diff options
author | Chia-I Wu <[email protected]> | 2010-01-12 11:25:02 +0800 |
---|---|---|
committer | Chia-I Wu <[email protected]> | 2010-01-12 11:25:02 +0800 |
commit | 562c127693200822f04a145db50add1be2425d7b (patch) | |
tree | 9441774fb212b17ddf2a364f06abc43f166cc00b /src/gallium/drivers/nv30 | |
parent | e5d351dcfde58777162552cf5cd2a9cd8299f4cd (diff) | |
parent | 077d6dd7508af88509dd0499c5dfbdaa186b4015 (diff) |
Merge branch 'master' into opengl-es-v2
Conflicts:
src/mesa/main/dd.h
Diffstat (limited to 'src/gallium/drivers/nv30')
-rw-r--r-- | src/gallium/drivers/nv30/nv30_context.c | 24 | ||||
-rw-r--r-- | src/gallium/drivers/nv30/nv30_context.h | 10 | ||||
-rw-r--r-- | src/gallium/drivers/nv30/nv30_fragprog.c | 131 | ||||
-rw-r--r-- | src/gallium/drivers/nv30/nv30_fragtex.c | 14 | ||||
-rw-r--r-- | src/gallium/drivers/nv30/nv30_miptree.c | 62 | ||||
-rw-r--r-- | src/gallium/drivers/nv30/nv30_query.c | 20 | ||||
-rw-r--r-- | src/gallium/drivers/nv30/nv30_screen.c | 17 | ||||
-rw-r--r-- | src/gallium/drivers/nv30/nv30_state.c | 21 | ||||
-rw-r--r-- | src/gallium/drivers/nv30/nv30_state.h | 1 | ||||
-rw-r--r-- | src/gallium/drivers/nv30/nv30_state_blend.c | 2 | ||||
-rw-r--r-- | src/gallium/drivers/nv30/nv30_state_emit.c | 10 | ||||
-rw-r--r-- | src/gallium/drivers/nv30/nv30_state_fb.c | 2 | ||||
-rw-r--r-- | src/gallium/drivers/nv30/nv30_state_scissor.c | 2 | ||||
-rw-r--r-- | src/gallium/drivers/nv30/nv30_state_stipple.c | 4 | ||||
-rw-r--r-- | src/gallium/drivers/nv30/nv30_state_viewport.c | 2 | ||||
-rw-r--r-- | src/gallium/drivers/nv30/nv30_transfer.c | 39 | ||||
-rw-r--r-- | src/gallium/drivers/nv30/nv30_vbo.c | 133 | ||||
-rw-r--r-- | src/gallium/drivers/nv30/nv30_vertprog.c | 91 |
18 files changed, 328 insertions, 257 deletions
diff --git a/src/gallium/drivers/nv30/nv30_context.c b/src/gallium/drivers/nv30/nv30_context.c index d8300fd69f6..54572e9ab3a 100644 --- a/src/gallium/drivers/nv30/nv30_context.c +++ b/src/gallium/drivers/nv30/nv30_context.c @@ -10,21 +10,32 @@ nv30_flush(struct pipe_context *pipe, unsigned flags, struct pipe_fence_handle **fence) { struct nv30_context *nv30 = nv30_context(pipe); + struct nv30_screen *screen = nv30->screen; + struct nouveau_channel *chan = screen->base.channel; + struct nouveau_grobj *rankine = screen->rankine; if (flags & PIPE_FLUSH_TEXTURE_CACHE) { - BEGIN_RING(rankine, 0x1fd8, 1); - OUT_RING (2); - BEGIN_RING(rankine, 0x1fd8, 1); - OUT_RING (1); + BEGIN_RING(chan, rankine, 0x1fd8, 1); + OUT_RING (chan, 2); + BEGIN_RING(chan, rankine, 0x1fd8, 1); + OUT_RING (chan, 1); } - FIRE_RING(fence); + FIRE_RING(chan); + if (fence) + *fence = NULL; } static void nv30_destroy(struct pipe_context *pipe) { struct nv30_context *nv30 = nv30_context(pipe); + unsigned i; + + for (i = 0; i < NV30_STATE_MAX; i++) { + if (nv30->state.hw[i]) + so_ref(NULL, &nv30->state.hw[i]); + } if (nv30->draw) draw_destroy(nv30->draw); @@ -58,6 +69,9 @@ nv30_create(struct pipe_screen *pscreen, unsigned pctx_id) nv30->pipe.is_texture_referenced = nouveau_is_texture_referenced; nv30->pipe.is_buffer_referenced = nouveau_is_buffer_referenced; + screen->base.channel->user_private = nv30; + screen->base.channel->flush_notify = nv30_state_flush_notify; + nv30_init_query_functions(nv30); nv30_init_surface_functions(nv30); nv30_init_state_functions(nv30); diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index 8d49366dfcb..e59449287b5 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -14,10 +14,6 @@ #include "nouveau/nouveau_winsys.h" #include "nouveau/nouveau_gldefs.h" #include "nouveau/nouveau_context.h" - -#define NOUVEAU_PUSH_CONTEXT(ctx) \ - struct nv30_screen *ctx = nv30->screen -#include "nouveau/nouveau_push.h" #include "nouveau/nouveau_stateobj.h" #include "nv30_state.h" @@ -144,7 +140,6 @@ struct nv30_context { unsigned vtxbuf_nr; struct pipe_vertex_element vtxelt[PIPE_MAX_ATTRIBS]; unsigned vtxelt_nr; - const unsigned *edgeflags; }; static INLINE struct nv30_context * @@ -184,6 +179,7 @@ extern void nv30_fragtex_bind(struct nv30_context *); /* nv30_state.c and friends */ extern boolean nv30_state_validate(struct nv30_context *nv30); extern void nv30_state_emit(struct nv30_context *nv30); +extern void nv30_state_flush_notify(struct nouveau_channel *chan); extern struct nv30_state_entry nv30_state_rasterizer; extern struct nv30_state_entry nv30_state_scissor; extern struct nv30_state_entry nv30_state_stipple; @@ -198,9 +194,9 @@ extern struct nv30_state_entry nv30_state_fragtex; extern struct nv30_state_entry nv30_state_vbo; /* nv30_vbo.c */ -extern boolean nv30_draw_arrays(struct pipe_context *, unsigned mode, +extern void nv30_draw_arrays(struct pipe_context *, unsigned mode, unsigned start, unsigned count); -extern boolean nv30_draw_elements(struct pipe_context *pipe, +extern void nv30_draw_elements(struct pipe_context *pipe, struct pipe_buffer *indexBuffer, unsigned indexSize, unsigned mode, unsigned start, diff --git a/src/gallium/drivers/nv30/nv30_fragprog.c b/src/gallium/drivers/nv30/nv30_fragprog.c index cc0385426c1..2d565cb631b 100644 --- a/src/gallium/drivers/nv30/nv30_fragprog.c +++ b/src/gallium/drivers/nv30/nv30_fragprog.c @@ -4,6 +4,7 @@ #include "pipe/p_inlines.h" #include "pipe/p_shader_tokens.h" +#include "tgsi/tgsi_dump.h" #include "tgsi/tgsi_parse.h" #include "tgsi/tgsi_util.h" @@ -131,7 +132,7 @@ emit_src(struct nv30_fpc *fpc, int pos, struct nv30_sreg src) sizeof(uint32_t) * 4); } - sr |= (NV30_FP_REG_TYPE_CONST << NV30_FP_REG_TYPE_SHIFT); + sr |= (NV30_FP_REG_TYPE_CONST << NV30_FP_REG_TYPE_SHIFT); break; case NV30SR_NONE: sr |= (NV30_FP_REG_TYPE_INPUT << NV30_FP_REG_TYPE_SHIFT); @@ -236,20 +237,20 @@ tgsi_src(struct nv30_fpc *fpc, const struct tgsi_full_src_register *fsrc) { struct nv30_sreg src; - switch (fsrc->SrcRegister.File) { + switch (fsrc->Register.File) { case TGSI_FILE_INPUT: src = nv30_sr(NV30SR_INPUT, - fpc->attrib_map[fsrc->SrcRegister.Index]); + fpc->attrib_map[fsrc->Register.Index]); break; case TGSI_FILE_CONSTANT: - src = constant(fpc, fsrc->SrcRegister.Index, NULL); + src = constant(fpc, fsrc->Register.Index, NULL); break; case TGSI_FILE_IMMEDIATE: - assert(fsrc->SrcRegister.Index < fpc->nr_imm); - src = fpc->imm[fsrc->SrcRegister.Index]; + assert(fsrc->Register.Index < fpc->nr_imm); + src = fpc->imm[fsrc->Register.Index]; break; case TGSI_FILE_TEMPORARY: - src = nv30_sr(NV30SR_TEMP, fsrc->SrcRegister.Index + 1); + src = nv30_sr(NV30SR_TEMP, fsrc->Register.Index + 1); if (fpc->high_temp < src.index) fpc->high_temp = src.index; break; @@ -257,7 +258,7 @@ tgsi_src(struct nv30_fpc *fpc, const struct tgsi_full_src_register *fsrc) * Luckily fragprog results are just temp regs.. */ case TGSI_FILE_OUTPUT: - if (fsrc->SrcRegister.Index == fpc->colour_id) + if (fsrc->Register.Index == fpc->colour_id) return nv30_sr(NV30SR_OUTPUT, 0); else return nv30_sr(NV30SR_OUTPUT, 1); @@ -267,12 +268,12 @@ tgsi_src(struct nv30_fpc *fpc, const struct tgsi_full_src_register *fsrc) break; } - src.abs = fsrc->SrcRegisterExtMod.Absolute; - src.negate = fsrc->SrcRegister.Negate; - src.swz[0] = fsrc->SrcRegister.SwizzleX; - src.swz[1] = fsrc->SrcRegister.SwizzleY; - src.swz[2] = fsrc->SrcRegister.SwizzleZ; - src.swz[3] = fsrc->SrcRegister.SwizzleW; + src.abs = fsrc->Register.Absolute; + src.negate = fsrc->Register.Negate; + src.swz[0] = fsrc->Register.SwizzleX; + src.swz[1] = fsrc->Register.SwizzleY; + src.swz[2] = fsrc->Register.SwizzleZ; + src.swz[3] = fsrc->Register.SwizzleW; return src; } @@ -280,22 +281,22 @@ static INLINE struct nv30_sreg tgsi_dst(struct nv30_fpc *fpc, const struct tgsi_full_dst_register *fdst) { int idx; - switch (fdst->DstRegister.File) { + switch (fdst->Register.File) { case TGSI_FILE_OUTPUT: - if (fdst->DstRegister.Index == fpc->colour_id) + if (fdst->Register.Index == fpc->colour_id) return nv30_sr(NV30SR_OUTPUT, 0); else return nv30_sr(NV30SR_OUTPUT, 1); break; case TGSI_FILE_TEMPORARY: - idx = fdst->DstRegister.Index + 1; + idx = fdst->Register.Index + 1; if (fpc->high_temp < idx) fpc->high_temp = idx; return nv30_sr(NV30SR_TEMP, idx); case TGSI_FILE_NULL: return nv30_sr(NV30SR_NONE, 0); default: - NOUVEAU_ERR("bad dst file %d\n", fdst->DstRegister.File); + NOUVEAU_ERR("bad dst file %d\n", fdst->Register.File); return nv30_sr(NV30SR_NONE, 0); } } @@ -362,8 +363,8 @@ nv30_fragprog_parse_instruction(struct nv30_fpc *fpc, for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { const struct tgsi_full_src_register *fsrc; - fsrc = &finst->FullSrcRegisters[i]; - if (fsrc->SrcRegister.File == TGSI_FILE_TEMPORARY) { + fsrc = &finst->Src[i]; + if (fsrc->Register.File == TGSI_FILE_TEMPORARY) { src[i] = tgsi_src(fpc, fsrc); } } @@ -371,9 +372,9 @@ nv30_fragprog_parse_instruction(struct nv30_fpc *fpc, for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { const struct tgsi_full_src_register *fsrc; - fsrc = &finst->FullSrcRegisters[i]; + fsrc = &finst->Src[i]; - switch (fsrc->SrcRegister.File) { + switch (fsrc->Register.File) { case TGSI_FILE_INPUT: case TGSI_FILE_CONSTANT: case TGSI_FILE_TEMPORARY: @@ -384,14 +385,14 @@ nv30_fragprog_parse_instruction(struct nv30_fpc *fpc, break; } - switch (fsrc->SrcRegister.File) { + switch (fsrc->Register.File) { case TGSI_FILE_INPUT: - if (ai == -1 || ai == fsrc->SrcRegister.Index) { - ai = fsrc->SrcRegister.Index; + if (ai == -1 || ai == fsrc->Register.Index) { + ai = fsrc->Register.Index; src[i] = tgsi_src(fpc, fsrc); } else { NOUVEAU_MSG("extra src attr %d\n", - fsrc->SrcRegister.Index); + fsrc->Register.Index); src[i] = temp(fpc); arith(fpc, 0, MOV, src[i], MASK_ALL, tgsi_src(fpc, fsrc), none, none); @@ -399,8 +400,8 @@ nv30_fragprog_parse_instruction(struct nv30_fpc *fpc, break; case TGSI_FILE_CONSTANT: case TGSI_FILE_IMMEDIATE: - if (ci == -1 || ci == fsrc->SrcRegister.Index) { - ci = fsrc->SrcRegister.Index; + if (ci == -1 || ci == fsrc->Register.Index) { + ci = fsrc->Register.Index; src[i] = tgsi_src(fpc, fsrc); } else { src[i] = temp(fpc); @@ -412,7 +413,7 @@ nv30_fragprog_parse_instruction(struct nv30_fpc *fpc, /* handled above */ break; case TGSI_FILE_SAMPLER: - unit = fsrc->SrcRegister.Index; + unit = fsrc->Register.Index; break; case TGSI_FILE_OUTPUT: break; @@ -422,8 +423,8 @@ nv30_fragprog_parse_instruction(struct nv30_fpc *fpc, } } - dst = tgsi_dst(fpc, &finst->FullDstRegisters[0]); - mask = tgsi_mask(finst->FullDstRegisters[0].DstRegister.WriteMask); + dst = tgsi_dst(fpc, &finst->Dst[0]); + mask = tgsi_mask(finst->Dst[0].Register.WriteMask); sat = (finst->Instruction.Saturate == TGSI_SAT_ZERO_ONE); switch (finst->Instruction.Opcode) { @@ -434,10 +435,11 @@ nv30_fragprog_parse_instruction(struct nv30_fpc *fpc, arith(fpc, sat, ADD, dst, mask, src[0], src[1], none); break; case TGSI_OPCODE_CMP: - tmp = temp(fpc); - arith(fpc, sat, MOV, dst, mask, src[2], none, none); + tmp = nv30_sr(NV30SR_NONE, 0); tmp.cc_update = 1; arith(fpc, 0, MOV, tmp, 0xf, src[0], none, none); + dst.cc_test = NV30_VP_INST_COND_GE; + arith(fpc, sat, MOV, dst, mask, src[2], none, none); dst.cc_test = NV30_VP_INST_COND_LT; arith(fpc, sat, MOV, dst, mask, src[1], none, none); break; @@ -516,13 +518,28 @@ nv30_fragprog_parse_instruction(struct nv30_fpc *fpc, arith(fpc, sat, RSQ, dst, mask, abs(swz(src[0], X, X, X, X)), none, none); break; case TGSI_OPCODE_SCS: - if (mask & MASK_X) { - arith(fpc, sat, COS, dst, MASK_X, - swz(src[0], X, X, X, X), none, none); + /* avoid overwriting the source */ + if(src[0].swz[SWZ_X] != SWZ_X) + { + if (mask & MASK_X) { + arith(fpc, sat, COS, dst, MASK_X, + swz(src[0], X, X, X, X), none, none); + } + if (mask & MASK_Y) { + arith(fpc, sat, SIN, dst, MASK_Y, + swz(src[0], X, X, X, X), none, none); + } } - if (mask & MASK_Y) { - arith(fpc, sat, SIN, dst, MASK_Y, - swz(src[0], X, X, X, X), none, none); + else + { + if (mask & MASK_Y) { + arith(fpc, sat, SIN, dst, MASK_Y, + swz(src[0], X, X, X, X), none, none); + } + if (mask & MASK_X) { + arith(fpc, sat, COS, dst, MASK_X, + swz(src[0], X, X, X, X), none, none); + } } break; case TGSI_OPCODE_SIN: @@ -571,15 +588,15 @@ nv30_fragprog_parse_decl_attrib(struct nv30_fpc *fpc, { int hw; - switch (fdec->Semantic.SemanticName) { + switch (fdec->Semantic.Name) { case TGSI_SEMANTIC_POSITION: hw = NV30_FP_OP_INPUT_SRC_POSITION; break; case TGSI_SEMANTIC_COLOR: - if (fdec->Semantic.SemanticIndex == 0) { + if (fdec->Semantic.Index == 0) { hw = NV30_FP_OP_INPUT_SRC_COL0; } else - if (fdec->Semantic.SemanticIndex == 1) { + if (fdec->Semantic.Index == 1) { hw = NV30_FP_OP_INPUT_SRC_COL1; } else { NOUVEAU_ERR("bad colour semantic index\n"); @@ -590,9 +607,9 @@ nv30_fragprog_parse_decl_attrib(struct nv30_fpc *fpc, hw = NV30_FP_OP_INPUT_SRC_FOGC; break; case TGSI_SEMANTIC_GENERIC: - if (fdec->Semantic.SemanticIndex <= 7) { + if (fdec->Semantic.Index <= 7) { hw = NV30_FP_OP_INPUT_SRC_TC(fdec->Semantic. - SemanticIndex); + Index); } else { NOUVEAU_ERR("bad generic semantic index\n"); return FALSE; @@ -603,7 +620,7 @@ nv30_fragprog_parse_decl_attrib(struct nv30_fpc *fpc, return FALSE; } - fpc->attrib_map[fdec->DeclarationRange.First] = hw; + fpc->attrib_map[fdec->Range.First] = hw; return TRUE; } @@ -611,12 +628,12 @@ static boolean nv30_fragprog_parse_decl_output(struct nv30_fpc *fpc, const struct tgsi_full_declaration *fdec) { - switch (fdec->Semantic.SemanticName) { + switch (fdec->Semantic.Name) { case TGSI_SEMANTIC_POSITION: - fpc->depth_id = fdec->DeclarationRange.First; + fpc->depth_id = fdec->Range.First; break; case TGSI_SEMANTIC_COLOR: - fpc->colour_id = fdec->DeclarationRange.First; + fpc->colour_id = fdec->Range.First; break; default: NOUVEAU_ERR("bad output semantic\n"); @@ -652,9 +669,9 @@ nv30_fragprog_prepare(struct nv30_fpc *fpc) goto out_err; break; /*case TGSI_FILE_TEMPORARY: - if (fdec->DeclarationRange.Last > high_temp) { + if (fdec->Range.Last > high_temp) { high_temp = - fdec->DeclarationRange.Last; + fdec->Range.Last; } break;*/ default: @@ -666,7 +683,7 @@ nv30_fragprog_prepare(struct nv30_fpc *fpc) { struct tgsi_full_immediate *imm; float vals[4]; - + imm = &p.FullToken.FullImmediate; assert(imm->Immediate.DataType == TGSI_IMM_FLOAT32); assert(fpc->nr_imm < MAX_IMM); @@ -754,7 +771,7 @@ nv30_fragprog_translate(struct nv30_context *nv30, fp->insn[fpc->inst_offset + 1] = 0x00000000; fp->insn[fpc->inst_offset + 2] = 0x00000000; fp->insn[fpc->inst_offset + 3] = 0x00000000; - + fp->translated = TRUE; fp->on_hw = FALSE; out_err: @@ -820,7 +837,7 @@ nv30_fragprog_validate(struct nv30_context *nv30) fp->buffer = pscreen->buffer_create(pscreen, 0x100, 0, fp->insn_len * 4); nv30_fragprog_upload(nv30, fp); - so = so_new(8, 1); + so = so_new(4, 4, 1); so_method(so, nv30->screen->rankine, NV34TCL_FP_ACTIVE_PROGRAM, 1); so_reloc (so, nouveau_bo(fp->buffer), 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | NOUVEAU_BO_LOW | @@ -838,7 +855,7 @@ nv30_fragprog_validate(struct nv30_context *nv30) update_constants: if (fp->nr_consts) { float *map; - + map = pipe_buffer_map(pscreen, constbuf, PIPE_BUFFER_USAGE_CPU_READ); for (i = 0; i < fp->nr_consts; i++) { @@ -869,6 +886,12 @@ void nv30_fragprog_destroy(struct nv30_context *nv30, struct nv30_fragment_program *fp) { + if (fp->buffer) + pipe_buffer_reference(&fp->buffer, NULL); + + if (fp->so) + so_ref(NULL, &fp->so); + if (fp->insn_len) FREE(fp->insn); } diff --git a/src/gallium/drivers/nv30/nv30_fragtex.c b/src/gallium/drivers/nv30/nv30_fragtex.c index dca760cae62..98935678911 100644 --- a/src/gallium/drivers/nv30/nv30_fragtex.c +++ b/src/gallium/drivers/nv30/nv30_fragtex.c @@ -74,9 +74,9 @@ nv30_fragtex_build(struct nv30_context *nv30, int unit) txf = tf->format; txf |= ((pt->last_level>0) ? NV34TCL_TX_FORMAT_MIPMAP : 0); - txf |= log2i(pt->width[0]) << NV34TCL_TX_FORMAT_BASE_SIZE_U_SHIFT; - txf |= log2i(pt->height[0]) << NV34TCL_TX_FORMAT_BASE_SIZE_V_SHIFT; - txf |= log2i(pt->depth[0]) << NV34TCL_TX_FORMAT_BASE_SIZE_W_SHIFT; + txf |= log2i(pt->width0) << NV34TCL_TX_FORMAT_BASE_SIZE_U_SHIFT; + txf |= log2i(pt->height0) << NV34TCL_TX_FORMAT_BASE_SIZE_V_SHIFT; + txf |= log2i(pt->depth0) << NV34TCL_TX_FORMAT_BASE_SIZE_W_SHIFT; txf |= NV34TCL_TX_FORMAT_NO_BORDER | 0x10000; switch (pt->target) { @@ -106,7 +106,7 @@ nv30_fragtex_build(struct nv30_context *nv30, int unit) txs = tf->swizzle; - so = so_new(16, 2); + so = so_new(1, 8, 2); so_method(so, nv30->screen->rankine, NV34TCL_TX_OFFSET(unit), 8); so_reloc (so, bo, 0, tex_flags | NOUVEAU_BO_LOW, 0, 0); so_reloc (so, bo, txf, tex_flags | NOUVEAU_BO_OR, @@ -115,8 +115,8 @@ nv30_fragtex_build(struct nv30_context *nv30, int unit) so_data (so, NV34TCL_TX_ENABLE_ENABLE | ps->en); so_data (so, txs); so_data (so, ps->filt | 0x2000 /*voodoo*/); - so_data (so, (pt->width[0] << NV34TCL_TX_NPOT_SIZE_W_SHIFT) | - pt->height[0]); + so_data (so, (pt->width0 << NV34TCL_TX_NPOT_SIZE_W_SHIFT) | + pt->height0); so_data (so, ps->bcol); return so; @@ -135,7 +135,7 @@ nv30_fragtex_validate(struct nv30_context *nv30) unit = ffs(samplers) - 1; samplers &= ~(1 << unit); - so = so_new(2, 0); + so = so_new(1, 1, 0); so_method(so, nv30->screen->rankine, NV34TCL_TX_ENABLE(unit), 1); so_data (so, 0); so_ref(so, &nv30->state.hw[NV30_STATE_FRAGTEX0 + unit]); diff --git a/src/gallium/drivers/nv30/nv30_miptree.c b/src/gallium/drivers/nv30/nv30_miptree.c index 280696d4503..8fbba38e78f 100644 --- a/src/gallium/drivers/nv30/nv30_miptree.c +++ b/src/gallium/drivers/nv30/nv30_miptree.c @@ -1,14 +1,17 @@ #include "pipe/p_state.h" #include "pipe/p_defines.h" #include "pipe/p_inlines.h" +#include "util/u_format.h" +#include "util/u_math.h" #include "nv30_context.h" +#include "../nv04/nv04_surface_2d.h" static void nv30_miptree_layout(struct nv30_miptree *nv30mt) { struct pipe_texture *pt = &nv30mt->base; - uint width = pt->width[0], height = pt->height[0], depth = pt->depth[0]; + uint width = pt->width0; uint offset = 0; int nr_faces, l, f; uint wide_pitch = pt->tex_usage & (PIPE_TEXTURE_USAGE_SAMPLER | @@ -21,29 +24,21 @@ nv30_miptree_layout(struct nv30_miptree *nv30mt) nr_faces = 6; } else if (pt->target == PIPE_TEXTURE_3D) { - nr_faces = pt->depth[0]; + nr_faces = pt->depth0; } else { nr_faces = 1; } for (l = 0; l <= pt->last_level; l++) { - pt->width[l] = width; - pt->height[l] = height; - pt->depth[l] = depth; - pt->nblocksx[l] = pf_get_nblocksx(&pt->block, width); - pt->nblocksy[l] = pf_get_nblocksy(&pt->block, height); - if (wide_pitch && (pt->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR)) - nv30mt->level[l].pitch = align(pt->width[0] * pt->block.size, 64); + nv30mt->level[l].pitch = align(util_format_get_stride(pt->format, pt->width0), 64); else - nv30mt->level[l].pitch = pt->width[l] * pt->block.size; + nv30mt->level[l].pitch = util_format_get_stride(pt->format, width); nv30mt->level[l].image_offset = CALLOC(nr_faces, sizeof(unsigned)); - width = MAX2(1, width >> 1); - height = MAX2(1, height >> 1); - depth = MAX2(1, depth >> 1); + width = u_minify(width, 1); } for (f = 0; f < nr_faces; f++) { @@ -51,14 +46,14 @@ nv30_miptree_layout(struct nv30_miptree *nv30mt) nv30mt->level[l].image_offset[f] = offset; if (!(pt->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR) && - pt->width[l + 1] > 1 && pt->height[l + 1] > 1) - offset += align(nv30mt->level[l].pitch * pt->height[l], 64); + u_minify(pt->width0, l + 1) > 1 && u_minify(pt->height0, l + 1) > 1) + offset += align(nv30mt->level[l].pitch * u_minify(pt->height0, l), 64); else - offset += nv30mt->level[l].pitch * pt->height[l]; + offset += nv30mt->level[l].pitch * u_minify(pt->height0, l); } nv30mt->level[l].image_offset[f] = offset; - offset += nv30mt->level[l].pitch * pt->height[l]; + offset += nv30mt->level[l].pitch * u_minify(pt->height0, l); } nv30mt->total_size = offset; @@ -79,8 +74,8 @@ nv30_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt) mt->base.screen = pscreen; /* Swizzled textures must be POT */ - if (pt->width[0] & (pt->width[0] - 1) || - pt->height[0] & (pt->height[0] - 1)) + if (pt->width0 & (pt->width0 - 1) || + pt->height0 & (pt->height0 - 1)) mt->base.tex_usage |= NOUVEAU_TEXTURE_USAGE_LINEAR; else if (pt->tex_usage & (PIPE_TEXTURE_USAGE_PRIMARY | @@ -114,6 +109,12 @@ nv30_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt) if (pt->tex_usage & PIPE_TEXTURE_USAGE_DYNAMIC) buf_usage |= PIPE_BUFFER_USAGE_CPU_READ_WRITE; + /* apparently we can't render to swizzled surfaces smaller than 64 bytes, so make them linear. + * If the user did not ask for a render target, they can still render to it, but it will cost them an extra copy. + * This also happens for small mipmaps of large textures. */ + if (pt->tex_usage & PIPE_TEXTURE_USAGE_RENDER_TARGET && util_format_get_stride(pt->format, pt->width0) < 64) + mt->base.tex_usage |= NOUVEAU_TEXTURE_USAGE_LINEAR; + nv30_miptree_layout(mt); mt->buffer = pscreen->buffer_create(pscreen, 256, buf_usage, @@ -122,6 +123,7 @@ nv30_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt) FREE(mt); return NULL; } + mt->bo = nouveau_bo(mt->buffer); return &mt->base; } @@ -134,7 +136,7 @@ nv30_miptree_blanket(struct pipe_screen *pscreen, const struct pipe_texture *pt, /* Only supports 2D, non-mipmapped textures for the moment */ if (pt->target != PIPE_TEXTURE_2D || pt->last_level != 0 || - pt->depth[0] != 1) + pt->depth0 != 1) return NULL; mt = CALLOC_STRUCT(nv30_miptree); @@ -151,6 +153,7 @@ nv30_miptree_blanket(struct pipe_screen *pscreen, const struct pipe_texture *pt, mt->base.tex_usage |= NOUVEAU_TEXTURE_USAGE_LINEAR; pipe_buffer_reference(&mt->buffer, pb); + mt->bo = nouveau_bo(mt->buffer); return &mt->base; } @@ -182,8 +185,8 @@ nv30_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt, return NULL; pipe_texture_reference(&ns->base.texture, pt); ns->base.format = pt->format; - ns->base.width = pt->width[level]; - ns->base.height = pt->height[level]; + ns->base.width = u_minify(pt->width0, level); + ns->base.height = u_minify(pt->height0, level); ns->base.usage = flags; pipe_reference_init(&ns->base.reference, 1); ns->base.face = face; @@ -200,12 +203,27 @@ nv30_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt, ns->base.offset = nv30mt->level[level].image_offset[0]; } + /* create a linear temporary that we can render into if necessary. + * Note that ns->pitch is always a multiple of 64 for linear surfaces and swizzled surfaces are POT, so + * ns->pitch & 63 is equivalent to (ns->pitch < 64 && swizzled)*/ + if((ns->pitch & 63) && (ns->base.usage & (PIPE_BUFFER_USAGE_GPU_WRITE | NOUVEAU_BUFFER_USAGE_NO_RENDER)) == PIPE_BUFFER_USAGE_GPU_WRITE) + return &nv04_surface_wrap_for_render(pscreen, ((struct nv30_screen*)pscreen)->eng2d, ns)->base; + return &ns->base; } static void nv30_miptree_surface_del(struct pipe_surface *ps) { + struct nv04_surface* ns = (struct nv04_surface*)ps; + if(ns->backing) + { + struct nv30_screen* screen = (struct nv30_screen*)ps->texture->screen; + if(ns->backing->base.usage & PIPE_BUFFER_USAGE_GPU_WRITE) + screen->eng2d->copy(screen->eng2d, &ns->backing->base, 0, 0, ps, 0, 0, ns->base.width, ns->base.height); + nv30_miptree_surface_del(&ns->backing->base); + } + pipe_texture_reference(&ps->texture, NULL); FREE(ps); } diff --git a/src/gallium/drivers/nv30/nv30_query.c b/src/gallium/drivers/nv30/nv30_query.c index 1d1c8a484e1..e27e9ccbf60 100644 --- a/src/gallium/drivers/nv30/nv30_query.c +++ b/src/gallium/drivers/nv30/nv30_query.c @@ -41,6 +41,9 @@ nv30_query_begin(struct pipe_context *pipe, struct pipe_query *pq) { struct nv30_context *nv30 = nv30_context(pipe); struct nv30_query *q = nv30_query(pq); + struct nv30_screen *screen = nv30->screen; + struct nouveau_channel *chan = screen->base.channel; + struct nouveau_grobj *rankine = screen->rankine; assert(q->type == PIPE_QUERY_OCCLUSION_COUNTER); @@ -57,10 +60,10 @@ nv30_query_begin(struct pipe_context *pipe, struct pipe_query *pq) assert(0); nouveau_notifier_reset(nv30->screen->query, q->object->start); - BEGIN_RING(rankine, NV34TCL_QUERY_RESET, 1); - OUT_RING (1); - BEGIN_RING(rankine, NV34TCL_QUERY_UNK17CC, 1); - OUT_RING (1); + BEGIN_RING(chan, rankine, NV34TCL_QUERY_RESET, 1); + OUT_RING (chan, 1); + BEGIN_RING(chan, rankine, NV34TCL_QUERY_UNK17CC, 1); + OUT_RING (chan, 1); q->ready = FALSE; } @@ -69,12 +72,15 @@ static void nv30_query_end(struct pipe_context *pipe, struct pipe_query *pq) { struct nv30_context *nv30 = nv30_context(pipe); + struct nv30_screen *screen = nv30->screen; + struct nouveau_channel *chan = screen->base.channel; + struct nouveau_grobj *rankine = screen->rankine; struct nv30_query *q = nv30_query(pq); - BEGIN_RING(rankine, NV34TCL_QUERY_GET, 1); - OUT_RING ((0x01 << NV34TCL_QUERY_GET_UNK24_SHIFT) | + BEGIN_RING(chan, rankine, NV34TCL_QUERY_GET, 1); + OUT_RING (chan, (0x01 << NV34TCL_QUERY_GET_UNK24_SHIFT) | ((q->object->start * 32) << NV34TCL_QUERY_GET_OFFSET_SHIFT)); - FIRE_RING(NULL); + FIRE_RING(chan); } static boolean diff --git a/src/gallium/drivers/nv30/nv30_screen.c b/src/gallium/drivers/nv30/nv30_screen.c index 221ae1b5f88..9ed48178dc2 100644 --- a/src/gallium/drivers/nv30/nv30_screen.c +++ b/src/gallium/drivers/nv30/nv30_screen.c @@ -116,7 +116,10 @@ nv30_screen_surface_format_supported(struct pipe_screen *pscreen, case PIPE_FORMAT_Z24X8_UNORM: return TRUE; case PIPE_FORMAT_Z16_UNORM: - return (front->format == PIPE_FORMAT_R5G6B5_UNORM); + if (front) { + return (front->format == PIPE_FORMAT_R5G6B5_UNORM); + } + return TRUE; default: break; } @@ -153,6 +156,12 @@ static void nv30_screen_destroy(struct pipe_screen *pscreen) { struct nv30_screen *screen = nv30_screen(pscreen); + unsigned i; + + for (i = 0; i < NV30_STATE_MAX; i++) { + if (screen->state[i]) + so_ref(NULL, &screen->state[i]); + } nouveau_resource_free(&screen->vp_exec_heap); nouveau_resource_free(&screen->vp_data_heap); @@ -160,6 +169,9 @@ nv30_screen_destroy(struct pipe_screen *pscreen) nouveau_notifier_free(&screen->query); nouveau_notifier_free(&screen->sync); nouveau_grobj_free(&screen->rankine); + nv04_surface_2d_takedown(&screen->eng2d); + + nouveau_screen_fini(&screen->base); FREE(pscreen); } @@ -221,7 +233,6 @@ nv30_screen_create(struct pipe_winsys *ws, struct nouveau_device *dev) NOUVEAU_ERR("Error creating 3D object: %d\n", ret); return FALSE; } - BIND_RING(chan, screen->rankine, 7); /* 2D engine setup */ screen->eng2d = nv04_surface_2d_init(&screen->base); @@ -258,7 +269,7 @@ nv30_screen_create(struct pipe_winsys *ws, struct nouveau_device *dev) } /* Static rankine initialisation */ - so = so_new(128, 0); + so = so_new(36, 60, 0); so_method(so, screen->rankine, NV34TCL_DMA_NOTIFY, 1); so_data (so, screen->sync->handle); so_method(so, screen->rankine, NV34TCL_DMA_TEXTURE0, 2); diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index b91e972c123..a80dfb04880 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -14,7 +14,7 @@ nv30_blend_state_create(struct pipe_context *pipe, struct nv30_context *nv30 = nv30_context(pipe); struct nouveau_grobj *rankine = nv30->screen->rankine; struct nv30_blend_state *bso = CALLOC(1, sizeof(*bso)); - struct nouveau_stateobj *so = so_new(16, 0); + struct nouveau_stateobj *so = so_new(5, 8, 0); if (cso->blend_enable) { so_method(so, rankine, NV34TCL_BLEND_FUNC_ENABLE, 3); @@ -300,7 +300,7 @@ nv30_rasterizer_state_create(struct pipe_context *pipe, { struct nv30_context *nv30 = nv30_context(pipe); struct nv30_rasterizer_state *rsso = CALLOC(1, sizeof(*rsso)); - struct nouveau_stateobj *so = so_new(32, 0); + struct nouveau_stateobj *so = so_new(9, 19, 0); struct nouveau_grobj *rankine = nv30->screen->rankine; /*XXX: ignored: @@ -435,7 +435,7 @@ nv30_depth_stencil_alpha_state_create(struct pipe_context *pipe, { struct nv30_context *nv30 = nv30_context(pipe); struct nv30_zsa_state *zsaso = CALLOC(1, sizeof(*zsaso)); - struct nouveau_stateobj *so = so_new(32, 0); + struct nouveau_stateobj *so = so_new(5, 21, 0); struct nouveau_grobj *rankine = nv30->screen->rankine; so_method(so, rankine, NV34TCL_DEPTH_FUNC, 3); @@ -672,16 +672,6 @@ nv30_set_vertex_elements(struct pipe_context *pipe, unsigned count, /*nv30->draw_dirty |= NV30_NEW_ARRAYS;*/ } -static void -nv30_set_edgeflags(struct pipe_context *pipe, const unsigned *bitfield) -{ - struct nv30_context *nv30 = nv30_context(pipe); - - nv30->edgeflags = bitfield; - nv30->dirty |= NV30_NEW_ARRAYS; - /*nv30->draw_dirty |= NV30_NEW_ARRAYS;*/ -} - void nv30_init_state_functions(struct nv30_context *nv30) { @@ -690,9 +680,9 @@ nv30_init_state_functions(struct nv30_context *nv30) nv30->pipe.delete_blend_state = nv30_blend_state_delete; nv30->pipe.create_sampler_state = nv30_sampler_state_create; - nv30->pipe.bind_sampler_states = nv30_sampler_state_bind; + nv30->pipe.bind_fragment_sampler_states = nv30_sampler_state_bind; nv30->pipe.delete_sampler_state = nv30_sampler_state_delete; - nv30->pipe.set_sampler_textures = nv30_set_sampler_texture; + nv30->pipe.set_fragment_sampler_textures = nv30_set_sampler_texture; nv30->pipe.create_rasterizer_state = nv30_rasterizer_state_create; nv30->pipe.bind_rasterizer_state = nv30_rasterizer_state_bind; @@ -721,7 +711,6 @@ nv30_init_state_functions(struct nv30_context *nv30) nv30->pipe.set_scissor_state = nv30_set_scissor_state; nv30->pipe.set_viewport_state = nv30_set_viewport_state; - nv30->pipe.set_edgeflags = nv30_set_edgeflags; nv30->pipe.set_vertex_buffers = nv30_set_vertex_buffers; nv30->pipe.set_vertex_elements = nv30_set_vertex_elements; } diff --git a/src/gallium/drivers/nv30/nv30_state.h b/src/gallium/drivers/nv30/nv30_state.h index e6f23bf1667..e42e872de75 100644 --- a/src/gallium/drivers/nv30/nv30_state.h +++ b/src/gallium/drivers/nv30/nv30_state.h @@ -72,6 +72,7 @@ struct nv30_fragment_program { struct nv30_miptree { struct pipe_texture base; + struct nouveau_bo *bo; struct pipe_buffer *buffer; uint total_size; diff --git a/src/gallium/drivers/nv30/nv30_state_blend.c b/src/gallium/drivers/nv30/nv30_state_blend.c index 64cf9ae93a0..c36d58c040c 100644 --- a/src/gallium/drivers/nv30/nv30_state_blend.c +++ b/src/gallium/drivers/nv30/nv30_state_blend.c @@ -18,7 +18,7 @@ struct nv30_state_entry nv30_state_blend = { static boolean nv30_state_blend_colour_validate(struct nv30_context *nv30) { - struct nouveau_stateobj *so = so_new(2, 0); + struct nouveau_stateobj *so = so_new(1, 1, 0); struct pipe_blend_color *bcol = &nv30->blend_colour; so_method(so, nv30->screen->rankine, NV34TCL_BLEND_COLOR, 1); diff --git a/src/gallium/drivers/nv30/nv30_state_emit.c b/src/gallium/drivers/nv30/nv30_state_emit.c index 621b8846c8e..ac52d946f02 100644 --- a/src/gallium/drivers/nv30/nv30_state_emit.c +++ b/src/gallium/drivers/nv30/nv30_state_emit.c @@ -41,7 +41,7 @@ nv30_state_emit(struct nv30_context *nv30) struct nouveau_channel *chan = nv30->screen->base.channel; struct nv30_state *state = &nv30->state; struct nv30_screen *screen = nv30->screen; - unsigned i, samplers; + unsigned i; uint64_t states; if (nv30->pctx_id != screen->cur_pctx) { @@ -63,6 +63,14 @@ nv30_state_emit(struct nv30_context *nv30) } state->dirty = 0; +} + +void +nv30_state_flush_notify(struct nouveau_channel *chan) +{ + struct nv30_context *nv30 = chan->user_private; + struct nv30_state *state = &nv30->state; + unsigned i, samplers; so_emit_reloc_markers(chan, state->hw[NV30_STATE_FB]); for (i = 0, samplers = state->fp_samplers; i < 16 && samplers; i++) { diff --git a/src/gallium/drivers/nv30/nv30_state_fb.c b/src/gallium/drivers/nv30/nv30_state_fb.c index 6f6d1740d6e..2ed2ea55e84 100644 --- a/src/gallium/drivers/nv30/nv30_state_fb.c +++ b/src/gallium/drivers/nv30/nv30_state_fb.c @@ -10,7 +10,7 @@ nv30_state_framebuffer_validate(struct nv30_context *nv30) struct nv04_surface *rt[2], *zeta = NULL; uint32_t rt_enable = 0, rt_format = 0; int i, colour_format = 0, zeta_format = 0, depth_only = 0; - struct nouveau_stateobj *so = so_new(64, 10); + struct nouveau_stateobj *so = so_new(12, 18, 10); unsigned rt_flags = NOUVEAU_BO_RDWR | NOUVEAU_BO_VRAM; unsigned w = fb->width; unsigned h = fb->height; diff --git a/src/gallium/drivers/nv30/nv30_state_scissor.c b/src/gallium/drivers/nv30/nv30_state_scissor.c index 3ac7a8471ea..ba61a9e24a4 100644 --- a/src/gallium/drivers/nv30/nv30_state_scissor.c +++ b/src/gallium/drivers/nv30/nv30_state_scissor.c @@ -12,7 +12,7 @@ nv30_state_scissor_validate(struct nv30_context *nv30) return FALSE; nv30->state.scissor_enabled = rast->scissor; - so = so_new(3, 0); + so = so_new(1, 2, 0); so_method(so, nv30->screen->rankine, NV34TCL_SCISSOR_HORIZ, 2); if (nv30->state.scissor_enabled) { so_data (so, ((s->maxx - s->minx) << 16) | s->minx); diff --git a/src/gallium/drivers/nv30/nv30_state_stipple.c b/src/gallium/drivers/nv30/nv30_state_stipple.c index d0c791ac082..ed520a4f439 100644 --- a/src/gallium/drivers/nv30/nv30_state_stipple.c +++ b/src/gallium/drivers/nv30/nv30_state_stipple.c @@ -14,14 +14,14 @@ nv30_state_stipple_validate(struct nv30_context *nv30) if (rast->poly_stipple_enable) { unsigned i; - so = so_new(35, 0); + so = so_new(2, 33, 0); so_method(so, rankine, NV34TCL_POLYGON_STIPPLE_ENABLE, 1); so_data (so, 1); so_method(so, rankine, NV34TCL_POLYGON_STIPPLE_PATTERN(0), 32); for (i = 0; i < 32; i++) so_data(so, nv30->stipple[i]); } else { - so = so_new(2, 0); + so = so_new(1, 1, 0); so_method(so, rankine, NV34TCL_POLYGON_STIPPLE_ENABLE, 1); so_data (so, 0); } diff --git a/src/gallium/drivers/nv30/nv30_state_viewport.c b/src/gallium/drivers/nv30/nv30_state_viewport.c index c3eb413dac6..2d7781292bd 100644 --- a/src/gallium/drivers/nv30/nv30_state_viewport.c +++ b/src/gallium/drivers/nv30/nv30_state_viewport.c @@ -19,7 +19,7 @@ nv30_state_viewport_validate(struct nv30_context *nv30) return FALSE; nv30->state.viewport_bypass = bypass; - so = so_new(11, 0); + so = so_new(3, 10, 0); if (!bypass) { so_method(so, nv30->screen->rankine, NV34TCL_VIEWPORT_TRANSLATE_X, 8); diff --git a/src/gallium/drivers/nv30/nv30_transfer.c b/src/gallium/drivers/nv30/nv30_transfer.c index 98011decf7c..65598991c68 100644 --- a/src/gallium/drivers/nv30/nv30_transfer.c +++ b/src/gallium/drivers/nv30/nv30_transfer.c @@ -1,7 +1,9 @@ #include <pipe/p_state.h> #include <pipe/p_defines.h> #include <pipe/p_inlines.h> +#include <util/u_format.h> #include <util/u_memory.h> +#include <util/u_math.h> #include <nouveau/nouveau_winsys.h> #include "nv30_context.h" #include "nv30_screen.h" @@ -10,22 +12,19 @@ struct nv30_transfer { struct pipe_transfer base; struct pipe_surface *surface; - bool direct; + boolean direct; }; static void -nv30_compatible_transfer_tex(struct pipe_texture *pt, unsigned level, +nv30_compatible_transfer_tex(struct pipe_texture *pt, unsigned width, unsigned height, struct pipe_texture *template) { memset(template, 0, sizeof(struct pipe_texture)); template->target = pt->target; template->format = pt->format; - template->width[0] = pt->width[level]; - template->height[0] = pt->height[level]; - template->depth[0] = 1; - template->block = pt->block; - template->nblocksx[0] = pt->nblocksx[level]; - template->nblocksy[0] = pt->nblocksx[level]; + template->width0 = width; + template->height0 = height; + template->depth0 = 1; template->last_level = 0; template->nr_samples = pt->nr_samples; @@ -48,14 +47,10 @@ nv30_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, return NULL; pipe_texture_reference(&tx->base.texture, pt); - tx->base.format = pt->format; tx->base.x = x; tx->base.y = y; tx->base.width = w; tx->base.height = h; - tx->base.block = pt->block; - tx->base.nblocksx = pt->nblocksx[level]; - tx->base.nblocksy = pt->nblocksy[level]; tx->base.stride = mt->level[level].pitch; tx->base.usage = usage; tx->base.face = face; @@ -76,7 +71,7 @@ nv30_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, tx->direct = false; - nv30_compatible_transfer_tex(pt, level, &tx_tex_template); + nv30_compatible_transfer_tex(pt, w, h, &tx_tex_template); tx_tex = pscreen->texture_create(pscreen, &tx_tex_template); if (!tx_tex) @@ -85,6 +80,8 @@ nv30_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, return NULL; } + tx->base.stride = ((struct nv30_miptree*)tx_tex)->level[0].pitch; + tx->surface = pscreen->get_tex_surface(pscreen, tx_tex, 0, 0, 0, pipe_transfer_buffer_flags(&tx->base)); @@ -110,8 +107,8 @@ nv30_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, /* TODO: Check if SIFM can un-swizzle */ nvscreen->eng2d->copy(nvscreen->eng2d, tx->surface, 0, 0, - src, 0, 0, - src->width, src->height); + src, x, y, + w, h); pipe_surface_reference(&src, NULL); } @@ -131,13 +128,13 @@ nv30_transfer_del(struct pipe_transfer *ptx) dst = pscreen->get_tex_surface(pscreen, ptx->texture, ptx->face, ptx->level, ptx->zslice, - PIPE_BUFFER_USAGE_GPU_WRITE); + PIPE_BUFFER_USAGE_GPU_WRITE | NOUVEAU_BUFFER_USAGE_NO_RENDER); /* TODO: Check if SIFM can deal with x,y,w,h when swizzling */ nvscreen->eng2d->copy(nvscreen->eng2d, - dst, 0, 0, + dst, tx->base.x, tx->base.y, tx->surface, 0, 0, - dst->width, dst->height); + tx->base.width, tx->base.height); pipe_surface_reference(&dst, NULL); } @@ -156,8 +153,10 @@ nv30_transfer_map(struct pipe_screen *pscreen, struct pipe_transfer *ptx) void *map = pipe_buffer_map(pscreen, mt->buffer, pipe_transfer_buffer_flags(ptx)); - return map + ns->base.offset + - ptx->y * ns->pitch + ptx->x * ptx->block.size; + if(!tx->direct) + return map + ns->base.offset; + else + return map + ns->base.offset + ptx->y * ns->pitch + ptx->x * util_format_get_blocksize(ptx->texture->format); } static void diff --git a/src/gallium/drivers/nv30/nv30_vbo.c b/src/gallium/drivers/nv30/nv30_vbo.c index 189656ec817..1c5db03ea24 100644 --- a/src/gallium/drivers/nv30/nv30_vbo.c +++ b/src/gallium/drivers/nv30/nv30_vbo.c @@ -163,19 +163,21 @@ nv30_vbo_static_attrib(struct nv30_context *nv30, struct nouveau_stateobj *so, return TRUE; } -boolean +void nv30_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, unsigned count) { struct nv30_context *nv30 = nv30_context(pipe); - struct nouveau_channel *chan = nv30->screen->base.channel; + struct nv30_screen *screen = nv30->screen; + struct nouveau_channel *chan = screen->base.channel; + struct nouveau_grobj *rankine = screen->rankine; unsigned restart = 0; nv30_vbo_set_idxbuf(nv30, NULL, 0); if (FORCE_SWTNL || !nv30_state_validate(nv30)) { /*return nv30_draw_elements_swtnl(pipe, NULL, 0, mode, start, count);*/ - return FALSE; + return; } while (count) { @@ -186,17 +188,17 @@ nv30_draw_arrays(struct pipe_context *pipe, vc = nouveau_vbuf_split(chan->pushbuf->remaining, 6, 256, mode, start, count, &restart); if (!vc) { - FIRE_RING(NULL); + FIRE_RING(chan); continue; } - BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); - OUT_RING (nvgl_primitive(mode)); + BEGIN_RING(chan, rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (chan, nvgl_primitive(mode)); nr = (vc & 0xff); if (nr) { - BEGIN_RING(rankine, NV34TCL_VB_VERTEX_BATCH, 1); - OUT_RING (((nr - 1) << 24) | start); + BEGIN_RING(chan, rankine, NV34TCL_VB_VERTEX_BATCH, 1); + OUT_RING (chan, ((nr - 1) << 24) | start); start += nr; } @@ -206,15 +208,15 @@ nv30_draw_arrays(struct pipe_context *pipe, nr -= push; - BEGIN_RING_NI(rankine, NV34TCL_VB_VERTEX_BATCH, push); + BEGIN_RING_NI(chan, rankine, NV34TCL_VB_VERTEX_BATCH, push); while (push--) { - OUT_RING(((0x100 - 1) << 24) | start); + OUT_RING(chan, ((0x100 - 1) << 24) | start); start += 0x100; } } - BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); - OUT_RING (0); + BEGIN_RING(chan, rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (chan, 0); count -= vc; start = restart; @@ -228,7 +230,9 @@ static INLINE void nv30_draw_elements_u08(struct nv30_context *nv30, void *ib, unsigned mode, unsigned start, unsigned count) { - struct nouveau_channel *chan = nv30->screen->base.channel; + struct nv30_screen *screen = nv30->screen; + struct nouveau_channel *chan = screen->base.channel; + struct nouveau_grobj *rankine = screen->rankine; while (count) { uint8_t *elts = (uint8_t *)ib + start; @@ -239,17 +243,17 @@ nv30_draw_elements_u08(struct nv30_context *nv30, void *ib, vc = nouveau_vbuf_split(chan->pushbuf->remaining, 6, 2, mode, start, count, &restart); if (vc == 0) { - FIRE_RING(NULL); + FIRE_RING(chan); continue; } count -= vc; - BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); - OUT_RING (nvgl_primitive(mode)); + BEGIN_RING(chan, rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (chan, nvgl_primitive(mode)); if (vc & 1) { - BEGIN_RING(rankine, NV34TCL_VB_ELEMENT_U32, 1); - OUT_RING (elts[0]); + BEGIN_RING(chan, rankine, NV34TCL_VB_ELEMENT_U32, 1); + OUT_RING (chan, elts[0]); elts++; vc--; } @@ -258,16 +262,16 @@ nv30_draw_elements_u08(struct nv30_context *nv30, void *ib, push = MIN2(vc, 2047 * 2); - BEGIN_RING_NI(rankine, NV34TCL_VB_ELEMENT_U16, push >> 1); + BEGIN_RING_NI(chan, rankine, NV34TCL_VB_ELEMENT_U16, push >> 1); for (i = 0; i < push; i+=2) - OUT_RING((elts[i+1] << 16) | elts[i]); + OUT_RING(chan, (elts[i+1] << 16) | elts[i]); vc -= push; elts += push; } - BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); - OUT_RING (0); + BEGIN_RING(chan, rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (chan, 0); start = restart; } @@ -277,7 +281,9 @@ static INLINE void nv30_draw_elements_u16(struct nv30_context *nv30, void *ib, unsigned mode, unsigned start, unsigned count) { - struct nouveau_channel *chan = nv30->screen->base.channel; + struct nv30_screen *screen = nv30->screen; + struct nouveau_channel *chan = screen->base.channel; + struct nouveau_grobj *rankine = screen->rankine; while (count) { uint16_t *elts = (uint16_t *)ib + start; @@ -288,17 +294,17 @@ nv30_draw_elements_u16(struct nv30_context *nv30, void *ib, vc = nouveau_vbuf_split(chan->pushbuf->remaining, 6, 2, mode, start, count, &restart); if (vc == 0) { - FIRE_RING(NULL); + FIRE_RING(chan); continue; } count -= vc; - BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); - OUT_RING (nvgl_primitive(mode)); + BEGIN_RING(chan, rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (chan, nvgl_primitive(mode)); if (vc & 1) { - BEGIN_RING(rankine, NV34TCL_VB_ELEMENT_U32, 1); - OUT_RING (elts[0]); + BEGIN_RING(chan, rankine, NV34TCL_VB_ELEMENT_U32, 1); + OUT_RING (chan, elts[0]); elts++; vc--; } @@ -307,16 +313,16 @@ nv30_draw_elements_u16(struct nv30_context *nv30, void *ib, push = MIN2(vc, 2047 * 2); - BEGIN_RING_NI(rankine, NV34TCL_VB_ELEMENT_U16, push >> 1); + BEGIN_RING_NI(chan, rankine, NV34TCL_VB_ELEMENT_U16, push >> 1); for (i = 0; i < push; i+=2) - OUT_RING((elts[i+1] << 16) | elts[i]); + OUT_RING(chan, (elts[i+1] << 16) | elts[i]); vc -= push; elts += push; } - BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); - OUT_RING (0); + BEGIN_RING(chan, rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (chan, 0); start = restart; } @@ -326,7 +332,9 @@ static INLINE void nv30_draw_elements_u32(struct nv30_context *nv30, void *ib, unsigned mode, unsigned start, unsigned count) { - struct nouveau_channel *chan = nv30->screen->base.channel; + struct nv30_screen *screen = nv30->screen; + struct nouveau_channel *chan = screen->base.channel; + struct nouveau_grobj *rankine = screen->rankine; while (count) { uint32_t *elts = (uint32_t *)ib + start; @@ -337,32 +345,32 @@ nv30_draw_elements_u32(struct nv30_context *nv30, void *ib, vc = nouveau_vbuf_split(chan->pushbuf->remaining, 5, 1, mode, start, count, &restart); if (vc == 0) { - FIRE_RING(NULL); + FIRE_RING(chan); continue; } count -= vc; - BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); - OUT_RING (nvgl_primitive(mode)); + BEGIN_RING(chan, rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (chan, nvgl_primitive(mode)); while (vc) { push = MIN2(vc, 2047); - BEGIN_RING_NI(rankine, NV34TCL_VB_ELEMENT_U32, push); - OUT_RINGp (elts, push); + BEGIN_RING_NI(chan, rankine, NV34TCL_VB_ELEMENT_U32, push); + OUT_RINGp (chan, elts, push); vc -= push; elts += push; } - BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); - OUT_RING (0); + BEGIN_RING(chan, rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (chan, 0); start = restart; } } -static boolean +static void nv30_draw_elements_inline(struct pipe_context *pipe, struct pipe_buffer *ib, unsigned ib_size, unsigned mode, unsigned start, unsigned count) @@ -393,15 +401,16 @@ nv30_draw_elements_inline(struct pipe_context *pipe, } pipe_buffer_unmap(pscreen, ib); - return TRUE; } -static boolean +static void nv30_draw_elements_vbo(struct pipe_context *pipe, unsigned mode, unsigned start, unsigned count) { struct nv30_context *nv30 = nv30_context(pipe); - struct nouveau_channel *chan = nv30->screen->base.channel; + struct nv30_screen *screen = nv30->screen; + struct nouveau_channel *chan = screen->base.channel; + struct nouveau_grobj *rankine = screen->rankine; unsigned restart = 0; while (count) { @@ -412,17 +421,17 @@ nv30_draw_elements_vbo(struct pipe_context *pipe, vc = nouveau_vbuf_split(chan->pushbuf->remaining, 6, 256, mode, start, count, &restart); if (!vc) { - FIRE_RING(NULL); + FIRE_RING(chan); continue; } - BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); - OUT_RING (nvgl_primitive(mode)); + BEGIN_RING(chan, rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (chan, nvgl_primitive(mode)); nr = (vc & 0xff); if (nr) { - BEGIN_RING(rankine, NV34TCL_VB_INDEX_BATCH, 1); - OUT_RING (((nr - 1) << 24) | start); + BEGIN_RING(chan, rankine, NV34TCL_VB_INDEX_BATCH, 1); + OUT_RING (chan, ((nr - 1) << 24) | start); start += nr; } @@ -432,24 +441,22 @@ nv30_draw_elements_vbo(struct pipe_context *pipe, nr -= push; - BEGIN_RING_NI(rankine, NV34TCL_VB_INDEX_BATCH, push); + BEGIN_RING_NI(chan, rankine, NV34TCL_VB_INDEX_BATCH, push); while (push--) { - OUT_RING(((0x100 - 1) << 24) | start); + OUT_RING(chan, ((0x100 - 1) << 24) | start); start += 0x100; } } - BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); - OUT_RING (0); + BEGIN_RING(chan, rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (chan, 0); count -= vc; start = restart; } - - return TRUE; } -boolean +void nv30_draw_elements(struct pipe_context *pipe, struct pipe_buffer *indexBuffer, unsigned indexSize, unsigned mode, unsigned start, unsigned count) @@ -461,7 +468,7 @@ nv30_draw_elements(struct pipe_context *pipe, if (FORCE_SWTNL || !nv30_state_validate(nv30)) { /*return nv30_draw_elements_swtnl(pipe, NULL, 0, mode, start, count);*/ - return FALSE; + return; } if (idxbuf) { @@ -472,7 +479,6 @@ nv30_draw_elements(struct pipe_context *pipe, } pipe->flush(pipe, 0, NULL); - return TRUE; } static boolean @@ -485,14 +491,9 @@ nv30_vbo_validate(struct nv30_context *nv30) unsigned vb_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD; int hw; - if (nv30->edgeflags) { - /*nv30->fallback_swtnl |= NV30_NEW_ARRAYS;*/ - return FALSE; - } - - vtxbuf = so_new(20, 18); + vtxbuf = so_new(3, 17, 18); so_method(vtxbuf, rankine, NV34TCL_VTXBUF_ADDRESS(0), nv30->vtxelt_nr); - vtxfmt = so_new(17, 0); + vtxfmt = so_new(1, 16, 0); so_method(vtxfmt, rankine, NV34TCL_VTXFMT(0), nv30->vtxelt_nr); for (hw = 0; hw < nv30->vtxelt_nr; hw++) { @@ -505,7 +506,7 @@ nv30_vbo_validate(struct nv30_context *nv30) if (!vb->stride) { if (!sattr) - sattr = so_new(16 * 5, 0); + sattr = so_new(16, 16 * 4, 0); if (nv30_vbo_static_attrib(nv30, sattr, hw, ve, vb)) { so_data(vtxbuf, 0); diff --git a/src/gallium/drivers/nv30/nv30_vertprog.c b/src/gallium/drivers/nv30/nv30_vertprog.c index 14a5c0260d0..e77a5be3f23 100644 --- a/src/gallium/drivers/nv30/nv30_vertprog.c +++ b/src/gallium/drivers/nv30/nv30_vertprog.c @@ -253,32 +253,32 @@ static INLINE struct nv30_sreg tgsi_src(struct nv30_vpc *vpc, const struct tgsi_full_src_register *fsrc) { struct nv30_sreg src; - switch (fsrc->SrcRegister.File) { + switch (fsrc->Register.File) { case TGSI_FILE_INPUT: - src = nv30_sr(NV30SR_INPUT, fsrc->SrcRegister.Index); + src = nv30_sr(NV30SR_INPUT, fsrc->Register.Index); break; case TGSI_FILE_CONSTANT: - src = constant(vpc, fsrc->SrcRegister.Index, 0, 0, 0, 0); + src = constant(vpc, fsrc->Register.Index, 0, 0, 0, 0); break; case TGSI_FILE_IMMEDIATE: - src = vpc->imm[fsrc->SrcRegister.Index]; + src = vpc->imm[fsrc->Register.Index]; break; case TGSI_FILE_TEMPORARY: - if (vpc->high_temp < fsrc->SrcRegister.Index) - vpc->high_temp = fsrc->SrcRegister.Index; - src = nv30_sr(NV30SR_TEMP, fsrc->SrcRegister.Index); + if (vpc->high_temp < fsrc->Register.Index) + vpc->high_temp = fsrc->Register.Index; + src = nv30_sr(NV30SR_TEMP, fsrc->Register.Index); break; default: NOUVEAU_ERR("bad src file\n"); break; } - src.abs = fsrc->SrcRegisterExtMod.Absolute; - src.negate = fsrc->SrcRegister.Negate; - src.swz[0] = fsrc->SrcRegister.SwizzleX; - src.swz[1] = fsrc->SrcRegister.SwizzleY; - src.swz[2] = fsrc->SrcRegister.SwizzleZ; - src.swz[3] = fsrc->SrcRegister.SwizzleW; + src.abs = fsrc->Register.Absolute; + src.negate = fsrc->Register.Negate; + src.swz[0] = fsrc->Register.SwizzleX; + src.swz[1] = fsrc->Register.SwizzleY; + src.swz[2] = fsrc->Register.SwizzleZ; + src.swz[3] = fsrc->Register.SwizzleW; return src; } @@ -286,14 +286,14 @@ static INLINE struct nv30_sreg tgsi_dst(struct nv30_vpc *vpc, const struct tgsi_full_dst_register *fdst) { struct nv30_sreg dst; - switch (fdst->DstRegister.File) { + switch (fdst->Register.File) { case TGSI_FILE_OUTPUT: dst = nv30_sr(NV30SR_OUTPUT, - vpc->output_map[fdst->DstRegister.Index]); + vpc->output_map[fdst->Register.Index]); break; case TGSI_FILE_TEMPORARY: - dst = nv30_sr(NV30SR_TEMP, fdst->DstRegister.Index); + dst = nv30_sr(NV30SR_TEMP, fdst->Register.Index); if (vpc->high_temp < dst.index) vpc->high_temp = dst.index; break; @@ -334,8 +334,8 @@ nv30_vertprog_parse_instruction(struct nv30_vpc *vpc, for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { const struct tgsi_full_src_register *fsrc; - fsrc = &finst->FullSrcRegisters[i]; - if (fsrc->SrcRegister.File == TGSI_FILE_TEMPORARY) { + fsrc = &finst->Src[i]; + if (fsrc->Register.File == TGSI_FILE_TEMPORARY) { src[i] = tgsi_src(vpc, fsrc); } } @@ -343,11 +343,11 @@ nv30_vertprog_parse_instruction(struct nv30_vpc *vpc, for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { const struct tgsi_full_src_register *fsrc; - fsrc = &finst->FullSrcRegisters[i]; - switch (fsrc->SrcRegister.File) { + fsrc = &finst->Src[i]; + switch (fsrc->Register.File) { case TGSI_FILE_INPUT: - if (ai == -1 || ai == fsrc->SrcRegister.Index) { - ai = fsrc->SrcRegister.Index; + if (ai == -1 || ai == fsrc->Register.Index) { + ai = fsrc->Register.Index; src[i] = tgsi_src(vpc, fsrc); } else { src[i] = temp(vpc); @@ -360,8 +360,8 @@ nv30_vertprog_parse_instruction(struct nv30_vpc *vpc, */ case TGSI_FILE_CONSTANT: case TGSI_FILE_IMMEDIATE: - if (ci == -1 || ci == fsrc->SrcRegister.Index) { - ci = fsrc->SrcRegister.Index; + if (ci == -1 || ci == fsrc->Register.Index) { + ci = fsrc->Register.Index; src[i] = tgsi_src(vpc, fsrc); } else { src[i] = temp(vpc); @@ -378,8 +378,8 @@ nv30_vertprog_parse_instruction(struct nv30_vpc *vpc, } } - dst = tgsi_dst(vpc, &finst->FullDstRegisters[0]); - mask = tgsi_mask(finst->FullDstRegisters[0].DstRegister.WriteMask); + dst = tgsi_dst(vpc, &finst->Dst[0]); + mask = tgsi_mask(finst->Dst[0].Register.WriteMask); switch (finst->Instruction.Opcode) { case TGSI_OPCODE_ABS: @@ -490,15 +490,15 @@ nv30_vertprog_parse_decl_output(struct nv30_vpc *vpc, { int hw; - switch (fdec->Semantic.SemanticName) { + switch (fdec->Semantic.Name) { case TGSI_SEMANTIC_POSITION: hw = NV30_VP_INST_DEST_POS; break; case TGSI_SEMANTIC_COLOR: - if (fdec->Semantic.SemanticIndex == 0) { + if (fdec->Semantic.Index == 0) { hw = NV30_VP_INST_DEST_COL0; } else - if (fdec->Semantic.SemanticIndex == 1) { + if (fdec->Semantic.Index == 1) { hw = NV30_VP_INST_DEST_COL1; } else { NOUVEAU_ERR("bad colour semantic index\n"); @@ -506,10 +506,10 @@ nv30_vertprog_parse_decl_output(struct nv30_vpc *vpc, } break; case TGSI_SEMANTIC_BCOLOR: - if (fdec->Semantic.SemanticIndex == 0) { + if (fdec->Semantic.Index == 0) { hw = NV30_VP_INST_DEST_BFC0; } else - if (fdec->Semantic.SemanticIndex == 1) { + if (fdec->Semantic.Index == 1) { hw = NV30_VP_INST_DEST_BFC1; } else { NOUVEAU_ERR("bad bcolour semantic index\n"); @@ -523,19 +523,22 @@ nv30_vertprog_parse_decl_output(struct nv30_vpc *vpc, hw = NV30_VP_INST_DEST_PSZ; break; case TGSI_SEMANTIC_GENERIC: - if (fdec->Semantic.SemanticIndex <= 7) { - hw = NV30_VP_INST_DEST_TC(fdec->Semantic.SemanticIndex); + if (fdec->Semantic.Index <= 7) { + hw = NV30_VP_INST_DEST_TC(fdec->Semantic.Index); } else { NOUVEAU_ERR("bad generic semantic index\n"); return FALSE; } break; + case TGSI_SEMANTIC_EDGEFLAG: + NOUVEAU_ERR("cannot handle edgeflag output\n"); + return FALSE; default: NOUVEAU_ERR("bad output semantic\n"); return FALSE; } - vpc->output_map[fdec->DeclarationRange.First] = hw; + vpc->output_map[fdec->Range.First] = hw; return TRUE; } @@ -647,7 +650,9 @@ static boolean nv30_vertprog_validate(struct nv30_context *nv30) { struct pipe_screen *pscreen = nv30->pipe.screen; - struct nouveau_grobj *rankine = nv30->screen->rankine; + struct nv30_screen *screen = nv30->screen; + struct nouveau_channel *chan = screen->base.channel; + struct nouveau_grobj *rankine = screen->rankine; struct nv30_vertex_program *vp; struct pipe_buffer *constbuf; boolean upload_code = FALSE, upload_data = FALSE; @@ -681,7 +686,7 @@ nv30_vertprog_validate(struct nv30_context *nv30) assert(0); } - so = so_new(2, 0); + so = so_new(1, 1, 0); so_method(so, rankine, NV34TCL_VP_START_FROM_ID, 1); so_data (so, vp->exec->start); so_ref(so, &vp->so); @@ -767,9 +772,9 @@ nv30_vertprog_validate(struct nv30_context *nv30) 4 * sizeof(float)); } - BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_CONST_ID, 5); - OUT_RING (i + vp->data->start); - OUT_RINGp ((uint32_t *)vpd->value, 4); + BEGIN_RING(chan, rankine, NV34TCL_VP_UPLOAD_CONST_ID, 5); + OUT_RING (chan, i + vp->data->start); + OUT_RINGp (chan, (uint32_t *)vpd->value, 4); } if (constbuf) @@ -785,11 +790,11 @@ nv30_vertprog_validate(struct nv30_context *nv30) vp->insns[i].data[2], vp->insns[i].data[3]); } #endif - BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_FROM_ID, 1); - OUT_RING (vp->exec->start); + BEGIN_RING(chan, rankine, NV34TCL_VP_UPLOAD_FROM_ID, 1); + OUT_RING (chan, vp->exec->start); for (i = 0; i < vp->nr_insns; i++) { - BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_INST(0), 4); - OUT_RINGp (vp->insns[i].data, 4); + BEGIN_RING(chan, rankine, NV34TCL_VP_UPLOAD_INST(0), 4); + OUT_RINGp (chan, vp->insns[i].data, 4); } } |