/* * Copyright 2002-2003 Jason Edmeades * Copyright 2002-2003 Raphael Junqueira * Copyright 2004 Christian Costa * Copyright 2005 Oliver Stieber * Copyright 2006 Ivan Gyurdiev * Copyright 2007-2008, 2013 Stefan Dösinger for CodeWeavers * Copyright 2009-2011 Henri Verbeet for CodeWeavers * * This library is free software; you can redistribute it and/or * modify it under the terms of the GNU Lesser General Public * License as published by the Free Software Foundation; either * version 2.1 of the License, or (at your option) any later version. * * This library is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU * Lesser General Public License for more details. * * You should have received a copy of the GNU Lesser General Public * License along with this library; if not, write to the Free Software * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA */ #include "config.h" #include #include #include "wined3d_private.h" WINE_DEFAULT_DEBUG_CHANNEL(d3d_shader); const struct wined3d_vec4 wined3d_srgb_const[] = { /* pow, mul_high, sub_high, mul_low */ {0.41666f, 1.055f, 0.055f, 12.92f}, /* cmp */ {0.0031308f, 0.0f, 0.0f, 0.0f}, }; static const char * const shader_opcode_names[] = { /* WINED3DSIH_ABS */ "abs", /* WINED3DSIH_ADD */ "add", /* WINED3DSIH_AND */ "and", /* WINED3DSIH_ATOMIC_AND */ "atomic_and", /* WINED3DSIH_ATOMIC_CMP_STORE */ "atomic_cmp_store", /* WINED3DSIH_ATOMIC_IADD */ "atomic_iadd", /* WINED3DSIH_ATOMIC_IMAX */ "atomic_imax", /* WINED3DSIH_ATOMIC_IMIN */ "atomic_imin", /* WINED3DSIH_ATOMIC_OR */ "atomic_or", /* WINED3DSIH_ATOMIC_UMAX */ "atomic_umax", /* WINED3DSIH_ATOMIC_UMIN */ "atomic_umin", /* WINED3DSIH_ATOMIC_XOR */ "atomic_xor", /* WINED3DSIH_BEM */ "bem", /* WINED3DSIH_BFI */ "bfi", /* WINED3DSIH_BFREV */ "bfrev", /* WINED3DSIH_BREAK */ "break", /* WINED3DSIH_BREAKC */ "breakc", /* WINED3DSIH_BREAKP */ "breakp", /* WINED3DSIH_BUFINFO */ "bufinfo", /* WINED3DSIH_CALL */ "call", /* WINED3DSIH_CALLNZ */ "callnz", /* WINED3DSIH_CASE */ "case", /* WINED3DSIH_CMP */ "cmp", /* WINED3DSIH_CND */ "cnd", /* WINED3DSIH_CONTINUE */ "continue", /* WINED3DSIH_CONTINUEP */ "continuec", /* WINED3DSIH_COUNTBITS */ "countbits", /* WINED3DSIH_CRS */ "crs", /* WINED3DSIH_CUT */ "cut", /* WINED3DSIH_CUT_STREAM */ "cut_stream", /* WINED3DSIH_DCL */ "dcl", /* WINED3DSIH_DCL_CONSTANT_BUFFER */ "dcl_constantBuffer", /* WINED3DSIH_DCL_FUNCTION_BODY */ "dcl_function_body", /* WINED3DSIH_DCL_FUNCTION_TABLE */ "dcl_function_table", /* WINED3DSIH_DCL_GLOBAL_FLAGS */ "dcl_globalFlags", /* WINED3DSIH_DCL_GS_INSTANCES */ "dcl_gs_instances", /* WINED3DSIH_DCL_HS_FORK_PHASE_INSTANCE_COUNT */ "dcl_hs_fork_phase_instance_count", /* WINED3DSIH_DCL_HS_JOIN_PHASE_INSTANCE_COUNT */ "dcl_hs_join_phase_instance_count", /* WINED3DSIH_DCL_HS_MAX_TESSFACTOR */ "dcl_hs_max_tessfactor", /* WINED3DSIH_DCL_IMMEDIATE_CONSTANT_BUFFER */ "dcl_immediateConstantBuffer", /* WINED3DSIH_DCL_INDEX_RANGE */ "dcl_index_range", /* WINED3DSIH_DCL_INDEXABLE_TEMP */ "dcl_indexableTemp", /* WINED3DSIH_DCL_INPUT */ "dcl_input", /* WINED3DSIH_DCL_INPUT_CONTROL_POINT_COUNT */ "dcl_input_control_point_count", /* WINED3DSIH_DCL_INPUT_PRIMITIVE */ "dcl_inputPrimitive", /* WINED3DSIH_DCL_INPUT_PS */ "dcl_input_ps", /* WINED3DSIH_DCL_INPUT_PS_SGV */ "dcl_input_ps_sgv", /* WINED3DSIH_DCL_INPUT_PS_SIV */ "dcl_input_ps_siv", /* WINED3DSIH_DCL_INPUT_SGV */ "dcl_input_sgv", /* WINED3DSIH_DCL_INPUT_SIV */ "dcl_input_siv", /* WINED3DSIH_DCL_INTERFACE */ "dcl_interface", /* WINED3DSIH_DCL_OUTPUT */ "dcl_output", /* WINED3DSIH_DCL_OUTPUT_CONTROL_POINT_COUNT */ "dcl_output_control_point_count", /* WINED3DSIH_DCL_OUTPUT_SIV */ "dcl_output_siv", /* WINED3DSIH_DCL_OUTPUT_TOPOLOGY */ "dcl_outputTopology", /* WINED3DSIH_DCL_RESOURCE_RAW */ "dcl_resource_raw", /* WINED3DSIH_DCL_RESOURCE_STRUCTURED */ "dcl_resource_structured", /* WINED3DSIH_DCL_SAMPLER */ "dcl_sampler", /* WINED3DSIH_DCL_STREAM */ "dcl_stream", /* WINED3DSIH_DCL_TEMPS */ "dcl_temps", /* WINED3DSIH_DCL_TESSELLATOR_DOMAIN */ "dcl_tessellator_domain", /* WINED3DSIH_DCL_TESSELLATOR_OUTPUT_PRIMITIVE */ "dcl_tessellator_output_primitive", /* WINED3DSIH_DCL_TESSELLATOR_PARTITIONING */ "dcl_tessellator_partitioning", /* WINED3DSIH_DCL_TGSM_RAW */ "dcl_tgsm_raw", /* WINED3DSIH_DCL_TGSM_STRUCTURED */ "dcl_tgsm_structured", /* WINED3DSIH_DCL_THREAD_GROUP */ "dcl_thread_group", /* WINED3DSIH_DCL_UAV_RAW */ "dcl_uav_raw", /* WINED3DSIH_DCL_UAV_STRUCTURED */ "dcl_uav_structured", /* WINED3DSIH_DCL_UAV_TYPED */ "dcl_uav_typed", /* WINED3DSIH_DCL_VERTICES_OUT */ "dcl_maxOutputVertexCount", /* WINED3DSIH_DEF */ "def", /* WINED3DSIH_DEFAULT */ "default", /* WINED3DSIH_DEFB */ "defb", /* WINED3DSIH_DEFI */ "defi", /* WINED3DSIH_DIV */ "div", /* WINED3DSIH_DP2 */ "dp2", /* WINED3DSIH_DP2ADD */ "dp2add", /* WINED3DSIH_DP3 */ "dp3", /* WINED3DSIH_DP4 */ "dp4", /* WINED3DSIH_DST */ "dst", /* WINED3DSIH_DSX */ "dsx", /* WINED3DSIH_DSX_COARSE */ "deriv_rtx_coarse", /* WINED3DSIH_DSX_FINE */ "deriv_rtx_fine", /* WINED3DSIH_DSY */ "dsy", /* WINED3DSIH_DSY_COARSE */ "deriv_rty_coarse", /* WINED3DSIH_DSY_FINE */ "deriv_rty_fine", /* WINED3DSIH_ELSE */ "else", /* WINED3DSIH_EMIT */ "emit", /* WINED3DSIH_EMIT_STREAM */ "emit_stream", /* WINED3DSIH_ENDIF */ "endif", /* WINED3DSIH_ENDLOOP */ "endloop", /* WINED3DSIH_ENDREP */ "endrep", /* WINED3DSIH_ENDSWITCH */ "endswitch", /* WINED3DSIH_EQ */ "eq", /* WINED3DSIH_EVAL_SAMPLE_INDEX */ "eval_sample_index", /* WINED3DSIH_EXP */ "exp", /* WINED3DSIH_EXPP */ "expp", /* WINED3DSIH_F16TOF32 */ "f16tof32", /* WINED3DSIH_F32TOF16 */ "f32tof16", /* WINED3DSIH_FCALL */ "fcall", /* WINED3DSIH_FIRSTBIT_HI */ "firstbit_hi", /* WINED3DSIH_FIRSTBIT_LO */ "firstbit_lo", /* WINED3DSIH_FIRSTBIT_SHI */ "firstbit_shi", /* WINED3DSIH_FRC */ "frc", /* WINED3DSIH_FTOI */ "ftoi", /* WINED3DSIH_FTOU */ "ftou", /* WINED3DSIH_GATHER4 */ "gather4", /* WINED3DSIH_GATHER4_C */ "gather4_c", /* WINED3DSIH_GATHER4_PO */ "gather4_po", /* WINED3DSIH_GATHER4_PO_C */ "gather4_po_c", /* WINED3DSIH_GE */ "ge", /* WINED3DSIH_HS_CONTROL_POINT_PHASE */ "hs_control_point_phase", /* WINED3DSIH_HS_DECLS */ "hs_decls", /* WINED3DSIH_HS_FORK_PHASE */ "hs_fork_phase", /* WINED3DSIH_HS_JOIN_PHASE */ "hs_join_phase", /* WINED3DSIH_IADD */ "iadd", /* WINED3DSIH_IBFE */ "ibfe", /* WINED3DSIH_IEQ */ "ieq", /* WINED3DSIH_IF */ "if", /* WINED3DSIH_IFC */ "ifc", /* WINED3DSIH_IGE */ "ige", /* WINED3DSIH_ILT */ "ilt", /* WINED3DSIH_IMAD */ "imad", /* WINED3DSIH_IMAX */ "imax", /* WINED3DSIH_IMIN */ "imin", /* WINED3DSIH_IMM_ATOMIC_ALLOC */ "imm_atomic_alloc", /* WINED3DSIH_IMM_ATOMIC_AND */ "imm_atomic_and", /* WINED3DSIH_IMM_ATOMIC_CMP_EXCH */ "imm_atomic_cmp_exch", /* WINED3DSIH_IMM_ATOMIC_CONSUME */ "imm_atomic_consume", /* WINED3DSIH_IMM_ATOMIC_EXCH */ "imm_atomic_exch", /* WINED3DSIH_IMM_ATOMIC_IADD */ "imm_atomic_iadd", /* WINED3DSIH_IMM_ATOMIC_IMAX */ "imm_atomic_imax", /* WINED3DSIH_IMM_ATOMIC_IMIN */ "imm_atomic_imin", /* WINED3DSIH_IMM_ATOMIC_OR */ "imm_atomic_or", /* WINED3DSIH_IMM_ATOMIC_UMAX */ "imm_atomic_umax", /* WINED3DSIH_IMM_ATOMIC_UMIN */ "imm_atomic_umin", /* WINED3DSIH_IMM_ATOMIC_XOR */ "imm_atomic_xor", /* WINED3DSIH_IMUL */ "imul", /* WINED3DSIH_INE */ "ine", /* WINED3DSIH_INEG */ "ineg", /* WINED3DSIH_ISHL */ "ishl", /* WINED3DSIH_ISHR */ "ishr", /* WINED3DSIH_ITOF */ "itof", /* WINED3DSIH_LABEL */ "label", /* WINED3DSIH_LD */ "ld", /* WINED3DSIH_LD2DMS */ "ld2dms", /* WINED3DSIH_LD_RAW */ "ld_raw", /* WINED3DSIH_LD_STRUCTURED */ "ld_structured", /* WINED3DSIH_LD_UAV_TYPED */ "ld_uav_typed", /* WINED3DSIH_LIT */ "lit", /* WINED3DSIH_LOD */ "lod", /* WINED3DSIH_LOG */ "log", /* WINED3DSIH_LOGP */ "logp", /* WINED3DSIH_LOOP */ "loop", /* WINED3DSIH_LRP */ "lrp", /* WINED3DSIH_LT */ "lt", /* WINED3DSIH_M3x2 */ "m3x2", /* WINED3DSIH_M3x3 */ "m3x3", /* WINED3DSIH_M3x4 */ "m3x4", /* WINED3DSIH_M4x3 */ "m4x3", /* WINED3DSIH_M4x4 */ "m4x4", /* WINED3DSIH_MAD */ "mad", /* WINED3DSIH_MAX */ "max", /* WINED3DSIH_MIN */ "min", /* WINED3DSIH_MOV */ "mov", /* WINED3DSIH_MOVA */ "mova", /* WINED3DSIH_MOVC */ "movc", /* WINED3DSIH_MUL */ "mul", /* WINED3DSIH_NE */ "ne", /* WINED3DSIH_NOP */ "nop", /* WINED3DSIH_NOT */ "not", /* WINED3DSIH_NRM */ "nrm", /* WINED3DSIH_OR */ "or", /* WINED3DSIH_PHASE */ "phase", /* WINED3DSIH_POW */ "pow", /* WINED3DSIH_RCP */ "rcp", /* WINED3DSIH_REP */ "rep", /* WINED3DSIH_RESINFO */ "resinfo", /* WINED3DSIH_RET */ "ret", /* WINED3DSIH_RETP */ "retp", /* WINED3DSIH_ROUND_NE */ "round_ne", /* WINED3DSIH_ROUND_NI */ "round_ni", /* WINED3DSIH_ROUND_PI */ "round_pi", /* WINED3DSIH_ROUND_Z */ "round_z", /* WINED3DSIH_RSQ */ "rsq", /* WINED3DSIH_SAMPLE */ "sample", /* WINED3DSIH_SAMPLE_B */ "sample_b", /* WINED3DSIH_SAMPLE_C */ "sample_c", /* WINED3DSIH_SAMPLE_C_LZ */ "sample_c_lz", /* WINED3DSIH_SAMPLE_GRAD */ "sample_d", /* WINED3DSIH_SAMPLE_INFO */ "sample_info", /* WINED3DSIH_SAMPLE_LOD */ "sample_l", /* WINED3DSIH_SAMPLE_POS */ "sample_pos", /* WINED3DSIH_SETP */ "setp", /* WINED3DSIH_SGE */ "sge", /* WINED3DSIH_SGN */ "sgn", /* WINED3DSIH_SINCOS */ "sincos", /* WINED3DSIH_SLT */ "slt", /* WINED3DSIH_SQRT */ "sqrt", /* WINED3DSIH_STORE_RAW */ "store_raw", /* WINED3DSIH_STORE_STRUCTURED */ "store_structured", /* WINED3DSIH_STORE_UAV_TYPED */ "store_uav_typed", /* WINED3DSIH_SUB */ "sub", /* WINED3DSIH_SWAPC */ "swapc", /* WINED3DSIH_SWITCH */ "switch", /* WINED3DSIH_SYNC */ "sync", /* WINED3DSIH_TEX */ "texld", /* WINED3DSIH_TEXBEM */ "texbem", /* WINED3DSIH_TEXBEML */ "texbeml", /* WINED3DSIH_TEXCOORD */ "texcrd", /* WINED3DSIH_TEXDEPTH */ "texdepth", /* WINED3DSIH_TEXDP3 */ "texdp3", /* WINED3DSIH_TEXDP3TEX */ "texdp3tex", /* WINED3DSIH_TEXKILL */ "texkill", /* WINED3DSIH_TEXLDD */ "texldd", /* WINED3DSIH_TEXLDL */ "texldl", /* WINED3DSIH_TEXM3x2DEPTH */ "texm3x2depth", /* WINED3DSIH_TEXM3x2PAD */ "texm3x2pad", /* WINED3DSIH_TEXM3x2TEX */ "texm3x2tex", /* WINED3DSIH_TEXM3x3 */ "texm3x3", /* WINED3DSIH_TEXM3x3DIFF */ "texm3x3diff", /* WINED3DSIH_TEXM3x3PAD */ "texm3x3pad", /* WINED3DSIH_TEXM3x3SPEC */ "texm3x3spec", /* WINED3DSIH_TEXM3x3TEX */ "texm3x3tex", /* WINED3DSIH_TEXM3x3VSPEC */ "texm3x3vspec", /* WINED3DSIH_TEXREG2AR */ "texreg2ar", /* WINED3DSIH_TEXREG2GB */ "texreg2gb", /* WINED3DSIH_TEXREG2RGB */ "texreg2rgb", /* WINED3DSIH_UBFE */ "ubfe", /* WINED3DSIH_UDIV */ "udiv", /* WINED3DSIH_UGE */ "uge", /* WINED3DSIH_ULT */ "ult", /* WINED3DSIH_UMAX */ "umax", /* WINED3DSIH_UMIN */ "umin", /* WINED3DSIH_UMUL */ "umul", /* WINED3DSIH_USHR */ "ushr", /* WINED3DSIH_UTOF */ "utof", /* WINED3DSIH_XOR */ "xor", }; static const char * const semantic_names[] = { /* WINED3D_DECL_USAGE_POSITION */ "SV_POSITION", /* WINED3D_DECL_USAGE_BLEND_WEIGHT */ "BLENDWEIGHT", /* WINED3D_DECL_USAGE_BLEND_INDICES */ "BLENDINDICES", /* WINED3D_DECL_USAGE_NORMAL */ "NORMAL", /* WINED3D_DECL_USAGE_PSIZE */ "PSIZE", /* WINED3D_DECL_USAGE_TEXCOORD */ "TEXCOORD", /* WINED3D_DECL_USAGE_TANGENT */ "TANGENT", /* WINED3D_DECL_USAGE_BINORMAL */ "BINORMAL", /* WINED3D_DECL_USAGE_TESS_FACTOR */ "TESSFACTOR", /* WINED3D_DECL_USAGE_POSITIONT */ "POSITIONT", /* WINED3D_DECL_USAGE_COLOR */ "COLOR", /* WINED3D_DECL_USAGE_FOG */ "FOG", /* WINED3D_DECL_USAGE_DEPTH */ "DEPTH", /* WINED3D_DECL_USAGE_SAMPLE */ "SAMPLE", }; static const struct { enum wined3d_shader_input_sysval_semantic sysval_semantic; const char *sysval_name; } shader_input_sysval_semantic_names[] = { {WINED3D_SIV_POSITION, "position"}, {WINED3D_SIV_CLIP_DISTANCE, "clip_distance"}, {WINED3D_SIV_CULL_DISTANCE, "cull_distance"}, {WINED3D_SIV_RENDER_TARGET_ARRAY_INDEX, "render_target_array_index"}, {WINED3D_SIV_VIEWPORT_ARRAY_INDEX, "viewport_array_index"}, {WINED3D_SIV_VERTEX_ID, "vertex_id"}, {WINED3D_SIV_INSTANCE_ID, "instance_id"}, {WINED3D_SIV_PRIMITIVE_ID, "primitive_id"}, {WINED3D_SIV_IS_FRONT_FACE, "is_front_face"}, {WINED3D_SIV_SAMPLE_INDEX, "sample_index"}, {WINED3D_SIV_QUAD_U0_TESS_FACTOR, "finalQuadUeq0EdgeTessFactor"}, {WINED3D_SIV_QUAD_V0_TESS_FACTOR, "finalQuadVeq0EdgeTessFactor"}, {WINED3D_SIV_QUAD_U1_TESS_FACTOR, "finalQuadUeq1EdgeTessFactor"}, {WINED3D_SIV_QUAD_V1_TESS_FACTOR, "finalQuadVeq1EdgeTessFactor"}, {WINED3D_SIV_QUAD_U_INNER_TESS_FACTOR, "finalQuadUInsideTessFactor"}, {WINED3D_SIV_QUAD_V_INNER_TESS_FACTOR, "finalQuadVInsideTessFactor"}, {WINED3D_SIV_TRIANGLE_U_TESS_FACTOR, "finalTriUeq0EdgeTessFactor"}, {WINED3D_SIV_TRIANGLE_V_TESS_FACTOR, "finalTriVeq0EdgeTessFactor"}, {WINED3D_SIV_TRIANGLE_W_TESS_FACTOR, "finalTriWeq0EdgeTessFactor"}, {WINED3D_SIV_TRIANGLE_INNER_TESS_FACTOR, "finalTriInsideTessFactor"}, {WINED3D_SIV_LINE_DETAIL_TESS_FACTOR, "finalLineDetailTessFactor"}, {WINED3D_SIV_LINE_DENSITY_TESS_FACTOR, "finalLineDensityTessFactor"}, }; static void shader_dump_src_param(struct wined3d_string_buffer *buffer, const struct wined3d_shader_src_param *param, const struct wined3d_shader_version *shader_version); const char *debug_d3dshaderinstructionhandler(enum WINED3D_SHADER_INSTRUCTION_HANDLER handler_idx) { if (handler_idx >= ARRAY_SIZE(shader_opcode_names)) return wine_dbg_sprintf("UNRECOGNIZED(%#x)", handler_idx); return shader_opcode_names[handler_idx]; } static const char *shader_semantic_name_from_usage(enum wined3d_decl_usage usage) { if (usage >= ARRAY_SIZE(semantic_names)) { FIXME("Unrecognized usage %#x.\n", usage); return "UNRECOGNIZED"; } return semantic_names[usage]; } static enum wined3d_decl_usage shader_usage_from_semantic_name(const char *name) { unsigned int i; for (i = 0; i < ARRAY_SIZE(semantic_names); ++i) { if (!strcmp(name, semantic_names[i])) return i; } return ~0U; } static enum wined3d_sysval_semantic shader_sysval_semantic_from_usage(enum wined3d_decl_usage usage) { switch (usage) { case WINED3D_DECL_USAGE_POSITION: return WINED3D_SV_POSITION; default: return 0; } } BOOL shader_match_semantic(const char *semantic_name, enum wined3d_decl_usage usage) { return !strcmp(semantic_name, shader_semantic_name_from_usage(usage)); } static void shader_signature_from_semantic(struct wined3d_shader_signature_element *e, const struct wined3d_shader_semantic *s) { e->semantic_name = shader_semantic_name_from_usage(s->usage); e->semantic_idx = s->usage_idx; e->stream_idx = 0; e->sysval_semantic = shader_sysval_semantic_from_usage(s->usage); e->component_type = WINED3D_TYPE_FLOAT; e->register_idx = s->reg.reg.idx[0].offset; e->mask = s->reg.write_mask; } static void shader_signature_from_usage(struct wined3d_shader_signature_element *e, enum wined3d_decl_usage usage, UINT usage_idx, UINT reg_idx, DWORD write_mask) { e->semantic_name = shader_semantic_name_from_usage(usage); e->semantic_idx = usage_idx; e->stream_idx = 0; e->sysval_semantic = shader_sysval_semantic_from_usage(usage); e->component_type = WINED3D_TYPE_FLOAT; e->register_idx = reg_idx; e->mask = write_mask; } static const struct wined3d_shader_frontend *shader_select_frontend(enum wined3d_shader_byte_code_format format) { switch (format) { case WINED3D_SHADER_BYTE_CODE_FORMAT_SM1: return &sm1_shader_frontend; case WINED3D_SHADER_BYTE_CODE_FORMAT_SM4: return &sm4_shader_frontend; default: WARN("Invalid byte code format %#x specified.\n", format); return NULL; } } void string_buffer_clear(struct wined3d_string_buffer *buffer) { buffer->buffer[0] = '\0'; buffer->content_size = 0; } BOOL string_buffer_init(struct wined3d_string_buffer *buffer) { buffer->buffer_size = 32; if (!(buffer->buffer = heap_alloc(buffer->buffer_size))) { ERR("Failed to allocate shader buffer memory.\n"); return FALSE; } string_buffer_clear(buffer); return TRUE; } void string_buffer_free(struct wined3d_string_buffer *buffer) { heap_free(buffer->buffer); } BOOL string_buffer_resize(struct wined3d_string_buffer *buffer, int rc) { char *new_buffer; unsigned int new_buffer_size = buffer->buffer_size * 2; while (rc > 0 && (unsigned int)rc >= new_buffer_size - buffer->content_size) new_buffer_size *= 2; if (!(new_buffer = heap_realloc(buffer->buffer, new_buffer_size))) { ERR("Failed to grow buffer.\n"); buffer->buffer[buffer->content_size] = '\0'; return FALSE; } buffer->buffer = new_buffer; buffer->buffer_size = new_buffer_size; return TRUE; } int shader_vaddline(struct wined3d_string_buffer *buffer, const char *format, va_list args) { unsigned int rem; int rc; rem = buffer->buffer_size - buffer->content_size; rc = vsnprintf(&buffer->buffer[buffer->content_size], rem, format, args); if (rc < 0 /* C89 */ || (unsigned int)rc >= rem /* C99 */) return rc; buffer->content_size += rc; return 0; } int shader_addline(struct wined3d_string_buffer *buffer, const char *format, ...) { va_list args; int ret; for (;;) { va_start(args, format); ret = shader_vaddline(buffer, format, args); va_end(args); if (!ret) return ret; if (!string_buffer_resize(buffer, ret)) return -1; } } struct wined3d_string_buffer *string_buffer_get(struct wined3d_string_buffer_list *list) { struct wined3d_string_buffer *buffer; if (list_empty(&list->list)) { buffer = heap_alloc(sizeof(*buffer)); if (!buffer || !string_buffer_init(buffer)) { ERR("Couldn't allocate buffer for temporary string.\n"); heap_free(buffer); return NULL; } } else { buffer = LIST_ENTRY(list_head(&list->list), struct wined3d_string_buffer, entry); list_remove(&buffer->entry); } string_buffer_clear(buffer); return buffer; } static int string_buffer_vsprintf(struct wined3d_string_buffer *buffer, const char *format, va_list args) { if (!buffer) return 0; string_buffer_clear(buffer); return shader_vaddline(buffer, format, args); } void string_buffer_sprintf(struct wined3d_string_buffer *buffer, const char *format, ...) { va_list args; int ret; for (;;) { va_start(args, format); ret = string_buffer_vsprintf(buffer, format, args); va_end(args); if (!ret) return; if (!string_buffer_resize(buffer, ret)) return; } } void string_buffer_release(struct wined3d_string_buffer_list *list, struct wined3d_string_buffer *buffer) { if (!buffer) return; list_add_head(&list->list, &buffer->entry); } void string_buffer_list_init(struct wined3d_string_buffer_list *list) { list_init(&list->list); } void string_buffer_list_cleanup(struct wined3d_string_buffer_list *list) { struct wined3d_string_buffer *buffer, *buffer_next; LIST_FOR_EACH_ENTRY_SAFE(buffer, buffer_next, &list->list, struct wined3d_string_buffer, entry) { string_buffer_free(buffer); heap_free(buffer); } list_init(&list->list); } /* Convert floating point offset relative to a register file to an absolute * offset for float constants. */ static unsigned int shader_get_float_offset(enum wined3d_shader_register_type register_type, UINT register_idx) { switch (register_type) { case WINED3DSPR_CONST: return register_idx; case WINED3DSPR_CONST2: return 2048 + register_idx; case WINED3DSPR_CONST3: return 4096 + register_idx; case WINED3DSPR_CONST4: return 6144 + register_idx; default: FIXME("Unsupported register type: %u.\n", register_type); return register_idx; } } static void shader_delete_constant_list(struct list *clist) { struct wined3d_shader_lconst *constant, *constant_next; LIST_FOR_EACH_ENTRY_SAFE(constant, constant_next, clist, struct wined3d_shader_lconst, entry) heap_free(constant); list_init(clist); } static void shader_set_limits(struct wined3d_shader *shader) { static const struct limits_entry { unsigned int min_version; unsigned int max_version; struct wined3d_shader_limits limits; } vs_limits[] = { /* min_version, max_version, sampler, constant_int, constant_float, constant_bool, packed_output, packed_input */ {WINED3D_SHADER_VERSION(1, 0), WINED3D_SHADER_VERSION(1, 1), { 0, 0, 256, 0, 12, 0}}, {WINED3D_SHADER_VERSION(2, 0), WINED3D_SHADER_VERSION(2, 1), { 0, 16, 256, 16, 12, 0}}, /* DX10 cards on Windows advertise a D3D9 constant limit of 256 * even though they are capable of supporting much more (GL * drivers advertise 1024). d3d9.dll and d3d8.dll clamp the * wined3d-advertised maximum. Clamp the constant limit for <= 3.0 * shaders to 256. */ {WINED3D_SHADER_VERSION(3, 0), WINED3D_SHADER_VERSION(3, 0), { 4, 16, 256, 16, 12, 0}}, {WINED3D_SHADER_VERSION(4, 0), WINED3D_SHADER_VERSION(4, 0), {16, 0, 0, 0, 16, 0}}, {WINED3D_SHADER_VERSION(4, 1), WINED3D_SHADER_VERSION(5, 0), {16, 0, 0, 0, 32, 0}}, {0} }, hs_limits[] = { /* min_version, max_version, sampler, constant_int, constant_float, constant_bool, packed_output, packet_input */ {WINED3D_SHADER_VERSION(5, 0), WINED3D_SHADER_VERSION(5, 0), {16, 0, 0, 0, 32, 32}}, }, ds_limits[] = { /* min_version, max_version, sampler, constant_int, constant_float, constant_bool, packed_output, packet_input */ {WINED3D_SHADER_VERSION(5, 0), WINED3D_SHADER_VERSION(5, 0), {16, 0, 0, 0, 32, 32}}, }, gs_limits[] = { /* min_version, max_version, sampler, constant_int, constant_float, constant_bool, packed_output, packed_input */ {WINED3D_SHADER_VERSION(4, 0), WINED3D_SHADER_VERSION(4, 0), {16, 0, 0, 0, 32, 16}}, {WINED3D_SHADER_VERSION(4, 1), WINED3D_SHADER_VERSION(5, 0), {16, 0, 0, 0, 32, 32}}, {0} }, ps_limits[] = { /* min_version, max_version, sampler, constant_int, constant_float, constant_bool, packed_output, packed_input */ {WINED3D_SHADER_VERSION(1, 0), WINED3D_SHADER_VERSION(1, 3), { 4, 0, 8, 0, 0, 0}}, {WINED3D_SHADER_VERSION(1, 4), WINED3D_SHADER_VERSION(1, 4), { 6, 0, 8, 0, 0, 0}}, {WINED3D_SHADER_VERSION(2, 0), WINED3D_SHADER_VERSION(2, 0), {16, 0, 32, 0, 0, 0}}, {WINED3D_SHADER_VERSION(2, 1), WINED3D_SHADER_VERSION(2, 1), {16, 16, 32, 16, 0, 0}}, {WINED3D_SHADER_VERSION(3, 0), WINED3D_SHADER_VERSION(3, 0), {16, 16, 224, 16, 0, 10}}, {WINED3D_SHADER_VERSION(4, 0), WINED3D_SHADER_VERSION(5, 0), {16, 0, 0, 0, 0, 32}}, {0} }, cs_limits[] = { /* min_version, max_version, sampler, constant_int, constant_float, constant_bool, packed_output, packed_input */ {WINED3D_SHADER_VERSION(5, 0), WINED3D_SHADER_VERSION(5, 0), {16, 0, 0, 0, 0, 0}}, }; const struct limits_entry *limits_array; DWORD shader_version = WINED3D_SHADER_VERSION(shader->reg_maps.shader_version.major, shader->reg_maps.shader_version.minor); int i = 0; switch (shader->reg_maps.shader_version.type) { default: FIXME("Unexpected shader type %u found.\n", shader->reg_maps.shader_version.type); /* Fall-through. */ case WINED3D_SHADER_TYPE_VERTEX: limits_array = vs_limits; break; case WINED3D_SHADER_TYPE_HULL: limits_array = hs_limits; break; case WINED3D_SHADER_TYPE_DOMAIN: limits_array = ds_limits; break; case WINED3D_SHADER_TYPE_GEOMETRY: limits_array = gs_limits; break; case WINED3D_SHADER_TYPE_PIXEL: limits_array = ps_limits; break; case WINED3D_SHADER_TYPE_COMPUTE: limits_array = cs_limits; break; } while (limits_array[i].min_version && limits_array[i].min_version <= shader_version) { if (shader_version <= limits_array[i].max_version) { shader->limits = &limits_array[i].limits; break; } ++i; } if (!shader->limits) { FIXME("Unexpected shader version \"%u.%u\".\n", shader->reg_maps.shader_version.major, shader->reg_maps.shader_version.minor); shader->limits = &limits_array[max(0, i - 1)].limits; } } static BOOL shader_record_register_usage(struct wined3d_shader *shader, struct wined3d_shader_reg_maps *reg_maps, const struct wined3d_shader_register *reg, enum wined3d_shader_type shader_type, unsigned int constf_size) { switch (reg->type) { case WINED3DSPR_TEXTURE: /* WINED3DSPR_ADDR */ if (shader_type == WINED3D_SHADER_TYPE_PIXEL) reg_maps->texcoord |= 1u << reg->idx[0].offset; else reg_maps->address |= 1u << reg->idx[0].offset; break; case WINED3DSPR_TEMP: reg_maps->temporary |= 1u << reg->idx[0].offset; break; case WINED3DSPR_INPUT: if (reg->idx[0].rel_addr) reg_maps->input_rel_addressing = 1; if (shader_type == WINED3D_SHADER_TYPE_PIXEL) { /* If relative addressing is used, we must assume that all * registers are used. Even if it is a construct like v3[aL], * we can't assume that v0, v1 and v2 aren't read because aL * can be negative. */ if (reg->idx[0].rel_addr) shader->u.ps.input_reg_used = ~0u; else shader->u.ps.input_reg_used |= 1u << reg->idx[0].offset; } else { reg_maps->input_registers |= 1u << reg->idx[0].offset; } break; case WINED3DSPR_RASTOUT: if (reg->idx[0].offset == 1) reg_maps->fog = 1; if (reg->idx[0].offset == 2) reg_maps->point_size = 1; break; case WINED3DSPR_MISCTYPE: if (shader_type == WINED3D_SHADER_TYPE_PIXEL) { if (!reg->idx[0].offset) reg_maps->vpos = 1; else if (reg->idx[0].offset == 1) reg_maps->usesfacing = 1; } break; case WINED3DSPR_CONST: if (reg->idx[0].rel_addr) { if (reg->idx[0].offset < reg_maps->min_rel_offset) reg_maps->min_rel_offset = reg->idx[0].offset; if (reg->idx[0].offset > reg_maps->max_rel_offset) reg_maps->max_rel_offset = reg->idx[0].offset; reg_maps->usesrelconstF = TRUE; } else { if (reg->idx[0].offset >= min(shader->limits->constant_float, constf_size)) { WARN("Shader using float constant %u which is not supported.\n", reg->idx[0].offset); return FALSE; } else { wined3d_insert_bits(reg_maps->constf, reg->idx[0].offset, 1, 0x1); } } break; case WINED3DSPR_CONSTINT: if (reg->idx[0].offset >= shader->limits->constant_int) { WARN("Shader using integer constant %u which is not supported.\n", reg->idx[0].offset); return FALSE; } else { reg_maps->integer_constants |= (1u << reg->idx[0].offset); } break; case WINED3DSPR_CONSTBOOL: if (reg->idx[0].offset >= shader->limits->constant_bool) { WARN("Shader using bool constant %u which is not supported.\n", reg->idx[0].offset); return FALSE; } else { reg_maps->boolean_constants |= (1u << reg->idx[0].offset); } break; case WINED3DSPR_COLOROUT: reg_maps->rt_mask |= (1u << reg->idx[0].offset); break; case WINED3DSPR_OUTCONTROLPOINT: reg_maps->vocp = 1; break; case WINED3DSPR_SAMPLEMASK: reg_maps->sample_mask = 1; break; default: TRACE("Not recording register of type %#x and [%#x][%#x].\n", reg->type, reg->idx[0].offset, reg->idx[1].offset); break; } return TRUE; } static void shader_record_sample(struct wined3d_shader_reg_maps *reg_maps, unsigned int resource_idx, unsigned int sampler_idx, unsigned int bind_idx) { struct wined3d_shader_sampler_map_entry *entries, *entry; struct wined3d_shader_sampler_map *map; unsigned int i; map = ®_maps->sampler_map; entries = map->entries; for (i = 0; i < map->count; ++i) { if (entries[i].resource_idx == resource_idx && entries[i].sampler_idx == sampler_idx) return; } if (!map->size) { if (!(entries = heap_calloc(4, sizeof(*entries)))) { ERR("Failed to allocate sampler map entries.\n"); return; } map->size = 4; map->entries = entries; } else if (map->count == map->size) { size_t new_size = map->size * 2; if (sizeof(*entries) * new_size <= sizeof(*entries) * map->size || !(entries = heap_realloc(entries, sizeof(*entries) * new_size))) { ERR("Failed to resize sampler map entries.\n"); return; } map->size = new_size; map->entries = entries; } entry = &entries[map->count++]; entry->resource_idx = resource_idx; entry->sampler_idx = sampler_idx; entry->bind_idx = bind_idx; } static unsigned int get_instr_extra_regcount(enum WINED3D_SHADER_INSTRUCTION_HANDLER instr, unsigned int param) { switch (instr) { case WINED3DSIH_M4x4: case WINED3DSIH_M3x4: return param == 1 ? 3 : 0; case WINED3DSIH_M4x3: case WINED3DSIH_M3x3: return param == 1 ? 2 : 0; case WINED3DSIH_M3x2: return param == 1 ? 1 : 0; default: return 0; } } static HRESULT shader_reg_maps_add_tgsm(struct wined3d_shader_reg_maps *reg_maps, unsigned int register_idx, unsigned int size, unsigned int stride) { struct wined3d_shader_tgsm *tgsm; if (register_idx >= MAX_TGSM_REGISTERS) { ERR("Invalid TGSM register index %u.\n", register_idx); return S_OK; } if (reg_maps->shader_version.type != WINED3D_SHADER_TYPE_COMPUTE) { FIXME("TGSM declarations are allowed only in compute shaders.\n"); return S_OK; } if (!wined3d_array_reserve((void **)®_maps->tgsm, ®_maps->tgsm_capacity, register_idx + 1, sizeof(*reg_maps->tgsm))) return E_OUTOFMEMORY; reg_maps->tgsm_count = max(register_idx + 1, reg_maps->tgsm_count); tgsm = ®_maps->tgsm[register_idx]; tgsm->size = size; tgsm->stride = stride; return S_OK; } static HRESULT shader_record_shader_phase(struct wined3d_shader *shader, struct wined3d_shader_phase **current_phase, const struct wined3d_shader_instruction *ins, const DWORD *current_instruction_ptr, const DWORD *previous_instruction_ptr) { struct wined3d_shader_phase *phase; if ((phase = *current_phase)) { phase->end = previous_instruction_ptr; *current_phase = NULL; } if (shader->reg_maps.shader_version.type != WINED3D_SHADER_TYPE_HULL) { ERR("Unexpected shader type %s.\n", debug_shader_type(shader->reg_maps.shader_version.type)); return E_FAIL; } switch (ins->handler_idx) { case WINED3DSIH_HS_CONTROL_POINT_PHASE: if (shader->u.hs.phases.control_point) { FIXME("Multiple control point phases.\n"); heap_free(shader->u.hs.phases.control_point); } if (!(shader->u.hs.phases.control_point = heap_alloc_zero(sizeof(*shader->u.hs.phases.control_point)))) return E_OUTOFMEMORY; phase = shader->u.hs.phases.control_point; break; case WINED3DSIH_HS_FORK_PHASE: if (!wined3d_array_reserve((void **)&shader->u.hs.phases.fork, &shader->u.hs.phases.fork_size, shader->u.hs.phases.fork_count + 1, sizeof(*shader->u.hs.phases.fork))) return E_OUTOFMEMORY; phase = &shader->u.hs.phases.fork[shader->u.hs.phases.fork_count++]; break; case WINED3DSIH_HS_JOIN_PHASE: if (!wined3d_array_reserve((void **)&shader->u.hs.phases.join, &shader->u.hs.phases.join_size, shader->u.hs.phases.join_count + 1, sizeof(*shader->u.hs.phases.join))) return E_OUTOFMEMORY; phase = &shader->u.hs.phases.join[shader->u.hs.phases.join_count++]; break; default: ERR("Unexpected opcode %s.\n", debug_d3dshaderinstructionhandler(ins->handler_idx)); return E_FAIL; } phase->start = current_instruction_ptr; *current_phase = phase; return WINED3D_OK; } static HRESULT shader_calculate_clip_or_cull_distance_mask( const struct wined3d_shader_signature_element *e, unsigned int *mask) { /* Clip and cull distances are packed in 4 component registers. 0 and 1 are * the only allowed semantic indices. */ if (e->semantic_idx >= WINED3D_MAX_CLIP_DISTANCES / 4) { *mask = 0; WARN("Invalid clip/cull distance index %u.\n", e->semantic_idx); return WINED3DERR_INVALIDCALL; } *mask = (e->mask & WINED3DSP_WRITEMASK_ALL) << (4 * e->semantic_idx); return WINED3D_OK; } static void wined3d_insert_interpolation_mode(DWORD *packed_interpolation_mode, unsigned int register_idx, enum wined3d_shader_interpolation_mode mode) { if (mode > WINED3DSIM_LINEAR_NOPERSPECTIVE_SAMPLE) FIXME("Unexpected interpolation mode %#x.\n", mode); wined3d_insert_bits(packed_interpolation_mode, register_idx * WINED3D_PACKED_INTERPOLATION_BIT_COUNT, WINED3D_PACKED_INTERPOLATION_BIT_COUNT, mode); } static HRESULT shader_scan_output_signature(struct wined3d_shader *shader) { const struct wined3d_shader_signature *output_signature = &shader->output_signature; struct wined3d_shader_reg_maps *reg_maps = &shader->reg_maps; unsigned int i; HRESULT hr; for (i = 0; i < output_signature->element_count; ++i) { const struct wined3d_shader_signature_element *e = &output_signature->elements[i]; unsigned int mask; reg_maps->output_registers |= 1u << e->register_idx; if (e->sysval_semantic == WINED3D_SV_CLIP_DISTANCE) { if (FAILED(hr = shader_calculate_clip_or_cull_distance_mask(e, &mask))) return hr; reg_maps->clip_distance_mask |= mask; } else if (e->sysval_semantic == WINED3D_SV_CULL_DISTANCE) { if (FAILED(hr = shader_calculate_clip_or_cull_distance_mask(e, &mask))) return hr; reg_maps->cull_distance_mask |= mask; } else if (e->sysval_semantic == WINED3D_SV_VIEWPORT_ARRAY_INDEX) { reg_maps->viewport_array = 1; } } return WINED3D_OK; } /* Note that this does not count the loop register as an address register. */ static HRESULT shader_get_registers_used(struct wined3d_shader *shader, DWORD constf_size) { struct wined3d_shader_signature_element input_signature_elements[max(MAX_ATTRIBS, MAX_REG_INPUT)]; struct wined3d_shader_signature_element output_signature_elements[MAX_REG_OUTPUT]; struct wined3d_shader_signature *output_signature = &shader->output_signature; struct wined3d_shader_signature *input_signature = &shader->input_signature; struct wined3d_shader_reg_maps *reg_maps = &shader->reg_maps; const struct wined3d_shader_frontend *fe = shader->frontend; unsigned int cur_loop_depth = 0, max_loop_depth = 0; struct wined3d_shader_version shader_version; struct wined3d_shader_phase *phase = NULL; const DWORD *ptr, *prev_ins, *current_ins; void *fe_data = shader->frontend_data; unsigned int i; HRESULT hr; memset(reg_maps, 0, sizeof(*reg_maps)); memset(input_signature_elements, 0, sizeof(input_signature_elements)); memset(output_signature_elements, 0, sizeof(output_signature_elements)); reg_maps->min_rel_offset = ~0U; list_init(®_maps->indexable_temps); fe->shader_read_header(fe_data, &ptr, &shader_version); prev_ins = current_ins = ptr; reg_maps->shader_version = shader_version; shader_set_limits(shader); if (!(reg_maps->constf = heap_calloc(((min(shader->limits->constant_float, constf_size) + 31) / 32), sizeof(*reg_maps->constf)))) { ERR("Failed to allocate constant map memory.\n"); return E_OUTOFMEMORY; } while (!fe->shader_is_end(fe_data, &ptr)) { struct wined3d_shader_instruction ins; current_ins = ptr; /* Fetch opcode. */ fe->shader_read_instruction(fe_data, &ptr, &ins); /* Unhandled opcode, and its parameters. */ if (ins.handler_idx == WINED3DSIH_TABLE_SIZE) { WARN("Encountered unrecognised or invalid instruction.\n"); return WINED3DERR_INVALIDCALL; } /* Handle declarations. */ if (ins.handler_idx == WINED3DSIH_DCL || ins.handler_idx == WINED3DSIH_DCL_UAV_TYPED) { struct wined3d_shader_semantic *semantic = &ins.declaration.semantic; unsigned int reg_idx = semantic->reg.reg.idx[0].offset; switch (semantic->reg.reg.type) { /* Mark input registers used. */ case WINED3DSPR_INPUT: if (reg_idx >= MAX_REG_INPUT) { ERR("Invalid input register index %u.\n", reg_idx); break; } if (shader_version.type == WINED3D_SHADER_TYPE_PIXEL && shader_version.major == 3 && semantic->usage == WINED3D_DECL_USAGE_POSITION && !semantic->usage_idx) return WINED3DERR_INVALIDCALL; reg_maps->input_registers |= 1u << reg_idx; shader_signature_from_semantic(&input_signature_elements[reg_idx], semantic); break; /* Vertex shader: mark 3.0 output registers used, save token. */ case WINED3DSPR_OUTPUT: if (reg_idx >= MAX_REG_OUTPUT) { ERR("Invalid output register index %u.\n", reg_idx); break; } reg_maps->output_registers |= 1u << reg_idx; shader_signature_from_semantic(&output_signature_elements[reg_idx], semantic); if (semantic->usage == WINED3D_DECL_USAGE_FOG) reg_maps->fog = 1; if (semantic->usage == WINED3D_DECL_USAGE_PSIZE) reg_maps->point_size = 1; break; case WINED3DSPR_SAMPLER: shader_record_sample(reg_maps, reg_idx, reg_idx, reg_idx); case WINED3DSPR_RESOURCE: if (reg_idx >= ARRAY_SIZE(reg_maps->resource_info)) { ERR("Invalid resource index %u.\n", reg_idx); break; } reg_maps->resource_info[reg_idx].type = semantic->resource_type; reg_maps->resource_info[reg_idx].data_type = semantic->resource_data_type; wined3d_bitmap_set(reg_maps->resource_map, reg_idx); break; case WINED3DSPR_UAV: if (reg_idx >= ARRAY_SIZE(reg_maps->uav_resource_info)) { ERR("Invalid UAV resource index %u.\n", reg_idx); break; } reg_maps->uav_resource_info[reg_idx].type = semantic->resource_type; reg_maps->uav_resource_info[reg_idx].data_type = semantic->resource_data_type; if (ins.flags) FIXME("Ignoring typed UAV flags %#x.\n", ins.flags); break; default: TRACE("Not recording DCL register type %#x.\n", semantic->reg.reg.type); break; } } else if (ins.handler_idx == WINED3DSIH_DCL_CONSTANT_BUFFER) { struct wined3d_shader_register *reg = &ins.declaration.src.reg; if (reg->idx[0].offset >= WINED3D_MAX_CBS) { ERR("Invalid CB index %u.\n", reg->idx[0].offset); } else { reg_maps->cb_sizes[reg->idx[0].offset] = reg->idx[1].offset; wined3d_bitmap_set(®_maps->cb_map, reg->idx[0].offset); } } else if (ins.handler_idx == WINED3DSIH_DCL_GLOBAL_FLAGS) { if (ins.flags & WINED3DSGF_FORCE_EARLY_DEPTH_STENCIL) { if (shader_version.type == WINED3D_SHADER_TYPE_PIXEL) shader->u.ps.force_early_depth_stencil = TRUE; else FIXME("Invalid instruction %#x for shader type %#x.\n", ins.handler_idx, shader_version.type); } else { WARN("Ignoring global flags %#x.\n", ins.flags); } } else if (ins.handler_idx == WINED3DSIH_DCL_GS_INSTANCES) { if (shader_version.type == WINED3D_SHADER_TYPE_GEOMETRY) shader->u.gs.instance_count = ins.declaration.count; else FIXME("Invalid instruction %#x for shader type %#x.\n", ins.handler_idx, shader_version.type); } else if (ins.handler_idx == WINED3DSIH_DCL_HS_FORK_PHASE_INSTANCE_COUNT || ins.handler_idx == WINED3DSIH_DCL_HS_JOIN_PHASE_INSTANCE_COUNT) { if (phase) phase->instance_count = ins.declaration.count; else FIXME("Instruction %s outside of shader phase.\n", debug_d3dshaderinstructionhandler(ins.handler_idx)); } else if (ins.handler_idx == WINED3DSIH_DCL_IMMEDIATE_CONSTANT_BUFFER) { if (reg_maps->icb) FIXME("Multiple immediate constant buffers.\n"); reg_maps->icb = ins.declaration.icb; } else if (ins.handler_idx == WINED3DSIH_DCL_INDEXABLE_TEMP) { if (phase) { FIXME("Indexable temporary registers not supported.\n"); } else { struct wined3d_shader_indexable_temp *reg; if (!(reg = heap_alloc(sizeof(*reg)))) return E_OUTOFMEMORY; *reg = ins.declaration.indexable_temp; list_add_tail(®_maps->indexable_temps, ®->entry); } } else if (ins.handler_idx == WINED3DSIH_DCL_INPUT_PRIMITIVE) { if (shader_version.type == WINED3D_SHADER_TYPE_GEOMETRY) shader->u.gs.input_type = ins.declaration.primitive_type.type; else FIXME("Invalid instruction %#x for shader type %#x.\n", ins.handler_idx, shader_version.type); } else if (ins.handler_idx == WINED3DSIH_DCL_INPUT_PS) { unsigned int reg_idx = ins.declaration.dst.reg.idx[0].offset; if (reg_idx >= MAX_REG_INPUT) { ERR("Invalid register index %u.\n", reg_idx); break; } if (shader_version.type == WINED3D_SHADER_TYPE_PIXEL) wined3d_insert_interpolation_mode(shader->u.ps.interpolation_mode, reg_idx, ins.flags); else FIXME("Invalid instruction %#x for shader type %#x.\n", ins.handler_idx, shader_version.type); } else if (ins.handler_idx == WINED3DSIH_DCL_OUTPUT) { if (ins.declaration.dst.reg.type == WINED3DSPR_DEPTHOUT || ins.declaration.dst.reg.type == WINED3DSPR_DEPTHOUTGE || ins.declaration.dst.reg.type == WINED3DSPR_DEPTHOUTLE) { if (shader_version.type == WINED3D_SHADER_TYPE_PIXEL) shader->u.ps.depth_output = ins.declaration.dst.reg.type; else FIXME("Invalid instruction %#x for shader type %#x.\n", ins.handler_idx, shader_version.type); } } else if (ins.handler_idx == WINED3DSIH_DCL_OUTPUT_CONTROL_POINT_COUNT) { if (shader_version.type == WINED3D_SHADER_TYPE_HULL) shader->u.hs.output_vertex_count = ins.declaration.count; else FIXME("Invalid instruction %#x for shader type %#x.\n", ins.handler_idx, shader_version.type); } else if (ins.handler_idx == WINED3DSIH_DCL_OUTPUT_TOPOLOGY) { if (shader_version.type == WINED3D_SHADER_TYPE_GEOMETRY) shader->u.gs.output_type = ins.declaration.primitive_type.type; else FIXME("Invalid instruction %#x for shader type %#x.\n", ins.handler_idx, shader_version.type); } else if (ins.handler_idx == WINED3DSIH_DCL_RESOURCE_RAW) { unsigned int reg_idx = ins.declaration.dst.reg.idx[0].offset; if (reg_idx >= ARRAY_SIZE(reg_maps->resource_info)) { ERR("Invalid resource index %u.\n", reg_idx); break; } reg_maps->resource_info[reg_idx].type = WINED3D_SHADER_RESOURCE_BUFFER; reg_maps->resource_info[reg_idx].data_type = WINED3D_DATA_UINT; reg_maps->resource_info[reg_idx].flags = WINED3D_VIEW_BUFFER_RAW; wined3d_bitmap_set(reg_maps->resource_map, reg_idx); } else if (ins.handler_idx == WINED3DSIH_DCL_RESOURCE_STRUCTURED) { unsigned int reg_idx = ins.declaration.structured_resource.reg.reg.idx[0].offset; if (reg_idx >= ARRAY_SIZE(reg_maps->resource_info)) { ERR("Invalid resource index %u.\n", reg_idx); break; } reg_maps->resource_info[reg_idx].type = WINED3D_SHADER_RESOURCE_BUFFER; reg_maps->resource_info[reg_idx].data_type = WINED3D_DATA_UINT; reg_maps->resource_info[reg_idx].flags = 0; reg_maps->resource_info[reg_idx].stride = ins.declaration.structured_resource.byte_stride / 4; wined3d_bitmap_set(reg_maps->resource_map, reg_idx); } else if (ins.handler_idx == WINED3DSIH_DCL_SAMPLER) { if (ins.flags & WINED3DSI_SAMPLER_COMPARISON_MODE) reg_maps->sampler_comparison_mode |= (1u << ins.declaration.dst.reg.idx[0].offset); } else if (ins.handler_idx == WINED3DSIH_DCL_TEMPS) { if (phase) phase->temporary_count = ins.declaration.count; else reg_maps->temporary_count = ins.declaration.count; } else if (ins.handler_idx == WINED3DSIH_DCL_TESSELLATOR_DOMAIN) { if (shader_version.type == WINED3D_SHADER_TYPE_DOMAIN) shader->u.ds.tessellator_domain = ins.declaration.tessellator_domain; else if (shader_version.type != WINED3D_SHADER_TYPE_HULL) FIXME("Invalid instruction %#x for shader type %#x.\n", ins.handler_idx, shader_version.type); } else if (ins.handler_idx == WINED3DSIH_DCL_TESSELLATOR_OUTPUT_PRIMITIVE) { if (shader_version.type == WINED3D_SHADER_TYPE_HULL) shader->u.hs.tessellator_output_primitive = ins.declaration.tessellator_output_primitive; else FIXME("Invalid instruction %#x for shader type %#x.\n", ins.handler_idx, shader_version.type); } else if (ins.handler_idx == WINED3DSIH_DCL_TESSELLATOR_PARTITIONING) { if (shader_version.type == WINED3D_SHADER_TYPE_HULL) shader->u.hs.tessellator_partitioning = ins.declaration.tessellator_partitioning; else FIXME("Invalid instruction %#x for shader type %#x.\n", ins.handler_idx, shader_version.type); } else if (ins.handler_idx == WINED3DSIH_DCL_TGSM_RAW) { if (FAILED(hr = shader_reg_maps_add_tgsm(reg_maps, ins.declaration.tgsm_raw.reg.reg.idx[0].offset, ins.declaration.tgsm_raw.byte_count / 4, 0))) return hr; } else if (ins.handler_idx == WINED3DSIH_DCL_TGSM_STRUCTURED) { unsigned int stride = ins.declaration.tgsm_structured.byte_stride / 4; unsigned int size = stride * ins.declaration.tgsm_structured.structure_count; if (FAILED(hr = shader_reg_maps_add_tgsm(reg_maps, ins.declaration.tgsm_structured.reg.reg.idx[0].offset, size, stride))) return hr; } else if (ins.handler_idx == WINED3DSIH_DCL_THREAD_GROUP) { if (shader_version.type == WINED3D_SHADER_TYPE_COMPUTE) { shader->u.cs.thread_group_size = ins.declaration.thread_group_size; } else { FIXME("Invalid instruction %#x for shader type %#x.\n", ins.handler_idx, shader_version.type); } } else if (ins.handler_idx == WINED3DSIH_DCL_UAV_RAW) { unsigned int reg_idx = ins.declaration.dst.reg.idx[0].offset; if (reg_idx >= ARRAY_SIZE(reg_maps->uav_resource_info)) { ERR("Invalid UAV resource index %u.\n", reg_idx); break; } if (ins.flags) FIXME("Ignoring raw UAV flags %#x.\n", ins.flags); reg_maps->uav_resource_info[reg_idx].type = WINED3D_SHADER_RESOURCE_BUFFER; reg_maps->uav_resource_info[reg_idx].data_type = WINED3D_DATA_UINT; reg_maps->uav_resource_info[reg_idx].flags = WINED3D_VIEW_BUFFER_RAW; } else if (ins.handler_idx == WINED3DSIH_DCL_UAV_STRUCTURED) { unsigned int reg_idx = ins.declaration.structured_resource.reg.reg.idx[0].offset; if (reg_idx >= ARRAY_SIZE(reg_maps->uav_resource_info)) { ERR("Invalid UAV resource index %u.\n", reg_idx); break; } if (ins.flags) FIXME("Ignoring structured UAV flags %#x.\n", ins.flags); reg_maps->uav_resource_info[reg_idx].type = WINED3D_SHADER_RESOURCE_BUFFER; reg_maps->uav_resource_info[reg_idx].data_type = WINED3D_DATA_UINT; reg_maps->uav_resource_info[reg_idx].flags = 0; reg_maps->uav_resource_info[reg_idx].stride = ins.declaration.structured_resource.byte_stride / 4; } else if (ins.handler_idx == WINED3DSIH_DCL_VERTICES_OUT) { if (shader_version.type == WINED3D_SHADER_TYPE_GEOMETRY) shader->u.gs.vertices_out = ins.declaration.count; else FIXME("Invalid instruction %#x for shader type %#x.\n", ins.handler_idx, shader_version.type); } else if (ins.handler_idx == WINED3DSIH_DEF) { struct wined3d_shader_lconst *lconst; float *value; if (!(lconst = heap_alloc(sizeof(*lconst)))) return E_OUTOFMEMORY; lconst->idx = ins.dst[0].reg.idx[0].offset; memcpy(lconst->value, ins.src[0].reg.u.immconst_data, 4 * sizeof(DWORD)); value = (float *)lconst->value; /* In pixel shader 1.X shaders, the constants are clamped between [-1;1] */ if (shader_version.major == 1 && shader_version.type == WINED3D_SHADER_TYPE_PIXEL) { if (value[0] < -1.0f) value[0] = -1.0f; else if (value[0] > 1.0f) value[0] = 1.0f; if (value[1] < -1.0f) value[1] = -1.0f; else if (value[1] > 1.0f) value[1] = 1.0f; if (value[2] < -1.0f) value[2] = -1.0f; else if (value[2] > 1.0f) value[2] = 1.0f; if (value[3] < -1.0f) value[3] = -1.0f; else if (value[3] > 1.0f) value[3] = 1.0f; } list_add_head(&shader->constantsF, &lconst->entry); if (isinf(value[0]) || isnan(value[0]) || isinf(value[1]) || isnan(value[1]) || isinf(value[2]) || isnan(value[2]) || isinf(value[3]) || isnan(value[3])) { shader->lconst_inf_or_nan = TRUE; } } else if (ins.handler_idx == WINED3DSIH_DEFI) { struct wined3d_shader_lconst *lconst; if (!(lconst = heap_alloc(sizeof(*lconst)))) return E_OUTOFMEMORY; lconst->idx = ins.dst[0].reg.idx[0].offset; memcpy(lconst->value, ins.src[0].reg.u.immconst_data, 4 * sizeof(DWORD)); list_add_head(&shader->constantsI, &lconst->entry); reg_maps->local_int_consts |= (1u << lconst->idx); } else if (ins.handler_idx == WINED3DSIH_DEFB) { struct wined3d_shader_lconst *lconst; if (!(lconst = heap_alloc(sizeof(*lconst)))) return E_OUTOFMEMORY; lconst->idx = ins.dst[0].reg.idx[0].offset; memcpy(lconst->value, ins.src[0].reg.u.immconst_data, sizeof(DWORD)); list_add_head(&shader->constantsB, &lconst->entry); reg_maps->local_bool_consts |= (1u << lconst->idx); } /* Handle shader phases. */ else if (ins.handler_idx == WINED3DSIH_HS_CONTROL_POINT_PHASE || ins.handler_idx == WINED3DSIH_HS_FORK_PHASE || ins.handler_idx == WINED3DSIH_HS_JOIN_PHASE) { if (FAILED(hr = shader_record_shader_phase(shader, &phase, &ins, current_ins, prev_ins))) return hr; } /* For subroutine prototypes. */ else if (ins.handler_idx == WINED3DSIH_LABEL) { reg_maps->labels |= 1u << ins.src[0].reg.idx[0].offset; } /* Set texture, address, temporary registers. */ else { BOOL color0_mov = FALSE; unsigned int i; /* This will loop over all the registers and try to * make a bitmask of the ones we're interested in. * * Relative addressing tokens are ignored, but that's * okay, since we'll catch any address registers when * they are initialized (required by spec). */ for (i = 0; i < ins.dst_count; ++i) { if (!shader_record_register_usage(shader, reg_maps, &ins.dst[i].reg, shader_version.type, constf_size)) return WINED3DERR_INVALIDCALL; if (shader_version.type == WINED3D_SHADER_TYPE_VERTEX) { UINT idx = ins.dst[i].reg.idx[0].offset; switch (ins.dst[i].reg.type) { case WINED3DSPR_RASTOUT: if (shader_version.major >= 3) break; switch (idx) { case 0: /* oPos */ reg_maps->output_registers |= 1u << 10; shader_signature_from_usage(&output_signature_elements[10], WINED3D_DECL_USAGE_POSITION, 0, 10, WINED3DSP_WRITEMASK_ALL); break; case 1: /* oFog */ reg_maps->output_registers |= 1u << 11; shader_signature_from_usage(&output_signature_elements[11], WINED3D_DECL_USAGE_FOG, 0, 11, WINED3DSP_WRITEMASK_0); break; case 2: /* oPts */ reg_maps->output_registers |= 1u << 11; shader_signature_from_usage(&output_signature_elements[11], WINED3D_DECL_USAGE_PSIZE, 0, 11, WINED3DSP_WRITEMASK_1); break; } break; case WINED3DSPR_ATTROUT: if (shader_version.major >= 3) break; if (idx < 2) { idx += 8; if (reg_maps->output_registers & (1u << idx)) { output_signature_elements[idx].mask |= ins.dst[i].write_mask; } else { reg_maps->output_registers |= 1u << idx; shader_signature_from_usage(&output_signature_elements[idx], WINED3D_DECL_USAGE_COLOR, idx - 8, idx, ins.dst[i].write_mask); } } break; case WINED3DSPR_TEXCRDOUT: /* WINED3DSPR_OUTPUT */ if (shader_version.major >= 3) { if (idx >= ARRAY_SIZE(reg_maps->u.output_registers_mask)) { WARN("Invalid output register index %u.\n", idx); break; } reg_maps->u.output_registers_mask[idx] |= ins.dst[i].write_mask; break; } if (idx >= ARRAY_SIZE(reg_maps->u.texcoord_mask)) { WARN("Invalid texcoord index %u.\n", idx); break; } reg_maps->u.texcoord_mask[idx] |= ins.dst[i].write_mask; if (reg_maps->output_registers & (1u << idx)) { output_signature_elements[idx].mask |= ins.dst[i].write_mask; } else { reg_maps->output_registers |= 1u << idx; shader_signature_from_usage(&output_signature_elements[idx], WINED3D_DECL_USAGE_TEXCOORD, idx, idx, ins.dst[i].write_mask); } break; default: break; } } if (shader_version.type == WINED3D_SHADER_TYPE_PIXEL) { if (ins.dst[i].reg.type == WINED3DSPR_COLOROUT && !ins.dst[i].reg.idx[0].offset) { /* Many 2.0 and 3.0 pixel shaders end with a MOV from a temp register to * COLOROUT 0. If we know this in advance, the ARB shader backend can skip * the mov and perform the sRGB write correction from the source register. * * However, if the mov is only partial, we can't do this, and if the write * comes from an instruction other than MOV it is hard to do as well. If * COLOROUT 0 is overwritten partially later, the marker is dropped again. */ shader->u.ps.color0_mov = FALSE; if (ins.handler_idx == WINED3DSIH_MOV && ins.dst[i].write_mask == WINED3DSP_WRITEMASK_ALL) { /* Used later when the source register is read. */ color0_mov = TRUE; } } /* Also drop the MOV marker if the source register is overwritten prior to the shader * end */ else if (ins.dst[i].reg.type == WINED3DSPR_TEMP && ins.dst[i].reg.idx[0].offset == shader->u.ps.color0_reg) { shader->u.ps.color0_mov = FALSE; } } /* Declare 1.x samplers implicitly, based on the destination reg. number. */ if (shader_version.major == 1 && (ins.handler_idx == WINED3DSIH_TEX || ins.handler_idx == WINED3DSIH_TEXBEM || ins.handler_idx == WINED3DSIH_TEXBEML || ins.handler_idx == WINED3DSIH_TEXDP3TEX || ins.handler_idx == WINED3DSIH_TEXM3x2TEX || ins.handler_idx == WINED3DSIH_TEXM3x3SPEC || ins.handler_idx == WINED3DSIH_TEXM3x3TEX || ins.handler_idx == WINED3DSIH_TEXM3x3VSPEC || ins.handler_idx == WINED3DSIH_TEXREG2AR || ins.handler_idx == WINED3DSIH_TEXREG2GB || ins.handler_idx == WINED3DSIH_TEXREG2RGB)) { unsigned int reg_idx = ins.dst[i].reg.idx[0].offset; if (reg_idx >= ARRAY_SIZE(reg_maps->resource_info)) { WARN("Invalid 1.x sampler index %u.\n", reg_idx); continue; } TRACE("Setting fake 2D resource for 1.x pixelshader.\n"); reg_maps->resource_info[reg_idx].type = WINED3D_SHADER_RESOURCE_TEXTURE_2D; reg_maps->resource_info[reg_idx].data_type = WINED3D_DATA_FLOAT; shader_record_sample(reg_maps, reg_idx, reg_idx, reg_idx); wined3d_bitmap_set(reg_maps->resource_map, reg_idx); /* texbem is only valid with < 1.4 pixel shaders */ if (ins.handler_idx == WINED3DSIH_TEXBEM || ins.handler_idx == WINED3DSIH_TEXBEML) { reg_maps->bumpmat |= 1u << reg_idx; if (ins.handler_idx == WINED3DSIH_TEXBEML) { reg_maps->luminanceparams |= 1u << reg_idx; } } } else if (ins.handler_idx == WINED3DSIH_BEM) { reg_maps->bumpmat |= 1u << ins.dst[i].reg.idx[0].offset; } } if (ins.handler_idx == WINED3DSIH_IMM_ATOMIC_ALLOC || ins.handler_idx == WINED3DSIH_IMM_ATOMIC_CONSUME) { unsigned int reg_idx = ins.src[0].reg.idx[0].offset; if (reg_idx >= MAX_UNORDERED_ACCESS_VIEWS) { ERR("Invalid UAV index %u.\n", reg_idx); break; } reg_maps->uav_counter_mask |= (1u << reg_idx); } else if ((WINED3DSIH_ATOMIC_AND <= ins.handler_idx && ins.handler_idx <= WINED3DSIH_ATOMIC_XOR) || (WINED3DSIH_IMM_ATOMIC_AND <= ins.handler_idx && ins.handler_idx <= WINED3DSIH_IMM_ATOMIC_XOR) || (ins.handler_idx == WINED3DSIH_BUFINFO && ins.src[0].reg.type == WINED3DSPR_UAV) || ins.handler_idx == WINED3DSIH_LD_UAV_TYPED || (ins.handler_idx == WINED3DSIH_LD_RAW && ins.src[1].reg.type == WINED3DSPR_UAV) || (ins.handler_idx == WINED3DSIH_LD_STRUCTURED && ins.src[2].reg.type == WINED3DSPR_UAV)) { const struct wined3d_shader_register *reg; if (ins.handler_idx == WINED3DSIH_LD_UAV_TYPED || ins.handler_idx == WINED3DSIH_LD_RAW) reg = &ins.src[1].reg; else if (ins.handler_idx == WINED3DSIH_LD_STRUCTURED) reg = &ins.src[2].reg; else if (WINED3DSIH_ATOMIC_AND <= ins.handler_idx && ins.handler_idx <= WINED3DSIH_ATOMIC_XOR) reg = &ins.dst[0].reg; else if (ins.handler_idx == WINED3DSIH_BUFINFO) reg = &ins.src[0].reg; else reg = &ins.dst[1].reg; if (reg->type == WINED3DSPR_UAV) { if (reg->idx[0].offset >= MAX_UNORDERED_ACCESS_VIEWS) { ERR("Invalid UAV index %u.\n", reg->idx[0].offset); break; } reg_maps->uav_read_mask |= (1u << reg->idx[0].offset); } } else if (ins.handler_idx == WINED3DSIH_NRM) { reg_maps->usesnrm = 1; } else if (ins.handler_idx == WINED3DSIH_DSY || ins.handler_idx == WINED3DSIH_DSY_COARSE || ins.handler_idx == WINED3DSIH_DSY_FINE) { reg_maps->usesdsy = 1; } else if (ins.handler_idx == WINED3DSIH_DSX || ins.handler_idx == WINED3DSIH_DSX_COARSE || ins.handler_idx == WINED3DSIH_DSX_FINE) { reg_maps->usesdsx = 1; } else if (ins.handler_idx == WINED3DSIH_TEXLDD) reg_maps->usestexldd = 1; else if (ins.handler_idx == WINED3DSIH_TEXLDL) reg_maps->usestexldl = 1; else if (ins.handler_idx == WINED3DSIH_MOVA) reg_maps->usesmova = 1; else if (ins.handler_idx == WINED3DSIH_IFC) reg_maps->usesifc = 1; else if (ins.handler_idx == WINED3DSIH_CALL) reg_maps->usescall = 1; else if (ins.handler_idx == WINED3DSIH_POW) reg_maps->usespow = 1; else if (ins.handler_idx == WINED3DSIH_LOOP || ins.handler_idx == WINED3DSIH_REP) { ++cur_loop_depth; if (cur_loop_depth > max_loop_depth) max_loop_depth = cur_loop_depth; } else if (ins.handler_idx == WINED3DSIH_ENDLOOP || ins.handler_idx == WINED3DSIH_ENDREP) { --cur_loop_depth; } else if (ins.handler_idx == WINED3DSIH_GATHER4 || ins.handler_idx == WINED3DSIH_GATHER4_C || ins.handler_idx == WINED3DSIH_SAMPLE || ins.handler_idx == WINED3DSIH_SAMPLE_B || ins.handler_idx == WINED3DSIH_SAMPLE_C || ins.handler_idx == WINED3DSIH_SAMPLE_C_LZ || ins.handler_idx == WINED3DSIH_SAMPLE_GRAD || ins.handler_idx == WINED3DSIH_SAMPLE_LOD) { shader_record_sample(reg_maps, ins.src[1].reg.idx[0].offset, ins.src[2].reg.idx[0].offset, reg_maps->sampler_map.count); } else if (ins.handler_idx == WINED3DSIH_GATHER4_PO || ins.handler_idx == WINED3DSIH_GATHER4_PO_C) { shader_record_sample(reg_maps, ins.src[2].reg.idx[0].offset, ins.src[3].reg.idx[0].offset, reg_maps->sampler_map.count); } else if ((ins.handler_idx == WINED3DSIH_BUFINFO && ins.src[0].reg.type == WINED3DSPR_RESOURCE) || (ins.handler_idx == WINED3DSIH_SAMPLE_INFO && ins.src[0].reg.type == WINED3DSPR_RESOURCE)) { shader_record_sample(reg_maps, ins.src[0].reg.idx[0].offset, WINED3D_SAMPLER_DEFAULT, reg_maps->sampler_map.count); } else if (ins.handler_idx == WINED3DSIH_LD || ins.handler_idx == WINED3DSIH_LD2DMS || (ins.handler_idx == WINED3DSIH_LD_RAW && ins.src[1].reg.type == WINED3DSPR_RESOURCE) || (ins.handler_idx == WINED3DSIH_RESINFO && ins.src[1].reg.type == WINED3DSPR_RESOURCE)) { shader_record_sample(reg_maps, ins.src[1].reg.idx[0].offset, WINED3D_SAMPLER_DEFAULT, reg_maps->sampler_map.count); } else if (ins.handler_idx == WINED3DSIH_LD_STRUCTURED && ins.src[2].reg.type == WINED3DSPR_RESOURCE) { shader_record_sample(reg_maps, ins.src[2].reg.idx[0].offset, WINED3D_SAMPLER_DEFAULT, reg_maps->sampler_map.count); } if (ins.predicate) if (!shader_record_register_usage(shader, reg_maps, &ins.predicate->reg, shader_version.type, constf_size)) return WINED3DERR_INVALIDCALL; for (i = 0; i < ins.src_count; ++i) { unsigned int count = get_instr_extra_regcount(ins.handler_idx, i); struct wined3d_shader_register reg = ins.src[i].reg; if (!shader_record_register_usage(shader, reg_maps, &ins.src[i].reg, shader_version.type, constf_size)) return WINED3DERR_INVALIDCALL; while (count) { ++reg.idx[0].offset; if (!shader_record_register_usage(shader, reg_maps, ®, shader_version.type, constf_size)) return WINED3DERR_INVALIDCALL; --count; } if (color0_mov) { if (ins.src[i].reg.type == WINED3DSPR_TEMP && ins.src[i].swizzle == WINED3DSP_NOSWIZZLE) { shader->u.ps.color0_mov = TRUE; shader->u.ps.color0_reg = ins.src[i].reg.idx[0].offset; } } } } prev_ins = current_ins; } reg_maps->loop_depth = max_loop_depth; if (phase) { phase->end = prev_ins; phase = NULL; } /* PS before 2.0 don't have explicit color outputs. Instead the value of * R0 is written to the render target. */ if (shader_version.major < 2 && shader_version.type == WINED3D_SHADER_TYPE_PIXEL) reg_maps->rt_mask |= (1u << 0); if (input_signature->elements) { for (i = 0; i < input_signature->element_count; ++i) { if (shader_version.type == WINED3D_SHADER_TYPE_VERTEX) { if (input_signature->elements[i].register_idx >= ARRAY_SIZE(shader->u.vs.attributes)) { WARN("Invalid input signature register index %u.\n", input_signature->elements[i].register_idx); return WINED3DERR_INVALIDCALL; } } else if (shader_version.type == WINED3D_SHADER_TYPE_PIXEL) { if (input_signature->elements[i].sysval_semantic == WINED3D_SV_POSITION) reg_maps->vpos = 1; else if (input_signature->elements[i].sysval_semantic == WINED3D_SV_IS_FRONT_FACE) reg_maps->usesfacing = 1; } reg_maps->input_registers |= 1u << input_signature->elements[i].register_idx; } } else if (!input_signature->elements && reg_maps->input_registers) { unsigned int count = wined3d_popcount(reg_maps->input_registers); struct wined3d_shader_signature_element *e; unsigned int i; if (!(input_signature->elements = heap_calloc(count, sizeof(*input_signature->elements)))) return E_OUTOFMEMORY; input_signature->element_count = count; e = input_signature->elements; for (i = 0; i < ARRAY_SIZE(input_signature_elements); ++i) { if (!(reg_maps->input_registers & (1u << i))) continue; input_signature_elements[i].register_idx = i; *e++ = input_signature_elements[i]; } } if (output_signature->elements) { if (FAILED(hr = shader_scan_output_signature(shader))) return hr; } else if (reg_maps->output_registers) { unsigned int count = wined3d_popcount(reg_maps->output_registers); struct wined3d_shader_signature_element *e; if (!(output_signature->elements = heap_calloc(count, sizeof(*output_signature->elements)))) return E_OUTOFMEMORY; output_signature->element_count = count; e = output_signature->elements; for (i = 0; i < ARRAY_SIZE(output_signature_elements); ++i) { if (!(reg_maps->output_registers & (1u << i))) continue; *e++ = output_signature_elements[i]; } } return WINED3D_OK; } static void shader_cleanup_reg_maps(struct wined3d_shader_reg_maps *reg_maps) { struct wined3d_shader_indexable_temp *reg, *reg_next; heap_free(reg_maps->constf); heap_free(reg_maps->sampler_map.entries); LIST_FOR_EACH_ENTRY_SAFE(reg, reg_next, ®_maps->indexable_temps, struct wined3d_shader_indexable_temp, entry) heap_free(reg); list_init(®_maps->indexable_temps); heap_free(reg_maps->tgsm); } unsigned int shader_find_free_input_register(const struct wined3d_shader_reg_maps *reg_maps, unsigned int max) { DWORD map = 1u << max; map |= map - 1; map &= reg_maps->shader_version.major < 3 ? ~reg_maps->texcoord : ~reg_maps->input_registers; return wined3d_log2i(map); } static void shader_dump_global_flags(struct wined3d_string_buffer *buffer, DWORD global_flags) { if (global_flags & WINED3DSGF_REFACTORING_ALLOWED) { shader_addline(buffer, "refactoringAllowed"); global_flags &= ~WINED3DSGF_REFACTORING_ALLOWED; if (global_flags) shader_addline(buffer, " | "); } if (global_flags & WINED3DSGF_FORCE_EARLY_DEPTH_STENCIL) { shader_addline(buffer, "forceEarlyDepthStencil"); global_flags &= ~WINED3DSGF_FORCE_EARLY_DEPTH_STENCIL; if (global_flags) shader_addline(buffer, " | "); } if (global_flags & WINED3DSGF_ENABLE_RAW_AND_STRUCTURED_BUFFERS) { shader_addline(buffer, "enableRawAndStructuredBuffers"); global_flags &= ~WINED3DSGF_ENABLE_RAW_AND_STRUCTURED_BUFFERS; } if (global_flags) shader_addline(buffer, "unknown_flags(%#x)", global_flags); } static void shader_dump_sync_flags(struct wined3d_string_buffer *buffer, DWORD sync_flags) { if (sync_flags & WINED3DSSF_GLOBAL_UAV) { shader_addline(buffer, "_uglobal"); sync_flags &= ~WINED3DSSF_GLOBAL_UAV; } if (sync_flags & WINED3DSSF_GROUP_SHARED_MEMORY) { shader_addline(buffer, "_g"); sync_flags &= ~WINED3DSSF_GROUP_SHARED_MEMORY; } if (sync_flags & WINED3DSSF_THREAD_GROUP) { shader_addline(buffer, "_t"); sync_flags &= ~WINED3DSSF_THREAD_GROUP; } if (sync_flags) shader_addline(buffer, "_unknown_flags(%#x)", sync_flags); } static void shader_dump_precise_flags(struct wined3d_string_buffer *buffer, DWORD flags) { if (!(flags & WINED3DSI_PRECISE_XYZW)) return; shader_addline(buffer, " [precise"); if (flags != WINED3DSI_PRECISE_XYZW) { shader_addline(buffer, "(%s%s%s%s)", flags & WINED3DSI_PRECISE_X ? "x" : "", flags & WINED3DSI_PRECISE_Y ? "y" : "", flags & WINED3DSI_PRECISE_Z ? "z" : "", flags & WINED3DSI_PRECISE_W ? "w" : ""); } shader_addline(buffer, "]"); } static void shader_dump_uav_flags(struct wined3d_string_buffer *buffer, DWORD uav_flags) { if (uav_flags & WINED3DSUF_GLOBALLY_COHERENT) { shader_addline(buffer, "_glc"); uav_flags &= ~WINED3DSUF_GLOBALLY_COHERENT; } if (uav_flags & WINED3DSUF_ORDER_PRESERVING_COUNTER) { shader_addline(buffer, "_opc"); uav_flags &= ~WINED3DSUF_ORDER_PRESERVING_COUNTER; } if (uav_flags) shader_addline(buffer, "_unknown_flags(%#x)", uav_flags); } static void shader_dump_tessellator_domain(struct wined3d_string_buffer *buffer, enum wined3d_tessellator_domain domain) { switch (domain) { case WINED3D_TESSELLATOR_DOMAIN_LINE: shader_addline(buffer, "line"); break; case WINED3D_TESSELLATOR_DOMAIN_TRIANGLE: shader_addline(buffer, "triangle"); break; case WINED3D_TESSELLATOR_DOMAIN_QUAD: shader_addline(buffer, "quad"); break; default: shader_addline(buffer, "unknown_tessellator_domain(%#x)", domain); break; } } static void shader_dump_tessellator_output_primitive(struct wined3d_string_buffer *buffer, enum wined3d_tessellator_output_primitive output_primitive) { switch (output_primitive) { case WINED3D_TESSELLATOR_OUTPUT_POINT: shader_addline(buffer, "point"); break; case WINED3D_TESSELLATOR_OUTPUT_LINE: shader_addline(buffer, "line"); break; case WINED3D_TESSELLATOR_OUTPUT_TRIANGLE_CW: shader_addline(buffer, "triangle_cw"); break; case WINED3D_TESSELLATOR_OUTPUT_TRIANGLE_CCW: shader_addline(buffer, "triangle_ccw"); break; default: shader_addline(buffer, "unknown_tessellator_output_primitive(%#x)", output_primitive); break; } } static void shader_dump_tessellator_partitioning(struct wined3d_string_buffer *buffer, enum wined3d_tessellator_partitioning partitioning) { switch (partitioning) { case WINED3D_TESSELLATOR_PARTITIONING_INTEGER: shader_addline(buffer, "integer"); break; case WINED3D_TESSELLATOR_PARTITIONING_POW2: shader_addline(buffer, "pow2"); break; case WINED3D_TESSELLATOR_PARTITIONING_FRACTIONAL_ODD: shader_addline(buffer, "fractional_odd"); break; case WINED3D_TESSELLATOR_PARTITIONING_FRACTIONAL_EVEN: shader_addline(buffer, "fractional_even"); break; default: shader_addline(buffer, "unknown_tessellator_partitioning(%#x)", partitioning); break; } } static void shader_dump_shader_input_sysval_semantic(struct wined3d_string_buffer *buffer, enum wined3d_shader_input_sysval_semantic semantic) { unsigned int i; for (i = 0; i < ARRAY_SIZE(shader_input_sysval_semantic_names); ++i) { if (shader_input_sysval_semantic_names[i].sysval_semantic == semantic) { shader_addline(buffer, "%s", shader_input_sysval_semantic_names[i].sysval_name); return; } } shader_addline(buffer, "unknown_shader_input_sysval_semantic(%#x)", semantic); } static void shader_dump_resource_type(struct wined3d_string_buffer *buffer, enum wined3d_shader_resource_type type) { static const char *const resource_type_names[] = { /* WINED3D_SHADER_RESOURCE_NONE */ "none", /* WINED3D_SHADER_RESOURCE_BUFFER */ "buffer", /* WINED3D_SHADER_RESOURCE_TEXTURE_1D */ "texture1d", /* WINED3D_SHADER_RESOURCE_TEXTURE_2D */ "texture2d", /* WINED3D_SHADER_RESOURCE_TEXTURE_2DMS */ "texture2dms", /* WINED3D_SHADER_RESOURCE_TEXTURE_3D */ "texture3d", /* WINED3D_SHADER_RESOURCE_TEXTURE_CUBE */ "texturecube", /* WINED3D_SHADER_RESOURCE_TEXTURE_1DARRAY */ "texture1darray", /* WINED3D_SHADER_RESOURCE_TEXTURE_2DARRAY */ "texture2darray", /* WINED3D_SHADER_RESOURCE_TEXTURE_2DMSARRAY */ "texture2dmsarray", /* WINED3D_SHADER_RESOURCE_TEXTURE_CUBEARRAY */ "texturecubearray", }; if (type < ARRAY_SIZE(resource_type_names)) shader_addline(buffer, "%s", resource_type_names[type]); else shader_addline(buffer, "unknown"); } static void shader_dump_data_type(struct wined3d_string_buffer *buffer, enum wined3d_data_type type) { static const char *const data_type_names[] = { /* WINED3D_DATA_FLOAT */ "(float)", /* WINED3D_DATA_INT */ "(int)", /* WINED3D_DATA_RESOURCE */ "(resource)", /* WINED3D_DATA_SAMPLER */ "(sampler)", /* WINED3D_DATA_UAV */ "(uav)", /* WINED3D_DATA_UINT */ "(uint)", /* WINED3D_DATA_UNORM */ "(unorm)", /* WINED3D_DATA_SNORM */ "(snorm)", /* WINED3D_DATA_OPAQUE */ "(opaque)", }; if (type < ARRAY_SIZE(data_type_names)) shader_addline(buffer, "%s", data_type_names[type]); else shader_addline(buffer, "(unknown)"); } static void shader_dump_decl_usage(struct wined3d_string_buffer *buffer, const struct wined3d_shader_semantic *semantic, unsigned int flags, const struct wined3d_shader_version *shader_version) { shader_addline(buffer, "dcl"); if (semantic->reg.reg.type == WINED3DSPR_SAMPLER) { switch (semantic->resource_type) { case WINED3D_SHADER_RESOURCE_TEXTURE_2D: shader_addline(buffer, "_2d"); break; case WINED3D_SHADER_RESOURCE_TEXTURE_3D: shader_addline(buffer, "_3d"); break; case WINED3D_SHADER_RESOURCE_TEXTURE_CUBE: shader_addline(buffer, "_cube"); break; default: shader_addline(buffer, "_unknown_resource_type(%#x)", semantic->resource_type); break; } } else if (semantic->reg.reg.type == WINED3DSPR_RESOURCE || semantic->reg.reg.type == WINED3DSPR_UAV) { if (semantic->reg.reg.type == WINED3DSPR_RESOURCE) shader_addline(buffer, "_resource_"); else shader_addline(buffer, "_uav_"); shader_dump_resource_type(buffer, semantic->resource_type); if (semantic->reg.reg.type == WINED3DSPR_UAV) shader_dump_uav_flags(buffer, flags); shader_dump_data_type(buffer, semantic->resource_data_type); } else { /* Pixel shaders 3.0 don't have usage semantics. */ if (shader_version->major < 3 && shader_version->type == WINED3D_SHADER_TYPE_PIXEL) return; else shader_addline(buffer, "_"); switch (semantic->usage) { case WINED3D_DECL_USAGE_POSITION: shader_addline(buffer, "position%u", semantic->usage_idx); break; case WINED3D_DECL_USAGE_BLEND_INDICES: shader_addline(buffer, "blend"); break; case WINED3D_DECL_USAGE_BLEND_WEIGHT: shader_addline(buffer, "weight"); break; case WINED3D_DECL_USAGE_NORMAL: shader_addline(buffer, "normal%u", semantic->usage_idx); break; case WINED3D_DECL_USAGE_PSIZE: shader_addline(buffer, "psize"); break; case WINED3D_DECL_USAGE_COLOR: if (!semantic->usage_idx) shader_addline(buffer, "color"); else shader_addline(buffer, "specular%u", (semantic->usage_idx - 1)); break; case WINED3D_DECL_USAGE_TEXCOORD: shader_addline(buffer, "texture%u", semantic->usage_idx); break; case WINED3D_DECL_USAGE_TANGENT: shader_addline(buffer, "tangent"); break; case WINED3D_DECL_USAGE_BINORMAL: shader_addline(buffer, "binormal"); break; case WINED3D_DECL_USAGE_TESS_FACTOR: shader_addline(buffer, "tessfactor"); break; case WINED3D_DECL_USAGE_POSITIONT: shader_addline(buffer, "positionT%u", semantic->usage_idx); break; case WINED3D_DECL_USAGE_FOG: shader_addline(buffer, "fog"); break; case WINED3D_DECL_USAGE_DEPTH: shader_addline(buffer, "depth"); break; case WINED3D_DECL_USAGE_SAMPLE: shader_addline(buffer, "sample"); break; default: shader_addline(buffer, "", semantic->usage); FIXME("Unrecognised semantic usage %#x.\n", semantic->usage); } } } static void shader_dump_register(struct wined3d_string_buffer *buffer, const struct wined3d_shader_register *reg, const struct wined3d_shader_version *shader_version) { static const char * const rastout_reg_names[] = {"oPos", "oFog", "oPts"}; static const char * const misctype_reg_names[] = {"vPos", "vFace"}; UINT offset = reg->idx[0].offset; switch (reg->type) { case WINED3DSPR_TEMP: shader_addline(buffer, "r"); break; case WINED3DSPR_INPUT: shader_addline(buffer, "v"); break; case WINED3DSPR_CONST: case WINED3DSPR_CONST2: case WINED3DSPR_CONST3: case WINED3DSPR_CONST4: shader_addline(buffer, "c"); offset = shader_get_float_offset(reg->type, offset); break; case WINED3DSPR_TEXTURE: /* vs: case WINED3DSPR_ADDR */ shader_addline(buffer, "%c", shader_version->type == WINED3D_SHADER_TYPE_PIXEL ? 't' : 'a'); break; case WINED3DSPR_RASTOUT: shader_addline(buffer, "%s", rastout_reg_names[offset]); break; case WINED3DSPR_COLOROUT: shader_addline(buffer, "oC"); break; case WINED3DSPR_DEPTHOUT: shader_addline(buffer, "oDepth"); break; case WINED3DSPR_DEPTHOUTGE: shader_addline(buffer, "oDepthGE"); break; case WINED3DSPR_DEPTHOUTLE: shader_addline(buffer, "oDepthLE"); break; case WINED3DSPR_ATTROUT: shader_addline(buffer, "oD"); break; case WINED3DSPR_TEXCRDOUT: /* Vertex shaders >= 3.0 use general purpose output registers * (WINED3DSPR_OUTPUT), which can include an address token. */ if (shader_version->major >= 3) shader_addline(buffer, "o"); else shader_addline(buffer, "oT"); break; case WINED3DSPR_CONSTINT: shader_addline(buffer, "i"); break; case WINED3DSPR_CONSTBOOL: shader_addline(buffer, "b"); break; case WINED3DSPR_LABEL: shader_addline(buffer, "l"); break; case WINED3DSPR_LOOP: shader_addline(buffer, "aL"); break; case WINED3DSPR_SAMPLER: shader_addline(buffer, "s"); break; case WINED3DSPR_MISCTYPE: if (offset > 1) { FIXME("Unhandled misctype register %u.\n", offset); shader_addline(buffer, "", offset); } else { shader_addline(buffer, "%s", misctype_reg_names[offset]); } break; case WINED3DSPR_PREDICATE: shader_addline(buffer, "p"); break; case WINED3DSPR_IMMCONST: shader_addline(buffer, "l"); break; case WINED3DSPR_CONSTBUFFER: shader_addline(buffer, "cb"); break; case WINED3DSPR_IMMCONSTBUFFER: shader_addline(buffer, "icb"); break; case WINED3DSPR_PRIMID: shader_addline(buffer, "primID"); break; case WINED3DSPR_NULL: shader_addline(buffer, "null"); break; case WINED3DSPR_RASTERIZER: shader_addline(buffer, "rasterizer"); break; case WINED3DSPR_RESOURCE: shader_addline(buffer, "t"); break; case WINED3DSPR_UAV: shader_addline(buffer, "u"); break; case WINED3DSPR_OUTPOINTID: shader_addline(buffer, "vOutputControlPointID"); break; case WINED3DSPR_FORKINSTID: shader_addline(buffer, "vForkInstanceId"); break; case WINED3DSPR_JOININSTID: shader_addline(buffer, "vJoinInstanceId"); break; case WINED3DSPR_INCONTROLPOINT: shader_addline(buffer, "vicp"); break; case WINED3DSPR_OUTCONTROLPOINT: shader_addline(buffer, "vocp"); break; case WINED3DSPR_PATCHCONST: shader_addline(buffer, "vpc"); break; case WINED3DSPR_TESSCOORD: shader_addline(buffer, "vDomainLocation"); break; case WINED3DSPR_GROUPSHAREDMEM: shader_addline(buffer, "g"); break; case WINED3DSPR_THREADID: shader_addline(buffer, "vThreadID"); break; case WINED3DSPR_THREADGROUPID: shader_addline(buffer, "vThreadGroupID"); break; case WINED3DSPR_LOCALTHREADID: shader_addline(buffer, "vThreadIDInGroup"); break; case WINED3DSPR_LOCALTHREADINDEX: shader_addline(buffer, "vThreadIDInGroupFlattened"); break; case WINED3DSPR_IDXTEMP: shader_addline(buffer, "x"); break; case WINED3DSPR_STREAM: shader_addline(buffer, "m"); break; case WINED3DSPR_FUNCTIONBODY: shader_addline(buffer, "fb"); break; case WINED3DSPR_FUNCTIONPOINTER: shader_addline(buffer, "fp"); break; case WINED3DSPR_COVERAGE: shader_addline(buffer, "vCoverage"); break; case WINED3DSPR_SAMPLEMASK: shader_addline(buffer, "oMask"); break; case WINED3DSPR_GSINSTID: shader_addline(buffer, "vGSInstanceID"); break; default: shader_addline(buffer, "", reg->type); break; } if (reg->type == WINED3DSPR_IMMCONST) { shader_addline(buffer, "("); switch (reg->immconst_type) { case WINED3D_IMMCONST_SCALAR: switch (reg->data_type) { case WINED3D_DATA_FLOAT: shader_addline(buffer, "%.8e", *(const float *)reg->u.immconst_data); break; case WINED3D_DATA_INT: shader_addline(buffer, "%d", reg->u.immconst_data[0]); break; case WINED3D_DATA_RESOURCE: case WINED3D_DATA_SAMPLER: case WINED3D_DATA_UINT: shader_addline(buffer, "%u", reg->u.immconst_data[0]); break; default: shader_addline(buffer, "", reg->data_type); break; } break; case WINED3D_IMMCONST_VEC4: switch (reg->data_type) { case WINED3D_DATA_FLOAT: shader_addline(buffer, "%.8e, %.8e, %.8e, %.8e", *(const float *)®->u.immconst_data[0], *(const float *)®->u.immconst_data[1], *(const float *)®->u.immconst_data[2], *(const float *)®->u.immconst_data[3]); break; case WINED3D_DATA_INT: shader_addline(buffer, "%d, %d, %d, %d", reg->u.immconst_data[0], reg->u.immconst_data[1], reg->u.immconst_data[2], reg->u.immconst_data[3]); break; case WINED3D_DATA_RESOURCE: case WINED3D_DATA_SAMPLER: case WINED3D_DATA_UINT: shader_addline(buffer, "%u, %u, %u, %u", reg->u.immconst_data[0], reg->u.immconst_data[1], reg->u.immconst_data[2], reg->u.immconst_data[3]); break; default: shader_addline(buffer, "", reg->data_type); break; } break; default: shader_addline(buffer, "", reg->immconst_type); break; } shader_addline(buffer, ")"); } else if (reg->type != WINED3DSPR_RASTOUT && reg->type != WINED3DSPR_MISCTYPE && reg->type != WINED3DSPR_NULL) { if (offset != ~0u) { shader_addline(buffer, "["); if (reg->idx[0].rel_addr) { shader_dump_src_param(buffer, reg->idx[0].rel_addr, shader_version); shader_addline(buffer, " + "); } shader_addline(buffer, "%u]", offset); if (reg->idx[1].offset != ~0u) { shader_addline(buffer, "["); if (reg->idx[1].rel_addr) { shader_dump_src_param(buffer, reg->idx[1].rel_addr, shader_version); shader_addline(buffer, " + "); } shader_addline(buffer, "%u]", reg->idx[1].offset); } } if (reg->type == WINED3DSPR_FUNCTIONPOINTER) shader_addline(buffer, "[%u]", reg->u.fp_body_idx); } } static void shader_dump_dst_param(struct wined3d_string_buffer *buffer, const struct wined3d_shader_dst_param *param, const struct wined3d_shader_version *shader_version) { DWORD write_mask = param->write_mask; shader_dump_register(buffer, ¶m->reg, shader_version); if (write_mask && write_mask != WINED3DSP_WRITEMASK_ALL) { static const char write_mask_chars[] = "xyzw"; shader_addline(buffer, "."); if (write_mask & WINED3DSP_WRITEMASK_0) shader_addline(buffer, "%c", write_mask_chars[0]); if (write_mask & WINED3DSP_WRITEMASK_1) shader_addline(buffer, "%c", write_mask_chars[1]); if (write_mask & WINED3DSP_WRITEMASK_2) shader_addline(buffer, "%c", write_mask_chars[2]); if (write_mask & WINED3DSP_WRITEMASK_3) shader_addline(buffer, "%c", write_mask_chars[3]); } } static void shader_dump_src_param(struct wined3d_string_buffer *buffer, const struct wined3d_shader_src_param *param, const struct wined3d_shader_version *shader_version) { enum wined3d_shader_src_modifier src_modifier = param->modifiers; DWORD swizzle = param->swizzle; if (src_modifier == WINED3DSPSM_NEG || src_modifier == WINED3DSPSM_BIASNEG || src_modifier == WINED3DSPSM_SIGNNEG || src_modifier == WINED3DSPSM_X2NEG || src_modifier == WINED3DSPSM_ABSNEG) shader_addline(buffer, "-"); else if (src_modifier == WINED3DSPSM_COMP) shader_addline(buffer, "1-"); else if (src_modifier == WINED3DSPSM_NOT) shader_addline(buffer, "!"); if (src_modifier == WINED3DSPSM_ABS || src_modifier == WINED3DSPSM_ABSNEG) shader_addline(buffer, "abs("); shader_dump_register(buffer, ¶m->reg, shader_version); switch (src_modifier) { case WINED3DSPSM_NONE: break; case WINED3DSPSM_NEG: break; case WINED3DSPSM_NOT: break; case WINED3DSPSM_BIAS: shader_addline(buffer, "_bias"); break; case WINED3DSPSM_BIASNEG: shader_addline(buffer, "_bias"); break; case WINED3DSPSM_SIGN: shader_addline(buffer, "_bx2"); break; case WINED3DSPSM_SIGNNEG: shader_addline(buffer, "_bx2"); break; case WINED3DSPSM_COMP: break; case WINED3DSPSM_X2: shader_addline(buffer, "_x2"); break; case WINED3DSPSM_X2NEG: shader_addline(buffer, "_x2"); break; case WINED3DSPSM_DZ: shader_addline(buffer, "_dz"); break; case WINED3DSPSM_DW: shader_addline(buffer, "_dw"); break; case WINED3DSPSM_ABSNEG: shader_addline(buffer, ")"); break; case WINED3DSPSM_ABS: shader_addline(buffer, ")"); break; default: shader_addline(buffer, "_unknown_modifier(%#x)", src_modifier); } if (swizzle != WINED3DSP_NOSWIZZLE) { static const char swizzle_chars[] = "xyzw"; DWORD swizzle_x = swizzle & 0x03; DWORD swizzle_y = (swizzle >> 2) & 0x03; DWORD swizzle_z = (swizzle >> 4) & 0x03; DWORD swizzle_w = (swizzle >> 6) & 0x03; if (swizzle_x == swizzle_y && swizzle_x == swizzle_z && swizzle_x == swizzle_w) { shader_addline(buffer, ".%c", swizzle_chars[swizzle_x]); } else { shader_addline(buffer, ".%c%c%c%c", swizzle_chars[swizzle_x], swizzle_chars[swizzle_y], swizzle_chars[swizzle_z], swizzle_chars[swizzle_w]); } } } /* Shared code in order to generate the bulk of the shader string. */ HRESULT shader_generate_code(const struct wined3d_shader *shader, struct wined3d_string_buffer *buffer, const struct wined3d_shader_reg_maps *reg_maps, void *backend_ctx, const DWORD *start, const DWORD *end) { struct wined3d_device *device = shader->device; const struct wined3d_shader_frontend *fe = shader->frontend; void *fe_data = shader->frontend_data; struct wined3d_shader_version shader_version; struct wined3d_shader_parser_state state; struct wined3d_shader_instruction ins; struct wined3d_shader_tex_mx tex_mx; struct wined3d_shader_context ctx; const DWORD *ptr; /* Initialize current parsing state. */ tex_mx.current_row = 0; state.current_loop_depth = 0; state.current_loop_reg = 0; state.in_subroutine = FALSE; ctx.shader = shader; ctx.reg_maps = reg_maps; ctx.buffer = buffer; ctx.tex_mx = &tex_mx; ctx.state = &state; ctx.backend_data = backend_ctx; ins.ctx = &ctx; fe->shader_read_header(fe_data, &ptr, &shader_version); if (start) ptr = start; while (!fe->shader_is_end(fe_data, &ptr) && ptr != end) { /* Read opcode. */ fe->shader_read_instruction(fe_data, &ptr, &ins); /* Unknown opcode and its parameters. */ if (ins.handler_idx == WINED3DSIH_TABLE_SIZE) { WARN("Encountered unrecognised or invalid instruction.\n"); return WINED3DERR_INVALIDCALL; } if (ins.predicate) FIXME("Predicates not implemented.\n"); /* Call appropriate function for output target */ device->shader_backend->shader_handle_instruction(&ins); } return WINED3D_OK; } static void shader_dump_ins_modifiers(struct wined3d_string_buffer *buffer, const struct wined3d_shader_dst_param *dst) { DWORD mmask = dst->modifiers; switch (dst->shift) { case 0: break; case 13: shader_addline(buffer, "_d8"); break; case 14: shader_addline(buffer, "_d4"); break; case 15: shader_addline(buffer, "_d2"); break; case 1: shader_addline(buffer, "_x2"); break; case 2: shader_addline(buffer, "_x4"); break; case 3: shader_addline(buffer, "_x8"); break; default: shader_addline(buffer, "_unhandled_shift(%d)", dst->shift); break; } if (mmask & WINED3DSPDM_SATURATE) shader_addline(buffer, "_sat"); if (mmask & WINED3DSPDM_PARTIALPRECISION) shader_addline(buffer, "_pp"); if (mmask & WINED3DSPDM_MSAMPCENTROID) shader_addline(buffer, "_centroid"); mmask &= ~(WINED3DSPDM_SATURATE | WINED3DSPDM_PARTIALPRECISION | WINED3DSPDM_MSAMPCENTROID); if (mmask) FIXME("Unrecognised modifier %#x.\n", mmask); } static void shader_dump_primitive_type(struct wined3d_string_buffer *buffer, const struct wined3d_shader_primitive_type *primitive_type) { switch (primitive_type->type) { case WINED3D_PT_UNDEFINED: shader_addline(buffer, "undefined"); break; case WINED3D_PT_POINTLIST: shader_addline(buffer, "pointlist"); break; case WINED3D_PT_LINELIST: shader_addline(buffer, "linelist"); break; case WINED3D_PT_LINESTRIP: shader_addline(buffer, "linestrip"); break; case WINED3D_PT_TRIANGLELIST: shader_addline(buffer, "trianglelist"); break; case WINED3D_PT_TRIANGLESTRIP: shader_addline(buffer, "trianglestrip"); break; case WINED3D_PT_TRIANGLEFAN: shader_addline(buffer, "trianglefan"); break; case WINED3D_PT_LINELIST_ADJ: shader_addline(buffer, "linelist_adj"); break; case WINED3D_PT_LINESTRIP_ADJ: shader_addline(buffer, "linestrip_adj"); break; case WINED3D_PT_TRIANGLELIST_ADJ: shader_addline(buffer, "trianglelist_adj"); break; case WINED3D_PT_TRIANGLESTRIP_ADJ: shader_addline(buffer, "trianglestrip_adj"); break; case WINED3D_PT_PATCH: shader_addline(buffer, "patch%u", primitive_type->patch_vertex_count); break; default: shader_addline(buffer, "", primitive_type->type); break; } } static void shader_dump_interpolation_mode(struct wined3d_string_buffer *buffer, enum wined3d_shader_interpolation_mode interpolation_mode) { switch (interpolation_mode) { case WINED3DSIM_CONSTANT: shader_addline(buffer, "constant"); break; case WINED3DSIM_LINEAR: shader_addline(buffer, "linear"); break; case WINED3DSIM_LINEAR_CENTROID: shader_addline(buffer, "linear centroid"); break; case WINED3DSIM_LINEAR_NOPERSPECTIVE: shader_addline(buffer, "linear noperspective"); break; case WINED3DSIM_LINEAR_SAMPLE: shader_addline(buffer, "linear sample"); break; case WINED3DSIM_LINEAR_NOPERSPECTIVE_CENTROID: shader_addline(buffer, "linear noperspective centroid"); break; case WINED3DSIM_LINEAR_NOPERSPECTIVE_SAMPLE: shader_addline(buffer, "linear noperspective sample"); break; default: shader_addline(buffer, "", interpolation_mode); break; } } static void shader_trace_init(const struct wined3d_shader_frontend *fe, void *fe_data) { struct wined3d_shader_version shader_version; struct wined3d_string_buffer buffer; const char *type_prefix; const char *p, *q; const DWORD *ptr; DWORD i; if (!string_buffer_init(&buffer)) { ERR("Failed to initialize string buffer.\n"); return; } fe->shader_read_header(fe_data, &ptr, &shader_version); TRACE("Parsing %p.\n", ptr); switch (shader_version.type) { case WINED3D_SHADER_TYPE_VERTEX: type_prefix = "vs"; break; case WINED3D_SHADER_TYPE_HULL: type_prefix = "hs"; break; case WINED3D_SHADER_TYPE_DOMAIN: type_prefix = "ds"; break; case WINED3D_SHADER_TYPE_GEOMETRY: type_prefix = "gs"; break; case WINED3D_SHADER_TYPE_PIXEL: type_prefix = "ps"; break; case WINED3D_SHADER_TYPE_COMPUTE: type_prefix = "cs"; break; default: FIXME("Unhandled shader type %#x.\n", shader_version.type); type_prefix = "unknown"; break; } shader_addline(&buffer, "%s_%u_%u\n", type_prefix, shader_version.major, shader_version.minor); while (!fe->shader_is_end(fe_data, &ptr)) { struct wined3d_shader_instruction ins; fe->shader_read_instruction(fe_data, &ptr, &ins); if (ins.handler_idx == WINED3DSIH_TABLE_SIZE) { WARN("Skipping unrecognized instruction.\n"); shader_addline(&buffer, "\n"); continue; } if (ins.handler_idx == WINED3DSIH_DCL || ins.handler_idx == WINED3DSIH_DCL_UAV_TYPED) { shader_dump_decl_usage(&buffer, &ins.declaration.semantic, ins.flags, &shader_version); shader_dump_ins_modifiers(&buffer, &ins.declaration.semantic.reg); shader_addline(&buffer, " "); shader_dump_dst_param(&buffer, &ins.declaration.semantic.reg, &shader_version); } else if (ins.handler_idx == WINED3DSIH_DCL_CONSTANT_BUFFER) { shader_addline(&buffer, "%s ", shader_opcode_names[ins.handler_idx]); shader_dump_src_param(&buffer, &ins.declaration.src, &shader_version); shader_addline(&buffer, ", %s", ins.flags & WINED3DSI_INDEXED_DYNAMIC ? "dynamicIndexed" : "immediateIndexed"); } else if (ins.handler_idx == WINED3DSIH_DCL_FUNCTION_BODY) { shader_addline(&buffer, "%s fb%u", shader_opcode_names[ins.handler_idx], ins.declaration.index); } else if (ins.handler_idx == WINED3DSIH_DCL_FUNCTION_TABLE) { shader_addline(&buffer, "%s ft%u = {...}", shader_opcode_names[ins.handler_idx], ins.declaration.index); } else if (ins.handler_idx == WINED3DSIH_DCL_GLOBAL_FLAGS) { shader_addline(&buffer, "%s ", shader_opcode_names[ins.handler_idx]); shader_dump_global_flags(&buffer, ins.flags); } else if (ins.handler_idx == WINED3DSIH_DCL_HS_MAX_TESSFACTOR) { shader_addline(&buffer, "%s %.8e", shader_opcode_names[ins.handler_idx], ins.declaration.max_tessellation_factor); } else if (ins.handler_idx == WINED3DSIH_DCL_IMMEDIATE_CONSTANT_BUFFER) { shader_addline(&buffer, "%s {\n", shader_opcode_names[ins.handler_idx]); for (i = 0; i < ins.declaration.icb->vec4_count; ++i) { shader_addline(&buffer, " {0x%08x, 0x%08x, 0x%08x, 0x%08x},\n", ins.declaration.icb->data[4 * i + 0], ins.declaration.icb->data[4 * i + 1], ins.declaration.icb->data[4 * i + 2], ins.declaration.icb->data[4 * i + 3]); } shader_addline(&buffer, "}"); } else if (ins.handler_idx == WINED3DSIH_DCL_INDEX_RANGE) { shader_addline(&buffer, "%s ", shader_opcode_names[ins.handler_idx]); shader_dump_dst_param(&buffer, &ins.declaration.index_range.first_register, &shader_version); shader_addline(&buffer, " %u", ins.declaration.index_range.last_register); } else if (ins.handler_idx == WINED3DSIH_DCL_INDEXABLE_TEMP) { shader_addline(&buffer, "%s x[%u][%u], %u", shader_opcode_names[ins.handler_idx], ins.declaration.indexable_temp.register_idx, ins.declaration.indexable_temp.register_size, ins.declaration.indexable_temp.component_count); } else if (ins.handler_idx == WINED3DSIH_DCL_INPUT_PS) { shader_addline(&buffer, "%s ", shader_opcode_names[ins.handler_idx]); shader_dump_interpolation_mode(&buffer, ins.flags); shader_addline(&buffer, " "); shader_dump_dst_param(&buffer, &ins.declaration.dst, &shader_version); } else if (ins.handler_idx == WINED3DSIH_DCL_INPUT_PS_SGV || ins.handler_idx == WINED3DSIH_DCL_INPUT_SGV || ins.handler_idx == WINED3DSIH_DCL_INPUT_SIV || ins.handler_idx == WINED3DSIH_DCL_OUTPUT_SIV) { shader_addline(&buffer, "%s ", shader_opcode_names[ins.handler_idx]); shader_dump_dst_param(&buffer, &ins.declaration.register_semantic.reg, &shader_version); shader_addline(&buffer, ", "); shader_dump_shader_input_sysval_semantic(&buffer, ins.declaration.register_semantic.sysval_semantic); } else if (ins.handler_idx == WINED3DSIH_DCL_INPUT_PS_SIV) { shader_addline(&buffer, "%s ", shader_opcode_names[ins.handler_idx]); shader_dump_interpolation_mode(&buffer, ins.flags); shader_addline(&buffer, " "); shader_dump_dst_param(&buffer, &ins.declaration.register_semantic.reg, &shader_version); shader_addline(&buffer, ", "); shader_dump_shader_input_sysval_semantic(&buffer, ins.declaration.register_semantic.sysval_semantic); } else if (ins.handler_idx == WINED3DSIH_DCL_INPUT || ins.handler_idx == WINED3DSIH_DCL_OUTPUT) { shader_addline(&buffer, "%s ", shader_opcode_names[ins.handler_idx]); shader_dump_dst_param(&buffer, &ins.declaration.dst, &shader_version); } else if (ins.handler_idx == WINED3DSIH_DCL_INPUT_PRIMITIVE || ins.handler_idx == WINED3DSIH_DCL_OUTPUT_TOPOLOGY) { shader_addline(&buffer, "%s ", shader_opcode_names[ins.handler_idx]); shader_dump_primitive_type(&buffer, &ins.declaration.primitive_type); } else if (ins.handler_idx == WINED3DSIH_DCL_INTERFACE) { shader_addline(&buffer, "%s fp[%u][%u][%u] = {...}", shader_opcode_names[ins.handler_idx], ins.declaration.fp.index, ins.declaration.fp.array_size, ins.declaration.fp.body_count); } else if (ins.handler_idx == WINED3DSIH_DCL_RESOURCE_RAW) { shader_addline(&buffer, "%s ", shader_opcode_names[ins.handler_idx]); shader_dump_dst_param(&buffer, &ins.declaration.dst, &shader_version); } else if (ins.handler_idx == WINED3DSIH_DCL_RESOURCE_STRUCTURED) { shader_addline(&buffer, "%s ", shader_opcode_names[ins.handler_idx]); shader_dump_dst_param(&buffer, &ins.declaration.structured_resource.reg, &shader_version); shader_addline(&buffer, ", %u", ins.declaration.structured_resource.byte_stride); } else if (ins.handler_idx == WINED3DSIH_DCL_SAMPLER) { shader_addline(&buffer, "%s ", shader_opcode_names[ins.handler_idx]); shader_dump_dst_param(&buffer, &ins.declaration.dst, &shader_version); if (ins.flags == WINED3DSI_SAMPLER_COMPARISON_MODE) shader_addline(&buffer, ", comparisonMode"); } else if (ins.handler_idx == WINED3DSIH_DCL_TEMPS || ins.handler_idx == WINED3DSIH_DCL_GS_INSTANCES || ins.handler_idx == WINED3DSIH_DCL_HS_FORK_PHASE_INSTANCE_COUNT || ins.handler_idx == WINED3DSIH_DCL_HS_JOIN_PHASE_INSTANCE_COUNT || ins.handler_idx == WINED3DSIH_DCL_INPUT_CONTROL_POINT_COUNT || ins.handler_idx == WINED3DSIH_DCL_OUTPUT_CONTROL_POINT_COUNT || ins.handler_idx == WINED3DSIH_DCL_VERTICES_OUT) { shader_addline(&buffer, "%s %u", shader_opcode_names[ins.handler_idx], ins.declaration.count); } else if (ins.handler_idx == WINED3DSIH_DCL_TESSELLATOR_DOMAIN) { shader_addline(&buffer, "%s ", shader_opcode_names[ins.handler_idx]); shader_dump_tessellator_domain(&buffer, ins.declaration.tessellator_domain); } else if (ins.handler_idx == WINED3DSIH_DCL_TESSELLATOR_OUTPUT_PRIMITIVE) { shader_addline(&buffer, "%s ", shader_opcode_names[ins.handler_idx]); shader_dump_tessellator_output_primitive(&buffer, ins.declaration.tessellator_output_primitive); } else if (ins.handler_idx == WINED3DSIH_DCL_TESSELLATOR_PARTITIONING) { shader_addline(&buffer, "%s ", shader_opcode_names[ins.handler_idx]); shader_dump_tessellator_partitioning(&buffer, ins.declaration.tessellator_partitioning); } else if (ins.handler_idx == WINED3DSIH_DCL_TGSM_RAW) { shader_addline(&buffer, "%s ", shader_opcode_names[ins.handler_idx]); shader_dump_dst_param(&buffer, &ins.declaration.tgsm_raw.reg, &shader_version); shader_addline(&buffer, ", %u", ins.declaration.tgsm_raw.byte_count); } else if (ins.handler_idx == WINED3DSIH_DCL_TGSM_STRUCTURED) { shader_addline(&buffer, "%s ", shader_opcode_names[ins.handler_idx]); shader_dump_dst_param(&buffer, &ins.declaration.tgsm_structured.reg, &shader_version); shader_addline(&buffer, ", %u, %u", ins.declaration.tgsm_structured.byte_stride, ins.declaration.tgsm_structured.structure_count); } else if (ins.handler_idx == WINED3DSIH_DCL_THREAD_GROUP) { shader_addline(&buffer, "%s %u, %u, %u", shader_opcode_names[ins.handler_idx], ins.declaration.thread_group_size.x, ins.declaration.thread_group_size.y, ins.declaration.thread_group_size.z); } else if (ins.handler_idx == WINED3DSIH_DCL_UAV_RAW) { shader_addline(&buffer, "%s", shader_opcode_names[ins.handler_idx]); shader_dump_uav_flags(&buffer, ins.flags); shader_addline(&buffer, " "); shader_dump_dst_param(&buffer, &ins.declaration.dst, &shader_version); } else if (ins.handler_idx == WINED3DSIH_DCL_UAV_STRUCTURED) { shader_addline(&buffer, "%s", shader_opcode_names[ins.handler_idx]); shader_dump_uav_flags(&buffer, ins.flags); shader_addline(&buffer, " "); shader_dump_dst_param(&buffer, &ins.declaration.structured_resource.reg, &shader_version); shader_addline(&buffer, ", %u", ins.declaration.structured_resource.byte_stride); } else if (ins.handler_idx == WINED3DSIH_DEF) { shader_addline(&buffer, "def c%u = %.8e, %.8e, %.8e, %.8e", shader_get_float_offset(ins.dst[0].reg.type, ins.dst[0].reg.idx[0].offset), *(const float *)&ins.src[0].reg.u.immconst_data[0], *(const float *)&ins.src[0].reg.u.immconst_data[1], *(const float *)&ins.src[0].reg.u.immconst_data[2], *(const float *)&ins.src[0].reg.u.immconst_data[3]); } else if (ins.handler_idx == WINED3DSIH_DEFI) { shader_addline(&buffer, "defi i%u = %d, %d, %d, %d", ins.dst[0].reg.idx[0].offset, ins.src[0].reg.u.immconst_data[0], ins.src[0].reg.u.immconst_data[1], ins.src[0].reg.u.immconst_data[2], ins.src[0].reg.u.immconst_data[3]); } else if (ins.handler_idx == WINED3DSIH_DEFB) { shader_addline(&buffer, "defb b%u = %s", ins.dst[0].reg.idx[0].offset, ins.src[0].reg.u.immconst_data[0] ? "true" : "false"); } else { if (ins.predicate) { shader_addline(&buffer, "("); shader_dump_src_param(&buffer, ins.predicate, &shader_version); shader_addline(&buffer, ") "); } /* PixWin marks instructions with the coissue flag with a '+' */ if (ins.coissue) shader_addline(&buffer, "+"); shader_addline(&buffer, "%s", shader_opcode_names[ins.handler_idx]); if (ins.handler_idx == WINED3DSIH_BREAKP || ins.handler_idx == WINED3DSIH_CONTINUEP || ins.handler_idx == WINED3DSIH_IF || ins.handler_idx == WINED3DSIH_RETP || ins.handler_idx == WINED3DSIH_TEXKILL) { switch (ins.flags) { case WINED3D_SHADER_CONDITIONAL_OP_NZ: shader_addline(&buffer, "_nz"); break; case WINED3D_SHADER_CONDITIONAL_OP_Z: shader_addline(&buffer, "_z"); break; default: shader_addline(&buffer, "_unrecognized(%#x)", ins.flags); break; } } else if (ins.handler_idx == WINED3DSIH_IFC || ins.handler_idx == WINED3DSIH_BREAKC) { switch (ins.flags) { case WINED3D_SHADER_REL_OP_GT: shader_addline(&buffer, "_gt"); break; case WINED3D_SHADER_REL_OP_EQ: shader_addline(&buffer, "_eq"); break; case WINED3D_SHADER_REL_OP_GE: shader_addline(&buffer, "_ge"); break; case WINED3D_SHADER_REL_OP_LT: shader_addline(&buffer, "_lt"); break; case WINED3D_SHADER_REL_OP_NE: shader_addline(&buffer, "_ne"); break; case WINED3D_SHADER_REL_OP_LE: shader_addline(&buffer, "_le"); break; default: shader_addline(&buffer, "_(%u)", ins.flags); } } else if (ins.handler_idx == WINED3DSIH_TEX && shader_version.major >= 2 && (ins.flags & WINED3DSI_TEXLD_PROJECT)) { shader_addline(&buffer, "p"); } else if (ins.handler_idx == WINED3DSIH_RESINFO && ins.flags) { switch (ins.flags) { case WINED3DSI_RESINFO_RCP_FLOAT: shader_addline(&buffer, "_rcpFloat"); break; case WINED3DSI_RESINFO_UINT: shader_addline(&buffer, "_uint"); break; default: shader_addline(&buffer, "_unrecognized(%#x)", ins.flags); } } else if (ins.handler_idx == WINED3DSIH_SAMPLE_INFO && ins.flags) { switch (ins.flags) { case WINED3DSI_SAMPLE_INFO_UINT: shader_addline(&buffer, "_uint"); break; default: shader_addline(&buffer, "_unrecognized(%#x)", ins.flags); } } else if (ins.handler_idx == WINED3DSIH_SYNC) { shader_dump_sync_flags(&buffer, ins.flags); } else { shader_dump_precise_flags(&buffer, ins.flags); } if (wined3d_shader_instruction_has_texel_offset(&ins)) shader_addline(&buffer, "(%d,%d,%d)", ins.texel_offset.u, ins.texel_offset.v, ins.texel_offset.w); if (ins.resource_type != WINED3D_SHADER_RESOURCE_NONE) { shader_addline(&buffer, "("); shader_dump_resource_type(&buffer, ins.resource_type); shader_addline(&buffer, ")"); } if (ins.resource_data_type != WINED3D_DATA_FLOAT) shader_dump_data_type(&buffer, ins.resource_data_type); for (i = 0; i < ins.dst_count; ++i) { shader_dump_ins_modifiers(&buffer, &ins.dst[i]); shader_addline(&buffer, !i ? " " : ", "); shader_dump_dst_param(&buffer, &ins.dst[i], &shader_version); } /* Other source tokens */ for (i = ins.dst_count; i < (ins.dst_count + ins.src_count); ++i) { shader_addline(&buffer, !i ? " " : ", "); shader_dump_src_param(&buffer, &ins.src[i - ins.dst_count], &shader_version); } } shader_addline(&buffer, "\n"); } for (p = buffer.buffer; *p; p = q) { if (!(q = strstr(p, "\n"))) q = p + strlen(p); else ++q; TRACE(" %.*s", (int)(q - p), p); } string_buffer_free(&buffer); } static void shader_cleanup(struct wined3d_shader *shader) { if (shader->reg_maps.shader_version.type == WINED3D_SHADER_TYPE_HULL) { heap_free(shader->u.hs.phases.control_point); heap_free(shader->u.hs.phases.fork); heap_free(shader->u.hs.phases.join); } heap_free(shader->patch_constant_signature.elements); heap_free(shader->output_signature.elements); heap_free(shader->input_signature.elements); shader->device->shader_backend->shader_destroy(shader); shader_cleanup_reg_maps(&shader->reg_maps); heap_free(shader->byte_code); shader_delete_constant_list(&shader->constantsF); shader_delete_constant_list(&shader->constantsB); shader_delete_constant_list(&shader->constantsI); list_remove(&shader->shader_list_entry); if (shader->frontend && shader->frontend_data) shader->frontend->shader_free(shader->frontend_data); } struct shader_none_priv { const struct wined3d_vertex_pipe_ops *vertex_pipe; const struct wined3d_fragment_pipe_ops *fragment_pipe; BOOL ffp_proj_control; }; static void shader_none_handle_instruction(const struct wined3d_shader_instruction *ins) {} static void shader_none_precompile(void *shader_priv, struct wined3d_shader *shader) {} static void shader_none_select_compute(void *shader_priv, struct wined3d_context *context, const struct wined3d_state *state) {} static void shader_none_update_float_vertex_constants(struct wined3d_device *device, UINT start, UINT count) {} static void shader_none_update_float_pixel_constants(struct wined3d_device *device, UINT start, UINT count) {} static void shader_none_load_constants(void *shader_priv, struct wined3d_context *context, const struct wined3d_state *state) {} static void shader_none_destroy(struct wined3d_shader *shader) {} static void shader_none_free_context_data(struct wined3d_context *context) {} static void shader_none_init_context_state(struct wined3d_context *context) {} /* Context activation is done by the caller. */ static void shader_none_select(void *shader_priv, struct wined3d_context *context, const struct wined3d_state *state) { struct shader_none_priv *priv = shader_priv; priv->vertex_pipe->vp_enable(context, !use_vs(state)); priv->fragment_pipe->fp_enable(context, !use_ps(state)); } /* Context activation is done by the caller. */ static void shader_none_disable(void *shader_priv, struct wined3d_context *context) { struct shader_none_priv *priv = shader_priv; priv->vertex_pipe->vp_enable(context, FALSE); priv->fragment_pipe->fp_enable(context, FALSE); context->shader_update_mask = (1u << WINED3D_SHADER_TYPE_PIXEL) | (1u << WINED3D_SHADER_TYPE_VERTEX) | (1u << WINED3D_SHADER_TYPE_GEOMETRY) | (1u << WINED3D_SHADER_TYPE_HULL) | (1u << WINED3D_SHADER_TYPE_DOMAIN) | (1u << WINED3D_SHADER_TYPE_COMPUTE); } static HRESULT shader_none_alloc(struct wined3d_device *device, const struct wined3d_vertex_pipe_ops *vertex_pipe, const struct wined3d_fragment_pipe_ops *fragment_pipe) { struct fragment_caps fragment_caps; void *vertex_priv, *fragment_priv; struct shader_none_priv *priv; if (!(priv = heap_alloc(sizeof(*priv)))) return E_OUTOFMEMORY; if (!(vertex_priv = vertex_pipe->vp_alloc(&none_shader_backend, priv))) { ERR("Failed to initialize vertex pipe.\n"); heap_free(priv); return E_FAIL; } if (!(fragment_priv = fragment_pipe->alloc_private(&none_shader_backend, priv))) { ERR("Failed to initialize fragment pipe.\n"); vertex_pipe->vp_free(device, NULL); heap_free(priv); return E_FAIL; } priv->vertex_pipe = vertex_pipe; priv->fragment_pipe = fragment_pipe; fragment_pipe->get_caps(device->adapter, &fragment_caps); priv->ffp_proj_control = fragment_caps.wined3d_caps & WINED3D_FRAGMENT_CAP_PROJ_CONTROL; device->vertex_priv = vertex_priv; device->fragment_priv = fragment_priv; device->shader_priv = priv; return WINED3D_OK; } static void shader_none_free(struct wined3d_device *device, struct wined3d_context *context) { struct shader_none_priv *priv = device->shader_priv; priv->fragment_pipe->free_private(device, context); priv->vertex_pipe->vp_free(device, context); heap_free(priv); } static BOOL shader_none_allocate_context_data(struct wined3d_context *context) { return TRUE; } static void shader_none_get_caps(const struct wined3d_adapter *adapter, struct shader_caps *caps) { /* Set the shader caps to 0 for the none shader backend */ memset(caps, 0, sizeof(*caps)); } static BOOL shader_none_color_fixup_supported(struct color_fixup_desc fixup) { /* We "support" every possible fixup, since we don't support any shader * model, and will never have to actually sample a texture. */ return TRUE; } static BOOL shader_none_has_ffp_proj_control(void *shader_priv) { struct shader_none_priv *priv = shader_priv; return priv->ffp_proj_control; } static uint64_t shader_none_shader_compile(struct wined3d_context *context, const struct wined3d_shader_desc *shader_desc, enum wined3d_shader_type shader_type) { return 0; } const struct wined3d_shader_backend_ops none_shader_backend = { shader_none_handle_instruction, shader_none_precompile, shader_none_select, shader_none_select_compute, shader_none_disable, shader_none_update_float_vertex_constants, shader_none_update_float_pixel_constants, shader_none_load_constants, shader_none_destroy, shader_none_alloc, shader_none_free, shader_none_allocate_context_data, shader_none_free_context_data, shader_none_init_context_state, shader_none_get_caps, shader_none_color_fixup_supported, shader_none_has_ffp_proj_control, shader_none_shader_compile, }; static unsigned int shader_max_version_from_feature_level(enum wined3d_feature_level level) { switch (level) { case WINED3D_FEATURE_LEVEL_11_1: case WINED3D_FEATURE_LEVEL_11: return 5; case WINED3D_FEATURE_LEVEL_10_1: case WINED3D_FEATURE_LEVEL_10: return 4; case WINED3D_FEATURE_LEVEL_9_3: return 3; case WINED3D_FEATURE_LEVEL_9_2: case WINED3D_FEATURE_LEVEL_9_1: return 2; default: return 1; } } static HRESULT shader_set_function(struct wined3d_shader *shader, struct wined3d_device *device, enum wined3d_shader_type type, unsigned int float_const_count) { const struct wined3d_d3d_info *d3d_info = &shader->device->adapter->d3d_info; struct wined3d_shader_reg_maps *reg_maps = &shader->reg_maps; const struct wined3d_shader_version *version = ®_maps->shader_version; const struct wined3d_shader_frontend *fe; unsigned int backend_version; HRESULT hr; TRACE("shader %p, device %p, type %s, float_const_count %u.\n", shader, device, debug_shader_type(type), float_const_count); fe = shader->frontend; if (!(shader->frontend_data = fe->shader_init(shader->function, shader->functionLength, &shader->output_signature))) { FIXME("Failed to initialize frontend.\n"); return WINED3DERR_INVALIDCALL; } /* First pass: trace shader. */ if (TRACE_ON(d3d_shader)) shader_trace_init(fe, shader->frontend_data); /* Second pass: figure out which registers are used, what the semantics are, etc. */ if (FAILED(hr = shader_get_registers_used(shader, float_const_count))) return hr; if (version->type != type) { WARN("Wrong shader type %s.\n", debug_shader_type(reg_maps->shader_version.type)); return WINED3DERR_INVALIDCALL; } if (version->major > shader_max_version_from_feature_level(device->cs->c.state->feature_level)) { WARN("Shader version %u not supported by this device.\n", version->major); return WINED3DERR_INVALIDCALL; } switch (type) { case WINED3D_SHADER_TYPE_VERTEX: backend_version = d3d_info->limits.vs_version; break; case WINED3D_SHADER_TYPE_HULL: backend_version = d3d_info->limits.hs_version; break; case WINED3D_SHADER_TYPE_DOMAIN: backend_version = d3d_info->limits.ds_version; break; case WINED3D_SHADER_TYPE_GEOMETRY: backend_version = d3d_info->limits.gs_version; break; case WINED3D_SHADER_TYPE_PIXEL: backend_version = d3d_info->limits.ps_version; break; case WINED3D_SHADER_TYPE_COMPUTE: backend_version = d3d_info->limits.cs_version; break; default: FIXME("No backend version-checking for this shader type.\n"); backend_version = 0; } if (version->major > backend_version) { WARN("Shader version %u.%u not supported by the current shader backend.\n", version->major, version->minor); return WINED3DERR_INVALIDCALL; } shader->load_local_constsF = shader->lconst_inf_or_nan; return WINED3D_OK; } ULONG CDECL wined3d_shader_incref(struct wined3d_shader *shader) { ULONG refcount = InterlockedIncrement(&shader->ref); TRACE("%p increasing refcount to %u.\n", shader, refcount); return refcount; } static void wined3d_shader_init_object(void *object) { struct wined3d_shader *shader = object; struct wined3d_device *device = shader->device; TRACE("shader %p.\n", shader); list_add_head(&device->shaders, &shader->shader_list_entry); device->shader_backend->shader_precompile(device->shader_priv, shader); } static void wined3d_shader_destroy_object(void *object) { TRACE("object %p.\n", object); shader_cleanup(object); heap_free(object); } ULONG CDECL wined3d_shader_decref(struct wined3d_shader *shader) { ULONG refcount = InterlockedDecrement(&shader->ref); TRACE("%p decreasing refcount to %u.\n", shader, refcount); if (!refcount) { wined3d_mutex_lock(); shader->parent_ops->wined3d_object_destroyed(shader->parent); wined3d_cs_destroy_object(shader->device->cs, wined3d_shader_destroy_object, shader); wined3d_mutex_unlock(); } return refcount; } void * CDECL wined3d_shader_get_parent(const struct wined3d_shader *shader) { TRACE("shader %p.\n", shader); return shader->parent; } HRESULT CDECL wined3d_shader_get_byte_code(const struct wined3d_shader *shader, void *byte_code, UINT *byte_code_size) { TRACE("shader %p, byte_code %p, byte_code_size %p.\n", shader, byte_code, byte_code_size); if (!byte_code) { *byte_code_size = shader->byte_code_size; return WINED3D_OK; } if (*byte_code_size < shader->byte_code_size) { /* MSDN claims (for d3d8 at least) that if *byte_code_size is smaller * than the required size we should write the required size and * return D3DERR_MOREDATA. That's not actually true. */ return WINED3DERR_INVALIDCALL; } memcpy(byte_code, shader->byte_code, shader->byte_code_size); return WINED3D_OK; } /* Set local constants for d3d8 shaders. */ HRESULT CDECL wined3d_shader_set_local_constants_float(struct wined3d_shader *shader, UINT start_idx, const float *src_data, UINT count) { UINT end_idx = start_idx + count; UINT i; TRACE("shader %p, start_idx %u, src_data %p, count %u.\n", shader, start_idx, src_data, count); if (end_idx > shader->limits->constant_float) { WARN("end_idx %u > float constants limit %u.\n", end_idx, shader->limits->constant_float); end_idx = shader->limits->constant_float; } for (i = start_idx; i < end_idx; ++i) { struct wined3d_shader_lconst *lconst; float *value; if (!(lconst = heap_alloc(sizeof(*lconst)))) return E_OUTOFMEMORY; lconst->idx = i; value = (float *)lconst->value; memcpy(value, src_data + (i - start_idx) * 4 /* 4 components */, 4 * sizeof(float)); list_add_head(&shader->constantsF, &lconst->entry); if (isinf(value[0]) || isnan(value[0]) || isinf(value[1]) || isnan(value[1]) || isinf(value[2]) || isnan(value[2]) || isinf(value[3]) || isnan(value[3])) { shader->lconst_inf_or_nan = TRUE; } } return WINED3D_OK; } static void init_interpolation_compile_args(DWORD *interpolation_args, const struct wined3d_shader *pixel_shader, const struct wined3d_d3d_info *d3d_info) { if (!d3d_info->shader_output_interpolation || !pixel_shader || pixel_shader->reg_maps.shader_version.major < 4) { memset(interpolation_args, 0, sizeof(pixel_shader->u.ps.interpolation_mode)); return; } memcpy(interpolation_args, pixel_shader->u.ps.interpolation_mode, sizeof(pixel_shader->u.ps.interpolation_mode)); } void find_vs_compile_args(const struct wined3d_state *state, const struct wined3d_shader *shader, struct vs_compile_args *args, const struct wined3d_context *context) { const struct wined3d_shader *geometry_shader = state->shader[WINED3D_SHADER_TYPE_GEOMETRY]; const struct wined3d_shader *pixel_shader = state->shader[WINED3D_SHADER_TYPE_PIXEL]; const struct wined3d_shader *hull_shader = state->shader[WINED3D_SHADER_TYPE_HULL]; const struct wined3d_d3d_info *d3d_info = context->d3d_info; WORD swizzle_map = context->stream_info.swizzle_map; args->fog_src = state->render_states[WINED3D_RS_FOGTABLEMODE] == WINED3D_FOG_NONE ? VS_FOG_COORD : VS_FOG_Z; args->clip_enabled = state->render_states[WINED3D_RS_CLIPPING] && state->render_states[WINED3D_RS_CLIPPLANEENABLE]; args->point_size = state->primitive_type == WINED3D_PT_POINTLIST; args->per_vertex_point_size = shader->reg_maps.point_size; args->next_shader_type = hull_shader ? WINED3D_SHADER_TYPE_HULL : geometry_shader ? WINED3D_SHADER_TYPE_GEOMETRY : WINED3D_SHADER_TYPE_PIXEL; if (shader->reg_maps.shader_version.major >= 4) args->next_shader_input_count = hull_shader ? hull_shader->limits->packed_input : geometry_shader ? geometry_shader->limits->packed_input : pixel_shader ? pixel_shader->limits->packed_input : 0; else args->next_shader_input_count = 0; args->swizzle_map = swizzle_map; if (d3d_info->emulated_flatshading) args->flatshading = state->render_states[WINED3D_RS_SHADEMODE] == WINED3D_SHADE_FLAT; else args->flatshading = 0; init_interpolation_compile_args(args->interpolation_mode, args->next_shader_type == WINED3D_SHADER_TYPE_PIXEL ? pixel_shader : NULL, d3d_info); } static BOOL match_usage(BYTE usage1, BYTE usage_idx1, BYTE usage2, BYTE usage_idx2) { if (usage_idx1 != usage_idx2) return FALSE; if (usage1 == usage2) return TRUE; if (usage1 == WINED3D_DECL_USAGE_POSITION && usage2 == WINED3D_DECL_USAGE_POSITIONT) return TRUE; if (usage2 == WINED3D_DECL_USAGE_POSITION && usage1 == WINED3D_DECL_USAGE_POSITIONT) return TRUE; return FALSE; } BOOL vshader_get_input(const struct wined3d_shader *shader, BYTE usage_req, BYTE usage_idx_req, unsigned int *regnum) { WORD map = shader->reg_maps.input_registers; unsigned int i; for (i = 0; map; map >>= 1, ++i) { if (!(map & 1)) continue; if (match_usage(shader->u.vs.attributes[i].usage, shader->u.vs.attributes[i].usage_idx, usage_req, usage_idx_req)) { *regnum = i; return TRUE; } } return FALSE; } static HRESULT shader_init(struct wined3d_shader *shader, struct wined3d_device *device, const struct wined3d_shader_desc *desc, void *parent, const struct wined3d_parent_ops *parent_ops) { HRESULT hr; TRACE("byte_code %p, byte_code_size %#lx.\n", desc->byte_code, (long)desc->byte_code_size); if (!desc->byte_code) return WINED3DERR_INVALIDCALL; shader->ref = 1; shader->device = device; shader->parent = parent; shader->parent_ops = parent_ops; list_init(&shader->linked_programs); list_init(&shader->constantsF); list_init(&shader->constantsB); list_init(&shader->constantsI); shader->lconst_inf_or_nan = FALSE; list_init(&shader->reg_maps.indexable_temps); list_init(&shader->shader_list_entry); if (desc->byte_code_size == ~(size_t)0) { struct wined3d_shader_version shader_version; const struct wined3d_shader_frontend *fe; struct wined3d_shader_instruction ins; const DWORD *ptr; void *fe_data; if (!(shader->frontend = shader_select_frontend(WINED3D_SHADER_BYTE_CODE_FORMAT_SM1))) { FIXME("Unable to find frontend for shader.\n"); hr = WINED3DERR_INVALIDCALL; goto fail; } fe = shader->frontend; if (!(fe_data = fe->shader_init(desc->byte_code, desc->byte_code_size, &shader->output_signature))) { WARN("Failed to initialise frontend data.\n"); hr = WINED3DERR_INVALIDCALL; goto fail; } fe->shader_read_header(fe_data, &ptr, &shader_version); while (!fe->shader_is_end(fe_data, &ptr)) fe->shader_read_instruction(fe_data, &ptr, &ins); fe->shader_free(fe_data); shader->byte_code_size = (ptr - desc->byte_code) * sizeof(*ptr); if (!(shader->byte_code = heap_alloc(shader->byte_code_size))) { hr = E_OUTOFMEMORY; goto fail; } memcpy(shader->byte_code, desc->byte_code, shader->byte_code_size); shader->function = shader->byte_code; shader->functionLength = shader->byte_code_size; } else { enum wined3d_shader_byte_code_format format; unsigned int max_version; if (!(shader->byte_code = heap_alloc(desc->byte_code_size))) { hr = E_OUTOFMEMORY; goto fail; } memcpy(shader->byte_code, desc->byte_code, desc->byte_code_size); shader->byte_code_size = desc->byte_code_size; max_version = shader_max_version_from_feature_level(device->cs->c.state->feature_level); if (FAILED(hr = shader_extract_from_dxbc(shader, max_version, &format))) goto fail; if (!(shader->frontend = shader_select_frontend(format))) { FIXME("Unable to find frontend for shader.\n"); hr = WINED3DERR_INVALIDCALL; goto fail; } } return WINED3D_OK; fail: shader_cleanup(shader); return hr; } static HRESULT vertex_shader_init(struct wined3d_shader *shader, struct wined3d_device *device, const struct wined3d_shader_desc *desc, void *parent, const struct wined3d_parent_ops *parent_ops) { struct wined3d_shader_reg_maps *reg_maps = &shader->reg_maps; unsigned int i; HRESULT hr; if (FAILED(hr = shader_init(shader, device, desc, parent, parent_ops))) return hr; if (FAILED(hr = shader_set_function(shader, device, WINED3D_SHADER_TYPE_VERTEX, device->adapter->d3d_info.limits.vs_uniform_count))) { shader_cleanup(shader); return hr; } for (i = 0; i < shader->input_signature.element_count; ++i) { const struct wined3d_shader_signature_element *input = &shader->input_signature.elements[i]; if (!(reg_maps->input_registers & (1u << input->register_idx)) || !input->semantic_name) continue; shader->u.vs.attributes[input->register_idx].usage = shader_usage_from_semantic_name(input->semantic_name); shader->u.vs.attributes[input->register_idx].usage_idx = input->semantic_idx; } if (reg_maps->usesrelconstF && !list_empty(&shader->constantsF)) shader->load_local_constsF = TRUE; return WINED3D_OK; } static struct wined3d_shader_signature_element *shader_find_signature_element(const struct wined3d_shader_signature *s, unsigned int stream_idx, const char *semantic_name, unsigned int semantic_idx) { struct wined3d_shader_signature_element *e = s->elements; unsigned int i; for (i = 0; i < s->element_count; ++i) { if (e[i].stream_idx == stream_idx && !_strnicmp(e[i].semantic_name, semantic_name, -1) && e[i].semantic_idx == semantic_idx) return &e[i]; } return NULL; } BOOL shader_get_stream_output_register_info(const struct wined3d_shader *shader, const struct wined3d_stream_output_element *so_element, unsigned int *register_idx, unsigned int *component_idx) { const struct wined3d_shader_signature_element *output; unsigned int idx; if (!(output = shader_find_signature_element(&shader->output_signature, so_element->stream_idx, so_element->semantic_name, so_element->semantic_idx))) return FALSE; for (idx = 0; idx < 4; ++idx) { if (output->mask & (1u << idx)) break; } idx += so_element->component_idx; *register_idx = output->register_idx; *component_idx = idx; return TRUE; } static HRESULT geometry_shader_init_so_desc(struct wined3d_geometry_shader *gs, struct wined3d_device *device, const struct wined3d_stream_output_desc *so_desc) { struct wined3d_so_desc_entry *s; struct wine_rb_entry *entry; unsigned int i; size_t size; char *name; if ((entry = wine_rb_get(&device->so_descs, so_desc))) { gs->so_desc = &WINE_RB_ENTRY_VALUE(entry, struct wined3d_so_desc_entry, entry)->desc; return WINED3D_OK; } size = FIELD_OFFSET(struct wined3d_so_desc_entry, elements[so_desc->element_count]); for (i = 0; i < so_desc->element_count; ++i) { const char *n = so_desc->elements[i].semantic_name; if (n) size += strlen(n) + 1; } if (!(s = heap_alloc(size))) return E_OUTOFMEMORY; s->desc = *so_desc; memcpy(s->elements, so_desc->elements, so_desc->element_count * sizeof(*s->elements)); s->desc.elements = s->elements; name = (char *)&s->elements[s->desc.element_count]; for (i = 0; i < so_desc->element_count; ++i) { struct wined3d_stream_output_element *e = &s->elements[i]; if (!e->semantic_name) continue; size = strlen(e->semantic_name) + 1; memcpy(name, e->semantic_name, size); e->semantic_name = name; name += size; } if (wine_rb_put(&device->so_descs, &s->desc, &s->entry) == -1) { heap_free(s); return E_FAIL; } gs->so_desc = &s->desc; return WINED3D_OK; } static HRESULT geometry_shader_init_stream_output(struct wined3d_shader *shader, const struct wined3d_stream_output_desc *so_desc) { const struct wined3d_shader_frontend *fe = shader->frontend; const struct wined3d_shader_signature_element *output; unsigned int i, component_idx, register_idx, mask; struct wined3d_shader_version shader_version; const DWORD *ptr; void *fe_data; HRESULT hr; if (!so_desc) return WINED3D_OK; if (!(fe_data = fe->shader_init(shader->function, shader->functionLength, &shader->output_signature))) { WARN("Failed to initialise frontend data.\n"); return WINED3DERR_INVALIDCALL; } fe->shader_read_header(fe_data, &ptr, &shader_version); fe->shader_free(fe_data); switch (shader_version.type) { case WINED3D_SHADER_TYPE_VERTEX: case WINED3D_SHADER_TYPE_DOMAIN: shader->function = NULL; shader->functionLength = 0; break; case WINED3D_SHADER_TYPE_GEOMETRY: break; default: WARN("Wrong shader type %s.\n", debug_shader_type(shader_version.type)); return E_INVALIDARG; } if (!shader->function) { shader->reg_maps.shader_version = shader_version; shader->reg_maps.shader_version.type = WINED3D_SHADER_TYPE_GEOMETRY; shader_set_limits(shader); if (FAILED(hr = shader_scan_output_signature(shader))) return hr; } for (i = 0; i < so_desc->element_count; ++i) { const struct wined3d_stream_output_element *e = &so_desc->elements[i]; if (!e->semantic_name) continue; if (!(output = shader_find_signature_element(&shader->output_signature, e->stream_idx, e->semantic_name, e->semantic_idx)) || !shader_get_stream_output_register_info(shader, e, ®ister_idx, &component_idx)) { WARN("Failed to find output signature element for stream output entry.\n"); return E_INVALIDARG; } mask = wined3d_mask_from_size(e->component_count) << component_idx; if ((output->mask & 0xff & mask) != mask) { WARN("Invalid component range %u-%u (mask %#x), output mask %#x.\n", component_idx, e->component_count, mask, output->mask & 0xff); return E_INVALIDARG; } } if (FAILED(hr = geometry_shader_init_so_desc(&shader->u.gs, shader->device, so_desc))) { WARN("Failed to initialise stream output description, hr %#x.\n", hr); return hr; } return WINED3D_OK; } static HRESULT geometry_shader_init(struct wined3d_shader *shader, struct wined3d_device *device, const struct wined3d_shader_desc *desc, const struct wined3d_stream_output_desc *so_desc, void *parent, const struct wined3d_parent_ops *parent_ops) { HRESULT hr; if (FAILED(hr = shader_init(shader, device, desc, parent, parent_ops))) return hr; if (FAILED(hr = geometry_shader_init_stream_output(shader, so_desc))) goto fail; if (shader->function && FAILED(hr = shader_set_function(shader, device, WINED3D_SHADER_TYPE_GEOMETRY, 0))) goto fail; return WINED3D_OK; fail: shader_cleanup(shader); return hr; } void find_ds_compile_args(const struct wined3d_state *state, const struct wined3d_shader *shader, struct ds_compile_args *args, const struct wined3d_context *context) { const struct wined3d_shader *geometry_shader = state->shader[WINED3D_SHADER_TYPE_GEOMETRY]; const struct wined3d_shader *pixel_shader = state->shader[WINED3D_SHADER_TYPE_PIXEL]; const struct wined3d_shader *hull_shader = state->shader[WINED3D_SHADER_TYPE_HULL]; args->tessellator_output_primitive = hull_shader->u.hs.tessellator_output_primitive; args->tessellator_partitioning = hull_shader->u.hs.tessellator_partitioning; args->output_count = geometry_shader ? geometry_shader->limits->packed_input : pixel_shader ? pixel_shader->limits->packed_input : shader->limits->packed_output; args->next_shader_type = geometry_shader ? WINED3D_SHADER_TYPE_GEOMETRY : WINED3D_SHADER_TYPE_PIXEL; args->render_offscreen = context->render_offscreen; init_interpolation_compile_args(args->interpolation_mode, args->next_shader_type == WINED3D_SHADER_TYPE_PIXEL ? pixel_shader : NULL, context->d3d_info); args->padding = 0; } void find_gs_compile_args(const struct wined3d_state *state, const struct wined3d_shader *shader, struct gs_compile_args *args, const struct wined3d_context *context) { const struct wined3d_shader *pixel_shader = state->shader[WINED3D_SHADER_TYPE_PIXEL]; args->output_count = pixel_shader ? pixel_shader->limits->packed_input : shader->limits->packed_output; if (!(args->primitive_type = shader->u.gs.input_type)) args->primitive_type = state->primitive_type; init_interpolation_compile_args(args->interpolation_mode, pixel_shader, context->d3d_info); } void find_ps_compile_args(const struct wined3d_state *state, const struct wined3d_shader *shader, BOOL position_transformed, struct ps_compile_args *args, const struct wined3d_context *context) { const struct wined3d_d3d_info *d3d_info = context->d3d_info; struct wined3d_texture *texture; unsigned int i; memset(args, 0, sizeof(*args)); /* FIXME: Make sure all bits are set. */ if (!d3d_info->srgb_write_control && needs_srgb_write(d3d_info, state, &state->fb)) { static unsigned int warned = 0; args->srgb_correction = 1; if (state->blend_state && state->blend_state->desc.rt[0].enable && !warned++) WARN("Blending into a sRGB render target with no GL_ARB_framebuffer_sRGB " "support, expect rendering artifacts.\n"); } if (shader->reg_maps.shader_version.major == 1 && shader->reg_maps.shader_version.minor <= 3) { for (i = 0; i < shader->limits->sampler; ++i) { DWORD flags = state->texture_states[i][WINED3D_TSS_TEXTURE_TRANSFORM_FLAGS]; if (flags & WINED3D_TTFF_PROJECTED) { DWORD tex_transform = flags & ~WINED3D_TTFF_PROJECTED; if (!state->shader[WINED3D_SHADER_TYPE_VERTEX]) { enum wined3d_shader_resource_type resource_type = shader->reg_maps.resource_info[i].type; unsigned int j; unsigned int index = state->texture_states[i][WINED3D_TSS_TEXCOORD_INDEX]; DWORD max_valid = WINED3D_TTFF_COUNT4; for (j = 0; j < state->vertex_declaration->element_count; ++j) { struct wined3d_vertex_declaration_element *element = &state->vertex_declaration->elements[j]; if (element->usage == WINED3D_DECL_USAGE_TEXCOORD && element->usage_idx == index) { max_valid = element->format->component_count; break; } } if (!tex_transform || tex_transform > max_valid) { WARN("Fixing up projected texture transform flags from %#x to %#x.\n", tex_transform, max_valid); tex_transform = max_valid; } if ((resource_type == WINED3D_SHADER_RESOURCE_TEXTURE_1D && tex_transform > WINED3D_TTFF_COUNT1) || (resource_type == WINED3D_SHADER_RESOURCE_TEXTURE_2D && tex_transform > WINED3D_TTFF_COUNT2) || (resource_type == WINED3D_SHADER_RESOURCE_TEXTURE_3D && tex_transform > WINED3D_TTFF_COUNT3)) tex_transform |= WINED3D_PSARGS_PROJECTED; else { WARN("Application requested projected texture with unsuitable texture coordinates.\n"); WARN("(texture unit %u, transform flags %#x, sampler type %u).\n", i, tex_transform, resource_type); } } else tex_transform = WINED3D_TTFF_COUNT4 | WINED3D_PSARGS_PROJECTED; args->tex_transform |= tex_transform << i * WINED3D_PSARGS_TEXTRANSFORM_SHIFT; } } } if (shader->reg_maps.shader_version.major == 1 && shader->reg_maps.shader_version.minor <= 4) { for (i = 0; i < shader->limits->sampler; ++i) { if (!shader->reg_maps.resource_info[i].type) continue; /* Treat unbound textures as 2D. The dummy texture will provide * the proper sample value. The tex_types bitmap defaults to * 2D because of the memset. */ if (!(texture = state->textures[i])) continue; switch (wined3d_texture_gl(texture)->target) { /* RECT textures are distinguished from 2D textures via np2_fixup */ default: break; case GL_TEXTURE_3D: args->tex_types |= WINED3D_SHADER_TEX_3D << i * WINED3D_PSARGS_TEXTYPE_SHIFT; break; case GL_TEXTURE_CUBE_MAP_ARB: args->tex_types |= WINED3D_SHADER_TEX_CUBE << i * WINED3D_PSARGS_TEXTYPE_SHIFT; break; } } } else if (shader->reg_maps.shader_version.major <= 3) { for (i = 0; i < shader->limits->sampler; ++i) { enum wined3d_shader_resource_type resource_type; enum wined3d_shader_tex_types tex_type; if (!(resource_type = shader->reg_maps.resource_info[i].type)) continue; switch (resource_type) { case WINED3D_SHADER_RESOURCE_TEXTURE_3D: tex_type = WINED3D_SHADER_TEX_3D; break; case WINED3D_SHADER_RESOURCE_TEXTURE_CUBE: tex_type = WINED3D_SHADER_TEX_CUBE; break; default: tex_type = WINED3D_SHADER_TEX_2D; break; } if ((texture = state->textures[i])) { if (texture->resource.type == WINED3D_RTYPE_TEXTURE_2D && resource_type == WINED3D_SHADER_RESOURCE_TEXTURE_3D && !(texture->resource.usage & WINED3DUSAGE_LEGACY_CUBEMAP)) tex_type = WINED3D_SHADER_TEX_2D; else if (texture->resource.type == WINED3D_RTYPE_TEXTURE_3D && resource_type == WINED3D_SHADER_RESOURCE_TEXTURE_2D) tex_type = WINED3D_SHADER_TEX_3D; } args->tex_types |= tex_type << i * WINED3D_PSARGS_TEXTYPE_SHIFT; } } if (shader->reg_maps.shader_version.major >= 4) { /* In SM4+ we use dcl_sampler in order to determine if we should use shadow sampler. */ args->shadow = 0; for (i = 0 ; i < WINED3D_MAX_FRAGMENT_SAMPLERS; ++i) args->color_fixup[i] = COLOR_FIXUP_IDENTITY; args->np2_fixup = 0; } else { for (i = 0; i < WINED3D_MAX_FRAGMENT_SAMPLERS; ++i) { if (!shader->reg_maps.resource_info[i].type) continue; texture = state->textures[i]; if (!texture) { args->color_fixup[i] = COLOR_FIXUP_IDENTITY; continue; } if (can_use_texture_swizzle(d3d_info, texture->resource.format)) args->color_fixup[i] = COLOR_FIXUP_IDENTITY; else args->color_fixup[i] = texture->resource.format->color_fixup; if (texture->resource.format_flags & WINED3DFMT_FLAG_SHADOW) args->shadow |= 1u << i; /* Flag samplers that need NP2 texcoord fixup. */ if (!(texture->flags & WINED3D_TEXTURE_POW2_MAT_IDENT)) args->np2_fixup |= (1u << i); } } if (shader->reg_maps.shader_version.major >= 3) { if (position_transformed) args->vp_mode = WINED3D_VP_MODE_NONE; else if (use_vs(state)) args->vp_mode = WINED3D_VP_MODE_SHADER; else args->vp_mode = WINED3D_VP_MODE_FF; args->fog = WINED3D_FFP_PS_FOG_OFF; } else { args->vp_mode = WINED3D_VP_MODE_SHADER; if (state->render_states[WINED3D_RS_FOGENABLE]) { switch (state->render_states[WINED3D_RS_FOGTABLEMODE]) { case WINED3D_FOG_NONE: if (position_transformed || use_vs(state)) { args->fog = WINED3D_FFP_PS_FOG_LINEAR; break; } switch (state->render_states[WINED3D_RS_FOGVERTEXMODE]) { case WINED3D_FOG_NONE: /* Fall through. */ case WINED3D_FOG_LINEAR: args->fog = WINED3D_FFP_PS_FOG_LINEAR; break; case WINED3D_FOG_EXP: args->fog = WINED3D_FFP_PS_FOG_EXP; break; case WINED3D_FOG_EXP2: args->fog = WINED3D_FFP_PS_FOG_EXP2; break; } break; case WINED3D_FOG_LINEAR: args->fog = WINED3D_FFP_PS_FOG_LINEAR; break; case WINED3D_FOG_EXP: args->fog = WINED3D_FFP_PS_FOG_EXP; break; case WINED3D_FOG_EXP2: args->fog = WINED3D_FFP_PS_FOG_EXP2; break; } } else { args->fog = WINED3D_FFP_PS_FOG_OFF; } } if (!d3d_info->full_ffp_varyings) { const struct wined3d_shader *vs = state->shader[WINED3D_SHADER_TYPE_VERTEX]; args->texcoords_initialized = 0; for (i = 0; i < WINED3D_MAX_TEXTURES; ++i) { if (vs) { if (state->shader[WINED3D_SHADER_TYPE_VERTEX]->reg_maps.output_registers & (1u << i)) args->texcoords_initialized |= 1u << i; } else { const struct wined3d_stream_info *si = &context->stream_info; unsigned int coord_idx = state->texture_states[i][WINED3D_TSS_TEXCOORD_INDEX]; if ((state->texture_states[i][WINED3D_TSS_TEXCOORD_INDEX] >> WINED3D_FFP_TCI_SHIFT) & WINED3D_FFP_TCI_MASK || (coord_idx < WINED3D_MAX_TEXTURES && (si->use_map & (1u << (WINED3D_FFP_TEXCOORD0 + coord_idx))))) args->texcoords_initialized |= 1u << i; } } } else { args->texcoords_initialized = wined3d_mask_from_size(WINED3D_MAX_TEXTURES); } args->pointsprite = state->render_states[WINED3D_RS_POINTSPRITEENABLE] && state->primitive_type == WINED3D_PT_POINTLIST; if (d3d_info->ffp_alpha_test) args->alpha_test_func = WINED3D_CMP_ALWAYS - 1; else args->alpha_test_func = (state->render_states[WINED3D_RS_ALPHATESTENABLE] ? wined3d_sanitize_cmp_func(state->render_states[WINED3D_RS_ALPHAFUNC]) : WINED3D_CMP_ALWAYS) - 1; if (d3d_info->emulated_flatshading) args->flatshading = state->render_states[WINED3D_RS_SHADEMODE] == WINED3D_SHADE_FLAT; args->y_correction = (shader->reg_maps.vpos && d3d_info->frag_coord_correction) || (shader->reg_maps.usesdsy && wined3d_settings.offscreen_rendering_mode != ORM_FBO) ? !context->render_offscreen : 0; for (i = 0; i < ARRAY_SIZE(state->fb.render_targets); ++i) { struct wined3d_rendertarget_view *rtv = state->fb.render_targets[i]; if (rtv && rtv->format->id == WINED3DFMT_A8_UNORM && !is_identity_fixup(rtv->format->color_fixup)) args->rt_alpha_swizzle |= 1u << i; } args->dual_source_blend = state->blend_state && state->blend_state->dual_source; } static HRESULT pixel_shader_init(struct wined3d_shader *shader, struct wined3d_device *device, const struct wined3d_shader_desc *desc, void *parent, const struct wined3d_parent_ops *parent_ops) { const struct wined3d_gl_info *gl_info = &device->adapter->gl_info; unsigned int i, highest_reg_used = 0, num_regs_used = 0; HRESULT hr; if (FAILED(hr = shader_init(shader, device, desc, parent, parent_ops))) return hr; if (FAILED(hr = shader_set_function(shader, device, WINED3D_SHADER_TYPE_PIXEL, device->adapter->d3d_info.limits.ps_uniform_count))) { shader_cleanup(shader); return hr; } for (i = 0; i < MAX_REG_INPUT; ++i) { if (shader->u.ps.input_reg_used & (1u << i)) { ++num_regs_used; highest_reg_used = i; } } /* Don't do any register mapping magic if it is not needed, or if we can't * achieve anything anyway */ if (highest_reg_used < (gl_info->limits.glsl_varyings / 4) || num_regs_used > (gl_info->limits.glsl_varyings / 4) || shader->reg_maps.shader_version.major >= 4) { if (num_regs_used > (gl_info->limits.glsl_varyings / 4)) { /* This happens with relative addressing. The input mapper function * warns about this if the higher registers are declared too, so * don't write a FIXME here */ WARN("More varying registers used than supported\n"); } for (i = 0; i < MAX_REG_INPUT; ++i) { shader->u.ps.input_reg_map[i] = i; } shader->u.ps.declared_in_count = highest_reg_used + 1; } else { shader->u.ps.declared_in_count = 0; for (i = 0; i < MAX_REG_INPUT; ++i) { if (shader->u.ps.input_reg_used & (1u << i)) shader->u.ps.input_reg_map[i] = shader->u.ps.declared_in_count++; else shader->u.ps.input_reg_map[i] = ~0U; } } return WINED3D_OK; } enum wined3d_shader_resource_type pixelshader_get_resource_type(const struct wined3d_shader_reg_maps *reg_maps, unsigned int resource_idx, DWORD tex_types) { static enum wined3d_shader_resource_type shader_resource_type_from_shader_tex_types[] = { WINED3D_SHADER_RESOURCE_TEXTURE_2D, /* WINED3D_SHADER_TEX_2D */ WINED3D_SHADER_RESOURCE_TEXTURE_3D, /* WINED3D_SHADER_TEX_3D */ WINED3D_SHADER_RESOURCE_TEXTURE_CUBE, /* WINED3D_SHADER_TEX_CUBE */ }; unsigned int idx; if (reg_maps->shader_version.major > 3) return reg_maps->resource_info[resource_idx].type; if (!reg_maps->resource_info[resource_idx].type) return 0; idx = (tex_types >> resource_idx * WINED3D_PSARGS_TEXTYPE_SHIFT) & WINED3D_PSARGS_TEXTYPE_MASK; assert(idx < ARRAY_SIZE(shader_resource_type_from_shader_tex_types)); return shader_resource_type_from_shader_tex_types[idx]; } HRESULT CDECL wined3d_shader_create_cs(struct wined3d_device *device, const struct wined3d_shader_desc *desc, void *parent, const struct wined3d_parent_ops *parent_ops, struct wined3d_shader **shader) { struct wined3d_shader *object; HRESULT hr; TRACE("device %p, desc %p, parent %p, parent_ops %p, shader %p.\n", device, desc, parent, parent_ops, shader); if (!(object = heap_alloc_zero(sizeof(*object)))) return E_OUTOFMEMORY; if (FAILED(hr = shader_init(object, device, desc, parent, parent_ops))) { WARN("Failed to initialize compute shader, hr %#x.\n", hr); heap_free(object); return hr; } if (FAILED(hr = shader_set_function(object, device, WINED3D_SHADER_TYPE_COMPUTE, 0))) { shader_cleanup(object); heap_free(object); return hr; } wined3d_cs_init_object(device->cs, wined3d_shader_init_object, object); TRACE("Created compute shader %p.\n", object); *shader = object; return WINED3D_OK; } HRESULT CDECL wined3d_shader_create_ds(struct wined3d_device *device, const struct wined3d_shader_desc *desc, void *parent, const struct wined3d_parent_ops *parent_ops, struct wined3d_shader **shader) { struct wined3d_shader *object; HRESULT hr; TRACE("device %p, desc %p, parent %p, parent_ops %p, shader %p.\n", device, desc, parent, parent_ops, shader); if (!(object = heap_alloc_zero(sizeof(*object)))) return E_OUTOFMEMORY; if (FAILED(hr = shader_init(object, device, desc, parent, parent_ops))) { WARN("Failed to initialize domain shader, hr %#x.\n", hr); heap_free(object); return hr; } if (FAILED(hr = shader_set_function(object, device, WINED3D_SHADER_TYPE_DOMAIN, 0))) { shader_cleanup(object); heap_free(object); return hr; } wined3d_cs_init_object(device->cs, wined3d_shader_init_object, object); TRACE("Created domain shader %p.\n", object); *shader = object; return WINED3D_OK; } HRESULT CDECL wined3d_shader_create_gs(struct wined3d_device *device, const struct wined3d_shader_desc *desc, const struct wined3d_stream_output_desc *so_desc, void *parent, const struct wined3d_parent_ops *parent_ops, struct wined3d_shader **shader) { struct wined3d_shader *object; HRESULT hr; TRACE("device %p, desc %p, so_desc %p, parent %p, parent_ops %p, shader %p.\n", device, desc, so_desc, parent, parent_ops, shader); if (!(object = heap_alloc_zero(sizeof(*object)))) return E_OUTOFMEMORY; if (FAILED(hr = geometry_shader_init(object, device, desc, so_desc, parent, parent_ops))) { WARN("Failed to initialize geometry shader, hr %#x.\n", hr); heap_free(object); return hr; } wined3d_cs_init_object(device->cs, wined3d_shader_init_object, object); TRACE("Created geometry shader %p.\n", object); *shader = object; return WINED3D_OK; } HRESULT CDECL wined3d_shader_create_hs(struct wined3d_device *device, const struct wined3d_shader_desc *desc, void *parent, const struct wined3d_parent_ops *parent_ops, struct wined3d_shader **shader) { struct wined3d_shader *object; HRESULT hr; TRACE("device %p, desc %p, parent %p, parent_ops %p, shader %p.\n", device, desc, parent, parent_ops, shader); if (!(object = heap_alloc_zero(sizeof(*object)))) return E_OUTOFMEMORY; if (FAILED(hr = shader_init(object, device, desc, parent, parent_ops))) { WARN("Failed to initialize hull shader, hr %#x.\n", hr); heap_free(object); return hr; } if (FAILED(hr = shader_set_function(object, device, WINED3D_SHADER_TYPE_HULL, 0))) { shader_cleanup(object); heap_free(object); return hr; } wined3d_cs_init_object(device->cs, wined3d_shader_init_object, object); TRACE("Created hull shader %p.\n", object); *shader = object; return WINED3D_OK; } HRESULT CDECL wined3d_shader_create_ps(struct wined3d_device *device, const struct wined3d_shader_desc *desc, void *parent, const struct wined3d_parent_ops *parent_ops, struct wined3d_shader **shader) { struct wined3d_shader *object; HRESULT hr; TRACE("device %p, desc %p, parent %p, parent_ops %p, shader %p.\n", device, desc, parent, parent_ops, shader); if (!(object = heap_alloc_zero(sizeof(*object)))) return E_OUTOFMEMORY; if (FAILED(hr = pixel_shader_init(object, device, desc, parent, parent_ops))) { WARN("Failed to initialize pixel shader, hr %#x.\n", hr); heap_free(object); return hr; } wined3d_cs_init_object(device->cs, wined3d_shader_init_object, object); TRACE("Created pixel shader %p.\n", object); *shader = object; return WINED3D_OK; } HRESULT CDECL wined3d_shader_create_vs(struct wined3d_device *device, const struct wined3d_shader_desc *desc, void *parent, const struct wined3d_parent_ops *parent_ops, struct wined3d_shader **shader) { struct wined3d_shader *object; HRESULT hr; TRACE("device %p, desc %p, parent %p, parent_ops %p, shader %p.\n", device, desc, parent, parent_ops, shader); if (!(object = heap_alloc_zero(sizeof(*object)))) return E_OUTOFMEMORY; if (FAILED(hr = vertex_shader_init(object, device, desc, parent, parent_ops))) { WARN("Failed to initialize vertex shader, hr %#x.\n", hr); heap_free(object); return hr; } wined3d_cs_init_object(device->cs, wined3d_shader_init_object, object); TRACE("Created vertex shader %p.\n", object); *shader = object; return WINED3D_OK; }