From 55b2fe1047b37d0d86641a252e1c745111030393 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 2 Jun 2008 23:52:58 +1000 Subject: nv50: drop the majority of the old shader code, reimplement, only MOV so far. --- src/gallium/drivers/nv50/nv50_program.c | 411 ++++++++++++-------------------- 1 file changed, 146 insertions(+), 265 deletions(-) (limited to 'src/gallium') diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 97e3eab906..aa848faa49 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -10,13 +10,8 @@ #include "nv50_context.h" #include "nv50_state.h" -#define OP_MOV 0x001 -#define OP_INTERP 0x008 -#define OP_RCP 0x009 -#define OP_ADD 0x00b -#define OP_MUL 0x00c -#define OP_MAD 0x00e #define NV50_SU_MAX_TEMP 64 +#define TX_FRAGPROG 0 struct nv50_reg { enum { @@ -106,181 +101,6 @@ free_temp(struct nv50_pc *pc, struct nv50_reg *r) } } -#if 0 -static struct nv50_reg * -constant(struct nv50_pc *pc, int pipe, int c, float v) -{ - struct nv50_reg *r = CALLOC_STRUCT(nv50_reg); - struct nv50_program *p = pc->p; - struct nv50_program_data *pd; - int idx; - - if (pipe >= 0) { - for (idx = 0; idx < p->nr_consts; idx++) { - if (p->consts[idx].index == pipe) - return nv40_sr(NV40SR_CONST, idx); - } - } - - idx = p->nr_consts++; - p->consts = realloc(p->consts, sizeof(*pd) * p->nr_consts); - pd = &p->consts[idx]; - - pd->index = pipe; - pd->component = c; - pd->value = v; - return nv40_sr(NV40SR_CONST, idx); -} -#endif - -static void -emit(struct nv50_pc *pc, unsigned op, struct nv50_reg *dst, - struct nv50_reg *src0, struct nv50_reg *src1, struct nv50_reg *src2) -{ - struct nv50_program *p = pc->p; - struct nv50_reg *tmp0 = NULL, *tmp = NULL, *tmp2 = NULL; - unsigned inst[2] = { 0, 0 }; - - /* Grr.. Fun restrictions on where attribs can be sourced from.. */ - if (src0 && (src0->type == P_CONST || src0->type == P_IMMD) && - (op == OP_MUL || op == OP_MAD)) { - tmp = src1; - src1 = src0; - src0 = tmp; - tmp = NULL; - } - - if (src1 && src1->type == P_ATTR) { - tmp = alloc_temp(pc, dst); - tmp->neg = src1->neg; src1->neg = 0; - emit(pc, 1, tmp, src1, NULL, NULL); - src1 = tmp; - } - - if (src2 && src2->type == P_ATTR) { - tmp2 = alloc_temp(pc, dst); - tmp2->neg = src2->neg; src2->neg = 0; - emit(pc, 1, tmp2, src2, NULL, NULL); - src2 = tmp2; - } - - /* Get this out of the way first. What type of opcode do we - * want/need to build? - */ - if ((op & 0x3f0) || dst->type == P_RESULT || - (src0 && src0->type == P_ATTR) || src1 || src2) - inst[0] |= 0x00000001; - - if (inst[0] & 0x00000001) { - inst[0] |= ((op & 0xf) << 28); - inst[1] |= ((op >> 4) << 26); - - alloc_reg(pc, dst); - if (dst->type == P_RESULT) - inst[1] |= 0x00000008; - inst[0] |= (dst->hw << 2); - - if (src0) { - if (src0->type == P_ATTR) - inst[1] |= 0x00200000; - else - if (src0->type == P_CONST || src0->type == P_IMMD) - assert(0); - alloc_reg(pc, src0); - inst[0] |= (src0->hw << 9); - } - - if (src1) { - if (src1->type == P_CONST || src1->type == P_IMMD) { - if (src1->type == P_IMMD) - inst[1] |= (NV50_CB_PMISC << 22); - else - inst[1] |= (NV50_CB_PVP << 22); - inst[0] |= 0x00800000; /* src1 is const */ - /*XXX: does src1 come from "src2" now? */ - alloc_reg(pc, src1); - inst[0] |= (src1->hw << 16); - } else { - alloc_reg(pc, src1); - if (op == OP_MUL || op == OP_MAD) - inst[0] |= (src1->hw << 16); - else - inst[1] |= (src1->hw << 14); - } - } else { - inst[1] |= 0x0003c000; /*XXX FIXME */ - } - - if (src2) { - if (src2->type == P_CONST || src2->type == P_IMMD) { - if (src2->type == P_IMMD) - inst[1] |= (NV50_CB_PMISC << 22); - else - inst[1] |= (NV50_CB_PVP << 22); - inst[0] |= 0x01000000; /* src2 is const */ - inst[1] |= (src2->hw << 14); - } else { - alloc_reg(pc, src2); - if (inst[0] & 0x00800000 || op == OP_MAD) - inst[1] |= (src2->hw << 14); - else - inst[0] |= (src2->hw << 16); - } - } - - /*XXX: FIXME */ - switch (op) { - case OP_ADD: - case OP_MUL: - case OP_RCP: - case OP_MAD: - /* 0x04000000 negates arg0 */ - /* 0x08000000 negates arg1 */ - /*XXX: true for !0xb also ? */ - if (src0 && src0->neg) - inst[1] |= 0x04000000; - if (src1 && src1->neg) - inst[1] |= 0x08000000; - inst[1] |= 0x00000780; - break; - default: - /* 0x04000000 == arg0 32 bit, otherwise 16 bit */ - inst[1] |= 0x04000780; - break; - } - } else { - inst[0] |= ((op & 0xf) << 28); - - alloc_reg(pc, dst); - inst[0] |= (dst->hw << 2); - - if (src0) { - alloc_reg(pc, src0); - inst[0] |= (src0->hw << 9); - } - - /*XXX: NFI if this even works - probably not.. */ - if (src1) { - alloc_reg(pc, src1); - inst[0] |= (src1->hw << 16); - } - } - - if (tmp0) free_temp(pc, tmp0); - if (tmp) free_temp(pc, tmp); - if (tmp2) free_temp(pc, tmp2); - - if (inst[0] & 1) { - p->insns_nr += 2; - p->insns = realloc(p->insns, sizeof(unsigned) * p->insns_nr); - memcpy(p->insns + (p->insns_nr - 2), inst, sizeof(unsigned)*2); - } else { - p->insns_nr += 1; - p->insns = realloc(p->insns, sizeof(unsigned) * p->insns_nr); - memcpy(p->insns + (p->insns_nr - 1), inst, sizeof(unsigned)); - } -} - static struct nv50_reg * tgsi_dst(struct nv50_pc *pc, int c, const struct tgsi_full_dst_register *dst) { @@ -327,11 +147,147 @@ tgsi_src(struct nv50_pc *pc, int c, const struct tgsi_full_src_register *src) return NULL; } +static void +emit(struct nv50_pc *pc, unsigned *inst) +{ + struct nv50_program *p = pc->p; + + if (inst[0] & 1) { + p->insns_nr += 2; + p->insns = realloc(p->insns, sizeof(unsigned) * p->insns_nr); + memcpy(p->insns + (p->insns_nr - 2), inst, sizeof(unsigned)*2); + } else { + p->insns_nr += 1; + p->insns = realloc(p->insns, sizeof(unsigned) * p->insns_nr); + memcpy(p->insns + (p->insns_nr - 1), inst, sizeof(unsigned)); + } +} + +static INLINE void set_long(struct nv50_pc *, unsigned *); + +static boolean +is_long(unsigned *inst) +{ + if (inst[0] & 1) + return TRUE; + return FALSE; +} + +static boolean +is_immd(unsigned *inst) +{ + if (is_long(inst) && (inst[1] & 3) == 3) + return TRUE; + return FALSE; +} + +static INLINE void +set_pred(struct nv50_pc *pc, unsigned pred, unsigned idx, unsigned *inst) +{ + set_long(pc, inst); + inst[1] &= ~((0x1f << 7) | (0x3 << 12)); + inst[1] |= (pred << 7) | (idx << 12); +} + +static INLINE void +set_pred_wr(struct nv50_pc *pc, unsigned on, unsigned idx, unsigned *inst) +{ + set_long(pc, inst); + inst[1] &= ~((0x3 << 4) | (1 << 6)); + inst[1] |= (idx << 4) | (on << 6); +} + +static INLINE void +set_long(struct nv50_pc *pc, unsigned *inst) +{ + if (is_long(inst)) + return; + + inst[0] |= 1; + set_pred(pc, 0xf, 0, inst); + set_pred_wr(pc, 0, 0, inst); +} + +static INLINE void +set_dst(struct nv50_pc *pc, struct nv50_reg *dst, unsigned *inst) +{ + if (dst->type == P_RESULT) { + set_long(pc, inst); + inst[1] |= 0x00000008; + } + + alloc_reg(pc, dst); + inst[0] |= (dst->hw << 2); +} + +static INLINE void +set_immd(struct nv50_pc *pc, struct nv50_reg *imm, unsigned *inst) +{ + unsigned val = fui(pc->immd_buf[imm->hw]); /* XXX */ + + set_long(pc, inst); + /*XXX: can't be predicated - bits overlap.. catch cases where both + * are required and avoid them. */ + set_pred(pc, 0, 0, inst); + set_pred_wr(pc, 0, 0, inst); + + inst[1] |= 0x00000002 | 0x00000001; + inst[0] |= (val & 0x3f) << 16; + inst[1] |= (val >> 6) << 2; +} + +static void +emit_mov(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) +{ + unsigned inst[2] = { 0, 0 }; + int i; + + inst[0] |= 0x10000000; + + set_dst(pc, dst, inst); + + if (dst->type != P_RESULT && src->type == P_IMMD) { + set_immd(pc, src, inst); + /*XXX: 32-bit, but steals part of "half" reg space - need to + * catch and handle this case if/when we do half-regs + */ + inst[0] |= 0x00008000; + } else + if (src->type == P_IMMD || src->type == P_CONST) { + set_long(pc, inst); + if (src->type == P_IMMD) + inst[1] |= (NV50_CB_PMISC << 22); + else + inst[1] |= (NV50_CB_PVP << 22); + inst[0] |= (src->hw << 9); + inst[1] |= 0x20000000; /* src0 const? */ + } else { + if (src->type == P_ATTR) { + set_long(pc, inst); + inst[1] |= 0x00200000; + } + + alloc_reg(pc, src); + inst[0] |= (src->hw << 9); + } + + /* We really should support "half" instructions here at some point, + * but I don't feel confident enough about them yet. + */ + set_long(pc, inst); + if (is_long(inst) && !is_immd(inst)) { + inst[1] |= 0x04000000; /* 32-bit */ + inst[1] |= 0x0003c000; /* "subsubop" 0xf == mov */ + } + + emit(pc, inst); +} + static boolean nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) { const struct tgsi_full_instruction *inst = &tok->FullInstruction; - struct nv50_reg *dst[4], *src[3][4], *none = NULL, *tmp; + struct nv50_reg *dst[4], *src[3][4]; unsigned mask; int i, c; @@ -352,85 +308,9 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) } switch (inst->Instruction.Opcode) { - case TGSI_OPCODE_ADD: - for (c = 0; c < 4; c++) { - if (mask & (1 << c)) { - emit(pc, OP_ADD, dst[c], - src[0][c], src[1][c], none); - } - } - break; - case TGSI_OPCODE_DP3: - tmp = alloc_temp(pc, NULL); - emit(pc, OP_MUL, tmp, src[0][0], src[1][0], NULL); - emit(pc, OP_MAD, tmp, src[0][1], src[1][1], tmp); - emit(pc, OP_MAD, tmp, src[0][2], src[1][2], tmp); - for (c = 0; c < 4; c++) { - if (mask & (1 << c)) - emit(pc, OP_MOV, dst[c], tmp, none, none); - } - free_temp(pc, tmp); - break; - case TGSI_OPCODE_DP4: - tmp = alloc_temp(pc, NULL); - emit(pc, OP_MUL, tmp, src[0][0], src[1][0], NULL); - emit(pc, OP_MAD, tmp, src[0][1], src[1][1], tmp); - emit(pc, OP_MAD, tmp, src[0][2], src[1][2], tmp); - emit(pc, OP_MAD, tmp, src[0][3], src[1][3], tmp); - for (c = 0; c < 4; c++) { - if (mask & (1 << c)) - emit(pc, OP_MOV, dst[c], tmp, none, none); - } - free_temp(pc, tmp); - break; - case TGSI_OPCODE_DPH: - tmp = alloc_temp(pc, NULL); - emit(pc, OP_MUL, tmp, src[0][0], src[1][0], NULL); - emit(pc, OP_MAD, tmp, src[0][1], src[1][1], tmp); - emit(pc, OP_MAD, tmp, src[0][2], src[1][2], tmp); - emit(pc, OP_ADD, tmp, src[1][3], tmp, NULL); - for (c = 0; c < 4; c++) { - if (mask & (1 << c)) - emit(pc, OP_MOV, dst[c], tmp, none, none); - } - free_temp(pc, tmp); - break; - case TGSI_OPCODE_MAD: - for (c = 0; c < 4; c++) { - if (mask & (1 << c)) - emit(pc, OP_MAD, dst[c], - src[0][c], src[1][c], src[2][c]); - } - break; case TGSI_OPCODE_MOV: - for (c = 0; c < 4; c++) { - if (mask & (1 << c)) - emit(pc, OP_MOV, dst[c], src[0][c], none, none); - } - break; - case TGSI_OPCODE_MUL: - for (c = 0; c < 4; c++) { - if (mask & (1 << c)) - emit(pc, OP_MUL, dst[c], - src[0][c], src[1][c], none); - } - break; - case TGSI_OPCODE_RCP: - for (c = 0; c < 4; c++) { - if (mask & (1 << c)) - emit(pc, OP_RCP, dst[c], - src[0][c], none, none); - } - break; - case TGSI_OPCODE_SUB: - for (c = 0; c < 4; c++) { - if (mask & (1 << c)) { - src[1][c]->neg = 1; - emit(pc, OP_ADD, dst[c], - src[0][c], src[1][c], none); - src[1][c]->neg = 0; - } - } + for (c = 0; c < 4; c++) + emit_mov(pc, dst[c], src[0][c]); break; case TGSI_OPCODE_END: break; @@ -648,6 +528,7 @@ out_cleanup: static void nv50_program_validate(struct nv50_context *nv50, struct nv50_program *p) { +#if TX_FRAGPROG == 0 struct tgsi_parse_context pc; tgsi_parse_init(&pc, p->pipe.tokens); @@ -665,6 +546,7 @@ nv50_program_validate(struct nv50_context *nv50, struct nv50_program *p) p->insns[7] = 0x00020781; } else if (pc.FullHeader.Processor.Processor == TGSI_PROCESSOR_VERTEX) { +#endif int i; if (nv50_program_tx(p) == FALSE) @@ -673,6 +555,7 @@ nv50_program_validate(struct nv50_context *nv50, struct nv50_program *p) for (i = 0; i < p->insns_nr; i++) NOUVEAU_ERR("%d 0x%08x\n", i, p->insns[i]); +#if TX_FRAGPROG == 0 } else { NOUVEAU_ERR("invalid TGSI processor\n"); tgsi_parse_free(&pc); @@ -680,6 +563,7 @@ nv50_program_validate(struct nv50_context *nv50, struct nv50_program *p) } tgsi_parse_free(&pc); +#endif p->translated = TRUE; } @@ -758,11 +642,8 @@ nv50_fragprog_validate(struct nv50_context *nv50) assert(0); } - if (!p->buffer) - p->buffer = ws->buffer_create(ws, 0x100, 0, p->insns_nr * 4); - map = ws->buffer_map(ws, p->buffer, PIPE_BUFFER_USAGE_CPU_WRITE); - memcpy(map, p->insns, p->insns_nr * 4); - ws->buffer_unmap(ws, p->buffer); + nv50_program_validate_data(nv50, p); + nv50_program_validate_code(nv50, p); so = so_new(3, 2); so_method(so, tesla, NV50TCL_FP_ADDRESS_HIGH, 2); -- cgit v1.2.3