From 1509d2ffbdf98ead108d672795b2e9a7a93f4ce6 Mon Sep 17 00:00:00 2001 From: Fernando Sahmkow Date: Tue, 22 Oct 2019 10:59:07 -0400 Subject: [PATCH] Shader_Ir: Fix TLD4S from using a component mask. TLD4S always outputs 4 values, the previous code checked a component mask and omitted those values that weren't part of it. This commit corrects that and makes sure all 4 values are set. --- src/video_core/shader/decode/texture.cpp | 8 ++++---- src/video_core/shader/shader_ir.h | 2 +- 2 files changed, 5 insertions(+), 5 deletions(-) diff --git a/src/video_core/shader/decode/texture.cpp b/src/video_core/shader/decode/texture.cpp index 0b934a0696..2954454988 100644 --- a/src/video_core/shader/decode/texture.cpp +++ b/src/video_core/shader/decode/texture.cpp @@ -150,7 +150,7 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) { values[element] = Operation(OperationCode::TextureGather, meta, std::move(coords_copy)); } - WriteTexsInstructionFloat(bb, instr, values); + WriteTexsInstructionFloat(bb, instr, values, true); break; } case OpCode::Id::TXQ_B: @@ -344,14 +344,14 @@ void ShaderIR::WriteTexInstructionFloat(NodeBlock& bb, Instruction instr, const } } -void ShaderIR::WriteTexsInstructionFloat(NodeBlock& bb, Instruction instr, - const Node4& components) { +void ShaderIR::WriteTexsInstructionFloat(NodeBlock& bb, Instruction instr, const Node4& components, + bool ignore_mask) { // TEXS has two destination registers and a swizzle. The first two elements in the swizzle // go into gpr0+0 and gpr0+1, and the rest goes into gpr28+0 and gpr28+1 u32 dest_elem = 0; for (u32 component = 0; component < 4; ++component) { - if (!instr.texs.IsComponentEnabled(component)) + if (!instr.texs.IsComponentEnabled(component) && !ignore_mask) continue; SetTemporary(bb, dest_elem++, components[component]); } diff --git a/src/video_core/shader/shader_ir.h b/src/video_core/shader/shader_ir.h index 105981d67a..df64845f50 100644 --- a/src/video_core/shader/shader_ir.h +++ b/src/video_core/shader/shader_ir.h @@ -322,7 +322,7 @@ private: const Node4& components); void WriteTexsInstructionFloat(NodeBlock& bb, Tegra::Shader::Instruction instr, - const Node4& components); + const Node4& components, bool ignore_mask = false); void WriteTexsInstructionHalfFloat(NodeBlock& bb, Tegra::Shader::Instruction instr, const Node4& components);