6 files changed, 17 insertions, 11 deletions
diff --git a/src/video_core/shader/async_shaders.cpp b/src/video_core/shader/async_shaders.cpp
index 3b40db9bc..02adcf9c7 100644
--- a/src/video_core/shader/async_shaders.cpp
+++ b/src/video_core/shader/async_shaders.cpp
@@ -64,6 +64,7 @@ void AsyncShaders::FreeWorkers() {
 
 void AsyncShaders::KillWorkers() {
     is_thread_exiting.store(true);
+    cv.notify_all();
     for (auto& thread : worker_threads) {
         thread.detach();
     }
diff --git a/src/video_core/shader/async_shaders.h b/src/video_core/shader/async_shaders.h
index 0dbb1a31f..7fdff6e56 100644
--- a/src/video_core/shader/async_shaders.h
+++ b/src/video_core/shader/async_shaders.h
@@ -9,16 +9,7 @@
 #include <shared_mutex>
 #include <thread>
 
-// This header includes both Vulkan and OpenGL headers, this has to be fixed
-// Unfortunately, including OpenGL will include Windows.h that defines macros that can cause issues.
-// Forcefully include glad early and undefine macros
 #include <glad/glad.h>
-#ifdef CreateEvent
-#undef CreateEvent
-#endif
-#ifdef CreateSemaphore
-#undef CreateSemaphore
-#endif
 
 #include "common/common_types.h"
 #include "video_core/renderer_opengl/gl_device.h"
diff --git a/src/video_core/shader/decode/memory.cpp b/src/video_core/shader/decode/memory.cpp
index 50f4e7d35..7728f600e 100644
--- a/src/video_core/shader/decode/memory.cpp
+++ b/src/video_core/shader/decode/memory.cpp
@@ -330,6 +330,7 @@ u32 ShaderIR::DecodeMemory(NodeBlock& bb, u32 pc) {
         case StoreType::Bits32:
             (this->*set_memory)(bb, GetAddress(0), GetRegister(instr.gpr0));
             break;
+        case StoreType::Unsigned16:
         case StoreType::Signed16: {
             Node address = GetAddress(0);
             Node memory = (this->*get_memory)(address);
diff --git a/src/video_core/shader/decode/other.cpp b/src/video_core/shader/decode/other.cpp
index d3ea07aac..5f88537bc 100644
--- a/src/video_core/shader/decode/other.cpp
+++ b/src/video_core/shader/decode/other.cpp
@@ -76,6 +76,7 @@ u32 ShaderIR::DecodeOther(NodeBlock& bb, u32 pc) {
             case SystemVariable::InvocationId:
                 return Operation(OperationCode::InvocationId);
             case SystemVariable::Ydirection:
+                uses_y_negate = true;
                 return Operation(OperationCode::YNegate);
             case SystemVariable::InvocationInfo:
                 LOG_WARNING(HW_GPU, "S2R instruction with InvocationInfo is incomplete");
diff --git a/src/video_core/shader/decode/texture.cpp b/src/video_core/shader/decode/texture.cpp
index 833fa2a39..c69681e8d 100644
--- a/src/video_core/shader/decode/texture.cpp
+++ b/src/video_core/shader/decode/texture.cpp
@@ -806,6 +806,7 @@ Node4 ShaderIR::GetTldsCode(Instruction instr, TextureType texture_type, bool is
 
     const std::size_t type_coord_count = GetCoordCount(texture_type);
     const bool lod_enabled = instr.tlds.GetTextureProcessMode() == TextureProcessMode::LL;
+    const bool aoffi_enabled = instr.tlds.UsesMiscMode(TextureMiscMode::AOFFI);
 
     // If enabled arrays index is always stored in the gpr8 field
     const u64 array_register = instr.gpr8.Value();
@@ -820,17 +821,23 @@ Node4 ShaderIR::GetTldsCode(Instruction instr, TextureType texture_type, bool is
     std::vector<Node> coords;
     for (std::size_t i = 0; i < type_coord_count; ++i) {
         const bool last = (i == (type_coord_count - 1)) && (type_coord_count > 1);
-        coords.push_back(GetRegister(last ? last_coord_register : coord_register + i));
+        coords.push_back(
+            GetRegister(last && !aoffi_enabled ? last_coord_register : coord_register + i));
     }
 
     const Node array = is_array ? GetRegister(array_register) : nullptr;
     // When lod is used always is in gpr20
     const Node lod = lod_enabled ? GetRegister(instr.gpr20) : Immediate(0);
 
+    std::vector<Node> aoffi;
+    if (aoffi_enabled) {
+        aoffi = GetAoffiCoordinates(GetRegister(instr.gpr20), type_coord_count, false);
+    }
+
     Node4 values;
     for (u32 element = 0; element < values.size(); ++element) {
         auto coords_copy = coords;
-        MetaTexture meta{*sampler, array, {}, {}, {}, {}, {}, lod, {}, element, {}};
+        MetaTexture meta{*sampler, array, {}, aoffi, {}, {}, {}, lod, {}, element, {}};
         values[element] = Operation(OperationCode::TexelFetch, meta, std::move(coords_copy));
     }
     return values;
diff --git a/src/video_core/shader/shader_ir.h b/src/video_core/shader/shader_ir.h
index 0c6ab0f07..1cd7c14d7 100644
--- a/src/video_core/shader/shader_ir.h
+++ b/src/video_core/shader/shader_ir.h
@@ -139,6 +139,10 @@ public:
         return uses_legacy_varyings;
     }
 
+    bool UsesYNegate() const {
+        return uses_y_negate;
+    }
+
     bool UsesWarps() const {
         return uses_warps;
     }
@@ -465,6 +469,7 @@ private:
     bool uses_instance_id{};
     bool uses_vertex_id{};
     bool uses_legacy_varyings{};
+    bool uses_y_negate{};
     bool uses_warps{};
     bool uses_indexed_samplers{};