_r500, rc_emulate_loops, NULL}, + {"dataflow optimize", 1, opt, rc_optimize, NULL}, + {"dataflow swizzles", 1, 1, rc_dataflow_swizzles, NULL}, + {"dead constants", 1, 1, rc_remove_unused_constants, &c->code->constants_remap_table}, + /* This pass makes it easier for the scheduler to group TEX + * instructions and reduces the chances of creating too + * many texture indirections.*/ + {"register rename", 1, !is_r500, rc_rename_regs, NULL}, + {"pair translate", 1, 1, rc_pair_translate, NULL}, + {"pair scheduling", 1, 1, rc_pair_schedule, NULL}, + {"dead sources", 1, 1, rc_pair_remove_dead_sources, NULL}, + {"register allocation", 1, 1, rc_pair_regalloc, &opt}, + {"final code validation", 0, 1, rc_validate_final_shader, NULL}, + {"machine code generation", 0, is_r500, r500BuildFragmentProgramHwCode, NULL}, + {"machine code generation", 0, !is_r500, r300BuildFragmentProgramHwCode, NULL}, + {"dump machine code", 0, is_r500 && (c->Base.Debug & RC_DBG_LOG), r500FragmentProgramDump, NULL}, + {"dump machine code", 0, !is_r500 && (c->Base.Debug & RC_DBG_LOG), r300FragmentProgramDump, NULL}, + {NULL, 0, 0, NULL, NULL} + }; + + c->Base.type = RC_FRAGMENT_PROGRAM; + c->Base.SwizzleCaps = c->Base.is_r500 ? &r500_swizzle_caps : &r300_swizzle_caps; + + rc_run_compiler(&c->Base, fs_list); + + rc_constants_copy(&c->code->constants, &c->Base.Program.Constants); +} diff --git a/src/gallium/drivers/r300/compiler/r3xx_vertprog.c b/src/gallium/drivers/r300/compiler/r3xx_vertprog.c new file mode 100644 index 00000000000..654f9a070d5 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/r3xx_vertprog.c @@ -0,0 +1,1045 @@ +/* + * Copyright 2009 Nicolai Hähnle <[email protected]> + * + * 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 + * on the rights to use, copy, modify, merge, publish, distribute, sub + * license, 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 NON-INFRINGEMENT. IN NO EVENT SHALL + * THE AUTHOR(S) AND/OR THEIR 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. */ + +#include "radeon_compiler.h" + +#include <stdio.h> + +#include "../r300_reg.h" + +#include "radeon_compiler_util.h" +#include "radeon_dataflow.h" +#include "radeon_program_alu.h" +#include "radeon_swizzle.h" +#include "radeon_emulate_branches.h" +#include "radeon_emulate_loops.h" +#include "radeon_remove_constants.h" + +struct loop { + int BgnLoop; + +}; + +/* + * Take an already-setup and valid source then swizzle it appropriately to + * obtain a constant ZERO or ONE source. + */ +#define __CONST(x, y) \ + (PVS_SRC_OPERAND(t_src_index(vp, &vpi->SrcReg[x]), \ + t_swizzle(y), \ + t_swizzle(y), \ + t_swizzle(y), \ + t_swizzle(y), \ + t_src_class(vpi->SrcReg[x].File), \ + RC_MASK_NONE) | (vpi->SrcReg[x].RelAddr << 4)) + + +static unsigned long t_dst_mask(unsigned int mask) +{ + /* RC_MASK_* is equivalent to VSF_FLAG_* */ + return mask & RC_MASK_XYZW; +} + +static unsigned long t_dst_class(rc_register_file file) +{ + switch (file) { + default: + fprintf(stderr, "%s: Bad register file %i\n", __FUNCTION__, file); + /* fall-through */ + case RC_FILE_TEMPORARY: + return PVS_DST_REG_TEMPORARY; + case RC_FILE_OUTPUT: + return PVS_DST_REG_OUT; + case RC_FILE_ADDRESS: + return PVS_DST_REG_A0; + } +} + +static unsigned long t_dst_index(struct r300_vertex_program_code *vp, + struct rc_dst_register *dst) +{ + if (dst->File == RC_FILE_OUTPUT) + return vp->outputs[dst->Index]; + + return dst->Index; +} + +static unsigned long t_src_class(rc_register_file file) +{ + switch (file) { + default: + fprintf(stderr, "%s: Bad register file %i\n", __FUNCTION__, file); + /* fall-through */ + case RC_FILE_NONE: + case RC_FILE_TEMPORARY: + return PVS_SRC_REG_TEMPORARY; + case RC_FILE_INPUT: + return PVS_SRC_REG_INPUT; + case RC_FILE_CONSTANT: + return PVS_SRC_REG_CONSTANT; + } +} + +static int t_src_conflict(struct rc_src_register a, struct rc_src_register b) +{ + unsigned long aclass = t_src_class(a.File); + unsigned long bclass = t_src_class(b.File); + + if (aclass != bclass) + return 0; + if (aclass == PVS_SRC_REG_TEMPORARY) + return 0; + + if (a.RelAddr || b.RelAddr) + return 1; + if (a.Index != b.Index) + return 1; + + return 0; +} + +static inline unsigned long t_swizzle(unsigned int swizzle) +{ + /* this is in fact a NOP as the Mesa RC_SWIZZLE_* are all identical to VSF_IN_COMPONENT_* */ + return swizzle; +} + +static unsigned long t_src_index(struct r300_vertex_program_code *vp, + struct rc_src_register *src) +{ + if (src->File == RC_FILE_INPUT) { + assert(vp->inputs[src->Index] != -1); + return vp->inputs[src->Index]; + } else { + if (src->Index < 0) { + fprintf(stderr, + "negative offsets for indirect addressing do not work.\n"); + return 0; + } + return src->Index; + } +} + +/* these two functions should probably be merged... */ + +static unsigned long t_src(struct r300_vertex_program_code *vp, + struct rc_src_register *src) +{ + /* src->Negate uses the RC_MASK_ flags from program_instruction.h, + * which equal our VSF_FLAGS_ values, so it's safe to just pass it here. + */ + return PVS_SRC_OPERAND(t_src_index(vp, src), + t_swizzle(GET_SWZ(src->Swizzle, 0)), + t_swizzle(GET_SWZ(src->Swizzle, 1)), + t_swizzle(GET_SWZ(src->Swizzle, 2)), + t_swizzle(GET_SWZ(src->Swizzle, 3)), + t_src_class(src->File), + src->Negate) | + (src->RelAddr << 4) | (src->Abs << 3); +} + +static unsigned long t_src_scalar(struct r300_vertex_program_code *vp, + struct rc_src_register *src) +{ + /* src->Negate uses the RC_MASK_ flags from program_instruction.h, + * which equal our VSF_FLAGS_ values, so it's safe to just pass it here. + */ + return PVS_SRC_OPERAND(t_src_index(vp, src), + t_swizzle(GET_SWZ(src->Swizzle, 0)), + t_swizzle(GET_SWZ(src->Swizzle, 0)), + t_swizzle(GET_SWZ(src->Swizzle, 0)), + t_swizzle(GET_SWZ(src->Swizzle, 0)), + t_src_class(src->File), + src->Negate ? RC_MASK_XYZW : RC_MASK_NONE) | + (src->RelAddr << 4) | (src->Abs << 3); +} + +static int valid_dst(struct r300_vertex_program_code *vp, + struct rc_dst_register *dst) +{ + if (dst->File == RC_FILE_OUTPUT && vp->outputs[dst->Index] == -1) { + return 0; + } else if (dst->File == RC_FILE_ADDRESS) { + assert(dst->Index == 0); + } + + return 1; +} + +static void ei_vector1(struct r300_vertex_program_code *vp, + unsigned int hw_opcode, + struct rc_sub_instruction *vpi, + unsigned int * inst) +{ + inst[0] = PVS_OP_DST_OPERAND(hw_opcode, + 0, + 0, + t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + inst[1] = t_src(vp, &vpi->SrcReg[0]); + inst[2] = __CONST(0, RC_SWIZZLE_ZERO); + inst[3] = __CONST(0, RC_SWIZZLE_ZERO); +} + +static void ei_vector2(struct r300_vertex_program_code *vp, + unsigned int hw_opcode, + struct rc_sub_instruction *vpi, + unsigned int * inst) +{ + inst[0] = PVS_OP_DST_OPERAND(hw_opcode, + 0, + 0, + t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + inst[1] = t_src(vp, &vpi->SrcReg[0]); + inst[2] = t_src(vp, &vpi->SrcReg[1]); + inst[3] = __CONST(1, RC_SWIZZLE_ZERO); +} + +static void ei_math1(struct r300_vertex_program_code *vp, + unsigned int hw_opcode, + struct rc_sub_instruction *vpi, + unsigned int * inst) +{ + inst[0] = PVS_OP_DST_OPERAND(hw_opcode, + 1, + 0, + t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + inst[1] = t_src_scalar(vp, &vpi->SrcReg[0]); + inst[2] = __CONST(0, RC_SWIZZLE_ZERO); + inst[3] = __CONST(0, RC_SWIZZLE_ZERO); +} + +static void ei_lit(struct r300_vertex_program_code *vp, + struct rc_sub_instruction *vpi, + unsigned int * inst) +{ + //LIT TMP 1.Y Z TMP 1{} {X W Z Y} TMP 1{} {Y W Z X} TMP 1{} {Y X Z W} + + inst[0] = PVS_OP_DST_OPERAND(ME_LIGHT_COEFF_DX, + 1, + 0, + t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + /* NOTE: Users swizzling might not work. */ + inst[1] = PVS_SRC_OPERAND(t_src_index(vp, &vpi->SrcReg[0]), t_swizzle(GET_SWZ(vpi->SrcReg[0].Swizzle, 0)), // X + t_swizzle(GET_SWZ(vpi->SrcReg[0].Swizzle, 3)), // W + PVS_SRC_SELECT_FORCE_0, // Z + t_swizzle(GET_SWZ(vpi->SrcReg[0].Swizzle, 1)), // Y + t_src_class(vpi->SrcReg[0].File), + vpi->SrcReg[0].Negate ? RC_MASK_XYZW : RC_MASK_NONE) | + (vpi->SrcReg[0].RelAddr << 4); + inst[2] = PVS_SRC_OPERAND(t_src_index(vp, &vpi->SrcReg[0]), t_swizzle(GET_SWZ(vpi->SrcReg[0].Swizzle, 1)), // Y + t_swizzle(GET_SWZ(vpi->SrcReg[0].Swizzle, 3)), // W + PVS_SRC_SELECT_FORCE_0, // Z + t_swizzle(GET_SWZ(vpi->SrcReg[0].Swizzle, 0)), // X + t_src_class(vpi->SrcReg[0].File), + vpi->SrcReg[0].Negate ? RC_MASK_XYZW : RC_MASK_NONE) | + (vpi->SrcReg[0].RelAddr << 4); + inst[3] = PVS_SRC_OPERAND(t_src_index(vp, &vpi->SrcReg[0]), t_swizzle(GET_SWZ(vpi->SrcReg[0].Swizzle, 1)), // Y + t_swizzle(GET_SWZ(vpi->SrcReg[0].Swizzle, 0)), // X + PVS_SRC_SELECT_FORCE_0, // Z + t_swizzle(GET_SWZ(vpi->SrcReg[0].Swizzle, 3)), // W + t_src_class(vpi->SrcReg[0].File), + vpi->SrcReg[0].Negate ? RC_MASK_XYZW : RC_MASK_NONE) | + (vpi->SrcReg[0].RelAddr << 4); +} + +static void ei_mad(struct r300_vertex_program_code *vp, + struct rc_sub_instruction *vpi, + unsigned int * inst) +{ + unsigned int i; + /* Remarks about hardware limitations of MAD + * (please preserve this comment, as this information is _NOT_ + * in the documentation provided by AMD). + * + * As described in the documentation, MAD with three unique temporary + * source registers requires the use of the macro version. + * + * However (and this is not mentioned in the documentation), apparently + * the macro version is _NOT_ a full superset of the normal version. + * In particular, the macro version does not always work when relative + * addressing is used in the source operands. + * + * This limitation caused incorrect rendering in Sauerbraten's OpenGL + * assembly shader path when using medium quality animations + * (i.e. animations with matrix blending instead of quaternion blending). + * + * Unfortunately, I (nha) have been unable to extract a Piglit regression + * test for this issue - for some reason, it is possible to have vertex + * programs whose prefix is *exactly* the same as the prefix of the + * offending program in Sauerbraten up to the offending instruction + * without causing any trouble. + * + * Bottom line: Only use the macro version only when really necessary; + * according to AMD docs, this should improve performance by one clock + * as a nice side bonus. + */ + if (vpi->SrcReg[0].File == RC_FILE_TEMPORARY && + vpi->SrcReg[1].File == RC_FILE_TEMPORARY && + vpi->SrcReg[2].File == RC_FILE_TEMPORARY && + vpi->SrcReg[0].Index != vpi->SrcReg[1].Index && + vpi->SrcReg[0].Index != vpi->SrcReg[2].Index && + vpi->SrcReg[1].Index != vpi->SrcReg[2].Index) { + inst[0] = PVS_OP_DST_OPERAND(PVS_MACRO_OP_2CLK_MADD, + 0, + 1, + t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + } else { + inst[0] = PVS_OP_DST_OPERAND(VE_MULTIPLY_ADD, + 0, + 0, + t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + + /* Arguments with constant swizzles still count as a unique + * temporary, so we should make sure these arguments share a + * register index with one of the other arguments. */ + for (i = 0; i < 3; i++) { + unsigned int j; + if (vpi->SrcReg[i].File != RC_FILE_NONE) + continue; + + for (j = 0; j < 3; j++) { + if (i != j) { + vpi->SrcReg[i].Index = + vpi->SrcReg[j].Index; + break; + } + } + } + } + inst[1] = t_src(vp, &vpi->SrcReg[0]); + inst[2] = t_src(vp, &vpi->SrcReg[1]); + inst[3] = t_src(vp, &vpi->SrcReg[2]); +} + +static void ei_pow(struct r300_vertex_program_code *vp, + struct rc_sub_instruction *vpi, + unsigned int * inst) +{ + inst[0] = PVS_OP_DST_OPERAND(ME_POWER_FUNC_FF, + 1, + 0, + t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + inst[1] = t_src_scalar(vp, &vpi->SrcReg[0]); + inst[2] = __CONST(0, RC_SWIZZLE_ZERO); + inst[3] = t_src_scalar(vp, &vpi->SrcReg[1]); +} + +static void mark_write(void * userdata, struct rc_instruction * inst, + rc_register_file file, unsigned int index, unsigned int mask) +{ + unsigned int * writemasks = userdata; + + if (file != RC_FILE_TEMPORARY) + return; + + if (index >= R300_VS_MAX_TEMPS) + return; + + writemasks[index] |= mask; +} + +static unsigned long t_pred_src(struct r300_vertex_program_compiler * compiler) +{ + return PVS_SRC_OPERAND(compiler->PredicateIndex, + t_swizzle(RC_SWIZZLE_ZERO), + t_swizzle(RC_SWIZZLE_ZERO), + t_swizzle(RC_SWIZZLE_ZERO), + t_swizzle(RC_SWIZZLE_W), + t_src_class(RC_FILE_TEMPORARY), + 0); +} + +static unsigned long t_pred_dst(struct r300_vertex_program_compiler * compiler, + unsigned int hw_opcode, int is_math) +{ + return PVS_OP_DST_OPERAND(hw_opcode, + is_math, + 0, + compiler->PredicateIndex, + RC_MASK_W, + t_dst_class(RC_FILE_TEMPORARY)); + +} + +static void ei_if(struct r300_vertex_program_compiler * compiler, + struct rc_instruction *rci, + unsigned int * inst, + unsigned int branch_depth) +{ + unsigned int predicate_opcode; + int is_math = 0; + + if (!compiler->Base.is_r500) { + rc_error(&compiler->Base,"Opcode IF not supported\n"); + return; + } + + /* Reserve a temporary to use as our predicate stack counter, if we + * don't already have one. */ + if (!compiler->PredicateMask) { + unsigned int writemasks[RC_REGISTER_MAX_INDEX]; + struct rc_instruction * inst; + unsigned int i; + memset(writemasks, 0, sizeof(writemasks)); + for(inst = compiler->Base.Program.Instructions.Next; + inst != &compiler->Base.Program.Instructions; + inst = inst->Next) { + rc_for_all_writes_mask(inst, mark_write, writemasks); + } + for(i = 0; i < compiler->Base.max_temp_regs; i++) { + unsigned int mask = ~writemasks[i] & RC_MASK_XYZW; + /* Only the W component can be used fo the predicate + * stack counter. */ + if (mask & RC_MASK_W) { + compiler->PredicateMask = RC_MASK_W; + compiler->PredicateIndex = i; + break; + } + } + if (i == compiler->Base.max_temp_regs) { + rc_error(&compiler->Base, "No free temporary to use for" + " predicate stack counter.\n"); + return; + } + } + predicate_opcode = + branch_depth ? VE_PRED_SET_NEQ_PUSH : ME_PRED_SET_NEQ; + + rci->U.I.SrcReg[0].Swizzle = RC_MAKE_SWIZZLE_SMEAR(GET_SWZ(rci->U.I.SrcReg[0].Swizzle,0)); + if (branch_depth == 0) { + is_math = 1; + predicate_opcode = ME_PRED_SET_NEQ; + inst[1] = t_src(compiler->code, &rci->U.I.SrcReg[0]); + inst[2] = 0; + } else { + predicate_opcode = VE_PRED_SET_NEQ_PUSH; + inst[1] = t_pred_src(compiler); + inst[2] = t_src(compiler->code, &rci->U.I.SrcReg[0]); + } + + inst[0] = t_pred_dst(compiler, predicate_opcode, is_math); + inst[3] = 0; + +} + +static void ei_else(struct r300_vertex_program_compiler * compiler, + unsigned int * inst) +{ + if (!compiler->Base.is_r500) { + rc_error(&compiler->Base,"Opcode ELSE not supported\n"); + return; + } + inst[0] = t_pred_dst(compiler, ME_PRED_SET_INV, 1); + inst[1] = t_pred_src(compiler); + inst[2] = 0; + inst[3] = 0; +} + +static void ei_endif(struct r300_vertex_program_compiler *compiler, + unsigned int * inst) +{ + if (!compiler->Base.is_r500) { + rc_error(&compiler->Base,"Opcode ENDIF not supported\n"); + return; + } + inst[0] = t_pred_dst(compiler, ME_PRED_SET_POP, 1); + inst[1] = t_pred_src(compiler); + inst[2] = 0; + inst[3] = 0; +} + +static void translate_vertex_program(struct radeon_compiler *c, void *user) +{ + struct r300_vertex_program_compiler *compiler = (struct r300_vertex_program_compiler*)c; + struct rc_instruction *rci; + + struct loop * loops = NULL; + int current_loop_depth = 0; + int loops_reserved = 0; + + unsigned int branch_depth = 0; + + compiler->code->pos_end = 0; /* Not supported yet */ + compiler->code->length = 0; + compiler->code->num_temporaries = 0; + + compiler->SetHwInputOutput(compiler); + + for(rci = compiler->Base.Program.Instructions.Next; rci != &compiler->Base.Program.Instructions; rci = rci->Next) { + struct rc_sub_instruction *vpi = &rci->U.I; + unsigned int *inst = compiler->code->body.d + compiler->code->length; + const struct rc_opcode_info *info = rc_get_opcode_info(vpi->Opcode); + + /* Skip instructions writing to non-existing destination */ + if (!valid_dst(compiler->code, &vpi->DstReg)) + continue; + + if (info->HasDstReg) { + /* Neither is Saturate. */ + if (vpi->SaturateMode != RC_SATURATE_NONE) { + rc_error(&compiler->Base, "Vertex program does not support the Saturate " + "modifier (yet).\n"); + } + } + + if (compiler->code->length >= c->max_alu_insts * 4) { + rc_error(&compiler->Base, "Vertex program has too many instructions\n"); + return; + } + + assert(compiler->Base.is_r500 || + (vpi->Opcode != RC_OPCODE_SEQ && + vpi->Opcode != RC_OPCODE_SNE)); + + switch (vpi->Opcode) { + case RC_OPCODE_ADD: ei_vector2(compiler->code, VE_ADD, vpi, inst); break; + case RC_OPCODE_ARL: ei_vector1(compiler->code, VE_FLT2FIX_DX, vpi, inst); break; + case RC_OPCODE_COS: ei_math1(compiler->code, ME_COS, vpi, inst); break; + case RC_OPCODE_DP4: ei_vector2(compiler->code, VE_DOT_PRODUCT, vpi, inst); break; + case RC_OPCODE_DST: ei_vector2(compiler->code, VE_DISTANCE_VECTOR, vpi, inst); break; + case RC_OPCODE_ELSE: ei_else(compiler, inst); break; + case RC_OPCODE_ENDIF: ei_endif(compiler, inst); branch_depth--; break; + case RC_OPCODE_EX2: ei_math1(compiler->code, ME_EXP_BASE2_FULL_DX, vpi, inst); break; + case RC_OPCODE_EXP: ei_math1(compiler->code, ME_EXP_BASE2_DX, vpi, inst); break; + case RC_OPCODE_FRC: ei_vector1(compiler->code, VE_FRACTION, vpi, inst); break; + case RC_OPCODE_IF: ei_if(compiler, rci, inst, branch_depth); branch_depth++; break; + case RC_OPCODE_LG2: ei_math1(compiler->code, ME_LOG_BASE2_FULL_DX, vpi, inst); break; + case RC_OPCODE_LIT: ei_lit(compiler->code, vpi, inst); break; + case RC_OPCODE_LOG: ei_math1(compiler->code, ME_LOG_BASE2_DX, vpi, inst); break; + case RC_OPCODE_MAD: ei_mad(compiler->code, vpi, inst); break; + case RC_OPCODE_MAX: ei_vector2(compiler->code, VE_MAXIMUM, vpi, inst); break; + case RC_OPCODE_MIN: ei_vector2(compiler->code, VE_MINIMUM, vpi, inst); break; + case RC_OPCODE_MOV: ei_vector1(compiler->code, VE_ADD, vpi, inst); break; + case RC_OPCODE_MUL: ei_vector2(compiler->code, VE_MULTIPLY, vpi, inst); break; + case RC_OPCODE_POW: ei_pow(compiler->code, vpi, inst); break; + case RC_OPCODE_RCP: ei_math1(compiler->code, ME_RECIP_DX, vpi, inst); break; + case RC_OPCODE_RSQ: ei_math1(compiler->code, ME_RECIP_SQRT_DX, vpi, inst); break; + case RC_OPCODE_SEQ: ei_vector2(compiler->code, VE_SET_EQUAL, vpi, inst); break; + case RC_OPCODE_SGE: ei_vector2(compiler->code, VE_SET_GREATER_THAN_EQUAL, vpi, inst); break; + case RC_OPCODE_SIN: ei_math1(compiler->code, ME_SIN, vpi, inst); break; + case RC_OPCODE_SLT: ei_vector2(compiler->code, VE_SET_LESS_THAN, vpi, inst); break; + case RC_OPCODE_SNE: ei_vector2(compiler->code, VE_SET_NOT_EQUAL, vpi, inst); break; + case RC_OPCODE_BGNLOOP: + { + struct loop * l; + + if ((!compiler->Base.is_r500 + && loops_reserved >= R300_VS_MAX_LOOP_DEPTH) + || loops_reserved >= R500_VS_MAX_FC_DEPTH) { + rc_error(&compiler->Base, + "Loops are nested too deep."); + return; + } + memory_pool_array_reserve(&compiler->Base.Pool, + struct loop, loops, current_loop_depth, + loops_reserved, 1); + l = &loops[current_loop_depth++]; + memset(l , 0, sizeof(struct loop)); + l->BgnLoop = (compiler->code->length / 4); + continue; + } + case RC_OPCODE_ENDLOOP: + { + struct loop * l; + unsigned int act_addr; + unsigned int last_addr; + unsigned int ret_addr; + + assert(loops); + l = &loops[current_loop_depth - 1]; + act_addr = l->BgnLoop - 1; + last_addr = (compiler->code->length / 4) - 1; + ret_addr = l->BgnLoop; + + if (loops_reserved >= R300_VS_MAX_FC_OPS) { + rc_error(&compiler->Base, + "Too many flow control instructions."); + return; + } + if (compiler->Base.is_r500) { + compiler->code->fc_op_addrs.r500 + [compiler->code->num_fc_ops].lw = + R500_PVS_FC_ACT_ADRS(act_addr) + | R500_PVS_FC_LOOP_CNT_JMP_INST(0xffff) + ; + compiler->code->fc_op_addrs.r500 + [compiler->code->num_fc_ops].uw = + R500_PVS_FC_LAST_INST(last_addr) + | R500_PVS_FC_RTN_INST(ret_addr) + ; + } else { + compiler->code->fc_op_addrs.r300 + [compiler->code->num_fc_ops] = + R300_PVS_FC_ACT_ADRS(act_addr) + | R300_PVS_FC_LOOP_CNT_JMP_INST(0xff) + | R300_PVS_FC_LAST_INST(last_addr) + | R300_PVS_FC_RTN_INST(ret_addr) + ; + } + compiler->code->fc_loop_index[compiler->code->num_fc_ops] = + R300_PVS_FC_LOOP_INIT_VAL(0x0) + | R300_PVS_FC_LOOP_STEP_VAL(0x1) + ; + compiler->code->fc_ops |= R300_VAP_PVS_FC_OPC_LOOP( + compiler->code->num_fc_ops); + compiler->code->num_fc_ops++; + current_loop_depth--; + continue; + } + + default: + rc_error(&compiler->Base, "Unknown opcode %s\n", info->Name); + return; + } + + /* Non-flow control instructions that are inside an if statement + * need to pay attention to the predicate bit. */ + if (branch_depth + && vpi->Opcode != RC_OPCODE_IF + && vpi->Opcode != RC_OPCODE_ELSE + && vpi->Opcode != RC_OPCODE_ENDIF) { + + inst[0] |= (PVS_DST_PRED_ENABLE_MASK + << PVS_DST_PRED_ENABLE_SHIFT); + inst[0] |= (PVS_DST_PRED_SENSE_MASK + << PVS_DST_PRED_SENSE_SHIFT); + } + + /* Update the number of temporaries. */ + if (info->HasDstReg && vpi->DstReg.File == RC_FILE_TEMPORARY && + vpi->DstReg.Index >= compiler->code->num_temporaries) + compiler->code->num_temporaries = vpi->DstReg.Index + 1; + + for (unsigned i = 0; i < info->NumSrcRegs; i++) + if (vpi->SrcReg[i].File == RC_FILE_TEMPORARY && + vpi->SrcReg[i].Index >= compiler->code->num_temporaries) + compiler->code->num_temporaries = vpi->SrcReg[i].Index + 1; + + if (compiler->PredicateMask) + if (compiler->PredicateIndex >= compiler->code->num_temporaries) + compiler->code->num_temporaries = compiler->PredicateIndex + 1; + + if (compiler->code->num_temporaries > compiler->Base.max_temp_regs) { + rc_error(&compiler->Base, "Too many temporaries.\n"); + return; + } + + compiler->code->length += 4; + + if (compiler->Base.Error) + return; + } +} + +struct temporary_allocation { + unsigned int Allocated:1; + unsigned int HwTemp:15; + struct rc_instruction * LastRead; +}; + +static void allocate_temporary_registers(struct radeon_compiler *c, void *user) +{ + struct r300_vertex_program_compiler *compiler = (struct r300_vertex_program_compiler*)c; + struct rc_instruction *inst; + struct rc_instruction *end_loop = NULL; + unsigned int num_orig_temps = 0; + char hwtemps[RC_REGISTER_MAX_INDEX]; + struct temporary_allocation * ta; + unsigned int i, j; + + memset(hwtemps, 0, sizeof(hwtemps)); + + rc_recompute_ips(c); + + /* Pass 1: Count original temporaries. */ + for(inst = compiler->Base.Program.Instructions.Next; inst != &compiler->Base.Program.Instructions; inst = inst->Next) { + const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode); + + for (i = 0; i < opcode->NumSrcRegs; ++i) { + if (inst->U.I.SrcReg[i].File == RC_FILE_TEMPORARY) { + if (inst->U.I.SrcReg[i].Index >= num_orig_temps) + num_orig_temps = inst->U.I.SrcReg[i].Index + 1; + } + } + + if (opcode->HasDstReg) { + if (inst->U.I.DstReg.File == RC_FILE_TEMPORARY) { + if (inst->U.I.DstReg.Index >= num_orig_temps) + num_orig_temps = inst->U.I.DstReg.Index + 1; + } + } + } + + ta = (struct temporary_allocation*)memory_pool_malloc(&compiler->Base.Pool, + sizeof(struct temporary_allocation) * num_orig_temps); + memset(ta, 0, sizeof(struct temporary_allocation) * num_orig_temps); + + /* Pass 2: Determine original temporary lifetimes */ + for(inst = compiler->Base.Program.Instructions.Next; inst != &compiler->Base.Program.Instructions; inst = inst->Next) { + const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode); + /* Instructions inside of loops need to use the ENDLOOP + * instruction as their LastRead. */ + if (!end_loop && inst->U.I.Opcode == RC_OPCODE_BGNLOOP) { + int endloops = 1; + struct rc_instruction * ptr; + for(ptr = inst->Next; + ptr != &compiler->Base.Program.Instructions; + ptr = ptr->Next){ + if (ptr->U.I.Opcode == RC_OPCODE_BGNLOOP) { + endloops++; + } else if (ptr->U.I.Opcode == RC_OPCODE_ENDLOOP) { + endloops--; + if (endloops <= 0) { + end_loop = ptr; + break; + } + } + } + } + + if (inst == end_loop) { + end_loop = NULL; + continue; + } + + for (i = 0; i < opcode->NumSrcRegs; ++i) { + if (inst->U.I.SrcReg[i].File == RC_FILE_TEMPORARY) { + ta[inst->U.I.SrcReg[i].Index].LastRead = end_loop ? end_loop : inst; + } + } + } + + /* Pass 3: Register allocation */ + for(inst = compiler->Base.Program.Instructions.Next; inst != &compiler->Base.Program.Instructions; inst = inst->Next) { + const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode); + + for (i = 0; i < opcode->NumSrcRegs; ++i) { + if (inst->U.I.SrcReg[i].File == RC_FILE_TEMPORARY) { + unsigned int orig = inst->U.I.SrcReg[i].Index; + inst->U.I.SrcReg[i].Index = ta[orig].HwTemp; + + if (ta[orig].Allocated && inst == ta[orig].LastRead) + hwtemps[ta[orig].HwTemp] = 0; + } + } + + if (opcode->HasDstReg) { + if (inst->U.I.DstReg.File == RC_FILE_TEMPORARY) { + unsigned int orig = inst->U.I.DstReg.Index; + + if (!ta[orig].Allocated) { + for(j = 0; j < c->max_temp_regs; ++j) { + if (!hwtemps[j]) + break; + } + ta[orig].Allocated = 1; + ta[orig].HwTemp = j; + hwtemps[ta[orig].HwTemp] = 1; + } + + inst->U.I.DstReg.Index = ta[orig].HwTemp; + } + } + } +} + +/** + * R3xx-R4xx vertex engine does not support the Absolute source operand modifier + * and the Saturate opcode modifier. Only Absolute is currently transformed. + */ +static int transform_nonnative_modifiers( + struct radeon_compiler *c, + struct rc_instruction *inst, + void* unused) +{ + const struct rc_opcode_info *opcode = rc_get_opcode_info(inst->U.I.Opcode); + unsigned i; + + /* Transform ABS(a) to MAX(a, -a). */ + for (i = 0; i < opcode->NumSrcRegs; i++) { + if (inst->U.I.SrcReg[i].Abs) { + struct rc_instruction *new_inst; + unsigned temp; + + inst->U.I.SrcReg[i].Abs = 0; + + temp = rc_find_free_temporary(c); + + new_inst = rc_insert_new_instruction(c, inst->Prev); + new_inst->U.I.Opcode = RC_OPCODE_MAX; + new_inst->U.I.DstReg.File = RC_FILE_TEMPORARY; + new_inst->U.I.DstReg.Index = temp; + new_inst->U.I.SrcReg[0] = inst->U.I.SrcReg[i]; + new_inst->U.I.SrcReg[1] = inst->U.I.SrcReg[i]; + new_inst->U.I.SrcReg[1].Negate ^= RC_MASK_XYZW; + + memset(&inst->U.I.SrcReg[i], 0, sizeof(inst->U.I.SrcReg[i])); + inst->U.I.SrcReg[i].File = RC_FILE_TEMPORARY; + inst->U.I.SrcReg[i].Index = temp; + inst->U.I.SrcReg[i].Swizzle = RC_SWIZZLE_XYZW; + } + } + return 1; +} + +/** + * Vertex engine cannot read two inputs or two constants at the same time. + * Introduce intermediate MOVs to temporary registers to account for this. + */ +static int transform_source_conflicts( + struct radeon_compiler *c, + struct rc_instruction* inst, + void* unused) +{ + const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode); + + if (opcode->NumSrcRegs == 3) { + if (t_src_conflict(inst->U.I.SrcReg[1], inst->U.I.SrcReg[2]) + || t_src_conflict(inst->U.I.SrcReg[0], inst->U.I.SrcReg[2])) { + int tmpreg = rc_find_free_temporary(c); + struct rc_instruction * inst_mov = rc_insert_new_instruction(c, inst->Prev); + inst_mov->U.I.Opcode = RC_OPCODE_MOV; + inst_mov->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst_mov->U.I.DstReg.Index = tmpreg; + inst_mov->U.I.SrcReg[0] = inst->U.I.SrcReg[2]; + + reset_srcreg(&inst->U.I.SrcReg[2]); + inst->U.I.SrcReg[2].File = RC_FILE_TEMPORARY; + inst->U.I.SrcReg[2].Index = tmpreg; + } + } + + if (opcode->NumSrcRegs >= 2) { + if (t_src_conflict(inst->U.I.SrcReg[1], inst->U.I.SrcReg[0])) { + int tmpreg = rc_find_free_temporary(c); + struct rc_instruction * inst_mov = rc_insert_new_instruction(c, inst->Prev); + inst_mov->U.I.Opcode = RC_OPCODE_MOV; + inst_mov->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst_mov->U.I.DstReg.Index = tmpreg; + inst_mov->U.I.SrcReg[0] = inst->U.I.SrcReg[1]; + + reset_srcreg(&inst->U.I.SrcReg[1]); + inst->U.I.SrcReg[1].File = RC_FILE_TEMPORARY; + inst->U.I.SrcReg[1].Index = tmpreg; + } + } + + return 1; +} + +static void rc_vs_add_artificial_outputs(struct radeon_compiler *c, void *user) +{ + struct r300_vertex_program_compiler * compiler = (struct r300_vertex_program_compiler*)c; + int i; + + for(i = 0; i < 32; ++i) { + if ((compiler->RequiredOutputs & (1 << i)) && + !(compiler->Base.Program.OutputsWritten & (1 << i))) { + struct rc_instruction * inst = rc_insert_new_instruction(&compiler->Base, compiler->Base.Program.Instructions.Prev); + inst->U.I.Opcode = RC_OPCODE_MOV; + + inst->U.I.DstReg.File = RC_FILE_OUTPUT; + inst->U.I.DstReg.Index = i; + inst->U.I.DstReg.WriteMask = RC_MASK_XYZW; + + inst->U.I.SrcReg[0].File = RC_FILE_CONSTANT; + inst->U.I.SrcReg[0].Index = 0; + inst->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_XYZW; + + compiler->Base.Program.OutputsWritten |= 1 << i; + } + } +} + +static void dataflow_outputs_mark_used(void * userdata, void * data, + void (*callback)(void *, unsigned int, unsigned int)) +{ + struct r300_vertex_program_compiler * c = userdata; + int i; + + for(i = 0; i < 32; ++i) { + if (c->RequiredOutputs & (1 << i)) + callback(data, i, RC_MASK_XYZW); + } +} + +static int swizzle_is_native(rc_opcode opcode, struct rc_src_register reg) +{ + (void) opcode; + (void) reg; + + return 1; +} + +static void transform_negative_addressing(struct r300_vertex_program_compiler *c, + struct rc_instruction *arl, + struct rc_instruction *end, + int min_offset) +{ + struct rc_instruction *inst, *add; + unsigned const_swizzle; + + /* Transform ARL */ + add = rc_insert_new_instruction(&c->Base, arl->Prev); + add->U.I.Opcode = RC_OPCODE_ADD; + add->U.I.DstReg.File = RC_FILE_TEMPORARY; + add->U.I.DstReg.Index = rc_find_free_temporary(&c->Base); + add->U.I.DstReg.WriteMask = RC_MASK_X; + add->U.I.SrcReg[0] = arl->U.I.SrcReg[0]; + add->U.I.SrcReg[1].File = RC_FILE_CONSTANT; + add->U.I.SrcReg[1].Index = rc_constants_add_immediate_scalar(&c->Base.Program.Constants, + min_offset, &const_swizzle); + add->U.I.SrcReg[1].Swizzle = const_swizzle; + + arl->U.I.SrcReg[0].File = RC_FILE_TEMPORARY; + arl->U.I.SrcReg[0].Index = add->U.I.DstReg.Index; + arl->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_XXXX; + + /* Rewrite offsets up to and excluding inst. */ + for (inst = arl->Next; inst != end; inst = inst->Next) { + const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode); + + for (unsigned i = 0; i < opcode->NumSrcRegs; i++) + if (inst->U.I.SrcReg[i].RelAddr) + inst->U.I.SrcReg[i].Index -= min_offset; + } +} + +static void rc_emulate_negative_addressing(struct radeon_compiler *compiler, void *user) +{ + struct r300_vertex_program_compiler * c = (struct r300_vertex_program_compiler*)compiler; + struct rc_instruction *inst, *lastARL = NULL; + int min_offset = 0; + + for (inst = c->Base.Program.Instructions.Next; inst != &c->Base.Program.Instructions; inst = inst->Next) { + const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode); + + if (inst->U.I.Opcode == RC_OPCODE_ARL) { + if (lastARL != NULL && min_offset < 0) + transform_negative_addressing(c, lastARL, inst, min_offset); + + lastARL = inst; + min_offset = 0; + continue; + } + + for (unsigned i = 0; i < opcode->NumSrcRegs; i++) { + if (inst->U.I.SrcReg[i].RelAddr && + inst->U.I.SrcReg[i].Index < 0) { + /* ARL must precede any indirect addressing. */ + if (lastARL == NULL) { + rc_error(&c->Base, "Vertex shader: Found relative addressing without ARL."); + return; + } + + if (inst->U.I.SrcReg[i].Index < min_offset) + min_offset = inst->U.I.SrcReg[i].Index; + } + } + } + + if (lastARL != NULL && min_offset < 0) + transform_negative_addressing(c, lastARL, inst, min_offset); +} + +static struct rc_swizzle_caps r300_vertprog_swizzle_caps = { + .IsNative = &swizzle_is_native, + .Split = 0 /* should never be called */ +}; + +void r3xx_compile_vertex_program(struct r300_vertex_program_compiler *c) +{ + int is_r500 = c->Base.is_r500; + int opt = !c->Base.disable_optimizations; + + /* Lists of instruction transformations. */ + struct radeon_program_transformation alu_rewrite_r500[] = { + { &r300_transform_vertex_alu, 0 }, + { &r300_transform_trig_scale_vertex, 0 }, + { 0, 0 } + }; + + struct radeon_program_transformation alu_rewrite_r300[] = { + { &r300_transform_vertex_alu, 0 }, + { &r300_transform_trig_simple, 0 }, + { 0, 0 } + }; + + /* Note: These passes have to be done seperately from ALU rewrite, + * otherwise non-native ALU instructions with source conflits + * or non-native modifiers will not be treated properly. + */ + struct radeon_program_transformation emulate_modifiers[] = { + { &transform_nonnative_modifiers, 0 }, + { 0, 0 } + }; + + struct radeon_program_transformation resolve_src_conflicts[] = { + { &transform_source_conflicts, 0 }, + { 0, 0 } + }; + + /* List of compiler passes. */ + struct radeon_compiler_pass vs_list[] = { + /* NAME DUMP PREDICATE FUNCTION PARAM */ + {"add artificial outputs", 0, 1, rc_vs_add_artificial_outputs, NULL}, + {"transform loops", 1, 1, rc_transform_loops, NULL}, + {"emulate branches", 1, !is_r500, rc_emulate_branches, NULL}, + {"emulate negative addressing", 1, 1, rc_emulate_negative_addressing, NULL}, + {"native rewrite", 1, is_r500, rc_local_transform, alu_rewrite_r500}, + {"native rewrite", 1, !is_r500, rc_local_transform, alu_rewrite_r300}, + {"emulate modifiers", 1, !is_r500, rc_local_transform, emulate_modifiers}, + {"deadcode", 1, opt, rc_dataflow_deadcode, dataflow_outputs_mark_used}, + {"dataflow optimize", 1, opt, rc_optimize, NULL}, + /* This pass must be done after optimizations. */ + {"source conflict resolve", 1, 1, rc_local_transform, resolve_src_conflicts}, + {"register allocation", 1, opt, allocate_temporary_registers, NULL}, + {"dead constants", 1, 1, rc_remove_unused_constants, &c->code->constants_remap_table}, + {"final code validation", 0, 1, rc_validate_final_shader, NULL}, + {"machine code generation", 0, 1, translate_vertex_program, NULL}, + {"dump machine code", 0, c->Base.Debug & RC_DBG_LOG, r300_vertex_program_dump, NULL}, + {NULL, 0, 0, NULL, NULL} + }; + + c->Base.type = RC_VERTEX_PROGRAM; + c->Base.SwizzleCaps = &r300_vertprog_swizzle_caps; + + rc_run_compiler(&c->Base, vs_list); + + c->code->InputsRead = c->Base.Program.InputsRead; + c->code->OutputsWritten = c->Base.Program.OutputsWritten; + rc_constants_copy(&c->code->constants, &c->Base.Program.Constants); +} diff --git a/src/gallium/drivers/r300/compiler/r3xx_vertprog_dump.c b/src/gallium/drivers/r300/compiler/r3xx_vertprog_dump.c new file mode 100644 index 00000000000..2bc0a87eed8 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/r3xx_vertprog_dump.c @@ -0,0 +1,207 @@ +/* + * Copyright 2009 Nicolai Hähnle <[email protected]> + * + * 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 + * on the rights to use, copy, modify, merge, publish, distribute, sub + * license, 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 NON-INFRINGEMENT. IN NO EVENT SHALL + * THE AUTHOR(S) AND/OR THEIR 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. */ + +#include "radeon_compiler.h" +#include "radeon_code.h" +#include "../r300_reg.h" + +#include <stdio.h> + +static char* r300_vs_ve_ops[] = { + /* R300 vector ops */ + " VE_NO_OP", + " VE_DOT_PRODUCT", + " VE_MULTIPLY", + " VE_ADD", + " VE_MULTIPLY_ADD", + " VE_DISTANCE_FACTOR", + " VE_FRACTION", + " VE_MAXIMUM", + " VE_MINIMUM", + "VE_SET_GREATER_THAN_EQUAL", + " VE_SET_LESS_THAN", + " VE_MULTIPLYX2_ADD", + " VE_MULTIPLY_CLAMP", + " VE_FLT2FIX_DX", + " VE_FLT2FIX_DX_RND", + /* R500 vector ops */ + " VE_PRED_SET_EQ_PUSH", + " VE_PRED_SET_GT_PUSH", + " VE_PRED_SET_GTE_PUSH", + " VE_PRED_SET_NEQ_PUSH", + " VE_COND_WRITE_EQ", + " VE_COND_WRITE_GT", + " VE_COND_WRITE_GTE", + " VE_COND_WRITE_NEQ", + " VE_COND_MUX_EQ", + " VE_COND_MUX_GT", + " VE_COND_MUX_GTE", + " VE_SET_GREATER_THAN", + " VE_SET_EQUAL", + " VE_SET_NOT_EQUAL", + " (reserved)", + " (reserved)", + " (reserved)", +}; + +static char* r300_vs_me_ops[] = { + /* R300 math ops */ + " ME_NO_OP", + " ME_EXP_BASE2_DX", + " ME_LOG_BASE2_DX", + " ME_EXP_BASEE_FF", + " ME_LIGHT_COEFF_DX", + " ME_POWER_FUNC_FF", + " ME_RECIP_DX", + " ME_RECIP_FF", + " ME_RECIP_SQRT_DX", + " ME_RECIP_SQRT_FF", + " ME_MULTIPLY", + " ME_EXP_BASE2_FULL_DX", + " ME_LOG_BASE2_FULL_DX", + " ME_POWER_FUNC_FF_CLAMP_B", + "ME_POWER_FUNC_FF_CLAMP_B1", + "ME_POWER_FUNC_FF_CLAMP_01", + " ME_SIN", + " ME_COS", + /* R500 math ops */ + " ME_LOG_BASE2_IEEE", + " ME_RECIP_IEEE", + " ME_RECIP_SQRT_IEEE", + " ME_PRED_SET_EQ", + " ME_PRED_SET_GT", + " ME_PRED_SET_GTE", + " ME_PRED_SET_NEQ", + " ME_PRED_SET_CLR", + " ME_PRED_SET_INV", + " ME_PRED_SET_POP", + " ME_PRED_SET_RESTORE", + " (reserved)", + " (reserved)", + " (reserved)", +}; + +/* XXX refactor to avoid clashing symbols */ +static char* r300_vs_src_debug[] = { + "t", + "i", + "c", + "a", +}; + +static char* r300_vs_dst_debug[] = { + "t", + "a0", + "o", + "ox", + "a", + "i", + "u", + "u", +}; + +static char* r300_vs_swiz_debug[] = { + "X", + "Y", + "Z", + "W", + "0", + "1", + "U", + "U", +}; + + +static void r300_vs_op_dump(uint32_t op) +{ + fprintf(stderr, " dst: %d%s op: ", + (op >> 13) & 0x7f, r300_vs_dst_debug[(op >> 8) & 0x7]); + if ((op >> PVS_DST_PRED_ENABLE_SHIFT) & 0x1) { + fprintf(stderr, "PRED %u", + (op >> PVS_DST_PRED_SENSE_SHIFT) & 0x1); + } + if (op & 0x80) { + if (op & 0x1) { + fprintf(stderr, "PVS_MACRO_OP_2CLK_M2X_ADD\n"); + } else { + fprintf(stderr, " PVS_MACRO_OP_2CLK_MADD\n"); + } + } else if (op & 0x40) { + fprintf(stderr, "%s\n", r300_vs_me_ops[op & 0x1f]); + } else { + fprintf(stderr, "%s\n", r300_vs_ve_ops[op & 0x1f]); + } +} + +static void r300_vs_src_dump(uint32_t src) +{ + fprintf(stderr, " reg: %d%s swiz: %s%s/%s%s/%s%s/%s%s\n", + (src >> 5) & 0xff, r300_vs_src_debug[src & 0x3], + src & (1 << 25) ? "-" : " ", + r300_vs_swiz_debug[(src >> 13) & 0x7], + src & (1 << 26) ? "-" : " ", + r300_vs_swiz_debug[(src >> 16) & 0x7], + src & (1 << 27) ? "-" : " ", + r300_vs_swiz_debug[(src >> 19) & 0x7], + src & (1 << 28) ? "-" : " ", + r300_vs_swiz_debug[(src >> 22) & 0x7]); +} + +void r300_vertex_program_dump(struct radeon_compiler *compiler, void *user) +{ + struct r300_vertex_program_compiler *c = (struct r300_vertex_program_compiler*)compiler; + struct r300_vertex_program_code * vs = c->code; + unsigned instrcount = vs->length / 4; + unsigned i; + + fprintf(stderr, "Final vertex program code:\n"); + + for(i = 0; i < instrcount; i++) { + unsigned offset = i*4; + unsigned src; + + fprintf(stderr, "%d: op: 0x%08x", i, vs->body.d[offset]); + r300_vs_op_dump(vs->body.d[offset]); + + for(src = 0; src < 3; ++src) { + fprintf(stderr, " src%i: 0x%08x", src, vs->body.d[offset+1+src]); + r300_vs_src_dump(vs->body.d[offset+1+src]); + } + } + + fprintf(stderr, "Flow Control Ops: 0x%08x\n",vs->fc_ops); + for(i = 0; i < vs->num_fc_ops; i++) { + switch((vs->fc_ops >> (i * 2)) & 0x3 ) { + case 0: fprintf(stderr, "NOP"); break; + case 1: fprintf(stderr, "JUMP"); break; + case 2: fprintf(stderr, "LOOP"); break; + case 3: fprintf(stderr, "JSR"); break; + } + if (c->Base.is_r500) { + fprintf(stderr,": uw-> 0x%08x lw-> 0x%08x\n", + vs->fc_op_addrs.r500[i].uw, + vs->fc_op_addrs.r500[i].lw); + } else { + fprintf(stderr,": 0x%08x\n", vs->fc_op_addrs.r300[i]); + } + } +} diff --git a/src/gallium/drivers/r300/compiler/r500_fragprog.c b/src/gallium/drivers/r300/compiler/r500_fragprog.c new file mode 100644 index 00000000000..cf99f5e4538 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/r500_fragprog.c @@ -0,0 +1,539 @@ +/* + * Copyright 2008 Corbin Simpson <[email protected]> + * + * 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. + * + */ + +#include "r500_fragprog.h" + +#include <stdio.h> + +#include "radeon_compiler_util.h" +#include "radeon_list.h" +#include "radeon_variable.h" +#include "../r300_reg.h" + +/** + * Rewrite IF instructions to use the ALU result special register. + */ +int r500_transform_IF( + struct radeon_compiler * c, + struct rc_instruction * inst_if, + void *data) +{ + struct rc_variable * writer; + struct rc_list * writer_list, * list_ptr; + struct rc_list * var_list = rc_get_variables(c); + unsigned int generic_if = 0; + unsigned int alu_chan; + + if (inst_if->U.I.Opcode != RC_OPCODE_IF) { + return 0; + } + + writer_list = rc_variable_list_get_writers( + var_list, inst_if->Type, &inst_if->U.I.SrcReg[0]); + if (!writer_list) { + generic_if = 1; + } else { + + /* Make sure it is safe for the writers to write to + * ALU Result */ + for (list_ptr = writer_list; list_ptr; + list_ptr = list_ptr->Next) { + struct rc_instruction * inst; + writer = list_ptr->Item; + /* We are going to modify the destination register + * of writer, so if it has a reader other than + * inst_if (aka ReaderCount > 1) we must fall back to + * our generic IF. + * If the writer has a lower IP than inst_if, this + * means that inst_if is above the writer in a loop. + * I'm not sure why this would ever happen, but + * if it does we want to make sure we fall back + * to our generic IF. */ + if (writer->ReaderCount > 1 || writer->Inst->IP < inst_if->IP) { + generic_if = 1; + break; + } + + /* The ALU Result is not preserved across IF + * instructions, so if there is another IF + * instruction between writer and inst_if, then + * we need to fall back to generic IF. */ + for (inst = writer->Inst; inst != inst_if; inst = inst->Next) { + const struct rc_opcode_info * info = + rc_get_opcode_info(inst->U.I.Opcode); + if (info->IsFlowControl) { + generic_if = 1; + break; + } + } + if (generic_if) { + break; + } + } + } + + if (GET_SWZ(inst_if->U.I.SrcReg[0].Swizzle, 0) == RC_SWIZZLE_X) { + alu_chan = RC_ALURESULT_X; + } else { + alu_chan = RC_ALURESULT_W; + } + if (generic_if) { + struct rc_instruction * inst_mov = + rc_insert_new_instruction(c, inst_if->Prev); + + inst_mov->U.I.Opcode = RC_OPCODE_MOV; + inst_mov->U.I.DstReg.WriteMask = 0; + inst_mov->U.I.DstReg.File = RC_FILE_NONE; + inst_mov->U.I.ALUResultCompare = RC_COMPARE_FUNC_NOTEQUAL; + inst_mov->U.I.WriteALUResult = alu_chan; + inst_mov->U.I.SrcReg[0] = inst_if->U.I.SrcReg[0]; + if (alu_chan == RC_ALURESULT_X) { + inst_mov->U.I.SrcReg[0].Swizzle = combine_swizzles4( + inst_mov->U.I.SrcReg[0].Swizzle, + RC_SWIZZLE_X, RC_SWIZZLE_UNUSED, + RC_SWIZZLE_UNUSED, RC_SWIZZLE_UNUSED); + } else { + inst_mov->U.I.SrcReg[0].Swizzle = combine_swizzles4( + inst_mov->U.I.SrcReg[0].Swizzle, + RC_SWIZZLE_UNUSED, RC_SWIZZLE_UNUSED, + RC_SWIZZLE_UNUSED, RC_SWIZZLE_Z); + } + } else { + rc_compare_func compare_func = RC_COMPARE_FUNC_NEVER; + unsigned int reverse_srcs = 0; + unsigned int preserve_opcode = 0; + for (list_ptr = writer_list; list_ptr; + list_ptr = list_ptr->Next) { + writer = list_ptr->Item; + switch(writer->Inst->U.I.Opcode) { + case RC_OPCODE_SEQ: + compare_func = RC_COMPARE_FUNC_EQUAL; + break; + case RC_OPCODE_SNE: + compare_func = RC_COMPARE_FUNC_NOTEQUAL; + break; + case RC_OPCODE_SLE: + reverse_srcs = 1; + /* Fall through */ + case RC_OPCODE_SGE: + compare_func = RC_COMPARE_FUNC_GEQUAL; + break; + case RC_OPCODE_SGT: + reverse_srcs = 1; + /* Fall through */ + case RC_OPCODE_SLT: + compare_func = RC_COMPARE_FUNC_LESS; + break; + default: + compare_func = RC_COMPARE_FUNC_NOTEQUAL; + preserve_opcode = 1; + break; + } + if (!preserve_opcode) { + writer->Inst->U.I.Opcode = RC_OPCODE_SUB; + } + writer->Inst->U.I.DstReg.WriteMask = 0; + writer->Inst->U.I.DstReg.File = RC_FILE_NONE; + writer->Inst->U.I.WriteALUResult = alu_chan; + writer->Inst->U.I.ALUResultCompare = compare_func; + if (reverse_srcs) { + struct rc_src_register temp_src; + temp_src = writer->Inst->U.I.SrcReg[0]; + writer->Inst->U.I.SrcReg[0] = + writer->Inst->U.I.SrcReg[1]; + writer->Inst->U.I.SrcReg[1] = temp_src; + } + } + } + + inst_if->U.I.SrcReg[0].File = RC_FILE_SPECIAL; + inst_if->U.I.SrcReg[0].Index = RC_SPECIAL_ALU_RESULT; + inst_if->U.I.SrcReg[0].Swizzle = RC_MAKE_SWIZZLE( + RC_SWIZZLE_X, RC_SWIZZLE_UNUSED, + RC_SWIZZLE_UNUSED, RC_SWIZZLE_UNUSED); + inst_if->U.I.SrcReg[0].Negate = 0; + + return 1; +} + +static int r500_swizzle_is_native(rc_opcode opcode, struct rc_src_register reg) +{ + unsigned int relevant; + int i; + + if (opcode == RC_OPCODE_TEX || + opcode == RC_OPCODE_TXB || + opcode == RC_OPCODE_TXP || + opcode == RC_OPCODE_TXD || + opcode == RC_OPCODE_TXL || + opcode == RC_OPCODE_KIL) { + if (reg.Abs) + return 0; + + if (opcode == RC_OPCODE_KIL && (reg.Swizzle != RC_SWIZZLE_XYZW || reg.Negate != RC_MASK_NONE)) + return 0; + + for(i = 0; i < 4; ++i) { + unsigned int swz = GET_SWZ(reg.Swizzle, i); + if (swz == RC_SWIZZLE_UNUSED) { + reg.Negate &= ~(1 << i); + continue; + } + if (swz >= 4) + return 0; + } + + if (reg.Negate) + return 0; + + return 1; + } else if (opcode == RC_OPCODE_DDX || opcode == RC_OPCODE_DDY) { + /* DDX/MDH and DDY/MDV explicitly ignore incoming swizzles; + * if it doesn't fit perfectly into a .xyzw case... */ + if (reg.Swizzle == RC_SWIZZLE_XYZW && !reg.Abs && !reg.Negate) + return 1; + + return 0; + } else { + /* ALU instructions support almost everything */ + relevant = 0; + for(i = 0; i < 3; ++i) { + unsigned int swz = GET_SWZ(reg.Swizzle, i); + if (swz != RC_SWIZZLE_UNUSED && swz != RC_SWIZZLE_ZERO) + relevant |= 1 << i; + } + if ((reg.Negate & relevant) && ((reg.Negate & relevant) != relevant)) + return 0; + + return 1; + } +} + +/** + * Split source register access. + * + * The only thing we *cannot* do in an ALU instruction is per-component + * negation. + */ +static void r500_swizzle_split(struct rc_src_register src, unsigned int usemask, + struct rc_swizzle_split * split) +{ + unsigned int negatebase[2] = { 0, 0 }; + int i; + + for(i = 0; i < 4; ++i) { + unsigned int swz = GET_SWZ(src.Swizzle, i); + if (swz == RC_SWIZZLE_UNUSED || !GET_BIT(usemask, i)) + continue; + negatebase[GET_BIT(src.Negate, i)] |= 1 << i; + } + + split->NumPhases = 0; + + for(i = 0; i <= 1; ++i) { + if (!negatebase[i]) + continue; + + split->Phase[split->NumPhases++] = negatebase[i]; + } +} + +struct rc_swizzle_caps r500_swizzle_caps = { + .IsNative = r500_swizzle_is_native, + .Split = r500_swizzle_split +}; + +static char *toswiz(int swiz_val) { + switch(swiz_val) { + case 0: return "R"; + case 1: return "G"; + case 2: return "B"; + case 3: return "A"; + case 4: return "0"; + case 5: return "H"; + case 6: return "1"; + case 7: return "U"; + } + return NULL; +} + +static char *toop(int op_val) +{ + char *str = NULL; + switch (op_val) { + case 0: str = "MAD"; break; + case 1: str = "DP3"; break; + case 2: str = "DP4"; break; + case 3: str = "D2A"; break; + case 4: str = "MIN"; break; + case 5: str = "MAX"; break; + case 6: str = "Reserved"; break; + case 7: str = "CND"; break; + case 8: str = "CMP"; break; + case 9: str = "FRC"; break; + case 10: str = "SOP"; break; + case 11: str = "MDH"; break; + case 12: str = "MDV"; break; + } + return str; +} + +static char *to_alpha_op(int op_val) +{ + char *str = NULL; + switch (op_val) { + case 0: str = "MAD"; break; + case 1: str = "DP"; break; + case 2: str = "MIN"; break; + case 3: str = "MAX"; break; + case 4: str = "Reserved"; break; + case 5: str = "CND"; break; + case 6: str = "CMP"; break; + case 7: str = "FRC"; break; + case 8: str = "EX2"; break; + case 9: str = "LN2"; break; + case 10: str = "RCP"; break; + case 11: str = "RSQ"; break; + case 12: str = "SIN"; break; + case 13: str = "COS"; break; + case 14: str = "MDH"; break; + case 15: str = "MDV"; break; + } + return str; +} + +static char *to_mask(int val) +{ + char *str = NULL; + switch(val) { + case 0: str = "NONE"; break; + case 1: str = "R"; break; + case 2: str = "G"; break; + case 3: str = "RG"; break; + case 4: str = "B"; break; + case 5: str = "RB"; break; + case 6: str = "GB"; break; + case 7: str = "RGB"; break; + case 8: str = "A"; break; + case 9: str = "AR"; break; + case 10: str = "AG"; break; + case 11: str = "ARG"; break; + case 12: str = "AB"; break; + case 13: str = "ARB"; break; + case 14: str = "AGB"; break; + case 15: str = "ARGB"; break; + } + return str; +} + +static char *to_texop(int val) +{ + switch(val) { + case 0: return "NOP"; + case 1: return "LD"; + case 2: return "TEXKILL"; + case 3: return "PROJ"; + case 4: return "LODBIAS"; + case 5: return "LOD"; + case 6: return "DXDY"; + } + return NULL; +} + +void r500FragmentProgramDump(struct radeon_compiler *c, void *user) +{ + struct r300_fragment_program_compiler *compiler = (struct r300_fragment_program_compiler*)c; + struct r500_fragment_program_code *code = &compiler->code->code.r500; + int n, i; + uint32_t inst; + uint32_t inst0; + char *str = NULL; + fprintf(stderr, "R500 Fragment Program:\n--------\n"); + + for (n = 0; n < code->inst_end+1; n++) { + inst0 = inst = code->inst[n].inst0; + fprintf(stderr,"%d\t0:CMN_INST 0x%08x:", n, inst); + switch(inst & 0x3) { + case R500_INST_TYPE_ALU: str = "ALU"; break; + case R500_INST_TYPE_OUT: str = "OUT"; break; + case R500_INST_TYPE_FC: str = "FC"; break; + case R500_INST_TYPE_TEX: str = "TEX"; break; + }; + fprintf(stderr,"%s %s %s %s %s ", str, + inst & R500_INST_TEX_SEM_WAIT ? "TEX_WAIT" : "", + inst & R500_INST_LAST ? "LAST" : "", + inst & R500_INST_NOP ? "NOP" : "", + inst & R500_INST_ALU_WAIT ? "ALU WAIT" : ""); + fprintf(stderr,"wmask: %s omask: %s\n", to_mask((inst >> 11) & 0xf), + to_mask((inst >> 15) & 0xf)); + + switch(inst0 & 0x3) { + case R500_INST_TYPE_ALU: + case R500_INST_TYPE_OUT: + fprintf(stderr,"\t1:RGB_ADDR 0x%08x:", code->inst[n].inst1); + inst = code->inst[n].inst1; + + fprintf(stderr,"Addr0: %d%c, Addr1: %d%c, Addr2: %d%c, srcp:%d\n", + inst & 0xff, (inst & (1<<8)) ? 'c' : 't', + (inst >> 10) & 0xff, (inst & (1<<18)) ? 'c' : 't', + (inst >> 20) & 0xff, (inst & (1<<28)) ? 'c' : 't', + (inst >> 30)); + + fprintf(stderr,"\t2:ALPHA_ADDR 0x%08x:", code->inst[n].inst2); + inst = code->inst[n].inst2; + fprintf(stderr,"Addr0: %d%c, Addr1: %d%c, Addr2: %d%c, srcp:%d\n", + inst & 0xff, (inst & (1<<8)) ? 'c' : 't', + (inst >> 10) & 0xff, (inst & (1<<18)) ? 'c' : 't', + (inst >> 20) & 0xff, (inst & (1<<28)) ? 'c' : 't', + (inst >> 30)); + fprintf(stderr,"\t3 RGB_INST: 0x%08x:", code->inst[n].inst3); + inst = code->inst[n].inst3; + fprintf(stderr,"rgb_A_src:%d %s/%s/%s %d rgb_B_src:%d %s/%s/%s %d targ: %d\n", + (inst) & 0x3, toswiz((inst >> 2) & 0x7), toswiz((inst >> 5) & 0x7), toswiz((inst >> 8) & 0x7), + (inst >> 11) & 0x3, + (inst >> 13) & 0x3, toswiz((inst >> 15) & 0x7), toswiz((inst >> 18) & 0x7), toswiz((inst >> 21) & 0x7), + (inst >> 24) & 0x3, (inst >> 29) & 0x3); + + + fprintf(stderr,"\t4 ALPHA_INST:0x%08x:", code->inst[n].inst4); + inst = code->inst[n].inst4; + fprintf(stderr,"%s dest:%d%s alp_A_src:%d %s %d alp_B_src:%d %s %d targ %d w:%d\n", to_alpha_op(inst & 0xf), + (inst >> 4) & 0x7f, inst & (1<<11) ? "(rel)":"", + (inst >> 12) & 0x3, toswiz((inst >> 14) & 0x7), (inst >> 17) & 0x3, + (inst >> 19) & 0x3, toswiz((inst >> 21) & 0x7), (inst >> 24) & 0x3, + (inst >> 29) & 0x3, + (inst >> 31) & 0x1); + + fprintf(stderr,"\t5 RGBA_INST: 0x%08x:", code->inst[n].inst5); + inst = code->inst[n].inst5; + fprintf(stderr,"%s dest:%d%s rgb_C_src:%d %s/%s/%s %d alp_C_src:%d %s %d\n", toop(inst & 0xf), + (inst >> 4) & 0x7f, inst & (1<<11) ? "(rel)":"", + (inst >> 12) & 0x3, toswiz((inst >> 14) & 0x7), toswiz((inst >> 17) & 0x7), toswiz((inst >> 20) & 0x7), + (inst >> 23) & 0x3, + (inst >> 25) & 0x3, toswiz((inst >> 27) & 0x7), (inst >> 30) & 0x3); + break; + case R500_INST_TYPE_FC: + fprintf(stderr, "\t2:FC_INST 0x%08x:", code->inst[n].inst2); + inst = code->inst[n].inst2; + /* JUMP_FUNC JUMP_ANY*/ + fprintf(stderr, "0x%02x %1x ", inst >> 8 & 0xff, + (inst & R500_FC_JUMP_ANY) >> 5); + + /* OP */ + switch(inst & 0x7){ + case R500_FC_OP_JUMP: + fprintf(stderr, "JUMP"); + break; + case R500_FC_OP_LOOP: + fprintf(stderr, "LOOP"); + break; + case R500_FC_OP_ENDLOOP: + fprintf(stderr, "ENDLOOP"); + break; + case R500_FC_OP_REP: + fprintf(stderr, "REP"); + break; + case R500_FC_OP_ENDREP: + fprintf(stderr, "ENDREP"); + break; + case R500_FC_OP_BREAKLOOP: + fprintf(stderr, "BREAKLOOP"); + break; + case R500_FC_OP_BREAKREP: + fprintf(stderr, "BREAKREP"); + break; + case R500_FC_OP_CONTINUE: + fprintf(stderr, "CONTINUE"); + break; + } + fprintf(stderr," "); + /* A_OP */ + switch(inst & (0x3 << 6)){ + case R500_FC_A_OP_NONE: + fprintf(stderr, "NONE"); + break; + case R500_FC_A_OP_POP: + fprintf(stderr, "POP"); + break; + case R500_FC_A_OP_PUSH: + fprintf(stderr, "PUSH"); + break; + } + /* B_OP0 B_OP1 */ + for(i=0; i<2; i++){ + fprintf(stderr, " "); + switch(inst & (0x3 << (24 + (i * 2)))){ + /* R500_FC_B_OP0_NONE + * R500_FC_B_OP1_NONE */ + case 0: + fprintf(stderr, "NONE"); + break; + case R500_FC_B_OP0_DECR: + case R500_FC_B_OP1_DECR: + fprintf(stderr, "DECR"); + break; + case R500_FC_B_OP0_INCR: + case R500_FC_B_OP1_INCR: + fprintf(stderr, "INCR"); + break; + } + } + /*POP_CNT B_ELSE */ + fprintf(stderr, " %d %1x", (inst >> 16) & 0x1f, (inst & R500_FC_B_ELSE) >> 4); + inst = code->inst[n].inst3; + /* JUMP_ADDR */ + fprintf(stderr, " %d", inst >> 16); + + if(code->inst[n].inst2 & R500_FC_IGNORE_UNCOVERED){ + fprintf(stderr, " IGN_UNC"); + } + inst = code->inst[n].inst3; + fprintf(stderr, "\n\t3:FC_ADDR 0x%08x:", inst); + fprintf(stderr, "BOOL: 0x%02x, INT: 0x%02x, JUMP_ADDR: %d, JMP_GLBL: %1x\n", + inst & 0x1f, (inst >> 8) & 0x1f, (inst >> 16) & 0x1ff, inst >> 31); + break; + case R500_INST_TYPE_TEX: + inst = code->inst[n].inst1; + fprintf(stderr,"\t1:TEX_INST: 0x%08x: id: %d op:%s, %s, %s %s\n", inst, (inst >> 16) & 0xf, + to_texop((inst >> 22) & 0x7), (inst & (1<<25)) ? "ACQ" : "", + (inst & (1<<26)) ? "IGNUNC" : "", (inst & (1<<27)) ? "UNSCALED" : "SCALED"); + inst = code->inst[n].inst2; + fprintf(stderr,"\t2:TEX_ADDR: 0x%08x: src: %d%s %s/%s/%s/%s dst: %d%s %s/%s/%s/%s\n", inst, + inst & 127, inst & (1<<7) ? "(rel)" : "", + toswiz((inst >> 8) & 0x3), toswiz((inst >> 10) & 0x3), + toswiz((inst >> 12) & 0x3), toswiz((inst >> 14) & 0x3), + (inst >> 16) & 127, inst & (1<<23) ? "(rel)" : "", + toswiz((inst >> 24) & 0x3), toswiz((inst >> 26) & 0x3), + toswiz((inst >> 28) & 0x3), toswiz((inst >> 30) & 0x3)); + + fprintf(stderr,"\t3:TEX_DXDY: 0x%08x\n", code->inst[n].inst3); + break; + } + fprintf(stderr,"\n"); + } + +} diff --git a/src/gallium/drivers/r300/compiler/r500_fragprog.h b/src/gallium/drivers/r300/compiler/r500_fragprog.h new file mode 100644 index 00000000000..6aa448cc6f7 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/r500_fragprog.h @@ -0,0 +1,50 @@ +/* + * Copyright (C) 2005 Ben Skeggs. + * + * 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. + * + */ + +/* + * Authors: + * Ben Skeggs <[email protected]> + * Jerome Glisse <[email protected]> + */ +#ifndef __R500_FRAGPROG_H_ +#define __R500_FRAGPROG_H_ + +#include "radeon_compiler.h" +#include "radeon_swizzle.h" + +extern void r500BuildFragmentProgramHwCode(struct radeon_compiler *c, void *user); + +extern void r500FragmentProgramDump(struct radeon_compiler *c, void *user); + +extern struct rc_swizzle_caps r500_swizzle_caps; + +extern int r500_transform_IF( + struct radeon_compiler * c, + struct rc_instruction * inst_if, + void* data); + +#endif diff --git a/src/gallium/drivers/r300/compiler/r500_fragprog_emit.c b/src/gallium/drivers/r300/compiler/r500_fragprog_emit.c new file mode 100644 index 00000000000..c30cd753d15 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/r500_fragprog_emit.c @@ -0,0 +1,678 @@ +/* + * Copyright (C) 2005 Ben Skeggs. + * + * Copyright 2008 Corbin Simpson <[email protected]> + * 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 <[email protected]> + * + * \author Jerome Glisse <[email protected]> + * + * \author Corbin Simpson <[email protected]> + * + */ + +#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; + } + + 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 |= R500_INST_TEX_SEM_WAIT; + + 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; + } + + 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) + | R500_INST_TEX_SEM_WAIT; + code->inst[ip].inst1 = R500_TEX_ID(inst->TexSrcUnit) + | R500_TEX_SEM_ACQUIRE; + + 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; + } + + /* 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; + } +} diff --git a/src/gallium/drivers/r300/compiler/radeon_code.c b/src/gallium/drivers/r300/compiler/radeon_code.c new file mode 100644 index 00000000000..6842fb873bc --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_code.c @@ -0,0 +1,187 @@ +/* + * Copyright (C) 2009 Nicolai Haehnle. + * + * 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. + * + */ + +#include "radeon_code.h" + +#include <stdlib.h> +#include <stdio.h> +#include <string.h> + +#include "radeon_program.h" + +void rc_constants_init(struct rc_constant_list * c) +{ + memset(c, 0, sizeof(*c)); +} + +/** + * Copy a constants structure, assuming that the destination structure + * is not initialized. + */ +void rc_constants_copy(struct rc_constant_list * dst, struct rc_constant_list * src) +{ + dst->Constants = malloc(sizeof(struct rc_constant) * src->Count); + memcpy(dst->Constants, src->Constants, sizeof(struct rc_constant) * src->Count); + dst->Count = src->Count; + dst->_Reserved = src->Count; +} + +void rc_constants_destroy(struct rc_constant_list * c) +{ + free(c->Constants); + memset(c, 0, sizeof(*c)); +} + +unsigned rc_constants_add(struct rc_constant_list * c, struct rc_constant * constant) +{ + unsigned index = c->Count; + + if (c->Count >= c->_Reserved) { + struct rc_constant * newlist; + + c->_Reserved = c->_Reserved * 2; + if (!c->_Reserved) + c->_Reserved = 16; + + newlist = malloc(sizeof(struct rc_constant) * c->_Reserved); + memcpy(newlist, c->Constants, sizeof(struct rc_constant) * c->Count); + + free(c->Constants); + c->Constants = newlist; + } + + c->Constants[index] = *constant; + c->Count++; + + return index; +} + + +/** + * Add a state vector to the constant list, while trying to avoid duplicates. + */ +unsigned rc_constants_add_state(struct rc_constant_list * c, unsigned state0, unsigned state1) +{ + unsigned index; + struct rc_constant constant; + + for(index = 0; index < c->Count; ++index) { + if (c->Constants[index].Type == RC_CONSTANT_STATE) { + if (c->Constants[index].u.State[0] == state0 && + c->Constants[index].u.State[1] == state1) + return index; + } + } + + memset(&constant, 0, sizeof(constant)); + constant.Type = RC_CONSTANT_STATE; + constant.Size = 4; + constant.u.State[0] = state0; + constant.u.State[1] = state1; + + return rc_constants_add(c, &constant); +} + + +/** + * Add an immediate vector to the constant list, while trying to avoid + * duplicates. + */ +unsigned rc_constants_add_immediate_vec4(struct rc_constant_list * c, const float * data) +{ + unsigned index; + struct rc_constant constant; + + for(index = 0; index < c->Count; ++index) { + if (c->Constants[index].Type == RC_CONSTANT_IMMEDIATE) { + if (!memcmp(c->Constants[index].u.Immediate, data, sizeof(float)*4)) + return index; + } + } + + memset(&constant, 0, sizeof(constant)); + constant.Type = RC_CONSTANT_IMMEDIATE; + constant.Size = 4; + memcpy(constant.u.Immediate, data, sizeof(float) * 4); + + return rc_constants_add(c, &constant); +} + + +/** + * Add an immediate scalar to the constant list, while trying to avoid + * duplicates. + */ +unsigned rc_constants_add_immediate_scalar(struct rc_constant_list * c, float data, unsigned * swizzle) +{ + unsigned index; + int free_index = -1; + struct rc_constant constant; + + for(index = 0; index < c->Count; ++index) { + if (c->Constants[index].Type == RC_CONSTANT_IMMEDIATE) { + unsigned comp; + for(comp = 0; comp < c->Constants[index].Size; ++comp) { + if (c->Constants[index].u.Immediate[comp] == data) { + *swizzle = RC_MAKE_SWIZZLE_SMEAR(comp); + return index; + } + } + + if (c->Constants[index].Size < 4) + free_index = index; + } + } + + if (free_index >= 0) { + unsigned comp = c->Constants[free_index].Size++; + c->Constants[free_index].u.Immediate[comp] = data; + *swizzle = RC_MAKE_SWIZZLE_SMEAR(comp); + return free_index; + } + + memset(&constant, 0, sizeof(constant)); + constant.Type = RC_CONSTANT_IMMEDIATE; + constant.Size = 1; + constant.u.Immediate[0] = data; + *swizzle = RC_SWIZZLE_XXXX; + + return rc_constants_add(c, &constant); +} + +void rc_constants_print(struct rc_constant_list * c) +{ + unsigned int i; + for(i = 0; i < c->Count; i++) { + if (c->Constants[i].Type == RC_CONSTANT_IMMEDIATE) { + float * values = c->Constants[i].u.Immediate; + fprintf(stderr, "CONST[%u] = " + "{ %10.4f %10.4f %10.4f %10.4f }\n", + i, values[0],values[1], values[2], values[3]); + } + } +} diff --git a/src/gallium/drivers/r300/compiler/radeon_code.h b/src/gallium/drivers/r300/compiler/radeon_code.h new file mode 100644 index 00000000000..67e6acf8b10 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_code.h @@ -0,0 +1,306 @@ +/* + * Copyright 2009 Nicolai Hähnle <[email protected]> + * + * 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 + * on the rights to use, copy, modify, merge, publish, distribute, sub + * license, 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 NON-INFRINGEMENT. IN NO EVENT SHALL + * THE AUTHOR(S) AND/OR THEIR 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. */ + +#ifndef RADEON_CODE_H +#define RADEON_CODE_H + +#include <stdint.h> + +#define R300_PFS_MAX_ALU_INST 64 +#define R300_PFS_MAX_TEX_INST 32 +#define R300_PFS_MAX_TEX_INDIRECT 4 +#define R300_PFS_NUM_TEMP_REGS 32 +#define R300_PFS_NUM_CONST_REGS 32 + +#define R400_PFS_MAX_ALU_INST 512 +#define R400_PFS_MAX_TEX_INST 512 + +#define R500_PFS_MAX_INST 512 +#define R500_PFS_NUM_TEMP_REGS 128 +#define R500_PFS_NUM_CONST_REGS 256 +#define R500_PFS_MAX_BRANCH_DEPTH_FULL 32 +#define R500_PFS_MAX_BRANCH_DEPTH_PARTIAL 4 + + +#define STATE_R300_WINDOW_DIMENSION (STATE_INTERNAL_DRIVER+0) + +enum { + /** + * External constants are constants whose meaning is unknown to this + * compiler. For example, a Mesa gl_program's constants are turned + * into external constants. + */ + RC_CONSTANT_EXTERNAL = 0, + + RC_CONSTANT_IMMEDIATE, + + /** + * Constant referring to state that is known by this compiler, + * see RC_STATE_xxx, i.e. *not* arbitrary Mesa (or other) state. + */ + RC_CONSTANT_STATE +}; + +enum { + RC_STATE_SHADOW_AMBIENT = 0, + + RC_STATE_R300_WINDOW_DIMENSION, + RC_STATE_R300_TEXRECT_FACTOR, + RC_STATE_R300_TEXSCALE_FACTOR, + RC_STATE_R300_VIEWPORT_SCALE, + RC_STATE_R300_VIEWPORT_OFFSET +}; + +struct rc_constant { + unsigned Type:2; /**< RC_CONSTANT_xxx */ + unsigned Size:3; + + union { + unsigned External; + float Immediate[4]; + unsigned State[2]; + } u; +}; + +struct rc_constant_list { + struct rc_constant * Constants; + unsigned Count; + + unsigned _Reserved; +}; + +void rc_constants_init(struct rc_constant_list * c); +void rc_constants_copy(struct rc_constant_list * dst, struct rc_constant_list * src); +void rc_constants_destroy(struct rc_constant_list * c); +unsigned rc_constants_add(struct rc_constant_list * c, struct rc_constant * constant); +unsigned rc_constants_add_state(struct rc_constant_list * c, unsigned state1, unsigned state2); +unsigned rc_constants_add_immediate_vec4(struct rc_constant_list * c, const float * data); +unsigned rc_constants_add_immediate_scalar(struct rc_constant_list * c, float data, unsigned * swizzle); +void rc_constants_print(struct rc_constant_list * c); + +/** + * Compare functions. + * + * \note By design, RC_COMPARE_FUNC_xxx + GL_NEVER gives you + * the correct GL compare function. + */ +typedef enum { + RC_COMPARE_FUNC_NEVER = 0, + RC_COMPARE_FUNC_LESS, + RC_COMPARE_FUNC_EQUAL, + RC_COMPARE_FUNC_LEQUAL, + RC_COMPARE_FUNC_GREATER, + RC_COMPARE_FUNC_NOTEQUAL, + RC_COMPARE_FUNC_GEQUAL, + RC_COMPARE_FUNC_ALWAYS +} rc_compare_func; + +/** + * Coordinate wrapping modes. + * + * These are not quite the same as their GL counterparts yet. + */ +typedef enum { + RC_WRAP_NONE = 0, + RC_WRAP_REPEAT, + RC_WRAP_MIRRORED_REPEAT, + RC_WRAP_MIRRORED_CLAMP +} rc_wrap_mode; + +/** + * Stores state that influences the compilation of a fragment program. + */ +struct r300_fragment_program_external_state { + struct { + /** + * This field contains swizzle for some lowering passes + * (shadow comparison, unorm->snorm conversion) + */ + unsigned texture_swizzle:12; + + /** + * If the sampler is used as a shadow sampler, + * this field specifies the compare function. + * + * Otherwise, this field is \ref RC_COMPARE_FUNC_NEVER (aka 0). + * \sa rc_compare_func + */ + unsigned texture_compare_func : 3; + + /** + * No matter what the sampler type is, + * this field turns it into a shadow sampler. + */ + unsigned compare_mode_enabled : 1; + + /** + * If the sampler will receive non-normalized coords, + * this field is set. The scaling factor is given by + * RC_STATE_R300_TEXRECT_FACTOR. + */ + unsigned non_normalized_coords : 1; + + /** + * This field specifies wrapping modes for the sampler. + * + * If this field is \ref RC_WRAP_NONE (aka 0), no wrapping maths + * will be performed on the coordinates. + */ + unsigned wrap_mode : 3; + + /** + * The coords are scaled after applying the wrap mode emulation + * and right before texture fetch. The scaling factor is given by + * RC_STATE_R300_TEXSCALE_FACTOR. */ + unsigned clamp_and_scale_before_fetch : 1; + + /** + * Fetch RGTC1_SNORM or LATC1_SNORM as UNORM and convert UNORM -> SNORM + * in the shader. + */ + unsigned convert_unorm_to_snorm:1; + } unit[16]; + + unsigned frag_clamp:1; +}; + + + +struct r300_fragment_program_node { + int tex_offset; /**< first tex instruction */ + int tex_end; /**< last tex instruction, relative to tex_offset */ + int alu_offset; /**< first ALU instruction */ + int alu_end; /**< last ALU instruction, relative to alu_offset */ + int flags; +}; + +/** + * Stores an R300 fragment program in its compiled-to-hardware form. + */ +struct r300_fragment_program_code { + struct { + unsigned int length; /**< total # of texture instructions used */ + uint32_t inst[R400_PFS_MAX_TEX_INST]; + } tex; + + struct { + unsigned int length; /**< total # of ALU instructions used */ + struct { + uint32_t rgb_inst; + uint32_t rgb_addr; + uint32_t alpha_inst; + uint32_t alpha_addr; + uint32_t r400_ext_addr; + } inst[R400_PFS_MAX_ALU_INST]; + } alu; + + uint32_t config; /* US_CONFIG */ + uint32_t pixsize; /* US_PIXSIZE */ + uint32_t code_offset; /* US_CODE_OFFSET */ + uint32_t r400_code_offset_ext; /* US_CODE_EXT */ + uint32_t code_addr[4]; /* US_CODE_ADDR */ + /*US_CODE_BANK.R390_MODE: Enables 512 instructions and 64 temporaries + * for r400 cards */ + unsigned int r390_mode:1; +}; + + +struct r500_fragment_program_code { + struct { + uint32_t inst0; + uint32_t inst1; + uint32_t inst2; + uint32_t inst3; + uint32_t inst4; + uint32_t inst5; + } inst[R500_PFS_MAX_INST]; + + int inst_end; /* Number of instructions - 1; also, last instruction to be executed */ + + int max_temp_idx; + + uint32_t us_fc_ctrl; + + uint32_t int_constants[32]; + uint32_t int_constant_count; +}; + +struct rX00_fragment_program_code { + union { + struct r300_fragment_program_code r300; + struct r500_fragment_program_code r500; + } code; + + unsigned writes_depth:1; + + struct rc_constant_list constants; + unsigned *constants_remap_table; +}; + + +#define R300_VS_MAX_ALU 256 +#define R300_VS_MAX_ALU_DWORDS (R300_VS_MAX_ALU * 4) +#define R500_VS_MAX_ALU 1024 +#define R500_VS_MAX_ALU_DWORDS (R500_VS_MAX_ALU * 4) +#define R300_VS_MAX_TEMPS 32 +/* This is the max for all chipsets (r300-r500) */ +#define R300_VS_MAX_FC_OPS 16 +/* The r500 maximum depth is not just for loops, but any combination of loops + * and subroutine jumps. */ +#define R500_VS_MAX_FC_DEPTH 8 +#define R300_VS_MAX_LOOP_DEPTH 1 + +#define VSF_MAX_INPUTS 32 +#define VSF_MAX_OUTPUTS 32 + +struct r300_vertex_program_code { + int length; + union { + uint32_t d[R500_VS_MAX_ALU_DWORDS]; + float f[R500_VS_MAX_ALU_DWORDS]; + } body; + + int pos_end; + int num_temporaries; /* Number of temp vars used by program */ + int inputs[VSF_MAX_INPUTS]; + int outputs[VSF_MAX_OUTPUTS]; + + struct rc_constant_list constants; + unsigned *constants_remap_table; + + uint32_t InputsRead; + uint32_t OutputsWritten; + + unsigned int num_fc_ops; + uint32_t fc_ops; + union { + uint32_t r300[R300_VS_MAX_FC_OPS]; + struct { + uint32_t lw; + uint32_t uw; + } r500[R300_VS_MAX_FC_OPS]; + } fc_op_addrs; + int32_t fc_loop_index[R300_VS_MAX_FC_OPS]; +}; + +#endif /* RADEON_CODE_H */ + diff --git a/src/gallium/drivers/r300/compiler/radeon_compiler.c b/src/gallium/drivers/r300/compiler/radeon_compiler.c new file mode 100644 index 00000000000..b7936725d85 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_compiler.c @@ -0,0 +1,489 @@ +/* + * Copyright 2009 Nicolai Hähnle <[email protected]> + * + * 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 + * on the rights to use, copy, modify, merge, publish, distribute, sub + * license, 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 NON-INFRINGEMENT. IN NO EVENT SHALL + * THE AUTHOR(S) AND/OR THEIR 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. */ + +#include "radeon_compiler.h" + +#include <stdarg.h> +#include <stdio.h> +#include <stdlib.h> + +#include "radeon_dataflow.h" +#include "radeon_program.h" +#include "radeon_program_pair.h" +#include "radeon_compiler_util.h" + + +void rc_init(struct radeon_compiler * c) +{ + memset(c, 0, sizeof(*c)); + + memory_pool_init(&c->Pool); + c->Program.Instructions.Prev = &c->Program.Instructions; + c->Program.Instructions.Next = &c->Program.Instructions; + c->Program.Instructions.U.I.Opcode = RC_OPCODE_ILLEGAL_OPCODE; +} + +void rc_destroy(struct radeon_compiler * c) +{ + rc_constants_destroy(&c->Program.Constants); + memory_pool_destroy(&c->Pool); + free(c->ErrorMsg); +} + +void rc_debug(struct radeon_compiler * c, const char * fmt, ...) +{ + va_list ap; + + if (!(c->Debug & RC_DBG_LOG)) + return; + + va_start(ap, fmt); + vfprintf(stderr, fmt, ap); + va_end(ap); +} + +void rc_error(struct radeon_compiler * c, const char * fmt, ...) +{ + va_list ap; + + c->Error = 1; + + if (!c->ErrorMsg) { + /* Only remember the first error */ + char buf[1024]; + int written; + + va_start(ap, fmt); + written = vsnprintf(buf, sizeof(buf), fmt, ap); + va_end(ap); + + if (written < sizeof(buf)) { + c->ErrorMsg = strdup(buf); + } else { + c->ErrorMsg = malloc(written + 1); + + va_start(ap, fmt); + vsnprintf(c->ErrorMsg, written + 1, fmt, ap); + va_end(ap); + } + } + + if (c->Debug & RC_DBG_LOG) { + fprintf(stderr, "r300compiler error: "); + + va_start(ap, fmt); + vfprintf(stderr, fmt, ap); + va_end(ap); + } +} + +int rc_if_fail_helper(struct radeon_compiler * c, const char * file, int line, const char * assertion) +{ + rc_error(c, "ICE at %s:%i: assertion failed: %s\n", file, line, assertion); + return 1; +} + +/** + * Recompute c->Program.InputsRead and c->Program.OutputsWritten + * based on which inputs and outputs are actually referenced + * in program instructions. + */ +void rc_calculate_inputs_outputs(struct radeon_compiler * c) +{ + struct rc_instruction *inst; + + c->Program.InputsRead = 0; + c->Program.OutputsWritten = 0; + + for(inst = c->Program.Instructions.Next; inst != &c->Program.Instructions; inst = inst->Next) + { + const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode); + int i; + + for (i = 0; i < opcode->NumSrcRegs; ++i) { + if (inst->U.I.SrcReg[i].File == RC_FILE_INPUT) + c->Program.InputsRead |= 1 << inst->U.I.SrcReg[i].Index; + } + + if (opcode->HasDstReg) { + if (inst->U.I.DstReg.File == RC_FILE_OUTPUT) + c->Program.OutputsWritten |= 1 << inst->U.I.DstReg.Index; + } + } +} + +/** + * Rewrite the program such that everything that source the given input + * register will source new_input instead. + */ +void rc_move_input(struct radeon_compiler * c, unsigned input, struct rc_src_register new_input) +{ + struct rc_instruction * inst; + + c->Program.InputsRead &= ~(1 << input); + + for(inst = c->Program.Instructions.Next; inst != &c->Program.Instructions; inst = inst->Next) { + const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode); + unsigned i; + + for(i = 0; i < opcode->NumSrcRegs; ++i) { + if (inst->U.I.SrcReg[i].File == RC_FILE_INPUT && inst->U.I.SrcReg[i].Index == input) { + inst->U.I.SrcReg[i].File = new_input.File; + inst->U.I.SrcReg[i].Index = new_input.Index; + inst->U.I.SrcReg[i].Swizzle = combine_swizzles(new_input.Swizzle, inst->U.I.SrcReg[i].Swizzle); + if (!inst->U.I.SrcReg[i].Abs) { + inst->U.I.SrcReg[i].Negate ^= new_input.Negate; + inst->U.I.SrcReg[i].Abs = new_input.Abs; + } + + c->Program.InputsRead |= 1 << new_input.Index; + } + } + } +} + + +/** + * Rewrite the program such that everything that writes into the given + * output register will instead write to new_output. The new_output + * writemask is honoured. + */ +void rc_move_output(struct radeon_compiler * c, unsigned output, unsigned new_output, unsigned writemask) +{ + struct rc_instruction * inst; + + c->Program.OutputsWritten &= ~(1 << output); + + for(inst = c->Program.Instructions.Next; inst != &c->Program.Instructions; inst = inst->Next) { + const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode); + + if (opcode->HasDstReg) { + if (inst->U.I.DstReg.File == RC_FILE_OUTPUT && inst->U.I.DstReg.Index == output) { + inst->U.I.DstReg.Index = new_output; + inst->U.I.DstReg.WriteMask &= writemask; + + c->Program.OutputsWritten |= 1 << new_output; + } + } + } +} + + +/** + * Rewrite the program such that a given output is duplicated. + */ +void rc_copy_output(struct radeon_compiler * c, unsigned output, unsigned dup_output) +{ + unsigned tempreg = rc_find_free_temporary(c); + struct rc_instruction * inst; + + for(inst = c->Program.Instructions.Next; inst != &c->Program.Instructions; inst = inst->Next) { + const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode); + + if (opcode->HasDstReg) { + if (inst->U.I.DstReg.File == RC_FILE_OUTPUT && inst->U.I.DstReg.Index == output) { + inst->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst->U.I.DstReg.Index = tempreg; + } + } + } + + inst = rc_insert_new_instruction(c, c->Program.Instructions.Prev); + inst->U.I.Opcode = RC_OPCODE_MOV; + inst->U.I.DstReg.File = RC_FILE_OUTPUT; + inst->U.I.DstReg.Index = output; + + inst->U.I.SrcReg[0].File = RC_FILE_TEMPORARY; + inst->U.I.SrcReg[0].Index = tempreg; + inst->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_XYZW; + + inst = rc_insert_new_instruction(c, c->Program.Instructions.Prev); + inst->U.I.Opcode = RC_OPCODE_MOV; + inst->U.I.DstReg.File = RC_FILE_OUTPUT; + inst->U.I.DstReg.Index = dup_output; + + inst->U.I.SrcReg[0].File = RC_FILE_TEMPORARY; + inst->U.I.SrcReg[0].Index = tempreg; + inst->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_XYZW; + + c->Program.OutputsWritten |= 1 << dup_output; +} + + +/** + * Introduce standard code fragment to deal with fragment.position. + */ +void rc_transform_fragment_wpos(struct radeon_compiler * c, unsigned wpos, unsigned new_input, + int full_vtransform) +{ + unsigned tempregi = rc_find_free_temporary(c); + struct rc_instruction * inst_rcp; + struct rc_instruction * inst_mul; + struct rc_instruction * inst_mad; + struct rc_instruction * inst; + + c->Program.InputsRead &= ~(1 << wpos); + c->Program.InputsRead |= 1 << new_input; + + /* perspective divide */ + inst_rcp = rc_insert_new_instruction(c, &c->Program.Instructions); + inst_rcp->U.I.Opcode = RC_OPCODE_RCP; + + inst_rcp->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst_rcp->U.I.DstReg.Index = tempregi; + inst_rcp->U.I.DstReg.WriteMask = RC_MASK_W; + + inst_rcp->U.I.SrcReg[0].File = RC_FILE_INPUT; + inst_rcp->U.I.SrcReg[0].Index = new_input; + inst_rcp->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_WWWW; + + inst_mul = rc_insert_new_instruction(c, inst_rcp); + inst_mul->U.I.Opcode = RC_OPCODE_MUL; + + inst_mul->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst_mul->U.I.DstReg.Index = tempregi; + inst_mul->U.I.DstReg.WriteMask = RC_MASK_XYZ; + + inst_mul->U.I.SrcReg[0].File = RC_FILE_INPUT; + inst_mul->U.I.SrcReg[0].Index = new_input; + + inst_mul->U.I.SrcReg[1].File = RC_FILE_TEMPORARY; + inst_mul->U.I.SrcReg[1].Index = tempregi; + inst_mul->U.I.SrcReg[1].Swizzle = RC_SWIZZLE_WWWW; + + /* viewport transformation */ + inst_mad = rc_insert_new_instruction(c, inst_mul); + inst_mad->U.I.Opcode = RC_OPCODE_MAD; + + inst_mad->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst_mad->U.I.DstReg.Index = tempregi; + inst_mad->U.I.DstReg.WriteMask = RC_MASK_XYZ; + + inst_mad->U.I.SrcReg[0].File = RC_FILE_TEMPORARY; + inst_mad->U.I.SrcReg[0].Index = tempregi; + inst_mad->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_XYZ0; + + inst_mad->U.I.SrcReg[1].File = RC_FILE_CONSTANT; + inst_mad->U.I.SrcReg[1].Swizzle = RC_SWIZZLE_XYZ0; + + inst_mad->U.I.SrcReg[2].File = RC_FILE_CONSTANT; + inst_mad->U.I.SrcReg[2].Swizzle = RC_SWIZZLE_XYZ0; + + if (full_vtransform) { + inst_mad->U.I.SrcReg[1].Index = rc_constants_add_state(&c->Program.Constants, RC_STATE_R300_VIEWPORT_SCALE, 0); + inst_mad->U.I.SrcReg[2].Index = rc_constants_add_state(&c->Program.Constants, RC_STATE_R300_VIEWPORT_OFFSET, 0); + } else { + inst_mad->U.I.SrcReg[1].Index = + inst_mad->U.I.SrcReg[2].Index = rc_constants_add_state(&c->Program.Constants, RC_STATE_R300_WINDOW_DIMENSION, 0); + } + + for (inst = inst_mad->Next; inst != &c->Program.Instructions; inst = inst->Next) { + const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode); + unsigned i; + + for(i = 0; i < opcode->NumSrcRegs; i++) { + if (inst->U.I.SrcReg[i].File == RC_FILE_INPUT && + inst->U.I.SrcReg[i].Index == wpos) { + inst->U.I.SrcReg[i].File = RC_FILE_TEMPORARY; + inst->U.I.SrcReg[i].Index = tempregi; + } + } + } +} + + +/** + * The FACE input in hardware contains 1 if it's a back face, 0 otherwise. + * Gallium and OpenGL define it the other way around. + * + * So let's just negate FACE at the beginning of the shader and rewrite the rest + * of the shader to read from the newly allocated temporary. + */ +void rc_transform_fragment_face(struct radeon_compiler *c, unsigned face) +{ + unsigned tempregi = rc_find_free_temporary(c); + struct rc_instruction *inst_add; + struct rc_instruction *inst; + + /* perspective divide */ + inst_add = rc_insert_new_instruction(c, &c->Program.Instructions); + inst_add->U.I.Opcode = RC_OPCODE_ADD; + + inst_add->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst_add->U.I.DstReg.Index = tempregi; + inst_add->U.I.DstReg.WriteMask = RC_MASK_X; + + inst_add->U.I.SrcReg[0].File = RC_FILE_NONE; + inst_add->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_1111; + + inst_add->U.I.SrcReg[1].File = RC_FILE_INPUT; + inst_add->U.I.SrcReg[1].Index = face; + inst_add->U.I.SrcReg[1].Swizzle = RC_SWIZZLE_XXXX; + inst_add->U.I.SrcReg[1].Negate = RC_MASK_XYZW; + + for (inst = inst_add->Next; inst != &c->Program.Instructions; inst = inst->Next) { + const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode); + unsigned i; + + for(i = 0; i < opcode->NumSrcRegs; i++) { + if (inst->U.I.SrcReg[i].File == RC_FILE_INPUT && + inst->U.I.SrcReg[i].Index == face) { + inst->U.I.SrcReg[i].File = RC_FILE_TEMPORARY; + inst->U.I.SrcReg[i].Index = tempregi; + } + } + } +} + +static void reg_count_callback(void * userdata, struct rc_instruction * inst, + rc_register_file file, unsigned int index, unsigned int mask) +{ + int *max_reg = userdata; + if (file == RC_FILE_TEMPORARY) + (int)index > *max_reg ? *max_reg = index : 0; +} + +void rc_get_stats(struct radeon_compiler *c, struct rc_program_stats *s) +{ + int max_reg = -1; + struct rc_instruction * tmp; + memset(s, 0, sizeof(*s)); + + for(tmp = c->Program.Instructions.Next; tmp != &c->Program.Instructions; + tmp = tmp->Next){ + const struct rc_opcode_info * info; + rc_for_all_reads_mask(tmp, reg_count_callback, &max_reg); + if (tmp->Type == RC_INSTRUCTION_NORMAL) { + info = rc_get_opcode_info(tmp->U.I.Opcode); + if (info->Opcode == RC_OPCODE_BEGIN_TEX) + continue; + if (tmp->U.I.PreSub.Opcode != RC_PRESUB_NONE) + s->num_presub_ops++; + } else { + if (tmp->U.P.RGB.Src[RC_PAIR_PRESUB_SRC].Used) + s->num_presub_ops++; + if (tmp->U.P.Alpha.Src[RC_PAIR_PRESUB_SRC].Used) + s->num_presub_ops++; + /* Assuming alpha will never be a flow control or + * a tex instruction. */ + if (tmp->U.P.Alpha.Opcode != RC_OPCODE_NOP) + s->num_alpha_insts++; + if (tmp->U.P.RGB.Opcode != RC_OPCODE_NOP) + s->num_rgb_insts++; + info = rc_get_opcode_info(tmp->U.P.RGB.Opcode); + } + if (info->IsFlowControl) + s->num_fc_insts++; + if (info->HasTexture) + s->num_tex_insts++; + s->num_insts++; + } + s->num_temp_regs = max_reg + 1; +} + +static void print_stats(struct radeon_compiler * c) +{ + struct rc_program_stats s; + + if (c->initial_num_insts <= 5) + return; + + rc_get_stats(c, &s); + + switch (c->type) { + case RC_VERTEX_PROGRAM: + fprintf(stderr,"~~~~~~~~~ VERTEX PROGRAM ~~~~~~~~\n" + "~%4u Instructions\n" + "~%4u Flow Control Instructions\n" + "~%4u Temporary Registers\n" + "~~~~~~~~~~~~~~ END ~~~~~~~~~~~~~~\n", + s.num_insts, s.num_fc_insts, s.num_temp_regs); + break; + + case RC_FRAGMENT_PROGRAM: + fprintf(stderr,"~~~~~~~~ FRAGMENT PROGRAM ~~~~~~~\n" + "~%4u Instructions\n" + "~%4u Vector Instructions (RGB)\n" + "~%4u Scalar Instructions (Alpha)\n" + "~%4u Flow Control Instructions\n" + "~%4u Texture Instructions\n" + "~%4u Presub Operations\n" + "~%4u Temporary Registers\n" + "~~~~~~~~~~~~~~ END ~~~~~~~~~~~~~~\n", + s.num_insts, s.num_rgb_insts, s.num_alpha_insts, + s.num_fc_insts, s.num_tex_insts, s.num_presub_ops, + s.num_temp_regs); + break; + default: + assert(0); + } +} + +static const char *shader_name[RC_NUM_PROGRAM_TYPES] = { + "Vertex Program", + "Fragment Program" +}; + +void rc_run_compiler_passes(struct radeon_compiler *c, struct radeon_compiler_pass *list) +{ + for (unsigned i = 0; list[i].name; i++) { + if (list[i].predicate) { + list[i].run(c, list[i].user); + + if (c->Error) + return; + + if ((c->Debug & RC_DBG_LOG) && list[i].dump) { + fprintf(stderr, "%s: after '%s'\n", shader_name[c->type], list[i].name); + rc_print_program(&c->Program); + } + } + } +} + +/* Executes a list of compiler passes given in the parameter 'list'. */ +void rc_run_compiler(struct radeon_compiler *c, struct radeon_compiler_pass *list) +{ + struct rc_program_stats s; + + rc_get_stats(c, &s); + c->initial_num_insts = s.num_insts; + + if (c->Debug & RC_DBG_LOG) { + fprintf(stderr, "%s: before compilation\n", shader_name[c->type]); + rc_print_program(&c->Program); + } + + rc_run_compiler_passes(c, list); + + if (c->Debug & RC_DBG_STATS) + print_stats(c); +} + +void rc_validate_final_shader(struct radeon_compiler *c, void *user) +{ + /* Check the number of constants. */ + if (c->Program.Constants.Count > c->max_constants) { + rc_error(c, "Too many constants. Max: %i, Got: %i\n", + c->max_constants, c->Program.Constants.Count); + } +} diff --git a/src/gallium/drivers/r300/compiler/radeon_compiler.h b/src/gallium/drivers/r300/compiler/radeon_compiler.h new file mode 100644 index 00000000000..74594af23c2 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_compiler.h @@ -0,0 +1,171 @@ +/* + * Copyright 2009 Nicolai Hähnle <[email protected]> + * + * 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 + * on the rights to use, copy, modify, merge, publish, distribute, sub + * license, 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 NON-INFRINGEMENT. IN NO EVENT SHALL + * THE AUTHOR(S) AND/OR THEIR 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. */ + +#ifndef RADEON_COMPILER_H +#define RADEON_COMPILER_H + +#include "main/compiler.h" + +#include "memory_pool.h" +#include "radeon_code.h" +#include "radeon_program.h" +#include "radeon_emulate_loops.h" + +#define RC_DBG_LOG (1 << 0) +#define RC_DBG_STATS (1 << 1) + +struct rc_swizzle_caps; + +enum rc_program_type { + RC_VERTEX_PROGRAM, + RC_FRAGMENT_PROGRAM, + RC_NUM_PROGRAM_TYPES +}; + +struct radeon_compiler { + struct memory_pool Pool; + struct rc_program Program; + enum rc_program_type type; + unsigned Debug:2; + unsigned Error:1; + char * ErrorMsg; + + /* Hardware specification. */ + unsigned is_r400:1; + unsigned is_r500:1; + unsigned has_half_swizzles:1; + unsigned has_presub:1; + unsigned disable_optimizations:1; + unsigned max_temp_regs; + unsigned max_constants; + int max_alu_insts; + unsigned max_tex_insts; + + /* Whether to remove unused constants and empty holes in constant space. */ + unsigned remove_unused_constants:1; + + /** + * Variables used internally, not be touched by callers + * of the compiler + */ + /*@{*/ + struct rc_swizzle_caps * SwizzleCaps; + /*@}*/ + + struct emulate_loop_state loop_state; + + unsigned initial_num_insts; /* Number of instructions at start. */ +}; + +void rc_init(struct radeon_compiler * c); +void rc_destroy(struct radeon_compiler * c); + +void rc_debug(struct radeon_compiler * c, const char * fmt, ...); +void rc_error(struct radeon_compiler * c, const char * fmt, ...); + +int rc_if_fail_helper(struct radeon_compiler * c, const char * file, int line, const char * assertion); + +/** + * This macro acts like an if-statement that can be used to implement + * non-aborting assertions in the compiler. + * + * It checks whether \p cond is true. If not, an internal compiler error is + * flagged and the if-clause is run. + * + * A typical use-case would be: + * + * if (rc_assert(c, condition-that-must-be-true)) + * return; + */ +#define rc_assert(c, cond) \ + (!(cond) && rc_if_fail_helper(c, __FILE__, __LINE__, #cond)) + +void rc_calculate_inputs_outputs(struct radeon_compiler * c); + +void rc_move_input(struct radeon_compiler * c, unsigned input, struct rc_src_register new_input); +void rc_move_output(struct radeon_compiler * c, unsigned output, unsigned new_output, unsigned writemask); +void rc_copy_output(struct radeon_compiler * c, unsigned output, unsigned dup_output); +void rc_transform_fragment_wpos(struct radeon_compiler * c, unsigned wpos, unsigned new_input, + int full_vtransform); +void rc_transform_fragment_face(struct radeon_compiler *c, unsigned face); + +struct r300_fragment_program_compiler { + struct radeon_compiler Base; + struct rX00_fragment_program_code *code; + /* Optional transformations and features. */ + struct r300_fragment_program_external_state state; + unsigned enable_shadow_ambient; + /* Register corresponding to the depthbuffer. */ + unsigned OutputDepth; + /* Registers corresponding to the four colorbuffers. */ + unsigned OutputColor[4]; + + void * UserData; + void (*AllocateHwInputs)( + struct r300_fragment_program_compiler * c, + void (*allocate)(void * data, unsigned input, unsigned hwreg), + void * mydata); +}; + +void r3xx_compile_fragment_program(struct r300_fragment_program_compiler* c); + +struct r300_vertex_program_compiler { + struct radeon_compiler Base; + struct r300_vertex_program_code *code; + uint32_t RequiredOutputs; + + void * UserData; + void (*SetHwInputOutput)(struct r300_vertex_program_compiler * c); + + int PredicateIndex; + unsigned int PredicateMask; +}; + +void r3xx_compile_vertex_program(struct r300_vertex_program_compiler* c); +void r300_vertex_program_dump(struct radeon_compiler *compiler, void *user); + +struct radeon_compiler_pass { + const char *name; /* Name of the pass. */ + int dump; /* Dump the program if Debug == 1? */ + int predicate; /* Run this pass? */ + void (*run)(struct radeon_compiler *c, void *user); /* The main entrypoint. */ + void *user; /* Optional parameter which is passed to the run function. */ +}; + +struct rc_program_stats { + unsigned num_insts; + unsigned num_fc_insts; + unsigned num_tex_insts; + unsigned num_rgb_insts; + unsigned num_alpha_insts; + unsigned num_presub_ops; + unsigned num_temp_regs; +}; + +void rc_get_stats(struct radeon_compiler *c, struct rc_program_stats *s); + +/* Executes a list of compiler passes given in the parameter 'list'. */ +void rc_run_compiler_passes(struct radeon_compiler *c, struct radeon_compiler_pass *list); +void rc_run_compiler(struct radeon_compiler *c, struct radeon_compiler_pass *list); +void rc_validate_final_shader(struct radeon_compiler *c, void *user); + +#endif /* RADEON_COMPILER_H */ diff --git a/src/gallium/drivers/r300/compiler/radeon_compiler_util.c b/src/gallium/drivers/r300/compiler/radeon_compiler_util.c new file mode 100644 index 00000000000..2742721f800 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_compiler_util.c @@ -0,0 +1,701 @@ +/* + * Copyright 2010 Tom Stellard <[email protected]> + * + * 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 + */ + +#include "radeon_compiler_util.h" + +#include "radeon_compiler.h" +#include "radeon_dataflow.h" +/** + */ +unsigned int rc_swizzle_to_writemask(unsigned int swz) +{ + unsigned int mask = 0; + unsigned int i; + + for(i = 0; i < 4; i++) { + mask |= 1 << GET_SWZ(swz, i); + } + mask &= RC_MASK_XYZW; + + return mask; +} + +rc_swizzle get_swz(unsigned int swz, rc_swizzle idx) +{ + if (idx & 0x4) + return idx; + return GET_SWZ(swz, idx); +} + +/** + * The purpose of this function is to standardize the number channels used by + * swizzles. All swizzles regardless of what instruction they are a part of + * should have 4 channels initialized with values. + * @param channels The number of channels in initial_value that have a + * meaningful value. + * @return An initialized swizzle that has all of the unused channels set to + * RC_SWIZZLE_UNUSED. + */ +unsigned int rc_init_swizzle(unsigned int initial_value, unsigned int channels) +{ + unsigned int i; + for (i = channels; i < 4; i++) { + SET_SWZ(initial_value, i, RC_SWIZZLE_UNUSED); + } + return initial_value; +} + +unsigned int combine_swizzles4(unsigned int src, + rc_swizzle swz_x, rc_swizzle swz_y, rc_swizzle swz_z, rc_swizzle swz_w) +{ + unsigned int ret = 0; + + ret |= get_swz(src, swz_x); + ret |= get_swz(src, swz_y) << 3; + ret |= get_swz(src, swz_z) << 6; + ret |= get_swz(src, swz_w) << 9; + + return ret; +} + +unsigned int combine_swizzles(unsigned int src, unsigned int swz) +{ + unsigned int ret = 0; + + ret |= get_swz(src, GET_SWZ(swz, RC_SWIZZLE_X)); + ret |= get_swz(src, GET_SWZ(swz, RC_SWIZZLE_Y)) << 3; + ret |= get_swz(src, GET_SWZ(swz, RC_SWIZZLE_Z)) << 6; + ret |= get_swz(src, GET_SWZ(swz, RC_SWIZZLE_W)) << 9; + + return ret; +} + +/** + * @param mask Must be either RC_MASK_X, RC_MASK_Y, RC_MASK_Z, or RC_MASK_W + */ +rc_swizzle rc_mask_to_swizzle(unsigned int mask) +{ + switch (mask) { + case RC_MASK_X: return RC_SWIZZLE_X; + case RC_MASK_Y: return RC_SWIZZLE_Y; + case RC_MASK_Z: return RC_SWIZZLE_Z; + case RC_MASK_W: return RC_SWIZZLE_W; + } + return RC_SWIZZLE_UNUSED; +} + +/* Reorder mask bits according to swizzle. */ +unsigned swizzle_mask(unsigned swizzle, unsigned mask) +{ + unsigned ret = 0; + for (unsigned chan = 0; chan < 4; ++chan) { + unsigned swz = GET_SWZ(swizzle, chan); + if (swz < 4) + ret |= GET_BIT(mask, swz) << chan; + } + return ret; +} + +static unsigned int srcs_need_rewrite(const struct rc_opcode_info * info) +{ + if (info->HasTexture) { + return 0; + } + switch (info->Opcode) { + case RC_OPCODE_DP2: + case RC_OPCODE_DP3: + case RC_OPCODE_DP4: + case RC_OPCODE_DDX: + case RC_OPCODE_DDY: + return 0; + default: + return 1; + } +} + +/** + * @return A swizzle the results from converting old_swizzle using + * conversion_swizzle + */ +unsigned int rc_adjust_channels( + unsigned int old_swizzle, + unsigned int conversion_swizzle) +{ + unsigned int i; + unsigned int new_swizzle = rc_init_swizzle(RC_SWIZZLE_UNUSED, 0); + for (i = 0; i < 4; i++) { + unsigned int new_chan = get_swz(conversion_swizzle, i); + if (new_chan == RC_SWIZZLE_UNUSED) { + continue; + } + SET_SWZ(new_swizzle, new_chan, GET_SWZ(old_swizzle, i)); + } + return new_swizzle; +} + +static unsigned int rewrite_writemask( + unsigned int old_mask, + unsigned int conversion_swizzle) +{ + unsigned int new_mask = 0; + unsigned int i; + + for (i = 0; i < 4; i++) { + if (!GET_BIT(old_mask, i) + || GET_SWZ(conversion_swizzle, i) == RC_SWIZZLE_UNUSED) { + continue; + } + new_mask |= (1 << GET_SWZ(conversion_swizzle, i)); + } + + return new_mask; +} + +/** + * This function rewrites the writemask of sub and adjusts the swizzles + * of all its source registers based on the conversion_swizzle. + * conversion_swizzle represents a mapping of the old writemask to the + * new writemask. For a detailed description of how conversion swizzles + * work see rc_rewrite_swizzle(). + */ +void rc_pair_rewrite_writemask( + struct rc_pair_sub_instruction * sub, + unsigned int conversion_swizzle) +{ + const struct rc_opcode_info * info = rc_get_opcode_info(sub->Opcode); + unsigned int i; + + sub->WriteMask = rewrite_writemask(sub->WriteMask, conversion_swizzle); + + if (!srcs_need_rewrite(info)) { + return ; + } + + for (i = 0; i < info->NumSrcRegs; i++) { + sub->Arg[i].Swizzle = + rc_adjust_channels(sub->Arg[i].Swizzle, + conversion_swizzle); + } +} + +static void normal_rewrite_writemask_cb( + void * userdata, + struct rc_instruction * inst, + struct rc_src_register * src) +{ + unsigned int * new_mask = (unsigned int *)userdata; + src->Swizzle = rc_adjust_channels(src->Swizzle, *new_mask); +} + +/** + * This function is the same as rc_pair_rewrite_writemask() except it + * operates on normal instructions. + */ +void rc_normal_rewrite_writemask( + struct rc_instruction * inst, + unsigned int conversion_swizzle) +{ + unsigned int new_mask; + struct rc_sub_instruction * sub = &inst->U.I; + const struct rc_opcode_info * info = rc_get_opcode_info(sub->Opcode); + sub->DstReg.WriteMask = + rewrite_writemask(sub->DstReg.WriteMask, conversion_swizzle); + + if (info->HasTexture) { + unsigned int i; + assert(sub->TexSwizzle == RC_SWIZZLE_XYZW); + for (i = 0; i < 4; i++) { + unsigned int swz = GET_SWZ(conversion_swizzle, i); + if (swz > 3) + continue; + SET_SWZ(sub->TexSwizzle, swz, i); + } + } + + if (!srcs_need_rewrite(info)) { + return; + } + + new_mask = sub->DstReg.WriteMask; + rc_for_all_reads_src(inst, normal_rewrite_writemask_cb, &new_mask); +} + +/** + * This function replaces each value 'swz' in swizzle with the value of + * GET_SWZ(conversion_swizzle, swz). So, if you want to change all the X's + * in swizzle to Y, then conversion_swizzle should be Y___ (0xff9). If you want + * to change all the Y's in swizzle to X, then conversion_swizzle should be + * _X__ (0xfc7). If you want to change the Y's to X and the X's to Y, then + * conversion swizzle should be YX__ (0xfc1). + * @param swizzle The swizzle to change + * @param conversion_swizzle Describes the conversion to perform on the swizzle + * @return A converted swizzle + */ +unsigned int rc_rewrite_swizzle( + unsigned int swizzle, + unsigned int conversion_swizzle) +{ + unsigned int chan; + unsigned int out_swizzle = swizzle; + + for (chan = 0; chan < 4; chan++) { + unsigned int swz = GET_SWZ(swizzle, chan); + unsigned int new_swz; + if (swz > 3) { + SET_SWZ(out_swizzle, chan, swz); + } else { + new_swz = GET_SWZ(conversion_swizzle, swz); + if (new_swz != RC_SWIZZLE_UNUSED) { + SET_SWZ(out_swizzle, chan, new_swz); + } else { + SET_SWZ(out_swizzle, chan, swz); + } + } + } + return out_swizzle; +} + +/** + * Left multiplication of a register with a swizzle + */ +struct rc_src_register lmul_swizzle(unsigned int swizzle, struct rc_src_register srcreg) +{ + struct rc_src_register tmp = srcreg; + int i; + tmp.Swizzle = 0; + tmp.Negate = 0; + for(i = 0; i < 4; ++i) { + rc_swizzle swz = GET_SWZ(swizzle, i); + if (swz < 4) { + tmp.Swizzle |= GET_SWZ(srcreg.Swizzle, swz) << (i*3); + tmp.Negate |= GET_BIT(srcreg.Negate, swz) << i; + } else { + tmp.Swizzle |= swz << (i*3); + } + } + return tmp; +} + +void reset_srcreg(struct rc_src_register* reg) +{ + memset(reg, 0, sizeof(struct rc_src_register)); + reg->Swizzle = RC_SWIZZLE_XYZW; +} + +unsigned int rc_src_reads_dst_mask( + rc_register_file src_file, + unsigned int src_idx, + unsigned int src_swz, + rc_register_file dst_file, + unsigned int dst_idx, + unsigned int dst_mask) +{ + if (src_file != dst_file || src_idx != dst_idx) { + return RC_MASK_NONE; + } + return dst_mask & rc_swizzle_to_writemask(src_swz); +} + +/** + * @return A bit mask specifying whether this swizzle will select from an RGB + * source, an Alpha source, or both. + */ +unsigned int rc_source_type_swz(unsigned int swizzle) +{ + unsigned int chan; + unsigned int swz = RC_SWIZZLE_UNUSED; + unsigned int ret = RC_SOURCE_NONE; + + for(chan = 0; chan < 4; chan++) { + swz = GET_SWZ(swizzle, chan); + if (swz == RC_SWIZZLE_W) { + ret |= RC_SOURCE_ALPHA; + } else if (swz == RC_SWIZZLE_X || swz == RC_SWIZZLE_Y + || swz == RC_SWIZZLE_Z) { + ret |= RC_SOURCE_RGB; + } + } + return ret; +} + +unsigned int rc_source_type_mask(unsigned int mask) +{ + unsigned int ret = RC_SOURCE_NONE; + + if (mask & RC_MASK_XYZ) + ret |= RC_SOURCE_RGB; + + if (mask & RC_MASK_W) + ret |= RC_SOURCE_ALPHA; + + return ret; +} + +struct src_select { + rc_register_file File; + int Index; + unsigned int SrcType; +}; + +struct can_use_presub_data { + struct src_select Selects[5]; + unsigned int SelectCount; + const struct rc_src_register * ReplaceReg; + unsigned int ReplaceRemoved; +}; + +static void can_use_presub_data_add_select( + struct can_use_presub_data * data, + rc_register_file file, + unsigned int index, + unsigned int src_type) +{ + struct src_select * select; + + select = &data->Selects[data->SelectCount++]; + select->File = file; + select->Index = index; + select->SrcType = src_type; +} + +/** + * This callback function counts the number of sources in inst that are + * different from the sources in can_use_presub_data->RemoveSrcs. + */ +static void can_use_presub_read_cb( + void * userdata, + struct rc_instruction * inst, + struct rc_src_register * src) +{ + struct can_use_presub_data * d = userdata; + + if (!d->ReplaceRemoved && src == d->ReplaceReg) { + d->ReplaceRemoved = 1; + return; + } + + if (src->File == RC_FILE_NONE) + return; + + can_use_presub_data_add_select(d, src->File, src->Index, + rc_source_type_swz(src->Swizzle)); +} + +unsigned int rc_inst_can_use_presub( + struct rc_instruction * inst, + rc_presubtract_op presub_op, + unsigned int presub_writemask, + const struct rc_src_register * replace_reg, + const struct rc_src_register * presub_src0, + const struct rc_src_register * presub_src1) +{ + struct can_use_presub_data d; + unsigned int num_presub_srcs; + unsigned int i; + const struct rc_opcode_info * info = + rc_get_opcode_info(inst->U.I.Opcode); + int rgb_count = 0, alpha_count = 0; + unsigned int src_type0, src_type1; + + if (presub_op == RC_PRESUB_NONE) { + return 1; + } + + if (info->HasTexture) { + return 0; + } + + /* We can't use more than one presubtract value in an + * instruction, unless the two prsubtract operations + * are the same and read from the same registers. + * XXX For now we will limit instructions to only one presubtract + * value.*/ + if (inst->U.I.PreSub.Opcode != RC_PRESUB_NONE) { + return 0; + } + + memset(&d, 0, sizeof(d)); + d.ReplaceReg = replace_reg; + + rc_for_all_reads_src(inst, can_use_presub_read_cb, &d); + + num_presub_srcs = rc_presubtract_src_reg_count(presub_op); + + src_type0 = rc_source_type_swz(presub_src0->Swizzle); + can_use_presub_data_add_select(&d, + presub_src0->File, + presub_src0->Index, + src_type0); + + if (num_presub_srcs > 1) { + src_type1 = rc_source_type_swz(presub_src1->Swizzle); + can_use_presub_data_add_select(&d, + presub_src1->File, + presub_src1->Index, + src_type1); + + /* Even if both of the presub sources read from the same + * register, we still need to use 2 different source selects + * for them, so we need to increment the count to compensate. + */ + if (presub_src0->File == presub_src1->File + && presub_src0->Index == presub_src1->Index) { + if (src_type0 & src_type1 & RC_SOURCE_RGB) { + rgb_count++; + } + if (src_type0 & src_type1 & RC_SOURCE_ALPHA) { + alpha_count++; + } + } + } + + /* Count the number of source selects for Alpha and RGB. If we + * encounter two of the same source selects then we can ignore the + * first one. */ + for (i = 0; i < d.SelectCount; i++) { + unsigned int j; + unsigned int src_type = d.Selects[i].SrcType; + for (j = i + 1; j < d.SelectCount; j++) { + if (d.Selects[i].File == d.Selects[j].File + && d.Selects[i].Index == d.Selects[j].Index) { + src_type &= ~d.Selects[j].SrcType; + } + } + if (src_type & RC_SOURCE_RGB) { + rgb_count++; + } + + if (src_type & RC_SOURCE_ALPHA) { + alpha_count++; + } + } + + if (rgb_count > 3 || alpha_count > 3) { + return 0; + } + + return 1; +} + +struct max_data { + unsigned int Max; + unsigned int HasFileType; + rc_register_file File; +}; + +static void max_callback( + void * userdata, + struct rc_instruction * inst, + rc_register_file file, + unsigned int index, + unsigned int mask) +{ + struct max_data * d = (struct max_data*)userdata; + if (file == d->File && (!d->HasFileType || index > d->Max)) { + d->Max = index; + d->HasFileType = 1; + } +} + +/** + * @return The maximum index of the specified register file used by the + * program. + */ +int rc_get_max_index( + struct radeon_compiler * c, + rc_register_file file) +{ + struct max_data data; + struct rc_instruction * inst; + data.Max = 0; + data.HasFileType = 0; + data.File = file; + for (inst = c->Program.Instructions.Next; + inst != &c->Program.Instructions; + inst = inst->Next) { + rc_for_all_reads_mask(inst, max_callback, &data); + rc_for_all_writes_mask(inst, max_callback, &data); + } + if (!data.HasFileType) { + return -1; + } else { + return data.Max; + } +} + +static unsigned int get_source_readmask( + struct rc_pair_sub_instruction * sub, + unsigned int source, + unsigned int src_type) +{ + unsigned int i; + unsigned int readmask = 0; + const struct rc_opcode_info * info = rc_get_opcode_info(sub->Opcode); + + for (i = 0; i < info->NumSrcRegs; i++) { + if (sub->Arg[i].Source != source + || src_type != rc_source_type_swz(sub->Arg[i].Swizzle)) { + continue; + } + readmask |= rc_swizzle_to_writemask(sub->Arg[i].Swizzle); + } + return readmask; +} + +/** + * This function attempts to remove a source from a pair instructions. + * @param inst + * @param src_type RC_SOURCE_RGB, RC_SOURCE_ALPHA, or both bitwise or'd + * @param source The index of the source to remove + * @param new_readmask A mask representing the components that are read by + * the source that is intended to replace the one you are removing. If you + * want to remove a source only and not replace it, this parameter should be + * zero. + * @return 1 if the source was successfully removed, 0 if it was not + */ +unsigned int rc_pair_remove_src( + struct rc_instruction * inst, + unsigned int src_type, + unsigned int source, + unsigned int new_readmask) +{ + unsigned int readmask = 0; + + readmask |= get_source_readmask(&inst->U.P.RGB, source, src_type); + readmask |= get_source_readmask(&inst->U.P.Alpha, source, src_type); + + if ((new_readmask & readmask) != readmask) + return 0; + + if (src_type & RC_SOURCE_RGB) { + memset(&inst->U.P.RGB.Src[source], 0, + sizeof(struct rc_pair_instruction_source)); + } + + if (src_type & RC_SOURCE_ALPHA) { + memset(&inst->U.P.Alpha.Src[source], 0, + sizeof(struct rc_pair_instruction_source)); + } + + return 1; +} + +/** + * @return RC_OPCODE_NOOP if inst is not a flow control instruction. + * @return The opcode of inst if it is a flow control instruction. + */ +rc_opcode rc_get_flow_control_inst(struct rc_instruction * inst) +{ + const struct rc_opcode_info * info; + if (inst->Type == RC_INSTRUCTION_NORMAL) { + info = rc_get_opcode_info(inst->U.I.Opcode); + } else { + info = rc_get_opcode_info(inst->U.P.RGB.Opcode); + /*A flow control instruction shouldn't have an alpha + * instruction.*/ + assert(!info->IsFlowControl || + inst->U.P.Alpha.Opcode == RC_OPCODE_NOP); + } + + if (info->IsFlowControl) + return info->Opcode; + else + return RC_OPCODE_NOP; + +} + +/** + * @return The BGNLOOP instruction that starts the loop ended by endloop. + */ +struct rc_instruction * rc_match_endloop(struct rc_instruction * endloop) +{ + unsigned int endloop_count = 0; + struct rc_instruction * inst; + for (inst = endloop->Prev; inst != endloop; inst = inst->Prev) { + rc_opcode op = rc_get_flow_control_inst(inst); + if (op == RC_OPCODE_ENDLOOP) { + endloop_count++; + } else if (op == RC_OPCODE_BGNLOOP) { + if (endloop_count == 0) { + return inst; + } else { + endloop_count--; + } + } + } + return NULL; +} + +/** + * @return The ENDLOOP instruction that ends the loop started by bgnloop. + */ +struct rc_instruction * rc_match_bgnloop(struct rc_instruction * bgnloop) +{ + unsigned int bgnloop_count = 0; + struct rc_instruction * inst; + for (inst = bgnloop->Next; inst!=bgnloop; inst = inst->Next) { + rc_opcode op = rc_get_flow_control_inst(inst); + if (op == RC_OPCODE_BGNLOOP) { + bgnloop_count++; + } else if (op == RC_OPCODE_ENDLOOP) { + if (bgnloop_count == 0) { + return inst; + } else { + bgnloop_count--; + } + } + } + return NULL; +} + +/** + * @return A conversion swizzle for converting from old_mask->new_mask + */ +unsigned int rc_make_conversion_swizzle( + unsigned int old_mask, + unsigned int new_mask) +{ + unsigned int conversion_swizzle = rc_init_swizzle(RC_SWIZZLE_UNUSED, 0); + unsigned int old_idx; + unsigned int new_idx = 0; + for (old_idx = 0; old_idx < 4; old_idx++) { + if (!GET_BIT(old_mask, old_idx)) + continue; + for ( ; new_idx < 4; new_idx++) { + if (GET_BIT(new_mask, new_idx)) { + SET_SWZ(conversion_swizzle, old_idx, new_idx); + new_idx++; + break; + } + } + } + return conversion_swizzle; +} diff --git a/src/gallium/drivers/r300/compiler/radeon_compiler_util.h b/src/gallium/drivers/r300/compiler/radeon_compiler_util.h new file mode 100644 index 00000000000..3730aa888c0 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_compiler_util.h @@ -0,0 +1,89 @@ +#include "radeon_program_constants.h" + +#ifndef RADEON_PROGRAM_UTIL_H +#define RADEON_PROGRAM_UTIL_H + +#include "radeon_opcodes.h" + +struct radeon_compiler; +struct rc_instruction; +struct rc_pair_instruction; +struct rc_pair_sub_instruction; +struct rc_src_register; + +unsigned int rc_swizzle_to_writemask(unsigned int swz); + +rc_swizzle get_swz(unsigned int swz, rc_swizzle idx); + +unsigned int rc_init_swizzle(unsigned int initial_value, unsigned int channels); + +unsigned int combine_swizzles4(unsigned int src, + rc_swizzle swz_x, rc_swizzle swz_y, + rc_swizzle swz_z, rc_swizzle swz_w); + +unsigned int combine_swizzles(unsigned int src, unsigned int swz); + +rc_swizzle rc_mask_to_swizzle(unsigned int mask); + +unsigned swizzle_mask(unsigned swizzle, unsigned mask); + +unsigned int rc_adjust_channels( + unsigned int old_swizzle, + unsigned int conversion_swizzle); + +void rc_pair_rewrite_writemask( + struct rc_pair_sub_instruction * sub, + unsigned int conversion_swizzle); + +void rc_normal_rewrite_writemask( + struct rc_instruction * inst, + unsigned int conversion_swizzle); + +unsigned int rc_rewrite_swizzle( + unsigned int swizzle, + unsigned int new_mask); + +struct rc_src_register lmul_swizzle(unsigned int swizzle, struct rc_src_register srcreg); + +void reset_srcreg(struct rc_src_register* reg); + +unsigned int rc_src_reads_dst_mask( + rc_register_file src_file, + unsigned int src_idx, + unsigned int src_swz, + rc_register_file dst_file, + unsigned int dst_idx, + unsigned int dst_mask); + +unsigned int rc_source_type_swz(unsigned int swizzle); + +unsigned int rc_source_type_mask(unsigned int mask); + +unsigned int rc_inst_can_use_presub( + struct rc_instruction * inst, + rc_presubtract_op presub_op, + unsigned int presub_writemask, + const struct rc_src_register * replace_reg, + const struct rc_src_register * presub_src0, + const struct rc_src_register * presub_src1); + +int rc_get_max_index( + struct radeon_compiler * c, + rc_register_file file); + +unsigned int rc_pair_remove_src( + struct rc_instruction * inst, + unsigned int src_type, + unsigned int source, + unsigned int new_readmask); + +rc_opcode rc_get_flow_control_inst(struct rc_instruction * inst); + +struct rc_instruction * rc_match_endloop(struct rc_instruction * endloop); +struct rc_instruction * rc_match_bgnloop(struct rc_instruction * bgnloop); + +unsigned int rc_make_conversion_swizzle( + unsigned int old_mask, + unsigned int new_mask); + +#endif /* RADEON_PROGRAM_UTIL_H */ diff --git a/src/gallium/drivers/r300/compiler/radeon_dataflow.c b/src/gallium/drivers/r300/compiler/radeon_dataflow.c new file mode 100644 index 00000000000..a8decacedaf --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_dataflow.c @@ -0,0 +1,892 @@ +/* + * Copyright (C) 2009 Nicolai Haehnle. + * Copyright 2010 Tom Stellard <[email protected]> + * + * 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. + * + */ + +#include "radeon_dataflow.h" + +#include "radeon_compiler.h" +#include "radeon_compiler_util.h" +#include "radeon_program.h" + +struct read_write_mask_data { + void * UserData; + rc_read_write_mask_fn Cb; +}; + +static void reads_normal_callback( + void * userdata, + struct rc_instruction * fullinst, + struct rc_src_register * src) +{ + struct read_write_mask_data * cb_data = userdata; + unsigned int refmask = 0; + unsigned int chan; + for(chan = 0; chan < 4; chan++) { + refmask |= 1 << GET_SWZ(src->Swizzle, chan); + } + refmask &= RC_MASK_XYZW; + + if (refmask) { + cb_data->Cb(cb_data->UserData, fullinst, src->File, + src->Index, refmask); + } + + if (refmask && src->RelAddr) { + cb_data->Cb(cb_data->UserData, fullinst, RC_FILE_ADDRESS, 0, + RC_MASK_X); + } +} + +static void pair_get_src_refmasks(unsigned int * refmasks, + struct rc_pair_instruction * inst, + unsigned int swz, unsigned int src) +{ + if (swz >= 4) + return; + + if (swz == RC_SWIZZLE_X || swz == RC_SWIZZLE_Y || swz == RC_SWIZZLE_Z) { + if(src == RC_PAIR_PRESUB_SRC) { + unsigned int i; + int srcp_regs = + rc_presubtract_src_reg_count( + inst->RGB.Src[src].Index); + for(i = 0; i < srcp_regs; i++) { + refmasks[i] |= 1 << swz; + } + } + else { + refmasks[src] |= 1 << swz; + } + } + + if (swz == RC_SWIZZLE_W) { + if (src == RC_PAIR_PRESUB_SRC) { + unsigned int i; + int srcp_regs = rc_presubtract_src_reg_count( + inst->Alpha.Src[src].Index); + for(i = 0; i < srcp_regs; i++) { + refmasks[i] |= 1 << swz; + } + } + else { + refmasks[src] |= 1 << swz; + } + } +} + +static void reads_pair(struct rc_instruction * fullinst, rc_read_write_mask_fn cb, void * userdata) +{ + struct rc_pair_instruction * inst = &fullinst->U.P; + unsigned int refmasks[3] = { 0, 0, 0 }; + + unsigned int arg; + + for(arg = 0; arg < 3; ++arg) { + unsigned int chan; + for(chan = 0; chan < 3; ++chan) { + unsigned int swz_rgb = + GET_SWZ(inst->RGB.Arg[arg].Swizzle, chan); + unsigned int swz_alpha = + GET_SWZ(inst->Alpha.Arg[arg].Swizzle, chan); + pair_get_src_refmasks(refmasks, inst, swz_rgb, + inst->RGB.Arg[arg].Source); + pair_get_src_refmasks(refmasks, inst, swz_alpha, + inst->Alpha.Arg[arg].Source); + } + } + + for(unsigned int src = 0; src < 3; ++src) { + if (inst->RGB.Src[src].Used && (refmasks[src] & RC_MASK_XYZ)) + cb(userdata, fullinst, inst->RGB.Src[src].File, inst->RGB.Src[src].Index, + refmasks[src] & RC_MASK_XYZ); + + if (inst->Alpha.Src[src].Used && (refmasks[src] & RC_MASK_W)) + cb(userdata, fullinst, inst->Alpha.Src[src].File, inst->Alpha.Src[src].Index, RC_MASK_W); + } +} + +static void pair_sub_for_all_args( + struct rc_instruction * fullinst, + struct rc_pair_sub_instruction * sub, + rc_pair_read_arg_fn cb, + void * userdata) +{ + int i; + const struct rc_opcode_info * info = rc_get_opcode_info(sub->Opcode); + + for(i = 0; i < info->NumSrcRegs; i++) { + unsigned int src_type; + + src_type = rc_source_type_swz(sub->Arg[i].Swizzle); + + if (src_type == RC_SOURCE_NONE) + continue; + + if (sub->Arg[i].Source == RC_PAIR_PRESUB_SRC) { + unsigned int presub_type; + unsigned int presub_src_count; + struct rc_pair_instruction_source * src_array; + unsigned int j; + + if (src_type & RC_SOURCE_RGB) { + presub_type = fullinst-> + U.P.RGB.Src[RC_PAIR_PRESUB_SRC].Index; + src_array = fullinst->U.P.RGB.Src; + } else { + presub_type = fullinst-> + U.P.Alpha.Src[RC_PAIR_PRESUB_SRC].Index; + src_array = fullinst->U.P.Alpha.Src; + } + presub_src_count + = rc_presubtract_src_reg_count(presub_type); + for(j = 0; j < presub_src_count; j++) { + cb(userdata, fullinst, &sub->Arg[i], + &src_array[j]); + } + } else { + struct rc_pair_instruction_source * src = + rc_pair_get_src(&fullinst->U.P, &sub->Arg[i]); + if (src) { + cb(userdata, fullinst, &sub->Arg[i], src); + } + } + } +} + +/* This function calls the callback function (cb) for each source used by + * the instruction. + * */ +void rc_for_all_reads_src( + struct rc_instruction * inst, + rc_read_src_fn cb, + void * userdata) +{ + const struct rc_opcode_info * opcode = + rc_get_opcode_info(inst->U.I.Opcode); + + /* This function only works with normal instructions. */ + if (inst->Type != RC_INSTRUCTION_NORMAL) { + assert(0); + return; + } + + for(unsigned int src = 0; src < opcode->NumSrcRegs; ++src) { + + if (inst->U.I.SrcReg[src].File == RC_FILE_NONE) + continue; + + if (inst->U.I.SrcReg[src].File == RC_FILE_PRESUB) { + unsigned int i; + unsigned int srcp_regs = rc_presubtract_src_reg_count( + inst->U.I.PreSub.Opcode); + for( i = 0; i < srcp_regs; i++) { + cb(userdata, inst, &inst->U.I.PreSub.SrcReg[i]); + } + } else { + cb(userdata, inst, &inst->U.I.SrcReg[src]); + } + } +} + +/** + * This function calls the callback function (cb) for each arg of the RGB and + * alpha components. + */ +void rc_pair_for_all_reads_arg( + struct rc_instruction * inst, + rc_pair_read_arg_fn cb, + void * userdata) +{ + /* This function only works with pair instructions. */ + if (inst->Type != RC_INSTRUCTION_PAIR) { + assert(0); + return; + } + + pair_sub_for_all_args(inst, &inst->U.P.RGB, cb, userdata); + pair_sub_for_all_args(inst, &inst->U.P.Alpha, cb, userdata); +} + +/** + * Calls a callback function for all register reads. + * + * This is conservative, i.e. if the same register is referenced multiple times, + * the callback may also be called multiple times. + * Also, the writemask of the instruction is not taken into account. + */ +void rc_for_all_reads_mask(struct rc_instruction * inst, rc_read_write_mask_fn cb, void * userdata) +{ + if (inst->Type == RC_INSTRUCTION_NORMAL) { + struct read_write_mask_data cb_data; + cb_data.UserData = userdata; + cb_data.Cb = cb; + + rc_for_all_reads_src(inst, reads_normal_callback, &cb_data); + } else { + reads_pair(inst, cb, userdata); + } +} + + + +static void writes_normal(struct rc_instruction * fullinst, rc_read_write_mask_fn cb, void * userdata) +{ + struct rc_sub_instruction * inst = &fullinst->U.I; + const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->Opcode); + + if (opcode->HasDstReg && inst->DstReg.WriteMask) + cb(userdata, fullinst, inst->DstReg.File, inst->DstReg.Index, inst->DstReg.WriteMask); + + if (inst->WriteALUResult) + cb(userdata, fullinst, RC_FILE_SPECIAL, RC_SPECIAL_ALU_RESULT, RC_MASK_X); +} + +static void writes_pair(struct rc_instruction * fullinst, rc_read_write_mask_fn cb, void * userdata) +{ + struct rc_pair_instruction * inst = &fullinst->U.P; + + if (inst->RGB.WriteMask) + cb(userdata, fullinst, RC_FILE_TEMPORARY, inst->RGB.DestIndex, inst->RGB.WriteMask); + + if (inst->Alpha.WriteMask) + cb(userdata, fullinst, RC_FILE_TEMPORARY, inst->Alpha.DestIndex, RC_MASK_W); + + if (inst->WriteALUResult) + cb(userdata, fullinst, RC_FILE_SPECIAL, RC_SPECIAL_ALU_RESULT, RC_MASK_X); +} + +/** + * Calls a callback function for all register writes in the instruction, + * reporting writemasks to the callback function. + * + * \warning Does not report output registers for paired instructions! + */ +void rc_for_all_writes_mask(struct rc_instruction * inst, rc_read_write_mask_fn cb, void * userdata) +{ + if (inst->Type == RC_INSTRUCTION_NORMAL) { + writes_normal(inst, cb, userdata); + } else { + writes_pair(inst, cb, userdata); + } +} + + +struct mask_to_chan_data { + void * UserData; + rc_read_write_chan_fn Fn; +}; + +static void mask_to_chan_cb(void * data, struct rc_instruction * inst, + rc_register_file file, unsigned int index, unsigned int mask) +{ + struct mask_to_chan_data * d = data; + for(unsigned int chan = 0; chan < 4; ++chan) { + if (GET_BIT(mask, chan)) + d->Fn(d->UserData, inst, file, index, chan); + } +} + +/** + * Calls a callback function for all sourced register channels. + * + * This is conservative, i.e. channels may be called multiple times, + * and the writemask of the instruction is not taken into account. + */ +void rc_for_all_reads_chan(struct rc_instruction * inst, rc_read_write_chan_fn cb, void * userdata) +{ + struct mask_to_chan_data d; + d.UserData = userdata; + d.Fn = cb; + rc_for_all_reads_mask(inst, &mask_to_chan_cb, &d); +} + +/** + * Calls a callback function for all written register channels. + * + * \warning Does not report output registers for paired instructions! + */ +void rc_for_all_writes_chan(struct rc_instruction * inst, rc_read_write_chan_fn cb, void * userdata) +{ + struct mask_to_chan_data d; + d.UserData = userdata; + d.Fn = cb; + rc_for_all_writes_mask(inst, &mask_to_chan_cb, &d); +} + +static void remap_normal_instruction(struct rc_instruction * fullinst, + rc_remap_register_fn cb, void * userdata) +{ + struct rc_sub_instruction * inst = &fullinst->U.I; + const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->Opcode); + unsigned int remapped_presub = 0; + + if (opcode->HasDstReg) { + rc_register_file file = inst->DstReg.File; + unsigned int index = inst->DstReg.Index; + + cb(userdata, fullinst, &file, &index); + + inst->DstReg.File = file; + inst->DstReg.Index = index; + } + + for(unsigned int src = 0; src < opcode->NumSrcRegs; ++src) { + rc_register_file file = inst->SrcReg[src].File; + unsigned int index = inst->SrcReg[src].Index; + + if (file == RC_FILE_PRESUB) { + unsigned int i; + unsigned int srcp_srcs = rc_presubtract_src_reg_count( + inst->PreSub.Opcode); + /* Make sure we only remap presubtract sources once in + * case more than one source register reads the + * presubtract result. */ + if (remapped_presub) + continue; + + for(i = 0; i < srcp_srcs; i++) { + file = inst->PreSub.SrcReg[i].File; + index = inst->PreSub.SrcReg[i].Index; + cb(userdata, fullinst, &file, &index); + inst->PreSub.SrcReg[i].File = file; + inst->PreSub.SrcReg[i].Index = index; + } + remapped_presub = 1; + } + else { + cb(userdata, fullinst, &file, &index); + + inst->SrcReg[src].File = file; + inst->SrcReg[src].Index = index; + } + } +} + +static void remap_pair_instruction(struct rc_instruction * fullinst, + rc_remap_register_fn cb, void * userdata) +{ + struct rc_pair_instruction * inst = &fullinst->U.P; + + if (inst->RGB.WriteMask) { + rc_register_file file = RC_FILE_TEMPORARY; + unsigned int index = inst->RGB.DestIndex; + + cb(userdata, fullinst, &file, &index); + + inst->RGB.DestIndex = index; + } + + if (inst->Alpha.WriteMask) { + rc_register_file file = RC_FILE_TEMPORARY; + unsigned int index = inst->Alpha.DestIndex; + + cb(userdata, fullinst, &file, &index); + + inst->Alpha.DestIndex = index; + } + + for(unsigned int src = 0; src < 3; ++src) { + if (inst->RGB.Src[src].Used) { + rc_register_file file = inst->RGB.Src[src].File; + unsigned int index = inst->RGB.Src[src].Index; + + cb(userdata, fullinst, &file, &index); + + inst->RGB.Src[src].File = file; + inst->RGB.Src[src].Index = index; + } + + if (inst->Alpha.Src[src].Used) { + rc_register_file file = inst->Alpha.Src[src].File; + unsigned int index = inst->Alpha.Src[src].Index; + + cb(userdata, fullinst, &file, &index); + + inst->Alpha.Src[src].File = file; + inst->Alpha.Src[src].Index = index; + } + } +} + + +/** + * Remap all register accesses according to the given function. + * That is, call the function \p cb for each referenced register (both read and written) + * and update the given instruction \p inst accordingly + * if it modifies its \ref pfile and \ref pindex contents. + */ +void rc_remap_registers(struct rc_instruction * inst, rc_remap_register_fn cb, void * userdata) +{ + if (inst->Type == RC_INSTRUCTION_NORMAL) + remap_normal_instruction(inst, cb, userdata); + else + remap_pair_instruction(inst, cb, userdata); +} + +struct branch_write_mask { + unsigned int IfWriteMask:4; + unsigned int ElseWriteMask:4; + unsigned int HasElse:1; +}; + +union get_readers_read_cb { + rc_read_src_fn I; + rc_pair_read_arg_fn P; +}; + +struct get_readers_callback_data { + struct radeon_compiler * C; + struct rc_reader_data * ReaderData; + rc_read_src_fn ReadNormalCB; + rc_pair_read_arg_fn ReadPairCB; + rc_read_write_mask_fn WriteCB; + rc_register_file DstFile; + unsigned int DstIndex; + unsigned int DstMask; + unsigned int AliveWriteMask; + /* For convenience, this is indexed starting at 1 */ + struct branch_write_mask BranchMasks[R500_PFS_MAX_BRANCH_DEPTH_FULL + 1]; +}; + +static struct rc_reader * add_reader( + struct memory_pool * pool, + struct rc_reader_data * data, + struct rc_instruction * inst, + unsigned int mask) +{ + struct rc_reader * new; + memory_pool_array_reserve(pool, struct rc_reader, data->Readers, + data->ReaderCount, data->ReadersReserved, 1); + new = &data->Readers[data->ReaderCount++]; + new->Inst = inst; + new->WriteMask = mask; + return new; +} + +static void add_reader_normal( + struct memory_pool * pool, + struct rc_reader_data * data, + struct rc_instruction * inst, + unsigned int mask, + struct rc_src_register * src) +{ + struct rc_reader * new = add_reader(pool, data, inst, mask); + new->U.I.Src = src; +} + + +static void add_reader_pair( + struct memory_pool * pool, + struct rc_reader_data * data, + struct rc_instruction * inst, + unsigned int mask, + struct rc_pair_instruction_arg * arg, + struct rc_pair_instruction_source * src) +{ + struct rc_reader * new = add_reader(pool, data, inst, mask); + new->U.P.Src = src; + new->U.P.Arg = arg; +} + +static unsigned int get_readers_read_callback( + struct get_readers_callback_data * cb_data, + unsigned int has_rel_addr, + rc_register_file file, + unsigned int index, + unsigned int swizzle) +{ + unsigned int shared_mask, read_mask; + + if (has_rel_addr) { + cb_data->ReaderData->Abort = 1; + return RC_MASK_NONE; + } + + shared_mask = rc_src_reads_dst_mask(file, index, swizzle, + cb_data->DstFile, cb_data->DstIndex, cb_data->AliveWriteMask); + + if (shared_mask == RC_MASK_NONE) + return shared_mask; + + /* If we make it this far, it means that this source reads from the + * same register written to by d->ReaderData->Writer. */ + + read_mask = rc_swizzle_to_writemask(swizzle); + if (cb_data->ReaderData->AbortOnRead & read_mask) { + cb_data->ReaderData->Abort = 1; + return shared_mask; + } + + if (cb_data->ReaderData->LoopDepth > 0) { + cb_data->ReaderData->AbortOnWrite |= + (read_mask & cb_data->AliveWriteMask); + } + + /* XXX The behavior in this case should be configurable. */ + if ((read_mask & cb_data->AliveWriteMask) != read_mask) { + cb_data->ReaderData->Abort = 1; + return shared_mask; + } + + return shared_mask; +} + +static void get_readers_pair_read_callback( + void * userdata, + struct rc_instruction * inst, + struct rc_pair_instruction_arg * arg, + struct rc_pair_instruction_source * src) +{ + unsigned int shared_mask; + struct get_readers_callback_data * d = userdata; + + shared_mask = get_readers_read_callback(d, + 0 /*Pair Instructions don't use RelAddr*/, + src->File, src->Index, arg->Swizzle); + + if (shared_mask == RC_MASK_NONE) + return; + + if (d->ReadPairCB) + d->ReadPairCB(d->ReaderData, inst, arg, src); + + if (d->ReaderData->ExitOnAbort && d->ReaderData->Abort) + return; + + add_reader_pair(&d->C->Pool, d->ReaderData, inst, shared_mask, arg, src); +} + +/** + * This function is used by rc_get_readers_normal() to determine whether inst + * is a reader of userdata->ReaderData->Writer + */ +static void get_readers_normal_read_callback( + void * userdata, + struct rc_instruction * inst, + struct rc_src_register * src) +{ + struct get_readers_callback_data * d = userdata; + unsigned int shared_mask; + + shared_mask = get_readers_read_callback(d, + src->RelAddr, src->File, src->Index, src->Swizzle); + + if (shared_mask == RC_MASK_NONE) + return; + /* The callback function could potentially clear d->ReaderData->Abort, + * so we need to call it before we return. */ + if (d->ReadNormalCB) + d->ReadNormalCB(d->ReaderData, inst, src); + + if (d->ReaderData->ExitOnAbort && d->ReaderData->Abort) + return; + + add_reader_normal(&d->C->Pool, d->ReaderData, inst, shared_mask, src); +} + +/** + * This function is used by rc_get_readers_normal() to determine when + * userdata->ReaderData->Writer is dead (i. e. All compontents of its + * destination register have been overwritten by other instructions). + */ +static void get_readers_write_callback( + void *userdata, + struct rc_instruction * inst, + rc_register_file file, + unsigned int index, + unsigned int mask) +{ + struct get_readers_callback_data * d = userdata; + + if (index == d->DstIndex && file == d->DstFile) { + unsigned int shared_mask = mask & d->DstMask; + d->ReaderData->AbortOnRead &= ~shared_mask; + d->AliveWriteMask &= ~shared_mask; + if (d->ReaderData->AbortOnWrite & shared_mask) { + d->ReaderData->Abort = 1; + } + } + + if(d->WriteCB) + d->WriteCB(d->ReaderData, inst, file, index, mask); +} + +static void push_branch_mask( + struct get_readers_callback_data * d, + unsigned int * branch_depth) +{ + (*branch_depth)++; + if (*branch_depth > R500_PFS_MAX_BRANCH_DEPTH_FULL) { + d->ReaderData->Abort = 1; + return; + } + d->BranchMasks[*branch_depth].IfWriteMask = + d->AliveWriteMask; +} + +static void pop_branch_mask( + struct get_readers_callback_data * d, + unsigned int * branch_depth) +{ + struct branch_write_mask * masks = &d->BranchMasks[*branch_depth]; + + if (masks->HasElse) { + /* Abort on read for components that were written in the IF + * block. */ + d->ReaderData->AbortOnRead |= + masks->IfWriteMask & ~masks->ElseWriteMask; + /* Abort on read for components that were written in the ELSE + * block. */ + d->ReaderData->AbortOnRead |= + masks->ElseWriteMask & ~d->AliveWriteMask; + + d->AliveWriteMask = masks->IfWriteMask + ^ ((masks->IfWriteMask ^ masks->ElseWriteMask) + & (masks->IfWriteMask ^ d->AliveWriteMask)); + } else { + d->ReaderData->AbortOnRead |= + masks->IfWriteMask & ~d->AliveWriteMask; + d->AliveWriteMask = masks->IfWriteMask; + + } + memset(masks, 0, sizeof(struct branch_write_mask)); + (*branch_depth)--; +} + +static void get_readers_for_single_write( + void * userdata, + struct rc_instruction * writer, + rc_register_file dst_file, + unsigned int dst_index, + unsigned int dst_mask) +{ + struct rc_instruction * tmp; + unsigned int branch_depth = 0; + struct rc_instruction * endloop = NULL; + unsigned int abort_on_read_at_endloop = 0; + struct get_readers_callback_data * d = userdata; + + d->ReaderData->Writer = writer; + d->ReaderData->AbortOnRead = 0; + d->ReaderData->AbortOnWrite = 0; + d->ReaderData->LoopDepth = 0; + d->ReaderData->InElse = 0; + d->DstFile = dst_file; + d->DstIndex = dst_index; + d->DstMask = dst_mask; + d->AliveWriteMask = dst_mask; + memset(d->BranchMasks, 0, sizeof(d->BranchMasks)); + + if (!dst_mask) + return; + + for(tmp = writer->Next; tmp != &d->C->Program.Instructions; + tmp = tmp->Next){ + rc_opcode opcode = rc_get_flow_control_inst(tmp); + switch(opcode) { + case RC_OPCODE_BGNLOOP: + d->ReaderData->LoopDepth++; + push_branch_mask(d, &branch_depth); + break; + case RC_OPCODE_ENDLOOP: + if (d->ReaderData->LoopDepth > 0) { + d->ReaderData->LoopDepth--; + if (d->ReaderData->LoopDepth == 0) { + d->ReaderData->AbortOnWrite = 0; + } + pop_branch_mask(d, &branch_depth); + } else { + /* Here we have reached an ENDLOOP without + * seeing its BGNLOOP. These means that + * the writer was written inside of a loop, + * so it could have readers that are above it + * (i.e. they have a lower IP). To find these + * readers we jump to the BGNLOOP instruction + * and check each instruction until we get + * back to the writer. + */ + endloop = tmp; + tmp = rc_match_endloop(tmp); + if (!tmp) { + rc_error(d->C, "Failed to match endloop.\n"); + d->ReaderData->Abort = 1; + return; + } + abort_on_read_at_endloop = d->ReaderData->AbortOnRead; + d->ReaderData->AbortOnRead |= d->AliveWriteMask; + continue; + } + break; + case RC_OPCODE_IF: + push_branch_mask(d, &branch_depth); + break; + case RC_OPCODE_ELSE: + if (branch_depth == 0) { + d->ReaderData->InElse = 1; + } else { + unsigned int temp_mask = d->AliveWriteMask; + d->AliveWriteMask = + d->BranchMasks[branch_depth].IfWriteMask; + d->BranchMasks[branch_depth].ElseWriteMask = + temp_mask; + d->BranchMasks[branch_depth].HasElse = 1; + } + break; + case RC_OPCODE_ENDIF: + if (branch_depth == 0) { + d->ReaderData->AbortOnRead = d->AliveWriteMask; + d->ReaderData->InElse = 0; + } + else { + pop_branch_mask(d, &branch_depth); + } + break; + default: + break; + } + + if (d->ReaderData->InElse) + continue; + + if (tmp->Type == RC_INSTRUCTION_NORMAL) { + rc_for_all_reads_src(tmp, + get_readers_normal_read_callback, d); + } else { + rc_pair_for_all_reads_arg(tmp, + get_readers_pair_read_callback, d); + } + + /* This can happen when we jump from an ENDLOOP to BGNLOOP */ + if (tmp == writer) { + tmp = endloop; + endloop = NULL; + d->ReaderData->AbortOnRead = abort_on_read_at_endloop; + continue; + } + rc_for_all_writes_mask(tmp, get_readers_write_callback, d); + + if (d->ReaderData->ExitOnAbort && d->ReaderData->Abort) + return; + + if (branch_depth == 0 && !d->AliveWriteMask) + return; + } +} + +static void init_get_readers_callback_data( + struct get_readers_callback_data * d, + struct rc_reader_data * reader_data, + struct radeon_compiler * c, + rc_read_src_fn read_normal_cb, + rc_pair_read_arg_fn read_pair_cb, + rc_read_write_mask_fn write_cb) +{ + reader_data->Abort = 0; + reader_data->ReaderCount = 0; + reader_data->ReadersReserved = 0; + reader_data->Readers = NULL; + + d->C = c; + d->ReaderData = reader_data; + d->ReadNormalCB = read_normal_cb; + d->ReadPairCB = read_pair_cb; + d->WriteCB = write_cb; +} + +/** + * This function will create a list of readers via the rc_reader_data struct. + * This function will abort (set the flag data->Abort) and return if it + * encounters an instruction that reads from @param writer and also a different + * instruction. Here are some examples: + * + * writer = instruction 0; + * 0 MOV TEMP[0].xy, TEMP[1].xy + * 1 MOV TEMP[0].zw, TEMP[2].xy + * 2 MOV TEMP[3], TEMP[0] + * The Abort flag will be set on instruction 2, because it reads values written + * by instructions 0 and 1. + * + * writer = instruction 1; + * 0 IF TEMP[0].x + * 1 MOV TEMP[1], TEMP[2] + * 2 ELSE + * 3 MOV TEMP[1], TEMP[2] + * 4 ENDIF + * 5 MOV TEMP[3], TEMP[1] + * The Abort flag will be set on instruction 5, because it could read from the + * value written by either instruction 1 or 3, depending on the jump decision + * made at instruction 0. + * + * writer = instruction 0; + * 0 MOV TEMP[0], TEMP[1] + * 2 BGNLOOP + * 3 ADD TEMP[0], TEMP[0], none.1 + * 4 ENDLOOP + * The Abort flag will be set on instruction 3, because in the first iteration + * of the loop it reads the value written by instruction 0 and in all other + * iterations it reads the value written by instruction 3. + * + * @param read_cb This function will be called for for every instruction that + * has been determined to be a reader of writer. + * @param write_cb This function will be called for every instruction after + * writer. + */ +void rc_get_readers( + struct radeon_compiler * c, + struct rc_instruction * writer, + struct rc_reader_data * data, + rc_read_src_fn read_normal_cb, + rc_pair_read_arg_fn read_pair_cb, + rc_read_write_mask_fn write_cb) +{ + struct get_readers_callback_data d; + + init_get_readers_callback_data(&d, data, c, read_normal_cb, + read_pair_cb, write_cb); + + rc_for_all_writes_mask(writer, get_readers_for_single_write, &d); +} + +void rc_get_readers_sub( + struct radeon_compiler * c, + struct rc_instruction * writer, + struct rc_pair_sub_instruction * sub_writer, + struct rc_reader_data * data, + rc_read_src_fn read_normal_cb, + rc_pair_read_arg_fn read_pair_cb, + rc_read_write_mask_fn write_cb) +{ + struct get_readers_callback_data d; + + init_get_readers_callback_data(&d, data, c, read_normal_cb, + read_pair_cb, write_cb); + + if (sub_writer->WriteMask) { + get_readers_for_single_write(&d, writer, RC_FILE_TEMPORARY, + sub_writer->DestIndex, sub_writer->WriteMask); + } +} diff --git a/src/gallium/drivers/r300/compiler/radeon_dataflow.h b/src/gallium/drivers/r300/compiler/radeon_dataflow.h new file mode 100644 index 00000000000..d8a627258ea --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_dataflow.h @@ -0,0 +1,134 @@ +/* + * Copyright (C) 2009 Nicolai Haehnle. + * Copyright 2010 Tom Stellard <[email protected]> + * + * 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. + * + */ + +#ifndef RADEON_DATAFLOW_H +#define RADEON_DATAFLOW_H + +#include "radeon_program_constants.h" + +struct radeon_compiler; +struct rc_instruction; +struct rc_swizzle_caps; +struct rc_src_register; +struct rc_pair_instruction_arg; +struct rc_pair_instruction_source; +struct rc_pair_sub_instruction; +struct rc_compiler; + + +/** + * Help analyze and modify the register accesses of instructions. + */ +/*@{*/ +typedef void (*rc_read_write_chan_fn)(void * userdata, struct rc_instruction * inst, + rc_register_file file, unsigned int index, unsigned int chan); +void rc_for_all_reads_chan(struct rc_instruction * inst, rc_read_write_chan_fn cb, void * userdata); +void rc_for_all_writes_chan(struct rc_instruction * inst, rc_read_write_chan_fn cb, void * userdata); + +typedef void (*rc_read_write_mask_fn)(void * userdata, struct rc_instruction * inst, + rc_register_file file, unsigned int index, unsigned int mask); +void rc_for_all_reads_mask(struct rc_instruction * inst, rc_read_write_mask_fn cb, void * userdata); +void rc_for_all_writes_mask(struct rc_instruction * inst, rc_read_write_mask_fn cb, void * userdata); + +typedef void (*rc_read_src_fn)(void * userdata, struct rc_instruction * inst, + struct rc_src_register * src); +void rc_for_all_reads_src(struct rc_instruction * inst, rc_read_src_fn cb, + void * userdata); + +typedef void (*rc_pair_read_arg_fn)(void * userdata, + struct rc_instruction * inst, struct rc_pair_instruction_arg * arg, + struct rc_pair_instruction_source * src); +void rc_pair_for_all_reads_arg(struct rc_instruction * inst, + rc_pair_read_arg_fn cb, void * userdata); + +typedef void (*rc_remap_register_fn)(void * userdata, struct rc_instruction * inst, + rc_register_file * pfile, unsigned int * pindex); +void rc_remap_registers(struct rc_instruction * inst, rc_remap_register_fn cb, void * userdata); +/*@}*/ + +struct rc_reader { + struct rc_instruction * Inst; + unsigned int WriteMask; + union { + struct { + struct rc_src_register * Src; + } I; + struct { + struct rc_pair_instruction_arg * Arg; + struct rc_pair_instruction_source * Src; + } P; + } U; +}; + +struct rc_reader_data { + unsigned int Abort; + unsigned int AbortOnRead; + unsigned int AbortOnWrite; + unsigned int LoopDepth; + unsigned int InElse; + struct rc_instruction * Writer; + + unsigned int ReaderCount; + unsigned int ReadersReserved; + struct rc_reader * Readers; + + /* If this flag is enabled, rc_get_readers will exit as soon possbile + * after the Abort flag is set.*/ + unsigned int ExitOnAbort; + void * CbData; +}; + +void rc_get_readers( + struct radeon_compiler * c, + struct rc_instruction * writer, + struct rc_reader_data * data, + rc_read_src_fn read_normal_cb, + rc_pair_read_arg_fn read_pair_cb, + rc_read_write_mask_fn write_cb); + +void rc_get_readers_sub( + struct radeon_compiler * c, + struct rc_instruction * writer, + struct rc_pair_sub_instruction * sub_writer, + struct rc_reader_data * data, + rc_read_src_fn read_normal_cb, + rc_pair_read_arg_fn read_pair_cb, + rc_read_write_mask_fn write_cb); +/** + * Compiler passes based on dataflow analysis. + */ +/*@{*/ +typedef void (*rc_dataflow_mark_outputs_fn)(void * userdata, void * data, + void (*mark_fn)(void * data, unsigned int index, unsigned int mask)); +void rc_dataflow_deadcode(struct radeon_compiler * c, void *user); +void rc_dataflow_swizzles(struct radeon_compiler * c, void *user); +/*@}*/ + +void rc_optimize(struct radeon_compiler * c, void *user); + +#endif /* RADEON_DATAFLOW_H */ diff --git a/src/gallium/drivers/r300/compiler/radeon_dataflow_deadcode.c b/src/gallium/drivers/r300/compiler/radeon_dataflow_deadcode.c new file mode 100644 index 00000000000..678e1475883 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_dataflow_deadcode.c @@ -0,0 +1,359 @@ +/* + * Copyright (C) 2009 Nicolai Haehnle. + * + * 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. + * + */ + +#include "radeon_dataflow.h" + +#include "radeon_compiler.h" + + +struct updatemask_state { + unsigned char Output[RC_REGISTER_MAX_INDEX]; + unsigned char Temporary[RC_REGISTER_MAX_INDEX]; + unsigned char Address; + unsigned char Special[RC_NUM_SPECIAL_REGISTERS]; +}; + +struct instruction_state { + unsigned char WriteMask:4; + unsigned char WriteALUResult:1; + unsigned char SrcReg[3]; +}; + +struct loopinfo { + struct updatemask_state * Breaks; + unsigned int BreakCount; + unsigned int BreaksReserved; +}; + +struct branchinfo { + unsigned int HaveElse:1; + + struct updatemask_state StoreEndif; + struct updatemask_state StoreElse; +}; + +struct deadcode_state { + struct radeon_compiler * C; + struct instruction_state * Instructions; + + struct updatemask_state R; + + struct branchinfo * BranchStack; + unsigned int BranchStackSize; + unsigned int BranchStackReserved; + + struct loopinfo * LoopStack; + unsigned int LoopStackSize; + unsigned int LoopStackReserved; +}; + + +static void or_updatemasks( + struct updatemask_state * dst, + struct updatemask_state * a, + struct updatemask_state * b) +{ + for(unsigned int i = 0; i < RC_REGISTER_MAX_INDEX; ++i) { + dst->Output[i] = a->Output[i] | b->Output[i]; + dst->Temporary[i] = a->Temporary[i] | b->Temporary[i]; + } + + for(unsigned int i = 0; i < RC_NUM_SPECIAL_REGISTERS; ++i) + dst->Special[i] = a->Special[i] | b->Special[i]; + + dst->Address = a->Address | b->Address; +} + +static void push_break(struct deadcode_state *s) +{ + struct loopinfo * loop = &s->LoopStack[s->LoopStackSize - 1]; + memory_pool_array_reserve(&s->C->Pool, struct updatemask_state, + loop->Breaks, loop->BreakCount, loop->BreaksReserved, 1); + + memcpy(&loop->Breaks[loop->BreakCount++], &s->R, sizeof(s->R)); +} + +static void push_loop(struct deadcode_state * s) +{ + memory_pool_array_reserve(&s->C->Pool, struct loopinfo, s->LoopStack, + s->LoopStackSize, s->LoopStackReserved, 1); + memset(&s->LoopStack[s->LoopStackSize++], 0, sizeof(struct loopinfo)); +} + +static void push_branch(struct deadcode_state * s) +{ + struct branchinfo * branch; + + memory_pool_array_reserve(&s->C->Pool, struct branchinfo, s->BranchStack, + s->BranchStackSize, s->BranchStackReserved, 1); + + branch = &s->BranchStack[s->BranchStackSize++]; + branch->HaveElse = 0; + memcpy(&branch->StoreEndif, &s->R, sizeof(s->R)); +} + +static unsigned char * get_used_ptr(struct deadcode_state *s, rc_register_file file, unsigned int index) +{ + if (file == RC_FILE_OUTPUT || file == RC_FILE_TEMPORARY) { + if (index >= RC_REGISTER_MAX_INDEX) { + rc_error(s->C, "%s: index %i is out of bounds for file %i\n", __FUNCTION__, index, file); + return 0; + } + + if (file == RC_FILE_OUTPUT) + return &s->R.Output[index]; + else + return &s->R.Temporary[index]; + } else if (file == RC_FILE_ADDRESS) { + return &s->R.Address; + } else if (file == RC_FILE_SPECIAL) { + if (index >= RC_NUM_SPECIAL_REGISTERS) { + rc_error(s->C, "%s: special file index %i out of bounds\n", __FUNCTION__, index); + return 0; + } + + return &s->R.Special[index]; + } + + return 0; +} + +static void mark_used(struct deadcode_state * s, rc_register_file file, unsigned int index, unsigned int mask) +{ + unsigned char * pused = get_used_ptr(s, file, index); + if (pused) + *pused |= mask; +} + +static void update_instruction(struct deadcode_state * s, struct rc_instruction * inst) +{ + const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode); + struct instruction_state * insts = &s->Instructions[inst->IP]; + unsigned int usedmask = 0; + unsigned int srcmasks[3]; + + if (opcode->HasDstReg) { + unsigned char * pused = get_used_ptr(s, inst->U.I.DstReg.File, inst->U.I.DstReg.Index); + if (pused) { + usedmask = *pused & inst->U.I.DstReg.WriteMask; + *pused &= ~usedmask; + } + } + + insts->WriteMask |= usedmask; + + if (inst->U.I.WriteALUResult) { + unsigned char * pused = get_used_ptr(s, RC_FILE_SPECIAL, RC_SPECIAL_ALU_RESULT); + if (pused && *pused) { + if (inst->U.I.WriteALUResult == RC_ALURESULT_X) + usedmask |= RC_MASK_X; + else if (inst->U.I.WriteALUResult == RC_ALURESULT_W) + usedmask |= RC_MASK_W; + + *pused = 0; + insts->WriteALUResult = 1; + } + } + + rc_compute_sources_for_writemask(inst, usedmask, srcmasks); + + for(unsigned int src = 0; src < opcode->NumSrcRegs; ++src) { + unsigned int refmask = 0; + unsigned int newsrcmask = srcmasks[src] & ~insts->SrcReg[src]; + insts->SrcReg[src] |= newsrcmask; + + for(unsigned int chan = 0; chan < 4; ++chan) { + if (GET_BIT(newsrcmask, chan)) + refmask |= 1 << GET_SWZ(inst->U.I.SrcReg[src].Swizzle, chan); + } + + /* get rid of spurious bits from ZERO, ONE, etc. swizzles */ + refmask &= RC_MASK_XYZW; + + if (!refmask) + continue; + + mark_used(s, inst->U.I.SrcReg[src].File, inst->U.I.SrcReg[src].Index, refmask); + + if (inst->U.I.SrcReg[src].RelAddr) + mark_used(s, RC_FILE_ADDRESS, 0, RC_MASK_X); + } +} + +static void mark_output_use(void * data, unsigned int index, unsigned int mask) +{ + struct deadcode_state * s = data; + + mark_used(s, RC_FILE_OUTPUT, index, mask); +} + +void rc_dataflow_deadcode(struct radeon_compiler * c, void *user) +{ + struct deadcode_state s; + unsigned int nr_instructions; + rc_dataflow_mark_outputs_fn dce = (rc_dataflow_mark_outputs_fn)user; + unsigned int ip; + + memset(&s, 0, sizeof(s)); + s.C = c; + + nr_instructions = rc_recompute_ips(c); + s.Instructions = memory_pool_malloc(&c->Pool, sizeof(struct instruction_state)*nr_instructions); + memset(s.Instructions, 0, sizeof(struct instruction_state)*nr_instructions); + + dce(c, &s, &mark_output_use); + + for(struct rc_instruction * inst = c->Program.Instructions.Prev; + inst != &c->Program.Instructions; + inst = inst->Prev) { + const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode); + + switch(opcode->Opcode){ + /* Mark all sources in the loop body as used before doing + * normal deadcode analysis. This is probably not optimal. + */ + case RC_OPCODE_ENDLOOP: + { + int endloops = 1; + struct rc_instruction *ptr; + for(ptr = inst->Prev; endloops > 0; ptr = ptr->Prev){ + opcode = rc_get_opcode_info(ptr->U.I.Opcode); + if(ptr->U.I.Opcode == RC_OPCODE_BGNLOOP){ + endloops--; + continue; + } + if(ptr->U.I.Opcode == RC_OPCODE_ENDLOOP){ + endloops++; + continue; + } + if(opcode->HasDstReg){ + int src = 0; + unsigned int srcmasks[3]; + rc_compute_sources_for_writemask(ptr, + ptr->U.I.DstReg.WriteMask, srcmasks); + for(src=0; src < opcode->NumSrcRegs; src++){ + mark_used(&s, + ptr->U.I.SrcReg[src].File, + ptr->U.I.SrcReg[src].Index, + srcmasks[src]); + } + } + } + push_loop(&s); + break; + } + case RC_OPCODE_BRK: + push_break(&s); + break; + case RC_OPCODE_BGNLOOP: + { + unsigned int i; + struct loopinfo * loop = &s.LoopStack[s.LoopStackSize-1]; + for(i = 0; i < loop->BreakCount; i++) { + or_updatemasks(&s.R, &s.R, &loop->Breaks[i]); + } + break; + } + case RC_OPCODE_CONT: + break; + case RC_OPCODE_ENDIF: + push_branch(&s); + break; + default: + if (opcode->IsFlowControl && s.BranchStackSize) { + struct branchinfo * branch = &s.BranchStack[s.BranchStackSize-1]; + if (opcode->Opcode == RC_OPCODE_IF) { + or_updatemasks(&s.R, + &s.R, + branch->HaveElse ? &branch->StoreElse : &branch->StoreEndif); + + s.BranchStackSize--; + } else if (opcode->Opcode == RC_OPCODE_ELSE) { + if (branch->HaveElse) { + rc_error(c, "%s: Multiple ELSE for one IF/ENDIF\n", __FUNCTION__); + } else { + memcpy(&branch->StoreElse, &s.R, sizeof(s.R)); + memcpy(&s.R, &branch->StoreEndif, sizeof(s.R)); + branch->HaveElse = 1; + } + } else { + rc_error(c, "%s: Unhandled control flow instruction %s\n", __FUNCTION__, opcode->Name); + } + } + } + + update_instruction(&s, inst); + } + + ip = 0; + for(struct rc_instruction * inst = c->Program.Instructions.Next; + inst != &c->Program.Instructions; + inst = inst->Next, ++ip) { + const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode); + int dead = 1; + unsigned int srcmasks[3]; + unsigned int usemask; + + if (!opcode->HasDstReg) { + dead = 0; + } else { + inst->U.I.DstReg.WriteMask = s.Instructions[ip].WriteMask; + if (s.Instructions[ip].WriteMask) + dead = 0; + + if (s.Instructions[ip].WriteALUResult) + dead = 0; + else + inst->U.I.WriteALUResult = RC_ALURESULT_NONE; + } + + if (dead) { + struct rc_instruction * todelete = inst; + inst = inst->Prev; + rc_remove_instruction(todelete); + continue; + } + + usemask = s.Instructions[ip].WriteMask; + + if (inst->U.I.WriteALUResult == RC_ALURESULT_X) + usemask |= RC_MASK_X; + else if (inst->U.I.WriteALUResult == RC_ALURESULT_W) + usemask |= RC_MASK_W; + + rc_compute_sources_for_writemask(inst, usemask, srcmasks); + + for(unsigned int src = 0; src < 3; ++src) { + for(unsigned int chan = 0; chan < 4; ++chan) { + if (!GET_BIT(srcmasks[src], chan)) + SET_SWZ(inst->U.I.SrcReg[src].Swizzle, chan, RC_SWIZZLE_UNUSED); + } + } + } + + rc_calculate_inputs_outputs(c); +} diff --git a/src/gallium/drivers/r300/compiler/radeon_dataflow_swizzles.c b/src/gallium/drivers/r300/compiler/radeon_dataflow_swizzles.c new file mode 100644 index 00000000000..133a9f72ec7 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_dataflow_swizzles.c @@ -0,0 +1,103 @@ +/* + * Copyright (C) 2009 Nicolai Haehnle. + * + * 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. + * + */ + +#include "radeon_dataflow.h" + +#include "radeon_compiler.h" +#include "radeon_swizzle.h" + + +static void rewrite_source(struct radeon_compiler * c, + struct rc_instruction * inst, unsigned src) +{ + struct rc_swizzle_split split; + unsigned int tempreg = rc_find_free_temporary(c); + unsigned int usemask; + + usemask = 0; + for(unsigned int chan = 0; chan < 4; ++chan) { + if (GET_SWZ(inst->U.I.SrcReg[src].Swizzle, chan) != RC_SWIZZLE_UNUSED) + usemask |= 1 << chan; + } + + c->SwizzleCaps->Split(inst->U.I.SrcReg[src], usemask, &split); + + for(unsigned int phase = 0; phase < split.NumPhases; ++phase) { + struct rc_instruction * mov = rc_insert_new_instruction(c, inst->Prev); + unsigned int phase_refmask; + unsigned int masked_negate; + + mov->U.I.Opcode = RC_OPCODE_MOV; + mov->U.I.DstReg.File = RC_FILE_TEMPORARY; + mov->U.I.DstReg.Index = tempreg; + mov->U.I.DstReg.WriteMask = split.Phase[phase]; + mov->U.I.SrcReg[0] = inst->U.I.SrcReg[src]; + mov->U.I.PreSub = inst->U.I.PreSub; + + phase_refmask = 0; + for(unsigned int chan = 0; chan < 4; ++chan) { + if (!GET_BIT(split.Phase[phase], chan)) + SET_SWZ(mov->U.I.SrcReg[0].Swizzle, chan, RC_SWIZZLE_UNUSED); + else + phase_refmask |= 1 << GET_SWZ(mov->U.I.SrcReg[0].Swizzle, chan); + } + + phase_refmask &= RC_MASK_XYZW; + + masked_negate = split.Phase[phase] & mov->U.I.SrcReg[0].Negate; + if (masked_negate == 0) + mov->U.I.SrcReg[0].Negate = 0; + else if (masked_negate == split.Phase[phase]) + mov->U.I.SrcReg[0].Negate = RC_MASK_XYZW; + + } + + inst->U.I.SrcReg[src].File = RC_FILE_TEMPORARY; + inst->U.I.SrcReg[src].Index = tempreg; + inst->U.I.SrcReg[src].Swizzle = 0; + inst->U.I.SrcReg[src].Negate = RC_MASK_NONE; + inst->U.I.SrcReg[src].Abs = 0; + for(unsigned int chan = 0; chan < 4; ++chan) { + SET_SWZ(inst->U.I.SrcReg[src].Swizzle, chan, + GET_BIT(usemask, chan) ? chan : RC_SWIZZLE_UNUSED); + } +} + +void rc_dataflow_swizzles(struct radeon_compiler * c, void *user) +{ + struct rc_instruction * inst; + + for(inst = c->Program.Instructions.Next; inst != &c->Program.Instructions; inst = inst->Next) { + const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode); + unsigned int src; + + for(src = 0; src < opcode->NumSrcRegs; ++src) { + if (!c->SwizzleCaps->IsNative(inst->U.I.Opcode, inst->U.I.SrcReg[src])) + rewrite_source(c, inst, src); + } + } +} diff --git a/src/gallium/drivers/r300/compiler/radeon_emulate_branches.c b/src/gallium/drivers/r300/compiler/radeon_emulate_branches.c new file mode 100644 index 00000000000..7bede344f30 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_emulate_branches.c @@ -0,0 +1,342 @@ +/* + * Copyright 2009 Nicolai Hähnle <[email protected]> + * + * 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 + * on the rights to use, copy, modify, merge, publish, distribute, sub + * license, 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 NON-INFRINGEMENT. IN NO EVENT SHALL + * THE AUTHOR(S) AND/OR THEIR 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. */ + +#include "radeon_emulate_branches.h" + +#include <stdio.h> + +#include "radeon_compiler.h" +#include "radeon_dataflow.h" + +#define VERBOSE 0 + +#define DBG(...) do { if (VERBOSE) fprintf(stderr, __VA_ARGS__); } while(0) + + +struct proxy_info { + unsigned int Proxied:1; + unsigned int Index:RC_REGISTER_INDEX_BITS; +}; + +struct register_proxies { + struct proxy_info Temporary[RC_REGISTER_MAX_INDEX]; +}; + +struct branch_info { + struct rc_instruction * If; + struct rc_instruction * Else; +}; + +struct emulate_branch_state { + struct radeon_compiler * C; + + struct branch_info * Branches; + unsigned int BranchCount; + unsigned int BranchReserved; +}; + + +static void handle_if(struct emulate_branch_state * s, struct rc_instruction * inst) +{ + struct branch_info * branch; + struct rc_instruction * inst_mov; + + memory_pool_array_reserve(&s->C->Pool, struct branch_info, + s->Branches, s->BranchCount, s->BranchReserved, 1); + + DBG("%s\n", __FUNCTION__); + + branch = &s->Branches[s->BranchCount++]; + memset(branch, 0, sizeof(struct branch_info)); + branch->If = inst; + + /* Make a safety copy of the decision register, because we will need + * it at ENDIF time and it might be overwritten in both branches. */ + inst_mov = rc_insert_new_instruction(s->C, inst->Prev); + inst_mov->U.I.Opcode = RC_OPCODE_MOV; + inst_mov->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst_mov->U.I.DstReg.Index = rc_find_free_temporary(s->C); + inst_mov->U.I.DstReg.WriteMask = RC_MASK_X; + inst_mov->U.I.SrcReg[0] = inst->U.I.SrcReg[0]; + + inst->U.I.SrcReg[0].File = RC_FILE_TEMPORARY; + inst->U.I.SrcReg[0].Index = inst_mov->U.I.DstReg.Index; + inst->U.I.SrcReg[0].Swizzle = 0; + inst->U.I.SrcReg[0].Abs = 0; + inst->U.I.SrcReg[0].Negate = 0; +} + +static void handle_else(struct emulate_branch_state * s, struct rc_instruction * inst) +{ + struct branch_info * branch; + + if (!s->BranchCount) { + rc_error(s->C, "Encountered ELSE outside of branches"); + return; + } + + DBG("%s\n", __FUNCTION__); + + branch = &s->Branches[s->BranchCount - 1]; + branch->Else = inst; +} + + +struct state_and_proxies { + struct emulate_branch_state * S; + struct register_proxies * Proxies; +}; + +static struct proxy_info * get_proxy_info(struct state_and_proxies * sap, + rc_register_file file, unsigned int index) +{ + if (file == RC_FILE_TEMPORARY) { + return &sap->Proxies->Temporary[index]; + } else { + return 0; + } +} + +static void scan_write(void * userdata, struct rc_instruction * inst, + rc_register_file file, unsigned int index, unsigned int comp) +{ + struct state_and_proxies * sap = userdata; + struct proxy_info * proxy = get_proxy_info(sap, file, index); + + if (proxy && !proxy->Proxied) { + proxy->Proxied = 1; + proxy->Index = rc_find_free_temporary(sap->S->C); + } +} + +static void remap_proxy_function(void * userdata, struct rc_instruction * inst, + rc_register_file * pfile, unsigned int * pindex) +{ + struct state_and_proxies * sap = userdata; + struct proxy_info * proxy = get_proxy_info(sap, *pfile, *pindex); + + if (proxy && proxy->Proxied) { + *pfile = RC_FILE_TEMPORARY; + *pindex = proxy->Index; + } +} + +/** + * Redirect all writes in the instruction range [begin, end) to proxy + * temporary registers. + */ +static void allocate_and_insert_proxies(struct emulate_branch_state * s, + struct register_proxies * proxies, + struct rc_instruction * begin, + struct rc_instruction * end) +{ + struct state_and_proxies sap; + + sap.S = s; + sap.Proxies = proxies; + + for(struct rc_instruction * inst = begin; inst != end; inst = inst->Next) { + rc_for_all_writes_mask(inst, scan_write, &sap); + rc_remap_registers(inst, remap_proxy_function, &sap); + } + + for(unsigned int index = 0; index < RC_REGISTER_MAX_INDEX; ++index) { + if (proxies->Temporary[index].Proxied) { + struct rc_instruction * inst_mov = rc_insert_new_instruction(s->C, begin->Prev); + inst_mov->U.I.Opcode = RC_OPCODE_MOV; + inst_mov->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst_mov->U.I.DstReg.Index = proxies->Temporary[index].Index; + inst_mov->U.I.DstReg.WriteMask = RC_MASK_XYZW; + inst_mov->U.I.SrcReg[0].File = RC_FILE_TEMPORARY; + inst_mov->U.I.SrcReg[0].Index = index; + } + } +} + + +static void inject_cmp(struct emulate_branch_state * s, + struct rc_instruction * inst_if, + struct rc_instruction * inst_endif, + rc_register_file file, unsigned int index, + struct proxy_info ifproxy, + struct proxy_info elseproxy) +{ + struct rc_instruction * inst_cmp = rc_insert_new_instruction(s->C, inst_endif); + inst_cmp->U.I.Opcode = RC_OPCODE_CMP; + inst_cmp->U.I.DstReg.File = file; + inst_cmp->U.I.DstReg.Index = index; + inst_cmp->U.I.DstReg.WriteMask = RC_MASK_XYZW; + inst_cmp->U.I.SrcReg[0] = inst_if->U.I.SrcReg[0]; + inst_cmp->U.I.SrcReg[0].Abs = 1; + inst_cmp->U.I.SrcReg[0].Negate = RC_MASK_XYZW; + inst_cmp->U.I.SrcReg[1].File = RC_FILE_TEMPORARY; + inst_cmp->U.I.SrcReg[1].Index = ifproxy.Proxied ? ifproxy.Index : index; + inst_cmp->U.I.SrcReg[2].File = RC_FILE_TEMPORARY; + inst_cmp->U.I.SrcReg[2].Index = elseproxy.Proxied ? elseproxy.Index : index; +} + +static void handle_endif(struct emulate_branch_state * s, struct rc_instruction * inst) +{ + struct branch_info * branch; + struct register_proxies IfProxies; + struct register_proxies ElseProxies; + + if (!s->BranchCount) { + rc_error(s->C, "Encountered ENDIF outside of branches"); + return; + } + + DBG("%s\n", __FUNCTION__); + + branch = &s->Branches[s->BranchCount - 1]; + + memset(&IfProxies, 0, sizeof(IfProxies)); + memset(&ElseProxies, 0, sizeof(ElseProxies)); + + allocate_and_insert_proxies(s, &IfProxies, branch->If->Next, branch->Else ? branch->Else : inst); + + if (branch->Else) + allocate_and_insert_proxies(s, &ElseProxies, branch->Else->Next, inst); + + /* Insert the CMP instructions at the end. */ + for(unsigned int index = 0; index < RC_REGISTER_MAX_INDEX; ++index) { + if (IfProxies.Temporary[index].Proxied || ElseProxies.Temporary[index].Proxied) { + inject_cmp(s, branch->If, inst, RC_FILE_TEMPORARY, index, + IfProxies.Temporary[index], ElseProxies.Temporary[index]); + } + } + + /* Remove all traces of the branch instructions */ + rc_remove_instruction(branch->If); + if (branch->Else) + rc_remove_instruction(branch->Else); + rc_remove_instruction(inst); + + s->BranchCount--; + + if (VERBOSE) { + DBG("Program after ENDIF handling:\n"); + rc_print_program(&s->C->Program); + } +} + + +struct remap_output_data { + unsigned int Output:RC_REGISTER_INDEX_BITS; + unsigned int Temporary:RC_REGISTER_INDEX_BITS; +}; + +static void remap_output_function(void * userdata, struct rc_instruction * inst, + rc_register_file * pfile, unsigned int * pindex) +{ + struct remap_output_data * data = userdata; + + if (*pfile == RC_FILE_OUTPUT && *pindex == data->Output) { + *pfile = RC_FILE_TEMPORARY; + *pindex = data->Temporary; + } +} + + +/** + * Output registers cannot be read from and so cannot be dealt with like + * temporary registers. + * + * We do the simplest thing: If an output registers is written within + * a branch, then *all* writes to this register are proxied to a + * temporary register, and a final MOV is appended to the end of + * the program. + */ +static void fix_output_writes(struct emulate_branch_state * s, struct rc_instruction * inst) +{ + const struct rc_opcode_info * opcode; + + if (!s->BranchCount) + return; + + opcode = rc_get_opcode_info(inst->U.I.Opcode); + + if (!opcode->HasDstReg) + return; + + if (inst->U.I.DstReg.File == RC_FILE_OUTPUT) { + struct remap_output_data remap; + struct rc_instruction * inst_mov; + + remap.Output = inst->U.I.DstReg.Index; + remap.Temporary = rc_find_free_temporary(s->C); + + for(struct rc_instruction * inst = s->C->Program.Instructions.Next; + inst != &s->C->Program.Instructions; + inst = inst->Next) { + rc_remap_registers(inst, &remap_output_function, &remap); + } + + inst_mov = rc_insert_new_instruction(s->C, s->C->Program.Instructions.Prev); + inst_mov->U.I.Opcode = RC_OPCODE_MOV; + inst_mov->U.I.DstReg.File = RC_FILE_OUTPUT; + inst_mov->U.I.DstReg.Index = remap.Output; + inst_mov->U.I.DstReg.WriteMask = RC_MASK_XYZW; + inst_mov->U.I.SrcReg[0].File = RC_FILE_TEMPORARY; + inst_mov->U.I.SrcReg[0].Index = remap.Temporary; + } +} + +/** + * Remove branch instructions; instead, execute both branches + * on different register sets and choose between their results + * using CMP instructions in place of the original ENDIF. + */ +void rc_emulate_branches(struct radeon_compiler *c, void *user) +{ + struct emulate_branch_state s; + struct rc_instruction * ptr; + + memset(&s, 0, sizeof(s)); + s.C = c; + + /* Untypical loop because we may remove the current instruction */ + ptr = c->Program.Instructions.Next; + while(ptr != &c->Program.Instructions) { + struct rc_instruction * inst = ptr; + ptr = ptr->Next; + + if (inst->Type == RC_INSTRUCTION_NORMAL) { + switch(inst->U.I.Opcode) { + case RC_OPCODE_IF: + handle_if(&s, inst); + break; + case RC_OPCODE_ELSE: + handle_else(&s, inst); + break; + case RC_OPCODE_ENDIF: + handle_endif(&s, inst); + break; + default: + fix_output_writes(&s, inst); + break; + } + } else { + rc_error(c, "%s: unhandled instruction type\n", __FUNCTION__); + } + } +} diff --git a/src/gallium/drivers/r300/compiler/radeon_emulate_branches.h b/src/gallium/drivers/r300/compiler/radeon_emulate_branches.h new file mode 100644 index 00000000000..818ab84d0cd --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_emulate_branches.h @@ -0,0 +1,30 @@ +/* + * Copyright 2009 Nicolai Hähnle <[email protected]> + * + * 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 + * on the rights to use, copy, modify, merge, publish, distribute, sub + * license, 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 NON-INFRINGEMENT. IN NO EVENT SHALL + * THE AUTHOR(S) AND/OR THEIR 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. */ + +#ifndef RADEON_EMULATE_BRANCHES_H +#define RADEON_EMULATE_BRANCHES_H + +struct radeon_compiler; + +void rc_emulate_branches(struct radeon_compiler *c, void *user); + +#endif /* RADEON_EMULATE_BRANCHES_H */ diff --git a/src/gallium/drivers/r300/compiler/radeon_emulate_loops.c b/src/gallium/drivers/r300/compiler/radeon_emulate_loops.c new file mode 100644 index 00000000000..205eecd1129 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_emulate_loops.c @@ -0,0 +1,522 @@ +/* + * Copyright 2010 Tom Stellard <[email protected]> + * + * 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 + */ + +#include "radeon_emulate_loops.h" + +#include "radeon_compiler.h" +#include "radeon_dataflow.h" + +#define VERBOSE 0 + +#define DBG(...) do { if (VERBOSE) fprintf(stderr, __VA_ARGS__); } while(0) + +struct const_value { + struct radeon_compiler * C; + struct rc_src_register * Src; + float Value; + int HasValue; +}; + +struct count_inst { + struct radeon_compiler * C; + int Index; + rc_swizzle Swz; + float Amount; + int Unknown; +}; + +static float get_constant_value(struct radeon_compiler * c, + struct rc_src_register * src, + int chan) +{ + float base = 1.0f; + int swz = GET_SWZ(src->Swizzle, chan); + if(swz >= 4 || src->Index >= c->Program.Constants.Count ){ + rc_error(c, "get_constant_value: Can't find a value.\n"); + return 0.0f; + } + if(GET_BIT(src->Negate, chan)){ + base = -1.0f; + } + return base * + c->Program.Constants.Constants[src->Index].u.Immediate[swz]; +} + +static int src_reg_is_immediate(struct rc_src_register * src, + struct radeon_compiler * c) +{ + return src->File == RC_FILE_CONSTANT && + c->Program.Constants.Constants[src->Index].Type==RC_CONSTANT_IMMEDIATE; +} + +static unsigned int loop_max_possible_iterations(struct radeon_compiler *c, + struct loop_info * loop) +{ + unsigned int total_i = rc_recompute_ips(c); + unsigned int loop_i = (loop->EndLoop->IP - loop->BeginLoop->IP) - 1; + /* +1 because the program already has one iteration of the loop. */ + return 1 + ((c->max_alu_insts - total_i) / loop_i); +} + +static void unroll_loop(struct radeon_compiler * c, struct loop_info * loop, + unsigned int iterations) +{ + unsigned int i; + struct rc_instruction * ptr; + struct rc_instruction * first = loop->BeginLoop->Next; + struct rc_instruction * last = loop->EndLoop->Prev; + struct rc_instruction * append_to = last; + rc_remove_instruction(loop->BeginLoop); + rc_remove_instruction(loop->EndLoop); + for( i = 1; i < iterations; i++){ + for(ptr = first; ptr != last->Next; ptr = ptr->Next){ + struct rc_instruction *new = rc_alloc_instruction(c); + memcpy(new, ptr, sizeof(struct rc_instruction)); + rc_insert_instruction(append_to, new); + append_to = new; + } + } +} + + +static void update_const_value(void * data, struct rc_instruction * inst, + rc_register_file file, unsigned int index, unsigned int mask) +{ + struct const_value * value = data; + if(value->Src->File != file || + value->Src->Index != index || + !(1 << GET_SWZ(value->Src->Swizzle, 0) & mask)){ + return; + } + switch(inst->U.I.Opcode){ + case RC_OPCODE_MOV: + if(!src_reg_is_immediate(&inst->U.I.SrcReg[0], value->C)){ + return; + } + value->HasValue = 1; + value->Value = + get_constant_value(value->C, &inst->U.I.SrcReg[0], 0); + break; + } +} + +static void get_incr_amount(void * data, struct rc_instruction * inst, + rc_register_file file, unsigned int index, unsigned int mask) +{ + struct count_inst * count_inst = data; + int amnt_src_index; + const struct rc_opcode_info * opcode; + float amount; + + if(file != RC_FILE_TEMPORARY || + count_inst->Index != index || + (1 << GET_SWZ(count_inst->Swz,0) != mask)){ + return; + } + /* Find the index of the counter register. */ + opcode = rc_get_opcode_info(inst->U.I.Opcode); + if(opcode->NumSrcRegs != 2){ + count_inst->Unknown = 1; + return; + } + if(inst->U.I.SrcReg[0].File == RC_FILE_TEMPORARY && + inst->U.I.SrcReg[0].Index == count_inst->Index && + inst->U.I.SrcReg[0].Swizzle == count_inst->Swz){ + amnt_src_index = 1; + } else if( inst->U.I.SrcReg[1].File == RC_FILE_TEMPORARY && + inst->U.I.SrcReg[1].Index == count_inst->Index && + inst->U.I.SrcReg[1].Swizzle == count_inst->Swz){ + amnt_src_index = 0; + } + else{ + count_inst->Unknown = 1; + return; + } + if(src_reg_is_immediate(&inst->U.I.SrcReg[amnt_src_index], + count_inst->C)){ + amount = get_constant_value(count_inst->C, + &inst->U.I.SrcReg[amnt_src_index], 0); + } + else{ + count_inst->Unknown = 1 ; + return; + } + switch(inst->U.I.Opcode){ + case RC_OPCODE_ADD: + count_inst->Amount += amount; + break; + case RC_OPCODE_SUB: + if(amnt_src_index == 0){ + count_inst->Unknown = 0; + return; + } + count_inst->Amount -= amount; + break; + default: + count_inst->Unknown = 1; + return; + } +} + +/** + * If c->max_alu_inst is -1, then all eligible loops will be unrolled regardless + * of how many iterations they have. + */ +static int try_unroll_loop(struct radeon_compiler * c, struct loop_info * loop) +{ + int end_loops; + int iterations; + struct count_inst count_inst; + float limit_value; + struct rc_src_register * counter; + struct rc_src_register * limit; + struct const_value counter_value; + struct rc_instruction * inst; + + /* Find the counter and the upper limit */ + + if(src_reg_is_immediate(&loop->Cond->U.I.SrcReg[0], c)){ + limit = &loop->Cond->U.I.SrcReg[0]; + counter = &loop->Cond->U.I.SrcReg[1]; + } + else if(src_reg_is_immediate(&loop->Cond->U.I.SrcReg[1], c)){ + limit = &loop->Cond->U.I.SrcReg[1]; + counter = &loop->Cond->U.I.SrcReg[0]; + } + else{ + DBG("No constant limit.\n"); + return 0; + } + + /* Find the initial value of the counter */ + counter_value.Src = counter; + counter_value.Value = 0.0f; + counter_value.HasValue = 0; + counter_value.C = c; + for(inst = c->Program.Instructions.Next; inst != loop->BeginLoop; + inst = inst->Next){ + rc_for_all_writes_mask(inst, update_const_value, &counter_value); + } + if(!counter_value.HasValue){ + DBG("Initial counter value cannot be determined.\n"); + return 0; + } + DBG("Initial counter value is %f\n", counter_value.Value); + /* Determine how the counter is modified each loop */ + count_inst.C = c; + count_inst.Index = counter->Index; + count_inst.Swz = counter->Swizzle; + count_inst.Amount = 0.0f; + count_inst.Unknown = 0; + end_loops = 1; + for(inst = loop->BeginLoop->Next; end_loops > 0; inst = inst->Next){ + switch(inst->U.I.Opcode){ + /* XXX In the future we might want to try to unroll nested + * loops here.*/ + case RC_OPCODE_BGNLOOP: + end_loops++; + break; + case RC_OPCODE_ENDLOOP: + loop->EndLoop = inst; + end_loops--; + break; + case RC_OPCODE_BRK: + /* Don't unroll loops if it has a BRK instruction + * other one used when testing the main conditional + * of the loop. */ + + /* Make sure we haven't entered a nested loops. */ + if(inst != loop->Brk && end_loops == 1) { + return 0; + } + break; + /* XXX Check if the counter is modified within an if statement. + */ + case RC_OPCODE_IF: + break; + default: + rc_for_all_writes_mask(inst, get_incr_amount, &count_inst); + if(count_inst.Unknown){ + return 0; + } + break; + } + } + /* Infinite loop */ + if(count_inst.Amount == 0.0f){ + return 0; + } + DBG("Counter is increased by %f each iteration.\n", count_inst.Amount); + /* Calculate the number of iterations of this loop. Keeping this + * simple, since we only support increment and decrement loops. + */ + limit_value = get_constant_value(c, limit, 0); + DBG("Limit is %f.\n", limit_value); + /* The iteration calculations are opposite of what you would expect. + * In a normal loop, if the condition is met, then loop continues, but + * with our loops, if the condition is met, the is exited. */ + switch(loop->Cond->U.I.Opcode){ + case RC_OPCODE_SGE: + case RC_OPCODE_SLE: + iterations = (int) ceilf((limit_value - counter_value.Value) / + count_inst.Amount); + break; + + case RC_OPCODE_SGT: + case RC_OPCODE_SLT: + iterations = (int) floorf((limit_value - counter_value.Value) / + count_inst.Amount) + 1; + break; + default: + return 0; + } + + if (c->max_alu_insts > 0 + && iterations > loop_max_possible_iterations(c, loop)) { + return 0; + } + + DBG("Loop will have %d iterations.\n", iterations); + + /* Prepare loop for unrolling */ + rc_remove_instruction(loop->Cond); + rc_remove_instruction(loop->If); + rc_remove_instruction(loop->Brk); + rc_remove_instruction(loop->EndIf); + + unroll_loop(c, loop, iterations); + loop->EndLoop = NULL; + return 1; +} + +/** + * @param c + * @param loop + * @param inst A pointer to a BGNLOOP instruction. + * @return 1 if all of the members of loop where set. + * @return 0 if there was an error and some members of loop are still NULL. + */ +static int build_loop_info(struct radeon_compiler * c, struct loop_info * loop, + struct rc_instruction * inst) +{ + struct rc_instruction * ptr; + + if(inst->U.I.Opcode != RC_OPCODE_BGNLOOP){ + rc_error(c, "%s: expected BGNLOOP", __FUNCTION__); + return 0; + } + + memset(loop, 0, sizeof(struct loop_info)); + + loop->BeginLoop = inst; + + for(ptr = loop->BeginLoop->Next; !loop->EndLoop; ptr = ptr->Next) { + + if (ptr == &c->Program.Instructions) { + rc_error(c, "%s: BGNLOOP without an ENDLOOOP.\n", + __FUNCTION__); + return 0; + } + + switch(ptr->U.I.Opcode){ + case RC_OPCODE_BGNLOOP: + { + /* Nested loop, skip ahead to the end. */ + unsigned int loop_depth = 1; + for(ptr = ptr->Next; ptr != &c->Program.Instructions; + ptr = ptr->Next){ + if (ptr->U.I.Opcode == RC_OPCODE_BGNLOOP) { + loop_depth++; + } else if (ptr->U.I.Opcode == RC_OPCODE_ENDLOOP) { + if (!--loop_depth) { + break; + } + } + } + if (ptr == &c->Program.Instructions) { + rc_error(c, "%s: BGNLOOP without an ENDLOOOP\n", + __FUNCTION__); + return 0; + } + break; + } + case RC_OPCODE_BRK: + if(ptr->Next->U.I.Opcode != RC_OPCODE_ENDIF + || ptr->Prev->U.I.Opcode != RC_OPCODE_IF + || loop->Brk){ + continue; + } + loop->Brk = ptr; + loop->If = ptr->Prev; + loop->EndIf = ptr->Next; + switch(loop->If->Prev->U.I.Opcode){ + case RC_OPCODE_SLT: + case RC_OPCODE_SGE: + case RC_OPCODE_SGT: + case RC_OPCODE_SLE: + case RC_OPCODE_SEQ: + case RC_OPCODE_SNE: + break; + default: + return 0; + } + loop->Cond = loop->If->Prev; + break; + + case RC_OPCODE_ENDLOOP: + loop->EndLoop = ptr; + break; + } + } + + if (loop->BeginLoop && loop->Brk && loop->If && loop->EndIf + && loop->Cond && loop->EndLoop) { + return 1; + } + return 0; +} + +/** + * This function prepares a loop to be unrolled by converting it into an if + * statement. Here is an outline of the conversion process: + * BGNLOOP; -> BGNLOOP; + * <Additional conditional code> -> <Additional conditional code> + * SGE/SLT temp[0], temp[1], temp[2]; -> SLT/SGE temp[0], temp[1], temp[2]; + * IF temp[0]; -> IF temp[0]; + * BRK; -> + * ENDIF; -> <Loop Body> + * <Loop Body> -> ENDIF; + * ENDLOOP; -> ENDLOOP + * + * @param inst A pointer to a BGNLOOP instruction. + * @return 1 for success, 0 for failure + */ +static int transform_loop(struct emulate_loop_state * s, + struct rc_instruction * inst) +{ + struct loop_info * loop; + + memory_pool_array_reserve(&s->C->Pool, struct loop_info, + s->Loops, s->LoopCount, s->LoopReserved, 1); + + loop = &s->Loops[s->LoopCount++]; + + if (!build_loop_info(s->C, loop, inst)) { + rc_error(s->C, "Failed to build loop info\n"); + return 0; + } + + if(try_unroll_loop(s->C, loop)){ + return 1; + } + + /* Reverse the conditional instruction */ + switch(loop->Cond->U.I.Opcode){ + case RC_OPCODE_SGE: + loop->Cond->U.I.Opcode = RC_OPCODE_SLT; + break; + case RC_OPCODE_SLT: + loop->Cond->U.I.Opcode = RC_OPCODE_SGE; + break; + case RC_OPCODE_SLE: + loop->Cond->U.I.Opcode = RC_OPCODE_SGT; + break; + case RC_OPCODE_SGT: + loop->Cond->U.I.Opcode = RC_OPCODE_SLE; + break; + case RC_OPCODE_SEQ: + loop->Cond->U.I.Opcode = RC_OPCODE_SNE; + break; + case RC_OPCODE_SNE: + loop->Cond->U.I.Opcode = RC_OPCODE_SEQ; + break; + default: + rc_error(s->C, "loop->Cond is not a conditional.\n"); + return 0; + } + + /* Prepare the loop to be emulated */ + rc_remove_instruction(loop->Brk); + rc_remove_instruction(loop->EndIf); + rc_insert_instruction(loop->EndLoop->Prev, loop->EndIf); + return 1; +} + +void rc_transform_loops(struct radeon_compiler *c, void *user) +{ + struct emulate_loop_state * s = &c->loop_state; + struct rc_instruction * ptr; + + memset(s, 0, sizeof(struct emulate_loop_state)); + s->C = c; + for(ptr = s->C->Program.Instructions.Next; + ptr != &s->C->Program.Instructions; ptr = ptr->Next) { + if(ptr->Type == RC_INSTRUCTION_NORMAL && + ptr->U.I.Opcode == RC_OPCODE_BGNLOOP){ + if (!transform_loop(s, ptr)) + return; + } + } +} + +void rc_unroll_loops(struct radeon_compiler *c, void *user) +{ + struct rc_instruction * inst; + struct loop_info loop; + + for(inst = c->Program.Instructions.Next; + inst != &c->Program.Instructions; inst = inst->Next) { + + if (inst->U.I.Opcode == RC_OPCODE_BGNLOOP) { + if (build_loop_info(c, &loop, inst)) { + try_unroll_loop(c, &loop); + } + } + } +} + +void rc_emulate_loops(struct radeon_compiler *c, void *user) +{ + struct emulate_loop_state * s = &c->loop_state; + int i; + /* Iterate backwards of the list of loops so that loops that nested + * loops are unrolled first. + */ + for( i = s->LoopCount - 1; i >= 0; i-- ){ + unsigned int iterations; + + if(!s->Loops[i].EndLoop){ + continue; + } + iterations = loop_max_possible_iterations(s->C, &s->Loops[i]); + unroll_loop(s->C, &s->Loops[i], iterations); + } +} diff --git a/src/gallium/drivers/r300/compiler/radeon_emulate_loops.h b/src/gallium/drivers/r300/compiler/radeon_emulate_loops.h new file mode 100644 index 00000000000..cd800c059d9 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_emulate_loops.h @@ -0,0 +1,32 @@ + + +#ifndef RADEON_EMULATE_LOOPS_H +#define RADEON_EMULATE_LOOPS_H + +#define MAX_ITERATIONS 8 + +struct radeon_compiler; + +struct loop_info { + struct rc_instruction * BeginLoop; + struct rc_instruction * Cond; + struct rc_instruction * If; + struct rc_instruction * Brk; + struct rc_instruction * EndIf; + struct rc_instruction * EndLoop; +}; + +struct emulate_loop_state { + struct radeon_compiler * C; + struct loop_info * Loops; + unsigned int LoopCount; + unsigned int LoopReserved; +}; + +void rc_transform_loops(struct radeon_compiler *c, void *user); + +void rc_unroll_loops(struct radeon_compiler * c, void *user); + +void rc_emulate_loops(struct radeon_compiler * c, void *user); + +#endif /* RADEON_EMULATE_LOOPS_H */ diff --git a/src/gallium/drivers/r300/compiler/radeon_list.c b/src/gallium/drivers/r300/compiler/radeon_list.c new file mode 100644 index 00000000000..811c908a81a --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_list.c @@ -0,0 +1,90 @@ +/* + * Copyright 2011 Tom Stellard <[email protected]> + * + * 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. + * + */ + +#include "radeon_list.h" + +#include <stdlib.h> +#include <stdio.h> + +#include "memory_pool.h" + +struct rc_list * rc_list(struct memory_pool * pool, void * item) +{ + struct rc_list * new = memory_pool_malloc(pool, sizeof(struct rc_list)); + new->Item = item; + new->Next = NULL; + new->Prev = NULL; + + return new; +} + +void rc_list_add(struct rc_list ** list, struct rc_list * new_value) +{ + struct rc_list * temp; + + if (*list == NULL) { + *list = new_value; + return; + } + + for (temp = *list; temp->Next; temp = temp->Next); + + temp->Next = new_value; + new_value->Prev = temp; +} + +void rc_list_remove(struct rc_list ** list, struct rc_list * rm_value) +{ + if (*list == rm_value) { + *list = rm_value->Next; + return; + } + + rm_value->Prev->Next = rm_value->Next; + if (rm_value->Next) { + rm_value->Next->Prev = rm_value->Prev; + } +} + +unsigned int rc_list_count(struct rc_list * list) +{ + unsigned int count = 0; + while (list) { + count++; + list = list->Next; + } + return count; +} + +void rc_list_print(struct rc_list * list) +{ + while(list) { + fprintf(stderr, "%p->", list->Item); + list = list->Next; + } + fprintf(stderr, "\n"); +} diff --git a/src/gallium/drivers/r300/compiler/radeon_list.h b/src/gallium/drivers/r300/compiler/radeon_list.h new file mode 100644 index 00000000000..b3c8f89cc68 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_list.h @@ -0,0 +1,46 @@ +/* + * Copyright 2011 Tom Stellard <[email protected]> + * + * 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. + * + */ + +#ifndef RADEON_LIST_H +#define RADEON_LIST_H + +struct memory_pool; + +struct rc_list { + void * Item; + struct rc_list * Prev; + struct rc_list * Next; +}; + +struct rc_list * rc_list(struct memory_pool * pool, void * item); +void rc_list_add(struct rc_list ** list, struct rc_list * new_value); +void rc_list_remove(struct rc_list ** list, struct rc_list * rm_value); +unsigned int rc_list_count(struct rc_list * list); +void rc_list_print(struct rc_list * list); + +#endif /* RADEON_LIST_H */ + diff --git a/src/gallium/drivers/r300/compiler/radeon_opcodes.c b/src/gallium/drivers/r300/compiler/radeon_opcodes.c new file mode 100644 index 00000000000..afd78ad79dd --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_opcodes.c @@ -0,0 +1,546 @@ +/* + * Copyright (C) 2009 Nicolai Haehnle. + * + * 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. + * + */ + +#include "radeon_opcodes.h" +#include "radeon_program.h" + +#include "radeon_program_constants.h" + +struct rc_opcode_info rc_opcodes[MAX_RC_OPCODE] = { + { + .Opcode = RC_OPCODE_NOP, + .Name = "NOP" + }, + { + .Opcode = RC_OPCODE_ILLEGAL_OPCODE, + .Name = "ILLEGAL OPCODE" + }, + { + .Opcode = RC_OPCODE_ABS, + .Name = "ABS", + .NumSrcRegs = 1, + .HasDstReg = 1, + .IsComponentwise = 1 + }, + { + .Opcode = RC_OPCODE_ADD, + .Name = "ADD", + .NumSrcRegs = 2, + .HasDstReg = 1, + .IsComponentwise = 1 + }, + { + .Opcode = RC_OPCODE_ARL, + .Name = "ARL", + .NumSrcRegs = 1, + .HasDstReg = 1 + }, + { + .Opcode = RC_OPCODE_CEIL, + .Name = "CEIL", + .NumSrcRegs = 1, + .HasDstReg = 1, + .IsComponentwise = 1 + }, + { + .Opcode = RC_OPCODE_CLAMP, + .Name = "CLAMP", + .NumSrcRegs = 3, + .HasDstReg = 1, + .IsComponentwise = 1 + }, + { + .Opcode = RC_OPCODE_CMP, + .Name = "CMP", + .NumSrcRegs = 3, + .HasDstReg = 1, + .IsComponentwise = 1 + }, + { + .Opcode = RC_OPCODE_CND, + .Name = "CND", + .NumSrcRegs = 3, + .HasDstReg = 1, + .IsComponentwise = 1 + }, + { + .Opcode = RC_OPCODE_COS, + .Name = "COS", + .NumSrcRegs = 1, + .HasDstReg = 1, + .IsStandardScalar = 1 + }, + { + .Opcode = RC_OPCODE_DDX, + .Name = "DDX", + .NumSrcRegs = 2, + .HasDstReg = 1, + .IsComponentwise = 1 + }, + { + .Opcode = RC_OPCODE_DDY, + .Name = "DDY", + .NumSrcRegs = 2, + .HasDstReg = 1, + .IsComponentwise = 1 + }, + { + .Opcode = RC_OPCODE_DP2, + .Name = "DP2", + .NumSrcRegs = 2, + .HasDstReg = 1 + }, + { + .Opcode = RC_OPCODE_DP3, + .Name = "DP3", + .NumSrcRegs = 2, + .HasDstReg = 1 + }, + { + .Opcode = RC_OPCODE_DP4, + .Name = "DP4", + .NumSrcRegs = 2, + .HasDstReg = 1 + }, + { + .Opcode = RC_OPCODE_DPH, + .Name = "DPH", + .NumSrcRegs = 2, + .HasDstReg = 1 + }, + { + .Opcode = RC_OPCODE_DST, + .Name = "DST", + .NumSrcRegs = 2, + .HasDstReg = 1 + }, + { + .Opcode = RC_OPCODE_EX2, + .Name = "EX2", + .NumSrcRegs = 1, + .HasDstReg = 1, + .IsStandardScalar = 1 + }, + { + .Opcode = RC_OPCODE_EXP, + .Name = "EXP", + .NumSrcRegs = 1, + .HasDstReg = 1 + }, + { + .Opcode = RC_OPCODE_FLR, + .Name = "FLR", + .NumSrcRegs = 1, + .HasDstReg = 1, + .IsComponentwise = 1 + }, + { + .Opcode = RC_OPCODE_FRC, + .Name = "FRC", + .NumSrcRegs = 1, + .HasDstReg = 1, + .IsComponentwise = 1 + }, + { + .Opcode = RC_OPCODE_KIL, + .Name = "KIL", + .NumSrcRegs = 1 + }, + { + .Opcode = RC_OPCODE_LG2, + .Name = "LG2", + .NumSrcRegs = 1, + .HasDstReg = 1, + .IsStandardScalar = 1 + }, + { + .Opcode = RC_OPCODE_LIT, + .Name = "LIT", + .NumSrcRegs = 1, + .HasDstReg = 1 + }, + { + .Opcode = RC_OPCODE_LOG, + .Name = "LOG", + .NumSrcRegs = 1, + .HasDstReg = 1 + }, + { + .Opcode = RC_OPCODE_LRP, + .Name = "LRP", + .NumSrcRegs = 3, + .HasDstReg = 1, + .IsComponentwise = 1 + }, + { + .Opcode = RC_OPCODE_MAD, + .Name = "MAD", + .NumSrcRegs = 3, + .HasDstReg = 1, + .IsComponentwise = 1 + }, + { + .Opcode = RC_OPCODE_MAX, + .Name = "MAX", + .NumSrcRegs = 2, + .HasDstReg = 1, + .IsComponentwise = 1 + }, + { + .Opcode = RC_OPCODE_MIN, + .Name = "MIN", + .NumSrcRegs = 2, + .HasDstReg = 1, + .IsComponentwise = 1 + }, + { + .Opcode = RC_OPCODE_MOV, + .Name = "MOV", + .NumSrcRegs = 1, + .HasDstReg = 1, + .IsComponentwise = 1 + }, + { + .Opcode = RC_OPCODE_MUL, + .Name = "MUL", + .NumSrcRegs = 2, + .HasDstReg = 1, + .IsComponentwise = 1 + }, + { + .Opcode = RC_OPCODE_POW, + .Name = "POW", + .NumSrcRegs = 2, + .HasDstReg = 1, + .IsStandardScalar = 1 + }, + { + .Opcode = RC_OPCODE_RCP, + .Name = "RCP", + .NumSrcRegs = 1, + .HasDstReg = 1, + .IsStandardScalar = 1 + }, + { + .Opcode = RC_OPCODE_RSQ, + .Name = "RSQ", + .NumSrcRegs = 1, + .HasDstReg = 1, + .IsStandardScalar = 1 + }, + { + .Opcode = RC_OPCODE_SCS, + .Name = "SCS", + .NumSrcRegs = 1, + .HasDstReg = 1 + }, + { + .Opcode = RC_OPCODE_SEQ, + .Name = "SEQ", + .NumSrcRegs = 2, + .HasDstReg = 1, + .IsComponentwise = 1 + }, + { + .Opcode = RC_OPCODE_SFL, + .Name = "SFL", + .NumSrcRegs = 0, + .HasDstReg = 1, + .IsComponentwise = 1 + }, + { + .Opcode = RC_OPCODE_SGE, + .Name = "SGE", + .NumSrcRegs = 2, + .HasDstReg = 1, + .IsComponentwise = 1 + }, + { + .Opcode = RC_OPCODE_SGT, + .Name = "SGT", + .NumSrcRegs = 2, + .HasDstReg = 1, + .IsComponentwise = 1 + }, + { + .Opcode = RC_OPCODE_SIN, + .Name = "SIN", + .NumSrcRegs = 1, + .HasDstReg = 1, + .IsStandardScalar = 1 + }, + { + .Opcode = RC_OPCODE_SLE, + .Name = "SLE", + .NumSrcRegs = 2, + .HasDstReg = 1, + .IsComponentwise = 1 + }, + { + .Opcode = RC_OPCODE_SLT, + .Name = "SLT", + .NumSrcRegs = 2, + .HasDstReg = 1, + .IsComponentwise = 1 + }, + { + .Opcode = RC_OPCODE_SNE, + .Name = "SNE", + .NumSrcRegs = 2, + .HasDstReg = 1, + .IsComponentwise = 1 + }, + { + .Opcode = RC_OPCODE_SSG, + .Name = "SSG", + .NumSrcRegs = 1, + .HasDstReg = 1, + .IsComponentwise = 1 + }, + { + .Opcode = RC_OPCODE_SUB, + .Name = "SUB", + .NumSrcRegs = 2, + .HasDstReg = 1, + .IsComponentwise = 1 + }, + { + .Opcode = RC_OPCODE_SWZ, + .Name = "SWZ", + .NumSrcRegs = 1, + .HasDstReg = 1, + .IsComponentwise = 1 + }, + { + .Opcode = RC_OPCODE_XPD, + .Name = "XPD", + .NumSrcRegs = 2, + .HasDstReg = 1 + }, + { + .Opcode = RC_OPCODE_TEX, + .Name = "TEX", + .HasTexture = 1, + .NumSrcRegs = 1, + .HasDstReg = 1 + }, + { + .Opcode = RC_OPCODE_TXB, + .Name = "TXB", + .HasTexture = 1, + .NumSrcRegs = 1, + .HasDstReg = 1 + }, + { + .Opcode = RC_OPCODE_TXD, + .Name = "TXD", + .HasTexture = 1, + .NumSrcRegs = 3, + .HasDstReg = 1 + }, + { + .Opcode = RC_OPCODE_TXL, + .Name = "TXL", + .HasTexture = 1, + .NumSrcRegs = 1, + .HasDstReg = 1 + }, + { + .Opcode = RC_OPCODE_TXP, + .Name = "TXP", + .HasTexture = 1, + .NumSrcRegs = 1, + .HasDstReg = 1 + }, + { + .Opcode = RC_OPCODE_IF, + .Name = "IF", + .IsFlowControl = 1, + .NumSrcRegs = 1 + }, + { + .Opcode = RC_OPCODE_ELSE, + .Name = "ELSE", + .IsFlowControl = 1, + .NumSrcRegs = 0 + }, + { + .Opcode = RC_OPCODE_ENDIF, + .Name = "ENDIF", + .IsFlowControl = 1, + .NumSrcRegs = 0 + }, + { + .Opcode = RC_OPCODE_BGNLOOP, + .Name = "BGNLOOP", + .IsFlowControl = 1, + .NumSrcRegs = 0 + }, + { + .Opcode = RC_OPCODE_BRK, + .Name = "BRK", + .IsFlowControl = 1, + .NumSrcRegs = 0 + }, + { + .Opcode = RC_OPCODE_ENDLOOP, + .Name = "ENDLOOP", + .IsFlowControl = 1, + .NumSrcRegs = 0, + }, + { + .Opcode = RC_OPCODE_CONT, + .Name = "CONT", + .IsFlowControl = 1, + .NumSrcRegs = 0 + }, + { + .Opcode = RC_OPCODE_REPL_ALPHA, + .Name = "REPL_ALPHA", + .HasDstReg = 1 + }, + { + .Opcode = RC_OPCODE_BEGIN_TEX, + .Name = "BEGIN_TEX" + }, + { + .Opcode = RC_OPCODE_KILP, + .Name = "KILP", + } +}; + +void rc_compute_sources_for_writemask( + const struct rc_instruction *inst, + unsigned int writemask, + unsigned int *srcmasks) +{ + const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode); + srcmasks[0] = 0; + srcmasks[1] = 0; + srcmasks[2] = 0; + + if (opcode->Opcode == RC_OPCODE_KIL) + srcmasks[0] |= RC_MASK_XYZW; + else if (opcode->Opcode == RC_OPCODE_IF) + srcmasks[0] |= RC_MASK_X; + + if (!writemask) + return; + + if (opcode->IsComponentwise) { + for(unsigned int src = 0; src < opcode->NumSrcRegs; ++src) + srcmasks[src] |= writemask; + } else if (opcode->IsStandardScalar) { + for(unsigned int src = 0; src < opcode->NumSrcRegs; ++src) + srcmasks[src] |= RC_MASK_X; + } else { + switch(opcode->Opcode) { + case RC_OPCODE_ARL: + srcmasks[0] |= RC_MASK_X; + break; + case RC_OPCODE_DP2: + srcmasks[0] |= RC_MASK_XY; + srcmasks[1] |= RC_MASK_XY; + break; + case RC_OPCODE_DP3: + case RC_OPCODE_XPD: + srcmasks[0] |= RC_MASK_XYZ; + srcmasks[1] |= RC_MASK_XYZ; + break; + case RC_OPCODE_DP4: + srcmasks[0] |= RC_MASK_XYZW; + srcmasks[1] |= RC_MASK_XYZW; + break; + case RC_OPCODE_DPH: + srcmasks[0] |= RC_MASK_XYZ; + srcmasks[1] |= RC_MASK_XYZW; + break; + case RC_OPCODE_TXB: + case RC_OPCODE_TXP: + case RC_OPCODE_TXL: + srcmasks[0] |= RC_MASK_W; + /* Fall through */ + case RC_OPCODE_TEX: + switch (inst->U.I.TexSrcTarget) { + case RC_TEXTURE_1D: + srcmasks[0] |= RC_MASK_X; + break; + case RC_TEXTURE_2D: + case RC_TEXTURE_RECT: + case RC_TEXTURE_1D_ARRAY: + srcmasks[0] |= RC_MASK_XY; + break; + case RC_TEXTURE_3D: + case RC_TEXTURE_CUBE: + case RC_TEXTURE_2D_ARRAY: + srcmasks[0] |= RC_MASK_XYZ; + break; + } + break; + case RC_OPCODE_TXD: + switch (inst->U.I.TexSrcTarget) { + case RC_TEXTURE_1D_ARRAY: + srcmasks[0] |= RC_MASK_Y; + /* Fall through. */ + case RC_TEXTURE_1D: + srcmasks[0] |= RC_MASK_X; + srcmasks[1] |= RC_MASK_X; + srcmasks[2] |= RC_MASK_X; + break; + case RC_TEXTURE_2D_ARRAY: + srcmasks[0] |= RC_MASK_Z; + /* Fall through. */ + case RC_TEXTURE_2D: + case RC_TEXTURE_RECT: + srcmasks[0] |= RC_MASK_XY; + srcmasks[1] |= RC_MASK_XY; + srcmasks[2] |= RC_MASK_XY; + break; + case RC_TEXTURE_3D: + case RC_TEXTURE_CUBE: + srcmasks[0] |= RC_MASK_XYZ; + srcmasks[1] |= RC_MASK_XYZ; + srcmasks[2] |= RC_MASK_XYZ; + break; + } + break; + case RC_OPCODE_DST: + srcmasks[0] |= RC_MASK_Y | RC_MASK_Z; + srcmasks[1] |= RC_MASK_Y | RC_MASK_W; + break; + case RC_OPCODE_EXP: + case RC_OPCODE_LOG: + srcmasks[0] |= RC_MASK_XY; + break; + case RC_OPCODE_LIT: + srcmasks[0] |= RC_MASK_X | RC_MASK_Y | RC_MASK_W; + break; + default: + break; + } + } +} diff --git a/src/gallium/drivers/r300/compiler/radeon_opcodes.h b/src/gallium/drivers/r300/compiler/radeon_opcodes.h new file mode 100644 index 00000000000..b5868820611 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_opcodes.h @@ -0,0 +1,263 @@ +/* + * Copyright (C) 2009 Nicolai Haehnle. + * + * 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. + * + */ + +#ifndef RADEON_OPCODES_H +#define RADEON_OPCODES_H + +#include <assert.h> + +/** + * Opcodes understood by the Radeon compiler. + */ +typedef enum { + RC_OPCODE_NOP = 0, + RC_OPCODE_ILLEGAL_OPCODE, + + /** vec4 instruction: dst.c = abs(src0.c); */ + RC_OPCODE_ABS, + + /** vec4 instruction: dst.c = src0.c + src1.c; */ + RC_OPCODE_ADD, + + /** special instruction: load address register + * dst.x = floor(src.x), where dst must be an address register */ + RC_OPCODE_ARL, + + /** vec4 instruction: dst.c = ceil(src0.c) */ + RC_OPCODE_CEIL, + + /** vec4 instruction: dst.c = clamp(src0.c, src1.c, src2.c) */ + RC_OPCODE_CLAMP, + + /** vec4 instruction: dst.c = src0.c < 0.0 ? src1.c : src2.c */ + RC_OPCODE_CMP, + + /** vec4 instruction: dst.c = src2.c > 0.5 ? src0.c : src1.c */ + RC_OPCODE_CND, + + /** scalar instruction: dst = cos(src0.x) */ + RC_OPCODE_COS, + + /** special instruction: take vec4 partial derivative in X direction + * dst.c = d src0.c / dx */ + RC_OPCODE_DDX, + + /** special instruction: take vec4 partial derivative in Y direction + * dst.c = d src0.c / dy */ + RC_OPCODE_DDY, + + /** scalar instruction: dst = src0.x*src1.x + src0.y*src1.y */ + RC_OPCODE_DP2, + + /** scalar instruction: dst = src0.x*src1.x + src0.y*src1.y + src0.z*src1.z */ + RC_OPCODE_DP3, + + /** scalar instruction: dst = src0.x*src1.x + src0.y*src1.y + src0.z*src1.z + src0.w*src1.w */ + RC_OPCODE_DP4, + + /** scalar instruction: dst = src0.x*src1.x + src0.y*src1.y + src0.z*src1.z + src1.w */ + RC_OPCODE_DPH, + + /** special instruction, see ARB_fragment_program */ + RC_OPCODE_DST, + + /** scalar instruction: dst = 2**src0.x */ + RC_OPCODE_EX2, + + /** special instruction, see ARB_vertex_program */ + RC_OPCODE_EXP, + + /** vec4 instruction: dst.c = floor(src0.c) */ + RC_OPCODE_FLR, + + /** vec4 instruction: dst.c = src0.c - floor(src0.c) */ + RC_OPCODE_FRC, + + /** special instruction: stop execution if any component of src0 is negative */ + RC_OPCODE_KIL, + + /** scalar instruction: dst = log_2(src0.x) */ + RC_OPCODE_LG2, + + /** special instruction, see ARB_vertex_program */ + RC_OPCODE_LIT, + + /** special instruction, see ARB_vertex_program */ + RC_OPCODE_LOG, + + /** vec4 instruction: dst.c = src0.c*src1.c + (1 - src0.c)*src2.c */ + RC_OPCODE_LRP, + + /** vec4 instruction: dst.c = src0.c*src1.c + src2.c */ + RC_OPCODE_MAD, + + /** vec4 instruction: dst.c = max(src0.c, src1.c) */ + RC_OPCODE_MAX, + + /** vec4 instruction: dst.c = min(src0.c, src1.c) */ + RC_OPCODE_MIN, + + /** vec4 instruction: dst.c = src0.c */ + RC_OPCODE_MOV, + + /** vec4 instruction: dst.c = src0.c*src1.c */ + RC_OPCODE_MUL, + + /** scalar instruction: dst = src0.x ** src1.x */ + RC_OPCODE_POW, + + /** scalar instruction: dst = 1 / src0.x */ + RC_OPCODE_RCP, + + /** scalar instruction: dst = 1 / sqrt(src0.x) */ + RC_OPCODE_RSQ, + + /** special instruction, see ARB_fragment_program */ + RC_OPCODE_SCS, + + /** vec4 instruction: dst.c = (src0.c == src1.c) ? 1.0 : 0.0 */ + RC_OPCODE_SEQ, + + /** vec4 instruction: dst.c = 0.0 */ + RC_OPCODE_SFL, + + /** vec4 instruction: dst.c = (src0.c >= src1.c) ? 1.0 : 0.0 */ + RC_OPCODE_SGE, + + /** vec4 instruction: dst.c = (src0.c > src1.c) ? 1.0 : 0.0 */ + RC_OPCODE_SGT, + + /** scalar instruction: dst = sin(src0.x) */ + RC_OPCODE_SIN, + + /** vec4 instruction: dst.c = (src0.c <= src1.c) ? 1.0 : 0.0 */ + RC_OPCODE_SLE, + + /** vec4 instruction: dst.c = (src0.c < src1.c) ? 1.0 : 0.0 */ + RC_OPCODE_SLT, + + /** vec4 instruction: dst.c = (src0.c != src1.c) ? 1.0 : 0.0 */ + RC_OPCODE_SNE, + + /** vec4 instruction: dst.c = (src0.c < 0 ?) -1 : ((src0.c > 0) : 1 : 0) */ + RC_OPCODE_SSG, + + /** vec4 instruction: dst.c = src0.c - src1.c */ + RC_OPCODE_SUB, + + /** vec4 instruction: dst.c = src0.c */ + RC_OPCODE_SWZ, + + /** special instruction, see ARB_fragment_program */ + RC_OPCODE_XPD, + + RC_OPCODE_TEX, + RC_OPCODE_TXB, + RC_OPCODE_TXD, + RC_OPCODE_TXL, + RC_OPCODE_TXP, + + /** branch instruction: + * If src0.x != 0.0, continue with the next instruction; + * otherwise, jump to matching RC_OPCODE_ELSE or RC_OPCODE_ENDIF. + */ + RC_OPCODE_IF, + + /** branch instruction: jump to matching RC_OPCODE_ENDIF */ + RC_OPCODE_ELSE, + + /** branch instruction: has no effect */ + RC_OPCODE_ENDIF, + + RC_OPCODE_BGNLOOP, + + RC_OPCODE_BRK, + + RC_OPCODE_ENDLOOP, + + RC_OPCODE_CONT, + + /** special instruction, used in R300-R500 fragment program pair instructions + * indicates that the result of the alpha operation shall be replicated + * across all other channels */ + RC_OPCODE_REPL_ALPHA, + + /** special instruction, used in R300-R500 fragment programs + * to indicate the start of a block of texture instructions that + * can run simultaneously. */ + RC_OPCODE_BEGIN_TEX, + + /** Stop execution of the shader (GLSL discard) */ + RC_OPCODE_KILP, + + MAX_RC_OPCODE +} rc_opcode; + + +struct rc_opcode_info { + rc_opcode Opcode; + const char * Name; + + /** true if the instruction reads from a texture. + * + * \note This is false for the KIL instruction, even though KIL is + * a texture instruction from a hardware point of view. */ + unsigned int HasTexture:1; + + unsigned int NumSrcRegs:2; + unsigned int HasDstReg:1; + + /** true if this instruction affects control flow */ + unsigned int IsFlowControl:1; + + /** true if this is a vector instruction that operates on components in parallel + * without any cross-component interaction */ + unsigned int IsComponentwise:1; + + /** true if this instruction sources only its operands X components + * to compute one result which is smeared across all output channels */ + unsigned int IsStandardScalar:1; +}; + +extern struct rc_opcode_info rc_opcodes[MAX_RC_OPCODE]; + +static inline const struct rc_opcode_info * rc_get_opcode_info(rc_opcode opcode) +{ + assert((unsigned int)opcode < MAX_RC_OPCODE); + assert(rc_opcodes[opcode].Opcode == opcode); + + return &rc_opcodes[opcode]; +} + +struct rc_instruction; + +void rc_compute_sources_for_writemask( + const struct rc_instruction *inst, + unsigned int writemask, + unsigned int *srcmasks); + +#endif /* RADEON_OPCODES_H */ diff --git a/src/gallium/drivers/r300/compiler/radeon_optimize.c b/src/gallium/drivers/r300/compiler/radeon_optimize.c new file mode 100644 index 00000000000..39dcb21d4f4 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_optimize.c @@ -0,0 +1,700 @@ +/* + * Copyright (C) 2009 Nicolai Haehnle. + * Copyright 2010 Tom Stellard <[email protected]> + * + * 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. + * + */ + +#include "radeon_dataflow.h" + +#include "radeon_compiler.h" +#include "radeon_compiler_util.h" +#include "radeon_swizzle.h" + +struct src_clobbered_reads_cb_data { + rc_register_file File; + unsigned int Index; + unsigned int Mask; + struct rc_reader_data * ReaderData; +}; + +typedef void (*rc_presub_replace_fn)(struct rc_instruction *, + struct rc_instruction *, + unsigned int); + +static struct rc_src_register chain_srcregs(struct rc_src_register outer, struct rc_src_register inner) +{ + struct rc_src_register combine; + combine.File = inner.File; + combine.Index = inner.Index; + combine.RelAddr = inner.RelAddr; + if (outer.Abs) { + combine.Abs = 1; + combine.Negate = outer.Negate; + } else { + combine.Abs = inner.Abs; + combine.Negate = swizzle_mask(outer.Swizzle, inner.Negate); + combine.Negate ^= outer.Negate; + } + combine.Swizzle = combine_swizzles(inner.Swizzle, outer.Swizzle); + return combine; +} + +static void copy_propagate_scan_read(void * data, struct rc_instruction * inst, + struct rc_src_register * src) +{ + rc_register_file file = src->File; + struct rc_reader_data * reader_data = data; + + if(!rc_inst_can_use_presub(inst, + reader_data->Writer->U.I.PreSub.Opcode, + rc_swizzle_to_writemask(src->Swizzle), + src, + &reader_data->Writer->U.I.PreSub.SrcReg[0], + &reader_data->Writer->U.I.PreSub.SrcReg[1])) { + reader_data->Abort = 1; + return; + } + + /* XXX This could probably be handled better. */ + if (file == RC_FILE_ADDRESS) { + reader_data->Abort = 1; + return; + } + + /* These instructions cannot read from the constants file. + * see radeonTransformTEX() + */ + if(reader_data->Writer->U.I.SrcReg[0].File != RC_FILE_TEMPORARY && + reader_data->Writer->U.I.SrcReg[0].File != RC_FILE_INPUT && + (inst->U.I.Opcode == RC_OPCODE_TEX || + inst->U.I.Opcode == RC_OPCODE_TXB || + inst->U.I.Opcode == RC_OPCODE_TXP || + inst->U.I.Opcode == RC_OPCODE_TXD || + inst->U.I.Opcode == RC_OPCODE_TXL || + inst->U.I.Opcode == RC_OPCODE_KIL)){ + reader_data->Abort = 1; + return; + } +} + +static void src_clobbered_reads_cb( + void * data, + struct rc_instruction * inst, + struct rc_src_register * src) +{ + struct src_clobbered_reads_cb_data * sc_data = data; + + if (src->File == sc_data->File + && src->Index == sc_data->Index + && (rc_swizzle_to_writemask(src->Swizzle) & sc_data->Mask)) { + + sc_data->ReaderData->AbortOnRead = RC_MASK_XYZW; + } + + if (src->RelAddr && sc_data->File == RC_FILE_ADDRESS) { + sc_data->ReaderData->AbortOnRead = RC_MASK_XYZW; + } +} + +static void is_src_clobbered_scan_write( + void * data, + struct rc_instruction * inst, + rc_register_file file, + unsigned int index, + unsigned int mask) +{ + struct src_clobbered_reads_cb_data sc_data; + struct rc_reader_data * reader_data = data; + sc_data.File = file; + sc_data.Index = index; + sc_data.Mask = mask; + sc_data.ReaderData = reader_data; + rc_for_all_reads_src(reader_data->Writer, + src_clobbered_reads_cb, &sc_data); +} + +static void copy_propagate(struct radeon_compiler * c, struct rc_instruction * inst_mov) +{ + struct rc_reader_data reader_data; + unsigned int i; + + if (inst_mov->U.I.DstReg.File != RC_FILE_TEMPORARY || + inst_mov->U.I.WriteALUResult || + inst_mov->U.I.SaturateMode) + return; + + /* Get a list of all the readers of this MOV instruction. */ + reader_data.ExitOnAbort = 1; + rc_get_readers(c, inst_mov, &reader_data, + copy_propagate_scan_read, NULL, + is_src_clobbered_scan_write); + + if (reader_data.Abort || reader_data.ReaderCount == 0) + return; + + /* Propagate the MOV instruction. */ + for (i = 0; i < reader_data.ReaderCount; i++) { + struct rc_instruction * inst = reader_data.Readers[i].Inst; + *reader_data.Readers[i].U.I.Src = chain_srcregs(*reader_data.Readers[i].U.I.Src, inst_mov->U.I.SrcReg[0]); + + if (inst_mov->U.I.SrcReg[0].File == RC_FILE_PRESUB) + inst->U.I.PreSub = inst_mov->U.I.PreSub; + } + + /* Finally, remove the original MOV instruction */ + rc_remove_instruction(inst_mov); +} + +/** + * Check if a source register is actually always the same + * swizzle constant. + */ +static int is_src_uniform_constant(struct rc_src_register src, + rc_swizzle * pswz, unsigned int * pnegate) +{ + int have_used = 0; + + if (src.File != RC_FILE_NONE) { + *pswz = 0; + return 0; + } + + for(unsigned int chan = 0; chan < 4; ++chan) { + unsigned int swz = GET_SWZ(src.Swizzle, chan); + if (swz < 4) { + *pswz = 0; + return 0; + } + if (swz == RC_SWIZZLE_UNUSED) + continue; + + if (!have_used) { + *pswz = swz; + *pnegate = GET_BIT(src.Negate, chan); + have_used = 1; + } else { + if (swz != *pswz || *pnegate != GET_BIT(src.Negate, chan)) { + *pswz = 0; + return 0; + } + } + } + + return 1; +} + +static void constant_folding_mad(struct rc_instruction * inst) +{ + rc_swizzle swz = 0; + unsigned int negate= 0; + + if (is_src_uniform_constant(inst->U.I.SrcReg[2], &swz, &negate)) { + if (swz == RC_SWIZZLE_ZERO) { + inst->U.I.Opcode = RC_OPCODE_MUL; + return; + } + } + + if (is_src_uniform_constant(inst->U.I.SrcReg[1], &swz, &negate)) { + if (swz == RC_SWIZZLE_ONE) { + inst->U.I.Opcode = RC_OPCODE_ADD; + if (negate) + inst->U.I.SrcReg[0].Negate ^= RC_MASK_XYZW; + inst->U.I.SrcReg[1] = inst->U.I.SrcReg[2]; + return; + } else if (swz == RC_SWIZZLE_ZERO) { + inst->U.I.Opcode = RC_OPCODE_MOV; + inst->U.I.SrcReg[0] = inst->U.I.SrcReg[2]; + return; + } + } + + if (is_src_uniform_constant(inst->U.I.SrcReg[0], &swz, &negate)) { + if (swz == RC_SWIZZLE_ONE) { + inst->U.I.Opcode = RC_OPCODE_ADD; + if (negate) + inst->U.I.SrcReg[1].Negate ^= RC_MASK_XYZW; + inst->U.I.SrcReg[0] = inst->U.I.SrcReg[2]; + return; + } else if (swz == RC_SWIZZLE_ZERO) { + inst->U.I.Opcode = RC_OPCODE_MOV; + inst->U.I.SrcReg[0] = inst->U.I.SrcReg[2]; + return; + } + } +} + +static void constant_folding_mul(struct rc_instruction * inst) +{ + rc_swizzle swz = 0; + unsigned int negate = 0; + + if (is_src_uniform_constant(inst->U.I.SrcReg[0], &swz, &negate)) { + if (swz == RC_SWIZZLE_ONE) { + inst->U.I.Opcode = RC_OPCODE_MOV; + inst->U.I.SrcReg[0] = inst->U.I.SrcReg[1]; + if (negate) + inst->U.I.SrcReg[0].Negate ^= RC_MASK_XYZW; + return; + } else if (swz == RC_SWIZZLE_ZERO) { + inst->U.I.Opcode = RC_OPCODE_MOV; + inst->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_0000; + return; + } + } + + if (is_src_uniform_constant(inst->U.I.SrcReg[1], &swz, &negate)) { + if (swz == RC_SWIZZLE_ONE) { + inst->U.I.Opcode = RC_OPCODE_MOV; + if (negate) + inst->U.I.SrcReg[0].Negate ^= RC_MASK_XYZW; + return; + } else if (swz == RC_SWIZZLE_ZERO) { + inst->U.I.Opcode = RC_OPCODE_MOV; + inst->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_0000; + return; + } + } +} + +static void constant_folding_add(struct rc_instruction * inst) +{ + rc_swizzle swz = 0; + unsigned int negate = 0; + + if (is_src_uniform_constant(inst->U.I.SrcReg[0], &swz, &negate)) { + if (swz == RC_SWIZZLE_ZERO) { + inst->U.I.Opcode = RC_OPCODE_MOV; + inst->U.I.SrcReg[0] = inst->U.I.SrcReg[1]; + return; + } + } + + if (is_src_uniform_constant(inst->U.I.SrcReg[1], &swz, &negate)) { + if (swz == RC_SWIZZLE_ZERO) { + inst->U.I.Opcode = RC_OPCODE_MOV; + return; + } + } +} + +/** + * Replace 0.0, 1.0 and 0.5 immediate constants by their + * respective swizzles. Simplify instructions like ADD dst, src, 0; + */ +static void constant_folding(struct radeon_compiler * c, struct rc_instruction * inst) +{ + const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode); + unsigned int i; + + /* Replace 0.0, 1.0 and 0.5 immediates by their explicit swizzles */ + for(unsigned int src = 0; src < opcode->NumSrcRegs; ++src) { + struct rc_constant * constant; + struct rc_src_register newsrc; + int have_real_reference; + unsigned int chan; + + /* If there are only 0, 0.5, 1, or _ swizzles, mark the source as a constant. */ + for (chan = 0; chan < 4; ++chan) + if (GET_SWZ(inst->U.I.SrcReg[src].Swizzle, chan) <= 3) + break; + if (chan == 4) { + inst->U.I.SrcReg[src].File = RC_FILE_NONE; + continue; + } + + /* Convert immediates to swizzles. */ + if (inst->U.I.SrcReg[src].File != RC_FILE_CONSTANT || + inst->U.I.SrcReg[src].RelAddr || + inst->U.I.SrcReg[src].Index >= c->Program.Constants.Count) + continue; + + constant = + &c->Program.Constants.Constants[inst->U.I.SrcReg[src].Index]; + + if (constant->Type != RC_CONSTANT_IMMEDIATE) + continue; + + newsrc = inst->U.I.SrcReg[src]; + have_real_reference = 0; + for (chan = 0; chan < 4; ++chan) { + unsigned int swz = GET_SWZ(newsrc.Swizzle, chan); + unsigned int newswz; + float imm; + float baseimm; + + if (swz >= 4) + continue; + + imm = constant->u.Immediate[swz]; + baseimm = imm; + if (imm < 0.0) + baseimm = -baseimm; + + if (baseimm == 0.0) { + newswz = RC_SWIZZLE_ZERO; + } else if (baseimm == 1.0) { + newswz = RC_SWIZZLE_ONE; + } else if (baseimm == 0.5 && c->has_half_swizzles) { + newswz = RC_SWIZZLE_HALF; + } else { + have_real_reference = 1; + continue; + } + + SET_SWZ(newsrc.Swizzle, chan, newswz); + if (imm < 0.0 && !newsrc.Abs) + newsrc.Negate ^= 1 << chan; + } + + if (!have_real_reference) { + newsrc.File = RC_FILE_NONE; + newsrc.Index = 0; + } + + /* don't make the swizzle worse */ + if (!c->SwizzleCaps->IsNative(inst->U.I.Opcode, newsrc) && + c->SwizzleCaps->IsNative(inst->U.I.Opcode, inst->U.I.SrcReg[src])) + continue; + + inst->U.I.SrcReg[src] = newsrc; + } + + /* Simplify instructions based on constants */ + if (inst->U.I.Opcode == RC_OPCODE_MAD) + constant_folding_mad(inst); + + /* note: MAD can simplify to MUL or ADD */ + if (inst->U.I.Opcode == RC_OPCODE_MUL) + constant_folding_mul(inst); + else if (inst->U.I.Opcode == RC_OPCODE_ADD) + constant_folding_add(inst); + + /* In case this instruction has been converted, make sure all of the + * registers that are no longer used are empty. */ + opcode = rc_get_opcode_info(inst->U.I.Opcode); + for(i = opcode->NumSrcRegs; i < 3; i++) { + memset(&inst->U.I.SrcReg[i], 0, sizeof(struct rc_src_register)); + } +} + +/** + * If src and dst use the same register, this function returns a writemask that + * indicates wich components are read by src. Otherwise zero is returned. + */ +static unsigned int src_reads_dst_mask(struct rc_src_register src, + struct rc_dst_register dst) +{ + if (dst.File != src.File || dst.Index != src.Index) { + return 0; + } + return rc_swizzle_to_writemask(src.Swizzle); +} + +/* Return 1 if the source registers has a constant swizzle (e.g. 0, 0.5, 1.0) + * in any of its channels. Return 0 otherwise. */ +static int src_has_const_swz(struct rc_src_register src) { + int chan; + for(chan = 0; chan < 4; chan++) { + unsigned int swz = GET_SWZ(src.Swizzle, chan); + if (swz == RC_SWIZZLE_ZERO || swz == RC_SWIZZLE_HALF + || swz == RC_SWIZZLE_ONE) { + return 1; + } + } + return 0; +} + +static void presub_scan_read( + void * data, + struct rc_instruction * inst, + struct rc_src_register * src) +{ + struct rc_reader_data * reader_data = data; + rc_presubtract_op * presub_opcode = reader_data->CbData; + + if (!rc_inst_can_use_presub(inst, *presub_opcode, + reader_data->Writer->U.I.DstReg.WriteMask, + src, + &reader_data->Writer->U.I.SrcReg[0], + &reader_data->Writer->U.I.SrcReg[1])) { + reader_data->Abort = 1; + return; + } +} + +static int presub_helper( + struct radeon_compiler * c, + struct rc_instruction * inst_add, + rc_presubtract_op presub_opcode, + rc_presub_replace_fn presub_replace) +{ + struct rc_reader_data reader_data; + unsigned int i; + rc_presubtract_op cb_op = presub_opcode; + + reader_data.CbData = &cb_op; + reader_data.ExitOnAbort = 1; + rc_get_readers(c, inst_add, &reader_data, presub_scan_read, NULL, + is_src_clobbered_scan_write); + + if (reader_data.Abort || reader_data.ReaderCount == 0) + return 0; + + for(i = 0; i < reader_data.ReaderCount; i++) { + unsigned int src_index; + struct rc_reader reader = reader_data.Readers[i]; + const struct rc_opcode_info * info = + rc_get_opcode_info(reader.Inst->U.I.Opcode); + + for (src_index = 0; src_index < info->NumSrcRegs; src_index++) { + if (&reader.Inst->U.I.SrcReg[src_index] == reader.U.I.Src) + presub_replace(inst_add, reader.Inst, src_index); + } + } + return 1; +} + +/* This function assumes that inst_add->U.I.SrcReg[0] and + * inst_add->U.I.SrcReg[1] aren't both negative. */ +static void presub_replace_add( + struct rc_instruction * inst_add, + struct rc_instruction * inst_reader, + unsigned int src_index) +{ + rc_presubtract_op presub_opcode; + if (inst_add->U.I.SrcReg[1].Negate || inst_add->U.I.SrcReg[0].Negate) + presub_opcode = RC_PRESUB_SUB; + else + presub_opcode = RC_PRESUB_ADD; + + if (inst_add->U.I.SrcReg[1].Negate) { + inst_reader->U.I.PreSub.SrcReg[0] = inst_add->U.I.SrcReg[1]; + inst_reader->U.I.PreSub.SrcReg[1] = inst_add->U.I.SrcReg[0]; + } else { + inst_reader->U.I.PreSub.SrcReg[0] = inst_add->U.I.SrcReg[0]; + inst_reader->U.I.PreSub.SrcReg[1] = inst_add->U.I.SrcReg[1]; + } + inst_reader->U.I.PreSub.SrcReg[0].Negate = 0; + inst_reader->U.I.PreSub.SrcReg[1].Negate = 0; + inst_reader->U.I.PreSub.Opcode = presub_opcode; + inst_reader->U.I.SrcReg[src_index] = + chain_srcregs(inst_reader->U.I.SrcReg[src_index], + inst_reader->U.I.PreSub.SrcReg[0]); + inst_reader->U.I.SrcReg[src_index].File = RC_FILE_PRESUB; + inst_reader->U.I.SrcReg[src_index].Index = presub_opcode; +} + +static int is_presub_candidate( + struct radeon_compiler * c, + struct rc_instruction * inst) +{ + const struct rc_opcode_info * info = rc_get_opcode_info(inst->U.I.Opcode); + unsigned int i; + unsigned int is_constant[2] = {0, 0}; + + assert(inst->U.I.Opcode == RC_OPCODE_ADD); + + if (inst->U.I.PreSub.Opcode != RC_PRESUB_NONE + || inst->U.I.SaturateMode + || inst->U.I.WriteALUResult) { + return 0; + } + + /* If both sources use a constant swizzle, then we can't convert it to + * a presubtract operation. In fact for the ADD and SUB presubtract + * operations neither source can contain a constant swizzle. This + * specific case is checked in peephole_add_presub_add() when + * we make sure the swizzles for both sources are equal, so we + * don't need to worry about it here. */ + for (i = 0; i < 2; i++) { + int chan; + for (chan = 0; chan < 4; chan++) { + rc_swizzle swz = + get_swz(inst->U.I.SrcReg[i].Swizzle, chan); + if (swz == RC_SWIZZLE_ONE + || swz == RC_SWIZZLE_ZERO + || swz == RC_SWIZZLE_HALF) { + is_constant[i] = 1; + } + } + } + if (is_constant[0] && is_constant[1]) + return 0; + + for(i = 0; i < info->NumSrcRegs; i++) { + struct rc_src_register src = inst->U.I.SrcReg[i]; + if (src_reads_dst_mask(src, inst->U.I.DstReg)) + return 0; + + src.File = RC_FILE_PRESUB; + if (!c->SwizzleCaps->IsNative(inst->U.I.Opcode, src)) + return 0; + } + return 1; +} + +static int peephole_add_presub_add( + struct radeon_compiler * c, + struct rc_instruction * inst_add) +{ + unsigned dstmask = inst_add->U.I.DstReg.WriteMask; + unsigned src0_neg = inst_add->U.I.SrcReg[0].Negate & dstmask; + unsigned src1_neg = inst_add->U.I.SrcReg[1].Negate & dstmask; + + if (inst_add->U.I.SrcReg[0].Swizzle != inst_add->U.I.SrcReg[1].Swizzle) + return 0; + + /* src0 and src1 can't have absolute values */ + if (inst_add->U.I.SrcReg[0].Abs || inst_add->U.I.SrcReg[1].Abs) + return 0; + + /* presub_replace_add() assumes only one is negative */ + if (inst_add->U.I.SrcReg[0].Negate && inst_add->U.I.SrcReg[1].Negate) + return 0; + + /* if src0 is negative, at least all bits of dstmask have to be set */ + if (inst_add->U.I.SrcReg[0].Negate && src0_neg != dstmask) + return 0; + + /* if src1 is negative, at least all bits of dstmask have to be set */ + if (inst_add->U.I.SrcReg[1].Negate && src1_neg != dstmask) + return 0; + + if (!is_presub_candidate(c, inst_add)) + return 0; + + if (presub_helper(c, inst_add, RC_PRESUB_ADD, presub_replace_add)) { + rc_remove_instruction(inst_add); + return 1; + } + return 0; +} + +static void presub_replace_inv( + struct rc_instruction * inst_add, + struct rc_instruction * inst_reader, + unsigned int src_index) +{ + /* We must be careful not to modify inst_add, since it + * is possible it will remain part of the program.*/ + inst_reader->U.I.PreSub.SrcReg[0] = inst_add->U.I.SrcReg[1]; + inst_reader->U.I.PreSub.SrcReg[0].Negate = 0; + inst_reader->U.I.PreSub.Opcode = RC_PRESUB_INV; + inst_reader->U.I.SrcReg[src_index] = chain_srcregs(inst_reader->U.I.SrcReg[src_index], + inst_reader->U.I.PreSub.SrcReg[0]); + + inst_reader->U.I.SrcReg[src_index].File = RC_FILE_PRESUB; + inst_reader->U.I.SrcReg[src_index].Index = RC_PRESUB_INV; +} + +/** + * PRESUB_INV: ADD TEMP[0], none.1, -TEMP[1] + * Use the presubtract 1 - src0 for all readers of TEMP[0]. The first source + * of the add instruction must have the constatnt 1 swizzle. This function + * does not check const registers to see if their value is 1.0, so it should + * be called after the constant_folding optimization. + * @return + * 0 if the ADD instruction is still part of the program. + * 1 if the ADD instruction is no longer part of the program. + */ +static int peephole_add_presub_inv( + struct radeon_compiler * c, + struct rc_instruction * inst_add) +{ + unsigned int i, swz; + + if (!is_presub_candidate(c, inst_add)) + return 0; + + /* Check if src0 is 1. */ + /* XXX It would be nice to use is_src_uniform_constant here, but that + * function only works if the register's file is RC_FILE_NONE */ + for(i = 0; i < 4; i++ ) { + swz = GET_SWZ(inst_add->U.I.SrcReg[0].Swizzle, i); + if(((1 << i) & inst_add->U.I.DstReg.WriteMask) + && swz != RC_SWIZZLE_ONE) { + return 0; + } + } + + /* Check src1. */ + if ((inst_add->U.I.SrcReg[1].Negate & inst_add->U.I.DstReg.WriteMask) != + inst_add->U.I.DstReg.WriteMask + || inst_add->U.I.SrcReg[1].Abs + || (inst_add->U.I.SrcReg[1].File != RC_FILE_TEMPORARY + && inst_add->U.I.SrcReg[1].File != RC_FILE_CONSTANT) + || src_has_const_swz(inst_add->U.I.SrcReg[1])) { + + return 0; + } + + if (presub_helper(c, inst_add, RC_PRESUB_INV, presub_replace_inv)) { + rc_remove_instruction(inst_add); + return 1; + } + return 0; +} + +/** + * @return + * 0 if inst is still part of the program. + * 1 if inst is no longer part of the program. + */ +static int peephole(struct radeon_compiler * c, struct rc_instruction * inst) +{ + switch(inst->U.I.Opcode){ + case RC_OPCODE_ADD: + if (c->has_presub) { + if(peephole_add_presub_inv(c, inst)) + return 1; + if(peephole_add_presub_add(c, inst)) + return 1; + } + break; + default: + break; + } + return 0; +} + +void rc_optimize(struct radeon_compiler * c, void *user) +{ + struct rc_instruction * inst = c->Program.Instructions.Next; + while(inst != &c->Program.Instructions) { + struct rc_instruction * cur = inst; + inst = inst->Next; + + constant_folding(c, cur); + + if(peephole(c, cur)) + continue; + + if (cur->U.I.Opcode == RC_OPCODE_MOV) { + copy_propagate(c, cur); + /* cur may no longer be part of the program */ + } + } +} diff --git a/src/gallium/drivers/r300/compiler/radeon_pair_dead_sources.c b/src/gallium/drivers/r300/compiler/radeon_pair_dead_sources.c new file mode 100644 index 00000000000..1e9a2c09d44 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_pair_dead_sources.c @@ -0,0 +1,62 @@ + +#include "radeon_compiler.h" +#include "radeon_compiler_util.h" +#include "radeon_opcodes.h" +#include "radeon_program_pair.h" + +static void mark_used_presub(struct rc_pair_sub_instruction * sub) +{ + if (sub->Src[RC_PAIR_PRESUB_SRC].Used) { + unsigned int presub_reg_count = rc_presubtract_src_reg_count( + sub->Src[RC_PAIR_PRESUB_SRC].Index); + unsigned int i; + for (i = 0; i < presub_reg_count; i++) { + sub->Src[i].Used = 1; + } + } +} + +static void mark_used( + struct rc_instruction * inst, + struct rc_pair_sub_instruction * sub) +{ + unsigned int i; + const struct rc_opcode_info * info = rc_get_opcode_info(sub->Opcode); + for (i = 0; i < info->NumSrcRegs; i++) { + unsigned int src_type = rc_source_type_swz(sub->Arg[i].Swizzle); + if (src_type & RC_SOURCE_RGB) { + inst->U.P.RGB.Src[sub->Arg[i].Source].Used = 1; + } + + if (src_type & RC_SOURCE_ALPHA) { + inst->U.P.Alpha.Src[sub->Arg[i].Source].Used = 1; + } + } +} + +/** + * This pass finds sources that are not used by their instruction and marks + * them as unused. + */ +void rc_pair_remove_dead_sources(struct radeon_compiler * c, void *user) +{ + struct rc_instruction * inst; + for (inst = c->Program.Instructions.Next; + inst != &c->Program.Instructions; + inst = inst->Next) { + unsigned int i; + if (inst->Type == RC_INSTRUCTION_NORMAL) + continue; + + /* Mark all sources as unused */ + for (i = 0; i < 4; i++) { + inst->U.P.RGB.Src[i].Used = 0; + inst->U.P.Alpha.Src[i].Used = 0; + } + mark_used(inst, &inst->U.P.RGB); + mark_used(inst, &inst->U.P.Alpha); + + mark_used_presub(&inst->U.P.RGB); + mark_used_presub(&inst->U.P.Alpha); + } +} diff --git a/src/gallium/drivers/r300/compiler/radeon_pair_regalloc.c b/src/gallium/drivers/r300/compiler/radeon_pair_regalloc.c new file mode 100644 index 00000000000..49983d6ce75 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_pair_regalloc.c @@ -0,0 +1,706 @@ +/* + * Copyright (C) 2009 Nicolai Haehnle. + * Copyright 2011 Tom Stellard <[email protected]> + * + * 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. + * + */ + +#include "radeon_program_pair.h" + +#include <stdio.h> + +#include "main/glheader.h" +#include "program/register_allocate.h" +#include "ralloc.h" + +#include "r300_fragprog_swizzle.h" +#include "radeon_compiler.h" +#include "radeon_compiler_util.h" +#include "radeon_dataflow.h" +#include "radeon_list.h" +#include "radeon_variable.h" + +#define VERBOSE 0 + +#define DBG(...) do { if (VERBOSE) fprintf(stderr, __VA_ARGS__); } while(0) + + + +struct register_info { + struct live_intervals Live[4]; + + unsigned int Used:1; + unsigned int Allocated:1; + unsigned int File:3; + unsigned int Index:RC_REGISTER_INDEX_BITS; + unsigned int Writemask; +}; + +struct regalloc_state { + struct radeon_compiler * C; + + struct register_info * Input; + unsigned int NumInputs; + + struct register_info * Temporary; + unsigned int NumTemporaries; + + unsigned int Simple; + int LoopEnd; +}; + +enum rc_reg_class { + RC_REG_CLASS_SINGLE, + RC_REG_CLASS_DOUBLE, + RC_REG_CLASS_TRIPLE, + RC_REG_CLASS_ALPHA, + RC_REG_CLASS_SINGLE_PLUS_ALPHA, + RC_REG_CLASS_DOUBLE_PLUS_ALPHA, + RC_REG_CLASS_TRIPLE_PLUS_ALPHA, + RC_REG_CLASS_X, + RC_REG_CLASS_Y, + RC_REG_CLASS_Z, + RC_REG_CLASS_XY, + RC_REG_CLASS_YZ, + RC_REG_CLASS_XZ, + RC_REG_CLASS_XW, + RC_REG_CLASS_YW, + RC_REG_CLASS_ZW, + RC_REG_CLASS_XYW, + RC_REG_CLASS_YZW, + RC_REG_CLASS_XZW, + RC_REG_CLASS_COUNT +}; + +struct rc_class { + enum rc_reg_class Class; + + unsigned int WritemaskCount; + + /** This is 1 if this class is being used by the register allocator + * and 0 otherwise */ + unsigned int Used; + + /** This is the ID number assigned to this class by ra. */ + unsigned int Id; + + /** List of writemasks that belong to this class */ + unsigned int Writemasks[3]; + + +}; + +static void print_live_intervals(struct live_intervals * src) +{ + if (!src || !src->Used) { + DBG("(null)"); + return; + } + + DBG("(%i,%i)", src->Start, src->End); +} + +static int overlap_live_intervals(struct live_intervals * a, struct live_intervals * b) +{ + if (VERBOSE) { + DBG("overlap_live_intervals: "); + print_live_intervals(a); + DBG(" to "); + print_live_intervals(b); + DBG("\n"); + } + + if (!a->Used || !b->Used) { + DBG(" unused interval\n"); + return 0; + } + + if (a->Start > b->Start) { + if (a->Start < b->End) { + DBG(" overlap\n"); + return 1; + } + } else if (b->Start > a->Start) { + if (b->Start < a->End) { + DBG(" overlap\n"); + return 1; + } + } else { /* a->Start == b->Start */ + if (a->Start != a->End && b->Start != b->End) { + DBG(" overlap\n"); + return 1; + } + } + + DBG(" no overlap\n"); + + return 0; +} + +static void scan_read_callback(void * data, struct rc_instruction * inst, + rc_register_file file, unsigned int index, unsigned int mask) +{ + struct regalloc_state * s = data; + struct register_info * reg; + unsigned int i; + + if (file != RC_FILE_INPUT) + return; + + s->Input[index].Used = 1; + reg = &s->Input[index]; + + for (i = 0; i < 4; i++) { + if (!((mask >> i) & 0x1)) { + continue; + } + reg->Live[i].Used = 1; + reg->Live[i].Start = 0; + reg->Live[i].End = + s->LoopEnd > inst->IP ? s->LoopEnd : inst->IP; + } +} + +static void remap_register(void * data, struct rc_instruction * inst, + rc_register_file * file, unsigned int * index) +{ + struct regalloc_state * s = data; + const struct register_info * reg; + + if (*file == RC_FILE_TEMPORARY && s->Simple) + reg = &s->Temporary[*index]; + else if (*file == RC_FILE_INPUT) + reg = &s->Input[*index]; + else + return; + + if (reg->Allocated) { + *index = reg->Index; + } +} + +static void alloc_input_simple(void * data, unsigned int input, + unsigned int hwreg) +{ + struct regalloc_state * s = data; + + if (input >= s->NumInputs) + return; + + s->Input[input].Allocated = 1; + s->Input[input].File = RC_FILE_TEMPORARY; + s->Input[input].Index = hwreg; +} + +/* This functions offsets the temporary register indices by the number + * of input registers, because input registers are actually temporaries and + * should not occupy the same space. + * + * This pass is supposed to be used to maintain correct allocation of inputs + * if the standard register allocation is disabled. */ +static void do_regalloc_inputs_only(struct regalloc_state * s) +{ + for (unsigned i = 0; i < s->NumTemporaries; i++) { + s->Temporary[i].Allocated = 1; + s->Temporary[i].File = RC_FILE_TEMPORARY; + s->Temporary[i].Index = i + s->NumInputs; + } +} + +static unsigned int is_derivative(rc_opcode op) +{ + return (op == RC_OPCODE_DDX || op == RC_OPCODE_DDY); +} + +static int find_class( + struct rc_class * classes, + unsigned int writemask, + unsigned int max_writemask_count) +{ + unsigned int i; + for (i = 0; i < RC_REG_CLASS_COUNT; i++) { + unsigned int j; + if (classes[i].WritemaskCount > max_writemask_count) { + continue; + } + for (j = 0; j < 3; j++) { + if (classes[i].Writemasks[j] == writemask) { + return i; + } + } + } + return -1; +} + +static enum rc_reg_class variable_get_class( + struct rc_variable * variable, + struct rc_class * classes) +{ + unsigned int i; + unsigned int can_change_writemask= 1; + unsigned int writemask = rc_variable_writemask_sum(variable); + struct rc_list * readers = rc_variable_readers_union(variable); + int class_index; + + if (!variable->C->is_r500) { + struct rc_class c; + /* The assumption here is that if an instruction has type + * RC_INSTRUCTION_NORMAL then it is a TEX instruction. + * r300 and r400 can't swizzle the result of a TEX lookup. */ + if (variable->Inst->Type == RC_INSTRUCTION_NORMAL) { + writemask = RC_MASK_XYZW; + } + + /* Check if it is possible to do swizzle packing for r300/r400 + * without creating non-native swizzles. */ + class_index = find_class(classes, writemask, 3); + if (class_index < 0) { + goto error; + } + c = classes[class_index]; + for (i = 0; i < c.WritemaskCount; i++) { + int j; + unsigned int conversion_swizzle = + rc_make_conversion_swizzle( + writemask, c.Writemasks[i]); + for (j = 0; j < variable->ReaderCount; j++) { + unsigned int old_swizzle; + unsigned int new_swizzle; + struct rc_reader r = variable->Readers[j]; + if (r.Inst->Type == RC_INSTRUCTION_PAIR ) { + old_swizzle = r.U.P.Arg->Swizzle; + } else { + old_swizzle = r.U.I.Src->Swizzle; + } + new_swizzle = rc_adjust_channels( + old_swizzle, conversion_swizzle); + if (!r300_swizzle_is_native_basic(new_swizzle)) { + can_change_writemask = 0; + break; + } + } + if (!can_change_writemask) { + break; + } + } + } + + if (variable->Inst->Type == RC_INSTRUCTION_PAIR) { + /* DDX/DDY seem to always fail when their writemasks are + * changed.*/ + if (is_derivative(variable->Inst->U.P.RGB.Opcode) + || is_derivative(variable->Inst->U.P.Alpha.Opcode)) { + can_change_writemask = 0; + } + } + for ( ; readers; readers = readers->Next) { + struct rc_reader * r = readers->Item; + if (r->Inst->Type == RC_INSTRUCTION_PAIR) { + if (r->U.P.Arg->Source == RC_PAIR_PRESUB_SRC) { + can_change_writemask = 0; + break; + } + /* DDX/DDY also fail when their swizzles are changed. */ + if (is_derivative(r->Inst->U.P.RGB.Opcode) + || is_derivative(r->Inst->U.P.Alpha.Opcode)) { + can_change_writemask = 0; + break; + } + } + } + + class_index = find_class(classes, writemask, + can_change_writemask ? 3 : 1); + if (class_index > -1) { + return classes[class_index].Class; + } else { +error: + rc_error(variable->C, + "Could not find class for index=%u mask=%u\n", + variable->Dst.Index, writemask); + return 0; + } +} + +static unsigned int overlap_live_intervals_array( + struct live_intervals * a, + struct live_intervals * b) +{ + unsigned int a_chan, b_chan; + for (a_chan = 0; a_chan < 4; a_chan++) { + for (b_chan = 0; b_chan < 4; b_chan++) { + if (overlap_live_intervals(&a[a_chan], &b[b_chan])) { + return 1; + } + } + } + return 0; +} + +static unsigned int reg_get_index(int reg) +{ + return reg / RC_MASK_XYZW; +} + +static unsigned int reg_get_writemask(int reg) +{ + return (reg % RC_MASK_XYZW) + 1; +} + +static int get_reg_id(unsigned int index, unsigned int writemask) +{ + assert(writemask); + if (writemask == 0) { + return 0; + } + return (index * RC_MASK_XYZW) + (writemask - 1); +} + +#if VERBOSE +static void print_reg(int reg) +{ + unsigned int index = reg_get_index(reg); + unsigned int mask = reg_get_writemask(reg); + fprintf(stderr, "Temp[%u].%c%c%c%c", index, + mask & RC_MASK_X ? 'x' : '_', + mask & RC_MASK_Y ? 'y' : '_', + mask & RC_MASK_Z ? 'z' : '_', + mask & RC_MASK_W ? 'w' : '_'); +} +#endif + +static void add_register_conflicts( + struct ra_regs * regs, + unsigned int max_temp_regs) +{ + unsigned int index, a_mask, b_mask; + for (index = 0; index < max_temp_regs; index++) { + for(a_mask = 1; a_mask <= RC_MASK_XYZW; a_mask++) { + for (b_mask = a_mask + 1; b_mask <= RC_MASK_XYZW; + b_mask++) { + if (a_mask & b_mask) { + ra_add_reg_conflict(regs, + get_reg_id(index, a_mask), + get_reg_id(index, b_mask)); + } + } + } + } +} + +static void do_advanced_regalloc(struct regalloc_state * s) +{ + struct rc_class rc_class_list [] = { + {RC_REG_CLASS_SINGLE, 3, 0, 0, + {RC_MASK_X, + RC_MASK_Y, + RC_MASK_Z}}, + {RC_REG_CLASS_DOUBLE, 3, 0, 0, + {RC_MASK_X | RC_MASK_Y, + RC_MASK_X | RC_MASK_Z, + RC_MASK_Y | RC_MASK_Z}}, + {RC_REG_CLASS_TRIPLE, 1, 0, 0, + {RC_MASK_X | RC_MASK_Y | RC_MASK_Z, + RC_MASK_NONE, + RC_MASK_NONE}}, + {RC_REG_CLASS_ALPHA, 1, 0, 0, + {RC_MASK_W, + RC_MASK_NONE, + RC_MASK_NONE}}, + {RC_REG_CLASS_SINGLE_PLUS_ALPHA, 3, 0, 0, + {RC_MASK_X | RC_MASK_W, + RC_MASK_Y | RC_MASK_W, + RC_MASK_Z | RC_MASK_W}}, + {RC_REG_CLASS_DOUBLE_PLUS_ALPHA, 3, 0, 0, + {RC_MASK_X | RC_MASK_Y | RC_MASK_W, + RC_MASK_X | RC_MASK_Z | RC_MASK_W, + RC_MASK_Y | RC_MASK_Z | RC_MASK_W}}, + {RC_REG_CLASS_TRIPLE_PLUS_ALPHA, 1, 0, 0, + {RC_MASK_X | RC_MASK_Y | RC_MASK_Z | RC_MASK_W, + RC_MASK_NONE, + RC_MASK_NONE}}, + {RC_REG_CLASS_X, 1, 0, 0, + {RC_MASK_X, + RC_MASK_NONE, + RC_MASK_NONE}}, + {RC_REG_CLASS_Y, 1, 0, 0, + {RC_MASK_Y, + RC_MASK_NONE, + RC_MASK_NONE}}, + {RC_REG_CLASS_Z, 1, 0, 0, + {RC_MASK_Z, + RC_MASK_NONE, + RC_MASK_NONE}}, + {RC_REG_CLASS_XY, 1, 0, 0, + {RC_MASK_X | RC_MASK_Y, + RC_MASK_NONE, + RC_MASK_NONE}}, + {RC_REG_CLASS_YZ, 1, 0, 0, + {RC_MASK_Y | RC_MASK_Z, + RC_MASK_NONE, + RC_MASK_NONE}}, + {RC_REG_CLASS_XZ, 1, 0, 0, + {RC_MASK_X | RC_MASK_Z, + RC_MASK_NONE, + RC_MASK_NONE}}, + {RC_REG_CLASS_XW, 1, 0, 0, + {RC_MASK_X | RC_MASK_W, + RC_MASK_NONE, + RC_MASK_NONE}}, + {RC_REG_CLASS_YW, 1, 0, 0, + {RC_MASK_Y | RC_MASK_W, + RC_MASK_NONE, + RC_MASK_NONE}}, + {RC_REG_CLASS_ZW, 1, 0, 0, + {RC_MASK_Z | RC_MASK_W, + RC_MASK_NONE, + RC_MASK_NONE}}, + {RC_REG_CLASS_XYW, 1, 0, 0, + {RC_MASK_X | RC_MASK_Y | RC_MASK_W, + RC_MASK_NONE, + RC_MASK_NONE}}, + {RC_REG_CLASS_YZW, 1, 0, 0, + {RC_MASK_Y | RC_MASK_Z | RC_MASK_W, + RC_MASK_NONE, + RC_MASK_NONE}}, + {RC_REG_CLASS_XZW, 1, 0, 0, + {RC_MASK_X | RC_MASK_Z | RC_MASK_W, + RC_MASK_NONE, + RC_MASK_NONE}} + }; + + unsigned int i, j, index, input_node, node_count, node_index; + unsigned int * node_classes; + unsigned int * input_classes; + struct rc_instruction * inst; + struct rc_list * var_ptr; + struct rc_list * variables; + struct ra_regs * regs; + struct ra_graph * graph; + + /* Allocate the main ra data structure */ + regs = ra_alloc_reg_set(s->C->max_temp_regs * RC_MASK_XYZW); + + /* Get list of program variables */ + variables = rc_get_variables(s->C); + node_count = rc_list_count(variables); + node_classes = memory_pool_malloc(&s->C->Pool, + node_count * sizeof(unsigned int)); + input_classes = memory_pool_malloc(&s->C->Pool, + s->NumInputs * sizeof(unsigned int)); + + for (var_ptr = variables, node_index = 0; var_ptr; + var_ptr = var_ptr->Next, node_index++) { + unsigned int class_index; + /* Compute the live intervals */ + rc_variable_compute_live_intervals(var_ptr->Item); + + class_index = variable_get_class(var_ptr->Item, rc_class_list); + + /* If we haven't used this register class yet, mark it + * as used and allocate space for it. */ + if (!rc_class_list[class_index].Used) { + rc_class_list[class_index].Used = 1; + rc_class_list[class_index].Id = ra_alloc_reg_class(regs); + } + + node_classes[node_index] = rc_class_list[class_index].Id; + } + + + /* Assign registers to the classes */ + for (i = 0; i < RC_REG_CLASS_COUNT; i++) { + struct rc_class class = rc_class_list[i]; + if (!class.Used) { + continue; + } + + for (index = 0; index < s->C->max_temp_regs; index++) { + for (j = 0; j < class.WritemaskCount; j++) { + int reg_id = get_reg_id(index, + class.Writemasks[j]); + ra_class_add_reg(regs, class.Id, reg_id); + } + } + } + + /* Add register conflicts */ + add_register_conflicts(regs, s->C->max_temp_regs); + + /* Calculate live intervals for input registers */ + for (inst = s->C->Program.Instructions.Next; + inst != &s->C->Program.Instructions; + inst = inst->Next) { + rc_opcode op = rc_get_flow_control_inst(inst); + if (op == RC_OPCODE_BGNLOOP) { + struct rc_instruction * endloop = + rc_match_bgnloop(inst); + if (endloop->IP > s->LoopEnd) { + s->LoopEnd = endloop->IP; + } + } + rc_for_all_reads_mask(inst, scan_read_callback, s); + } + + /* Create classes for input registers */ + for (i = 0; i < s->NumInputs; i++) { + unsigned int chan, class_id, writemask = 0; + for (chan = 0; chan < 4; chan++) { + if (s->Input[i].Live[chan].Used) { + writemask |= (1 << chan); + } + } + s->Input[i].Writemask = writemask; + if (!writemask) { + continue; + } + + class_id = ra_alloc_reg_class(regs); + input_classes[i] = class_id; + ra_class_add_reg(regs, class_id, + get_reg_id(s->Input[i].Index, writemask)); + } + + ra_set_finalize(regs); + + graph = ra_alloc_interference_graph(regs, node_count + s->NumInputs); + + /* Build the interference graph */ + for (var_ptr = variables, node_index = 0; var_ptr; + var_ptr = var_ptr->Next,node_index++) { + struct rc_list * a, * b; + unsigned int b_index; + + ra_set_node_class(graph, node_index, node_classes[node_index]); + + for (a = var_ptr, b = var_ptr->Next, b_index = node_index + 1; + b; b = b->Next, b_index++) { + struct rc_variable * var_a = a->Item; + while (var_a) { + struct rc_variable * var_b = b->Item; + while (var_b) { + if (overlap_live_intervals_array(var_a->Live, var_b->Live)) { + ra_add_node_interference(graph, + node_index, b_index); + } + var_b = var_b->Friend; + } + var_a = var_a->Friend; + } + } + } + + /* Add input registers to the interference graph */ + for (i = 0, input_node = 0; i< s->NumInputs; i++) { + if (!s->Input[i].Writemask) { + continue; + } + ra_set_node_class(graph, node_count + input_node, + input_classes[i]); + for (var_ptr = variables, node_index = 0; + var_ptr; var_ptr = var_ptr->Next, node_index++) { + struct rc_variable * var = var_ptr->Item; + if (overlap_live_intervals_array(s->Input[i].Live, + var->Live)) { + ra_add_node_interference(graph, node_index, + node_count + input_node); + } + } + /* Manually allocate a register for this input */ + ra_set_node_reg(graph, node_count + input_node, get_reg_id( + s->Input[i].Index, s->Input[i].Writemask)); + input_node++; + } + + if (!ra_allocate_no_spills(graph)) { + rc_error(s->C, "Ran out of hardware temporaries\n"); + return; + } + + /* Rewrite the registers */ + for (var_ptr = variables, node_index = 0; var_ptr; + var_ptr = var_ptr->Next, node_index++) { + int reg = ra_get_node_reg(graph, node_index); + unsigned int writemask = reg_get_writemask(reg); + unsigned int index = reg_get_index(reg); + struct rc_variable * var = var_ptr->Item; + + if (!s->C->is_r500 && var->Inst->Type == RC_INSTRUCTION_NORMAL) { + writemask = rc_variable_writemask_sum(var); + } + + if (var->Dst.File == RC_FILE_INPUT) { + continue; + } + rc_variable_change_dst(var, index, writemask); + } + + ralloc_free(graph); + ralloc_free(regs); +} + +/** + * @param user This parameter should be a pointer to an integer value. If this + * integer value is zero, then a simple register allocator will be used that + * only allocates space for input registers (\sa do_regalloc_inputs_only). If + * user is non-zero, then the regular register allocator will be used + * (\sa do_regalloc). + */ +void rc_pair_regalloc(struct radeon_compiler *cc, void *user) +{ + struct r300_fragment_program_compiler *c = + (struct r300_fragment_program_compiler*)cc; + struct regalloc_state s; + int * do_full_regalloc = (int*)user; + + memset(&s, 0, sizeof(s)); + s.C = cc; + s.NumInputs = rc_get_max_index(cc, RC_FILE_INPUT) + 1; + s.Input = memory_pool_malloc(&cc->Pool, + s.NumInputs * sizeof(struct register_info)); + memset(s.Input, 0, s.NumInputs * sizeof(struct register_info)); + + s.NumTemporaries = rc_get_max_index(cc, RC_FILE_TEMPORARY) + 1; + s.Temporary = memory_pool_malloc(&cc->Pool, + s.NumTemporaries * sizeof(struct register_info)); + memset(s.Temporary, 0, s.NumTemporaries * sizeof(struct register_info)); + + rc_recompute_ips(s.C); + + c->AllocateHwInputs(c, &alloc_input_simple, &s); + if (*do_full_regalloc) { + do_advanced_regalloc(&s); + } else { + s.Simple = 1; + do_regalloc_inputs_only(&s); + } + + /* Rewrite inputs and if we are doing the simple allocation, rewrite + * temporaries too. */ + for (struct rc_instruction *inst = s.C->Program.Instructions.Next; + inst != &s.C->Program.Instructions; + inst = inst->Next) { + rc_remap_registers(inst, &remap_register, &s); + } +} diff --git a/src/gallium/drivers/r300/compiler/radeon_pair_schedule.c b/src/gallium/drivers/r300/compiler/radeon_pair_schedule.c new file mode 100644 index 00000000000..25cd52c9cd4 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_pair_schedule.c @@ -0,0 +1,1010 @@ +/* + * Copyright (C) 2009 Nicolai Haehnle. + * + * 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. + * + */ + +#include "radeon_program_pair.h" + +#include <stdio.h> + +#include "radeon_compiler.h" +#include "radeon_compiler_util.h" +#include "radeon_dataflow.h" + + +#define VERBOSE 0 + +#define DBG(...) do { if (VERBOSE) fprintf(stderr, __VA_ARGS__); } while(0) + +struct schedule_instruction { + struct rc_instruction * Instruction; + + /** Next instruction in the linked list of ready instructions. */ + struct schedule_instruction *NextReady; + + /** Values that this instruction reads and writes */ + struct reg_value * WriteValues[4]; + struct reg_value * ReadValues[12]; + unsigned int NumWriteValues:3; + unsigned int NumReadValues:4; + + /** + * Number of (read and write) dependencies that must be resolved before + * this instruction can be scheduled. + */ + unsigned int NumDependencies:5; + + /** List of all readers (see rc_get_readers() for the definition of + * "all readers"), even those outside the basic block this instruction + * lives in. */ + struct rc_reader_data GlobalReaders; +}; + + +/** + * Used to keep track of which instructions read a value. + */ +struct reg_value_reader { + struct schedule_instruction *Reader; + struct reg_value_reader *Next; +}; + +/** + * Used to keep track which values are stored in each component of a + * RC_FILE_TEMPORARY. + */ +struct reg_value { + struct schedule_instruction * Writer; + + /** + * Unordered linked list of instructions that read from this value. + * When this value becomes available, we increase all readers' + * dependency count. + */ + struct reg_value_reader *Readers; + + /** + * Number of readers of this value. This is decremented each time + * a reader of the value is committed. + * When the reader cound reaches zero, the dependency count + * of the instruction writing \ref Next is decremented. + */ + unsigned int NumReaders; + + struct reg_value *Next; /**< Pointer to the next value to be written to the same register */ +}; + +struct register_state { + struct reg_value * Values[4]; +}; + +struct remap_reg { + struct rc_instruciont * Inst; + unsigned int OldIndex:(RC_REGISTER_INDEX_BITS+1); + unsigned int OldSwizzle:3; + unsigned int NewIndex:(RC_REGISTER_INDEX_BITS+1); + unsigned int NewSwizzle:3; + unsigned int OnlyTexReads:1; + struct remap_reg * Next; +}; + +struct schedule_state { + struct radeon_compiler * C; + struct schedule_instruction * Current; + + struct register_state Temporary[RC_REGISTER_MAX_INDEX]; + + /** + * Linked lists of instructions that can be scheduled right now, + * based on which ALU/TEX resources they require. + */ + /*@{*/ + struct schedule_instruction *ReadyFullALU; + struct schedule_instruction *ReadyRGB; + struct schedule_instruction *ReadyAlpha; + struct schedule_instruction *ReadyTEX; + /*@}*/ +}; + +static struct reg_value ** get_reg_valuep(struct schedule_state * s, + rc_register_file file, unsigned int index, unsigned int chan) +{ + if (file != RC_FILE_TEMPORARY) + return 0; + + if (index >= RC_REGISTER_MAX_INDEX) { + rc_error(s->C, "%s: index %i out of bounds\n", __FUNCTION__, index); + return 0; + } + + return &s->Temporary[index].Values[chan]; +} + +static void add_inst_to_list(struct schedule_instruction ** list, struct schedule_instruction * inst) +{ + inst->NextReady = *list; + *list = inst; +} + +static void add_inst_to_list_end(struct schedule_instruction ** list, + struct schedule_instruction * inst) +{ + if(!*list){ + *list = inst; + }else{ + struct schedule_instruction * temp = *list; + while(temp->NextReady){ + temp = temp->NextReady; + } + temp->NextReady = inst; + } +} + +static void instruction_ready(struct schedule_state * s, struct schedule_instruction * sinst) +{ + DBG("%i is now ready\n", sinst->Instruction->IP); + + /* Adding Ready TEX instructions to the end of the "Ready List" helps + * us emit TEX instructions in blocks without losing our place. */ + if (sinst->Instruction->Type == RC_INSTRUCTION_NORMAL) + add_inst_to_list_end(&s->ReadyTEX, sinst); + else if (sinst->Instruction->U.P.Alpha.Opcode == RC_OPCODE_NOP) + add_inst_to_list(&s->ReadyRGB, sinst); + else if (sinst->Instruction->U.P.RGB.Opcode == RC_OPCODE_NOP) + add_inst_to_list(&s->ReadyAlpha, sinst); + else + add_inst_to_list(&s->ReadyFullALU, sinst); +} + +static void decrease_dependencies(struct schedule_state * s, struct schedule_instruction * sinst) +{ + assert(sinst->NumDependencies > 0); + sinst->NumDependencies--; + if (!sinst->NumDependencies) + instruction_ready(s, sinst); +} + +/** + * This function decreases the dependencies of the next instruction that + * wants to write to each of sinst's read values. + */ +static void commit_update_reads(struct schedule_state * s, + struct schedule_instruction * sinst){ + unsigned int i; + for(i = 0; i < sinst->NumReadValues; ++i) { + struct reg_value * v = sinst->ReadValues[i]; + assert(v->NumReaders > 0); + v->NumReaders--; + if (!v->NumReaders) { + if (v->Next) + decrease_dependencies(s, v->Next->Writer); + } + } +} + +static void commit_update_writes(struct schedule_state * s, + struct schedule_instruction * sinst){ + unsigned int i; + for(i = 0; i < sinst->NumWriteValues; ++i) { + struct reg_value * v = sinst->WriteValues[i]; + if (v->NumReaders) { + for(struct reg_value_reader * r = v->Readers; r; r = r->Next) { + decrease_dependencies(s, r->Reader); + } + } else { + /* This happens in instruction sequences of the type + * OP r.x, ...; + * OP r.x, r.x, ...; + * See also the subtlety in how instructions that both + * read and write the same register are scanned. + */ + if (v->Next) + decrease_dependencies(s, v->Next->Writer); + } + } +} + +static void commit_alu_instruction(struct schedule_state * s, struct schedule_instruction * sinst) +{ + DBG("%i: commit\n", sinst->Instruction->IP); + + commit_update_reads(s, sinst); + + commit_update_writes(s, sinst); +} + +/** + * Emit all ready texture instructions in a single block. + * + * Emit as a single block to (hopefully) sample many textures in parallel, + * and to avoid hardware indirections on R300. + */ +static void emit_all_tex(struct schedule_state * s, struct rc_instruction * before) +{ + struct schedule_instruction *readytex; + struct rc_instruction * inst_begin; + + assert(s->ReadyTEX); + + /* Node marker for R300 */ + inst_begin = rc_insert_new_instruction(s->C, before->Prev); + inst_begin->U.I.Opcode = RC_OPCODE_BEGIN_TEX; + + /* Link texture instructions back in */ + readytex = s->ReadyTEX; + while(readytex) { + rc_insert_instruction(before->Prev, readytex->Instruction); + DBG("%i: commit TEX reads\n", readytex->Instruction->IP); + + /* All of the TEX instructions in the same TEX block have + * their source registers read from before any of the + * instructions in that block write to their destination + * registers. This means that when we commit a TEX + * instruction, any other TEX instruction that wants to write + * to one of the committed instruction's source register can be + * marked as ready and should be emitted in the same TEX + * block. This prevents the following sequence from being + * emitted in two different TEX blocks: + * 0: TEX temp[0].xyz, temp[1].xy__, 2D[0]; + * 1: TEX temp[1].xyz, temp[2].xy__, 2D[0]; + */ + commit_update_reads(s, readytex); + readytex = readytex->NextReady; + } + readytex = s->ReadyTEX; + s->ReadyTEX = 0; + while(readytex){ + DBG("%i: commit TEX writes\n", readytex->Instruction->IP); + commit_update_writes(s, readytex); + readytex = readytex->NextReady; + } +} + +/* This is a helper function for destructive_merge_instructions(). It helps + * merge presubtract sources from two instructions and makes sure the + * presubtract sources end up in the correct spot. This function assumes that + * dst_full is an rgb instruction, meaning that it has a vector instruction(rgb) + * but no scalar instruction (alpha). + * @return 0 if merging the presubtract sources fails. + * @retrun 1 if merging the presubtract sources succeeds. + */ +static int merge_presub_sources( + struct rc_pair_instruction * dst_full, + struct rc_pair_sub_instruction src, + unsigned int type) +{ + unsigned int srcp_src, srcp_regs, is_rgb, is_alpha; + struct rc_pair_sub_instruction * dst_sub; + const struct rc_opcode_info * info; + + assert(dst_full->Alpha.Opcode == RC_OPCODE_NOP); + + switch(type) { + case RC_SOURCE_RGB: + is_rgb = 1; + is_alpha = 0; + dst_sub = &dst_full->RGB; + break; + case RC_SOURCE_ALPHA: + is_rgb = 0; + is_alpha = 1; + dst_sub = &dst_full->Alpha; + break; + default: + assert(0); + return 0; + } + + info = rc_get_opcode_info(dst_full->RGB.Opcode); + + if (dst_sub->Src[RC_PAIR_PRESUB_SRC].Used) + return 0; + + srcp_regs = rc_presubtract_src_reg_count( + src.Src[RC_PAIR_PRESUB_SRC].Index); + for(srcp_src = 0; srcp_src < srcp_regs; srcp_src++) { + unsigned int arg; + int free_source; + unsigned int one_way = 0; + struct rc_pair_instruction_source srcp = src.Src[srcp_src]; + struct rc_pair_instruction_source temp; + + free_source = rc_pair_alloc_source(dst_full, is_rgb, is_alpha, + srcp.File, srcp.Index); + + /* If free_source < 0 then there are no free source + * slots. */ + if (free_source < 0) + return 0; + + temp = dst_sub->Src[srcp_src]; + dst_sub->Src[srcp_src] = dst_sub->Src[free_source]; + + /* srcp needs src0 and src1 to be the same */ + if (free_source < srcp_src) { + if (!temp.Used) + continue; + free_source = rc_pair_alloc_source(dst_full, is_rgb, + is_alpha, temp.File, temp.Index); + if (free_source < 0) + return 0; + one_way = 1; + } else { + dst_sub->Src[free_source] = temp; + } + + /* If free_source == srcp_src, then the presubtract + * source is already in the correct place. */ + if (free_source == srcp_src) + continue; + + /* Shuffle the sources, so we can put the + * presubtract source in the correct place. */ + for(arg = 0; arg < info->NumSrcRegs; arg++) { + /*If this arg does not read from an rgb source, + * do nothing. */ + if (!(rc_source_type_swz(dst_full->RGB.Arg[arg].Swizzle) + & type)) { + continue; + } + + if (dst_full->RGB.Arg[arg].Source == srcp_src) + dst_full->RGB.Arg[arg].Source = free_source; + /* We need to do this just in case register + * is one of the sources already, but in the + * wrong spot. */ + else if(dst_full->RGB.Arg[arg].Source == free_source + && !one_way) { + dst_full->RGB.Arg[arg].Source = srcp_src; + } + } + } + return 1; +} + + +/* This function assumes that rgb.Alpha and alpha.RGB are unused */ +static int destructive_merge_instructions( + struct rc_pair_instruction * rgb, + struct rc_pair_instruction * alpha) +{ + const struct rc_opcode_info * opcode; + + assert(rgb->Alpha.Opcode == RC_OPCODE_NOP); + assert(alpha->RGB.Opcode == RC_OPCODE_NOP); + + /* Presubtract registers need to be merged first so that registers + * needed by the presubtract operation can be placed in src0 and/or + * src1. */ + + /* Merge the rgb presubtract registers. */ + if (alpha->RGB.Src[RC_PAIR_PRESUB_SRC].Used) { + if (!merge_presub_sources(rgb, alpha->RGB, RC_SOURCE_RGB)) { + return 0; + } + } + /* Merge the alpha presubtract registers */ + if (alpha->Alpha.Src[RC_PAIR_PRESUB_SRC].Used) { + if(!merge_presub_sources(rgb, alpha->Alpha, RC_SOURCE_ALPHA)){ + return 0; + } + } + + /* Copy alpha args into rgb */ + opcode = rc_get_opcode_info(alpha->Alpha.Opcode); + + for(unsigned int arg = 0; arg < opcode->NumSrcRegs; ++arg) { + unsigned int srcrgb = 0; + unsigned int srcalpha = 0; + unsigned int oldsrc = alpha->Alpha.Arg[arg].Source; + rc_register_file file = 0; + unsigned int index = 0; + int source; + + if (GET_SWZ(alpha->Alpha.Arg[arg].Swizzle, 0) < 3) { + srcrgb = 1; + file = alpha->RGB.Src[oldsrc].File; + index = alpha->RGB.Src[oldsrc].Index; + } else if (GET_SWZ(alpha->Alpha.Arg[arg].Swizzle, 0) < 4) { + srcalpha = 1; + file = alpha->Alpha.Src[oldsrc].File; + index = alpha->Alpha.Src[oldsrc].Index; + } + + source = rc_pair_alloc_source(rgb, srcrgb, srcalpha, file, index); + if (source < 0) + return 0; + + rgb->Alpha.Arg[arg].Source = source; + rgb->Alpha.Arg[arg].Swizzle = alpha->Alpha.Arg[arg].Swizzle; + rgb->Alpha.Arg[arg].Abs = alpha->Alpha.Arg[arg].Abs; + rgb->Alpha.Arg[arg].Negate = alpha->Alpha.Arg[arg].Negate; + } + + /* Copy alpha opcode into rgb */ + rgb->Alpha.Opcode = alpha->Alpha.Opcode; + rgb->Alpha.DestIndex = alpha->Alpha.DestIndex; + rgb->Alpha.WriteMask = alpha->Alpha.WriteMask; + rgb->Alpha.OutputWriteMask = alpha->Alpha.OutputWriteMask; + rgb->Alpha.DepthWriteMask = alpha->Alpha.DepthWriteMask; + rgb->Alpha.Saturate = alpha->Alpha.Saturate; + + /* Merge ALU result writing */ + if (alpha->WriteALUResult) { + if (rgb->WriteALUResult) + return 0; + + rgb->WriteALUResult = alpha->WriteALUResult; + rgb->ALUResultCompare = alpha->ALUResultCompare; + } + + return 1; +} + +/** + * Try to merge the given instructions into the rgb instructions. + * + * Return true on success; on failure, return false, and keep + * the instructions untouched. + */ +static int merge_instructions(struct rc_pair_instruction * rgb, struct rc_pair_instruction * alpha) +{ + struct rc_pair_instruction backup; + + /*Instructions can't write output registers and ALU result at the + * same time. */ + if ((rgb->WriteALUResult && alpha->Alpha.OutputWriteMask) + || (rgb->RGB.OutputWriteMask && alpha->WriteALUResult)) { + return 0; + } + memcpy(&backup, rgb, sizeof(struct rc_pair_instruction)); + + if (destructive_merge_instructions(rgb, alpha)) + return 1; + + memcpy(rgb, &backup, sizeof(struct rc_pair_instruction)); + return 0; +} + +static void presub_nop(struct rc_instruction * emitted) { + int prev_rgb_index, prev_alpha_index, i, num_src; + + /* We don't need a nop if the previous instruction is a TEX. */ + if (emitted->Prev->Type != RC_INSTRUCTION_PAIR) { + return; + } + if (emitted->Prev->U.P.RGB.WriteMask) + prev_rgb_index = emitted->Prev->U.P.RGB.DestIndex; + else + prev_rgb_index = -1; + if (emitted->Prev->U.P.Alpha.WriteMask) + prev_alpha_index = emitted->Prev->U.P.Alpha.DestIndex; + else + prev_alpha_index = 1; + + /* Check the previous rgb instruction */ + if (emitted->U.P.RGB.Src[RC_PAIR_PRESUB_SRC].Used) { + num_src = rc_presubtract_src_reg_count( + emitted->U.P.RGB.Src[RC_PAIR_PRESUB_SRC].Index); + for (i = 0; i < num_src; i++) { + unsigned int index = emitted->U.P.RGB.Src[i].Index; + if (emitted->U.P.RGB.Src[i].File == RC_FILE_TEMPORARY + && (index == prev_rgb_index + || index == prev_alpha_index)) { + emitted->Prev->U.P.Nop = 1; + return; + } + } + } + + /* Check the previous alpha instruction. */ + if (!emitted->U.P.Alpha.Src[RC_PAIR_PRESUB_SRC].Used) + return; + + num_src = rc_presubtract_src_reg_count( + emitted->U.P.Alpha.Src[RC_PAIR_PRESUB_SRC].Index); + for (i = 0; i < num_src; i++) { + unsigned int index = emitted->U.P.Alpha.Src[i].Index; + if(emitted->U.P.Alpha.Src[i].File == RC_FILE_TEMPORARY + && (index == prev_rgb_index || index == prev_alpha_index)) { + emitted->Prev->U.P.Nop = 1; + return; + } + } +} + +static void rgb_to_alpha_remap ( + struct rc_instruction * inst, + struct rc_pair_instruction_arg * arg, + rc_register_file old_file, + rc_swizzle old_swz, + unsigned int new_index) +{ + int new_src_index; + unsigned int i; + + for (i = 0; i < 3; i++) { + if (get_swz(arg->Swizzle, i) == old_swz) { + SET_SWZ(arg->Swizzle, i, RC_SWIZZLE_W); + } + } + new_src_index = rc_pair_alloc_source(&inst->U.P, 0, 1, + old_file, new_index); + /* This conversion is not possible, we must have made a mistake in + * is_rgb_to_alpha_possible. */ + if (new_src_index < 0) { + assert(0); + return; + } + + arg->Source = new_src_index; +} + +static int can_remap(unsigned int opcode) +{ + switch(opcode) { + case RC_OPCODE_DDX: + case RC_OPCODE_DDY: + return 0; + default: + return 1; + } +} + +static int can_convert_opcode_to_alpha(unsigned int opcode) +{ + switch(opcode) { + case RC_OPCODE_DDX: + case RC_OPCODE_DDY: + case RC_OPCODE_DP2: + case RC_OPCODE_DP3: + case RC_OPCODE_DP4: + case RC_OPCODE_DPH: + return 0; + default: + return 1; + } +} + +static void is_rgb_to_alpha_possible( + void * userdata, + struct rc_instruction * inst, + struct rc_pair_instruction_arg * arg, + struct rc_pair_instruction_source * src) +{ + unsigned int chan_count = 0; + unsigned int alpha_sources = 0; + unsigned int i; + struct rc_reader_data * reader_data = userdata; + + if (!can_remap(inst->U.P.RGB.Opcode) + || !can_remap(inst->U.P.Alpha.Opcode)) { + reader_data->Abort = 1; + return; + } + + if (!src) + return; + + /* XXX There are some cases where we can still do the conversion if + * a reader reads from a presubtract source, but for now we'll prevent + * it. */ + if (arg->Source == RC_PAIR_PRESUB_SRC) { + reader_data->Abort = 1; + return; + } + + /* Make sure the source only reads from one component. + * XXX We should allow the source to read from the same component twice. + * XXX If the index we will be converting to is the same as the + * current index, then it is OK to read from more than one component. + */ + for (i = 0; i < 3; i++) { + rc_swizzle swz = get_swz(arg->Swizzle, i); + switch(swz) { + case RC_SWIZZLE_X: + case RC_SWIZZLE_Y: + case RC_SWIZZLE_Z: + case RC_SWIZZLE_W: + chan_count++; + break; + default: + break; + } + } + if (chan_count > 1) { + reader_data->Abort = 1; + return; + } + + /* Make sure there are enough alpha sources. + * XXX If we know what register all the readers are going + * to be remapped to, then in some situations we can still do + * the subsitution, even if all 3 alpha sources are being used.*/ + for (i = 0; i < 3; i++) { + if (inst->U.P.Alpha.Src[i].Used) { + alpha_sources++; + } + } + if (alpha_sources > 2) { + reader_data->Abort = 1; + return; + } +} + +static int convert_rgb_to_alpha( + struct schedule_state * s, + struct schedule_instruction * sched_inst) +{ + struct rc_pair_instruction * pair_inst = &sched_inst->Instruction->U.P; + unsigned int old_mask = pair_inst->RGB.WriteMask; + unsigned int old_swz = rc_mask_to_swizzle(old_mask); + const struct rc_opcode_info * info = + rc_get_opcode_info(pair_inst->RGB.Opcode); + int new_index = -1; + unsigned int i; + + if (sched_inst->GlobalReaders.Abort) + return 0; + + if (!pair_inst->RGB.WriteMask) + return 0; + + if (!can_convert_opcode_to_alpha(pair_inst->RGB.Opcode) + || !can_convert_opcode_to_alpha(pair_inst->Alpha.Opcode)) { + return 0; + } + + assert(sched_inst->NumWriteValues == 1); + + if (!sched_inst->WriteValues[0]) { + assert(0); + return 0; + } + + /* We start at the old index, because if we can reuse the same + * register and just change the swizzle then it is more likely we + * will be able to convert all the readers. */ + for (i = pair_inst->RGB.DestIndex; i < RC_REGISTER_MAX_INDEX; i++) { + struct reg_value ** new_regvalp = get_reg_valuep( + s, RC_FILE_TEMPORARY, i, 3); + if (!*new_regvalp) { + struct reg_value ** old_regvalp = + get_reg_valuep(s, + RC_FILE_TEMPORARY, + pair_inst->RGB.DestIndex, + rc_mask_to_swizzle(old_mask)); + new_index = i; + *new_regvalp = *old_regvalp; + *old_regvalp = NULL; + new_regvalp = get_reg_valuep(s, RC_FILE_TEMPORARY, i, 3); + break; + } + } + if (new_index < 0) { + return 0; + } + + pair_inst->Alpha.Opcode = pair_inst->RGB.Opcode; + pair_inst->Alpha.DestIndex = new_index; + pair_inst->Alpha.WriteMask = RC_MASK_W; + pair_inst->Alpha.Target = pair_inst->RGB.Target; + pair_inst->Alpha.OutputWriteMask = pair_inst->RGB.OutputWriteMask; + pair_inst->Alpha.DepthWriteMask = pair_inst->RGB.DepthWriteMask; + pair_inst->Alpha.Saturate = pair_inst->RGB.Saturate; + memcpy(pair_inst->Alpha.Arg, pair_inst->RGB.Arg, + sizeof(pair_inst->Alpha.Arg)); + /* Move the swizzles into the first chan */ + for (i = 0; i < info->NumSrcRegs; i++) { + unsigned int j; + for (j = 0; j < 3; j++) { + unsigned int swz = get_swz(pair_inst->Alpha.Arg[i].Swizzle, j); + if (swz != RC_SWIZZLE_UNUSED) { + pair_inst->Alpha.Arg[i].Swizzle = + rc_init_swizzle(swz, 1); + break; + } + } + } + pair_inst->RGB.Opcode = RC_OPCODE_NOP; + pair_inst->RGB.DestIndex = 0; + pair_inst->RGB.WriteMask = 0; + pair_inst->RGB.Target = 0; + pair_inst->RGB.OutputWriteMask = 0; + pair_inst->RGB.DepthWriteMask = 0; + pair_inst->RGB.Saturate = 0; + memset(pair_inst->RGB.Arg, 0, sizeof(pair_inst->RGB.Arg)); + + for(i = 0; i < sched_inst->GlobalReaders.ReaderCount; i++) { + struct rc_reader reader = sched_inst->GlobalReaders.Readers[i]; + rgb_to_alpha_remap(reader.Inst, reader.U.P.Arg, + RC_FILE_TEMPORARY, old_swz, new_index); + } + return 1; +} + +/** + * Find a good ALU instruction or pair of ALU instruction and emit it. + * + * Prefer emitting full ALU instructions, so that when we reach a point + * where no full ALU instruction can be emitted, we have more candidates + * for RGB/Alpha pairing. + */ +static void emit_one_alu(struct schedule_state *s, struct rc_instruction * before) +{ + struct schedule_instruction * sinst; + + if (s->ReadyFullALU) { + sinst = s->ReadyFullALU; + s->ReadyFullALU = s->ReadyFullALU->NextReady; + rc_insert_instruction(before->Prev, sinst->Instruction); + commit_alu_instruction(s, sinst); + } else { + struct schedule_instruction **prgb; + struct schedule_instruction **palpha; + struct schedule_instruction *prev; +pair: + /* Some pairings might fail because they require too + * many source slots; try all possible pairings if necessary */ + for(prgb = &s->ReadyRGB; *prgb; prgb = &(*prgb)->NextReady) { + for(palpha = &s->ReadyAlpha; *palpha; palpha = &(*palpha)->NextReady) { + struct schedule_instruction * psirgb = *prgb; + struct schedule_instruction * psialpha = *palpha; + + if (!merge_instructions(&psirgb->Instruction->U.P, &psialpha->Instruction->U.P)) + continue; + + *prgb = (*prgb)->NextReady; + *palpha = (*palpha)->NextReady; + rc_insert_instruction(before->Prev, psirgb->Instruction); + commit_alu_instruction(s, psirgb); + commit_alu_instruction(s, psialpha); + goto success; + } + } + prev = NULL; + /* No success in pairing, now try to convert one of the RGB + * instructions to an Alpha so we can pair it with another RGB. + */ + if (s->ReadyRGB && s->ReadyRGB->NextReady) { + for(prgb = &s->ReadyRGB; *prgb; prgb = &(*prgb)->NextReady) { + if ((*prgb)->NumWriteValues == 1) { + struct schedule_instruction * prgb_next; + if (!convert_rgb_to_alpha(s, *prgb)) + goto cont_loop; + prgb_next = (*prgb)->NextReady; + /* Add instruction to the Alpha ready list. */ + (*prgb)->NextReady = s->ReadyAlpha; + s->ReadyAlpha = *prgb; + /* Remove instruction from the RGB ready list.*/ + if (prev) + prev->NextReady = prgb_next; + else + s->ReadyRGB = prgb_next; + goto pair; + } +cont_loop: + prev = *prgb; + } + } + /* Still no success in pairing, just take the first RGB + * or alpha instruction. */ + if (s->ReadyRGB) { + sinst = s->ReadyRGB; + s->ReadyRGB = s->ReadyRGB->NextReady; + } else if (s->ReadyAlpha) { + sinst = s->ReadyAlpha; + s->ReadyAlpha = s->ReadyAlpha->NextReady; + } else { + /*XXX Something real bad has happened. */ + assert(0); + } + + rc_insert_instruction(before->Prev, sinst->Instruction); + commit_alu_instruction(s, sinst); + success: ; + } + /* If the instruction we just emitted uses a presubtract value, and + * the presubtract sources were written by the previous intstruction, + * the previous instruction needs a nop. */ + presub_nop(before->Prev); +} + +static void scan_read(void * data, struct rc_instruction * inst, + rc_register_file file, unsigned int index, unsigned int chan) +{ + struct schedule_state * s = data; + struct reg_value ** v = get_reg_valuep(s, file, index, chan); + struct reg_value_reader * reader; + + if (!v) + return; + + if (*v && (*v)->Writer == s->Current) { + /* The instruction reads and writes to a register component. + * In this case, we only want to increment dependencies by one. + */ + return; + } + + DBG("%i: read %i[%i] chan %i\n", s->Current->Instruction->IP, file, index, chan); + + reader = memory_pool_malloc(&s->C->Pool, sizeof(*reader)); + reader->Reader = s->Current; + if (!*v) { + /* In this situation, the instruction reads from a register + * that hasn't been written to or read from in the current + * block. */ + *v = memory_pool_malloc(&s->C->Pool, sizeof(struct reg_value)); + memset(*v, 0, sizeof(struct reg_value)); + (*v)->Readers = reader; + } else { + reader->Next = (*v)->Readers; + (*v)->Readers = reader; + /* Only update the current instruction's dependencies if the + * register it reads from has been written to in this block. */ + if ((*v)->Writer) { + s->Current->NumDependencies++; + } + } + (*v)->NumReaders++; + + if (s->Current->NumReadValues >= 12) { + rc_error(s->C, "%s: NumReadValues overflow\n", __FUNCTION__); + } else { + s->Current->ReadValues[s->Current->NumReadValues++] = *v; + } +} + +static void scan_write(void * data, struct rc_instruction * inst, + rc_register_file file, unsigned int index, unsigned int chan) +{ + struct schedule_state * s = data; + struct reg_value ** pv = get_reg_valuep(s, file, index, chan); + struct reg_value * newv; + + if (!pv) + return; + + DBG("%i: write %i[%i] chan %i\n", s->Current->Instruction->IP, file, index, chan); + + newv = memory_pool_malloc(&s->C->Pool, sizeof(*newv)); + memset(newv, 0, sizeof(*newv)); + + newv->Writer = s->Current; + + if (*pv) { + (*pv)->Next = newv; + s->Current->NumDependencies++; + } + + *pv = newv; + + if (s->Current->NumWriteValues >= 4) { + rc_error(s->C, "%s: NumWriteValues overflow\n", __FUNCTION__); + } else { + s->Current->WriteValues[s->Current->NumWriteValues++] = newv; + } +} + +static void is_rgb_to_alpha_possible_normal( + void * userdata, + struct rc_instruction * inst, + struct rc_src_register * src) +{ + struct rc_reader_data * reader_data = userdata; + reader_data->Abort = 1; + +} + +static void schedule_block(struct r300_fragment_program_compiler * c, + struct rc_instruction * begin, struct rc_instruction * end) +{ + struct schedule_state s; + unsigned int ip; + + memset(&s, 0, sizeof(s)); + s.C = &c->Base; + + /* Scan instructions for data dependencies */ + ip = 0; + for(struct rc_instruction * inst = begin; inst != end; inst = inst->Next) { + s.Current = memory_pool_malloc(&c->Base.Pool, sizeof(*s.Current)); + memset(s.Current, 0, sizeof(struct schedule_instruction)); + + s.Current->Instruction = inst; + inst->IP = ip++; + + DBG("%i: Scanning\n", inst->IP); + + /* The order of things here is subtle and maybe slightly + * counter-intuitive, to account for the case where an + * instruction writes to the same register as it reads + * from. */ + rc_for_all_writes_chan(inst, &scan_write, &s); + rc_for_all_reads_chan(inst, &scan_read, &s); + + DBG("%i: Has %i dependencies\n", inst->IP, s.Current->NumDependencies); + + if (!s.Current->NumDependencies) + instruction_ready(&s, s.Current); + + /* Get global readers for possible RGB->Alpha conversion. */ + s.Current->GlobalReaders.ExitOnAbort = 1; + rc_get_readers(s.C, inst, &s.Current->GlobalReaders, + is_rgb_to_alpha_possible_normal, + is_rgb_to_alpha_possible, NULL); + } + + /* Temporarily unlink all instructions */ + begin->Prev->Next = end; + end->Prev = begin->Prev; + + /* Schedule instructions back */ + while(!s.C->Error && + (s.ReadyTEX || s.ReadyRGB || s.ReadyAlpha || s.ReadyFullALU)) { + if (s.ReadyTEX) + emit_all_tex(&s, end); + + while(!s.C->Error && (s.ReadyFullALU || s.ReadyRGB || s.ReadyAlpha)) + emit_one_alu(&s, end); + } +} + +static int is_controlflow(struct rc_instruction * inst) +{ + if (inst->Type == RC_INSTRUCTION_NORMAL) { + const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode); + return opcode->IsFlowControl; + } + return 0; +} + +void rc_pair_schedule(struct radeon_compiler *cc, void *user) +{ + struct schedule_state s; + + struct r300_fragment_program_compiler *c = (struct r300_fragment_program_compiler*)cc; + struct rc_instruction * inst = c->Base.Program.Instructions.Next; + + memset(&s, 0, sizeof(s)); + s.C = &c->Base; + while(inst != &c->Base.Program.Instructions) { + struct rc_instruction * first; + + if (is_controlflow(inst)) { + inst = inst->Next; + continue; + } + + first = inst; + + while(inst != &c->Base.Program.Instructions && !is_controlflow(inst)) + inst = inst->Next; + + DBG("Schedule one block\n"); + schedule_block(c, first, inst); + } +} diff --git a/src/gallium/drivers/r300/compiler/radeon_pair_translate.c b/src/gallium/drivers/r300/compiler/radeon_pair_translate.c new file mode 100644 index 00000000000..2dae56a2428 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_pair_translate.c @@ -0,0 +1,359 @@ +/* + * Copyright (C) 2009 Nicolai Haehnle. + * + * 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. + * + */ + +#include "radeon_program_pair.h" + +#include "radeon_compiler.h" +#include "radeon_compiler_util.h" + + +/** + * Finally rewrite ADD, MOV, MUL as the appropriate native instruction + * and reverse the order of arguments for CMP. + */ +static void final_rewrite(struct rc_sub_instruction *inst) +{ + struct rc_src_register tmp; + + switch(inst->Opcode) { + case RC_OPCODE_ADD: + inst->SrcReg[2] = inst->SrcReg[1]; + inst->SrcReg[1].File = RC_FILE_NONE; + inst->SrcReg[1].Swizzle = RC_SWIZZLE_1111; + inst->SrcReg[1].Negate = RC_MASK_NONE; + inst->Opcode = RC_OPCODE_MAD; + break; + case RC_OPCODE_CMP: + tmp = inst->SrcReg[2]; + inst->SrcReg[2] = inst->SrcReg[0]; + inst->SrcReg[0] = tmp; + break; + case RC_OPCODE_MOV: + /* AMD say we should use CMP. + * However, when we transform + * KIL -r0; + * into + * CMP tmp, -r0, -r0, 0; + * KIL tmp; + * we get incorrect behaviour on R500 when r0 == 0.0. + * It appears that the R500 KIL hardware treats -0.0 as less + * than zero. + */ + inst->SrcReg[1].File = RC_FILE_NONE; + inst->SrcReg[1].Swizzle = RC_SWIZZLE_1111; + inst->SrcReg[2].File = RC_FILE_NONE; + inst->SrcReg[2].Swizzle = RC_SWIZZLE_0000; + inst->Opcode = RC_OPCODE_MAD; + break; + case RC_OPCODE_MUL: + inst->SrcReg[2].File = RC_FILE_NONE; + inst->SrcReg[2].Swizzle = RC_SWIZZLE_0000; + inst->Opcode = RC_OPCODE_MAD; + break; + default: + /* nothing to do */ + break; + } +} + + +/** + * Classify an instruction according to which ALUs etc. it needs + */ +static void classify_instruction(struct rc_sub_instruction * inst, + int * needrgb, int * needalpha, int * istranscendent) +{ + *needrgb = (inst->DstReg.WriteMask & RC_MASK_XYZ) ? 1 : 0; + *needalpha = (inst->DstReg.WriteMask & RC_MASK_W) ? 1 : 0; + *istranscendent = 0; + + if (inst->WriteALUResult == RC_ALURESULT_X) + *needrgb = 1; + else if (inst->WriteALUResult == RC_ALURESULT_W) + *needalpha = 1; + + switch(inst->Opcode) { + case RC_OPCODE_ADD: + case RC_OPCODE_CMP: + case RC_OPCODE_CND: + case RC_OPCODE_DDX: + case RC_OPCODE_DDY: + case RC_OPCODE_FRC: + case RC_OPCODE_MAD: + case RC_OPCODE_MAX: + case RC_OPCODE_MIN: + case RC_OPCODE_MOV: + case RC_OPCODE_MUL: + break; + case RC_OPCODE_COS: + case RC_OPCODE_EX2: + case RC_OPCODE_LG2: + case RC_OPCODE_RCP: + case RC_OPCODE_RSQ: + case RC_OPCODE_SIN: + *istranscendent = 1; + *needalpha = 1; + break; + case RC_OPCODE_DP4: + *needalpha = 1; + /* fall through */ + case RC_OPCODE_DP3: + *needrgb = 1; + break; + default: + break; + } +} + +static void src_uses(struct rc_src_register src, unsigned int * rgb, + unsigned int * alpha) +{ + int j; + for(j = 0; j < 4; ++j) { + unsigned int swz = GET_SWZ(src.Swizzle, j); + if (swz < 3) + *rgb = 1; + else if (swz < 4) + *alpha = 1; + } +} + +/** + * Fill the given ALU instruction's opcodes and source operands into the given pair, + * if possible. + */ +static void set_pair_instruction(struct r300_fragment_program_compiler *c, + struct rc_pair_instruction * pair, + struct rc_sub_instruction * inst) +{ + int needrgb, needalpha, istranscendent; + const struct rc_opcode_info * opcode; + int i; + + memset(pair, 0, sizeof(struct rc_pair_instruction)); + + classify_instruction(inst, &needrgb, &needalpha, &istranscendent); + + if (needrgb) { + if (istranscendent) + pair->RGB.Opcode = RC_OPCODE_REPL_ALPHA; + else + pair->RGB.Opcode = inst->Opcode; + if (inst->SaturateMode == RC_SATURATE_ZERO_ONE) + pair->RGB.Saturate = 1; + } + if (needalpha) { + pair->Alpha.Opcode = inst->Opcode; + if (inst->SaturateMode == RC_SATURATE_ZERO_ONE) + pair->Alpha.Saturate = 1; + } + + opcode = rc_get_opcode_info(inst->Opcode); + + /* Presubtract handling: + * We need to make sure that the values used by the presubtract + * operation end up in src0 or src1. */ + if(inst->PreSub.Opcode != RC_PRESUB_NONE) { + /* rc_pair_alloc_source() will fill in data for + * pair->{RGB,ALPHA}.Src[RC_PAIR_PRESUB_SRC] */ + int j; + for(j = 0; j < 3; j++) { + int src_regs; + if(inst->SrcReg[j].File != RC_FILE_PRESUB) + continue; + + src_regs = rc_presubtract_src_reg_count( + inst->PreSub.Opcode); + for(i = 0; i < src_regs; i++) { + unsigned int rgb = 0; + unsigned int alpha = 0; + src_uses(inst->SrcReg[j], &rgb, &alpha); + if(rgb) { + pair->RGB.Src[i].File = + inst->PreSub.SrcReg[i].File; + pair->RGB.Src[i].Index = + inst->PreSub.SrcReg[i].Index; + pair->RGB.Src[i].Used = 1; + } + if(alpha) { + pair->Alpha.Src[i].File = + inst->PreSub.SrcReg[i].File; + pair->Alpha.Src[i].Index = + inst->PreSub.SrcReg[i].Index; + pair->Alpha.Src[i].Used = 1; + } + } + } + } + + for(i = 0; i < opcode->NumSrcRegs; ++i) { + int source; + if (needrgb && !istranscendent) { + unsigned int srcrgb = 0; + unsigned int srcalpha = 0; + unsigned int srcmask = 0; + int j; + /* We don't care about the alpha channel here. We only + * want the part of the swizzle that writes to rgb, + * since we are creating an rgb instruction. */ + for(j = 0; j < 3; ++j) { + unsigned int swz = GET_SWZ(inst->SrcReg[i].Swizzle, j); + + if (swz < RC_SWIZZLE_W) + srcrgb = 1; + else if (swz == RC_SWIZZLE_W) + srcalpha = 1; + + if (swz < RC_SWIZZLE_UNUSED) + srcmask |= 1 << j; + } + source = rc_pair_alloc_source(pair, srcrgb, srcalpha, + inst->SrcReg[i].File, inst->SrcReg[i].Index); + if (source < 0) { + rc_error(&c->Base, "Failed to translate " + "rgb instruction.\n"); + return; + } + pair->RGB.Arg[i].Source = source; + pair->RGB.Arg[i].Swizzle = + rc_init_swizzle(inst->SrcReg[i].Swizzle, 3); + pair->RGB.Arg[i].Abs = inst->SrcReg[i].Abs; + pair->RGB.Arg[i].Negate = !!(srcmask & inst->SrcReg[i].Negate & (RC_MASK_X | RC_MASK_Y | RC_MASK_Z)); + } + if (needalpha) { + unsigned int srcrgb = 0; + unsigned int srcalpha = 0; + unsigned int swz = GET_SWZ(inst->SrcReg[i].Swizzle, istranscendent ? 0 : 3); + if (swz < 3) + srcrgb = 1; + else if (swz < 4) + srcalpha = 1; + source = rc_pair_alloc_source(pair, srcrgb, srcalpha, + inst->SrcReg[i].File, inst->SrcReg[i].Index); + if (source < 0) { + rc_error(&c->Base, "Failed to translate " + "alpha instruction.\n"); + return; + } + pair->Alpha.Arg[i].Source = source; + pair->Alpha.Arg[i].Swizzle = rc_init_swizzle(swz, 1); + pair->Alpha.Arg[i].Abs = inst->SrcReg[i].Abs; + pair->Alpha.Arg[i].Negate = !!(inst->SrcReg[i].Negate & RC_MASK_W); + } + } + + /* Destination handling */ + if (inst->DstReg.File == RC_FILE_OUTPUT) { + if (inst->DstReg.Index == c->OutputDepth) { + pair->Alpha.DepthWriteMask |= GET_BIT(inst->DstReg.WriteMask, 3); + } else { + for (i = 0; i < 4; i++) { + if (inst->DstReg.Index == c->OutputColor[i]) { + pair->RGB.Target = i; + pair->Alpha.Target = i; + pair->RGB.OutputWriteMask |= + inst->DstReg.WriteMask & RC_MASK_XYZ; + pair->Alpha.OutputWriteMask |= + GET_BIT(inst->DstReg.WriteMask, 3); + break; + } + } + } + } else { + if (needrgb) { + pair->RGB.DestIndex = inst->DstReg.Index; + pair->RGB.WriteMask |= inst->DstReg.WriteMask & RC_MASK_XYZ; + } + + if (needalpha) { + pair->Alpha.WriteMask |= (GET_BIT(inst->DstReg.WriteMask, 3) << 3); + if (pair->Alpha.WriteMask) { + pair->Alpha.DestIndex = inst->DstReg.Index; + } + } + } + + if (inst->WriteALUResult) { + pair->WriteALUResult = inst->WriteALUResult; + pair->ALUResultCompare = inst->ALUResultCompare; + } +} + + +static void check_opcode_support(struct r300_fragment_program_compiler *c, + struct rc_sub_instruction *inst) +{ + const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->Opcode); + + if (opcode->HasDstReg) { + if (inst->SaturateMode == RC_SATURATE_MINUS_PLUS_ONE) { + rc_error(&c->Base, "Fragment program does not support signed Saturate.\n"); + return; + } + } + + for (unsigned i = 0; i < opcode->NumSrcRegs; i++) { + if (inst->SrcReg[i].RelAddr) { + rc_error(&c->Base, "Fragment program does not support relative addressing " + " of source operands.\n"); + return; + } + } +} + + +/** + * Translate all ALU instructions into corresponding pair instructions, + * performing no other changes. + */ +void rc_pair_translate(struct radeon_compiler *cc, void *user) +{ + struct r300_fragment_program_compiler *c = (struct r300_fragment_program_compiler*)cc; + + for(struct rc_instruction * inst = c->Base.Program.Instructions.Next; + inst != &c->Base.Program.Instructions; + inst = inst->Next) { + const struct rc_opcode_info * opcode; + struct rc_sub_instruction copy; + + if (inst->Type != RC_INSTRUCTION_NORMAL) + continue; + + opcode = rc_get_opcode_info(inst->U.I.Opcode); + + if (opcode->HasTexture || opcode->IsFlowControl || opcode->Opcode == RC_OPCODE_KIL) + continue; + + copy = inst->U.I; + + check_opcode_support(c, ©); + + final_rewrite(©); + inst->Type = RC_INSTRUCTION_PAIR; + set_pair_instruction(c, &inst->U.P, ©); + } +} diff --git a/src/gallium/drivers/r300/compiler/radeon_program.c b/src/gallium/drivers/r300/compiler/radeon_program.c new file mode 100644 index 00000000000..fe5756ebc45 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_program.c @@ -0,0 +1,225 @@ +/* + * Copyright (C) 2008 Nicolai Haehnle. + * + * 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. + * + */ + +#include "radeon_program.h" + +#include <stdio.h> + +#include "radeon_compiler.h" +#include "radeon_dataflow.h" + + +/** + * Transform the given clause in the following way: + * 1. Replace it with an empty clause + * 2. For every instruction in the original clause, try the given + * transformations in order. + * 3. If one of the transformations returns GL_TRUE, assume that it + * has emitted the appropriate instruction(s) into the new clause; + * otherwise, copy the instruction verbatim. + * + * \note The transformation is currently not recursive; in other words, + * instructions emitted by transformations are not transformed. + * + * \note The transform is called 'local' because it can only look at + * one instruction at a time. + */ +void rc_local_transform( + struct radeon_compiler * c, + void *user) +{ + struct radeon_program_transformation *transformations = + (struct radeon_program_transformation*)user; + struct rc_instruction * inst = c->Program.Instructions.Next; + + while(inst != &c->Program.Instructions) { + struct rc_instruction * current = inst; + int i; + + inst = inst->Next; + + for(i = 0; transformations[i].function; ++i) { + struct radeon_program_transformation* t = transformations + i; + + if (t->function(c, current, t->userData)) + break; + } + } +} + +struct get_used_temporaries_data { + unsigned char * Used; + unsigned int UsedLength; +}; + +static void get_used_temporaries_cb( + void * userdata, + struct rc_instruction * inst, + rc_register_file file, + unsigned int index, + unsigned int mask) +{ + struct get_used_temporaries_data * d = userdata; + + if (file != RC_FILE_TEMPORARY) + return; + + if (index >= d->UsedLength) + return; + + d->Used[index] |= mask; +} + +/** + * This function fills in the parameter 'used' with a writemask that + * represent which components of each temporary register are used by the + * program. This is meant to be combined with rc_find_free_temporary_list as a + * more efficient version of rc_find_free_temporary. + * @param used The function does not initialize this parameter. + */ +void rc_get_used_temporaries( + struct radeon_compiler * c, + unsigned char * used, + unsigned int used_length) +{ + struct rc_instruction * inst; + struct get_used_temporaries_data d; + d.Used = used; + d.UsedLength = used_length; + + for(inst = c->Program.Instructions.Next; + inst != &c->Program.Instructions; inst = inst->Next) { + + rc_for_all_reads_mask(inst, get_used_temporaries_cb, &d); + rc_for_all_writes_mask(inst, get_used_temporaries_cb, &d); + } +} + +/* Search a list of used temporaries for a free one + * \sa rc_get_used_temporaries + * @note If this functions finds a free temporary, it will mark it as used + * in the used temporary list (param 'used') + * @param used list of used temporaries + * @param used_length number of items in param 'used' + * @param mask which components must be free in the temporary index that is + * returned. + * @return -1 If there are no more free temporaries, otherwise the index of + * a temporary register where the components specified in param 'mask' are + * not being used. + */ +int rc_find_free_temporary_list( + struct radeon_compiler * c, + unsigned char * used, + unsigned int used_length, + unsigned int mask) +{ + int i; + for(i = 0; i < used_length; i++) { + if ((~used[i] & mask) == mask) { + used[i] |= mask; + return i; + } + } + return -1; +} + +unsigned int rc_find_free_temporary(struct radeon_compiler * c) +{ + unsigned char used[RC_REGISTER_MAX_INDEX]; + int free; + + memset(used, 0, sizeof(used)); + + rc_get_used_temporaries(c, used, RC_REGISTER_MAX_INDEX); + + free = rc_find_free_temporary_list(c, used, RC_REGISTER_MAX_INDEX, + RC_MASK_XYZW); + if (free < 0) { + rc_error(c, "Ran out of temporary registers\n"); + return 0; + } + return free; +} + + +struct rc_instruction *rc_alloc_instruction(struct radeon_compiler * c) +{ + struct rc_instruction * inst = memory_pool_malloc(&c->Pool, sizeof(struct rc_instruction)); + + memset(inst, 0, sizeof(struct rc_instruction)); + + inst->U.I.Opcode = RC_OPCODE_ILLEGAL_OPCODE; + inst->U.I.DstReg.WriteMask = RC_MASK_XYZW; + inst->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_XYZW; + inst->U.I.SrcReg[1].Swizzle = RC_SWIZZLE_XYZW; + inst->U.I.SrcReg[2].Swizzle = RC_SWIZZLE_XYZW; + + return inst; +} + +void rc_insert_instruction(struct rc_instruction * after, struct rc_instruction * inst) +{ + inst->Prev = after; + inst->Next = after->Next; + + inst->Prev->Next = inst; + inst->Next->Prev = inst; +} + +struct rc_instruction *rc_insert_new_instruction(struct radeon_compiler * c, struct rc_instruction * after) +{ + struct rc_instruction * inst = rc_alloc_instruction(c); + + rc_insert_instruction(after, inst); + + return inst; +} + +void rc_remove_instruction(struct rc_instruction * inst) +{ + inst->Prev->Next = inst->Next; + inst->Next->Prev = inst->Prev; +} + +/** + * Return the number of instructions in the program. + */ +unsigned int rc_recompute_ips(struct radeon_compiler * c) +{ + unsigned int ip = 0; + struct rc_instruction * inst; + + for(inst = c->Program.Instructions.Next; + inst != &c->Program.Instructions; + inst = inst->Next) { + inst->IP = ip++; + } + + c->Program.Instructions.IP = 0xcafedead; + + return ip; +} diff --git a/src/gallium/drivers/r300/compiler/radeon_program.h b/src/gallium/drivers/r300/compiler/radeon_program.h new file mode 100644 index 00000000000..b899eccbf53 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_program.h @@ -0,0 +1,206 @@ +/* + * Copyright (C) 2008 Nicolai Haehnle. + * + * 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. + * + */ + +#ifndef __RADEON_PROGRAM_H_ +#define __RADEON_PROGRAM_H_ + +#include <stdint.h> +#include <string.h> + +#include "radeon_opcodes.h" +#include "radeon_code.h" +#include "radeon_program_constants.h" +#include "radeon_program_pair.h" + +struct radeon_compiler; + +struct rc_src_register { + unsigned int File:4; + + /** Negative values may be used for relative addressing. */ + signed int Index:(RC_REGISTER_INDEX_BITS+1); + unsigned int RelAddr:1; + + unsigned int Swizzle:12; + + /** Take the component-wise absolute value */ + unsigned int Abs:1; + + /** Post-Abs negation. */ + unsigned int Negate:4; +}; + +struct rc_dst_register { + unsigned int File:3; + unsigned int Index:RC_REGISTER_INDEX_BITS; + unsigned int WriteMask:4; +}; + +struct rc_presub_instruction { + rc_presubtract_op Opcode; + struct rc_src_register SrcReg[2]; +}; + +/** + * Instructions are maintained by the compiler in a doubly linked list + * of these structures. + * + * This instruction format is intended to be expanded for hardware-specific + * trickery. At different stages of compilation, a different set of + * instruction types may be valid. + */ +struct rc_sub_instruction { + struct rc_src_register SrcReg[3]; + struct rc_dst_register DstReg; + + /** + * Opcode of this instruction, according to \ref rc_opcode enums. + */ + unsigned int Opcode:8; + + /** + * Saturate each value of the result to the range [0,1] or [-1,1], + * according to \ref rc_saturate_mode enums. + */ + unsigned int SaturateMode:2; + + /** + * Writing to the special register RC_SPECIAL_ALU_RESULT + */ + /*@{*/ + unsigned int WriteALUResult:2; + unsigned int ALUResultCompare:3; + /*@}*/ + + /** + * \name Extra fields for TEX, TXB, TXD, TXL, TXP instructions. + */ + /*@{*/ + /** Source texture unit. */ + unsigned int TexSrcUnit:5; + + /** Source texture target, one of the \ref rc_texture_target enums */ + unsigned int TexSrcTarget:3; + + /** True if tex instruction should do shadow comparison */ + unsigned int TexShadow:1; + + /**R500 Only. How to swizzle the result of a TEX lookup*/ + unsigned int TexSwizzle:12; + /*@}*/ + + /** This holds information about the presubtract operation used by + * this instruction. */ + struct rc_presub_instruction PreSub; +}; + +typedef enum { + RC_INSTRUCTION_NORMAL = 0, + RC_INSTRUCTION_PAIR +} rc_instruction_type; + +struct rc_instruction { + struct rc_instruction * Prev; + struct rc_instruction * Next; + + rc_instruction_type Type; + union { + struct rc_sub_instruction I; + struct rc_pair_instruction P; + } U; + + /** + * Warning: IPs are not stable. If you want to use them, + * you need to recompute them at the beginning of each pass + * using \ref rc_recompute_ips + */ + unsigned int IP; +}; + +struct rc_program { + /** + * Instructions.Next points to the first instruction, + * Instructions.Prev points to the last instruction. + */ + struct rc_instruction Instructions; + + /* Long term, we should probably remove InputsRead & OutputsWritten, + * since updating dependent state can be fragile, and they aren't + * actually used very often. */ + uint32_t InputsRead; + uint32_t OutputsWritten; + uint32_t ShadowSamplers; /**< Texture units used for shadow sampling. */ + + struct rc_constant_list Constants; +}; + +/** + * A transformation that can be passed to \ref rc_local_transform. + * + * The function will be called once for each instruction. + * It has to either emit the appropriate transformed code for the instruction + * and return true, or return false if it doesn't understand the + * instruction. + * + * The function gets passed the userData as last parameter. + */ +struct radeon_program_transformation { + int (*function)( + struct radeon_compiler*, + struct rc_instruction*, + void*); + void *userData; +}; + +void rc_local_transform( + struct radeon_compiler *c, + void *user); + +void rc_get_used_temporaries( + struct radeon_compiler * c, + unsigned char * used, + unsigned int used_length); + +int rc_find_free_temporary_list( + struct radeon_compiler * c, + unsigned char * used, + unsigned int used_length, + unsigned int mask); + +unsigned int rc_find_free_temporary(struct radeon_compiler * c); + +struct rc_instruction *rc_alloc_instruction(struct radeon_compiler * c); +struct rc_instruction *rc_insert_new_instruction(struct radeon_compiler * c, struct rc_instruction * after); +void rc_insert_instruction(struct rc_instruction * after, struct rc_instruction * inst); +void rc_remove_instruction(struct rc_instruction * inst); + +unsigned int rc_recompute_ips(struct radeon_compiler * c); + +void rc_print_program(const struct rc_program *prog); + +rc_swizzle rc_mask_to_swizzle(unsigned int mask); +#endif diff --git a/src/gallium/drivers/r300/compiler/radeon_program_alu.c b/src/gallium/drivers/r300/compiler/radeon_program_alu.c new file mode 100644 index 00000000000..e273bc40c26 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_program_alu.c @@ -0,0 +1,1154 @@ +/* + * Copyright (C) 2008 Nicolai Haehnle. + * + * 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 + * + * Shareable transformations that transform "special" ALU instructions + * into ALU instructions that are supported by hardware. + * + */ + +#include "radeon_program_alu.h" + +#include "radeon_compiler.h" +#include "radeon_compiler_util.h" + + +static struct rc_instruction *emit1( + struct radeon_compiler * c, struct rc_instruction * after, + rc_opcode Opcode, rc_saturate_mode Saturate, struct rc_dst_register DstReg, + struct rc_src_register SrcReg) +{ + struct rc_instruction *fpi = rc_insert_new_instruction(c, after); + + fpi->U.I.Opcode = Opcode; + fpi->U.I.SaturateMode = Saturate; + fpi->U.I.DstReg = DstReg; + fpi->U.I.SrcReg[0] = SrcReg; + return fpi; +} + +static struct rc_instruction *emit2( + struct radeon_compiler * c, struct rc_instruction * after, + rc_opcode Opcode, rc_saturate_mode Saturate, struct rc_dst_register DstReg, + struct rc_src_register SrcReg0, struct rc_src_register SrcReg1) +{ + struct rc_instruction *fpi = rc_insert_new_instruction(c, after); + + fpi->U.I.Opcode = Opcode; + fpi->U.I.SaturateMode = Saturate; + fpi->U.I.DstReg = DstReg; + fpi->U.I.SrcReg[0] = SrcReg0; + fpi->U.I.SrcReg[1] = SrcReg1; + return fpi; +} + +static struct rc_instruction *emit3( + struct radeon_compiler * c, struct rc_instruction * after, + rc_opcode Opcode, rc_saturate_mode Saturate, struct rc_dst_register DstReg, + struct rc_src_register SrcReg0, struct rc_src_register SrcReg1, + struct rc_src_register SrcReg2) +{ + struct rc_instruction *fpi = rc_insert_new_instruction(c, after); + + fpi->U.I.Opcode = Opcode; + fpi->U.I.SaturateMode = Saturate; + fpi->U.I.DstReg = DstReg; + fpi->U.I.SrcReg[0] = SrcReg0; + fpi->U.I.SrcReg[1] = SrcReg1; + fpi->U.I.SrcReg[2] = SrcReg2; + return fpi; +} + +static struct rc_dst_register dstregtmpmask(int index, int mask) +{ + struct rc_dst_register dst = {0, 0, 0}; + dst.File = RC_FILE_TEMPORARY; + dst.Index = index; + dst.WriteMask = mask; + return dst; +} + +static const struct rc_src_register builtin_zero = { + .File = RC_FILE_NONE, + .Index = 0, + .Swizzle = RC_SWIZZLE_0000 +}; +static const struct rc_src_register builtin_one = { + .File = RC_FILE_NONE, + .Index = 0, + .Swizzle = RC_SWIZZLE_1111 +}; +static const struct rc_src_register srcreg_undefined = { + .File = RC_FILE_NONE, + .Index = 0, + .Swizzle = RC_SWIZZLE_XYZW +}; + +static struct rc_src_register srcreg(int file, int index) +{ + struct rc_src_register src = srcreg_undefined; + src.File = file; + src.Index = index; + return src; +} + +static struct rc_src_register srcregswz(int file, int index, int swz) +{ + struct rc_src_register src = srcreg_undefined; + src.File = file; + src.Index = index; + src.Swizzle = swz; + return src; +} + +static struct rc_src_register absolute(struct rc_src_register reg) +{ + struct rc_src_register newreg = reg; + newreg.Abs = 1; + newreg.Negate = RC_MASK_NONE; + return newreg; +} + +static struct rc_src_register negate(struct rc_src_register reg) +{ + struct rc_src_register newreg = reg; + newreg.Negate = newreg.Negate ^ RC_MASK_XYZW; + return newreg; +} + +static struct rc_src_register swizzle(struct rc_src_register reg, + rc_swizzle x, rc_swizzle y, rc_swizzle z, rc_swizzle w) +{ + struct rc_src_register swizzled = reg; + swizzled.Swizzle = combine_swizzles4(reg.Swizzle, x, y, z, w); + return swizzled; +} + +static struct rc_src_register swizzle_smear(struct rc_src_register reg, + rc_swizzle x) +{ + return swizzle(reg, x, x, x, x); +} + +static struct rc_src_register swizzle_xxxx(struct rc_src_register reg) +{ + return swizzle_smear(reg, RC_SWIZZLE_X); +} + +static struct rc_src_register swizzle_yyyy(struct rc_src_register reg) +{ + return swizzle_smear(reg, RC_SWIZZLE_Y); +} + +static struct rc_src_register swizzle_zzzz(struct rc_src_register reg) +{ + return swizzle_smear(reg, RC_SWIZZLE_Z); +} + +static struct rc_src_register swizzle_wwww(struct rc_src_register reg) +{ + return swizzle_smear(reg, RC_SWIZZLE_W); +} + +static int is_dst_safe_to_reuse(struct rc_instruction *inst) +{ + const struct rc_opcode_info *info = rc_get_opcode_info(inst->U.I.Opcode); + unsigned i; + + assert(info->HasDstReg); + + if (inst->U.I.DstReg.File != RC_FILE_TEMPORARY) + return 0; + + for (i = 0; i < info->NumSrcRegs; i++) { + if (inst->U.I.SrcReg[i].File == RC_FILE_TEMPORARY && + inst->U.I.SrcReg[i].Index == inst->U.I.DstReg.Index) + return 0; + } + + return 1; +} + +static struct rc_dst_register try_to_reuse_dst(struct radeon_compiler *c, + struct rc_instruction *inst) +{ + unsigned tmp; + + if (is_dst_safe_to_reuse(inst)) + tmp = inst->U.I.DstReg.Index; + else + tmp = rc_find_free_temporary(c); + + return dstregtmpmask(tmp, inst->U.I.DstReg.WriteMask); +} + +static void transform_ABS(struct radeon_compiler* c, + struct rc_instruction* inst) +{ + struct rc_src_register src = inst->U.I.SrcReg[0]; + src.Abs = 1; + src.Negate = RC_MASK_NONE; + emit1(c, inst->Prev, RC_OPCODE_MOV, inst->U.I.SaturateMode, inst->U.I.DstReg, src); + rc_remove_instruction(inst); +} + +static void transform_CEIL(struct radeon_compiler* c, + struct rc_instruction* inst) +{ + /* Assuming: + * ceil(x) = -floor(-x) + * + * After inlining floor: + * ceil(x) = -(-x-frac(-x)) + * + * After simplification: + * ceil(x) = x+frac(-x) + */ + + struct rc_dst_register dst = try_to_reuse_dst(c, inst); + emit1(c, inst->Prev, RC_OPCODE_FRC, 0, dst, negate(inst->U.I.SrcReg[0])); + emit2(c, inst->Prev, RC_OPCODE_ADD, inst->U.I.SaturateMode, inst->U.I.DstReg, + inst->U.I.SrcReg[0], srcreg(RC_FILE_TEMPORARY, dst.Index)); + rc_remove_instruction(inst); +} + +static void transform_CLAMP(struct radeon_compiler *c, + struct rc_instruction *inst) +{ + /* CLAMP dst, src, min, max + * into: + * MIN tmp, src, max + * MAX dst, tmp, min + */ + struct rc_dst_register dst = try_to_reuse_dst(c, inst); + emit2(c, inst->Prev, RC_OPCODE_MIN, 0, dst, + inst->U.I.SrcReg[0], inst->U.I.SrcReg[2]); + emit2(c, inst->Prev, RC_OPCODE_MAX, inst->U.I.SaturateMode, inst->U.I.DstReg, + srcreg(RC_FILE_TEMPORARY, dst.Index), inst->U.I.SrcReg[1]); + rc_remove_instruction(inst); +} + +static void transform_DP2(struct radeon_compiler* c, + struct rc_instruction* inst) +{ + struct rc_src_register src0 = inst->U.I.SrcReg[0]; + struct rc_src_register src1 = inst->U.I.SrcReg[1]; + src0.Negate &= ~(RC_MASK_Z | RC_MASK_W); + src0.Swizzle &= ~(63 << (3 * 2)); + src0.Swizzle |= (RC_SWIZZLE_ZERO << (3 * 2)) | (RC_SWIZZLE_ZERO << (3 * 3)); + src1.Negate &= ~(RC_MASK_Z | RC_MASK_W); + src1.Swizzle &= ~(63 << (3 * 2)); + src1.Swizzle |= (RC_SWIZZLE_ZERO << (3 * 2)) | (RC_SWIZZLE_ZERO << (3 * 3)); + emit2(c, inst->Prev, RC_OPCODE_DP3, inst->U.I.SaturateMode, inst->U.I.DstReg, src0, src1); + rc_remove_instruction(inst); +} + +static void transform_DPH(struct radeon_compiler* c, + struct rc_instruction* inst) +{ + struct rc_src_register src0 = inst->U.I.SrcReg[0]; + src0.Negate &= ~RC_MASK_W; + src0.Swizzle &= ~(7 << (3 * 3)); + src0.Swizzle |= RC_SWIZZLE_ONE << (3 * 3); + emit2(c, inst->Prev, RC_OPCODE_DP4, inst->U.I.SaturateMode, inst->U.I.DstReg, src0, inst->U.I.SrcReg[1]); + rc_remove_instruction(inst); +} + +/** + * [1, src0.y*src1.y, src0.z, src1.w] + * So basically MUL with lotsa swizzling. + */ +static void transform_DST(struct radeon_compiler* c, + struct rc_instruction* inst) +{ + emit2(c, inst->Prev, RC_OPCODE_MUL, inst->U.I.SaturateMode, inst->U.I.DstReg, + swizzle(inst->U.I.SrcReg[0], RC_SWIZZLE_ONE, RC_SWIZZLE_Y, RC_SWIZZLE_Z, RC_SWIZZLE_ONE), + swizzle(inst->U.I.SrcReg[1], RC_SWIZZLE_ONE, RC_SWIZZLE_Y, RC_SWIZZLE_ONE, RC_SWIZZLE_W)); + rc_remove_instruction(inst); +} + +static void transform_FLR(struct radeon_compiler* c, + struct rc_instruction* inst) +{ + struct rc_dst_register dst = try_to_reuse_dst(c, inst); + emit1(c, inst->Prev, RC_OPCODE_FRC, 0, dst, inst->U.I.SrcReg[0]); + emit2(c, inst->Prev, RC_OPCODE_ADD, inst->U.I.SaturateMode, inst->U.I.DstReg, + inst->U.I.SrcReg[0], negate(srcreg(RC_FILE_TEMPORARY, dst.Index))); + rc_remove_instruction(inst); +} + +/** + * Definition of LIT (from ARB_fragment_program): + * + * tmp = VectorLoad(op0); + * if (tmp.x < 0) tmp.x = 0; + * if (tmp.y < 0) tmp.y = 0; + * if (tmp.w < -(128.0-epsilon)) tmp.w = -(128.0-epsilon); + * else if (tmp.w > 128-epsilon) tmp.w = 128-epsilon; + * result.x = 1.0; + * result.y = tmp.x; + * result.z = (tmp.x > 0) ? RoughApproxPower(tmp.y, tmp.w) : 0.0; + * result.w = 1.0; + * + * The longest path of computation is the one leading to result.z, + * consisting of 5 operations. This implementation of LIT takes + * 5 slots, if the subsequent optimization passes are clever enough + * to pair instructions correctly. + */ +static void transform_LIT(struct radeon_compiler* c, + struct rc_instruction* inst) +{ + unsigned int constant; + unsigned int constant_swizzle; + unsigned int temp; + struct rc_src_register srctemp; + + constant = rc_constants_add_immediate_scalar(&c->Program.Constants, -127.999999, &constant_swizzle); + + if (inst->U.I.DstReg.WriteMask != RC_MASK_XYZW || inst->U.I.DstReg.File != RC_FILE_TEMPORARY) { + struct rc_instruction * inst_mov; + + inst_mov = emit1(c, inst, + RC_OPCODE_MOV, 0, inst->U.I.DstReg, + srcreg(RC_FILE_TEMPORARY, rc_find_free_temporary(c))); + + inst->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst->U.I.DstReg.Index = inst_mov->U.I.SrcReg[0].Index; + inst->U.I.DstReg.WriteMask = RC_MASK_XYZW; + } + + temp = inst->U.I.DstReg.Index; + srctemp = srcreg(RC_FILE_TEMPORARY, temp); + + /* tmp.x = max(0.0, Src.x); */ + /* tmp.y = max(0.0, Src.y); */ + /* tmp.w = clamp(Src.z, -128+eps, 128-eps); */ + emit2(c, inst->Prev, RC_OPCODE_MAX, 0, + dstregtmpmask(temp, RC_MASK_XYW), + inst->U.I.SrcReg[0], + swizzle(srcreg(RC_FILE_CONSTANT, constant), + RC_SWIZZLE_ZERO, RC_SWIZZLE_ZERO, RC_SWIZZLE_ZERO, constant_swizzle&3)); + emit2(c, inst->Prev, RC_OPCODE_MIN, 0, + dstregtmpmask(temp, RC_MASK_Z), + swizzle_wwww(srctemp), + negate(srcregswz(RC_FILE_CONSTANT, constant, constant_swizzle))); + + /* tmp.w = Pow(tmp.y, tmp.w) */ + emit1(c, inst->Prev, RC_OPCODE_LG2, 0, + dstregtmpmask(temp, RC_MASK_W), + swizzle_yyyy(srctemp)); + emit2(c, inst->Prev, RC_OPCODE_MUL, 0, + dstregtmpmask(temp, RC_MASK_W), + swizzle_wwww(srctemp), + swizzle_zzzz(srctemp)); + emit1(c, inst->Prev, RC_OPCODE_EX2, 0, + dstregtmpmask(temp, RC_MASK_W), + swizzle_wwww(srctemp)); + + /* tmp.z = (tmp.x > 0) ? tmp.w : 0.0 */ + emit3(c, inst->Prev, RC_OPCODE_CMP, inst->U.I.SaturateMode, + dstregtmpmask(temp, RC_MASK_Z), + negate(swizzle_xxxx(srctemp)), + swizzle_wwww(srctemp), + builtin_zero); + + /* tmp.x, tmp.y, tmp.w = 1.0, tmp.x, 1.0 */ + emit1(c, inst->Prev, RC_OPCODE_MOV, inst->U.I.SaturateMode, + dstregtmpmask(temp, RC_MASK_XYW), + swizzle(srctemp, RC_SWIZZLE_ONE, RC_SWIZZLE_X, RC_SWIZZLE_ONE, RC_SWIZZLE_ONE)); + + rc_remove_instruction(inst); +} + +static void transform_LRP(struct radeon_compiler* c, + struct rc_instruction* inst) +{ + struct rc_dst_register dst = try_to_reuse_dst(c, inst); + + emit2(c, inst->Prev, RC_OPCODE_ADD, 0, + dst, + inst->U.I.SrcReg[1], negate(inst->U.I.SrcReg[2])); + emit3(c, inst->Prev, RC_OPCODE_MAD, inst->U.I.SaturateMode, + inst->U.I.DstReg, + inst->U.I.SrcReg[0], srcreg(RC_FILE_TEMPORARY, dst.Index), inst->U.I.SrcReg[2]); + + rc_remove_instruction(inst); +} + +static void transform_POW(struct radeon_compiler* c, + struct rc_instruction* inst) +{ + struct rc_dst_register tempdst = try_to_reuse_dst(c, inst); + struct rc_src_register tempsrc = srcreg(RC_FILE_TEMPORARY, tempdst.Index); + tempdst.WriteMask = RC_MASK_W; + tempsrc.Swizzle = RC_SWIZZLE_WWWW; + + emit1(c, inst->Prev, RC_OPCODE_LG2, 0, tempdst, swizzle_xxxx(inst->U.I.SrcReg[0])); + emit2(c, inst->Prev, RC_OPCODE_MUL, 0, tempdst, tempsrc, swizzle_xxxx(inst->U.I.SrcReg[1])); + emit1(c, inst->Prev, RC_OPCODE_EX2, inst->U.I.SaturateMode, inst->U.I.DstReg, tempsrc); + + rc_remove_instruction(inst); +} + +static void transform_RSQ(struct radeon_compiler* c, + struct rc_instruction* inst) +{ + inst->U.I.SrcReg[0] = absolute(inst->U.I.SrcReg[0]); +} + +static void transform_SEQ(struct radeon_compiler* c, + struct rc_instruction* inst) +{ + struct rc_dst_register dst = try_to_reuse_dst(c, inst); + + emit2(c, inst->Prev, RC_OPCODE_ADD, 0, dst, inst->U.I.SrcReg[0], negate(inst->U.I.SrcReg[1])); + emit3(c, inst->Prev, RC_OPCODE_CMP, inst->U.I.SaturateMode, inst->U.I.DstReg, + negate(absolute(srcreg(RC_FILE_TEMPORARY, dst.Index))), builtin_zero, builtin_one); + + rc_remove_instruction(inst); +} + +static void transform_SFL(struct radeon_compiler* c, + struct rc_instruction* inst) +{ + emit1(c, inst->Prev, RC_OPCODE_MOV, inst->U.I.SaturateMode, inst->U.I.DstReg, builtin_zero); + rc_remove_instruction(inst); +} + +static void transform_SGE(struct radeon_compiler* c, + struct rc_instruction* inst) +{ + struct rc_dst_register dst = try_to_reuse_dst(c, inst); + + emit2(c, inst->Prev, RC_OPCODE_ADD, 0, dst, inst->U.I.SrcReg[0], negate(inst->U.I.SrcReg[1])); + emit3(c, inst->Prev, RC_OPCODE_CMP, inst->U.I.SaturateMode, inst->U.I.DstReg, + srcreg(RC_FILE_TEMPORARY, dst.Index), builtin_zero, builtin_one); + + rc_remove_instruction(inst); +} + +static void transform_SGT(struct radeon_compiler* c, + struct rc_instruction* inst) +{ + struct rc_dst_register dst = try_to_reuse_dst(c, inst); + + emit2(c, inst->Prev, RC_OPCODE_ADD, 0, dst, negate(inst->U.I.SrcReg[0]), inst->U.I.SrcReg[1]); + emit3(c, inst->Prev, RC_OPCODE_CMP, inst->U.I.SaturateMode, inst->U.I.DstReg, + srcreg(RC_FILE_TEMPORARY, dst.Index), builtin_one, builtin_zero); + + rc_remove_instruction(inst); +} + +static void transform_SLE(struct radeon_compiler* c, + struct rc_instruction* inst) +{ + struct rc_dst_register dst = try_to_reuse_dst(c, inst); + + emit2(c, inst->Prev, RC_OPCODE_ADD, 0, dst, negate(inst->U.I.SrcReg[0]), inst->U.I.SrcReg[1]); + emit3(c, inst->Prev, RC_OPCODE_CMP, inst->U.I.SaturateMode, inst->U.I.DstReg, + srcreg(RC_FILE_TEMPORARY, dst.Index), builtin_zero, builtin_one); + + rc_remove_instruction(inst); +} + +static void transform_SLT(struct radeon_compiler* c, + struct rc_instruction* inst) +{ + struct rc_dst_register dst = try_to_reuse_dst(c, inst); + + emit2(c, inst->Prev, RC_OPCODE_ADD, 0, dst, inst->U.I.SrcReg[0], negate(inst->U.I.SrcReg[1])); + emit3(c, inst->Prev, RC_OPCODE_CMP, inst->U.I.SaturateMode, inst->U.I.DstReg, + srcreg(RC_FILE_TEMPORARY, dst.Index), builtin_one, builtin_zero); + + rc_remove_instruction(inst); +} + +static void transform_SNE(struct radeon_compiler* c, + struct rc_instruction* inst) +{ + struct rc_dst_register dst = try_to_reuse_dst(c, inst); + + emit2(c, inst->Prev, RC_OPCODE_ADD, 0, dst, inst->U.I.SrcReg[0], negate(inst->U.I.SrcReg[1])); + emit3(c, inst->Prev, RC_OPCODE_CMP, inst->U.I.SaturateMode, inst->U.I.DstReg, + negate(absolute(srcreg(RC_FILE_TEMPORARY, dst.Index))), builtin_one, builtin_zero); + + rc_remove_instruction(inst); +} + +static void transform_SSG(struct radeon_compiler* c, + struct rc_instruction* inst) +{ + /* result = sign(x) + * + * CMP tmp0, -x, 1, 0 + * CMP tmp1, x, 1, 0 + * ADD result, tmp0, -tmp1; + */ + struct rc_dst_register dst0; + unsigned tmp1; + + /* 0 < x */ + dst0 = try_to_reuse_dst(c, inst); + emit3(c, inst->Prev, RC_OPCODE_CMP, 0, + dst0, + negate(inst->U.I.SrcReg[0]), + builtin_one, + builtin_zero); + + /* x < 0 */ + tmp1 = rc_find_free_temporary(c); + emit3(c, inst->Prev, RC_OPCODE_CMP, 0, + dstregtmpmask(tmp1, inst->U.I.DstReg.WriteMask), + inst->U.I.SrcReg[0], + builtin_one, + builtin_zero); + + /* Either both are zero, or one of them is one and the other is zero. */ + /* result = tmp0 - tmp1 */ + emit2(c, inst->Prev, RC_OPCODE_ADD, 0, + inst->U.I.DstReg, + srcreg(RC_FILE_TEMPORARY, dst0.Index), + negate(srcreg(RC_FILE_TEMPORARY, tmp1))); + + rc_remove_instruction(inst); +} + +static void transform_SUB(struct radeon_compiler* c, + struct rc_instruction* inst) +{ + inst->U.I.Opcode = RC_OPCODE_ADD; + inst->U.I.SrcReg[1] = negate(inst->U.I.SrcReg[1]); +} + +static void transform_SWZ(struct radeon_compiler* c, + struct rc_instruction* inst) +{ + inst->U.I.Opcode = RC_OPCODE_MOV; +} + +static void transform_XPD(struct radeon_compiler* c, + struct rc_instruction* inst) +{ + struct rc_dst_register dst = try_to_reuse_dst(c, inst); + + emit2(c, inst->Prev, RC_OPCODE_MUL, 0, dst, + swizzle(inst->U.I.SrcReg[0], RC_SWIZZLE_Z, RC_SWIZZLE_X, RC_SWIZZLE_Y, RC_SWIZZLE_W), + swizzle(inst->U.I.SrcReg[1], RC_SWIZZLE_Y, RC_SWIZZLE_Z, RC_SWIZZLE_X, RC_SWIZZLE_W)); + emit3(c, inst->Prev, RC_OPCODE_MAD, inst->U.I.SaturateMode, inst->U.I.DstReg, + swizzle(inst->U.I.SrcReg[0], RC_SWIZZLE_Y, RC_SWIZZLE_Z, RC_SWIZZLE_X, RC_SWIZZLE_W), + swizzle(inst->U.I.SrcReg[1], RC_SWIZZLE_Z, RC_SWIZZLE_X, RC_SWIZZLE_Y, RC_SWIZZLE_W), + negate(srcreg(RC_FILE_TEMPORARY, dst.Index))); + + rc_remove_instruction(inst); +} + + +/** + * Can be used as a transformation for @ref radeonClauseLocalTransform, + * no userData necessary. + * + * Eliminates the following ALU instructions: + * ABS, CEIL, DPH, DST, FLR, LIT, LRP, POW, SEQ, SFL, SGE, SGT, SLE, SLT, SNE, SUB, SWZ, XPD + * using: + * MOV, ADD, MUL, MAD, FRC, DP3, LG2, EX2, CMP + * + * Transforms RSQ to Radeon's native RSQ by explicitly setting + * absolute value. + * + * @note should be applicable to R300 and R500 fragment programs. + */ +int radeonTransformALU( + struct radeon_compiler * c, + struct rc_instruction* inst, + void* unused) +{ + switch(inst->U.I.Opcode) { + case RC_OPCODE_ABS: transform_ABS(c, inst); return 1; + case RC_OPCODE_CEIL: transform_CEIL(c, inst); return 1; + case RC_OPCODE_CLAMP: transform_CLAMP(c, inst); return 1; + case RC_OPCODE_DP2: transform_DP2(c, inst); return 1; + case RC_OPCODE_DPH: transform_DPH(c, inst); return 1; + case RC_OPCODE_DST: transform_DST(c, inst); return 1; + case RC_OPCODE_FLR: transform_FLR(c, inst); return 1; + case RC_OPCODE_LIT: transform_LIT(c, inst); return 1; + case RC_OPCODE_LRP: transform_LRP(c, inst); return 1; + case RC_OPCODE_POW: transform_POW(c, inst); return 1; + case RC_OPCODE_RSQ: transform_RSQ(c, inst); return 1; + case RC_OPCODE_SEQ: transform_SEQ(c, inst); return 1; + case RC_OPCODE_SFL: transform_SFL(c, inst); return 1; + case RC_OPCODE_SGE: transform_SGE(c, inst); return 1; + case RC_OPCODE_SGT: transform_SGT(c, inst); return 1; + case RC_OPCODE_SLE: transform_SLE(c, inst); return 1; + case RC_OPCODE_SLT: transform_SLT(c, inst); return 1; + case RC_OPCODE_SNE: transform_SNE(c, inst); return 1; + case RC_OPCODE_SSG: transform_SSG(c, inst); return 1; + case RC_OPCODE_SUB: transform_SUB(c, inst); return 1; + case RC_OPCODE_SWZ: transform_SWZ(c, inst); return 1; + case RC_OPCODE_XPD: transform_XPD(c, inst); return 1; + default: + return 0; + } +} + + +static void transform_r300_vertex_ABS(struct radeon_compiler* c, + struct rc_instruction* inst) +{ + /* Note: r500 can take absolute values, but r300 cannot. */ + inst->U.I.Opcode = RC_OPCODE_MAX; + inst->U.I.SrcReg[1] = inst->U.I.SrcReg[0]; + inst->U.I.SrcReg[1].Negate ^= RC_MASK_XYZW; +} + +static void transform_r300_vertex_CMP(struct radeon_compiler* c, + struct rc_instruction* inst) +{ + /* There is no decent CMP available, so let's rig one up. + * CMP is defined as dst = src0 < 0.0 ? src1 : src2 + * The following sequence consumes zero to two temps and two extra slots + * (the second temp and the second slot is consumed by transform_LRP), + * but should be equivalent: + * + * SLT tmp0, src0, 0.0 + * LRP dst, tmp0, src1, src2 + * + * Yes, I know, I'm a mad scientist. ~ C. & M. */ + struct rc_dst_register dst = try_to_reuse_dst(c, inst); + + /* SLT tmp0, src0, 0.0 */ + emit2(c, inst->Prev, RC_OPCODE_SLT, 0, + dst, + inst->U.I.SrcReg[0], builtin_zero); + + /* LRP dst, tmp0, src1, src2 */ + transform_LRP(c, + emit3(c, inst->Prev, RC_OPCODE_LRP, 0, + inst->U.I.DstReg, + srcreg(RC_FILE_TEMPORARY, dst.Index), inst->U.I.SrcReg[1], inst->U.I.SrcReg[2])); + + rc_remove_instruction(inst); +} + +static void transform_r300_vertex_DP2(struct radeon_compiler* c, + struct rc_instruction* inst) +{ + struct rc_instruction *next_inst = inst->Next; + transform_DP2(c, inst); + next_inst->Prev->U.I.Opcode = RC_OPCODE_DP4; +} + +static void transform_r300_vertex_DP3(struct radeon_compiler* c, + struct rc_instruction* inst) +{ + struct rc_src_register src0 = inst->U.I.SrcReg[0]; + struct rc_src_register src1 = inst->U.I.SrcReg[1]; + src0.Negate &= ~RC_MASK_W; + src0.Swizzle &= ~(7 << (3 * 3)); + src0.Swizzle |= RC_SWIZZLE_ZERO << (3 * 3); + src1.Negate &= ~RC_MASK_W; + src1.Swizzle &= ~(7 << (3 * 3)); + src1.Swizzle |= RC_SWIZZLE_ZERO << (3 * 3); + emit2(c, inst->Prev, RC_OPCODE_DP4, inst->U.I.SaturateMode, inst->U.I.DstReg, src0, src1); + rc_remove_instruction(inst); +} + +static void transform_r300_vertex_fix_LIT(struct radeon_compiler* c, + struct rc_instruction* inst) +{ + struct rc_dst_register dst = try_to_reuse_dst(c, inst); + unsigned constant_swizzle; + int constant = rc_constants_add_immediate_scalar(&c->Program.Constants, + 0.0000000000000000001, + &constant_swizzle); + + /* MOV dst, src */ + dst.WriteMask = RC_MASK_XYZW; + emit1(c, inst->Prev, RC_OPCODE_MOV, 0, + dst, + inst->U.I.SrcReg[0]); + + /* MAX dst.y, src, 0.00...001 */ + emit2(c, inst->Prev, RC_OPCODE_MAX, 0, + dstregtmpmask(dst.Index, RC_MASK_Y), + srcreg(RC_FILE_TEMPORARY, dst.Index), + srcregswz(RC_FILE_CONSTANT, constant, constant_swizzle)); + + inst->U.I.SrcReg[0] = srcreg(RC_FILE_TEMPORARY, dst.Index); +} + +static void transform_r300_vertex_SEQ(struct radeon_compiler *c, + struct rc_instruction *inst) +{ + /* x = y <==> x >= y && y >= x */ + int tmp = rc_find_free_temporary(c); + + /* x <= y */ + emit2(c, inst->Prev, RC_OPCODE_SGE, 0, + dstregtmpmask(tmp, inst->U.I.DstReg.WriteMask), + inst->U.I.SrcReg[0], + inst->U.I.SrcReg[1]); + + /* y <= x */ + emit2(c, inst->Prev, RC_OPCODE_SGE, 0, + inst->U.I.DstReg, + inst->U.I.SrcReg[1], + inst->U.I.SrcReg[0]); + + /* x && y = x * y */ + emit2(c, inst->Prev, RC_OPCODE_MUL, 0, + inst->U.I.DstReg, + srcreg(RC_FILE_TEMPORARY, tmp), + srcreg(inst->U.I.DstReg.File, inst->U.I.DstReg.Index)); + + rc_remove_instruction(inst); +} + +static void transform_r300_vertex_SNE(struct radeon_compiler *c, + struct rc_instruction *inst) +{ + /* x != y <==> x < y || y < x */ + int tmp = rc_find_free_temporary(c); + + /* x < y */ + emit2(c, inst->Prev, RC_OPCODE_SLT, 0, + dstregtmpmask(tmp, inst->U.I.DstReg.WriteMask), + inst->U.I.SrcReg[0], + inst->U.I.SrcReg[1]); + + /* y < x */ + emit2(c, inst->Prev, RC_OPCODE_SLT, 0, + inst->U.I.DstReg, + inst->U.I.SrcReg[1], + inst->U.I.SrcReg[0]); + + /* x || y = max(x, y) */ + emit2(c, inst->Prev, RC_OPCODE_MAX, 0, + inst->U.I.DstReg, + srcreg(RC_FILE_TEMPORARY, tmp), + srcreg(inst->U.I.DstReg.File, inst->U.I.DstReg.Index)); + + rc_remove_instruction(inst); +} + +static void transform_r300_vertex_SGT(struct radeon_compiler* c, + struct rc_instruction* inst) +{ + /* x > y <==> -x < -y */ + inst->U.I.Opcode = RC_OPCODE_SLT; + inst->U.I.SrcReg[0].Negate ^= RC_MASK_XYZW; + inst->U.I.SrcReg[1].Negate ^= RC_MASK_XYZW; +} + +static void transform_r300_vertex_SLE(struct radeon_compiler* c, + struct rc_instruction* inst) +{ + /* x <= y <==> -x >= -y */ + inst->U.I.Opcode = RC_OPCODE_SGE; + inst->U.I.SrcReg[0].Negate ^= RC_MASK_XYZW; + inst->U.I.SrcReg[1].Negate ^= RC_MASK_XYZW; +} + +static void transform_r300_vertex_SSG(struct radeon_compiler* c, + struct rc_instruction* inst) +{ + /* result = sign(x) + * + * SLT tmp0, 0, x; + * SLT tmp1, x, 0; + * ADD result, tmp0, -tmp1; + */ + struct rc_dst_register dst0 = try_to_reuse_dst(c, inst); + unsigned tmp1; + + /* 0 < x */ + dst0 = try_to_reuse_dst(c, inst); + emit2(c, inst->Prev, RC_OPCODE_SLT, 0, + dst0, + builtin_zero, + inst->U.I.SrcReg[0]); + + /* x < 0 */ + tmp1 = rc_find_free_temporary(c); + emit2(c, inst->Prev, RC_OPCODE_SLT, 0, + dstregtmpmask(tmp1, inst->U.I.DstReg.WriteMask), + inst->U.I.SrcReg[0], + builtin_zero); + + /* Either both are zero, or one of them is one and the other is zero. */ + /* result = tmp0 - tmp1 */ + emit2(c, inst->Prev, RC_OPCODE_ADD, 0, + inst->U.I.DstReg, + srcreg(RC_FILE_TEMPORARY, dst0.Index), + negate(srcreg(RC_FILE_TEMPORARY, tmp1))); + + rc_remove_instruction(inst); +} + +/** + * For use with rc_local_transform, this transforms non-native ALU + * instructions of the r300 up to r500 vertex engine. + */ +int r300_transform_vertex_alu( + struct radeon_compiler * c, + struct rc_instruction* inst, + void* unused) +{ + switch(inst->U.I.Opcode) { + case RC_OPCODE_ABS: transform_r300_vertex_ABS(c, inst); return 1; + case RC_OPCODE_CEIL: transform_CEIL(c, inst); return 1; + case RC_OPCODE_CLAMP: transform_CLAMP(c, inst); return 1; + case RC_OPCODE_CMP: transform_r300_vertex_CMP(c, inst); return 1; + case RC_OPCODE_DP2: transform_r300_vertex_DP2(c, inst); return 1; + case RC_OPCODE_DP3: transform_r300_vertex_DP3(c, inst); return 1; + case RC_OPCODE_DPH: transform_DPH(c, inst); return 1; + case RC_OPCODE_FLR: transform_FLR(c, inst); return 1; + case RC_OPCODE_LIT: transform_r300_vertex_fix_LIT(c, inst); return 1; + case RC_OPCODE_LRP: transform_LRP(c, inst); return 1; + case RC_OPCODE_SEQ: + if (!c->is_r500) { + transform_r300_vertex_SEQ(c, inst); + return 1; + } + return 0; + case RC_OPCODE_SFL: transform_SFL(c, inst); return 1; + case RC_OPCODE_SGT: transform_r300_vertex_SGT(c, inst); return 1; + case RC_OPCODE_SLE: transform_r300_vertex_SLE(c, inst); return 1; + case RC_OPCODE_SNE: + if (!c->is_r500) { + transform_r300_vertex_SNE(c, inst); + return 1; + } + return 0; + case RC_OPCODE_SSG: transform_r300_vertex_SSG(c, inst); return 1; + case RC_OPCODE_SUB: transform_SUB(c, inst); return 1; + case RC_OPCODE_SWZ: transform_SWZ(c, inst); return 1; + case RC_OPCODE_XPD: transform_XPD(c, inst); return 1; + default: + return 0; + } +} + +static void sincos_constants(struct radeon_compiler* c, unsigned int *constants) +{ + static const float SinCosConsts[2][4] = { + { + 1.273239545, /* 4/PI */ + -0.405284735, /* -4/(PI*PI) */ + 3.141592654, /* PI */ + 0.2225 /* weight */ + }, + { + 0.75, + 0.5, + 0.159154943, /* 1/(2*PI) */ + 6.283185307 /* 2*PI */ + } + }; + int i; + + for(i = 0; i < 2; ++i) + constants[i] = rc_constants_add_immediate_vec4(&c->Program.Constants, SinCosConsts[i]); +} + +/** + * Approximate sin(x), where x is clamped to (-pi/2, pi/2). + * + * MUL tmp.xy, src, { 4/PI, -4/(PI^2) } + * MAD tmp.x, tmp.y, |src|, tmp.x + * MAD tmp.y, tmp.x, |tmp.x|, -tmp.x + * MAD dest, tmp.y, weight, tmp.x + */ +static void sin_approx( + struct radeon_compiler* c, struct rc_instruction * inst, + struct rc_dst_register dst, struct rc_src_register src, const unsigned int* constants) +{ + unsigned int tempreg = rc_find_free_temporary(c); + + emit2(c, inst->Prev, RC_OPCODE_MUL, 0, dstregtmpmask(tempreg, RC_MASK_XY), + swizzle_xxxx(src), + srcreg(RC_FILE_CONSTANT, constants[0])); + emit3(c, inst->Prev, RC_OPCODE_MAD, 0, dstregtmpmask(tempreg, RC_MASK_X), + swizzle_yyyy(srcreg(RC_FILE_TEMPORARY, tempreg)), + absolute(swizzle_xxxx(src)), + swizzle_xxxx(srcreg(RC_FILE_TEMPORARY, tempreg))); + emit3(c, inst->Prev, RC_OPCODE_MAD, 0, dstregtmpmask(tempreg, RC_MASK_Y), + swizzle_xxxx(srcreg(RC_FILE_TEMPORARY, tempreg)), + absolute(swizzle_xxxx(srcreg(RC_FILE_TEMPORARY, tempreg))), + negate(swizzle_xxxx(srcreg(RC_FILE_TEMPORARY, tempreg)))); + emit3(c, inst->Prev, RC_OPCODE_MAD, 0, dst, + swizzle_yyyy(srcreg(RC_FILE_TEMPORARY, tempreg)), + swizzle_wwww(srcreg(RC_FILE_CONSTANT, constants[0])), + swizzle_xxxx(srcreg(RC_FILE_TEMPORARY, tempreg))); +} + +/** + * Translate the trigonometric functions COS, SIN, and SCS + * using only the basic instructions + * MOV, ADD, MUL, MAD, FRC + */ +int r300_transform_trig_simple(struct radeon_compiler* c, + struct rc_instruction* inst, + void* unused) +{ + unsigned int constants[2]; + unsigned int tempreg; + + if (inst->U.I.Opcode != RC_OPCODE_COS && + inst->U.I.Opcode != RC_OPCODE_SIN && + inst->U.I.Opcode != RC_OPCODE_SCS) + return 0; + + tempreg = rc_find_free_temporary(c); + + sincos_constants(c, constants); + + if (inst->U.I.Opcode == RC_OPCODE_COS) { + /* MAD tmp.x, src, 1/(2*PI), 0.75 */ + /* FRC tmp.x, tmp.x */ + /* MAD tmp.z, tmp.x, 2*PI, -PI */ + emit3(c, inst->Prev, RC_OPCODE_MAD, 0, dstregtmpmask(tempreg, RC_MASK_W), + swizzle_xxxx(inst->U.I.SrcReg[0]), + swizzle_zzzz(srcreg(RC_FILE_CONSTANT, constants[1])), + swizzle_xxxx(srcreg(RC_FILE_CONSTANT, constants[1]))); + emit1(c, inst->Prev, RC_OPCODE_FRC, 0, dstregtmpmask(tempreg, RC_MASK_W), + swizzle_wwww(srcreg(RC_FILE_TEMPORARY, tempreg))); + emit3(c, inst->Prev, RC_OPCODE_MAD, 0, dstregtmpmask(tempreg, RC_MASK_W), + swizzle_wwww(srcreg(RC_FILE_TEMPORARY, tempreg)), + swizzle_wwww(srcreg(RC_FILE_CONSTANT, constants[1])), + negate(swizzle_zzzz(srcreg(RC_FILE_CONSTANT, constants[0])))); + + sin_approx(c, inst, inst->U.I.DstReg, + swizzle_wwww(srcreg(RC_FILE_TEMPORARY, tempreg)), + constants); + } else if (inst->U.I.Opcode == RC_OPCODE_SIN) { + emit3(c, inst->Prev, RC_OPCODE_MAD, 0, dstregtmpmask(tempreg, RC_MASK_W), + swizzle_xxxx(inst->U.I.SrcReg[0]), + swizzle_zzzz(srcreg(RC_FILE_CONSTANT, constants[1])), + swizzle_yyyy(srcreg(RC_FILE_CONSTANT, constants[1]))); + emit1(c, inst->Prev, RC_OPCODE_FRC, 0, dstregtmpmask(tempreg, RC_MASK_W), + swizzle_wwww(srcreg(RC_FILE_TEMPORARY, tempreg))); + emit3(c, inst->Prev, RC_OPCODE_MAD, 0, dstregtmpmask(tempreg, RC_MASK_W), + swizzle_wwww(srcreg(RC_FILE_TEMPORARY, tempreg)), + swizzle_wwww(srcreg(RC_FILE_CONSTANT, constants[1])), + negate(swizzle_zzzz(srcreg(RC_FILE_CONSTANT, constants[0])))); + + sin_approx(c, inst, inst->U.I.DstReg, + swizzle_wwww(srcreg(RC_FILE_TEMPORARY, tempreg)), + constants); + } else { + struct rc_dst_register dst; + + emit3(c, inst->Prev, RC_OPCODE_MAD, 0, dstregtmpmask(tempreg, RC_MASK_XY), + swizzle_xxxx(inst->U.I.SrcReg[0]), + swizzle_zzzz(srcreg(RC_FILE_CONSTANT, constants[1])), + swizzle(srcreg(RC_FILE_CONSTANT, constants[1]), RC_SWIZZLE_X, RC_SWIZZLE_Y, RC_SWIZZLE_Z, RC_SWIZZLE_W)); + emit1(c, inst->Prev, RC_OPCODE_FRC, 0, dstregtmpmask(tempreg, RC_MASK_XY), + srcreg(RC_FILE_TEMPORARY, tempreg)); + emit3(c, inst->Prev, RC_OPCODE_MAD, 0, dstregtmpmask(tempreg, RC_MASK_XY), + srcreg(RC_FILE_TEMPORARY, tempreg), + swizzle_wwww(srcreg(RC_FILE_CONSTANT, constants[1])), + negate(swizzle_zzzz(srcreg(RC_FILE_CONSTANT, constants[0])))); + + dst = inst->U.I.DstReg; + + dst.WriteMask = inst->U.I.DstReg.WriteMask & RC_MASK_X; + sin_approx(c, inst, dst, + swizzle_xxxx(srcreg(RC_FILE_TEMPORARY, tempreg)), + constants); + + dst.WriteMask = inst->U.I.DstReg.WriteMask & RC_MASK_Y; + sin_approx(c, inst, dst, + swizzle_yyyy(srcreg(RC_FILE_TEMPORARY, tempreg)), + constants); + } + + rc_remove_instruction(inst); + + return 1; +} + +static void r300_transform_SIN_COS_SCS(struct radeon_compiler *c, + struct rc_instruction *inst, + unsigned srctmp) +{ + if (inst->U.I.Opcode == RC_OPCODE_COS) { + emit1(c, inst->Prev, RC_OPCODE_COS, inst->U.I.SaturateMode, inst->U.I.DstReg, + srcregswz(RC_FILE_TEMPORARY, srctmp, RC_SWIZZLE_WWWW)); + } else if (inst->U.I.Opcode == RC_OPCODE_SIN) { + emit1(c, inst->Prev, RC_OPCODE_SIN, inst->U.I.SaturateMode, + inst->U.I.DstReg, srcregswz(RC_FILE_TEMPORARY, srctmp, RC_SWIZZLE_WWWW)); + } else if (inst->U.I.Opcode == RC_OPCODE_SCS) { + struct rc_dst_register moddst = inst->U.I.DstReg; + + if (inst->U.I.DstReg.WriteMask & RC_MASK_X) { + moddst.WriteMask = RC_MASK_X; + emit1(c, inst->Prev, RC_OPCODE_COS, inst->U.I.SaturateMode, moddst, + srcregswz(RC_FILE_TEMPORARY, srctmp, RC_SWIZZLE_WWWW)); + } + if (inst->U.I.DstReg.WriteMask & RC_MASK_Y) { + moddst.WriteMask = RC_MASK_Y; + emit1(c, inst->Prev, RC_OPCODE_SIN, inst->U.I.SaturateMode, moddst, + srcregswz(RC_FILE_TEMPORARY, srctmp, RC_SWIZZLE_WWWW)); + } + } + + rc_remove_instruction(inst); +} + + +/** + * Transform the trigonometric functions COS, SIN, and SCS + * to include pre-scaling by 1/(2*PI) and taking the fractional + * part, so that the input to COS and SIN is always in the range [0,1). + * SCS is replaced by one COS and one SIN instruction. + * + * @warning This transformation implicitly changes the semantics of SIN and COS! + */ +int radeonTransformTrigScale(struct radeon_compiler* c, + struct rc_instruction* inst, + void* unused) +{ + static const float RCP_2PI = 0.15915494309189535; + unsigned int temp; + unsigned int constant; + unsigned int constant_swizzle; + + if (inst->U.I.Opcode != RC_OPCODE_COS && + inst->U.I.Opcode != RC_OPCODE_SIN && + inst->U.I.Opcode != RC_OPCODE_SCS) + return 0; + + temp = rc_find_free_temporary(c); + constant = rc_constants_add_immediate_scalar(&c->Program.Constants, RCP_2PI, &constant_swizzle); + + emit2(c, inst->Prev, RC_OPCODE_MUL, 0, dstregtmpmask(temp, RC_MASK_W), + swizzle_xxxx(inst->U.I.SrcReg[0]), + srcregswz(RC_FILE_CONSTANT, constant, constant_swizzle)); + emit1(c, inst->Prev, RC_OPCODE_FRC, 0, dstregtmpmask(temp, RC_MASK_W), + srcreg(RC_FILE_TEMPORARY, temp)); + + r300_transform_SIN_COS_SCS(c, inst, temp); + return 1; +} + +/** + * Transform the trigonometric functions COS, SIN, and SCS + * so that the input to COS and SIN is always in the range [-PI, PI]. + * SCS is replaced by one COS and one SIN instruction. + */ +int r300_transform_trig_scale_vertex(struct radeon_compiler *c, + struct rc_instruction *inst, + void *unused) +{ + static const float cons[4] = {0.15915494309189535, 0.5, 6.28318530717959, -3.14159265358979}; + unsigned int temp; + unsigned int constant; + + if (inst->U.I.Opcode != RC_OPCODE_COS && + inst->U.I.Opcode != RC_OPCODE_SIN && + inst->U.I.Opcode != RC_OPCODE_SCS) + return 0; + + /* Repeat x in the range [-PI, PI]: + * + * repeat(x) = frac(x / 2PI + 0.5) * 2PI - PI + */ + + temp = rc_find_free_temporary(c); + constant = rc_constants_add_immediate_vec4(&c->Program.Constants, cons); + + emit3(c, inst->Prev, RC_OPCODE_MAD, 0, dstregtmpmask(temp, RC_MASK_W), + swizzle_xxxx(inst->U.I.SrcReg[0]), + srcregswz(RC_FILE_CONSTANT, constant, RC_SWIZZLE_XXXX), + srcregswz(RC_FILE_CONSTANT, constant, RC_SWIZZLE_YYYY)); + emit1(c, inst->Prev, RC_OPCODE_FRC, 0, dstregtmpmask(temp, RC_MASK_W), + srcreg(RC_FILE_TEMPORARY, temp)); + emit3(c, inst->Prev, RC_OPCODE_MAD, 0, dstregtmpmask(temp, RC_MASK_W), + srcreg(RC_FILE_TEMPORARY, temp), + srcregswz(RC_FILE_CONSTANT, constant, RC_SWIZZLE_ZZZZ), + srcregswz(RC_FILE_CONSTANT, constant, RC_SWIZZLE_WWWW)); + + r300_transform_SIN_COS_SCS(c, inst, temp); + return 1; +} + +/** + * Rewrite DDX/DDY instructions to properly work with r5xx shaders. + * The r5xx MDH/MDV instruction provides per-quad partial derivatives. + * It takes the form A*B+C. A and C are set by setting src0. B should be -1. + * + * @warning This explicitly changes the form of DDX and DDY! + */ + +int radeonTransformDeriv(struct radeon_compiler* c, + struct rc_instruction* inst, + void* unused) +{ + if (inst->U.I.Opcode != RC_OPCODE_DDX && inst->U.I.Opcode != RC_OPCODE_DDY) + return 0; + + inst->U.I.SrcReg[1].Swizzle = RC_SWIZZLE_1111; + inst->U.I.SrcReg[1].Negate = RC_MASK_XYZW; + + return 1; +} + +/** + * IF Temp[0].x -\ + * KILP - > KIL -abs(Temp[0].x) + * ENDIF -/ + * + * This needs to be done in its own pass, because it modifies the instructions + * before and after KILP. + */ +void rc_transform_KILP(struct radeon_compiler * c, void *user) +{ + struct rc_instruction * inst; + for (inst = c->Program.Instructions.Next; + inst != &c->Program.Instructions; inst = inst->Next) { + + if (inst->U.I.Opcode != RC_OPCODE_KILP) + continue; + + inst->U.I.Opcode = RC_OPCODE_KIL; + + if (inst->Prev->U.I.Opcode != RC_OPCODE_IF + || inst->Next->U.I.Opcode != RC_OPCODE_ENDIF) { + inst->U.I.SrcReg[0] = negate(builtin_one); + } else { + + inst->U.I.SrcReg[0] = + negate(absolute(inst->Prev->U.I.SrcReg[0])); + /* Remove IF */ + rc_remove_instruction(inst->Prev); + /* Remove ENDIF */ + rc_remove_instruction(inst->Next); + } + } +} diff --git a/src/gallium/drivers/r300/compiler/radeon_program_alu.h b/src/gallium/drivers/r300/compiler/radeon_program_alu.h new file mode 100644 index 00000000000..b5f361e624f --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_program_alu.h @@ -0,0 +1,66 @@ +/* + * Copyright (C) 2008 Nicolai Haehnle. + * + * 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. + * + */ + +#ifndef __RADEON_PROGRAM_ALU_H_ +#define __RADEON_PROGRAM_ALU_H_ + +#include "radeon_program.h" + +int radeonTransformALU( + struct radeon_compiler * c, + struct rc_instruction * inst, + void*); + +int r300_transform_vertex_alu( + struct radeon_compiler * c, + struct rc_instruction * inst, + void*); + +int r300_transform_trig_simple( + struct radeon_compiler * c, + struct rc_instruction * inst, + void*); + +int radeonTransformTrigScale( + struct radeon_compiler * c, + struct rc_instruction * inst, + void*); + +int r300_transform_trig_scale_vertex( + struct radeon_compiler *c, + struct rc_instruction *inst, + void*); + +int radeonTransformDeriv( + struct radeon_compiler * c, + struct rc_instruction * inst, + void*); + +void rc_transform_KILP(struct radeon_compiler * c, + void *user); + +#endif /* __RADEON_PROGRAM_ALU_H_ */ diff --git a/src/gallium/drivers/r300/compiler/radeon_program_constants.h b/src/gallium/drivers/r300/compiler/radeon_program_constants.h new file mode 100644 index 00000000000..24577333450 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_program_constants.h @@ -0,0 +1,190 @@ +/* + * Copyright (C) 2009 Nicolai Haehnle. + * + * 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. + * + */ + +#ifndef RADEON_PROGRAM_CONSTANTS_H +#define RADEON_PROGRAM_CONSTANTS_H + +typedef enum { + RC_SATURATE_NONE = 0, + RC_SATURATE_ZERO_ONE, + RC_SATURATE_MINUS_PLUS_ONE +} rc_saturate_mode; + +typedef enum { + RC_TEXTURE_2D_ARRAY, + RC_TEXTURE_1D_ARRAY, + RC_TEXTURE_CUBE, + RC_TEXTURE_3D, + RC_TEXTURE_RECT, + RC_TEXTURE_2D, + RC_TEXTURE_1D +} rc_texture_target; + +typedef enum { + /** + * Used to indicate unused register descriptions and + * source register that use a constant swizzle. + */ + RC_FILE_NONE = 0, + RC_FILE_TEMPORARY, + + /** + * Input register. + * + * \note The compiler attaches no implicit semantics to input registers. + * Fragment/vertex program specific semantics must be defined explicitly + * using the appropriate compiler interfaces. + */ + RC_FILE_INPUT, + + /** + * Output register. + * + * \note The compiler attaches no implicit semantics to input registers. + * Fragment/vertex program specific semantics must be defined explicitly + * using the appropriate compiler interfaces. + */ + RC_FILE_OUTPUT, + RC_FILE_ADDRESS, + + /** + * Indicates a constant from the \ref rc_constant_list . + */ + RC_FILE_CONSTANT, + + /** + * Indicates a special register, see RC_SPECIAL_xxx. + */ + RC_FILE_SPECIAL, + + /** + * Indicates this register should use the result of the presubtract + * operation. + */ + RC_FILE_PRESUB +} rc_register_file; + +enum { + /** R500 fragment program ALU result "register" */ + RC_SPECIAL_ALU_RESULT = 0, + + /** Must be last */ + RC_NUM_SPECIAL_REGISTERS +}; + +#define RC_REGISTER_INDEX_BITS 10 +#define RC_REGISTER_MAX_INDEX (1 << RC_REGISTER_INDEX_BITS) + +typedef enum { + RC_SWIZZLE_X = 0, + RC_SWIZZLE_Y, + RC_SWIZZLE_Z, + RC_SWIZZLE_W, + RC_SWIZZLE_ZERO, + RC_SWIZZLE_ONE, + RC_SWIZZLE_HALF, + RC_SWIZZLE_UNUSED +} rc_swizzle; + +#define RC_MAKE_SWIZZLE(a,b,c,d) (((a)<<0) | ((b)<<3) | ((c)<<6) | ((d)<<9)) +#define RC_MAKE_SWIZZLE_SMEAR(a) RC_MAKE_SWIZZLE((a),(a),(a),(a)) +#define GET_SWZ(swz, idx) (((swz) >> ((idx)*3)) & 0x7) +#define GET_BIT(msk, idx) (((msk) >> (idx)) & 0x1) +#define SET_SWZ(swz, idx, newv) \ + do { \ + (swz) = ((swz) & ~(7 << ((idx)*3))) | ((newv) << ((idx)*3)); \ + } while(0) + +#define RC_SWIZZLE_XYZW RC_MAKE_SWIZZLE(RC_SWIZZLE_X, RC_SWIZZLE_Y, RC_SWIZZLE_Z, RC_SWIZZLE_W) +#define RC_SWIZZLE_XYZ0 RC_MAKE_SWIZZLE(RC_SWIZZLE_X, RC_SWIZZLE_Y, RC_SWIZZLE_Z, RC_SWIZZLE_ZERO) +#define RC_SWIZZLE_XYZZ RC_MAKE_SWIZZLE(RC_SWIZZLE_X, RC_SWIZZLE_Y, RC_SWIZZLE_Z, RC_SWIZZLE_Z) +#define RC_SWIZZLE_XXXX RC_MAKE_SWIZZLE_SMEAR(RC_SWIZZLE_X) +#define RC_SWIZZLE_YYYY RC_MAKE_SWIZZLE_SMEAR(RC_SWIZZLE_Y) +#define RC_SWIZZLE_ZZZZ RC_MAKE_SWIZZLE_SMEAR(RC_SWIZZLE_Z) +#define RC_SWIZZLE_WWWW RC_MAKE_SWIZZLE_SMEAR(RC_SWIZZLE_W) +#define RC_SWIZZLE_0000 RC_MAKE_SWIZZLE_SMEAR(RC_SWIZZLE_ZERO) +#define RC_SWIZZLE_1111 RC_MAKE_SWIZZLE_SMEAR(RC_SWIZZLE_ONE) +#define RC_SWIZZLE_HHHH RC_MAKE_SWIZZLE_SMEAR(RC_SWIZZLE_HALF) +#define RC_SWIZZLE_UUUU RC_MAKE_SWIZZLE_SMEAR(RC_SWIZZLE_UNUSED) + +/** + * \name Bitmasks for components of vectors. + * + * Used for write masks, negation masks, etc. + */ +/*@{*/ +#define RC_MASK_NONE 0 +#define RC_MASK_X 1 +#define RC_MASK_Y 2 +#define RC_MASK_Z 4 +#define RC_MASK_W 8 +#define RC_MASK_XY (RC_MASK_X|RC_MASK_Y) +#define RC_MASK_XYZ (RC_MASK_X|RC_MASK_Y|RC_MASK_Z) +#define RC_MASK_XYW (RC_MASK_X|RC_MASK_Y|RC_MASK_W) +#define RC_MASK_XYZW (RC_MASK_X|RC_MASK_Y|RC_MASK_Z|RC_MASK_W) +/*@}*/ + +typedef enum { + RC_ALURESULT_NONE = 0, + RC_ALURESULT_X, + RC_ALURESULT_W +} rc_write_aluresult; + +typedef enum { + RC_PRESUB_NONE = 0, + + /** 1 - 2 * src0 */ + RC_PRESUB_BIAS, + + /** src1 - src0 */ + RC_PRESUB_SUB, + + /** src1 + src0 */ + RC_PRESUB_ADD, + + /** 1 - src0 */ + RC_PRESUB_INV +} rc_presubtract_op; + +static inline int rc_presubtract_src_reg_count(rc_presubtract_op op){ + switch(op){ + case RC_PRESUB_BIAS: + case RC_PRESUB_INV: + return 1; + case RC_PRESUB_ADD: + case RC_PRESUB_SUB: + return 2; + default: + return 0; + } +} + +#define RC_SOURCE_NONE 0x0 +#define RC_SOURCE_RGB 0x1 +#define RC_SOURCE_ALPHA 0x2 + +#endif /* RADEON_PROGRAM_CONSTANTS_H */ diff --git a/src/gallium/drivers/r300/compiler/radeon_program_pair.c b/src/gallium/drivers/r300/compiler/radeon_program_pair.c new file mode 100644 index 00000000000..52315957520 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_program_pair.c @@ -0,0 +1,239 @@ +/* + * Copyright (C) 2008-2009 Nicolai Haehnle. + * + * 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. + * + */ + +#include "radeon_program_pair.h" + +#include "radeon_compiler_util.h" + +#include <stdlib.h> + +/** + * Return the source slot where we installed the given register access, + * or -1 if no slot was free anymore. + */ +int rc_pair_alloc_source(struct rc_pair_instruction *pair, + unsigned int rgb, unsigned int alpha, + rc_register_file file, unsigned int index) +{ + int candidate = -1; + int candidate_quality = -1; + unsigned int alpha_used = 0; + unsigned int rgb_used = 0; + int i; + + if ((!rgb && !alpha) || file == RC_FILE_NONE) + return 0; + + /* Make sure only one presubtract operation is used per instruction. */ + if (file == RC_FILE_PRESUB) { + if (rgb && pair->RGB.Src[RC_PAIR_PRESUB_SRC].Used + && index != pair->RGB.Src[RC_PAIR_PRESUB_SRC].Index) { + return -1; + } + + if (alpha && pair->Alpha.Src[RC_PAIR_PRESUB_SRC].Used + && index != pair->Alpha.Src[RC_PAIR_PRESUB_SRC].Index) { + return -1; + } + } + + for(i = 0; i < 3; ++i) { + int q = 0; + if (rgb) { + if (pair->RGB.Src[i].Used) { + if (pair->RGB.Src[i].File != file || + pair->RGB.Src[i].Index != index) { + rgb_used++; + continue; + } + q++; + } + } + if (alpha) { + if (pair->Alpha.Src[i].Used) { + if (pair->Alpha.Src[i].File != file || + pair->Alpha.Src[i].Index != index) { + alpha_used++; + continue; + } + q++; + } + } + if (q > candidate_quality) { + candidate_quality = q; + candidate = i; + } + } + + if (file == RC_FILE_PRESUB) { + candidate = RC_PAIR_PRESUB_SRC; + } else if (candidate < 0 || (rgb && rgb_used > 2) + || (alpha && alpha_used > 2)) { + return -1; + } + + /* candidate >= 0 */ + + if (rgb) { + pair->RGB.Src[candidate].Used = 1; + pair->RGB.Src[candidate].File = file; + pair->RGB.Src[candidate].Index = index; + if (candidate == RC_PAIR_PRESUB_SRC) { + /* For registers with the RC_FILE_PRESUB file, + * the index stores the presubtract op. */ + int src_regs = rc_presubtract_src_reg_count(index); + for(i = 0; i < src_regs; i++) { + pair->RGB.Src[i].Used = 1; + } + } + } + if (alpha) { + pair->Alpha.Src[candidate].Used = 1; + pair->Alpha.Src[candidate].File = file; + pair->Alpha.Src[candidate].Index = index; + if (candidate == RC_PAIR_PRESUB_SRC) { + /* For registers with the RC_FILE_PRESUB file, + * the index stores the presubtract op. */ + int src_regs = rc_presubtract_src_reg_count(index); + for(i=0; i < src_regs; i++) { + pair->Alpha.Src[i].Used = 1; + } + } + } + + return candidate; +} + +static void pair_foreach_source_callback( + struct rc_pair_instruction * pair, + void * data, + rc_pair_foreach_src_fn cb, + unsigned int swz, + unsigned int src) +{ + /* swz > 3 means that the swizzle is either not used, or a constant + * swizzle (e.g. 0, 1, 0.5). */ + if(swz > 3) + return; + + if(swz == RC_SWIZZLE_W) { + if (src == RC_PAIR_PRESUB_SRC) { + unsigned int i; + unsigned int src_count = rc_presubtract_src_reg_count( + pair->Alpha.Src[RC_PAIR_PRESUB_SRC].Index); + for(i = 0; i < src_count; i++) { + cb(data, &pair->Alpha.Src[i]); + } + } else { + cb(data, &pair->Alpha.Src[src]); + } + } else { + if (src == RC_PAIR_PRESUB_SRC) { + unsigned int i; + unsigned int src_count = rc_presubtract_src_reg_count( + pair->RGB.Src[RC_PAIR_PRESUB_SRC].Index); + for(i = 0; i < src_count; i++) { + cb(data, &pair->RGB.Src[i]); + } + } + else { + cb(data, &pair->RGB.Src[src]); + } + } +} + +void rc_pair_foreach_source_that_alpha_reads( + struct rc_pair_instruction * pair, + void * data, + rc_pair_foreach_src_fn cb) +{ + unsigned int i; + const struct rc_opcode_info * info = + rc_get_opcode_info(pair->Alpha.Opcode); + for(i = 0; i < info->NumSrcRegs; i++) { + pair_foreach_source_callback(pair, data, cb, + GET_SWZ(pair->Alpha.Arg[i].Swizzle, 0), + pair->Alpha.Arg[i].Source); + } +} + +void rc_pair_foreach_source_that_rgb_reads( + struct rc_pair_instruction * pair, + void * data, + rc_pair_foreach_src_fn cb) +{ + unsigned int i; + const struct rc_opcode_info * info = + rc_get_opcode_info(pair->RGB.Opcode); + for(i = 0; i < info->NumSrcRegs; i++) { + unsigned int chan; + unsigned int swz = RC_SWIZZLE_UNUSED; + /* Find a swizzle that is either X,Y,Z,or W. We assume here + * that if one channel swizzles X,Y, or Z, then none of the + * other channels swizzle W, and vice-versa. */ + for(chan = 0; chan < 4; chan++) { + swz = GET_SWZ(pair->RGB.Arg[i].Swizzle, chan); + if(swz == RC_SWIZZLE_X || swz == RC_SWIZZLE_Y + || swz == RC_SWIZZLE_Z || swz == RC_SWIZZLE_W) + continue; + } + pair_foreach_source_callback(pair, data, cb, + swz, + pair->RGB.Arg[i].Source); + } +} + +struct rc_pair_instruction_source * rc_pair_get_src( + struct rc_pair_instruction * pair_inst, + struct rc_pair_instruction_arg * arg) +{ + unsigned int type; + + type = rc_source_type_swz(arg->Swizzle); + + if (type & RC_SOURCE_RGB) { + return &pair_inst->RGB.Src[arg->Source]; + } else if (type & RC_SOURCE_ALPHA) { + return &pair_inst->Alpha.Src[arg->Source]; + } else { + return NULL; + } +} + +int rc_pair_get_src_index( + struct rc_pair_instruction * pair_inst, + struct rc_pair_instruction_source * src) +{ + int i; + for (i = 0; i < 3; i++) { + if (&pair_inst->RGB.Src[i] == src + || &pair_inst->Alpha.Src[i] == src) { + return i; + } + } + return -1; +} diff --git a/src/gallium/drivers/r300/compiler/radeon_program_pair.h b/src/gallium/drivers/r300/compiler/radeon_program_pair.h new file mode 100644 index 00000000000..a957ea9f7a0 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_program_pair.h @@ -0,0 +1,137 @@ +/* + * Copyright (C) 2008 Nicolai Haehnle. + * + * 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. + * + */ + +#ifndef __RADEON_PROGRAM_PAIR_H_ +#define __RADEON_PROGRAM_PAIR_H_ + +#include "radeon_code.h" +#include "radeon_opcodes.h" +#include "radeon_program_constants.h" + +struct radeon_compiler; + + +/** + * \file + * Represents a paired ALU instruction, as found in R300 and R500 + * fragment programs. + * + * Note that this representation is taking some liberties as far + * as register files are concerned, to allow separate register + * allocation. + * + * Also note that there are some subtleties in that the semantics + * of certain opcodes are implicitly changed in this representation; + * see \ref rc_pair_translate + */ + +/* For rgb and alpha instructions when arg[n].Source = RC_PAIR_PRESUB_SRC, then + * the presubtract value will be used, and + * {RGB,Alpha}.Src[RC_PAIR_PRESUB_SRC].File will be set to RC_FILE_PRESUB. + */ +#define RC_PAIR_PRESUB_SRC 3 + +struct rc_pair_instruction_source { + unsigned int Used:1; + unsigned int File:3; + unsigned int Index:RC_REGISTER_INDEX_BITS; +}; + +struct rc_pair_instruction_arg { + unsigned int Source:2; + unsigned int Swizzle:12; + unsigned int Abs:1; + unsigned int Negate:1; +}; + +struct rc_pair_sub_instruction { + unsigned int Opcode:8; + unsigned int DestIndex:RC_REGISTER_INDEX_BITS; + unsigned int WriteMask:4; + unsigned int Target:2; + unsigned int OutputWriteMask:3; + unsigned int DepthWriteMask:1; + unsigned int Saturate:1; + + struct rc_pair_instruction_source Src[4]; + struct rc_pair_instruction_arg Arg[3]; +}; + +struct rc_pair_instruction { + struct rc_pair_sub_instruction RGB; + struct rc_pair_sub_instruction Alpha; + + unsigned int WriteALUResult:2; + unsigned int ALUResultCompare:3; + unsigned int Nop:1; +}; + +typedef void (*rc_pair_foreach_src_fn) + (void *, struct rc_pair_instruction_source *); + +/** + * General helper functions for dealing with the paired instruction format. + */ +/*@{*/ +int rc_pair_alloc_source(struct rc_pair_instruction *pair, + unsigned int rgb, unsigned int alpha, + rc_register_file file, unsigned int index); + +void rc_pair_foreach_source_that_alpha_reads( + struct rc_pair_instruction * pair, + void * data, + rc_pair_foreach_src_fn cb); + +void rc_pair_foreach_source_that_rgb_reads( + struct rc_pair_instruction * pair, + void * data, + rc_pair_foreach_src_fn cb); + +struct rc_pair_instruction_source * rc_pair_get_src( + struct rc_pair_instruction * pair_inst, + struct rc_pair_instruction_arg * arg); + +int rc_pair_get_src_index( + struct rc_pair_instruction * pair_inst, + struct rc_pair_instruction_source * src); +/*@}*/ + + +/** + * Compiler passes that operate with the paired format. + */ +/*@{*/ +struct radeon_pair_handler; + +void rc_pair_translate(struct radeon_compiler *cc, void *user); +void rc_pair_schedule(struct radeon_compiler *cc, void *user); +void rc_pair_regalloc(struct radeon_compiler *cc, void *user); +void rc_pair_regalloc_inputs_only(struct radeon_compiler *cc, void *user); +void rc_pair_remove_dead_sources(struct radeon_compiler *c, void *user); +/*@}*/ + +#endif /* __RADEON_PROGRAM_PAIR_H_ */ diff --git a/src/gallium/drivers/r300/compiler/radeon_program_print.c b/src/gallium/drivers/r300/compiler/radeon_program_print.c new file mode 100644 index 00000000000..390d1319460 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_program_print.c @@ -0,0 +1,418 @@ +/* + * Copyright 2009 Nicolai Hähnle <[email protected]> + * + * 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 + * on the rights to use, copy, modify, merge, publish, distribute, sub + * license, 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 NON-INFRINGEMENT. IN NO EVENT SHALL + * THE AUTHOR(S) AND/OR THEIR 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. */ + +#include "radeon_program.h" + +#include <stdio.h> + +static const char * textarget_to_string(rc_texture_target target) +{ + switch(target) { + case RC_TEXTURE_2D_ARRAY: return "2D_ARRAY"; + case RC_TEXTURE_1D_ARRAY: return "1D_ARRAY"; + case RC_TEXTURE_CUBE: return "CUBE"; + case RC_TEXTURE_3D: return "3D"; + case RC_TEXTURE_RECT: return "RECT"; + case RC_TEXTURE_2D: return "2D"; + case RC_TEXTURE_1D: return "1D"; + default: return "BAD_TEXTURE_TARGET"; + } +} + +static const char * presubtract_op_to_string(rc_presubtract_op op) +{ + switch(op) { + case RC_PRESUB_NONE: + return "NONE"; + case RC_PRESUB_BIAS: + return "(1 - 2 * src0)"; + case RC_PRESUB_SUB: + return "(src1 - src0)"; + case RC_PRESUB_ADD: + return "(src1 + src0)"; + case RC_PRESUB_INV: + return "(1 - src0)"; + default: + return "BAD_PRESUBTRACT_OP"; + } +} + +static void rc_print_comparefunc(FILE * f, const char * lhs, rc_compare_func func, const char * rhs) +{ + if (func == RC_COMPARE_FUNC_NEVER) { + fprintf(f, "false"); + } else if (func == RC_COMPARE_FUNC_ALWAYS) { + fprintf(f, "true"); + } else { + const char * op; + switch(func) { + case RC_COMPARE_FUNC_LESS: op = "<"; break; + case RC_COMPARE_FUNC_EQUAL: op = "=="; break; + case RC_COMPARE_FUNC_LEQUAL: op = "<="; break; + case RC_COMPARE_FUNC_GREATER: op = ">"; break; + case RC_COMPARE_FUNC_NOTEQUAL: op = "!="; break; + case RC_COMPARE_FUNC_GEQUAL: op = ">="; break; + default: op = "???"; break; + } + fprintf(f, "%s %s %s", lhs, op, rhs); + } +} + +static void rc_print_register(FILE * f, rc_register_file file, int index, unsigned int reladdr) +{ + if (file == RC_FILE_NONE) { + fprintf(f, "none"); + } else if (file == RC_FILE_SPECIAL) { + switch(index) { + case RC_SPECIAL_ALU_RESULT: fprintf(f, "aluresult"); break; + default: fprintf(f, "special[%i]", index); break; + } + } else { + const char * filename; + switch(file) { + case RC_FILE_TEMPORARY: filename = "temp"; break; + case RC_FILE_INPUT: filename = "input"; break; + case RC_FILE_OUTPUT: filename = "output"; break; + case RC_FILE_ADDRESS: filename = "addr"; break; + case RC_FILE_CONSTANT: filename = "const"; break; + default: filename = "BAD FILE"; break; + } + fprintf(f, "%s[%i%s]", filename, index, reladdr ? " + addr[0]" : ""); + } +} + +static void rc_print_mask(FILE * f, unsigned int mask) +{ + if (mask & RC_MASK_X) fprintf(f, "x"); + if (mask & RC_MASK_Y) fprintf(f, "y"); + if (mask & RC_MASK_Z) fprintf(f, "z"); + if (mask & RC_MASK_W) fprintf(f, "w"); +} + +static void rc_print_dst_register(FILE * f, struct rc_dst_register dst) +{ + rc_print_register(f, dst.File, dst.Index, 0); + if (dst.WriteMask != RC_MASK_XYZW) { + fprintf(f, "."); + rc_print_mask(f, dst.WriteMask); + } +} + +static char rc_swizzle_char(unsigned int swz) +{ + switch(swz) { + case RC_SWIZZLE_X: return 'x'; + case RC_SWIZZLE_Y: return 'y'; + case RC_SWIZZLE_Z: return 'z'; + case RC_SWIZZLE_W: return 'w'; + case RC_SWIZZLE_ZERO: return '0'; + case RC_SWIZZLE_ONE: return '1'; + case RC_SWIZZLE_HALF: return 'H'; + case RC_SWIZZLE_UNUSED: return '_'; + } + fprintf(stderr, "bad swz: %u\n", swz); + return '?'; +} + +static void rc_print_swizzle(FILE * f, unsigned int swizzle, unsigned int negate) +{ + unsigned int comp; + for(comp = 0; comp < 4; ++comp) { + rc_swizzle swz = GET_SWZ(swizzle, comp); + if (GET_BIT(negate, comp)) + fprintf(f, "-"); + fprintf(f, "%c", rc_swizzle_char(swz)); + } +} + +static void rc_print_presub_instruction(FILE * f, + struct rc_presub_instruction inst) +{ + fprintf(f,"("); + switch(inst.Opcode){ + case RC_PRESUB_BIAS: + fprintf(f, "1 - 2 * "); + rc_print_register(f, inst.SrcReg[0].File, + inst.SrcReg[0].Index,inst.SrcReg[0].RelAddr); + break; + case RC_PRESUB_SUB: + rc_print_register(f, inst.SrcReg[1].File, + inst.SrcReg[1].Index,inst.SrcReg[1].RelAddr); + fprintf(f, " - "); + rc_print_register(f, inst.SrcReg[0].File, + inst.SrcReg[0].Index,inst.SrcReg[0].RelAddr); + break; + case RC_PRESUB_ADD: + rc_print_register(f, inst.SrcReg[1].File, + inst.SrcReg[1].Index,inst.SrcReg[1].RelAddr); + fprintf(f, " + "); + rc_print_register(f, inst.SrcReg[0].File, + inst.SrcReg[0].Index,inst.SrcReg[0].RelAddr); + break; + case RC_PRESUB_INV: + fprintf(f, "1 - "); + rc_print_register(f, inst.SrcReg[0].File, + inst.SrcReg[0].Index,inst.SrcReg[0].RelAddr); + break; + default: + break; + } + fprintf(f, ")"); +} + +static void rc_print_src_register(FILE * f, struct rc_instruction * inst, + struct rc_src_register src) +{ + int trivial_negate = (src.Negate == RC_MASK_NONE || src.Negate == RC_MASK_XYZW); + + if (src.Negate == RC_MASK_XYZW) + fprintf(f, "-"); + if (src.Abs) + fprintf(f, "|"); + + if(src.File == RC_FILE_PRESUB) + rc_print_presub_instruction(f, inst->U.I.PreSub); + else + rc_print_register(f, src.File, src.Index, src.RelAddr); + + if (src.Abs && !trivial_negate) + fprintf(f, "|"); + + if (src.Swizzle != RC_SWIZZLE_XYZW || !trivial_negate) { + fprintf(f, "."); + rc_print_swizzle(f, src.Swizzle, trivial_negate ? 0 : src.Negate); + } + + if (src.Abs && trivial_negate) + fprintf(f, "|"); +} + +static unsigned update_branch_depth(rc_opcode opcode, unsigned *branch_depth) +{ + switch (opcode) { + case RC_OPCODE_IF: + case RC_OPCODE_BGNLOOP: + return (*branch_depth)++ * 2; + + case RC_OPCODE_ENDIF: + case RC_OPCODE_ENDLOOP: + assert(*branch_depth > 0); + return --(*branch_depth) * 2; + + case RC_OPCODE_ELSE: + assert(*branch_depth > 0); + return (*branch_depth - 1) * 2; + + default: + return *branch_depth * 2; + } +} + +static void rc_print_normal_instruction(FILE * f, struct rc_instruction * inst, unsigned *branch_depth) +{ + const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode); + unsigned int reg; + unsigned spaces = update_branch_depth(inst->U.I.Opcode, branch_depth); + + for (unsigned i = 0; i < spaces; i++) + fprintf(f, " "); + + fprintf(f, "%s", opcode->Name); + + switch(inst->U.I.SaturateMode) { + case RC_SATURATE_NONE: break; + case RC_SATURATE_ZERO_ONE: fprintf(f, "_SAT"); break; + case RC_SATURATE_MINUS_PLUS_ONE: fprintf(f, "_SAT2"); break; + default: fprintf(f, "_BAD_SAT"); break; + } + + if (opcode->HasDstReg) { + fprintf(f, " "); + rc_print_dst_register(f, inst->U.I.DstReg); + if (opcode->NumSrcRegs) + fprintf(f, ","); + } + + for(reg = 0; reg < opcode->NumSrcRegs; ++reg) { + if (reg > 0) + fprintf(f, ","); + fprintf(f, " "); + rc_print_src_register(f, inst, inst->U.I.SrcReg[reg]); + } + + if (opcode->HasTexture) { + fprintf(f, ", %s%s[%u]", + textarget_to_string(inst->U.I.TexSrcTarget), + inst->U.I.TexShadow ? "SHADOW" : "", + inst->U.I.TexSrcUnit); + } + + fprintf(f, ";"); + + if (inst->U.I.WriteALUResult) { + fprintf(f, " [aluresult = ("); + rc_print_comparefunc(f, + (inst->U.I.WriteALUResult == RC_ALURESULT_X) ? "x" : "w", + inst->U.I.ALUResultCompare, "0"); + fprintf(f, ")]"); + } + + fprintf(f, "\n"); +} + +static void rc_print_pair_instruction(FILE * f, struct rc_instruction * fullinst, unsigned *branch_depth) +{ + struct rc_pair_instruction * inst = &fullinst->U.P; + int printedsrc = 0; + unsigned spaces = update_branch_depth(inst->RGB.Opcode != RC_OPCODE_NOP ? + inst->RGB.Opcode : inst->Alpha.Opcode, branch_depth); + + for (unsigned i = 0; i < spaces; i++) + fprintf(f, " "); + + for(unsigned int src = 0; src < 3; ++src) { + if (inst->RGB.Src[src].Used) { + if (printedsrc) + fprintf(f, ", "); + fprintf(f, "src%i.xyz = ", src); + rc_print_register(f, inst->RGB.Src[src].File, inst->RGB.Src[src].Index, 0); + printedsrc = 1; + } + if (inst->Alpha.Src[src].Used) { + if (printedsrc) + fprintf(f, ", "); + fprintf(f, "src%i.w = ", src); + rc_print_register(f, inst->Alpha.Src[src].File, inst->Alpha.Src[src].Index, 0); + printedsrc = 1; + } + } + if(inst->RGB.Src[RC_PAIR_PRESUB_SRC].Used) { + fprintf(f, ", srcp.xyz = %s", + presubtract_op_to_string( + inst->RGB.Src[RC_PAIR_PRESUB_SRC].Index)); + } + if(inst->Alpha.Src[RC_PAIR_PRESUB_SRC].Used) { + fprintf(f, ", srcp.w = %s", + presubtract_op_to_string( + inst->Alpha.Src[RC_PAIR_PRESUB_SRC].Index)); + } + fprintf(f, "\n"); + + if (inst->RGB.Opcode != RC_OPCODE_NOP) { + const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->RGB.Opcode); + + for (unsigned i = 0; i < spaces; i++) + fprintf(f, " "); + + fprintf(f, " %s%s", opcode->Name, inst->RGB.Saturate ? "_SAT" : ""); + if (inst->RGB.WriteMask) + fprintf(f, " temp[%i].%s%s%s", inst->RGB.DestIndex, + (inst->RGB.WriteMask & 1) ? "x" : "", + (inst->RGB.WriteMask & 2) ? "y" : "", + (inst->RGB.WriteMask & 4) ? "z" : ""); + if (inst->RGB.OutputWriteMask) + fprintf(f, " color[%i].%s%s%s", inst->RGB.Target, + (inst->RGB.OutputWriteMask & 1) ? "x" : "", + (inst->RGB.OutputWriteMask & 2) ? "y" : "", + (inst->RGB.OutputWriteMask & 4) ? "z" : ""); + if (inst->WriteALUResult == RC_ALURESULT_X) + fprintf(f, " aluresult"); + + for(unsigned int arg = 0; arg < opcode->NumSrcRegs; ++arg) { + const char* abs = inst->RGB.Arg[arg].Abs ? "|" : ""; + const char* neg = inst->RGB.Arg[arg].Negate ? "-" : ""; + fprintf(f, ", %s%ssrc", neg, abs); + if(inst->RGB.Arg[arg].Source == RC_PAIR_PRESUB_SRC) + fprintf(f,"p"); + else + fprintf(f,"%d", inst->RGB.Arg[arg].Source); + fprintf(f,".%c%c%c%s", + rc_swizzle_char(GET_SWZ(inst->RGB.Arg[arg].Swizzle, 0)), + rc_swizzle_char(GET_SWZ(inst->RGB.Arg[arg].Swizzle, 1)), + rc_swizzle_char(GET_SWZ(inst->RGB.Arg[arg].Swizzle, 2)), + abs); + } + fprintf(f, "\n"); + } + + if (inst->Alpha.Opcode != RC_OPCODE_NOP) { + const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->Alpha.Opcode); + + for (unsigned i = 0; i < spaces; i++) + fprintf(f, " "); + + fprintf(f, " %s%s", opcode->Name, inst->Alpha.Saturate ? "_SAT" : ""); + if (inst->Alpha.WriteMask) + fprintf(f, " temp[%i].w", inst->Alpha.DestIndex); + if (inst->Alpha.OutputWriteMask) + fprintf(f, " color[%i].w", inst->Alpha.Target); + if (inst->Alpha.DepthWriteMask) + fprintf(f, " depth.w"); + if (inst->WriteALUResult == RC_ALURESULT_W) + fprintf(f, " aluresult"); + + for(unsigned int arg = 0; arg < opcode->NumSrcRegs; ++arg) { + const char* abs = inst->Alpha.Arg[arg].Abs ? "|" : ""; + const char* neg = inst->Alpha.Arg[arg].Negate ? "-" : ""; + fprintf(f, ", %s%ssrc", neg, abs); + if(inst->Alpha.Arg[arg].Source == RC_PAIR_PRESUB_SRC) + fprintf(f,"p"); + else + fprintf(f,"%d", inst->Alpha.Arg[arg].Source); + fprintf(f,".%c%s", + rc_swizzle_char(GET_SWZ(inst->Alpha.Arg[arg].Swizzle, 0)), abs); + } + fprintf(f, "\n"); + } + + if (inst->WriteALUResult) { + for (unsigned i = 0; i < spaces; i++) + fprintf(f, " "); + + fprintf(f, " [aluresult = ("); + rc_print_comparefunc(f, "result", inst->ALUResultCompare, "0"); + fprintf(f, ")]\n"); + } +} + +/** + * Print program to stderr, default options. + */ +void rc_print_program(const struct rc_program *prog) +{ + unsigned int linenum = 0; + unsigned branch_depth = 0; + struct rc_instruction *inst; + + fprintf(stderr, "# Radeon Compiler Program\n"); + + for(inst = prog->Instructions.Next; inst != &prog->Instructions; inst = inst->Next) { + fprintf(stderr, "%3d: ", linenum); + + if (inst->Type == RC_INSTRUCTION_PAIR) + rc_print_pair_instruction(stderr, inst, &branch_depth); + else + rc_print_normal_instruction(stderr, inst, &branch_depth); + + linenum++; + } +} diff --git a/src/gallium/drivers/r300/compiler/radeon_program_tex.c b/src/gallium/drivers/r300/compiler/radeon_program_tex.c new file mode 100644 index 00000000000..9d69ebd18fb --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_program_tex.c @@ -0,0 +1,528 @@ +/* + * Copyright (C) 2010 Corbin Simpson + * Copyright (C) 2010 Marek Olšák <[email protected]> + * + * 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. + * + */ + +#include "radeon_program_tex.h" + +#include "radeon_compiler_util.h" + +/* Series of transformations to be done on textures. */ + +static struct rc_src_register shadow_fail_value(struct r300_fragment_program_compiler *compiler, + int tmu) +{ + struct rc_src_register reg = { 0, 0, 0, 0, 0, 0 }; + + if (compiler->enable_shadow_ambient) { + reg.File = RC_FILE_CONSTANT; + reg.Index = rc_constants_add_state(&compiler->Base.Program.Constants, + RC_STATE_SHADOW_AMBIENT, tmu); + reg.Swizzle = RC_SWIZZLE_WWWW; + } else { + reg.File = RC_FILE_NONE; + reg.Swizzle = RC_SWIZZLE_0000; + } + + reg.Swizzle = combine_swizzles(reg.Swizzle, + compiler->state.unit[tmu].texture_swizzle); + return reg; +} + +static struct rc_src_register shadow_pass_value(struct r300_fragment_program_compiler *compiler, + int tmu) +{ + struct rc_src_register reg = { 0, 0, 0, 0, 0, 0 }; + + reg.File = RC_FILE_NONE; + reg.Swizzle = combine_swizzles(RC_SWIZZLE_1111, + compiler->state.unit[tmu].texture_swizzle); + return reg; +} + +static void scale_texcoords(struct r300_fragment_program_compiler *compiler, + struct rc_instruction *inst, + unsigned state_constant) +{ + struct rc_instruction *inst_mov; + + unsigned temp = rc_find_free_temporary(&compiler->Base); + + inst_mov = rc_insert_new_instruction(&compiler->Base, inst->Prev); + + inst_mov->U.I.Opcode = RC_OPCODE_MUL; + inst_mov->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst_mov->U.I.DstReg.Index = temp; + inst_mov->U.I.SrcReg[0] = inst->U.I.SrcReg[0]; + inst_mov->U.I.SrcReg[1].File = RC_FILE_CONSTANT; + inst_mov->U.I.SrcReg[1].Index = + rc_constants_add_state(&compiler->Base.Program.Constants, + state_constant, inst->U.I.TexSrcUnit); + + reset_srcreg(&inst->U.I.SrcReg[0]); + inst->U.I.SrcReg[0].File = RC_FILE_TEMPORARY; + inst->U.I.SrcReg[0].Index = temp; +} + +static void projective_divide(struct r300_fragment_program_compiler *compiler, + struct rc_instruction *inst) +{ + struct rc_instruction *inst_mul, *inst_rcp; + + unsigned temp = rc_find_free_temporary(&compiler->Base); + + inst_rcp = rc_insert_new_instruction(&compiler->Base, inst->Prev); + inst_rcp->U.I.Opcode = RC_OPCODE_RCP; + inst_rcp->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst_rcp->U.I.DstReg.Index = temp; + inst_rcp->U.I.DstReg.WriteMask = RC_MASK_W; + inst_rcp->U.I.SrcReg[0] = inst->U.I.SrcReg[0]; + /* Because the input can be arbitrarily swizzled, + * read the component mapped to W. */ + inst_rcp->U.I.SrcReg[0].Swizzle = + RC_MAKE_SWIZZLE_SMEAR(GET_SWZ(inst->U.I.SrcReg[0].Swizzle, 3)); + + inst_mul = rc_insert_new_instruction(&compiler->Base, inst->Prev); + inst_mul->U.I.Opcode = RC_OPCODE_MUL; + inst_mul->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst_mul->U.I.DstReg.Index = temp; + inst_mul->U.I.SrcReg[0] = inst->U.I.SrcReg[0]; + inst_mul->U.I.SrcReg[1].File = RC_FILE_TEMPORARY; + inst_mul->U.I.SrcReg[1].Index = temp; + inst_mul->U.I.SrcReg[1].Swizzle = RC_SWIZZLE_WWWW; + + reset_srcreg(&inst->U.I.SrcReg[0]); + inst->U.I.Opcode = RC_OPCODE_TEX; + inst->U.I.SrcReg[0].File = RC_FILE_TEMPORARY; + inst->U.I.SrcReg[0].Index = temp; +} + +/** + * Transform TEX, TXP, TXB, and KIL instructions in the following ways: + * - implement texture compare (shadow extensions) + * - extract non-native source / destination operands + * - premultiply texture coordinates for RECT + * - extract operand swizzles + * - introduce a temporary register when write masks are needed + */ +int radeonTransformTEX( + struct radeon_compiler * c, + struct rc_instruction * inst, + void* data) +{ + struct r300_fragment_program_compiler *compiler = + (struct r300_fragment_program_compiler*)data; + rc_wrap_mode wrapmode = compiler->state.unit[inst->U.I.TexSrcUnit].wrap_mode; + int is_rect = inst->U.I.TexSrcTarget == RC_TEXTURE_RECT || + compiler->state.unit[inst->U.I.TexSrcUnit].non_normalized_coords; + + if (inst->U.I.Opcode != RC_OPCODE_TEX && + inst->U.I.Opcode != RC_OPCODE_TXB && + inst->U.I.Opcode != RC_OPCODE_TXP && + inst->U.I.Opcode != RC_OPCODE_TXD && + inst->U.I.Opcode != RC_OPCODE_TXL && + inst->U.I.Opcode != RC_OPCODE_KIL) + return 0; + + /* ARB_shadow & EXT_shadow_funcs */ + if (inst->U.I.Opcode != RC_OPCODE_KIL && + ((c->Program.ShadowSamplers & (1 << inst->U.I.TexSrcUnit)) || + (compiler->state.unit[inst->U.I.TexSrcUnit].compare_mode_enabled))) { + rc_compare_func comparefunc = compiler->state.unit[inst->U.I.TexSrcUnit].texture_compare_func; + + if (comparefunc == RC_COMPARE_FUNC_NEVER || comparefunc == RC_COMPARE_FUNC_ALWAYS) { + inst->U.I.Opcode = RC_OPCODE_MOV; + + if (comparefunc == RC_COMPARE_FUNC_ALWAYS) { + inst->U.I.SrcReg[0] = shadow_pass_value(compiler, inst->U.I.TexSrcUnit); + } else { + inst->U.I.SrcReg[0] = shadow_fail_value(compiler, inst->U.I.TexSrcUnit); + } + + return 1; + } else { + struct rc_instruction * inst_rcp = NULL; + struct rc_instruction *inst_mul, *inst_add, *inst_cmp; + unsigned tmp_texsample; + unsigned tmp_sum; + int pass, fail; + + /* Save the output register. */ + struct rc_dst_register output_reg = inst->U.I.DstReg; + unsigned saturate_mode = inst->U.I.SaturateMode; + + /* Redirect TEX to a new temp. */ + tmp_texsample = rc_find_free_temporary(c); + inst->U.I.SaturateMode = 0; + inst->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst->U.I.DstReg.Index = tmp_texsample; + inst->U.I.DstReg.WriteMask = RC_MASK_XYZW; + + tmp_sum = rc_find_free_temporary(c); + + if (inst->U.I.Opcode == RC_OPCODE_TXP) { + /* Compute 1/W. */ + inst_rcp = rc_insert_new_instruction(c, inst); + inst_rcp->U.I.Opcode = RC_OPCODE_RCP; + inst_rcp->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst_rcp->U.I.DstReg.Index = tmp_sum; + inst_rcp->U.I.DstReg.WriteMask = RC_MASK_W; + inst_rcp->U.I.SrcReg[0] = inst->U.I.SrcReg[0]; + inst_rcp->U.I.SrcReg[0].Swizzle = + RC_MAKE_SWIZZLE_SMEAR(GET_SWZ(inst->U.I.SrcReg[0].Swizzle, 3)); + } + + /* Divide Z by W (if it's TXP) and saturate. */ + inst_mul = rc_insert_new_instruction(c, inst_rcp ? inst_rcp : inst); + inst_mul->U.I.Opcode = inst->U.I.Opcode == RC_OPCODE_TXP ? RC_OPCODE_MUL : RC_OPCODE_MOV; + inst_mul->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst_mul->U.I.DstReg.Index = tmp_sum; + inst_mul->U.I.DstReg.WriteMask = RC_MASK_W; + inst_mul->U.I.SaturateMode = RC_SATURATE_ZERO_ONE; + inst_mul->U.I.SrcReg[0] = inst->U.I.SrcReg[0]; + inst_mul->U.I.SrcReg[0].Swizzle = + RC_MAKE_SWIZZLE_SMEAR(GET_SWZ(inst->U.I.SrcReg[0].Swizzle, 2)); + if (inst->U.I.Opcode == RC_OPCODE_TXP) { + inst_mul->U.I.SrcReg[1].File = RC_FILE_TEMPORARY; + inst_mul->U.I.SrcReg[1].Index = tmp_sum; + inst_mul->U.I.SrcReg[1].Swizzle = RC_SWIZZLE_WWWW; + } + + /* Add the depth texture value. */ + inst_add = rc_insert_new_instruction(c, inst_mul); + inst_add->U.I.Opcode = RC_OPCODE_ADD; + inst_add->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst_add->U.I.DstReg.Index = tmp_sum; + inst_add->U.I.DstReg.WriteMask = RC_MASK_W; + inst_add->U.I.SrcReg[0].File = RC_FILE_TEMPORARY; + inst_add->U.I.SrcReg[0].Index = tmp_sum; + inst_add->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_WWWW; + inst_add->U.I.SrcReg[1].File = RC_FILE_TEMPORARY; + inst_add->U.I.SrcReg[1].Index = tmp_texsample; + inst_add->U.I.SrcReg[1].Swizzle = RC_SWIZZLE_XXXX; + + /* Note that SrcReg[0] is r, SrcReg[1] is tex and: + * LESS: r < tex <=> -tex+r < 0 + * GEQUAL: r >= tex <=> not (-tex+r < 0) + * GREATER: r > tex <=> tex-r < 0 + * LEQUAL: r <= tex <=> not ( tex-r < 0) + * EQUAL: GEQUAL + * NOTEQUAL:LESS + */ + + /* This negates either r or tex: */ + if (comparefunc == RC_COMPARE_FUNC_LESS || comparefunc == RC_COMPARE_FUNC_GEQUAL || + comparefunc == RC_COMPARE_FUNC_EQUAL || comparefunc == RC_COMPARE_FUNC_NOTEQUAL) + inst_add->U.I.SrcReg[1].Negate = inst_add->U.I.SrcReg[1].Negate ^ RC_MASK_XYZW; + else + inst_add->U.I.SrcReg[0].Negate = inst_add->U.I.SrcReg[0].Negate ^ RC_MASK_XYZW; + + /* This negates the whole expresion: */ + if (comparefunc == RC_COMPARE_FUNC_LESS || comparefunc == RC_COMPARE_FUNC_GREATER || + comparefunc == RC_COMPARE_FUNC_NOTEQUAL) { + pass = 1; + fail = 2; + } else { + pass = 2; + fail = 1; + } + + inst_cmp = rc_insert_new_instruction(c, inst_add); + inst_cmp->U.I.Opcode = RC_OPCODE_CMP; + inst_cmp->U.I.SaturateMode = saturate_mode; + inst_cmp->U.I.DstReg = output_reg; + inst_cmp->U.I.SrcReg[0].File = RC_FILE_TEMPORARY; + inst_cmp->U.I.SrcReg[0].Index = tmp_sum; + inst_cmp->U.I.SrcReg[0].Swizzle = + combine_swizzles(RC_SWIZZLE_WWWW, + compiler->state.unit[inst->U.I.TexSrcUnit].texture_swizzle); + inst_cmp->U.I.SrcReg[pass] = shadow_pass_value(compiler, inst->U.I.TexSrcUnit); + inst_cmp->U.I.SrcReg[fail] = shadow_fail_value(compiler, inst->U.I.TexSrcUnit); + + assert(tmp_texsample != tmp_sum); + } + } + + /* R300 cannot sample from rectangles and the wrap mode fallback needs + * normalized coordinates anyway. */ + if (inst->U.I.Opcode != RC_OPCODE_KIL && + is_rect && (!c->is_r500 || wrapmode != RC_WRAP_NONE)) { + scale_texcoords(compiler, inst, RC_STATE_R300_TEXRECT_FACTOR); + inst->U.I.TexSrcTarget = RC_TEXTURE_2D; + } + + /* Divide by W if needed. */ + if (inst->U.I.Opcode == RC_OPCODE_TXP && + (wrapmode == RC_WRAP_REPEAT || wrapmode == RC_WRAP_MIRRORED_REPEAT || + compiler->state.unit[inst->U.I.TexSrcUnit].clamp_and_scale_before_fetch)) { + projective_divide(compiler, inst); + } + + /* Texture wrap modes don't work on NPOT textures. + * + * Non-wrapped/clamped texcoords with NPOT are free in HW. Repeat and + * mirroring are not. If we need to repeat, we do: + * + * MUL temp, texcoord, <scaling factor constant> + * FRC temp, temp ; Discard integer portion of coords + * + * This gives us coords in [0, 1]. + * + * Mirroring is trickier. We're going to start out like repeat: + * + * MUL temp, texcoord, <scaling factor constant> ; De-mirror across axes + * MUL temp, temp, 0.5 ; Pattern repeats in [0, 2] + * ; so scale to [0, 1] + * FRC temp, temp ; Make the pattern repeat + * MAD temp, temp, 2, -1 ; Move the pattern to [-1, 1] + * ADD temp, 1, -abs(temp) ; Now comes a neat trick: use abs to mirror the pattern. + * ; The pattern is backwards, so reverse it (1-x). + * + * This gives us coords in [0, 1]. + * + * ~ C & M. ;) + */ + if (inst->U.I.Opcode != RC_OPCODE_KIL && + wrapmode != RC_WRAP_NONE) { + struct rc_instruction *inst_mov; + unsigned temp = rc_find_free_temporary(c); + + if (wrapmode == RC_WRAP_REPEAT) { + /* Both instructions will be paired up. */ + struct rc_instruction *inst_frc = rc_insert_new_instruction(c, inst->Prev); + + inst_frc->U.I.Opcode = RC_OPCODE_FRC; + inst_frc->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst_frc->U.I.DstReg.Index = temp; + inst_frc->U.I.DstReg.WriteMask = RC_MASK_XYZ; + inst_frc->U.I.SrcReg[0] = inst->U.I.SrcReg[0]; + } else if (wrapmode == RC_WRAP_MIRRORED_REPEAT) { + /* + * Function: + * f(v) = 1 - abs(frac(v * 0.5) * 2 - 1) + * + * Code: + * MUL temp, src0, 0.5 + * FRC temp, temp + * MAD temp, temp, 2, -1 + * ADD temp, 1, -abs(temp) + */ + + struct rc_instruction *inst_mul, *inst_frc, *inst_mad, *inst_add; + unsigned two, two_swizzle; + + inst_mul = rc_insert_new_instruction(c, inst->Prev); + + inst_mul->U.I.Opcode = RC_OPCODE_MUL; + inst_mul->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst_mul->U.I.DstReg.Index = temp; + inst_mul->U.I.DstReg.WriteMask = RC_MASK_XYZ; + inst_mul->U.I.SrcReg[0] = inst->U.I.SrcReg[0]; + inst_mul->U.I.SrcReg[1].Swizzle = RC_SWIZZLE_HHHH; + + inst_frc = rc_insert_new_instruction(c, inst->Prev); + + inst_frc->U.I.Opcode = RC_OPCODE_FRC; + inst_frc->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst_frc->U.I.DstReg.Index = temp; + inst_frc->U.I.DstReg.WriteMask = RC_MASK_XYZ; + inst_frc->U.I.SrcReg[0].File = RC_FILE_TEMPORARY; + inst_frc->U.I.SrcReg[0].Index = temp; + inst_frc->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_XYZ0; + + two = rc_constants_add_immediate_scalar(&c->Program.Constants, 2, &two_swizzle); + inst_mad = rc_insert_new_instruction(c, inst->Prev); + + inst_mad->U.I.Opcode = RC_OPCODE_MAD; + inst_mad->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst_mad->U.I.DstReg.Index = temp; + inst_mad->U.I.DstReg.WriteMask = RC_MASK_XYZ; + inst_mad->U.I.SrcReg[0].File = RC_FILE_TEMPORARY; + inst_mad->U.I.SrcReg[0].Index = temp; + inst_mad->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_XYZ0; + inst_mad->U.I.SrcReg[1].File = RC_FILE_CONSTANT; + inst_mad->U.I.SrcReg[1].Index = two; + inst_mad->U.I.SrcReg[1].Swizzle = two_swizzle; + inst_mad->U.I.SrcReg[2].Swizzle = RC_SWIZZLE_1111; + inst_mad->U.I.SrcReg[2].Negate = RC_MASK_XYZ; + + inst_add = rc_insert_new_instruction(c, inst->Prev); + + inst_add->U.I.Opcode = RC_OPCODE_ADD; + inst_add->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst_add->U.I.DstReg.Index = temp; + inst_add->U.I.DstReg.WriteMask = RC_MASK_XYZ; + inst_add->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_1111; + inst_add->U.I.SrcReg[1].File = RC_FILE_TEMPORARY; + inst_add->U.I.SrcReg[1].Index = temp; + inst_add->U.I.SrcReg[1].Swizzle = RC_SWIZZLE_XYZ0; + inst_add->U.I.SrcReg[1].Abs = 1; + inst_add->U.I.SrcReg[1].Negate = RC_MASK_XYZ; + } else if (wrapmode == RC_WRAP_MIRRORED_CLAMP) { + /* + * Mirrored clamp modes are bloody simple, we just use abs + * to mirror [0, 1] into [-1, 0]. This works for + * all modes i.e. CLAMP, CLAMP_TO_EDGE, and CLAMP_TO_BORDER. + */ + struct rc_instruction *inst_mov; + + inst_mov = rc_insert_new_instruction(c, inst->Prev); + + inst_mov->U.I.Opcode = RC_OPCODE_MOV; + inst_mov->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst_mov->U.I.DstReg.Index = temp; + inst_mov->U.I.DstReg.WriteMask = RC_MASK_XYZ; + inst_mov->U.I.SrcReg[0] = inst->U.I.SrcReg[0]; + inst_mov->U.I.SrcReg[0].Abs = 1; + } + + /* Preserve W for TXP/TXB. */ + inst_mov = rc_insert_new_instruction(c, inst->Prev); + + inst_mov->U.I.Opcode = RC_OPCODE_MOV; + inst_mov->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst_mov->U.I.DstReg.Index = temp; + inst_mov->U.I.DstReg.WriteMask = RC_MASK_W; + inst_mov->U.I.SrcReg[0] = inst->U.I.SrcReg[0]; + + reset_srcreg(&inst->U.I.SrcReg[0]); + inst->U.I.SrcReg[0].File = RC_FILE_TEMPORARY; + inst->U.I.SrcReg[0].Index = temp; + } + + /* NPOT -> POT conversion for 3D textures. */ + if (inst->U.I.Opcode != RC_OPCODE_KIL && + compiler->state.unit[inst->U.I.TexSrcUnit].clamp_and_scale_before_fetch) { + struct rc_instruction *inst_mov; + unsigned temp = rc_find_free_temporary(c); + + /* Saturate XYZ. */ + inst_mov = rc_insert_new_instruction(c, inst->Prev); + inst_mov->U.I.Opcode = RC_OPCODE_MOV; + inst_mov->U.I.SaturateMode = RC_SATURATE_ZERO_ONE; + inst_mov->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst_mov->U.I.DstReg.Index = temp; + inst_mov->U.I.DstReg.WriteMask = RC_MASK_XYZ; + inst_mov->U.I.SrcReg[0] = inst->U.I.SrcReg[0]; + + /* Copy W. */ + inst_mov = rc_insert_new_instruction(c, inst->Prev); + inst_mov->U.I.Opcode = RC_OPCODE_MOV; + inst_mov->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst_mov->U.I.DstReg.Index = temp; + inst_mov->U.I.DstReg.WriteMask = RC_MASK_W; + inst_mov->U.I.SrcReg[0] = inst->U.I.SrcReg[0]; + + reset_srcreg(&inst->U.I.SrcReg[0]); + inst->U.I.SrcReg[0].File = RC_FILE_TEMPORARY; + inst->U.I.SrcReg[0].Index = temp; + + scale_texcoords(compiler, inst, RC_STATE_R300_TEXSCALE_FACTOR); + } + + /* Convert SNORM-encoded ATI1N sampled as UNORM to SNORM. + * Formula: dst = tex > 0.5 ? tex*2-2 : tex*2 + */ + if (inst->U.I.Opcode != RC_OPCODE_KIL && + compiler->state.unit[inst->U.I.TexSrcUnit].convert_unorm_to_snorm) { + unsigned two, two_swizzle; + struct rc_instruction *inst_mul, *inst_mad, *inst_cnd; + + two = rc_constants_add_immediate_scalar(&c->Program.Constants, 2.35, &two_swizzle); + + inst_mul = rc_insert_new_instruction(c, inst); + inst_mul->U.I.Opcode = RC_OPCODE_MUL; + inst_mul->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst_mul->U.I.DstReg.Index = rc_find_free_temporary(c); + inst_mul->U.I.SrcReg[0].File = RC_FILE_TEMPORARY; + inst_mul->U.I.SrcReg[0].Index = rc_find_free_temporary(c); /* redirected TEX output */ + inst_mul->U.I.SrcReg[1].File = RC_FILE_CONSTANT; /* 2 */ + inst_mul->U.I.SrcReg[1].Index = two; + inst_mul->U.I.SrcReg[1].Swizzle = two_swizzle; + + inst_mad = rc_insert_new_instruction(c, inst_mul); + inst_mad->U.I.Opcode = RC_OPCODE_MAD; + inst_mad->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst_mad->U.I.DstReg.Index = rc_find_free_temporary(c); + inst_mad->U.I.SrcReg[0] = inst_mul->U.I.SrcReg[0]; /* redirected TEX output */ + inst_mad->U.I.SrcReg[1] = inst_mul->U.I.SrcReg[1]; /* 2 */ + inst_mad->U.I.SrcReg[2] = inst_mul->U.I.SrcReg[1]; /* 2 */ + inst_mad->U.I.SrcReg[2].Negate = RC_MASK_XYZW; + + inst_cnd = rc_insert_new_instruction(c, inst_mad); + inst_cnd->U.I.Opcode = RC_OPCODE_CND; + inst_cnd->U.I.SaturateMode = inst->U.I.SaturateMode; + inst_cnd->U.I.DstReg = inst->U.I.DstReg; + inst_cnd->U.I.SrcReg[0].File = RC_FILE_TEMPORARY; + inst_cnd->U.I.SrcReg[0].Index = inst_mad->U.I.DstReg.Index; + inst_cnd->U.I.SrcReg[0].Swizzle = compiler->state.unit[inst->U.I.TexSrcUnit].texture_swizzle; + inst_cnd->U.I.SrcReg[1].File = RC_FILE_TEMPORARY; + inst_cnd->U.I.SrcReg[1].Index = inst_mul->U.I.DstReg.Index; + inst_cnd->U.I.SrcReg[1].Swizzle = compiler->state.unit[inst->U.I.TexSrcUnit].texture_swizzle; + inst_cnd->U.I.SrcReg[2] = inst_mul->U.I.SrcReg[0]; /* redirected TEX output */ + + inst->U.I.SaturateMode = 0; + inst->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst->U.I.DstReg.Index = inst_mul->U.I.SrcReg[0].Index; + inst->U.I.DstReg.WriteMask = RC_MASK_XYZW; + } + + /* Cannot write texture to output registers or with saturate (all chips), + * or with masks (non-r500). */ + if (inst->U.I.Opcode != RC_OPCODE_KIL && + (inst->U.I.DstReg.File != RC_FILE_TEMPORARY || + inst->U.I.SaturateMode || + (!c->is_r500 && inst->U.I.DstReg.WriteMask != RC_MASK_XYZW))) { + struct rc_instruction * inst_mov = rc_insert_new_instruction(c, inst); + + inst_mov->U.I.Opcode = RC_OPCODE_MOV; + inst_mov->U.I.SaturateMode = inst->U.I.SaturateMode; + inst_mov->U.I.DstReg = inst->U.I.DstReg; + inst_mov->U.I.SrcReg[0].File = RC_FILE_TEMPORARY; + inst_mov->U.I.SrcReg[0].Index = rc_find_free_temporary(c); + + inst->U.I.SaturateMode = 0; + inst->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst->U.I.DstReg.Index = inst_mov->U.I.SrcReg[0].Index; + inst->U.I.DstReg.WriteMask = RC_MASK_XYZW; + } + + /* Cannot read texture coordinate from constants file */ + if (inst->U.I.SrcReg[0].File != RC_FILE_TEMPORARY && inst->U.I.SrcReg[0].File != RC_FILE_INPUT) { + struct rc_instruction * inst_mov = rc_insert_new_instruction(c, inst->Prev); + + inst_mov->U.I.Opcode = RC_OPCODE_MOV; + inst_mov->U.I.DstReg.File = RC_FILE_TEMPORARY; + inst_mov->U.I.DstReg.Index = rc_find_free_temporary(c); + inst_mov->U.I.SrcReg[0] = inst->U.I.SrcReg[0]; + + reset_srcreg(&inst->U.I.SrcReg[0]); + inst->U.I.SrcReg[0].File = RC_FILE_TEMPORARY; + inst->U.I.SrcReg[0].Index = inst_mov->U.I.DstReg.Index; + } + + return 1; +} diff --git a/src/gallium/drivers/r300/compiler/radeon_program_tex.h b/src/gallium/drivers/r300/compiler/radeon_program_tex.h new file mode 100644 index 00000000000..a0105051ac4 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_program_tex.h @@ -0,0 +1,39 @@ +/* + * Copyright (C) 2010 Corbin Simpson + * + * 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. + * + */ + +#ifndef __RADEON_PROGRAM_TEX_H_ +#define __RADEON_PROGRAM_TEX_H_ + +#include "radeon_compiler.h" +#include "radeon_program.h" + +int radeonTransformTEX( + struct radeon_compiler * c, + struct rc_instruction * inst, + void* data); + +#endif /* __RADEON_PROGRAM_TEX_H_ */ diff --git a/src/gallium/drivers/r300/compiler/radeon_remove_constants.c b/src/gallium/drivers/r300/compiler/radeon_remove_constants.c new file mode 100644 index 00000000000..7d76585a593 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_remove_constants.c @@ -0,0 +1,150 @@ +/* + * Copyright (C) 2010 Marek Olšák <[email protected]> + * + * 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. + * + */ + +#include "radeon_remove_constants.h" +#include "radeon_dataflow.h" + +struct mark_used_data { + unsigned char * const_used; + unsigned * has_rel_addr; +}; + +static void remap_regs(void * userdata, struct rc_instruction * inst, + rc_register_file * pfile, unsigned int * pindex) +{ + unsigned *inv_remap_table = userdata; + + if (*pfile == RC_FILE_CONSTANT) { + *pindex = inv_remap_table[*pindex]; + } +} + +static void mark_used(void * userdata, struct rc_instruction * inst, + struct rc_src_register * src) +{ + struct mark_used_data * d = userdata; + + if (src->File == RC_FILE_CONSTANT) { + if (src->RelAddr) { + *d->has_rel_addr = 1; + } else { + d->const_used[src->Index] = 1; + } + } +} + +void rc_remove_unused_constants(struct radeon_compiler *c, void *user) +{ + unsigned **out_remap_table = (unsigned**)user; + unsigned char *const_used; + unsigned *remap_table; + unsigned *inv_remap_table; + unsigned has_rel_addr = 0; + unsigned is_identity = 1; + unsigned are_externals_remapped = 0; + struct rc_constant *constants = c->Program.Constants.Constants; + struct mark_used_data d; + unsigned new_count; + + if (!c->Program.Constants.Count) { + *out_remap_table = NULL; + return; + } + + const_used = malloc(c->Program.Constants.Count); + memset(const_used, 0, c->Program.Constants.Count); + + d.const_used = const_used; + d.has_rel_addr = &has_rel_addr; + + /* Pass 1: Mark used constants. */ + for (struct rc_instruction *inst = c->Program.Instructions.Next; + inst != &c->Program.Instructions; inst = inst->Next) { + rc_for_all_reads_src(inst, mark_used, &d); + } + + /* Pass 2: If there is relative addressing or dead constant elimination + * is disabled, mark all externals as used. */ + if (has_rel_addr || !c->remove_unused_constants) { + for (unsigned i = 0; i < c->Program.Constants.Count; i++) + if (constants[i].Type == RC_CONSTANT_EXTERNAL) + const_used[i] = 1; + } + + /* Pass 3: Make the remapping table and remap constants. + * This pass removes unused constants simply by overwriting them by other constants. */ + remap_table = malloc(c->Program.Constants.Count * sizeof(unsigned)); + inv_remap_table = malloc(c->Program.Constants.Count * sizeof(unsigned)); + new_count = 0; + + for (unsigned i = 0; i < c->Program.Constants.Count; i++) { + if (const_used[i]) { + remap_table[new_count] = i; + inv_remap_table[i] = new_count; + + if (i != new_count) { + if (constants[i].Type == RC_CONSTANT_EXTERNAL) + are_externals_remapped = 1; + + constants[new_count] = constants[i]; + is_identity = 0; + } + new_count++; + } + } + + /* is_identity ==> new_count == old_count + * !is_identity ==> new_count < old_count */ + assert( is_identity || new_count < c->Program.Constants.Count); + assert(!((has_rel_addr || !c->remove_unused_constants) && are_externals_remapped)); + + /* Pass 4: Redirect reads of all constants to their new locations. */ + if (!is_identity) { + for (struct rc_instruction *inst = c->Program.Instructions.Next; + inst != &c->Program.Instructions; inst = inst->Next) { + rc_remap_registers(inst, remap_regs, inv_remap_table); + } + } + + /* Set the new constant count. Note that new_count may be less than + * Count even though the remapping function is identity. In that case, + * the constants have been removed at the end of the array. */ + c->Program.Constants.Count = new_count; + + if (are_externals_remapped) { + *out_remap_table = remap_table; + } else { + *out_remap_table = NULL; + free(remap_table); + } + + free(const_used); + free(inv_remap_table); + + if (c->Debug & RC_DBG_LOG) + rc_constants_print(&c->Program.Constants); +} diff --git a/src/gallium/drivers/r300/compiler/radeon_remove_constants.h b/src/gallium/drivers/r300/compiler/radeon_remove_constants.h new file mode 100644 index 00000000000..f29113b922b --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_remove_constants.h @@ -0,0 +1,35 @@ +/* + * Copyright (C) 2010 Marek Olšák <[email protected]> + * + * 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. + * + */ + +#ifndef RADEON_REMOVE_CONSTANTS_H +#define RADEON_REMOVE_CONSTANTS_H + +#include "radeon_compiler.h" + +void rc_remove_unused_constants(struct radeon_compiler *c, void *user); + +#endif diff --git a/src/gallium/drivers/r300/compiler/radeon_rename_regs.c b/src/gallium/drivers/r300/compiler/radeon_rename_regs.c new file mode 100644 index 00000000000..cafa0579734 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_rename_regs.c @@ -0,0 +1,92 @@ +/* + * Copyright 2010 Tom Stellard <[email protected]> + * + * 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 + */ + +#include "radeon_rename_regs.h" + +#include "radeon_compiler.h" +#include "radeon_dataflow.h" +#include "radeon_program.h" + +/** + * This function renames registers in an attempt to get the code close to + * SSA form. After this function has completed, most of the register are only + * written to one time, with a few exceptions. + * + * This function assumes all the instructions are still of type + * RC_INSTRUCTION_NORMAL. + */ +void rc_rename_regs(struct radeon_compiler *c, void *user) +{ + unsigned int i, used_length; + int new_index; + struct rc_instruction * inst; + struct rc_reader_data reader_data; + unsigned char * used; + + /* XXX Remove this once the register allocation works with flow control. */ + for(inst = c->Program.Instructions.Next; + inst != &c->Program.Instructions; + inst = inst->Next) { + if (inst->U.I.Opcode == RC_OPCODE_BGNLOOP) + return; + } + + used_length = 2 * rc_recompute_ips(c); + used = memory_pool_malloc(&c->Pool, sizeof(unsigned char) * used_length); + memset(used, 0, sizeof(unsigned char) * used_length); + + rc_get_used_temporaries(c, used, used_length); + for(inst = c->Program.Instructions.Next; + inst != &c->Program.Instructions; + inst = inst->Next) { + + if (inst->U.I.DstReg.File != RC_FILE_TEMPORARY) + continue; + + reader_data.ExitOnAbort = 1; + rc_get_readers(c, inst, &reader_data, NULL, NULL, NULL); + + if (reader_data.Abort || reader_data.ReaderCount == 0) + continue; + + new_index = rc_find_free_temporary_list(c, used, used_length, + RC_MASK_XYZW); + if (new_index < 0) { + rc_error(c, "Ran out of temporary registers\n"); + return; + } + + reader_data.Writer->U.I.DstReg.Index = new_index; + for(i = 0; i < reader_data.ReaderCount; i++) { + reader_data.Readers[i].U.I.Src->Index = new_index; + } + } +} diff --git a/src/gallium/drivers/r300/compiler/radeon_rename_regs.h b/src/gallium/drivers/r300/compiler/radeon_rename_regs.h new file mode 100644 index 00000000000..3baf29f6120 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_rename_regs.h @@ -0,0 +1,9 @@ + +#ifndef RADEON_RENAME_REGS_H +#define RADEON_RENAME_REGS_H + +struct radeon_compiler; + +void rc_rename_regs(struct radeon_compiler *c, void *user); + +#endif /* RADEON_RENAME_REGS_H */ diff --git a/src/gallium/drivers/r300/compiler/radeon_swizzle.h b/src/gallium/drivers/r300/compiler/radeon_swizzle.h new file mode 100644 index 00000000000..c81d5f7a5e9 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_swizzle.h @@ -0,0 +1,57 @@ +/* + * Copyright (C) 2009 Nicolai Haehnle. + * + * 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. + * + */ + +#ifndef RADEON_SWIZZLE_H +#define RADEON_SWIZZLE_H + +#include "radeon_program.h" + +struct rc_swizzle_split { + unsigned char NumPhases; + unsigned char Phase[4]; +}; + +/** + * Describe the swizzling capability of target hardware. + */ +struct rc_swizzle_caps { + /** + * Check whether the given swizzle, absolute and negate combination + * can be implemented natively by the hardware for this opcode. + * + * \return 1 if the swizzle is native for the given opcode + */ + int (*IsNative)(rc_opcode opcode, struct rc_src_register reg); + + /** + * Determine how to split access to the masked channels of the + * given source register to obtain ALU-native swizzles. + */ + void (*Split)(struct rc_src_register reg, unsigned int mask, struct rc_swizzle_split * split); +}; + +#endif /* RADEON_SWIZZLE_H */ diff --git a/src/gallium/drivers/r300/compiler/radeon_variable.c b/src/gallium/drivers/r300/compiler/radeon_variable.c new file mode 100644 index 00000000000..938fb8421f2 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_variable.c @@ -0,0 +1,517 @@ +/* + * Copyright 2011 Tom Stellard <[email protected]> + * + * 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. + * + */ + +#include "radeon_variable.h" + +#include "memory_pool.h" +#include "radeon_compiler_util.h" +#include "radeon_dataflow.h" +#include "radeon_list.h" +#include "radeon_opcodes.h" +#include "radeon_program.h" + +/** + * Rewrite the index and writemask for the destination register of var + * and its friends to new_index and new_writemask. This function also takes + * care of rewriting the swizzles for the sources of var. + */ +void rc_variable_change_dst( + struct rc_variable * var, + unsigned int new_index, + unsigned int new_writemask) +{ + struct rc_variable * var_ptr; + struct rc_list * readers; + unsigned int old_mask = rc_variable_writemask_sum(var); + unsigned int conversion_swizzle = + rc_make_conversion_swizzle(old_mask, new_writemask); + + for (var_ptr = var; var_ptr; var_ptr = var_ptr->Friend) { + if (var_ptr->Inst->Type == RC_INSTRUCTION_NORMAL) { + rc_normal_rewrite_writemask(var_ptr->Inst, + conversion_swizzle); + var_ptr->Inst->U.I.DstReg.Index = new_index; + } else { + struct rc_pair_sub_instruction * sub; + if (var_ptr->Dst.WriteMask == RC_MASK_W) { + assert(new_writemask & RC_MASK_W); + sub = &var_ptr->Inst->U.P.Alpha; + } else { + sub = &var_ptr->Inst->U.P.RGB; + rc_pair_rewrite_writemask(sub, + conversion_swizzle); + } + sub->DestIndex = new_index; + } + } + + readers = rc_variable_readers_union(var); + + for ( ; readers; readers = readers->Next) { + struct rc_reader * reader = readers->Item; + if (reader->Inst->Type == RC_INSTRUCTION_NORMAL) { + reader->U.I.Src->Index = new_index; + reader->U.I.Src->Swizzle = rc_rewrite_swizzle( + reader->U.I.Src->Swizzle, conversion_swizzle); + } else { + struct rc_pair_instruction * pair_inst = + &reader->Inst->U.P; + unsigned int src_type = rc_source_type_swz( + reader->U.P.Arg->Swizzle); + + int src_index = reader->U.P.Arg->Source; + if (src_index == RC_PAIR_PRESUB_SRC) { + src_index = rc_pair_get_src_index( + pair_inst, reader->U.P.Src); + } + /* Try to delete the old src, it is OK if this fails, + * because rc_pair_alloc_source might be able to + * find a source the ca be reused. + */ + if (rc_pair_remove_src(reader->Inst, src_type, + src_index, old_mask)) { + /* Reuse the source index of the source that + * was just deleted and set its register + * index. We can't use rc_pair_alloc_source + * for this becuase it might return a source + * index that is already being used. */ + if (src_type & RC_SOURCE_RGB) { + pair_inst->RGB.Src[src_index] + .Used = 1; + pair_inst->RGB.Src[src_index] + .Index = new_index; + pair_inst->RGB.Src[src_index] + .File = RC_FILE_TEMPORARY; + } + if (src_type & RC_SOURCE_ALPHA) { + pair_inst->Alpha.Src[src_index] + .Used = 1; + pair_inst->Alpha.Src[src_index] + .Index = new_index; + pair_inst->Alpha.Src[src_index] + .File = RC_FILE_TEMPORARY; + } + } else { + src_index = rc_pair_alloc_source( + &reader->Inst->U.P, + src_type & RC_SOURCE_RGB, + src_type & RC_SOURCE_ALPHA, + RC_FILE_TEMPORARY, + new_index); + if (src_index < 0) { + rc_error(var->C, "Rewrite of inst %u failed " + "Can't allocate source for " + "Inst %u src_type=%x " + "new_index=%u new_mask=%u\n", + var->Inst->IP, reader->Inst->IP, src_type, new_index, new_writemask); + continue; + } + } + reader->U.P.Arg->Swizzle = rc_rewrite_swizzle( + reader->U.P.Arg->Swizzle, conversion_swizzle); + if (reader->U.P.Arg->Source != RC_PAIR_PRESUB_SRC) { + reader->U.P.Arg->Source = src_index; + } + } + } +} + +/** + * Compute the live intervals for var and its friends. + */ +void rc_variable_compute_live_intervals(struct rc_variable * var) +{ + while(var) { + unsigned int i; + unsigned int start = var->Inst->IP; + + for (i = 0; i < var->ReaderCount; i++) { + unsigned int chan; + unsigned int chan_start = start; + unsigned int chan_end = var->Readers[i].Inst->IP; + unsigned int mask = var->Readers[i].WriteMask; + struct rc_instruction * inst; + + /* Extend the live interval of T0 to the start of the + * loop for sequences like: + * BGNLOOP + * read T0 + * ... + * write T0 + * ENDLOOP + */ + if (var->Readers[i].Inst->IP < start) { + struct rc_instruction * bgnloop = + rc_match_endloop(var->Readers[i].Inst); + chan_start = bgnloop->IP; + } + + /* Extend the live interval of T0 to the start of the + * loop in case there is a BRK instruction in the loop + * (we don't actually check for a BRK instruction we + * assume there is one somewhere in the loop, which + * there usually is) for sequences like: + * BGNLOOP + * ... + * conditional BRK + * ... + * write T0 + * ENDLOOP + * read T0 + *************************************************** + * Extend the live interval of T0 to the end of the + * loop for sequences like: + * write T0 + * BGNLOOP + * ... + * read T0 + * ENDLOOP + */ + for (inst = var->Inst; inst != var->Readers[i].Inst; + inst = inst->Next) { + rc_opcode op = rc_get_flow_control_inst(inst); + if (op == RC_OPCODE_ENDLOOP) { + struct rc_instruction * bgnloop = + rc_match_endloop(inst); + if (bgnloop->IP < chan_start) { + chan_start = bgnloop->IP; + } + } else if (op == RC_OPCODE_BGNLOOP) { + struct rc_instruction * endloop = + rc_match_bgnloop(inst); + if (endloop->IP > chan_end) { + chan_end = endloop->IP; + } + } + } + + for (chan = 0; chan < 4; chan++) { + if ((mask >> chan) & 0x1) { + if (!var->Live[chan].Used + || chan_start < var->Live[chan].Start) { + var->Live[chan].Start = + chan_start; + } + if (!var->Live[chan].Used + || chan_end > var->Live[chan].End) { + var->Live[chan].End = chan_end; + } + var->Live[chan].Used = 1; + } + } + } + var = var->Friend; + } +} + +/** + * @return 1 if a and b share a reader + * @return 0 if they do not + */ +static unsigned int readers_intersect( + struct rc_variable * a, + struct rc_variable * b) +{ + unsigned int a_index, b_index; + for (a_index = 0; a_index < a->ReaderCount; a_index++) { + struct rc_reader reader_a = a->Readers[a_index]; + for (b_index = 0; b_index < b->ReaderCount; b_index++) { + struct rc_reader reader_b = b->Readers[b_index]; + if (reader_a.Inst->Type == RC_INSTRUCTION_NORMAL + && reader_b.Inst->Type == RC_INSTRUCTION_NORMAL + && reader_a.U.I.Src == reader_b.U.I.Src) { + + return 1; + } + if (reader_a.Inst->Type == RC_INSTRUCTION_PAIR + && reader_b.Inst->Type == RC_INSTRUCTION_PAIR + && reader_a.U.P.Src == reader_b.U.P.Src) { + + return 1; + } + } + } + return 0; +} + +void rc_variable_add_friend( + struct rc_variable * var, + struct rc_variable * friend) +{ + assert(var->Dst.Index == friend->Dst.Index); + while(var->Friend) { + var = var->Friend; + } + var->Friend = friend; +} + +struct rc_variable * rc_variable( + struct radeon_compiler * c, + unsigned int DstFile, + unsigned int DstIndex, + unsigned int DstWriteMask, + struct rc_reader_data * reader_data) +{ + struct rc_variable * new = + memory_pool_malloc(&c->Pool, sizeof(struct rc_variable)); + memset(new, 0, sizeof(struct rc_variable)); + new->C = c; + new->Dst.File = DstFile; + new->Dst.Index = DstIndex; + new->Dst.WriteMask = DstWriteMask; + if (reader_data) { + new->Inst = reader_data->Writer; + new->ReaderCount = reader_data->ReaderCount; + new->Readers = reader_data->Readers; + } + return new; +} + +static void get_variable_helper( + struct rc_list ** variable_list, + struct rc_variable * variable) +{ + struct rc_list * list_ptr; + for (list_ptr = *variable_list; list_ptr; list_ptr = list_ptr->Next) { + if (readers_intersect(variable, list_ptr->Item)) { + rc_variable_add_friend(list_ptr->Item, variable); + return; + } + } + rc_list_add(variable_list, rc_list(&variable->C->Pool, variable)); +} + +static void get_variable_pair_helper( + struct rc_list ** variable_list, + struct radeon_compiler * c, + struct rc_instruction * inst, + struct rc_pair_sub_instruction * sub_inst) +{ + struct rc_reader_data reader_data; + struct rc_variable * new_var; + rc_register_file file; + unsigned int writemask; + + if (sub_inst->Opcode == RC_OPCODE_NOP) { + return; + } + memset(&reader_data, 0, sizeof(struct rc_reader_data)); + rc_get_readers_sub(c, inst, sub_inst, &reader_data, NULL, NULL, NULL); + + if (reader_data.ReaderCount == 0) { + return; + } + + if (sub_inst->WriteMask) { + file = RC_FILE_TEMPORARY; + writemask = sub_inst->WriteMask; + } else if (sub_inst->OutputWriteMask) { + file = RC_FILE_OUTPUT; + writemask = sub_inst->OutputWriteMask; + } else { + writemask = 0; + file = RC_FILE_NONE; + } + new_var = rc_variable(c, file, sub_inst->DestIndex, writemask, + &reader_data); + get_variable_helper(variable_list, new_var); +} + +/** + * Generate a list of variables used by the shader program. Each instruction + * that writes to a register is considered a variable. The struct rc_variable + * data structure includes a list of readers and is essentially a + * definition-use chain. Any two variables that share a reader are considered + * "friends" and they are linked together via the Friend attribute. + */ +struct rc_list * rc_get_variables(struct radeon_compiler * c) +{ + struct rc_instruction * inst; + struct rc_list * variable_list = NULL; + + for (inst = c->Program.Instructions.Next; + inst != &c->Program.Instructions; + inst = inst->Next) { + struct rc_reader_data reader_data; + struct rc_variable * new_var; + memset(&reader_data, 0, sizeof(reader_data)); + + if (inst->Type == RC_INSTRUCTION_NORMAL) { + rc_get_readers(c, inst, &reader_data, NULL, NULL, NULL); + if (reader_data.ReaderCount == 0) { + continue; + } + new_var = rc_variable(c, inst->U.I.DstReg.File, + inst->U.I.DstReg.Index, + inst->U.I.DstReg.WriteMask, &reader_data); + get_variable_helper(&variable_list, new_var); + } else { + get_variable_pair_helper(&variable_list, c, inst, + &inst->U.P.RGB); + get_variable_pair_helper(&variable_list, c, inst, + &inst->U.P.Alpha); + } + } + + return variable_list; +} + +/** + * @return The bitwise or of the writemasks of a variable and all of its + * friends. + */ +unsigned int rc_variable_writemask_sum(struct rc_variable * var) +{ + unsigned int writemask = 0; + while(var) { + writemask |= var->Dst.WriteMask; + var = var->Friend; + } + return writemask; +} + +/* + * @return A list of readers for a variable and its friends. Readers + * that read from two different variable friends are only included once in + * this list. + */ +struct rc_list * rc_variable_readers_union(struct rc_variable * var) +{ + struct rc_list * list = NULL; + while (var) { + unsigned int i; + for (i = 0; i < var->ReaderCount; i++) { + struct rc_list * temp; + struct rc_reader * a = &var->Readers[i]; + unsigned int match = 0; + for (temp = list; temp; temp = temp->Next) { + struct rc_reader * b = temp->Item; + if (a->Inst->Type != b->Inst->Type) { + continue; + } + if (a->Inst->Type == RC_INSTRUCTION_NORMAL) { + if (a->U.I.Src == b->U.I.Src) { + match = 1; + break; + } + } + if (a->Inst->Type == RC_INSTRUCTION_PAIR) { + if (a->U.P.Arg == b->U.P.Arg + && a->U.P.Src == b->U.P.Src) { + match = 1; + break; + } + } + } + if (match) { + continue; + } + rc_list_add(&list, rc_list(&var->C->Pool, a)); + } + var = var->Friend; + } + return list; +} + +static unsigned int reader_equals_src( + struct rc_reader reader, + unsigned int src_type, + void * src) +{ + if (reader.Inst->Type != src_type) { + return 0; + } + if (src_type == RC_INSTRUCTION_NORMAL) { + return reader.U.I.Src == src; + } else { + return reader.U.P.Src == src; + } +} + +static unsigned int variable_writes_src( + struct rc_variable * var, + unsigned int src_type, + void * src) +{ + unsigned int i; + for (i = 0; i < var->ReaderCount; i++) { + if (reader_equals_src(var->Readers[i], src_type, src)) { + return 1; + } + } + return 0; +} + + +struct rc_list * rc_variable_list_get_writers( + struct rc_list * var_list, + unsigned int src_type, + void * src) +{ + struct rc_list * list_ptr; + struct rc_list * writer_list = NULL; + for (list_ptr = var_list; list_ptr; list_ptr = list_ptr->Next) { + struct rc_variable * var = list_ptr->Item; + if (variable_writes_src(var, src_type, src)) { + struct rc_variable * friend; + rc_list_add(&writer_list, rc_list(&var->C->Pool, var)); + for (friend = var->Friend; friend; + friend = friend->Friend) { + if (variable_writes_src(friend, src_type, src)) { + rc_list_add(&writer_list, + rc_list(&var->C->Pool, friend)); + } + } + /* Once we have indentifed the variable and its + * friends that write this source, we can stop + * stop searching, because we know know of the + * other variables in the list will write this source. + * If they did they would be friends of var. + */ + break; + } + } + return writer_list; +} + +void rc_variable_print(struct rc_variable * var) +{ + unsigned int i; + while (var) { + fprintf(stderr, "%u: TEMP[%u].%u: ", + var->Inst->IP, var->Dst.Index, var->Dst.WriteMask); + for (i = 0; i < 4; i++) { + fprintf(stderr, "chan %u: start=%u end=%u ", i, + var->Live[i].Start, var->Live[i].End); + } + fprintf(stderr, "%u readers\n", var->ReaderCount); + if (var->Friend) { + fprintf(stderr, "Friend: \n\t"); + } + var = var->Friend; + } +} diff --git a/src/gallium/drivers/r300/compiler/radeon_variable.h b/src/gallium/drivers/r300/compiler/radeon_variable.h new file mode 100644 index 00000000000..9427bee18a7 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/radeon_variable.h @@ -0,0 +1,89 @@ +/* + * Copyright 2011 Tom Stellard <[email protected]> + * + * 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. + * + */ + +#ifndef RADEON_VARIABLE_H +#define RADEON_VARIABLE_H + +#include "radeon_compiler.h" + +struct radeon_compiler; +struct rc_list; +struct rc_reader_data; +struct rc_readers; + +struct live_intervals { + int Start; + int End; + int Used; +}; + +struct rc_variable { + struct radeon_compiler * C; + struct rc_dst_register Dst; + + struct rc_instruction * Inst; + unsigned int ReaderCount; + struct rc_reader * Readers; + struct live_intervals Live[4]; + + /* A friend is a variable that shares a reader with another variable. + */ + struct rc_variable * Friend; +}; + +void rc_variable_change_dst( + struct rc_variable * var, + unsigned int new_index, + unsigned int new_writemask); + +void rc_variable_compute_live_intervals(struct rc_variable * var); + +void rc_variable_add_friend( + struct rc_variable * var, + struct rc_variable * friend); + +struct rc_variable * rc_variable( + struct radeon_compiler * c, + unsigned int DstFile, + unsigned int DstIndex, + unsigned int DstWriteMask, + struct rc_reader_data * reader_data); + +struct rc_list * rc_get_variables(struct radeon_compiler * c); + +unsigned int rc_variable_writemask_sum(struct rc_variable * var); + +struct rc_list * rc_variable_readers_union(struct rc_variable * var); + +struct rc_list * rc_variable_list_get_writers( + struct rc_list * var_list, + unsigned int src_type, + void * src); + +void rc_variable_print(struct rc_variable * var); + +#endif /* RADEON_VARIABLE_H */ diff --git a/src/gallium/drivers/r300/compiler/tests/.gitignore b/src/gallium/drivers/r300/compiler/tests/.gitignore new file mode 100644 index 00000000000..85672fed777 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/tests/.gitignore @@ -0,0 +1 @@ +radeon_compiler_util_tests diff --git a/src/gallium/drivers/r300/compiler/tests/Makefile b/src/gallium/drivers/r300/compiler/tests/Makefile new file mode 100644 index 00000000000..6eda34a2c00 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/tests/Makefile @@ -0,0 +1,53 @@ +TOP = ../../../../../.. +include $(TOP)/configs/current + +CFLAGS += -Wall -Werror + +### Basic defines ### +TESTS = radeon_compiler_util_tests + +TEST_SOURCES := $(TESTS:=.c) + +SHARED_SOURCES = \ + rc_test_helpers.c \ + unit_test.c + +C_SOURCES = $(SHARED_SOURCES) $(TEST_SOURCES) + +INCLUDES = \ + -I. \ + -I.. + +COMPILER_LIB = ../../libr300.a + +##### TARGETS ##### + +default: depend run_tests + +depend: $(C_SOURCES) + rm -f depend + touch depend + $(MKDEP) $(MKDEP_OPTIONS) $(INCLUDES) $^ 2> /dev/null + +# Remove .o and backup files +clean: + rm -f $(TESTS) depend depend.bak + +$(TESTS): $(TESTS:=.o) $(SHARED_SOURCES:.c=.o) $(COMPILER_LIB) + $(APP_CC) -o $@ $^ + +run_tests: $(TESTS) + @echo "RUNNING TESTS:" + @echo "" + $(foreach test, $^, @./$(test)) + +.PHONY: $(COMPILER_LIB) +$(COMPILER_LIB): + $(MAKE) -C ../.. + +##### RULES ##### +.c.o: + $(CC) -c $(INCLUDES) $(CFLAGS) $(LIBRARY_DEFINES) $< -o $@ + + +sinclude depend diff --git a/src/gallium/drivers/r300/compiler/tests/radeon_compiler_util_tests.c b/src/gallium/drivers/r300/compiler/tests/radeon_compiler_util_tests.c new file mode 100644 index 00000000000..a2e3f2ab2e5 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/tests/radeon_compiler_util_tests.c @@ -0,0 +1,76 @@ +#include <stdlib.h> +#include <string.h> +#include <sys/types.h> + +#include "radeon_compiler_util.h" +#include "radeon_program.h" + +#include "rc_test_helpers.h" +#include "unit_test.h" + +static void test_rc_inst_can_use_presub( + struct test_result * result, + int expected, + const char * add_str, + const char * replace_str) +{ + struct rc_instruction add_inst, replace_inst; + int ret; + + test_begin(result); + init_rc_normal_instruction(&add_inst, add_str); + init_rc_normal_instruction(&replace_inst, replace_str); + + ret = rc_inst_can_use_presub(&replace_inst, RC_PRESUB_ADD, 0, + &replace_inst.U.I.SrcReg[0], + &add_inst.U.I.SrcReg[0], &add_inst.U.I.SrcReg[1]); + + test_check(result, ret == expected); +} + +static void test_runner_rc_inst_can_use_presub(struct test_result * result) +{ + + /* This tests the case where the source being replace has the same + * register file and register index as another source register in the + * CMP instruction. A previous version of this function was ignoring + * all registers that shared the same file and index as the replacement + * register when counting the number of source selects. + * + * https://bugs.freedesktop.org/show_bug.cgi?id=36527 + */ + test_rc_inst_can_use_presub(result, 0, + "ADD temp[0].z, temp[6].__x_, const[1].__x_;", + "CMP temp[0].y, temp[0]._z__, const[0]._z__, temp[0]._y__;"); + + + /* Testing a random case that should fail + * + * https://bugs.freedesktop.org/show_bug.cgi?id=36527 + */ + test_rc_inst_can_use_presub(result, 0, + "ADD temp[3], temp[1], temp[2];", + "MAD temp[1], temp[0], const[0].xxxx, -temp[3];"); + + /* This tests the case where the arguments of the ADD + * instruction share the same register file and index. Normally, we + * would need only one source select for these two arguments, but since + * they will be part of a presubtract operation we need to use the two + * source selects that the presubtract instruction expects + * (src0 and src1). + * + * https://bugs.freedesktop.org/show_bug.cgi?id=36527 + */ + test_rc_inst_can_use_presub(result, 0, + "ADD temp[3].x, temp[0].x___, temp[0].x___;", + "MAD temp[0].xyz, temp[2].xyz_, -temp[3].xxx_, input[5].xyz_;"); +} + +int main(int argc, char ** argv) +{ + struct test tests[] = { + {"rc_inst_can_use_presub()", test_runner_rc_inst_can_use_presub}, + {NULL, NULL} + }; + run_tests(tests); +} diff --git a/src/gallium/drivers/r300/compiler/tests/rc_test_helpers.c b/src/gallium/drivers/r300/compiler/tests/rc_test_helpers.c new file mode 100644 index 00000000000..ca4738af54d --- /dev/null +++ b/src/gallium/drivers/r300/compiler/tests/rc_test_helpers.c @@ -0,0 +1,380 @@ +#include <errno.h> +#include <regex.h> +#include <stdlib.h> +#include <stdio.h> +#include <string.h> +#include <sys/types.h> + +#include "../radeon_compiler_util.h" +#include "../radeon_opcodes.h" +#include "../radeon_program.h" + +#include "rc_test_helpers.h" + +/* This file contains some helper functions for filling out the rc_instruction + * data structures. These functions take a string as input based on the format + * output by rc_program_print(). + */ + +#define VERBOSE 0 + +#define DBG(...) do { if (VERBOSE) fprintf(stderr, __VA_ARGS__); } while(0) + +#define REGEX_ERR_BUF_SIZE 50 + +struct match_info { + const char * String; + int Length; +}; + +static int match_length(regmatch_t * matches, int index) +{ + return matches[index].rm_eo - matches[index].rm_so; +} + +static int regex_helper( + const char * regex_str, + const char * search_str, + regmatch_t * matches, + int num_matches) +{ + char err_buf[REGEX_ERR_BUF_SIZE]; + regex_t regex; + int err_code; + unsigned int i; + + err_code = regcomp(®ex, regex_str, REG_EXTENDED); + if (err_code) { + regerror(err_code, ®ex, err_buf, REGEX_ERR_BUF_SIZE); + fprintf(stderr, "Failed to compile regex: %s\n", err_buf); + return 0; + } + + err_code = regexec(®ex, search_str, num_matches, matches, 0); + DBG("Search string: '%s'\n", search_str); + for (i = 0; i < num_matches; i++) { + DBG("Match %u start = %d end = %d\n", i, + matches[i].rm_so, matches[i].rm_eo); + } + if (err_code) { + regerror(err_code, ®ex, err_buf, REGEX_ERR_BUF_SIZE); + fprintf(stderr, "Failed to match regex: %s\n", err_buf); + return 0; + } + return 1; +} + +#define REGEX_SRC_MATCHES 6 + +struct src_tokens { + struct match_info Negate; + struct match_info Abs; + struct match_info File; + struct match_info Index; + struct match_info Swizzle; +}; + +/** + * Initialize the source register at index src_index for the instruction based + * on src_str. + * + * NOTE: Warning in init_rc_normal_instruction() applies to this function as + * well. + * + * @param src_str A string that represents the source register. The format for + * this string is the same that is output by rc_program_print. + * @return 1 On success, 0 on failure + */ +int init_rc_normal_src( + struct rc_instruction * inst, + unsigned int src_index, + const char * src_str) +{ + const char * regex_str = "(-*)(\\|*)([[:lower:]]*)\\[([[:digit:]])\\](\\.*[[:lower:]-]*)"; + regmatch_t matches[REGEX_SRC_MATCHES]; + struct src_tokens tokens; + struct rc_src_register * src_reg = &inst->U.I.SrcReg[src_index]; + unsigned int i; + + /* Execute the regex */ + if (!regex_helper(regex_str, src_str, matches, REGEX_SRC_MATCHES)) { + fprintf(stderr, "Failed to execute regex for src register.\n"); + return 0; + } + + /* Create Tokens */ + tokens.Negate.String = src_str + matches[1].rm_so; + tokens.Negate.Length = match_length(matches, 1); + tokens.Abs.String = src_str + matches[2].rm_so; + tokens.Abs.Length = match_length(matches, 2); + tokens.File.String = src_str + matches[3].rm_so; + tokens.File.Length = match_length(matches, 3); + tokens.Index.String = src_str + matches[4].rm_so; + tokens.Index.Length = match_length(matches, 4); + tokens.Swizzle.String = src_str + matches[5].rm_so; + tokens.Swizzle.Length = match_length(matches, 5); + + /* Negate */ + if (tokens.Negate.Length > 0) { + src_reg->Negate = RC_MASK_XYZW; + } + + /* Abs */ + if (tokens.Abs.Length > 0) { + src_reg->Abs = 1; + } + + /* File */ + if (!strncmp(tokens.File.String, "temp", tokens.File.Length)) { + src_reg->File = RC_FILE_TEMPORARY; + } else if (!strncmp(tokens.File.String, "input", tokens.File.Length)) { + src_reg->File = RC_FILE_INPUT; + } else if (!strncmp(tokens.File.String, "const", tokens.File.Length)) { + src_reg->File = RC_FILE_CONSTANT; + } else if (!strncmp(tokens.File.String, "none", tokens.File.Length)) { + src_reg->File = RC_FILE_NONE; + } + + /* Index */ + errno = 0; + src_reg->Index = strtol(tokens.Index.String, NULL, 10); + if (errno > 0) { + fprintf(stderr, "Could not convert src register index.\n"); + return 0; + } + + /* Swizzle */ + if (tokens.Swizzle.Length == 0) { + src_reg->Swizzle = RC_SWIZZLE_XYZW; + } else { + int str_index = 1; + src_reg->Swizzle = RC_MAKE_SWIZZLE_SMEAR(RC_SWIZZLE_UNUSED); + if (tokens.Swizzle.String[0] != '.') { + fprintf(stderr, "First char of swizzle is not valid.\n"); + return 0; + } + for (i = 0; i < 4; i++, str_index++) { + if (tokens.Swizzle.String[str_index] == '-') { + src_reg->Negate |= (1 << i); + str_index++; + } + switch(tokens.Swizzle.String[str_index]) { + case 'x': + SET_SWZ(src_reg->Swizzle, i, RC_SWIZZLE_X); + break; + case 'y': + SET_SWZ(src_reg->Swizzle, i, RC_SWIZZLE_Y); + break; + case 'z': + SET_SWZ(src_reg->Swizzle, i, RC_SWIZZLE_Z); + break; + case 'w': + SET_SWZ(src_reg->Swizzle, i, RC_SWIZZLE_W); + break; + case '1': + SET_SWZ(src_reg->Swizzle, i, RC_SWIZZLE_ONE); + break; + case '0': + SET_SWZ(src_reg->Swizzle, i, RC_SWIZZLE_ZERO); + break; + case 'H': + SET_SWZ(src_reg->Swizzle, i, RC_SWIZZLE_HALF); + break; + case '_': + SET_SWZ(src_reg->Swizzle, i, RC_SWIZZLE_UNUSED); + break; + default: + fprintf(stderr, "Unknown src register swizzle.\n"); + return 0; + } + } + } + DBG("File=%u index=%u swizzle=%x negate=%u abs=%u\n", + src_reg->File, src_reg->Index, src_reg->Swizzle, + src_reg->Negate, src_reg->Abs); + return 1; +} + +#define REGEX_DST_MATCHES 4 + +struct dst_tokens { + struct match_info File; + struct match_info Index; + struct match_info WriteMask; +}; + +/** + * Initialize the destination for the instruction based on dst_str. + * + * NOTE: Warning in init_rc_normal_instruction() applies to this function as + * well. + * + * @param dst_str A string that represents the destination register. The format + * for this string is the same that is output by rc_program_print. + * @return 1 On success, 0 on failure + */ +int init_rc_normal_dst( + struct rc_instruction * inst, + const char * dst_str) +{ + const char * regex_str = "([[:lower:]]*)\\[([[:digit:]]*)\\](\\.*[[:lower:]]*)"; + regmatch_t matches[REGEX_DST_MATCHES]; + struct dst_tokens tokens; + unsigned int i; + + /* Execute the regex */ + if (!regex_helper(regex_str, dst_str, matches, REGEX_DST_MATCHES)) { + fprintf(stderr, "Failed to execute regex for dst register.\n"); + return 0; + } + + /* Create Tokens */ + tokens.File.String = dst_str + matches[1].rm_so; + tokens.File.Length = match_length(matches, 1); + tokens.Index.String = dst_str + matches[2].rm_so; + tokens.Index.Length = match_length(matches, 2); + tokens.WriteMask.String = dst_str + matches[3].rm_so; + tokens.WriteMask.Length = match_length(matches, 3); + + /* File Type */ + if (!strncmp(tokens.File.String, "temp", tokens.File.Length)) { + inst->U.I.DstReg.File = RC_FILE_TEMPORARY; + } else if (!strncmp(tokens.File.String, "output", tokens.File.Length)) { + inst->U.I.DstReg.File = RC_FILE_OUTPUT; + } else { + fprintf(stderr, "Unknown dst register file type.\n"); + return 0; + } + + /* File Index */ + errno = 0; + inst->U.I.DstReg.Index = strtol(tokens.Index.String, NULL, 10); + + if (errno > 0) { + fprintf(stderr, "Could not convert dst register index\n"); + return 0; + } + + /* WriteMask */ + if (tokens.WriteMask.Length == 0) { + inst->U.I.DstReg.WriteMask = RC_MASK_XYZW; + } else { + /* The first character should be '.' */ + if (tokens.WriteMask.String[0] != '.') { + fprintf(stderr, "1st char of writemask is not valid.\n"); + return 0; + } + for (i = 1; i < tokens.WriteMask.Length; i++) { + switch(tokens.WriteMask.String[i]) { + case 'x': + inst->U.I.DstReg.WriteMask |= RC_MASK_X; + break; + case 'y': + inst->U.I.DstReg.WriteMask |= RC_MASK_Y; + break; + case 'z': + inst->U.I.DstReg.WriteMask |= RC_MASK_Z; + break; + case 'w': + inst->U.I.DstReg.WriteMask |= RC_MASK_W; + break; + default: + fprintf(stderr, "Unknown swizzle in writemask.\n"); + return 0; + } + } + } + DBG("Dst Reg File=%u Index=%d Writemask=%d\n", + inst->U.I.DstReg.File, + inst->U.I.DstReg.Index, + inst->U.I.DstReg.WriteMask); + return 1; +} + +#define REGEX_INST_MATCHES 7 + +struct inst_tokens { + struct match_info Opcode; + struct match_info Sat; + struct match_info Dst; + struct match_info Srcs[3]; +}; + +/** + * Initialize a normal instruction based on inst_str. + * + * WARNING: This function might not be able to handle every kind of format that + * rc_program_print() can output. If you are having problems with a + * particular string, you may need to add support for it to this functions. + * + * @param inst_str A string that represents the source register. The format for + * this string is the same that is output by rc_program_print. + * @return 1 On success, 0 on failure + */ +int init_rc_normal_instruction( + struct rc_instruction * inst, + const char * inst_str) +{ + const char * regex_str = "([[:upper:]]+)(_SAT)* ([^,]*)[, ]*([^,]*)[, ]*([^,]*)[, ]*([^;]*)"; + int i; + regmatch_t matches[REGEX_INST_MATCHES]; + struct inst_tokens tokens; + + /* Initialize inst */ + memset(inst, 0, sizeof(struct rc_instruction)); + inst->Type = RC_INSTRUCTION_NORMAL; + + /* Execute the regex */ + if (!regex_helper(regex_str, inst_str, matches, REGEX_INST_MATCHES)) { + return 0; + } + memset(&tokens, 0, sizeof(tokens)); + + /* Create Tokens */ + tokens.Opcode.String = inst_str + matches[1].rm_so; + tokens.Opcode.Length = match_length(matches, 1); + if (matches[2].rm_so > -1) { + tokens.Sat.String = inst_str + matches[2].rm_so; + tokens.Sat.Length = match_length(matches, 2); + } + + + /* Fill out the rest of the instruction. */ + for (i = 0; i < MAX_RC_OPCODE; i++) { + const struct rc_opcode_info * info = rc_get_opcode_info(i); + unsigned int first_src = 3; + unsigned int j; + if (strncmp(tokens.Opcode.String, info->Name, tokens.Opcode.Length)) { + continue; + } + inst->U.I.Opcode = info->Opcode; + if (info->HasDstReg) { + char * dst_str; + tokens.Dst.String = inst_str + matches[3].rm_so; + tokens.Dst.Length = match_length(matches, 3); + first_src++; + + dst_str = malloc(sizeof(char) * (tokens.Dst.Length + 1)); + strncpy(dst_str, tokens.Dst.String, tokens.Dst.Length); + dst_str[tokens.Dst.Length] = '\0'; + init_rc_normal_dst(inst, dst_str); + free(dst_str); + } + for (j = 0; j < info->NumSrcRegs; j++) { + char * src_str; + tokens.Srcs[j].String = + inst_str + matches[first_src + j].rm_so; + tokens.Srcs[j].Length = + match_length(matches, first_src + j); + + src_str = malloc(sizeof(char) * + (tokens.Srcs[j].Length + 1)); + strncpy(src_str, tokens.Srcs[j].String, + tokens.Srcs[j].Length); + src_str[tokens.Srcs[j].Length] = '\0'; + init_rc_normal_src(inst, j, src_str); + } + break; + } + return 1; +} diff --git a/src/gallium/drivers/r300/compiler/tests/rc_test_helpers.h b/src/gallium/drivers/r300/compiler/tests/rc_test_helpers.h new file mode 100644 index 00000000000..1a6bf9699ba --- /dev/null +++ b/src/gallium/drivers/r300/compiler/tests/rc_test_helpers.h @@ -0,0 +1,13 @@ + +int init_rc_normal_src( + struct rc_instruction * inst, + unsigned int src_index, + const char * src_str); + +int init_rc_normal_dst( + struct rc_instruction * inst, + const char * dst_str); + +int init_rc_normal_instruction( + struct rc_instruction * inst, + const char * inst_str); diff --git a/src/gallium/drivers/r300/compiler/tests/unit_test.c b/src/gallium/drivers/r300/compiler/tests/unit_test.c new file mode 100644 index 00000000000..266f3365c58 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/tests/unit_test.c @@ -0,0 +1,35 @@ +#include <stdlib.h> +#include <stdio.h> +#include <string.h> + +#include "unit_test.h" + +void run_tests(struct test tests[]) +{ + int i; + for (i = 0; tests[i].name; i++) { + printf("Test %s\n", tests[i].name); + memset(&tests[i].result, 0, sizeof(tests[i].result)); + tests[i].test_func(&tests[i].result); + printf("Test %s (%d/%d) pass\n", tests[i].name, + tests[i].result.pass, tests[i].result.test_count); + } +} + +void test_begin(struct test_result * result) +{ + result->test_count++; +} + +void test_check(struct test_result * result, int cond) +{ + printf("Subtest %u -> ", result->test_count); + if (cond) { + result->pass++; + printf("Pass"); + } else { + result->fail++; + printf("Fail"); + } + printf("\n"); +} diff --git a/src/gallium/drivers/r300/compiler/tests/unit_test.h b/src/gallium/drivers/r300/compiler/tests/unit_test.h new file mode 100644 index 00000000000..441e8b655a5 --- /dev/null +++ b/src/gallium/drivers/r300/compiler/tests/unit_test.h @@ -0,0 +1,17 @@ + +struct test_result { + unsigned int test_count; + unsigned int pass; + unsigned int fail; +}; + +struct test { + const char * name; + void (*test_func)(struct test_result * result); + struct test_result result; +}; + +void run_tests(struct test tests[]); + +void test_begin(struct test_result * result); +void test_check(struct test_result * result, int cond); |