externals: Update dynarmic to 28714ee7

Reduces compilation times on MSVC.

CMakeLists.txt

@@ -229,7 +229,7 @@ elseif (${CMAKE_SYSTEM_NAME} STREQUAL "Linux" OR YUZU_USE_BUNDLED_BOOST)
     include_directories(SYSTEM "${Boost_INCLUDE_DIRS}")
 else()
     message(STATUS "Boost 1.73.0 or newer not found, falling back to Conan")
-    list(APPEND CONAN_REQUIRED_LIBS "boost/1.73.0")
+    list(APPEND CONAN_REQUIRED_LIBS "boost/1.78.0")
 endif()
 
 # Attempt to locate any packages that are required and report the missing ones in CONAN_REQUIRED_LIBS

src/core/hle/kernel/k_page_table.cpp

@@ -806,6 +806,33 @@ ResultCode KPageTable::ResetTransferMemory(VAddr addr, std::size_t size) {
                          KMemoryAttribute::Locked, KMemoryAttribute::IpcAndDeviceMapped));
 
     block_manager->Update(addr, size / PageSize, state, KMemoryPermission::ReadAndWrite);
+    return ResultSuccess;
+}
+
+ResultCode KPageTable::SetMemoryPermission(VAddr addr, std::size_t size,
+                                           Svc::MemoryPermission svc_perm) {
+    const size_t num_pages = size / PageSize;
+
+    // Lock the table.
+    std::lock_guard lock{page_table_lock};
+
+    // Verify we can change the memory permission.
+    KMemoryState old_state;
+    KMemoryPermission old_perm;
+    R_TRY(this->CheckMemoryState(
+        std::addressof(old_state), std::addressof(old_perm), nullptr, addr, size,
+        KMemoryState::FlagCanReprotect, KMemoryState::FlagCanReprotect, KMemoryPermission::None,
+        KMemoryPermission::None, KMemoryAttribute::All, KMemoryAttribute::None));
+
+    // Determine new perm.
+    const KMemoryPermission new_perm = ConvertToKMemoryPermission(svc_perm);
+    R_SUCCEED_IF(old_perm == new_perm);
+
+    // Perform mapping operation.
+    R_TRY(Operate(addr, num_pages, new_perm, OperationType::ChangePermissions));
+
+    // Update the blocks.
+    block_manager->Update(addr, num_pages, old_state, new_perm, KMemoryAttribute::None);
 
     return ResultSuccess;
 }

src/core/hle/kernel/k_page_table.h

@@ -47,6 +47,7 @@ public:
     KMemoryInfo QueryInfo(VAddr addr);
     ResultCode ReserveTransferMemory(VAddr addr, std::size_t size, KMemoryPermission perm);
     ResultCode ResetTransferMemory(VAddr addr, std::size_t size);
+    ResultCode SetMemoryPermission(VAddr addr, std::size_t size, Svc::MemoryPermission perm);
     ResultCode SetMemoryAttribute(VAddr addr, std::size_t size, KMemoryAttribute mask,
                                   KMemoryAttribute value);
     ResultCode SetHeapCapacity(std::size_t new_heap_capacity);

src/core/hle/kernel/k_thread.cpp

@@ -26,6 +26,7 @@
 #include "core/hle/kernel/k_resource_limit.h"
 #include "core/hle/kernel/k_scheduler.h"
 #include "core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h"
+#include "core/hle/kernel/k_system_control.h"
 #include "core/hle/kernel/k_thread.h"
 #include "core/hle/kernel/k_thread_queue.h"
 #include "core/hle/kernel/kernel.h"
@@ -50,6 +51,7 @@ static void ResetThreadContext64(Core::ARM_Interface::ThreadContext64& context,
                                  VAddr entry_point, u64 arg) {
     context = {};
     context.cpu_registers[0] = arg;
+    context.cpu_registers[18] = Kernel::KSystemControl::GenerateRandomU64() | 1;
     context.pc = entry_point;
     context.sp = stack_top;
     // TODO(merry): Perform a hardware test to determine the below value.

src/core/hle/kernel/svc.cpp

@@ -164,6 +164,36 @@ static ResultCode SetHeapSize32(Core::System& system, u32* heap_addr, u32 heap_s
     return result;
 }
 
+constexpr bool IsValidSetMemoryPermission(MemoryPermission perm) {
+    switch (perm) {
+    case MemoryPermission::None:
+    case MemoryPermission::Read:
+    case MemoryPermission::ReadWrite:
+        return true;
+    default:
+        return false;
+    }
+}
+
+static ResultCode SetMemoryPermission(Core::System& system, VAddr address, u64 size,
+                                      MemoryPermission perm) {
+    // Validate address / size.
+    R_UNLESS(Common::IsAligned(address, PageSize), ResultInvalidAddress);
+    R_UNLESS(Common::IsAligned(size, PageSize), ResultInvalidSize);
+    R_UNLESS(size > 0, ResultInvalidSize);
+    R_UNLESS((address < address + size), ResultInvalidCurrentMemory);
+
+    // Validate the permission.
+    R_UNLESS(IsValidSetMemoryPermission(perm), ResultInvalidNewMemoryPermission);
+
+    // Validate that the region is in range for the current process.
+    auto& page_table = system.Kernel().CurrentProcess()->PageTable();
+    R_UNLESS(page_table.Contains(address, size), ResultInvalidCurrentMemory);
+
+    // Set the memory attribute.
+    return page_table.SetMemoryPermission(address, size, perm);
+}
+
 static ResultCode SetMemoryAttribute(Core::System& system, VAddr address, u64 size, u32 mask,
                                      u32 attribute) {
     LOG_DEBUG(Kernel_SVC,
@@ -2724,7 +2754,7 @@ static const FunctionDef SVC_Table_32[] = {
 static const FunctionDef SVC_Table_64[] = {
     {0x00, nullptr, "Unknown"},
     {0x01, SvcWrap64<SetHeapSize>, "SetHeapSize"},
-    {0x02, nullptr, "SetMemoryPermission"},
+    {0x02, SvcWrap64<SetMemoryPermission>, "SetMemoryPermission"},
     {0x03, SvcWrap64<SetMemoryAttribute>, "SetMemoryAttribute"},
     {0x04, SvcWrap64<MapMemory>, "MapMemory"},
     {0x05, SvcWrap64<UnmapMemory>, "UnmapMemory"},

src/core/hle/kernel/svc_wrap.h

@@ -249,6 +249,14 @@ void SvcWrap64(Core::System& system) {
         func(system, Param(system, 0), Param(system, 1), static_cast<u32>(Param(system, 2))).raw);
 }
 
+// Used by SetMemoryPermission
+template <ResultCode func(Core::System&, u64, u64, Svc::MemoryPermission)>
+void SvcWrap64(Core::System& system) {
+    FuncReturn(system, func(system, Param(system, 0), Param(system, 1),
+                            static_cast<Svc::MemoryPermission>(Param(system, 2)))
+                           .raw);
+}
+
 // Used by MapSharedMemory
 template <ResultCode func(Core::System&, Handle, u64, u64, Svc::MemoryPermission)>
 void SvcWrap64(Core::System& system) {

src/shader_recompiler/backend/glasm/emit_glasm_context_get_set.cpp

@@ -86,7 +86,7 @@ void EmitGetAttribute(EmitContext& ctx, IR::Inst& inst, IR::Attribute attr, Scal
     }
     switch (attr) {
     case IR::Attribute::PrimitiveId:
-        ctx.Add("MOV.S {}.x,primitive.id;", inst);
+        ctx.Add("MOV.F {}.x,primitive.id;", inst);
         break;
     case IR::Attribute::PositionX:
     case IR::Attribute::PositionY:
@@ -113,35 +113,19 @@ void EmitGetAttribute(EmitContext& ctx, IR::Inst& inst, IR::Attribute attr, Scal
         ctx.Add("MOV.F {}.x,vertex.tesscoord.{};", inst, swizzle);
         break;
     case IR::Attribute::InstanceId:
-        ctx.Add("MOV.S {}.x,{}.instance;", inst, ctx.attrib_name);
+        ctx.Add("MOV.F {}.x,{}.instance;", inst, ctx.attrib_name);
         break;
     case IR::Attribute::VertexId:
-        ctx.Add("MOV.S {}.x,{}.id;", inst, ctx.attrib_name);
+        ctx.Add("MOV.F {}.x,{}.id;", inst, ctx.attrib_name);
         break;
     case IR::Attribute::FrontFace:
-        ctx.Add("CMP.S {}.x,{}.facing.x,0,-1;", inst, ctx.attrib_name);
+        ctx.Add("CMP.F {}.x,{}.facing.x,0,-1;", inst, ctx.attrib_name);
         break;
     default:
         throw NotImplementedException("Get attribute {}", attr);
     }
 }
 
-void EmitGetAttributeU32(EmitContext& ctx, IR::Inst& inst, IR::Attribute attr, ScalarU32) {
-    switch (attr) {
-    case IR::Attribute::PrimitiveId:
-        ctx.Add("MOV.S {}.x,primitive.id;", inst);
-        break;
-    case IR::Attribute::InstanceId:
-        ctx.Add("MOV.S {}.x,{}.instance;", inst, ctx.attrib_name);
-        break;
-    case IR::Attribute::VertexId:
-        ctx.Add("MOV.S {}.x,{}.id;", inst, ctx.attrib_name);
-        break;
-    default:
-        throw NotImplementedException("Get U32 attribute {}", attr);
-    }
-}
-
 void EmitSetAttribute(EmitContext& ctx, IR::Attribute attr, ScalarF32 value,
                       [[maybe_unused]] ScalarU32 vertex) {
     const u32 element{static_cast<u32>(attr) % 4};

src/shader_recompiler/backend/glasm/emit_glasm_instructions.h

@@ -50,7 +50,6 @@ void EmitGetCbufU32(EmitContext& ctx, IR::Inst& inst, const IR::Value& binding,
 void EmitGetCbufF32(EmitContext& ctx, IR::Inst& inst, const IR::Value& binding, ScalarU32 offset);
 void EmitGetCbufU32x2(EmitContext& ctx, IR::Inst& inst, const IR::Value& binding, ScalarU32 offset);
 void EmitGetAttribute(EmitContext& ctx, IR::Inst& inst, IR::Attribute attr, ScalarU32 vertex);
-void EmitGetAttributeU32(EmitContext& ctx, IR::Inst& inst, IR::Attribute attr, ScalarU32 vertex);
 void EmitSetAttribute(EmitContext& ctx, IR::Attribute attr, ScalarF32 value, ScalarU32 vertex);
 void EmitGetAttributeIndexed(EmitContext& ctx, IR::Inst& inst, ScalarS32 offset, ScalarU32 vertex);
 void EmitSetAttributeIndexed(EmitContext& ctx, ScalarU32 offset, ScalarF32 value, ScalarU32 vertex);

src/shader_recompiler/backend/glsl/emit_glsl_bitwise_conversion.cpp

@@ -7,7 +7,6 @@
 #include "shader_recompiler/backend/glsl/emit_glsl_instructions.h"
 #include "shader_recompiler/backend/glsl/glsl_emit_context.h"
 #include "shader_recompiler/frontend/ir/value.h"
-#include "shader_recompiler/profile.h"
 
 namespace Shader::Backend::GLSL {
 namespace {
@@ -31,9 +30,8 @@ void EmitConditionRef(EmitContext& ctx, IR::Inst& inst, const IR::Value& value)
     inst.DestructiveAddUsage(1);
     const auto ret{ctx.var_alloc.Define(inst, GlslVarType::U1)};
     const auto input{ctx.var_alloc.Consume(value)};
-    const auto suffix{ctx.profile.has_gl_bool_ref_bug ? "?true:false" : ""};
     if (ret != input) {
-        ctx.Add("{}={}{};", ret, input, suffix);
+        ctx.Add("{}={};", ret, input);
     }
 }
 

src/shader_recompiler/backend/glsl/emit_glsl_context_get_set.cpp

@@ -102,46 +102,39 @@ void GetCbuf16(EmitContext& ctx, IR::Inst& inst, const IR::Value& binding, const
 
 void EmitGetCbufU8(EmitContext& ctx, IR::Inst& inst, const IR::Value& binding,
                    const IR::Value& offset) {
-    const auto cast{ctx.profile.has_gl_cbuf_ftou_bug ? "" : "ftou"};
-    GetCbuf8(ctx, inst, binding, offset, cast);
+    GetCbuf8(ctx, inst, binding, offset, "ftou");
 }
 
 void EmitGetCbufS8(EmitContext& ctx, IR::Inst& inst, const IR::Value& binding,
                    const IR::Value& offset) {
-    const auto cast{ctx.profile.has_gl_cbuf_ftou_bug ? "int" : "ftoi"};
-    GetCbuf8(ctx, inst, binding, offset, cast);
+    GetCbuf8(ctx, inst, binding, offset, "ftoi");
 }
 
 void EmitGetCbufU16(EmitContext& ctx, IR::Inst& inst, const IR::Value& binding,
                     const IR::Value& offset) {
-    const auto cast{ctx.profile.has_gl_cbuf_ftou_bug ? "" : "ftou"};
-    GetCbuf16(ctx, inst, binding, offset, cast);
+    GetCbuf16(ctx, inst, binding, offset, "ftou");
 }
 
 void EmitGetCbufS16(EmitContext& ctx, IR::Inst& inst, const IR::Value& binding,
                     const IR::Value& offset) {
-    const auto cast{ctx.profile.has_gl_cbuf_ftou_bug ? "int" : "ftoi"};
-    GetCbuf16(ctx, inst, binding, offset, cast);
+    GetCbuf16(ctx, inst, binding, offset, "ftoi");
 }
 
 void EmitGetCbufU32(EmitContext& ctx, IR::Inst& inst, const IR::Value& binding,
                     const IR::Value& offset) {
     const auto ret{ctx.var_alloc.Define(inst, GlslVarType::U32)};
-    const auto cast{ctx.profile.has_gl_cbuf_ftou_bug ? "" : "ftou"};
-    GetCbuf(ctx, ret, binding, offset, 32, cast);
+    GetCbuf(ctx, ret, binding, offset, 32, "ftou");
 }
 
 void EmitGetCbufF32(EmitContext& ctx, IR::Inst& inst, const IR::Value& binding,
                     const IR::Value& offset) {
     const auto ret{ctx.var_alloc.Define(inst, GlslVarType::F32)};
-    const auto cast{ctx.profile.has_gl_cbuf_ftou_bug ? "utof" : ""};
-    GetCbuf(ctx, ret, binding, offset, 32, cast);
+    GetCbuf(ctx, ret, binding, offset, 32);
 }
 
 void EmitGetCbufU32x2(EmitContext& ctx, IR::Inst& inst, const IR::Value& binding,
                       const IR::Value& offset) {
     const auto cbuf{fmt::format("{}_cbuf{}", ctx.stage_name, binding.U32())};
-    const auto cast{ctx.profile.has_gl_cbuf_ftou_bug ? "" : "ftou"};
     if (offset.IsImmediate()) {
         static constexpr u32 cbuf_size{0x10000};
         const u32 u32_offset{offset.U32()};
@@ -152,26 +145,26 @@ void EmitGetCbufU32x2(EmitContext& ctx, IR::Inst& inst, const IR::Value& binding
             return;
         }
         if (u32_offset % 2 == 0) {
-            ctx.AddU32x2("{}={}({}[{}].{}{});", inst, cast, cbuf, u32_offset / 16,
+            ctx.AddU32x2("{}=ftou({}[{}].{}{});", inst, cbuf, u32_offset / 16,
                          OffsetSwizzle(u32_offset), OffsetSwizzle(u32_offset + 4));
         } else {
-            ctx.AddU32x2("{}=uvec2({}({}[{}].{}),{}({}[{}].{}));", inst, cast, cbuf,
-                         u32_offset / 16, OffsetSwizzle(u32_offset), cast, cbuf,
-                         (u32_offset + 4) / 16, OffsetSwizzle(u32_offset + 4));
+            ctx.AddU32x2("{}=uvec2(ftou({}[{}].{}),ftou({}[{}].{}));", inst, cbuf, u32_offset / 16,
+                         OffsetSwizzle(u32_offset), cbuf, (u32_offset + 4) / 16,
+                         OffsetSwizzle(u32_offset + 4));
         }
         return;
     }
     const auto offset_var{ctx.var_alloc.Consume(offset)};
     if (!ctx.profile.has_gl_component_indexing_bug) {
-        ctx.AddU32x2("{}=uvec2({}({}[{}>>4][({}>>2)%4]),{}({}[({}+4)>>4][(({}+4)>>2)%4]));", inst,
-                     cast, cbuf, offset_var, offset_var, cast, cbuf, offset_var, offset_var);
+        ctx.AddU32x2("{}=uvec2(ftou({}[{}>>4][({}>>2)%4]),ftou({}[({}+4)>>4][(({}+4)>>2)%4]));",
+                     inst, cbuf, offset_var, offset_var, cbuf, offset_var, offset_var);
         return;
     }
     const auto ret{ctx.var_alloc.Define(inst, GlslVarType::U32x2)};
     const auto cbuf_offset{fmt::format("{}>>2", offset_var)};
     for (u32 swizzle = 0; swizzle < 4; ++swizzle) {
-        ctx.Add("if(({}&3)=={}){}=uvec2({}({}[{}>>4].{}),{}({}[({}+4)>>4].{}));", cbuf_offset,
-                swizzle, ret, cast, cbuf, offset_var, "xyzw"[swizzle], cast, cbuf, offset_var,
+        ctx.Add("if(({}&3)=={}){}=uvec2(ftou({}[{}>>4].{}),ftou({}[({}+4)>>4].{}));", cbuf_offset,
+                swizzle, ret, cbuf, offset_var, "xyzw"[swizzle], cbuf, offset_var,
                 "xyzw"[(swizzle + 1) % 4]);
     }
 }
@@ -228,22 +221,6 @@ void EmitGetAttribute(EmitContext& ctx, IR::Inst& inst, IR::Attribute attr,
     }
 }
 
-void EmitGetAttributeU32(EmitContext& ctx, IR::Inst& inst, IR::Attribute attr, std::string_view) {
-    switch (attr) {
-    case IR::Attribute::PrimitiveId:
-        ctx.AddU32("{}=uint(gl_PrimitiveID);", inst);
-        break;
-    case IR::Attribute::InstanceId:
-        ctx.AddU32("{}=uint(gl_InstanceID);", inst);
-        break;
-    case IR::Attribute::VertexId:
-        ctx.AddU32("{}=uint(gl_VertexID);", inst);
-        break;
-    default:
-        throw NotImplementedException("Get U32 attribute {}", attr);
-    }
-}
-
 void EmitSetAttribute(EmitContext& ctx, IR::Attribute attr, std::string_view value,
                       [[maybe_unused]] std::string_view vertex) {
     if (IR::IsGeneric(attr)) {

src/shader_recompiler/backend/glsl/emit_glsl_floating_point.cpp

@@ -125,11 +125,11 @@ void EmitFPNeg16([[maybe_unused]] EmitContext& ctx, [[maybe_unused]] IR::Inst& i
 }
 
 void EmitFPNeg32(EmitContext& ctx, IR::Inst& inst, std::string_view value) {
-    ctx.AddF32("{}=0.f-({});", inst, value);
+    ctx.AddF32("{}=-({});", inst, value);
 }
 
 void EmitFPNeg64(EmitContext& ctx, IR::Inst& inst, std::string_view value) {
-    ctx.AddF64("{}=double(0.)-({});", inst, value);
+    ctx.AddF64("{}=-({});", inst, value);
 }
 
 void EmitFPSin(EmitContext& ctx, IR::Inst& inst, std::string_view value) {

src/shader_recompiler/backend/glsl/emit_glsl_instructions.h

@@ -60,8 +60,6 @@ void EmitGetCbufU32x2(EmitContext& ctx, IR::Inst& inst, const IR::Value& binding
                       const IR::Value& offset);
 void EmitGetAttribute(EmitContext& ctx, IR::Inst& inst, IR::Attribute attr,
                       std::string_view vertex);
-void EmitGetAttributeU32(EmitContext& ctx, IR::Inst& inst, IR::Attribute attr,
-                         std::string_view vertex);
 void EmitSetAttribute(EmitContext& ctx, IR::Attribute attr, std::string_view value,
                       std::string_view vertex);
 void EmitGetAttributeIndexed(EmitContext& ctx, IR::Inst& inst, std::string_view offset,

src/shader_recompiler/backend/glsl/emit_glsl_integer.cpp

@@ -87,11 +87,11 @@ void EmitUDiv32(EmitContext& ctx, IR::Inst& inst, std::string_view a, std::strin
 }
 
 void EmitINeg32(EmitContext& ctx, IR::Inst& inst, std::string_view value) {
-    ctx.AddU32("{}=uint(int(0)-int({}));", inst, value);
+    ctx.AddU32("{}=uint(-({}));", inst, value);
 }
 
 void EmitINeg64(EmitContext& ctx, IR::Inst& inst, std::string_view value) {
-    ctx.AddU64("{}=uint64_t(int64_t(0)-int64_t({}));", inst, value);
+    ctx.AddU64("{}=-({});", inst, value);
 }
 
 void EmitIAbs32(EmitContext& ctx, IR::Inst& inst, std::string_view value) {

src/shader_recompiler/backend/glsl/emit_glsl_special.cpp

@@ -90,9 +90,7 @@ void EmitPhiMove(EmitContext& ctx, const IR::Value& phi_value, const IR::Value&
     if (phi_reg == val_reg) {
         return;
     }
-    const bool needs_workaround{ctx.profile.has_gl_bool_ref_bug && phi_type == IR::Type::U1};
-    const auto suffix{needs_workaround ? "?true:false" : ""};
-    ctx.Add("{}={}{};", phi_reg, val_reg, suffix);
+    ctx.Add("{}={};", phi_reg, val_reg);
 }
 
 void EmitPrologue(EmitContext& ctx) {

src/shader_recompiler/backend/glsl/glsl_emit_context.cpp

@@ -428,10 +428,9 @@ void EmitContext::DefineConstantBuffers(Bindings& bindings) {
         return;
     }
     for (const auto& desc : info.constant_buffer_descriptors) {
-        const auto cbuf_type{profile.has_gl_cbuf_ftou_bug ? "uvec4" : "vec4"};
-        header += fmt::format("layout(std140,binding={}) uniform {}_cbuf_{}{{{} {}_cbuf{}[{}];}};",
-                              bindings.uniform_buffer, stage_name, desc.index, cbuf_type,
-                              stage_name, desc.index, 4 * 1024);
+        header += fmt::format(
+            "layout(std140,binding={}) uniform {}_cbuf_{}{{vec4 {}_cbuf{}[{}];}};",
+            bindings.uniform_buffer, stage_name, desc.index, stage_name, desc.index, 4 * 1024);
         bindings.uniform_buffer += desc.count;
     }
 }

src/shader_recompiler/backend/spirv/emit_spirv.cpp

@@ -30,11 +30,20 @@ struct FuncTraits<ReturnType_ (*)(Args...)> {
     using ArgType = std::tuple_element_t<I, std::tuple<Args...>>;
 };
 
+#ifdef _MSC_VER
+#pragma warning(push)
+#pragma warning(disable : 4702) // Ignore unreachable code warning
+#endif
+
 template <auto func, typename... Args>
 void SetDefinition(EmitContext& ctx, IR::Inst* inst, Args... args) {
     inst->SetDefinition<Id>(func(ctx, std::forward<Args>(args)...));
 }
 
+#ifdef _MSC_VER
+#pragma warning(pop)
+#endif
+
 template <typename ArgType>
 ArgType Arg(EmitContext& ctx, const IR::Value& arg) {
     if constexpr (std::is_same_v<ArgType, Id>) {

src/shader_recompiler/backend/spirv/emit_spirv_context_get_set.cpp

@@ -355,31 +355,6 @@ Id EmitGetAttribute(EmitContext& ctx, IR::Attribute attr, Id vertex) {
     }
 }
 
-Id EmitGetAttributeU32(EmitContext& ctx, IR::Attribute attr, Id) {
-    switch (attr) {
-    case IR::Attribute::PrimitiveId:
-        return ctx.OpLoad(ctx.U32[1], ctx.primitive_id);
-    case IR::Attribute::InstanceId:
-        if (ctx.profile.support_vertex_instance_id) {
-            return ctx.OpLoad(ctx.U32[1], ctx.instance_id);
-        } else {
-            const Id index{ctx.OpLoad(ctx.U32[1], ctx.instance_index)};
-            const Id base{ctx.OpLoad(ctx.U32[1], ctx.base_instance)};
-            return ctx.OpISub(ctx.U32[1], index, base);
-        }
-    case IR::Attribute::VertexId:
-        if (ctx.profile.support_vertex_instance_id) {
-            return ctx.OpLoad(ctx.U32[1], ctx.vertex_id);
-        } else {
-            const Id index{ctx.OpLoad(ctx.U32[1], ctx.vertex_index)};
-            const Id base{ctx.OpLoad(ctx.U32[1], ctx.base_vertex)};
-            return ctx.OpISub(ctx.U32[1], index, base);
-        }
-    default:
-        throw NotImplementedException("Read U32 attribute {}", attr);
-    }
-}
-
 void EmitSetAttribute(EmitContext& ctx, IR::Attribute attr, Id value, [[maybe_unused]] Id vertex) {
     const std::optional<OutAttr> output{OutputAttrPointer(ctx, attr)};
     if (!output) {

src/shader_recompiler/backend/spirv/emit_spirv_instructions.h

@@ -53,7 +53,6 @@ Id EmitGetCbufU32(EmitContext& ctx, const IR::Value& binding, const IR::Value& o
 Id EmitGetCbufF32(EmitContext& ctx, const IR::Value& binding, const IR::Value& offset);
 Id EmitGetCbufU32x2(EmitContext& ctx, const IR::Value& binding, const IR::Value& offset);
 Id EmitGetAttribute(EmitContext& ctx, IR::Attribute attr, Id vertex);
-Id EmitGetAttributeU32(EmitContext& ctx, IR::Attribute attr, Id vertex);
 void EmitSetAttribute(EmitContext& ctx, IR::Attribute attr, Id value, Id vertex);
 Id EmitGetAttributeIndexed(EmitContext& ctx, Id offset, Id vertex);
 void EmitSetAttributeIndexed(EmitContext& ctx, Id offset, Id value, Id vertex);

src/shader_recompiler/frontend/ir/opcodes.inc

@@ -40,7 +40,6 @@ OPCODE(GetCbufU32,                                          U32,            U32,
 OPCODE(GetCbufF32,                                          F32,            U32,            U32,                                                            )
 OPCODE(GetCbufU32x2,                                        U32x2,          U32,            U32,                                                            )
 OPCODE(GetAttribute,                                        F32,            Attribute,      U32,                                                            )
-OPCODE(GetAttributeU32,                                     U32,            Attribute,      U32,                                                            )
 OPCODE(SetAttribute,                                        Void,           Attribute,      F32,            U32,                                            )
 OPCODE(GetAttributeIndexed,                                 F32,            U32,            U32,                                                            )
 OPCODE(SetAttributeIndexed,                                 Void,           U32,            F32,            U32,                                            )

src/shader_recompiler/ir_opt/collect_shader_info_pass.cpp

@@ -389,7 +389,6 @@ void VisitUsages(Info& info, IR::Inst& inst) {
         info.uses_demote_to_helper_invocation = true;
         break;
     case IR::Opcode::GetAttribute:
-    case IR::Opcode::GetAttributeU32:
         info.loads.mask[static_cast<size_t>(inst.Arg(0).Attribute())] = true;
         break;
     case IR::Opcode::SetAttribute:

src/shader_recompiler/ir_opt/constant_propagation_pass.cpp

@@ -505,29 +505,6 @@ void FoldBitCast(IR::Inst& inst, IR::Opcode reverse) {
             return;
         }
     }
-    if constexpr (op == IR::Opcode::BitCastU32F32) {
-        // Workaround for new NVIDIA driver bug, where:
-        // uint attr = ftou(itof(gl_InstanceID));
-        // always returned 0.
-        // We can instead manually optimize this and work around the driver bug:
-        // uint attr = uint(gl_InstanceID);
-        if (arg_inst->GetOpcode() == IR::Opcode::GetAttribute) {
-            const IR::Attribute attr{arg_inst->Arg(0).Attribute()};
-            switch (attr) {
-            case IR::Attribute::PrimitiveId:
-            case IR::Attribute::InstanceId:
-            case IR::Attribute::VertexId:
-                break;
-            default:
-                return;
-            }
-            // Replace the bitcasts with an integer attribute get
-            inst.ReplaceOpcode(IR::Opcode::GetAttributeU32);
-            inst.SetArg(0, arg_inst->Arg(0));
-            inst.SetArg(1, arg_inst->Arg(1));
-            return;
-        }
-    }
 }
 
 void FoldInverseFunc(IR::Inst& inst, IR::Opcode reverse) {

src/shader_recompiler/profile.h

@@ -65,10 +65,6 @@ struct Profile {
     bool has_gl_component_indexing_bug{};
     /// The precise type qualifier is broken in the fragment stage of some drivers
     bool has_gl_precise_bug{};
-    /// Some drivers do not properly support floatBitsToUint when used on cbufs
-    bool has_gl_cbuf_ftou_bug{};
-    /// Some drivers poorly optimize boolean variable references
-    bool has_gl_bool_ref_bug{};
     /// Ignores SPIR-V ordered vs unordered using GLSL semantics
     bool ignore_nan_fp_comparisons{};
 

src/video_core/renderer_opengl/gl_device.cpp

@@ -182,13 +182,17 @@ Device::Device() {
         shader_backend = Settings::ShaderBackend::GLSL;
     }
 
-    if (shader_backend == Settings::ShaderBackend::GLSL && is_nvidia) {
+    if (shader_backend == Settings::ShaderBackend::GLSL && is_nvidia &&
+        !Settings::values.renderer_debug) {
         const std::string_view driver_version = version.substr(13);
         const int version_major =
             std::atoi(driver_version.substr(0, driver_version.find(".")).data());
+
         if (version_major >= 495) {
-            has_cbuf_ftou_bug = true;
-            has_bool_ref_bug = true;
+            LOG_WARNING(Render_OpenGL, "NVIDIA drivers 495 and later causes significant problems "
+                                       "with yuzu. Forcing GLASM as a mitigation.");
+            shader_backend = Settings::ShaderBackend::GLASM;
+            use_assembly_shaders = true;
         }
     }
 

src/video_core/renderer_opengl/gl_device.h

@@ -152,14 +152,6 @@ public:
         return need_fastmath_off;
     }
 
-    bool HasCbufFtouBug() const {
-        return has_cbuf_ftou_bug;
-    }
-
-    bool HasBoolRefBug() const {
-        return has_bool_ref_bug;
-    }
-
     Settings::ShaderBackend GetShaderBackend() const {
         return shader_backend;
     }
@@ -208,8 +200,6 @@ private:
     bool has_sparse_texture_2{};
     bool warp_size_potentially_larger_than_guest{};
     bool need_fastmath_off{};
-    bool has_cbuf_ftou_bug{};
-    bool has_bool_ref_bug{};
 
     std::string vendor_name;
 };

src/video_core/renderer_opengl/gl_shader_cache.cpp

@@ -214,8 +214,6 @@ ShaderCache::ShaderCache(RasterizerOpenGL& rasterizer_, Core::Frontend::EmuWindo
           .has_broken_fp16_float_controls = false,
           .has_gl_component_indexing_bug = device.HasComponentIndexingBug(),
           .has_gl_precise_bug = device.HasPreciseBug(),
-          .has_gl_cbuf_ftou_bug = device.HasCbufFtouBug(),
-          .has_gl_bool_ref_bug = device.HasBoolRefBug(),
           .ignore_nan_fp_comparisons = true,
           .gl_max_compute_smem_size = device.GetMaxComputeSharedMemorySize(),
       },

src/video_core/renderer_opengl/gl_texture_cache.cpp

@@ -1047,7 +1047,7 @@ bool Image::ScaleDown(bool ignore) {
 }
 
 ImageView::ImageView(TextureCacheRuntime& runtime, const VideoCommon::ImageViewInfo& info,
-                     ImageId image_id_, Image& image)
+                     ImageId image_id_, Image& image, const SlotVector<Image>&)
     : VideoCommon::ImageViewBase{info, image.info, image_id_}, views{runtime.null_image_views} {
     const Device& device = runtime.device;
     if (True(image.flags & ImageFlagBits::Converted)) {

src/video_core/renderer_opengl/gl_texture_cache.h

@@ -36,6 +36,7 @@ using VideoCommon::ImageViewType;
 using VideoCommon::NUM_RT;
 using VideoCommon::Region2D;
 using VideoCommon::RenderTargets;
+using VideoCommon::SlotVector;
 
 struct ImageBufferMap {
     ~ImageBufferMap();
@@ -234,7 +235,8 @@ class ImageView : public VideoCommon::ImageViewBase {
     friend Image;
 
 public:
-    explicit ImageView(TextureCacheRuntime&, const VideoCommon::ImageViewInfo&, ImageId, Image&);
+    explicit ImageView(TextureCacheRuntime&, const VideoCommon::ImageViewInfo&, ImageId, Image&,
+                       const SlotVector<Image>&);
     explicit ImageView(TextureCacheRuntime&, const VideoCommon::ImageInfo&,
                        const VideoCommon::ImageViewInfo&, GPUVAddr);
     explicit ImageView(TextureCacheRuntime&, const VideoCommon::ImageInfo& info,

src/video_core/renderer_vulkan/vk_texture_cache.cpp

@@ -1473,8 +1473,7 @@ bool Image::BlitScaleHelper(bool scale_up) {
 ImageView::ImageView(TextureCacheRuntime& runtime, const VideoCommon::ImageViewInfo& info,
                      ImageId image_id_, Image& image)
     : VideoCommon::ImageViewBase{info, image.info, image_id_}, device{&runtime.device},
-      src_image{&image}, image_handle{image.Handle()},
-      samples(ConvertSampleCount(image.info.num_samples)) {
+      image_handle{image.Handle()}, samples(ConvertSampleCount(image.info.num_samples)) {
     using Shader::TextureType;
 
     const VkImageAspectFlags aspect_mask = ImageViewAspectMask(info);
@@ -1557,6 +1556,12 @@ ImageView::ImageView(TextureCacheRuntime& runtime, const VideoCommon::ImageViewI
     }
 }
 
+ImageView::ImageView(TextureCacheRuntime& runtime, const VideoCommon::ImageViewInfo& info,
+                     ImageId image_id_, Image& image, const SlotVector<Image>& slot_imgs)
+    : ImageView{runtime, info, image_id_, image} {
+    slot_images = &slot_imgs;
+}
+
 ImageView::ImageView(TextureCacheRuntime&, const VideoCommon::ImageInfo& info,
                      const VideoCommon::ImageViewInfo& view_info, GPUVAddr gpu_addr_)
     : VideoCommon::ImageViewBase{info, view_info}, gpu_addr{gpu_addr_},
@@ -1613,10 +1618,12 @@ VkImageView ImageView::StorageView(Shader::TextureType texture_type,
 }
 
 bool ImageView::IsRescaled() const noexcept {
-    if (!src_image) {
+    if (!slot_images) {
         return false;
     }
-    return src_image->IsRescaled();
+    const auto& slots = *slot_images;
+    const auto& src_image = slots[image_id];
+    return src_image.IsRescaled();
 }
 
 vk::ImageView ImageView::MakeView(VkFormat vk_format, VkImageAspectFlags aspect_mask) {

src/video_core/renderer_vulkan/vk_texture_cache.h

@@ -23,6 +23,7 @@ using VideoCommon::ImageId;
 using VideoCommon::NUM_RT;
 using VideoCommon::Region2D;
 using VideoCommon::RenderTargets;
+using VideoCommon::SlotVector;
 using VideoCore::Surface::PixelFormat;
 
 class ASTCDecoderPass;
@@ -170,6 +171,8 @@ private:
 class ImageView : public VideoCommon::ImageViewBase {
 public:
     explicit ImageView(TextureCacheRuntime&, const VideoCommon::ImageViewInfo&, ImageId, Image&);
+    explicit ImageView(TextureCacheRuntime&, const VideoCommon::ImageViewInfo&, ImageId, Image&,
+                       const SlotVector<Image>&);
     explicit ImageView(TextureCacheRuntime&, const VideoCommon::ImageInfo&,
                        const VideoCommon::ImageViewInfo&, GPUVAddr);
     explicit ImageView(TextureCacheRuntime&, const VideoCommon::NullImageViewParams&);
@@ -226,7 +229,7 @@ private:
     [[nodiscard]] vk::ImageView MakeView(VkFormat vk_format, VkImageAspectFlags aspect_mask);
 
     const Device* device = nullptr;
-    const Image* src_image{};
+    const SlotVector<Image>* slot_images = nullptr;
 
     std::array<vk::ImageView, Shader::NUM_TEXTURE_TYPES> image_views;
     std::unique_ptr<StorageViews> storage_views;

src/video_core/texture_cache/texture_cache.h

@@ -1397,7 +1397,8 @@ ImageViewId TextureCache<P>::FindOrEmplaceImageView(ImageId image_id, const Imag
     if (const ImageViewId image_view_id = image.FindView(info); image_view_id) {
         return image_view_id;
     }
-    const ImageViewId image_view_id = slot_image_views.insert(runtime, info, image_id, image);
+    const ImageViewId image_view_id =
+        slot_image_views.insert(runtime, info, image_id, image, slot_images);
     image.InsertView(info, image_view_id);
     return image_view_id;
 }