From: Francisco Casas fcasas@codeweavers.com
According to the documentation, if_comp is available from 2_x pixel and vertex shaders and, unlike "if bool" it doesn't expect a constant boolean register (from the input signature), so:
if_neq cond -cond
seems like a convenient way to write these, for profiles above 2.0. --- libs/vkd3d-shader/d3dbc.c | 54 ++++++++++++++++++++++++++++++ tests/hlsl/conditional.shader_test | 32 +++++++++--------- 2 files changed, 70 insertions(+), 16 deletions(-)
diff --git a/libs/vkd3d-shader/d3dbc.c b/libs/vkd3d-shader/d3dbc.c index fe4f9ec27..93ddb0d5e 100644 --- a/libs/vkd3d-shader/d3dbc.c +++ b/libs/vkd3d-shader/d3dbc.c @@ -1793,6 +1793,7 @@ static uint32_t sm1_encode_register_type(D3DSHADER_PARAM_REGISTER_TYPE type) struct sm1_instruction { D3DSHADER_INSTRUCTION_OPCODE_TYPE opcode; + unsigned int flags;
struct sm1_dst_register { @@ -1832,6 +1833,8 @@ static void write_sm1_instruction(struct hlsl_ctx *ctx, struct vkd3d_bytecode_bu uint32_t token = instr->opcode; unsigned int i;
+ token |= VKD3D_SM1_INSTRUCTION_FLAGS_MASK & (instr->flags << VKD3D_SM1_INSTRUCTION_FLAGS_SHIFT); + if (ctx->profile->major_version > 1) token |= (instr->has_dst + instr->src_count) << D3DSI_INSTLENGTH_SHIFT; put_u32(buffer, token); @@ -2394,6 +2397,49 @@ static void write_sm1_expr(struct hlsl_ctx *ctx, struct vkd3d_bytecode_buffer *b } }
+static void write_sm1_block(struct hlsl_ctx *ctx, struct vkd3d_bytecode_buffer *buffer, + const struct hlsl_block *block); + +static void write_sm1_if(struct hlsl_ctx *ctx, struct vkd3d_bytecode_buffer *buffer, const struct hlsl_ir_node *instr) +{ + const struct hlsl_ir_if *iff = hlsl_ir_if(instr); + const struct hlsl_ir_node *condition; + struct sm1_instruction sm1_ifc, sm1_else, sm1_endif; + + condition = iff->condition.node; + assert(condition->data_type->dimx == 1 && condition->data_type->dimy == 1); + + sm1_ifc = (struct sm1_instruction) + { + .opcode = D3DSIO_IFC, + .flags = VKD3D_SHADER_REL_OP_NE, /* Make it a "if_ne" instruction. */ + + .srcs[0].type = D3DSPR_TEMP, + .srcs[0].swizzle = hlsl_swizzle_from_writemask(condition->reg.writemask), + .srcs[0].reg = condition->reg.id, + .srcs[0].mod = 0, + + .srcs[1].type = D3DSPR_TEMP, + .srcs[1].swizzle = hlsl_swizzle_from_writemask(condition->reg.writemask), + .srcs[1].reg = condition->reg.id, + .srcs[1].mod = D3DSPSM_NEG, + + .src_count = 2, + }; + write_sm1_instruction(ctx, buffer, &sm1_ifc); + write_sm1_block(ctx, buffer, &iff->then_block); + + if (!list_empty(&iff->else_block.instrs)) + { + sm1_else = (struct sm1_instruction){.opcode = D3DSIO_ELSE}; + write_sm1_instruction(ctx, buffer, &sm1_else); + write_sm1_block(ctx, buffer, &iff->else_block); + } + + sm1_endif = (struct sm1_instruction){.opcode = D3DSIO_ENDIF}; + write_sm1_instruction(ctx, buffer, &sm1_endif); +} + static void write_sm1_jump(struct hlsl_ctx *ctx, struct vkd3d_bytecode_buffer *buffer, const struct hlsl_ir_node *instr) { const struct hlsl_ir_jump *jump = hlsl_ir_jump(instr); @@ -2623,6 +2669,14 @@ static void write_sm1_block(struct hlsl_ctx *ctx, struct vkd3d_bytecode_buffer * write_sm1_expr(ctx, buffer, instr); break;
+ case HLSL_IR_IF: + if (ctx->profile->major_version > 2 + || (ctx->profile->major_version == 2 && ctx->profile->minor_version >= 1)) + write_sm1_if(ctx, buffer, instr); + else + hlsl_fixme(ctx, &instr->loc, "Flatten "if" conditionals branches."); + break; + case HLSL_IR_JUMP: write_sm1_jump(ctx, buffer, instr); break; diff --git a/tests/hlsl/conditional.shader_test b/tests/hlsl/conditional.shader_test index a73b42898..2b9ca66bd 100644 --- a/tests/hlsl/conditional.shader_test +++ b/tests/hlsl/conditional.shader_test @@ -75,7 +75,7 @@ float main() : sv_target shader model >= 3.0
-[pixel shader todo(sm<4)] +[pixel shader] uniform float a;
float4 main() : sv_target @@ -87,14 +87,14 @@ float4 main() : sv_target
[test] uniform 0 float -2 -todo(sm<4 | glsl) draw quad +todo(glsl) draw quad probe all rgba (1, 2, 3, 4) uniform 0 float 10 -todo(sm<4 | glsl) draw quad +todo(glsl) draw quad probe all rgba (10, 20, 30, 40)
-[pixel shader todo(sm<4)] +[pixel shader] uniform float4 u;
float4 main() : sv_target @@ -108,7 +108,7 @@ float4 main() : sv_target
[test] uniform 0 float4 0.0 0.0 0.0 0.0 -todo(sm<4 | glsl) draw quad +todo(glsl) draw quad probe all rgba (0.9, 0.8, 0.7, 0.6)
[pixel shader] @@ -129,7 +129,7 @@ todo(glsl) draw quad probe all rgba (9.0, 10.0, 11.0, 12.0)
-[pixel shader todo(sm<4)] +[pixel shader] int a, b;
float4 main() : sv_target @@ -145,23 +145,23 @@ if(sm<4) uniform 0 float 8 if(sm<4) uniform 4 float 9 if(sm>=4) uniform 0 int 8 if(sm>=4) uniform 1 int 9 -todo(sm<4 | glsl) draw quad +todo(glsl) draw quad probe all rgba (-1.0, -1.0, -1.0, -1.0) if(sm<4) uniform 0 float -3 if(sm<4) uniform 4 float -4 if(sm>=4) uniform 0 int -3 if(sm>=4) uniform 1 int -4 -todo(sm<4 | glsl) draw quad +todo(glsl) draw quad probe all rgba (1.0, 1.0, 1.0, 1.0) if(sm<4) uniform 0 float 7 if(sm<4) uniform 4 float 7 if(sm>=4) uniform 0 int 7 if(sm>=4) uniform 1 int 7 -todo(sm<4 | glsl) draw quad +todo(glsl) draw quad probe all rgba (1.0, 1.0, 1.0, 1.0)
-[pixel shader todo(sm<4)] +[pixel shader] int a, b;
float4 main() : sv_target @@ -177,23 +177,23 @@ if(sm<4) uniform 0 float 8 if(sm<4) uniform 4 float 9 if(sm>=4) uniform 0 int 8 if(sm>=4) uniform 1 int 9 -todo(sm<4 | glsl) draw quad +todo(glsl) draw quad probe all rgba (-1.0, -1.0, -1.0, -1.0) if(sm<4) uniform 0 float -3 if(sm<4) uniform 4 float -4 if(sm>=4) uniform 0 int -3 if(sm>=4) uniform 1 int -4 -todo(sm<4 | glsl) draw quad +todo(glsl) draw quad probe all rgba (1.0, 1.0, 1.0, 1.0) if(sm<4) uniform 0 float 7 if(sm<4) uniform 4 float 7 if(sm>=4) uniform 0 int 7 if(sm>=4) uniform 1 int 7 -todo(sm<4 | glsl) draw quad +todo(glsl) draw quad probe all rgba (-1.0, -1.0, -1.0, -1.0)
-[pixel shader todo(sm<4)] +[pixel shader] int a, b;
float4 main() : sv_target @@ -209,13 +209,13 @@ if(sm<4) uniform 0 float -3 if(sm<4) uniform 4 float -2 if(sm>=4) uniform 0 int -3 if(sm>=4) uniform 1 int -2 -todo(sm<4 | glsl) draw quad +todo(glsl) draw quad probe all rgba (-1.0, -1.0, -1.0, -1.0) if(sm<4) uniform 0 float 4 if(sm<4) uniform 4 float 4 if(sm>=4) uniform 0 int 4 if(sm>=4) uniform 1 int 4 -todo(sm<4 | glsl) draw quad +todo(glsl) draw quad probe all rgba (1.0, 1.0, 1.0, 1.0)