[PATCH 0/13] MR441: Draft: vkd3d-shader/dxil: Implement the DXIL CMP2 instruction.
Goes atop MRs 409 and 436. The last three commits belong to this MR. -- https://gitlab.winehq.org/wine/vkd3d/-/merge_requests/441
From: Conor McCarthy <cmccarthy(a)codeweavers.com> --- libs/vkd3d-shader/d3d_asm.c | 1 + libs/vkd3d-shader/spirv.c | 2 ++ libs/vkd3d-shader/vkd3d_shader_private.h | 1 + 3 files changed, 4 insertions(+) diff --git a/libs/vkd3d-shader/d3d_asm.c b/libs/vkd3d-shader/d3d_asm.c index eae7e6b70..7a810819c 100644 --- a/libs/vkd3d-shader/d3d_asm.c +++ b/libs/vkd3d-shader/d3d_asm.c @@ -152,6 +152,7 @@ static const char * const shader_opcode_names[] = [VKD3DSIH_FIRSTBIT_LO ] = "firstbit_lo", [VKD3DSIH_FIRSTBIT_SHI ] = "firstbit_shi", [VKD3DSIH_FRC ] = "frc", + [VKD3DSIH_FREM ] = "frem", [VKD3DSIH_FTOD ] = "ftod", [VKD3DSIH_FTOI ] = "ftoi", [VKD3DSIH_FTOU ] = "ftou", diff --git a/libs/vkd3d-shader/spirv.c b/libs/vkd3d-shader/spirv.c index 6faac0ccd..4fc2348bc 100644 --- a/libs/vkd3d-shader/spirv.c +++ b/libs/vkd3d-shader/spirv.c @@ -6602,6 +6602,7 @@ static SpvOp spirv_compiler_map_alu_instruction(const struct vkd3d_shader_instru {VKD3DSIH_DTOF, SpvOpFConvert}, {VKD3DSIH_DTOI, SpvOpConvertFToS}, {VKD3DSIH_DTOU, SpvOpConvertFToU}, + {VKD3DSIH_FREM, SpvOpFRem}, {VKD3DSIH_FTOD, SpvOpFConvert}, {VKD3DSIH_IADD, SpvOpIAdd}, {VKD3DSIH_INEG, SpvOpSNegate}, @@ -9413,6 +9414,7 @@ static int spirv_compiler_handle_instruction(struct spirv_compiler *compiler, case VKD3DSIH_DTOF: case VKD3DSIH_DTOI: case VKD3DSIH_DTOU: + case VKD3DSIH_FREM: case VKD3DSIH_FTOD: case VKD3DSIH_IADD: case VKD3DSIH_INEG: diff --git a/libs/vkd3d-shader/vkd3d_shader_private.h b/libs/vkd3d-shader/vkd3d_shader_private.h index 912267e0e..39107ddbb 100644 --- a/libs/vkd3d-shader/vkd3d_shader_private.h +++ b/libs/vkd3d-shader/vkd3d_shader_private.h @@ -336,6 +336,7 @@ enum vkd3d_shader_opcode VKD3DSIH_FIRSTBIT_LO, VKD3DSIH_FIRSTBIT_SHI, VKD3DSIH_FRC, + VKD3DSIH_FREM, VKD3DSIH_FTOD, VKD3DSIH_FTOI, VKD3DSIH_FTOU, -- GitLab https://gitlab.winehq.org/wine/vkd3d/-/merge_requests/441
From: Conor McCarthy <cmccarthy(a)codeweavers.com> --- libs/vkd3d-shader/d3d_asm.c | 1 + libs/vkd3d-shader/spirv.c | 20 +++++++++----------- libs/vkd3d-shader/vkd3d_shader_private.h | 1 + 3 files changed, 11 insertions(+), 11 deletions(-) diff --git a/libs/vkd3d-shader/d3d_asm.c b/libs/vkd3d-shader/d3d_asm.c index 7a810819c..c9e75c90a 100644 --- a/libs/vkd3d-shader/d3d_asm.c +++ b/libs/vkd3d-shader/d3d_asm.c @@ -171,6 +171,7 @@ static const char * const shader_opcode_names[] = [VKD3DSIH_HS_JOIN_PHASE ] = "hs_join_phase", [VKD3DSIH_IADD ] = "iadd", [VKD3DSIH_IBFE ] = "ibfe", + [VKD3DSIH_IDIV ] = "idiv", [VKD3DSIH_IEQ ] = "ieq", [VKD3DSIH_IF ] = "if", [VKD3DSIH_IFC ] = "ifc", diff --git a/libs/vkd3d-shader/spirv.c b/libs/vkd3d-shader/spirv.c index 4fc2348bc..d9e11c82d 100644 --- a/libs/vkd3d-shader/spirv.c +++ b/libs/vkd3d-shader/spirv.c @@ -1431,13 +1431,6 @@ static uint32_t vkd3d_spirv_build_op_udiv(struct vkd3d_spirv_builder *builder, SpvOpUDiv, result_type, operand0, operand1); } -static uint32_t vkd3d_spirv_build_op_umod(struct vkd3d_spirv_builder *builder, - uint32_t result_type, uint32_t operand0, uint32_t operand1) -{ - return vkd3d_spirv_build_op_tr2(builder, &builder->function_stream, - SpvOpUMod, result_type, operand0, operand1); -} - static uint32_t vkd3d_spirv_build_op_isub(struct vkd3d_spirv_builder *builder, uint32_t result_type, uint32_t operand0, uint32_t operand1) { @@ -7017,7 +7010,7 @@ static void spirv_compiler_emit_imad(struct spirv_compiler *compiler, spirv_compiler_emit_store_dst(compiler, dst, val_id); } -static void spirv_compiler_emit_udiv(struct spirv_compiler *compiler, +static void spirv_compiler_emit_int_div(struct spirv_compiler *compiler, const struct vkd3d_shader_instruction *instruction) { uint32_t type_id, val_id, src0_id, src1_id, condition_id, uint_max_id; @@ -7025,6 +7018,10 @@ static void spirv_compiler_emit_udiv(struct spirv_compiler *compiler, const struct vkd3d_shader_dst_param *dst = instruction->dst; const struct vkd3d_shader_src_param *src = instruction->src; unsigned int component_count = 0; + SpvOp div_op, mod_op; + + div_op = instruction->handler_idx == VKD3DSIH_IDIV ? SpvOpSDiv : SpvOpUDiv; + mod_op = instruction->handler_idx == VKD3DSIH_IDIV ? SpvOpSRem : SpvOpUMod; if (dst[0].reg.type != VKD3DSPR_NULL) { @@ -7039,7 +7036,7 @@ static void spirv_compiler_emit_udiv(struct spirv_compiler *compiler, uint_max_id = spirv_compiler_get_constant_uint_vector(compiler, 0xffffffff, component_count); - val_id = vkd3d_spirv_build_op_udiv(builder, type_id, src0_id, src1_id); + val_id = vkd3d_spirv_build_op_tr2(builder, &builder->function_stream, div_op, type_id, src0_id, src1_id); /* The SPIR-V spec says: "The resulting value is undefined if Operand 2 is 0." */ val_id = vkd3d_spirv_build_op_select(builder, type_id, condition_id, val_id, uint_max_id); @@ -7062,7 +7059,7 @@ static void spirv_compiler_emit_udiv(struct spirv_compiler *compiler, 0xffffffff, component_count); } - val_id = vkd3d_spirv_build_op_umod(builder, type_id, src0_id, src1_id); + val_id = vkd3d_spirv_build_op_tr2(builder, &builder->function_stream, mod_op, type_id, src0_id, src1_id); /* The SPIR-V spec says: "The resulting value is undefined if Operand 2 is 0." */ val_id = vkd3d_spirv_build_op_select(builder, type_id, condition_id, val_id, uint_max_id); @@ -9473,8 +9470,9 @@ static int spirv_compiler_handle_instruction(struct spirv_compiler *compiler, case VKD3DSIH_IMAD: spirv_compiler_emit_imad(compiler, instruction); break; + case VKD3DSIH_IDIV: case VKD3DSIH_UDIV: - spirv_compiler_emit_udiv(compiler, instruction); + spirv_compiler_emit_int_div(compiler, instruction); break; case VKD3DSIH_FTOI: spirv_compiler_emit_ftoi(compiler, instruction); diff --git a/libs/vkd3d-shader/vkd3d_shader_private.h b/libs/vkd3d-shader/vkd3d_shader_private.h index 39107ddbb..1afcf98c5 100644 --- a/libs/vkd3d-shader/vkd3d_shader_private.h +++ b/libs/vkd3d-shader/vkd3d_shader_private.h @@ -355,6 +355,7 @@ enum vkd3d_shader_opcode VKD3DSIH_HS_JOIN_PHASE, VKD3DSIH_IADD, VKD3DSIH_IBFE, + VKD3DSIH_IDIV, VKD3DSIH_IEQ, VKD3DSIH_IF, VKD3DSIH_IFC, -- GitLab https://gitlab.winehq.org/wine/vkd3d/-/merge_requests/441
From: Conor McCarthy <cmccarthy(a)codeweavers.com> --- libs/vkd3d-shader/spirv.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/libs/vkd3d-shader/spirv.c b/libs/vkd3d-shader/spirv.c index d9e11c82d..e24431d32 100644 --- a/libs/vkd3d-shader/spirv.c +++ b/libs/vkd3d-shader/spirv.c @@ -6974,7 +6974,7 @@ static void spirv_compiler_emit_imul(struct spirv_compiler *compiler, uint32_t type_id, val_id, src0_id, src1_id; if (dst[0].reg.type != VKD3DSPR_NULL) - FIXME("Extended multiplies not implemented.\n"); /* SpvOpSMulExtended */ + FIXME("Extended multiplies not implemented.\n"); /* SpvOpSMulExtended/SpvOpUMulExtended */ if (dst[1].reg.type == VKD3DSPR_NULL) return; @@ -9465,6 +9465,7 @@ static int spirv_compiler_handle_instruction(struct spirv_compiler *compiler, spirv_compiler_emit_sincos(compiler, instruction); break; case VKD3DSIH_IMUL: + case VKD3DSIH_UMUL: spirv_compiler_emit_imul(compiler, instruction); break; case VKD3DSIH_IMAD: -- GitLab https://gitlab.winehq.org/wine/vkd3d/-/merge_requests/441
From: Conor McCarthy <cmccarthy(a)codeweavers.com> The DXIL parser uses unsigned types even if the source code uses signed, so unsigned values may be negated. --- libs/vkd3d-shader/spirv.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/libs/vkd3d-shader/spirv.c b/libs/vkd3d-shader/spirv.c index e24431d32..099da0058 100644 --- a/libs/vkd3d-shader/spirv.c +++ b/libs/vkd3d-shader/spirv.c @@ -3865,7 +3865,7 @@ static uint32_t spirv_compiler_emit_neg(struct spirv_compiler *compiler, type_id = spirv_compiler_get_type_id_for_reg(compiler, reg, write_mask); if (reg->data_type == VKD3D_DATA_FLOAT || reg->data_type == VKD3D_DATA_DOUBLE) return vkd3d_spirv_build_op_fnegate(builder, type_id, val_id); - else if (reg->data_type == VKD3D_DATA_INT) + else if (reg->data_type == VKD3D_DATA_INT || reg->data_type == VKD3D_DATA_UINT) return vkd3d_spirv_build_op_snegate(builder, type_id, val_id); FIXME("Unhandled data type %#x.\n", reg->data_type); -- GitLab https://gitlab.winehq.org/wine/vkd3d/-/merge_requests/441
From: Conor McCarthy <cmccarthy(a)codeweavers.com> --- libs/vkd3d-shader/dxil.c | 222 ++++++++++++++++++ libs/vkd3d-shader/vkd3d_shader_private.h | 1 + .../hlsl/arithmetic-float-uniform.shader_test | 14 +- tests/hlsl/cbuffer.shader_test | 6 +- tests/hlsl/const.shader_test | 2 +- tests/hlsl/cross.shader_test | 4 +- tests/hlsl/dot.shader_test | 2 +- tests/hlsl/expr-indexing.shader_test | 2 +- tests/hlsl/lerp.shader_test | 2 +- tests/hlsl/loop.shader_test | 10 +- tests/hlsl/math.shader_test | 2 +- tests/hlsl/object-references.shader_test | 2 +- tests/hlsl/reflect.shader_test | 2 +- tests/hlsl/writemask-assignop-0.shader_test | 2 +- tests/hlsl/writemask-assignop-1.shader_test | 2 +- 15 files changed, 250 insertions(+), 25 deletions(-) diff --git a/libs/vkd3d-shader/dxil.c b/libs/vkd3d-shader/dxil.c index 94f5b8622..f6237b5cb 100644 --- a/libs/vkd3d-shader/dxil.c +++ b/libs/vkd3d-shader/dxil.c @@ -242,6 +242,45 @@ enum dxil_shader_properties_tag SHADER_PROPERTIES_ENTRY_ROOT_SIG = 12, }; +enum dxil_binop_code +{ + BINOP_ADD = 0, + BINOP_SUB = 1, + BINOP_MUL = 2, + BINOP_UDIV = 3, + BINOP_SDIV = 4, + BINOP_UREM = 5, + BINOP_SREM = 6, + BINOP_SHL = 7, + BINOP_LSHR = 8, + BINOP_ASHR = 9, + BINOP_AND = 10, + BINOP_OR = 11, + BINOP_XOR = 12 +}; + +enum dxil_fast_fp_flags +{ + FP_ALLOW_UNSAFE_ALGEBRA = 0x1, + FP_NO_NAN = 0x2, + FP_NO_INF = 0x4, + FP_NO_SIGNED_ZEROS = 0x8, + FP_ALLOW_RECIPROCAL = 0x10, +}; + +enum dxil_overflowing_binop_flags +{ + /* Operation is known to never overflow. */ + OB_NO_UNSIGNED_WRAP = 0x1, + OB_NO_SIGNED_WRAP = 0x2, +}; + +enum dxil_possibly_exact_binop_flags +{ + /* "A udiv or sdiv instruction, which can be marked as "exact", indicating that no bits are destroyed." */ + PEB_EXACT = 0x1, +}; + enum dx_intrinsic_opcode { DX_LOAD_INPUT = 4, @@ -1463,6 +1502,11 @@ static inline bool sm6_type_is_integer(const struct sm6_type *type) return type->class == TYPE_CLASS_INTEGER; } +static bool sm6_type_is_bool_i16_i32_i64(const struct sm6_type *type) +{ + return type->class == TYPE_CLASS_INTEGER && (type->u.width == 1 || type->u.width >= 16); +} + static bool sm6_type_is_bool(const struct sm6_type *type) { return type->class == TYPE_CLASS_INTEGER && type->u.width == 1; @@ -2634,6 +2678,181 @@ static struct sm6_block *sm6_block_create() return block; } +static enum vkd3d_shader_opcode sm6_parser_decode_binary_op(struct sm6_parser *sm6, + uint64_t code, const struct sm6_type *type_a, const struct sm6_type *type_b) +{ + bool is_int = sm6_type_is_bool_i16_i32_i64(type_a); + bool is_bool = sm6_type_is_bool(type_a); + enum vkd3d_shader_opcode op; + bool is_valid; + + if (!is_int && !sm6_type_is_floating_point(type_a)) + { + WARN("Argument type %u is not bool, int16/32/64 or floating point.\n", type_a->class); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "An argument to a binary operation is not bool, int16/32/64 or floating point."); + return VKD3DSIH_INVALID; + } + if (type_a != type_b) + { + WARN("Type mismatch, type %u width %u vs type %u width %u.\n", type_a->class, + type_a->u.width, type_b->class, type_b->u.width); + vkd3d_shader_parser_warning(&sm6->p, VKD3D_SHADER_WARNING_DXIL_TYPE_MISMATCH, + "Type mismatch in binary operation arguments."); + } + + switch (code) + { + case BINOP_ADD: + case BINOP_SUB: + /* NEG is applied later for subtraction. */ + op = is_int ? VKD3DSIH_IADD : VKD3DSIH_ADD; + is_valid = !is_bool; + break; + case BINOP_AND: + op = VKD3DSIH_AND; + is_valid = is_int; + break; + case BINOP_ASHR: + op = VKD3DSIH_ISHR; + is_valid = is_int && !is_bool; + break; + case BINOP_LSHR: + op = VKD3DSIH_USHR; + is_valid = is_int && !is_bool; + break; + case BINOP_MUL: + op = is_int ? VKD3DSIH_UMUL : VKD3DSIH_MUL; + is_valid = !is_bool; + break; + case BINOP_OR: + op = VKD3DSIH_OR; + is_valid = is_int; + break; + case BINOP_SDIV: + op = is_int ? VKD3DSIH_IDIV : VKD3DSIH_DIV; + is_valid = !is_bool; + break; + case BINOP_SREM: + op = is_int ? VKD3DSIH_IDIV : VKD3DSIH_FREM; + is_valid = !is_bool; + break; + case BINOP_SHL: + op = VKD3DSIH_ISHL; + is_valid = is_int && !is_bool; + break; + case BINOP_UDIV: + case BINOP_UREM: + op = VKD3DSIH_UDIV; + is_valid = is_int && !is_bool; + break; + case BINOP_XOR: + op = VKD3DSIH_XOR; + is_valid = is_int; + break; + default: + FIXME("Unhandled binary op %#"PRIx64".\n", code); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Binary operation %#"PRIx64" is unhandled.", code); + return VKD3DSIH_INVALID; + } + + if (!is_valid) + { + WARN("Invalid operation %u for type %u, width %u.\n", op, type_a->class, type_a->u.width); + vkd3d_shader_parser_warning(&sm6->p, VKD3D_SHADER_WARNING_DXIL_INVALID_OPERATION, + "Binary operation %u is invalid on type class %u, width %u.", op, type_a->class, type_a->u.width); + } + + return op; +} + +static void sm6_parser_emit_binop(struct sm6_parser *sm6, const struct dxil_record *record, + struct vkd3d_shader_instruction *ins, struct sm6_value *dst) +{ + struct vkd3d_shader_src_param *src_params; + enum vkd3d_shader_opcode handler_idx; + const struct sm6_value *a, *b; + unsigned int i = 0; + uint64_t code; + + a = sm6_parser_get_value_by_ref(sm6, record, NULL, &i); + b = sm6_parser_get_value_by_ref(sm6, record, a->type, &i); + if (!a || !b) + return; + + if (!dxil_record_validate_operand_count(record, i + 1, i + 2, sm6)) + return; + + code = record->operands[i++]; + if ((handler_idx = sm6_parser_decode_binary_op(sm6, code, a->type, b->type)) == VKD3DSIH_INVALID) + return; + + vsir_instruction_init(ins, &sm6->p.location, handler_idx); + + if (record->operand_count > i && record->operands[i]) + { + uint64_t flags = record->operands[i]; + bool silence_warning = false; + + switch (handler_idx) + { + case VKD3DSIH_ADD: + case VKD3DSIH_MUL: + case VKD3DSIH_DIV: + case VKD3DSIH_FREM: + if (!(flags & FP_ALLOW_UNSAFE_ALGEBRA)) + ins->flags |= VKD3DSI_PRECISE_X; + flags &= ~FP_ALLOW_UNSAFE_ALGEBRA; + /* SPIR-V FPFastMathMode is only available in the Kernel executon model. */ + silence_warning = !(flags & ~(FP_NO_NAN | FP_NO_INF | FP_NO_SIGNED_ZEROS | FP_ALLOW_RECIPROCAL)); + break; + case VKD3DSIH_IADD: + case VKD3DSIH_UMUL: + case VKD3DSIH_ISHL: + silence_warning = !(flags & ~(OB_NO_UNSIGNED_WRAP | OB_NO_SIGNED_WRAP)); + break; + case VKD3DSIH_ISHR: + case VKD3DSIH_USHR: + case VKD3DSIH_IDIV: + case VKD3DSIH_UDIV: + silence_warning = !(flags & ~PEB_EXACT); + break; + default: + break; + } + /* The above flags are very common and cause warning spam. */ + if (flags && silence_warning) + TRACE("Ignoring flags %#"PRIx64".\n", flags); + else if (flags) + WARN("Ignoring flags %#"PRIx64".\n", flags); + } + + src_params = instruction_src_params_alloc(ins, 2, sm6); + src_param_init_from_value(&src_params[0], a); + src_param_init_from_value(&src_params[1], b); + if (code == BINOP_SUB) + src_params[1].modifiers = VKD3DSPSM_NEG; + + dst->type = a->type; + + if (handler_idx == VKD3DSIH_UMUL || handler_idx == VKD3DSIH_UDIV || handler_idx == VKD3DSIH_IDIV) + { + struct vkd3d_shader_dst_param *dst_params = instruction_dst_params_alloc(ins, 2, sm6); + unsigned int index = code != BINOP_UDIV && code != BINOP_SDIV; + + dst_param_init(&dst_params[0]); + dst_param_init(&dst_params[1]); + register_init_ssa_scalar(&dst_params[index].reg, a->type, sm6); + dst_params[index ^ 1].reg.type = VKD3DSPR_NULL; + dst->u.reg = dst_params[index].reg; + } + else + { + instruction_dst_param_init_ssa_scalar(ins, sm6); + } +} + static void sm6_parser_emit_dx_cbuffer_load(struct sm6_parser *sm6, struct sm6_block *code_block, enum dx_intrinsic_opcode op, const struct sm6_value **operands, struct vkd3d_shader_instruction *ins) { @@ -3233,6 +3452,9 @@ static enum vkd3d_result sm6_parser_function_init(struct sm6_parser *sm6, const record = block->records[i]; switch (record->code) { + case FUNC_CODE_INST_BINOP: + sm6_parser_emit_binop(sm6, record, ins, dst); + break; case FUNC_CODE_INST_CALL: sm6_parser_emit_call(sm6, record, code_block, ins, dst); break; diff --git a/libs/vkd3d-shader/vkd3d_shader_private.h b/libs/vkd3d-shader/vkd3d_shader_private.h index 1afcf98c5..7f350d7ac 100644 --- a/libs/vkd3d-shader/vkd3d_shader_private.h +++ b/libs/vkd3d-shader/vkd3d_shader_private.h @@ -192,6 +192,7 @@ enum vkd3d_shader_error VKD3D_SHADER_WARNING_DXIL_TYPE_MISMATCH = 8305, VKD3D_SHADER_WARNING_DXIL_ENTRY_POINT_MISMATCH = 8306, VKD3D_SHADER_WARNING_DXIL_INVALID_MASK = 8307, + VKD3D_SHADER_WARNING_DXIL_INVALID_OPERATION = 8308, VKD3D_SHADER_ERROR_VSIR_NOT_IMPLEMENTED = 9000, VKD3D_SHADER_ERROR_VSIR_INVALID_HANDLER = 9001, diff --git a/tests/hlsl/arithmetic-float-uniform.shader_test b/tests/hlsl/arithmetic-float-uniform.shader_test index 8aaca621a..4812d053a 100644 --- a/tests/hlsl/arithmetic-float-uniform.shader_test +++ b/tests/hlsl/arithmetic-float-uniform.shader_test @@ -10,7 +10,7 @@ float4 main() : SV_TARGET [test] uniform 0 float4 5.0 15.0 0.0 0.0 -todo(sm>=6) draw quad +draw quad probe all rgba (20.0, -10.0, 75.0, 0.33333333) 1 [pixel shader] @@ -25,7 +25,7 @@ float4 main() : SV_TARGET [test] uniform 0 float4 5.0 15.0 0.0 0.0 -todo(sm>=6) draw quad +draw quad probe all rgba (5.0, 5.0, -5.0, 3.0) 1 [pixel shader] @@ -40,7 +40,7 @@ float4 main() : SV_TARGET [test] uniform 0 float4 42.0 5.0 0.0 0.0 -todo(sm>=6) draw quad +draw quad probe all rgba (2.0, -2.0, 2.0, -2.0) 16 [pixel shader] @@ -55,7 +55,7 @@ float4 main() : SV_TARGET [test] uniform 0 float4 45.0 5.0 0.0 0.0 -todo(sm>=6) draw quad +draw quad probe all rgba (0.0, 0.0, 0.0, 0.0) [pixel shader] @@ -69,8 +69,8 @@ float4 main() : sv_target [test] uniform 0 float4 5.0 -42.1 4.0 45.0 uniform 4 float4 15.0 -5.0 4.1 5.0 -todo(sm>=6) draw quad -probe all rgba (5.0, -2.1, 4.0, 0.0) 4 +draw quad +probe all rgba (5.0, -2.1, 4.0, 0.0) 6 [require] % Infinities are not allowed in SM1 @@ -88,5 +88,5 @@ float4 main() : SV_TARGET [test] uniform 0 float4 1.0 0.0 0.0 0.0 -todo(sm>=6) draw quad +draw quad probe all rgba (1e99, 1e99, 1e99, 1e99) diff --git a/tests/hlsl/cbuffer.shader_test b/tests/hlsl/cbuffer.shader_test index dbb49ea7e..b4dc01edd 100644 --- a/tests/hlsl/cbuffer.shader_test +++ b/tests/hlsl/cbuffer.shader_test @@ -176,7 +176,7 @@ float4 main() : sv_target uniform 0 float4 1.0 2.0 3.0 4.0 uniform 4 float4 5.0 6.0 7.0 8.0 uniform 8 float4 9.0 10.0 11.0 12.0 -todo(sm>=6) draw quad +draw quad probe all rgba (509, 610, 711, 812) @@ -204,7 +204,7 @@ uniform 0 float4 0.0 1.0 2.0 3.0 uniform 4 float4 4.0 5.0 6.0 7.0 uniform 8 float4 8.0 9.0 10.0 11.0 uniform 12 float4 12.0 13.0 14.0 15.0 -todo(sm>=6) draw quad +draw quad probe all rgba (12468.0, 13509.0, 14010.0, 15011.0) @@ -326,7 +326,7 @@ float4 main() : sv_target uniform 0 float 1.0 uniform 1 float 2.0 uniform 4 float4 5.0 6.0 7.0 8.0 -todo(sm>=6) draw quad +draw quad probe all rgba (512.0, 612.0, 712.0, 812.0) diff --git a/tests/hlsl/const.shader_test b/tests/hlsl/const.shader_test index 17427c385..9541203dc 100644 --- a/tests/hlsl/const.shader_test +++ b/tests/hlsl/const.shader_test @@ -10,7 +10,7 @@ float4 main() : sv_target [test] uniform 0 float4 1.0 2.0 3.0 4.0 uniform 4 float4 0.1 0.2 0.3 0.4 -todo(sm>=6) draw quad +draw quad probe all rgba (1.1, 2.2, 3.3, 4.4) [pixel shader fail(sm<6)] diff --git a/tests/hlsl/cross.shader_test b/tests/hlsl/cross.shader_test index b84e2eaca..22f373a6d 100644 --- a/tests/hlsl/cross.shader_test +++ b/tests/hlsl/cross.shader_test @@ -12,7 +12,7 @@ float4 main() : sv_target [test] uniform 0 float4 1 -2 3 4 uniform 4 float4 10 100 1000 10000 -todo(sm>=6) draw quad +draw quad probe all rgba (-2300, -970, 120, 0) @@ -29,5 +29,5 @@ float4 main() : sv_target [test] uniform 0 float4 1 -2 3 4 -todo(sm>=6) draw quad +draw quad probe all rgba (-20, 8, 12, 3.5) diff --git a/tests/hlsl/dot.shader_test b/tests/hlsl/dot.shader_test index bb71919ce..c620e5fac 100644 --- a/tests/hlsl/dot.shader_test +++ b/tests/hlsl/dot.shader_test @@ -71,7 +71,7 @@ float4 main() : SV_TARGET % Account for both the SM1 and SM4 uniform layout uniform 0 float4 2.0 3.0 0.0 0.0 uniform 4 float4 3.0 0.0 0.0 0.0 -todo(sm>=6) draw quad +draw quad probe all rgba (6.0, 6.0, 6.0, 6.0) [pixel shader] diff --git a/tests/hlsl/expr-indexing.shader_test b/tests/hlsl/expr-indexing.shader_test index 1c598816e..c11fa6540 100644 --- a/tests/hlsl/expr-indexing.shader_test +++ b/tests/hlsl/expr-indexing.shader_test @@ -9,7 +9,7 @@ float4 main() : sv_target [test] uniform 0 float4 1.0 2.0 3.0 4.0 uniform 4 float4 5.0 6.0 7.0 8.0 -todo(sm>=6) draw quad +draw quad probe all rgba (8.0, 8.0, 8.0, 8.0) diff --git a/tests/hlsl/lerp.shader_test b/tests/hlsl/lerp.shader_test index 27c45fe7d..413bfe833 100644 --- a/tests/hlsl/lerp.shader_test +++ b/tests/hlsl/lerp.shader_test @@ -12,7 +12,7 @@ float4 main() : SV_TARGET uniform 0 float4 2.0 3.0 4.0 5.0 uniform 4 float4 0.0 -10.0 10.0 100.0 uniform 8 float4 0.0 1.0 -1.0 0.75 -todo(sm>=6) draw quad +draw quad probe all rgba (2.0, -10.0, -2.0, 76.25) [require] diff --git a/tests/hlsl/loop.shader_test b/tests/hlsl/loop.shader_test index c15e48798..35a303595 100644 --- a/tests/hlsl/loop.shader_test +++ b/tests/hlsl/loop.shader_test @@ -1,3 +1,5 @@ +% TODO: dxcompiler emits no loops for any of these test shaders. + [pixel shader] float a; @@ -16,7 +18,7 @@ float4 main() : sv_target [test] uniform 0 float 5.0 -todo(sm>=6) draw quad +draw quad probe all rgba (50.0, 50.0, 50.0, 50.0) @@ -39,7 +41,7 @@ float4 main() : sv_target [test] uniform 0 float 4.0 -todo(sm>=6) draw quad +draw quad probe all rgba (20.0, 20.0, 20.0, 20.0) [pixel shader] @@ -68,7 +70,7 @@ float4 main() : sv_target [test] uniform 0 float 4.0 -todo(sm>=6) draw quad +draw quad probe all rgba (409.1, 409.1, 409.1, 409.1) [pixel shader] @@ -98,7 +100,7 @@ float4 main() : sv_target [test] uniform 0 float 4.0 -todo(sm>=6) draw quad +draw quad probe all rgba (410.1, 410.1, 410.1, 410.1) % loop attribute by itself diff --git a/tests/hlsl/math.shader_test b/tests/hlsl/math.shader_test index 8f0592275..15f579331 100644 --- a/tests/hlsl/math.shader_test +++ b/tests/hlsl/math.shader_test @@ -14,5 +14,5 @@ float4 main() : SV_TARGET [test] uniform 0 float4 2.5 0.3 0.2 0.7 uniform 4 float4 0.1 1.5 0.0 0.0 -todo(sm>=6) draw quad +draw quad probe all rgba (-12.43, 9.833333, 1.6, 35.0) 1 diff --git a/tests/hlsl/object-references.shader_test b/tests/hlsl/object-references.shader_test index ff405559c..c857f3885 100644 --- a/tests/hlsl/object-references.shader_test +++ b/tests/hlsl/object-references.shader_test @@ -202,7 +202,7 @@ float4 main() : sv_target [test] uniform 0 float 10.0 -todo(sm>=6) draw quad +draw quad probe (0, 0) rgba (11.0, 12.0, 13.0, 11.0) diff --git a/tests/hlsl/reflect.shader_test b/tests/hlsl/reflect.shader_test index 808b4b772..25890086b 100644 --- a/tests/hlsl/reflect.shader_test +++ b/tests/hlsl/reflect.shader_test @@ -62,7 +62,7 @@ float4 main() : sv_target [test] uniform 0 float4 0.5 0.0 0.0 0.0 uniform 4 float4 0.6 0.0 0.0 0.0 -todo(sm>=6) draw quad +draw quad probe all rgba (0.14, 0.14, 0.14, 0.14) 4 [pixel shader] diff --git a/tests/hlsl/writemask-assignop-0.shader_test b/tests/hlsl/writemask-assignop-0.shader_test index fa8ecc2e4..374a38bb4 100644 --- a/tests/hlsl/writemask-assignop-0.shader_test +++ b/tests/hlsl/writemask-assignop-0.shader_test @@ -11,5 +11,5 @@ float4 main() : SV_target [test] uniform 0 float4 0.0303 0.08 0.07 0.0202 -todo(sm>=6) draw quad +draw quad probe all rgba (-0.4697, -0.02, 0.57, 0.3202) 2 diff --git a/tests/hlsl/writemask-assignop-1.shader_test b/tests/hlsl/writemask-assignop-1.shader_test index 3bebcce61..61993257c 100644 --- a/tests/hlsl/writemask-assignop-1.shader_test +++ b/tests/hlsl/writemask-assignop-1.shader_test @@ -12,5 +12,5 @@ float4 main() : SV_target [test] uniform 0 float4 0.0303 0.08 0.07 0.0202 -todo(sm>=6) draw quad +draw quad probe all rgba (0.5697, -0.08, -0.27, -0.4202) -- GitLab https://gitlab.winehq.org/wine/vkd3d/-/merge_requests/441
From: Conor McCarthy <cmccarthy(a)codeweavers.com> --- libs/vkd3d-shader/dxil.c | 1 + libs/vkd3d-shader/spirv.c | 10 ++++++++++ libs/vkd3d-shader/vkd3d_shader_private.h | 1 + 3 files changed, 12 insertions(+) diff --git a/libs/vkd3d-shader/dxil.c b/libs/vkd3d-shader/dxil.c index f6237b5cb..ce2c61b00 100644 --- a/libs/vkd3d-shader/dxil.c +++ b/libs/vkd3d-shader/dxil.c @@ -1937,6 +1937,7 @@ static void register_init_ssa_vector(struct vkd3d_shader_register *reg, const st id = sm6_parser_alloc_ssa_id(sm6); data_type = vkd3d_data_type_from_sm6_type(sm6_type_get_scalar_type(type, 0)); register_init_with_id(reg, VKD3DSPR_SSA, data_type, id); + reg->u.ssa_data_type = data_type; reg->dimension = component_count > 1 ? VSIR_DIMENSION_VEC4 : VSIR_DIMENSION_SCALAR; } diff --git a/libs/vkd3d-shader/spirv.c b/libs/vkd3d-shader/spirv.c index 099da0058..69169368d 100644 --- a/libs/vkd3d-shader/spirv.c +++ b/libs/vkd3d-shader/spirv.c @@ -3751,6 +3751,7 @@ static uint32_t spirv_compiler_emit_load_ssa_reg(struct spirv_compiler *compiler unsigned int swizzle) { struct vkd3d_spirv_builder *builder = &compiler->spirv_builder; + enum vkd3d_shader_component_type reg_component_type; unsigned int component_idx; uint32_t type_id, val_id; @@ -3759,7 +3760,16 @@ static uint32_t spirv_compiler_emit_load_ssa_reg(struct spirv_compiler *compiler assert(vkd3d_swizzle_is_scalar(swizzle)); if (reg->dimension == VSIR_DIMENSION_SCALAR) + { + reg_component_type = vkd3d_component_type_from_data_type(reg->u.ssa_data_type); + if (component_type != reg_component_type) + { + type_id = vkd3d_spirv_get_type_id(builder, component_type, 1); + val_id = vkd3d_spirv_build_op_bitcast(builder, type_id, val_id); + } + return val_id; + } type_id = vkd3d_spirv_get_type_id(builder, component_type, 1); component_idx = vkd3d_swizzle_get_component(swizzle, 0); diff --git a/libs/vkd3d-shader/vkd3d_shader_private.h b/libs/vkd3d-shader/vkd3d_shader_private.h index 7f350d7ac..d3ee57052 100644 --- a/libs/vkd3d-shader/vkd3d_shader_private.h +++ b/libs/vkd3d-shader/vkd3d_shader_private.h @@ -818,6 +818,7 @@ struct vkd3d_shader_register uint64_t immconst_uint64[VKD3D_DVEC2_SIZE]; double immconst_double[VKD3D_DVEC2_SIZE]; unsigned fp_body_idx; + enum vkd3d_data_type ssa_data_type; } u; }; -- GitLab https://gitlab.winehq.org/wine/vkd3d/-/merge_requests/441
From: Conor McCarthy <cmccarthy(a)codeweavers.com> In DXIL, bool can be the subject of a logical op, or converted to an integer or float type. --- libs/vkd3d-shader/spirv.c | 111 ++++++++++++++++++++++- libs/vkd3d-shader/vkd3d_shader_private.h | 1 + 2 files changed, 107 insertions(+), 5 deletions(-) diff --git a/libs/vkd3d-shader/spirv.c b/libs/vkd3d-shader/spirv.c index 69169368d..198984cb0 100644 --- a/libs/vkd3d-shader/spirv.c +++ b/libs/vkd3d-shader/spirv.c @@ -1150,6 +1150,20 @@ static uint32_t vkd3d_spirv_get_op_type_pointer(struct vkd3d_spirv_builder *buil vkd3d_spirv_build_op_type_pointer); } +static uint32_t vkd3d_spirv_build_op_constant_bool(struct vkd3d_spirv_builder *builder, + uint32_t result_type, uint32_t value) +{ + return vkd3d_spirv_build_op_tr(builder, &builder->global_stream, + value ? SpvOpConstantTrue : SpvOpConstantFalse, result_type); +} + +static uint32_t vkd3d_spirv_get_op_constant_bool(struct vkd3d_spirv_builder *builder, + uint32_t result_type, uint32_t value) +{ + return vkd3d_spirv_build_once2(builder, value ? SpvOpConstantTrue : SpvOpConstantFalse, result_type, value, + vkd3d_spirv_build_op_constant_bool); +} + /* Types larger than 32-bits are not supported. */ static uint32_t vkd3d_spirv_build_op_constant(struct vkd3d_spirv_builder *builder, uint32_t result_type, uint32_t value) @@ -1802,6 +1816,8 @@ static uint32_t vkd3d_spirv_get_type_id_for_data_type(struct vkd3d_spirv_builder break; case VKD3D_DATA_DOUBLE: return vkd3d_spirv_get_op_type_float(builder, 64); + case VKD3D_DATA_BOOL: + return vkd3d_spirv_get_op_type_bool(builder); default: FIXME("Unhandled data type %#x.\n", data_type); return 0; @@ -2880,6 +2896,13 @@ static uint32_t spirv_compiler_get_constant(struct spirv_compiler *compiler, case VKD3D_SHADER_COMPONENT_INT: case VKD3D_SHADER_COMPONENT_FLOAT: break; + case VKD3D_SHADER_COMPONENT_BOOL: + if (component_count == 1) + return vkd3d_spirv_get_op_constant_bool(builder, type_id, *values); + FIXME("Unsupported vector of bool.\n"); + spirv_compiler_error(compiler, VKD3D_SHADER_ERROR_SPV_INVALID_TYPE, + "Vectors of bool type are not supported."); + return vkd3d_spirv_get_op_undef(builder, type_id); default: FIXME("Unhandled component_type %#x.\n", component_type); return vkd3d_spirv_get_op_undef(builder, type_id); @@ -4212,17 +4235,41 @@ static uint32_t spirv_compiler_emit_int_to_bool(struct spirv_compiler *compiler, } static uint32_t spirv_compiler_emit_bool_to_int(struct spirv_compiler *compiler, - unsigned int component_count, uint32_t val_id) + unsigned int component_count, uint32_t val_id, bool signedness) { struct vkd3d_spirv_builder *builder = &compiler->spirv_builder; uint32_t type_id, true_id, false_id; - true_id = spirv_compiler_get_constant_uint_vector(compiler, 0xffffffff, component_count); + true_id = spirv_compiler_get_constant_uint_vector(compiler, signedness ? 0xffffffff : 1, component_count); false_id = spirv_compiler_get_constant_uint_vector(compiler, 0, component_count); type_id = vkd3d_spirv_get_type_id(builder, VKD3D_SHADER_COMPONENT_UINT, component_count); return vkd3d_spirv_build_op_select(builder, type_id, val_id, true_id, false_id); } +static uint32_t spirv_compiler_emit_bool_to_float(struct spirv_compiler *compiler, + unsigned int component_count, uint32_t val_id, bool signedness) +{ + struct vkd3d_spirv_builder *builder = &compiler->spirv_builder; + uint32_t type_id, true_id, false_id; + + true_id = spirv_compiler_get_constant_float_vector(compiler, signedness ? -1.0f : 1.0f, component_count); + false_id = spirv_compiler_get_constant_float_vector(compiler, 0.0f, component_count); + type_id = vkd3d_spirv_get_type_id(builder, VKD3D_SHADER_COMPONENT_FLOAT, component_count); + return vkd3d_spirv_build_op_select(builder, type_id, val_id, true_id, false_id); +} + +static uint32_t spirv_compiler_emit_bool_to_double(struct spirv_compiler *compiler, + unsigned int component_count, uint32_t val_id, bool signedness) +{ + struct vkd3d_spirv_builder *builder = &compiler->spirv_builder; + uint32_t type_id, true_id, false_id; + + true_id = spirv_compiler_get_constant_double_vector(compiler, signedness ? -1.0 : 1.0, component_count); + false_id = spirv_compiler_get_constant_double_vector(compiler, 0.0, component_count); + type_id = vkd3d_spirv_get_type_id(builder, VKD3D_SHADER_COMPONENT_DOUBLE, component_count); + return vkd3d_spirv_build_op_select(builder, type_id, val_id, true_id, false_id); +} + typedef uint32_t (*vkd3d_spirv_builtin_fixup_pfn)(struct spirv_compiler *compiler, uint32_t val_id); @@ -4265,7 +4312,7 @@ static uint32_t sv_instance_id_fixup(struct spirv_compiler *compiler, static uint32_t sv_front_face_fixup(struct spirv_compiler *compiler, uint32_t front_facing_id) { - return spirv_compiler_emit_bool_to_int(compiler, 1, front_facing_id); + return spirv_compiler_emit_bool_to_int(compiler, 1, front_facing_id, true); } /* frag_coord.w = 1.0f / frag_coord.w */ @@ -6632,6 +6679,49 @@ static SpvOp spirv_compiler_map_alu_instruction(const struct vkd3d_shader_instru return SpvOpMax; } +static SpvOp spirv_compiler_map_logical_instruction(const struct vkd3d_shader_instruction *instruction) +{ + switch (instruction->handler_idx) + { + case VKD3DSIH_AND: + return SpvOpLogicalAnd; + case VKD3DSIH_OR: + return SpvOpLogicalOr; + case VKD3DSIH_XOR: + return SpvOpLogicalNotEqual; + default: + return SpvOpMax; + } +} + +static void spirv_compiler_emit_bool_cast(struct spirv_compiler *compiler, + const struct vkd3d_shader_instruction *instruction) +{ + const struct vkd3d_shader_dst_param *dst = instruction->dst; + const struct vkd3d_shader_src_param *src = instruction->src; + uint32_t val_id; + + assert(src->reg.data_type == VKD3D_DATA_BOOL && dst->reg.data_type != VKD3D_DATA_BOOL); + + val_id = spirv_compiler_emit_load_src(compiler, src, dst->write_mask); + if (dst->reg.data_type == VKD3D_DATA_FLOAT) + { + val_id = spirv_compiler_emit_bool_to_float(compiler, 1, val_id, instruction->handler_idx == VKD3DSIH_ITOF); + } + else if (dst->reg.data_type == VKD3D_DATA_DOUBLE) + { + val_id = spirv_compiler_emit_bool_to_double(compiler, 1, val_id, instruction->handler_idx == VKD3DSIH_ITOD); + } + else + { + WARN("Unhandled data type %u.\n", dst->reg.data_type); + spirv_compiler_error(compiler, VKD3D_SHADER_ERROR_SPV_INVALID_TYPE, + "Register data type %u is unhandled.", dst->reg.data_type); + } + + spirv_compiler_emit_store_dst(compiler, dst, val_id); +} + static void spirv_compiler_emit_alu_instruction(struct spirv_compiler *compiler, const struct vkd3d_shader_instruction *instruction) { @@ -6643,7 +6733,18 @@ static void spirv_compiler_emit_alu_instruction(struct spirv_compiler *compiler, unsigned int i; SpvOp op; - op = spirv_compiler_map_alu_instruction(instruction); + if (src->reg.data_type == VKD3D_DATA_BOOL) + { + if (dst->reg.data_type == VKD3D_DATA_BOOL) + op = spirv_compiler_map_logical_instruction(instruction); + else + return spirv_compiler_emit_bool_cast(compiler, instruction); + } + else + { + op = spirv_compiler_map_alu_instruction(instruction); + } + if (op == SpvOpMax) { ERR("Unexpected instruction %#x.\n", instruction->handler_idx); @@ -7318,7 +7419,7 @@ static void spirv_compiler_emit_comparison_instruction(struct spirv_compiler *co result_id = vkd3d_spirv_build_op_tr2(builder, &builder->function_stream, op, type_id, src0_id, src1_id); - result_id = spirv_compiler_emit_bool_to_int(compiler, component_count, result_id); + result_id = spirv_compiler_emit_bool_to_int(compiler, component_count, result_id, true); spirv_compiler_emit_store_reg(compiler, &dst->reg, dst->write_mask, result_id); } diff --git a/libs/vkd3d-shader/vkd3d_shader_private.h b/libs/vkd3d-shader/vkd3d_shader_private.h index d3ee57052..6bc7dd0c7 100644 --- a/libs/vkd3d-shader/vkd3d_shader_private.h +++ b/libs/vkd3d-shader/vkd3d_shader_private.h @@ -93,6 +93,7 @@ enum vkd3d_shader_error VKD3D_SHADER_ERROR_SPV_DESCRIPTOR_IDX_UNSUPPORTED = 2003, VKD3D_SHADER_ERROR_SPV_STENCIL_EXPORT_UNSUPPORTED = 2004, VKD3D_SHADER_ERROR_SPV_OUT_OF_MEMORY = 2005, + VKD3D_SHADER_ERROR_SPV_INVALID_TYPE = 2006, VKD3D_SHADER_WARNING_SPV_INVALID_SWIZZLE = 2300, -- GitLab https://gitlab.winehq.org/wine/vkd3d/-/merge_requests/441
From: Conor McCarthy <cmccarthy(a)codeweavers.com> ITOI and UTOU may cast from a bool to a 32-bit integer. Cast to a 64-bit integer from a smaller type will be added later. --- libs/vkd3d-shader/d3d_asm.c | 2 ++ libs/vkd3d-shader/spirv.c | 6 ++++++ libs/vkd3d-shader/vkd3d_shader_private.h | 2 ++ 3 files changed, 10 insertions(+) diff --git a/libs/vkd3d-shader/d3d_asm.c b/libs/vkd3d-shader/d3d_asm.c index c9e75c90a..c0779d718 100644 --- a/libs/vkd3d-shader/d3d_asm.c +++ b/libs/vkd3d-shader/d3d_asm.c @@ -199,6 +199,7 @@ static const char * const shader_opcode_names[] = [VKD3DSIH_ISHR ] = "ishr", [VKD3DSIH_ITOD ] = "itod", [VKD3DSIH_ITOF ] = "itof", + [VKD3DSIH_ITOI ] = "itoi", [VKD3DSIH_LABEL ] = "label", [VKD3DSIH_LD ] = "ld", [VKD3DSIH_LD2DMS ] = "ld2dms", @@ -306,6 +307,7 @@ static const char * const shader_opcode_names[] = [VKD3DSIH_USHR ] = "ushr", [VKD3DSIH_UTOD ] = "utod", [VKD3DSIH_UTOF ] = "utof", + [VKD3DSIH_UTOU ] = "utou", [VKD3DSIH_XOR ] = "xor", }; diff --git a/libs/vkd3d-shader/spirv.c b/libs/vkd3d-shader/spirv.c index 198984cb0..35a341787 100644 --- a/libs/vkd3d-shader/spirv.c +++ b/libs/vkd3d-shader/spirv.c @@ -6712,6 +6712,10 @@ static void spirv_compiler_emit_bool_cast(struct spirv_compiler *compiler, { val_id = spirv_compiler_emit_bool_to_double(compiler, 1, val_id, instruction->handler_idx == VKD3DSIH_ITOD); } + else if (dst->reg.data_type == VKD3D_DATA_UINT) + { + val_id = spirv_compiler_emit_bool_to_int(compiler, 1, val_id, instruction->handler_idx == VKD3DSIH_ITOI); + } else { WARN("Unhandled data type %u.\n", dst->reg.data_type); @@ -9530,12 +9534,14 @@ static int spirv_compiler_handle_instruction(struct spirv_compiler *compiler, case VKD3DSIH_ISHR: case VKD3DSIH_ITOD: case VKD3DSIH_ITOF: + case VKD3DSIH_ITOI: case VKD3DSIH_MUL: case VKD3DSIH_NOT: case VKD3DSIH_OR: case VKD3DSIH_USHR: case VKD3DSIH_UTOD: case VKD3DSIH_UTOF: + case VKD3DSIH_UTOU: case VKD3DSIH_XOR: spirv_compiler_emit_alu_instruction(compiler, instruction); break; diff --git a/libs/vkd3d-shader/vkd3d_shader_private.h b/libs/vkd3d-shader/vkd3d_shader_private.h index 6bc7dd0c7..c10e77f53 100644 --- a/libs/vkd3d-shader/vkd3d_shader_private.h +++ b/libs/vkd3d-shader/vkd3d_shader_private.h @@ -385,6 +385,7 @@ enum vkd3d_shader_opcode VKD3DSIH_ISHR, VKD3DSIH_ITOD, VKD3DSIH_ITOF, + VKD3DSIH_ITOI, VKD3DSIH_LABEL, VKD3DSIH_LD, VKD3DSIH_LD2DMS, @@ -492,6 +493,7 @@ enum vkd3d_shader_opcode VKD3DSIH_USHR, VKD3DSIH_UTOD, VKD3DSIH_UTOF, + VKD3DSIH_UTOU, VKD3DSIH_XOR, VKD3DSIH_INVALID, -- GitLab https://gitlab.winehq.org/wine/vkd3d/-/merge_requests/441
From: Conor McCarthy <cmccarthy(a)codeweavers.com> --- libs/vkd3d-shader/spirv.c | 14 ++++++-------- 1 file changed, 6 insertions(+), 8 deletions(-) diff --git a/libs/vkd3d-shader/spirv.c b/libs/vkd3d-shader/spirv.c index 35a341787..ee9cca1bc 100644 --- a/libs/vkd3d-shader/spirv.c +++ b/libs/vkd3d-shader/spirv.c @@ -2969,12 +2969,6 @@ static uint32_t spirv_compiler_get_constant_vector(struct spirv_compiler *compil return spirv_compiler_get_constant(compiler, component_type, component_count, values); } -static uint32_t spirv_compiler_get_constant_int_vector(struct spirv_compiler *compiler, - uint32_t value, unsigned int component_count) -{ - return spirv_compiler_get_constant_vector(compiler, VKD3D_SHADER_COMPONENT_INT, component_count, value); -} - static uint32_t spirv_compiler_get_constant_uint_vector(struct spirv_compiler *compiler, uint32_t value, unsigned int component_count) { @@ -7190,6 +7184,7 @@ static void spirv_compiler_emit_ftoi(struct spirv_compiler *compiler, const struct vkd3d_shader_dst_param *dst = instruction->dst; const struct vkd3d_shader_src_param *src = instruction->src; uint32_t src_type_id, dst_type_id, condition_type_id; + enum vkd3d_shader_component_type component_type; unsigned int component_count; assert(instruction->dst_count == 1); @@ -7207,8 +7202,11 @@ static void spirv_compiler_emit_ftoi(struct spirv_compiler *compiler, int_min_id = spirv_compiler_get_constant_float_vector(compiler, -2147483648.0f, component_count); val_id = vkd3d_spirv_build_op_glsl_std450_max(builder, src_type_id, src_id, int_min_id); + /* DXIL uses unsigned int but requires a signed conversion. */ + component_type = vkd3d_component_type_from_data_type(dst->reg.data_type); + float_max_id = spirv_compiler_get_constant_float_vector(compiler, 2147483648.0f, component_count); - int_max_id = spirv_compiler_get_constant_int_vector(compiler, INT_MAX, component_count); + int_max_id = spirv_compiler_get_constant_vector(compiler, component_type, component_count, INT_MAX); condition_type_id = vkd3d_spirv_get_type_id(builder, VKD3D_SHADER_COMPONENT_BOOL, component_count); condition_id = vkd3d_spirv_build_op_tr2(builder, &builder->function_stream, SpvOpFOrdGreaterThanEqual, condition_type_id, val_id, float_max_id); @@ -7216,7 +7214,7 @@ static void spirv_compiler_emit_ftoi(struct spirv_compiler *compiler, val_id = vkd3d_spirv_build_op_tr1(builder, &builder->function_stream, SpvOpConvertFToS, dst_type_id, val_id); val_id = vkd3d_spirv_build_op_select(builder, dst_type_id, condition_id, int_max_id, val_id); - zero_id = spirv_compiler_get_constant_int_vector(compiler, 0, component_count); + zero_id = spirv_compiler_get_constant_vector(compiler, component_type, component_count, 0); condition_id = vkd3d_spirv_build_op_tr1(builder, &builder->function_stream, SpvOpIsNan, condition_type_id, src_id); val_id = vkd3d_spirv_build_op_select(builder, dst_type_id, condition_id, zero_id, val_id); -- GitLab https://gitlab.winehq.org/wine/vkd3d/-/merge_requests/441
From: Conor McCarthy <cmccarthy(a)codeweavers.com> --- libs/vkd3d-shader/dxil.c | 172 ++++++++++++++++++ tests/hlsl/arithmetic-int-uniform.shader_test | 14 +- tests/hlsl/array-index-expr.shader_test | 16 +- tests/hlsl/asfloat.shader_test | 2 +- tests/hlsl/asuint.shader_test | 4 +- tests/hlsl/d3dcolor-to-ubyte4.shader_test | 4 +- tests/hlsl/function-cast.shader_test | 4 +- tests/hlsl/lerp.shader_test | 2 +- .../shader-interstage-interface.shader_test | 2 +- 9 files changed, 196 insertions(+), 24 deletions(-) diff --git a/libs/vkd3d-shader/dxil.c b/libs/vkd3d-shader/dxil.c index ce2c61b00..5f7232325 100644 --- a/libs/vkd3d-shader/dxil.c +++ b/libs/vkd3d-shader/dxil.c @@ -289,6 +289,23 @@ enum dx_intrinsic_opcode DX_CBUFFER_LOAD_LEGACY = 59, }; +enum dxil_cast_code +{ + CAST_TRUNC = 0, + CAST_ZEXT = 1, + CAST_SEXT = 2, + CAST_FPTOUI = 3, + CAST_FPTOSI = 4, + CAST_UITOFP = 5, + CAST_SITOFP = 6, + CAST_FPTRUNC = 7, + CAST_FPEXT = 8, + CAST_PTRTOINT = 9, + CAST_INTTOPTR = 10, + CAST_BITCAST = 11, + CAST_ADDRSPACECAST = 12, +}; + struct sm6_pointer_info { const struct sm6_type *type; @@ -3243,6 +3260,158 @@ static void sm6_parser_emit_call(struct sm6_parser *sm6, const struct dxil_recor fn_value->u.function.name, &operands[1], operand_count - 1, ins, dst); } +static enum vkd3d_shader_opcode sm6_map_cast_op(uint64_t code, const struct sm6_type *from, + const struct sm6_type *to, struct sm6_parser *sm6) +{ + enum vkd3d_shader_opcode op = VKD3DSIH_INVALID; + bool from_int, to_int; + bool is_valid = false; + + /* NOTE: DXIL currently doesn't use vectors here. */ + if (!sm6_type_is_scalar(from) || !sm6_type_is_scalar(to)) + { + FIXME("Unhandled non-scalar type class %u and/or %u.\n", from->class, to->class); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Cast of non-scalar types is not implemented."); + /* width is not available for non-scalar types so we can't attempt to compile it. */ + return VKD3DSIH_INVALID; + } + if (to->u.width == 8 || from->u.width == 8) + { + FIXME("Unhandled 8-bit value.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Cast to/from an 8-bit type is not implemented."); + return VKD3DSIH_INVALID; + } + + from_int = sm6_type_is_integer(from); + to_int = sm6_type_is_integer(to); + + /* DXC emits minimum precision types as 16-bit. These must be emitted + * as 32-bit in VSIR, so all width extensions to 32 bits are no-ops. */ + switch (code) + { + case CAST_TRUNC: + /* nop or min precision. TODO: native 16-bit */ + if (to->u.width == from->u.width || (to->u.width == 16 && from->u.width == 32)) + op = VKD3DSIH_NOP; + else + op = VKD3DSIH_UTOU; + is_valid = from_int && to_int && to->u.width <= from->u.width; + break; + case CAST_ZEXT: + case CAST_SEXT: + /* nop or min precision. TODO: native 16-bit */ + if (to->u.width == from->u.width || (to->u.width == 32 && from->u.width == 16)) + { + op = VKD3DSIH_NOP; + is_valid = from_int && to_int; + } + else if (to->u.width > from->u.width) + { + op = (code == CAST_ZEXT) ? VKD3DSIH_UTOU : VKD3DSIH_ITOI; + assert(from->u.width == 1 || to->u.width == 64); + is_valid = from_int && to_int; + } + break; + case CAST_FPTOUI: + op = (from->u.width == 64) ? VKD3DSIH_DTOU : VKD3DSIH_FTOU; + is_valid = !from_int && to_int && to->u.width > 1; + break; + case CAST_FPTOSI: + op = (from->u.width == 64) ? VKD3DSIH_DTOI : VKD3DSIH_FTOI; + is_valid = !from_int && to_int && to->u.width > 1; + break; + case CAST_UITOFP: + op = (to->u.width == 64) ? VKD3DSIH_UTOD : VKD3DSIH_UTOF; + is_valid = from_int && !to_int; + break; + case CAST_SITOFP: + op = (to->u.width == 64) ? VKD3DSIH_ITOD : VKD3DSIH_ITOF; + is_valid = from_int && !to_int; + break; + case CAST_FPTRUNC: + /* TODO: native 16-bit */ + op = (from->u.width == 64) ? VKD3DSIH_DTOF : VKD3DSIH_NOP; + is_valid = !from_int && !to_int; + break; + case CAST_FPEXT: + /* TODO: native 16-bit */ + op = (to->u.width == 64) ? VKD3DSIH_FTOD : VKD3DSIH_NOP; + is_valid = !from_int && !to_int; + break; + case CAST_BITCAST: + op = VKD3DSIH_MOV; + is_valid = to->u.width == from->u.width; + break; + default: + FIXME("Unhandled cast op %"PRIu64".\n", code); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Cast operation %"PRIu64" is unhandled.\n", code); + return VKD3DSIH_INVALID; + } + + if (!is_valid) + { + FIXME("Invalid types %u and/or %u for op %"PRIu64".\n", from->class, to->class, code); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Cast operation %"PRIu64" from type class %u, width %u to type class %u, width %u is invalid.\n", + code, from->class, from->u.width, to->class, to->u.width); + return VKD3DSIH_INVALID; + } + + return op; +} + +static void sm6_parser_emit_cast(struct sm6_parser *sm6, const struct dxil_record *record, + struct vkd3d_shader_instruction *ins, struct sm6_value *dst) +{ + struct vkd3d_shader_src_param *src_param; + enum vkd3d_shader_opcode handler_idx; + const struct sm6_value *value; + const struct sm6_type *type; + unsigned int i = 0; + + if (!(value = sm6_parser_get_value_by_ref(sm6, record, NULL, &i))) + return; + + if (!dxil_record_validate_operand_count(record, i + 2, i + 2, sm6)) + return; + + if (!(type = sm6_parser_get_type(sm6, record->operands[i++]))) + return; + + dst->type = type; + + if (sm6_type_is_pointer(type)) + { + *dst = *value; + dst->type = type; + ins->handler_idx = VKD3DSIH_NOP; + return; + } + + if ((handler_idx = sm6_map_cast_op(record->operands[i], value->type, type, sm6)) == VKD3DSIH_INVALID) + return; + + vsir_instruction_init(ins, &sm6->p.location, handler_idx); + + if (handler_idx == VKD3DSIH_NOP) + { + dst->u.reg = value->u.reg; + return; + } + + src_param = instruction_src_params_alloc(ins, 1, sm6); + src_param_init_from_value(src_param, value); + + instruction_dst_param_init_ssa_scalar(ins, sm6); + + /* bitcast */ + if (handler_idx == VKD3DSIH_MOV) + src_param->reg.data_type = dst->u.reg.data_type; +} + static void sm6_parser_emit_extractval(struct sm6_parser *sm6, const struct dxil_record *record, struct vkd3d_shader_instruction *ins, struct sm6_value *dst) { @@ -3459,6 +3628,9 @@ static enum vkd3d_result sm6_parser_function_init(struct sm6_parser *sm6, const case FUNC_CODE_INST_CALL: sm6_parser_emit_call(sm6, record, code_block, ins, dst); break; + case FUNC_CODE_INST_CAST: + sm6_parser_emit_cast(sm6, record, ins, dst); + break; case FUNC_CODE_INST_EXTRACTVAL: sm6_parser_emit_extractval(sm6, record, ins, dst); break; diff --git a/tests/hlsl/arithmetic-int-uniform.shader_test b/tests/hlsl/arithmetic-int-uniform.shader_test index 726a191a7..7f5cdaaa6 100644 --- a/tests/hlsl/arithmetic-int-uniform.shader_test +++ b/tests/hlsl/arithmetic-int-uniform.shader_test @@ -10,7 +10,7 @@ float4 main() : SV_TARGET [test] uniform 0 float4 5.0 16.0 0.0 0.0 -todo(sm>=6) draw quad +draw quad probe all rgba (21.0, -11.0, 80.0, 0.0) [pixel shader] @@ -25,7 +25,7 @@ float4 main() : SV_TARGET [test] uniform 0 float4 5.0 16.0 0.0 0.0 -todo(sm>=6) draw quad +draw quad probe all rgba (5.0, 5.0, -5.0, 3.0) [pixel shader] @@ -40,7 +40,7 @@ float4 main() : SV_TARGET [test] uniform 0 float4 42.0 5.0 0.0 0.0 -todo(sm>=6) draw quad +draw quad probe all rgba (8.0, -8.0, -8.0, 8.0) [pixel shader] @@ -55,7 +55,7 @@ float4 main() : SV_TARGET [test] uniform 0 float4 42.0 5.0 0.0 0.0 -todo(sm>=6) draw quad +draw quad probe all rgba (2.0, -2.0, 2.0, -2.0) [pixel shader] @@ -70,7 +70,7 @@ float4 main() : SV_TARGET [test] uniform 0 float4 45.0 5.0 0.0 0.0 -todo(sm>=6) draw quad +draw quad probe all rgba (9.0, -9.0, -9.0, 9.0) [pixel shader] @@ -85,7 +85,7 @@ float4 main() : SV_TARGET [test] uniform 0 float4 45.0 5.0 0.0 0.0 -todo(sm>=6) draw quad +draw quad probe all rgba (0.0, 0.0, 0.0, 0.0) [pixel shader] @@ -117,5 +117,5 @@ float4 main() : sv_target [test] uniform 0 float4 45.0 5.0 50.0 10.0 uniform 4 float4 3.0 8.0 2.0 5.0 -todo(sm>=6) draw quad +draw quad probe all rgba (9.0, 5.0, 1.0, 3.0) diff --git a/tests/hlsl/array-index-expr.shader_test b/tests/hlsl/array-index-expr.shader_test index b7d91ea09..0b45df927 100644 --- a/tests/hlsl/array-index-expr.shader_test +++ b/tests/hlsl/array-index-expr.shader_test @@ -12,17 +12,17 @@ uniform 0 float4 1.0 2.0 3.0 4.0 uniform 4 float4 5.0 6.0 7.0 8.0 uniform 8 float4 9.0 10.0 11.0 12.0 uniform 12 float4 0 0 0 0 -todo draw quad -todo probe all rgba (1.0, 2.0, 3.0, 4.0) +todo(sm<6) draw quad +todo(sm<6) probe all rgba (1.0, 2.0, 3.0, 4.0) uniform 12 float4 1 0 0 0 -todo draw quad -todo probe all rgba (5.0, 6.0, 7.0, 8.0) +todo(sm<6) draw quad +todo(sm<6) probe all rgba (5.0, 6.0, 7.0, 8.0) uniform 12 float4 0 1 0 0 -todo draw quad -todo probe all rgba (5.0, 6.0, 7.0, 8.0) +todo(sm<6) draw quad +todo(sm<6) probe all rgba (5.0, 6.0, 7.0, 8.0) uniform 12 float4 1 1 0 0 -todo draw quad -todo probe all rgba (9.0, 10.0, 11.0, 12.0) +todo(sm<6) draw quad +todo(sm<6) probe all rgba (9.0, 10.0, 11.0, 12.0) [pixel shader] diff --git a/tests/hlsl/asfloat.shader_test b/tests/hlsl/asfloat.shader_test index f43f2fdff..2184c1869 100644 --- a/tests/hlsl/asfloat.shader_test +++ b/tests/hlsl/asfloat.shader_test @@ -20,7 +20,7 @@ float4 main() : sv_target [test] uniform 0 float4 123.0 -2.0 456 0.01 -todo(sm>=6) draw quad +draw quad probe (320,240) rgba (123.0, -2.0, 456.0, 0.01) [pixel shader] diff --git a/tests/hlsl/asuint.shader_test b/tests/hlsl/asuint.shader_test index 0a2e39e53..50b0895e9 100644 --- a/tests/hlsl/asuint.shader_test +++ b/tests/hlsl/asuint.shader_test @@ -20,7 +20,7 @@ float4 main() : sv_target [test] uniform 0 uint4 123 0xc0000000 456 0x7fd69345 -todo(sm>=6) draw quad +draw quad probe (320,240) rgba (123.0, 3221225472.0, 456.0, 2144768896.0) @@ -37,7 +37,7 @@ float4 main() : sv_target uniform 0 uint4 11 12 0 0 uniform 4 uint4 13 14 0 0 uniform 8 uint4 20 21 22 23 -todo(sm>=6) draw quad +draw quad probe (320,240) rgba (13.0, 21.0, 0.0, 0.0) diff --git a/tests/hlsl/d3dcolor-to-ubyte4.shader_test b/tests/hlsl/d3dcolor-to-ubyte4.shader_test index 01b7f2f64..dca307cdb 100644 --- a/tests/hlsl/d3dcolor-to-ubyte4.shader_test +++ b/tests/hlsl/d3dcolor-to-ubyte4.shader_test @@ -11,7 +11,7 @@ float4 main() : sv_target [test] uniform 0 float4 -0.5 6.5 7.5 3.4 -todo(sm>=6) draw quad +draw quad probe all rgba (1912.0, 1657.0, -127.0, 867.0) 1 [pixel shader] @@ -24,5 +24,5 @@ float4 main() : sv_target [test] uniform 0 float4 -0.5 6.5 7.5 3.4 -todo(sm>=6) draw quad +draw quad probe all rgba (-127.0, -127.0, -127.0, -127.0) 1 diff --git a/tests/hlsl/function-cast.shader_test b/tests/hlsl/function-cast.shader_test index a2b9cf4f4..c92289863 100644 --- a/tests/hlsl/function-cast.shader_test +++ b/tests/hlsl/function-cast.shader_test @@ -18,7 +18,7 @@ float4 main() : sv_target [test] uniform 0 float4 -1.9 -1.0 2.9 4.0 -todo draw quad +todo(sm<6) draw quad probe all rgba (-1.0, -1.0, 2.0, 4.0) % As above, but cast "x" to float4 first. @@ -89,5 +89,5 @@ float4 main() : sv_target [test] uniform 0 int4 -2 0 1 -3000000 -todo draw quad +todo(sm<6) draw quad probe all rgba (-1.0, 0.0, 1.0, -3000000.0) diff --git a/tests/hlsl/lerp.shader_test b/tests/hlsl/lerp.shader_test index 413bfe833..15e90cef9 100644 --- a/tests/hlsl/lerp.shader_test +++ b/tests/hlsl/lerp.shader_test @@ -32,7 +32,7 @@ float4 main() : SV_TARGET uniform 0 int4 2 3 4 0 uniform 4 int4 0 -10 10 1000000 uniform 8 int4 0 1 -1 1000000 -todo(sm>=6) draw quad +draw quad probe all rgba (2.0, -10.0, -2.0, 1e12) diff --git a/tests/hlsl/shader-interstage-interface.shader_test b/tests/hlsl/shader-interstage-interface.shader_test index 271eb3bbe..584b88cf9 100644 --- a/tests/hlsl/shader-interstage-interface.shader_test +++ b/tests/hlsl/shader-interstage-interface.shader_test @@ -52,5 +52,5 @@ void main(float4 position : SV_Position, float2 t0 : TEXCOORD0, } [test] -todo(sm>=6) draw triangle strip 4 +draw triangle strip 4 probe all rgba (10.0, 8.0, 7.0, 3.0) -- GitLab https://gitlab.winehq.org/wine/vkd3d/-/merge_requests/441
From: Conor McCarthy <cmccarthy(a)codeweavers.com> --- libs/vkd3d-shader/spirv.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/libs/vkd3d-shader/spirv.c b/libs/vkd3d-shader/spirv.c index ee9cca1bc..b4ff4a027 100644 --- a/libs/vkd3d-shader/spirv.c +++ b/libs/vkd3d-shader/spirv.c @@ -7421,7 +7421,8 @@ static void spirv_compiler_emit_comparison_instruction(struct spirv_compiler *co result_id = vkd3d_spirv_build_op_tr2(builder, &builder->function_stream, op, type_id, src0_id, src1_id); - result_id = spirv_compiler_emit_bool_to_int(compiler, component_count, result_id, true); + if (dst->reg.data_type != VKD3D_DATA_BOOL) + result_id = spirv_compiler_emit_bool_to_int(compiler, component_count, result_id, true); spirv_compiler_emit_store_reg(compiler, &dst->reg, dst->write_mask, result_id); } -- GitLab https://gitlab.winehq.org/wine/vkd3d/-/merge_requests/441
From: Conor McCarthy <cmccarthy(a)codeweavers.com> --- libs/vkd3d-shader/spirv.c | 9 +++++---- libs/vkd3d-shader/vkd3d_shader_private.h | 1 + 2 files changed, 6 insertions(+), 4 deletions(-) diff --git a/libs/vkd3d-shader/spirv.c b/libs/vkd3d-shader/spirv.c index b4ff4a027..f9b3fb688 100644 --- a/libs/vkd3d-shader/spirv.c +++ b/libs/vkd3d-shader/spirv.c @@ -7384,6 +7384,7 @@ static void spirv_compiler_emit_f32tof16(struct spirv_compiler *compiler, static void spirv_compiler_emit_comparison_instruction(struct spirv_compiler *compiler, const struct vkd3d_shader_instruction *instruction) { + bool inv_is_ord = instruction->flags & VKD3DSI_FCMP_INVERT_IS_ORDERED; struct vkd3d_spirv_builder *builder = &compiler->spirv_builder; const struct vkd3d_shader_dst_param *dst = instruction->dst; const struct vkd3d_shader_src_param *src = instruction->src; @@ -7394,17 +7395,17 @@ static void spirv_compiler_emit_comparison_instruction(struct spirv_compiler *co switch (instruction->handler_idx) { case VKD3DSIH_DEQ: - case VKD3DSIH_EQ: op = SpvOpFOrdEqual; break; + case VKD3DSIH_EQ: op = inv_is_ord ? SpvOpFUnordEqual : SpvOpFOrdEqual; break; case VKD3DSIH_DGE: - case VKD3DSIH_GE: op = SpvOpFOrdGreaterThanEqual; break; + case VKD3DSIH_GE: op = inv_is_ord ? SpvOpFUnordGreaterThanEqual : SpvOpFOrdGreaterThanEqual; break; case VKD3DSIH_IEQ: op = SpvOpIEqual; break; case VKD3DSIH_IGE: op = SpvOpSGreaterThanEqual; break; case VKD3DSIH_ILT: op = SpvOpSLessThan; break; case VKD3DSIH_INE: op = SpvOpINotEqual; break; case VKD3DSIH_DLT: - case VKD3DSIH_LT: op = SpvOpFOrdLessThan; break; + case VKD3DSIH_LT: op = inv_is_ord ? SpvOpFUnordLessThan : SpvOpFOrdLessThan; break; case VKD3DSIH_DNE: - case VKD3DSIH_NE: op = SpvOpFUnordNotEqual; break; + case VKD3DSIH_NE: op = inv_is_ord ? SpvOpFOrdNotEqual : SpvOpFUnordNotEqual; break; case VKD3DSIH_UGE: op = SpvOpUGreaterThanEqual; break; case VKD3DSIH_ULT: op = SpvOpULessThan; break; default: diff --git a/libs/vkd3d-shader/vkd3d_shader_private.h b/libs/vkd3d-shader/vkd3d_shader_private.h index c10e77f53..618e15bc8 100644 --- a/libs/vkd3d-shader/vkd3d_shader_private.h +++ b/libs/vkd3d-shader/vkd3d_shader_private.h @@ -731,6 +731,7 @@ enum vkd3d_tessellator_domain #define VKD3DSI_RESINFO_UINT 0x2 #define VKD3DSI_SAMPLE_INFO_UINT 0x1 #define VKD3DSI_SAMPLER_COMPARISON_MODE 0x1 +#define VKD3DSI_FCMP_INVERT_IS_ORDERED 0x1 #define VKD3DSI_PRECISE_X 0x100 #define VKD3DSI_PRECISE_Y 0x200 -- GitLab https://gitlab.winehq.org/wine/vkd3d/-/merge_requests/441
From: Conor McCarthy <cmccarthy(a)codeweavers.com> --- libs/vkd3d-shader/dxil.c | 155 +++++++++++++++++++++++ libs/vkd3d-shader/vkd3d_shader_private.h | 2 + tests/hlsl/all.shader_test | 14 +- tests/hlsl/any.shader_test | 36 +++--- tests/hlsl/bool-cast.shader_test | 4 +- tests/hlsl/cast-to-float.shader_test | 2 +- tests/hlsl/cast-to-half.shader_test | 2 +- tests/hlsl/cast-to-int.shader_test | 2 +- tests/hlsl/cast-to-uint.shader_test | 2 +- tests/hlsl/sign.shader_test | 20 +-- 10 files changed, 198 insertions(+), 41 deletions(-) diff --git a/libs/vkd3d-shader/dxil.c b/libs/vkd3d-shader/dxil.c index 5f7232325..f00449c6c 100644 --- a/libs/vkd3d-shader/dxil.c +++ b/libs/vkd3d-shader/dxil.c @@ -306,6 +306,36 @@ enum dxil_cast_code CAST_ADDRSPACECAST = 12, }; +enum dxil_predicate +{ + FCMP_FALSE = 0, + FCMP_OEQ = 1, + FCMP_OGT = 2, + FCMP_OGE = 3, + FCMP_OLT = 4, + FCMP_OLE = 5, + FCMP_ONE = 6, + FCMP_ORD = 7, + FCMP_UNO = 8, + FCMP_UEQ = 9, + FCMP_UGT = 10, + FCMP_UGE = 11, + FCMP_ULT = 12, + FCMP_ULE = 13, + FCMP_UNE = 14, + FCMP_TRUE = 15, + ICMP_EQ = 32, + ICMP_NE = 33, + ICMP_UGT = 34, + ICMP_UGE = 35, + ICMP_ULT = 36, + ICMP_ULE = 37, + ICMP_SGT = 38, + ICMP_SGE = 39, + ICMP_SLT = 40, + ICMP_SLE = 41, +}; + struct sm6_pointer_info { const struct sm6_type *type; @@ -514,6 +544,7 @@ struct sm6_parser struct sm6_type *types; size_t type_count; + struct sm6_type *bool_type; struct sm6_type *metadata_type; struct sm6_type *handle_type; @@ -1388,6 +1419,8 @@ static enum vkd3d_result sm6_parser_type_table_init(struct sm6_parser *sm6) switch ((width = record->operands[0])) { case 1: + sm6->bool_type = type; + break; case 8: case 16: case 32: @@ -3412,6 +3445,125 @@ static void sm6_parser_emit_cast(struct sm6_parser *sm6, const struct dxil_recor src_param->reg.data_type = dst->u.reg.data_type; } +struct sm6_cmp_info +{ + enum vkd3d_shader_opcode handler_idx_32; + enum vkd3d_shader_opcode handler_idx_64; + bool src_swap; + unsigned int flags; +}; + +static const struct sm6_cmp_info sm6_cmp_op_table[] = +{ + [FCMP_FALSE] = {VKD3DSIH_INVALID, VKD3DSIH_INVALID}, + [FCMP_OEQ] = {VKD3DSIH_EQ, VKD3DSIH_DEQ}, + [FCMP_OGT] = {VKD3DSIH_LT, VKD3DSIH_DLT, true}, + [FCMP_OGE] = {VKD3DSIH_GE, VKD3DSIH_DGE}, + [FCMP_OLT] = {VKD3DSIH_LT, VKD3DSIH_DLT}, + [FCMP_OLE] = {VKD3DSIH_GE, VKD3DSIH_DGE, true}, + [FCMP_ONE] = {VKD3DSIH_NE, VKD3DSIH_DNE, false, VKD3DSI_FCMP_INVERT_IS_ORDERED}, + [FCMP_ORD] = {VKD3DSIH_INVALID, VKD3DSIH_INVALID}, + [FCMP_UNO] = {VKD3DSIH_INVALID, VKD3DSIH_INVALID}, + [FCMP_UEQ] = {VKD3DSIH_EQ, VKD3DSIH_DEQ, false, VKD3DSI_FCMP_INVERT_IS_ORDERED}, + [FCMP_UGT] = {VKD3DSIH_LT, VKD3DSIH_DLT, true, VKD3DSI_FCMP_INVERT_IS_ORDERED}, + [FCMP_UGE] = {VKD3DSIH_GE, VKD3DSIH_DGE, false, VKD3DSI_FCMP_INVERT_IS_ORDERED}, + [FCMP_ULT] = {VKD3DSIH_LT, VKD3DSIH_DLT, false, VKD3DSI_FCMP_INVERT_IS_ORDERED}, + [FCMP_ULE] = {VKD3DSIH_GE, VKD3DSIH_DGE, true, VKD3DSI_FCMP_INVERT_IS_ORDERED}, + [FCMP_UNE] = {VKD3DSIH_NE, VKD3DSIH_DNE}, + [FCMP_TRUE] = {VKD3DSIH_INVALID, VKD3DSIH_INVALID}, + + [ICMP_EQ] = {VKD3DSIH_IEQ, VKD3DSIH_INVALID}, + [ICMP_NE] = {VKD3DSIH_INE, VKD3DSIH_INVALID}, + [ICMP_UGT] = {VKD3DSIH_ULT, VKD3DSIH_INVALID, true}, + [ICMP_UGE] = {VKD3DSIH_UGE, VKD3DSIH_INVALID}, + [ICMP_ULT] = {VKD3DSIH_ULT, VKD3DSIH_INVALID}, + [ICMP_ULE] = {VKD3DSIH_UGE, VKD3DSIH_INVALID, true}, + [ICMP_SGT] = {VKD3DSIH_ILT, VKD3DSIH_INVALID, true}, + [ICMP_SGE] = {VKD3DSIH_IGE, VKD3DSIH_INVALID}, + [ICMP_SLT] = {VKD3DSIH_ILT, VKD3DSIH_INVALID}, + [ICMP_SLE] = {VKD3DSIH_IGE, VKD3DSIH_INVALID, true}, +}; + +static void sm6_parser_emit_cmp2(struct sm6_parser *sm6, const struct dxil_record *record, + struct vkd3d_shader_instruction *ins, struct sm6_value *dst) +{ + struct vkd3d_shader_src_param *src_params; + const struct sm6_type *type_a, *type_b; + enum vkd3d_shader_opcode handler_idx; + const struct sm6_cmp_info *cmp; + const struct sm6_value *a, *b; + unsigned int idx = 0; + uint64_t op; + bool is_int; + + if (!(dst->type = sm6->bool_type)) + { + FIXME("Bool type not found.\n"); + return; + } + + a = sm6_parser_get_value_by_ref(sm6, record, NULL, &idx); + b = sm6_parser_get_value_by_ref(sm6, record, a->type, &idx); + if (!a || !b) + return; + + if (!dxil_record_validate_operand_count(record, idx + 1, idx + 2, sm6)) + return; + + type_a = a->type; + type_b = b->type; + is_int = sm6_type_is_1wil(type_a); + + if (!is_int && !sm6_type_is_floating_point(type_a)) + { + FIXME("Argument type %u is not bool, int16/32/64 or floating point.\n", type_a->class); + return; + } + + op = record->operands[idx++]; + + if (is_int != (op >= ICMP_EQ)) + WARN("Invalid type %u for operation %"PRIu64".\n", type_a->class, op); + if (type_a != type_b) + WARN("Type mismatch.\n"); + + if (record->operand_count > idx) + { + if (sm6_type_is_floating_point(a->type)) + TRACE("Ignoring fast FP modifier %#"PRIx64".\n", record->operands[idx]); + else + FIXME("Unexpected extra value %#"PRIx64".\n", record->operands[idx]); + } + + handler_idx = VKD3DSIH_INVALID; + if (op < ARRAY_SIZE(sm6_cmp_op_table)) + { + cmp = &sm6_cmp_op_table[op]; + handler_idx = (!is_int && type_a->u.width == 64) ? cmp->handler_idx_64 : cmp->handler_idx_32; + } + if (!handler_idx || handler_idx == VKD3DSIH_INVALID) + { + FIXME("Unhandled comparison op %"PRIu64".\n", op); + return; + } + vsir_instruction_init(ins, &sm6->p.location, handler_idx); + ins->flags = cmp->flags; + + src_params = instruction_src_params_alloc(ins, 2, sm6); + if (cmp->src_swap) + { + src_param_init_from_value(&src_params[0], b); + src_param_init_from_value(&src_params[1], a); + } + else + { + src_param_init_from_value(&src_params[0], a); + src_param_init_from_value(&src_params[1], b); + } + + instruction_dst_param_init_ssa_scalar(ins, sm6); +} + static void sm6_parser_emit_extractval(struct sm6_parser *sm6, const struct dxil_record *record, struct vkd3d_shader_instruction *ins, struct sm6_value *dst) { @@ -3631,6 +3783,9 @@ static enum vkd3d_result sm6_parser_function_init(struct sm6_parser *sm6, const case FUNC_CODE_INST_CAST: sm6_parser_emit_cast(sm6, record, ins, dst); break; + case FUNC_CODE_INST_CMP2: + sm6_parser_emit_cmp2(sm6, record, ins, dst); + break; case FUNC_CODE_INST_EXTRACTVAL: sm6_parser_emit_extractval(sm6, record, ins, dst); break; diff --git a/libs/vkd3d-shader/vkd3d_shader_private.h b/libs/vkd3d-shader/vkd3d_shader_private.h index 618e15bc8..73b1278dc 100644 --- a/libs/vkd3d-shader/vkd3d_shader_private.h +++ b/libs/vkd3d-shader/vkd3d_shader_private.h @@ -1416,6 +1416,8 @@ static inline enum vkd3d_shader_component_type vkd3d_component_type_from_data_ty return VKD3D_SHADER_COMPONENT_INT; case VKD3D_DATA_DOUBLE: return VKD3D_SHADER_COMPONENT_DOUBLE; + case VKD3D_DATA_BOOL: + return VKD3D_SHADER_COMPONENT_BOOL; default: FIXME("Unhandled data type %#x.\n", data_type); /* fall-through */ diff --git a/tests/hlsl/all.shader_test b/tests/hlsl/all.shader_test index 564cc5b00..7bdb0dc82 100644 --- a/tests/hlsl/all.shader_test +++ b/tests/hlsl/all.shader_test @@ -11,17 +11,17 @@ float4 main() : sv_target [test] uniform 0 float4 -1.1 1.6 1.3 0.5 -todo(sm>=6) draw quad +draw quad probe all rgba (1.0, 1.0, 1.0, 1.0) [test] uniform 0 float4 0.0 1.6 1.3 0.5 -todo(sm>=6) draw quad +draw quad probe all rgba (0.0, 0.0, 0.0, 0.0) [test] uniform 0 float4 1.0 0.0 1.3 0.5 -todo(sm>=6) draw quad +draw quad probe all rgba (0.0, 0.0, 0.0, 0.0) [pixel shader] @@ -34,12 +34,12 @@ float4 main() : sv_target [test] uniform 0 float4 1.0 0.0 0.0 0.0 -todo(sm>=6) draw quad +draw quad probe all rgba (1.0, 1.0, 1.0, 1.0) [test] uniform 0 float4 0.0 0.0 0.0 0.0 -todo(sm>=6) draw quad +draw quad probe all rgba (0.0, 0.0, 0.0, 0.0) [pixel shader] @@ -53,11 +53,11 @@ float4 main() : sv_target [test] uniform 0 float4 1.0 2.0 0.0 0.0 uniform 4 float4 3.0 4.0 0.0 0.0 -todo(sm>=6) draw quad +draw quad probe all rgba (1.0, 1.0, 1.0, 1.0) [test] uniform 0 float4 1.0 2.0 0.0 0.0 uniform 4 float4 0.0 4.0 0.0 0.0 -todo(sm>=6) draw quad +draw quad probe all rgba (0.0, 0.0, 0.0, 0.0) diff --git a/tests/hlsl/any.shader_test b/tests/hlsl/any.shader_test index 9b8b922c9..f2298d3a3 100644 --- a/tests/hlsl/any.shader_test +++ b/tests/hlsl/any.shader_test @@ -8,25 +8,25 @@ float4 main() : sv_target [test] uniform 0 float4 1.0 1.0 1.0 1.0 -todo(sm>=6) draw quad +draw quad probe all rgba (1.0, 1.0, 1.0, 1.0) uniform 0 float4 1.0 0.0 0.0 0.0 -todo(sm>=6) draw quad +draw quad probe all rgba (1.0, 1.0, 1.0, 1.0) uniform 0 float4 0.0 1.0 0.0 0.0 -todo(sm>=6) draw quad +draw quad probe all rgba (1.0, 1.0, 1.0, 1.0) uniform 0 float4 0.0 0.0 1.0 0.0 -todo(sm>=6) draw quad +draw quad probe all rgba (1.0, 1.0, 1.0, 1.0) uniform 0 float4 0.0 0.0 0.0 1.0 -todo(sm>=6) draw quad +draw quad probe all rgba (1.0, 1.0, 1.0, 1.0) uniform 0 float4 0.0 0.0 0.0 0.0 -todo(sm>=6) draw quad +draw quad probe all rgba (0.0, 0.0, 0.0, 0.0) uniform 0 float4 -1.0 -1.0 -1.0 -1.0 -todo(sm>=6) draw quad +draw quad probe all rgba (1.0, 1.0, 1.0, 1.0) [pixel shader] @@ -39,13 +39,13 @@ float4 main() : sv_target [test] uniform 0 float4 1.0 0.0 0.0 0.0 -todo(sm>=6) draw quad +draw quad probe all rgba (1.0, 1.0, 1.0, 1.0) uniform 0 float4 0.0 0.0 0.0 0.0 -todo(sm>=6) draw quad +draw quad probe all rgba (0.0, 0.0, 0.0, 0.0) uniform 0 float4 -1.0 0.0 0.0 0.0 -todo(sm>=6) draw quad +draw quad probe all rgba (1.0, 1.0, 1.0, 1.0) [require] @@ -61,22 +61,22 @@ float4 main() : sv_target [test] uniform 0 uint4 1 1 1 1 -todo(sm>=6) draw quad +draw quad probe all rgba (1.0, 1.0, 1.0, 1.0) uniform 0 uint4 1 0 0 0 -todo(sm>=6) draw quad +draw quad probe all rgba (1.0, 1.0, 1.0, 1.0) uniform 0 uint4 0 1 0 0 -todo(sm>=6) draw quad +draw quad probe all rgba (1.0, 1.0, 1.0, 1.0) uniform 0 uint4 0 0 1 0 -todo(sm>=6) draw quad +draw quad probe all rgba (1.0, 1.0, 1.0, 1.0) uniform 0 uint4 0 0 0 1 -todo(sm>=6) draw quad +draw quad probe all rgba (1.0, 1.0, 1.0, 1.0) uniform 0 uint4 0 0 0 0 -todo(sm>=6) draw quad +draw quad probe all rgba (0.0, 0.0, 0.0, 0.0) [pixel shader] @@ -89,8 +89,8 @@ float4 main() : sv_target [test] uniform 0 uint4 1 0 0 0 -todo(sm>=6) draw quad +draw quad probe all rgba (1.0, 1.0, 1.0, 1.0) uniform 0 uint4 0 0 0 0 -todo(sm>=6) draw quad +draw quad probe all rgba (0.0, 0.0, 0.0, 0.0) diff --git a/tests/hlsl/bool-cast.shader_test b/tests/hlsl/bool-cast.shader_test index dc75ea376..09ca12e2b 100644 --- a/tests/hlsl/bool-cast.shader_test +++ b/tests/hlsl/bool-cast.shader_test @@ -30,7 +30,7 @@ float4 main() : SV_TARGET [test] uniform 0 float4 0.0 0.0 2.0 4.0 uniform 4 int4 0 1 0 10 -todo(sm>=6) draw quad +draw quad probe all rgba (0.0, 10.0, 1.0, 11.0) @@ -44,5 +44,5 @@ float4 main() : sv_target [test] uniform 0 uint4 0x00000001 0x00000002 0x80000000 0x00000000 -todo(sm>=6) draw quad +draw quad probe all rgba (2.0, 2.0, 2.0, 0.0) diff --git a/tests/hlsl/cast-to-float.shader_test b/tests/hlsl/cast-to-float.shader_test index 7c32acfa3..caaf98c02 100644 --- a/tests/hlsl/cast-to-float.shader_test +++ b/tests/hlsl/cast-to-float.shader_test @@ -17,7 +17,7 @@ uniform 0 int -1 uniform 1 uint 3 uniform 2 int -2 uniform 3 float 0.5 -todo(sm>=6) draw quad +draw quad probe all rgba (0.5, 0.5, 0.5, 0.5) [pixel shader] diff --git a/tests/hlsl/cast-to-half.shader_test b/tests/hlsl/cast-to-half.shader_test index 1579c63f3..b8feb6760 100644 --- a/tests/hlsl/cast-to-half.shader_test +++ b/tests/hlsl/cast-to-half.shader_test @@ -17,7 +17,7 @@ uniform 0 int -1 uniform 1 uint 3 uniform 2 int -2 uniform 3 float 0.5 -todo(sm>=6) draw quad +draw quad probe all rgba (0.5, 0.5, 0.5, 0.5) [pixel shader] diff --git a/tests/hlsl/cast-to-int.shader_test b/tests/hlsl/cast-to-int.shader_test index ae566a039..3e850fb5b 100644 --- a/tests/hlsl/cast-to-int.shader_test +++ b/tests/hlsl/cast-to-int.shader_test @@ -23,7 +23,7 @@ uniform 0 float 2.6 uniform 1 int -2 uniform 2 int -2 uniform 3 float -3.6 -todo(sm>=6) draw quad +draw quad probe all rgba (0.5, 0.5, 0.5, 0.5) [pixel shader] diff --git a/tests/hlsl/cast-to-uint.shader_test b/tests/hlsl/cast-to-uint.shader_test index dece99b75..07479984a 100644 --- a/tests/hlsl/cast-to-uint.shader_test +++ b/tests/hlsl/cast-to-uint.shader_test @@ -23,7 +23,7 @@ uniform 0 float 2.6 uniform 1 int 2 uniform 2 int -2 uniform 3 float -3.6 -todo(sm>=6) draw quad +draw quad probe all rgba (0.5, 0.5, 0.5, 0.5) [pixel shader] diff --git a/tests/hlsl/sign.shader_test b/tests/hlsl/sign.shader_test index 54b953681..5d8b43168 100644 --- a/tests/hlsl/sign.shader_test +++ b/tests/hlsl/sign.shader_test @@ -8,13 +8,13 @@ float4 main() : sv_target [test] uniform 0 float4 1.0 0.0 0.0 0.0 -todo(sm>=6) draw quad +draw quad probe all rgba (1.0, 1.0, 1.0, 1.0) uniform 0 float4 -1.0 0.0 0.0 0.0 -todo(sm>=6) draw quad +draw quad probe all rgba (-1.0, -1.0, -1.0, -1.0) uniform 0 float4 0.0 0.0 0.0 0.0 -todo(sm>=6) draw quad +draw quad probe all rgba (0.0, 0.0, 0.0, 0.0) [pixel shader] @@ -27,7 +27,7 @@ float4 main() : sv_target [test] uniform 0 float4 1.0 2.0 3.0 4.0 -todo(sm>=6) draw quad +draw quad probe all rgba (1.0, 1.0, 1.0, 1.0) [pixel shader] @@ -41,7 +41,7 @@ float4 main() : sv_target [test] uniform 0 float4 1.0 2.0 0.0 0.0 uniform 4 float4 3.0 4.0 0.0 0.0 -todo(sm>=6) draw quad +draw quad probe all rgba (1.0, 1.0, 1.0, 1.0) [require] @@ -58,13 +58,13 @@ float4 main() : sv_target [test] uniform 0 int4 1 0 0 0 -todo(sm>=6) draw quad +draw quad probe all rgba (1, 1, 1, 1) uniform 0 int4 -1 0 0 0 -todo(sm>=6) draw quad +draw quad probe all rgba (-1, -1, -1, -1) uniform 0 int4 0 0 0 0 -todo(sm>=6) draw quad +draw quad probe all rgba (0, 0, 0, 0) [pixel shader] @@ -77,7 +77,7 @@ float4 main() : sv_target [test] uniform 0 int4 1 2 3 4 -todo(sm>=6) draw quad +draw quad probe all rgba (1, 1, 1, 1) [pixel shader] @@ -91,5 +91,5 @@ float4 main() : sv_target [test] uniform 0 int4 1 2 0 0 uniform 4 int4 3 4 0 0 -todo(sm>=6) draw quad +draw quad probe all rgba (1, 1, 1, 1) -- GitLab https://gitlab.winehq.org/wine/vkd3d/-/merge_requests/441
participants (2)
-
Conor McCarthy -
Conor McCarthy (＠cmccarthy)