"Merge Tagged PR 12173"

src/core/core.cpp

@@ -242,7 +242,7 @@ struct System::Impl {
     void Run() {
         std::unique_lock<std::mutex> lk(suspend_guard);
 
-        kernel.SuspendApplication(false);
+        kernel.SuspendEmulation(false);
         core_timing.SyncPause(false);
         is_paused.store(false, std::memory_order_relaxed);
     }
@@ -251,7 +251,7 @@ struct System::Impl {
         std::unique_lock<std::mutex> lk(suspend_guard);
 
         core_timing.SyncPause(true);
-        kernel.SuspendApplication(true);
+        kernel.SuspendEmulation(true);
         is_paused.store(true, std::memory_order_relaxed);
     }
 
@@ -261,7 +261,7 @@ struct System::Impl {
 
     std::unique_lock<std::mutex> StallApplication() {
         std::unique_lock<std::mutex> lk(suspend_guard);
-        kernel.SuspendApplication(true);
+        kernel.SuspendEmulation(true);
         core_timing.SyncPause(true);
         return lk;
     }
@@ -269,7 +269,7 @@ struct System::Impl {
     void UnstallApplication() {
         if (!IsPaused()) {
             core_timing.SyncPause(false);
-            kernel.SuspendApplication(false);
+            kernel.SuspendEmulation(false);
         }
     }
 
@@ -459,7 +459,7 @@ struct System::Impl {
         }
 
         Network::CancelPendingSocketOperations();
-        kernel.SuspendApplication(true);
+        kernel.SuspendEmulation(true);
         if (services) {
             services->KillNVNFlinger();
         }

src/core/device_memory_manager.inc

@@ -522,13 +522,17 @@ void DeviceMemoryManager<Traits>::UpdatePagesCachedCount(DAddr addr, size_t size
     auto* memory_device_inter = registered_processes[asid.id];
     const auto release_pending = [&] {
         if (uncache_bytes > 0) {
-            MarkRegionCaching(memory_device_inter, uncache_begin << Memory::YUZU_PAGEBITS,
-                              uncache_bytes, false);
+            if (memory_device_inter != nullptr) {
+                MarkRegionCaching(memory_device_inter, uncache_begin << Memory::YUZU_PAGEBITS,
+                                  uncache_bytes, false);
+            }
             uncache_bytes = 0;
         }
         if (cache_bytes > 0) {
-            MarkRegionCaching(memory_device_inter, cache_begin << Memory::YUZU_PAGEBITS,
-                              cache_bytes, true);
+            if (memory_device_inter != nullptr) {
+                MarkRegionCaching(memory_device_inter, cache_begin << Memory::YUZU_PAGEBITS,
+                                  cache_bytes, true);
+            }
             cache_bytes = 0;
         }
     };

src/core/hle/kernel/k_thread.h

@@ -66,6 +66,7 @@ enum class SuspendType : u32 {
     Debug = 2,
     Backtrace = 3,
     Init = 4,
+    System = 5,
 
     Count,
 };
@@ -84,8 +85,9 @@ enum class ThreadState : u16 {
     DebugSuspended = (1 << (2 + SuspendShift)),
     BacktraceSuspended = (1 << (3 + SuspendShift)),
     InitSuspended = (1 << (4 + SuspendShift)),
+    SystemSuspended = (1 << (5 + SuspendShift)),
 
-    SuspendFlagMask = ((1 << 5) - 1) << SuspendShift,
+    SuspendFlagMask = ((1 << 6) - 1) << SuspendShift,
 };
 DECLARE_ENUM_FLAG_OPERATORS(ThreadState);
 

src/core/hle/kernel/kernel.cpp

@@ -1204,39 +1204,48 @@ const Kernel::KSharedMemory& KernelCore::GetHidBusSharedMem() const {
     return *impl->hidbus_shared_mem;
 }
 
-void KernelCore::SuspendApplication(bool suspended) {
+void KernelCore::SuspendEmulation(bool suspended) {
     const bool should_suspend{exception_exited || suspended};
-    const auto activity =
-        should_suspend ? Svc::ProcessActivity::Paused : Svc::ProcessActivity::Runnable;
+    auto processes = GetProcessList();
 
-    // Get the application process.
-    KScopedAutoObject<KProcess> process = ApplicationProcess();
-    if (process.IsNull()) {
+    for (auto& process : processes) {
+        KScopedLightLock ll{process->GetListLock()};
+
+        for (auto& thread : process->GetThreadList()) {
+            if (should_suspend) {
+                thread.RequestSuspend(SuspendType::System);
+            } else {
+                thread.Resume(SuspendType::System);
+            }
+        }
+    }
+
+    if (!should_suspend) {
         return;
     }
 
-    // Set the new activity.
-    process->SetActivity(activity);
-
     // Wait for process execution to stop.
-    bool must_wait{should_suspend};
-
-    // KernelCore::SuspendApplication must be called from locked context,
-    // or we could race another call to SetActivity, interfering with waiting.
-    while (must_wait) {
+    // KernelCore::SuspendEmulation must be called from locked context,
+    // or we could race another call, interfering with waiting.
+    const auto TryWait = [&]() {
         KScopedSchedulerLock sl{*this};
 
-        // Assume that all threads have finished running.
-        must_wait = false;
-
-        for (auto i = 0; i < static_cast<s32>(Core::Hardware::NUM_CPU_CORES); ++i) {
-            if (Scheduler(i).GetSchedulerCurrentThread()->GetOwnerProcess() ==
-                process.GetPointerUnsafe()) {
-                // A thread has not finished running yet.
-                // Continue waiting.
-                must_wait = true;
+        for (auto& process : processes) {
+            for (auto i = 0; i < static_cast<s32>(Core::Hardware::NUM_CPU_CORES); ++i) {
+                if (Scheduler(i).GetSchedulerCurrentThread()->GetOwnerProcess() ==
+                    process.GetPointerUnsafe()) {
+                    // A thread has not finished running yet.
+                    // Continue waiting.
+                    return false;
+                }
             }
         }
+
+        return true;
+    };
+
+    while (!TryWait()) {
+        // ...
     }
 }
 
@@ -1260,7 +1269,7 @@ bool KernelCore::IsShuttingDown() const {
 
 void KernelCore::ExceptionalExitApplication() {
     exception_exited = true;
-    SuspendApplication(true);
+    SuspendEmulation(true);
 }
 
 void KernelCore::EnterSVCProfile() {

src/core/hle/kernel/kernel.h

@@ -258,8 +258,8 @@ public:
     /// Gets the shared memory object for HIDBus services.
     const Kernel::KSharedMemory& GetHidBusSharedMem() const;
 
-    /// Suspend/unsuspend application process.
-    void SuspendApplication(bool suspend);
+    /// Suspend/unsuspend emulated processes.
+    void SuspendEmulation(bool suspend);
 
     /// Exceptional exit application process.
     void ExceptionalExitApplication();

src/shader_recompiler/backend/spirv/emit_spirv_image.cpp

@@ -60,11 +60,10 @@ public:
         Add(spv::ImageOperandsMask::ConstOffsets, offsets);
     }
 
-    explicit ImageOperands(EmitContext& ctx, const IR::Value& offset, Id lod, Id ms) {
+    explicit ImageOperands(Id lod, Id ms) {
         if (Sirit::ValidId(lod)) {
             Add(spv::ImageOperandsMask::Lod, lod);
         }
-        AddOffset(ctx, offset, ImageFetchOffsetAllowed);
         if (Sirit::ValidId(ms)) {
             Add(spv::ImageOperandsMask::Sample, ms);
         }
@@ -312,6 +311,43 @@ Id ImageGatherSubpixelOffset(EmitContext& ctx, const IR::TextureInstInfo& info,
         return coords;
     }
 }
+
+void AddOffsetToCoordinates(EmitContext& ctx, const IR::TextureInstInfo& info, Id& coords,
+                            Id offset) {
+    if (!Sirit::ValidId(offset)) {
+        return;
+    }
+
+    Id result_type{};
+    switch (info.type) {
+    case TextureType::Buffer:
+    case TextureType::Color1D: {
+        result_type = ctx.U32[1];
+        break;
+    }
+    case TextureType::ColorArray1D:
+        offset = ctx.OpCompositeConstruct(ctx.U32[2], offset, ctx.u32_zero_value);
+        [[fallthrough]];
+    case TextureType::Color2D:
+    case TextureType::Color2DRect: {
+        result_type = ctx.U32[2];
+        break;
+    }
+    case TextureType::ColorArray2D:
+        offset = ctx.OpCompositeConstruct(ctx.U32[3], ctx.OpCompositeExtract(ctx.U32[1], coords, 0),
+                                          ctx.OpCompositeExtract(ctx.U32[1], coords, 1),
+                                          ctx.u32_zero_value);
+        [[fallthrough]];
+    case TextureType::Color3D: {
+        result_type = ctx.U32[3];
+        break;
+    }
+    case TextureType::ColorCube:
+    case TextureType::ColorArrayCube:
+        return;
+    }
+    coords = ctx.OpIAdd(result_type, coords, offset);
+}
 } // Anonymous namespace
 
 Id EmitBindlessImageSampleImplicitLod(EmitContext&) {
@@ -494,9 +530,10 @@ Id EmitImageGatherDref(EmitContext& ctx, IR::Inst* inst, const IR::Value& index,
                 operands.Span());
 }
 
-Id EmitImageFetch(EmitContext& ctx, IR::Inst* inst, const IR::Value& index, Id coords,
-                  const IR::Value& offset, Id lod, Id ms) {
+Id EmitImageFetch(EmitContext& ctx, IR::Inst* inst, const IR::Value& index, Id coords, Id offset,
+                  Id lod, Id ms) {
     const auto info{inst->Flags<IR::TextureInstInfo>()};
+    AddOffsetToCoordinates(ctx, info, coords, offset);
     if (info.type == TextureType::Buffer) {
         lod = Id{};
     }
@@ -504,7 +541,7 @@ Id EmitImageFetch(EmitContext& ctx, IR::Inst* inst, const IR::Value& index, Id c
         // This image is multisampled, lod must be implicit
         lod = Id{};
     }
-    const ImageOperands operands(ctx, offset, lod, ms);
+    const ImageOperands operands(lod, ms);
     return Emit(&EmitContext::OpImageSparseFetch, &EmitContext::OpImageFetch, ctx, inst, ctx.F32[4],
                 TextureImage(ctx, info, index), coords, operands.MaskOptional(), operands.Span());
 }

src/shader_recompiler/backend/spirv/emit_spirv_instructions.h

@@ -537,8 +537,8 @@ Id EmitImageGather(EmitContext& ctx, IR::Inst* inst, const IR::Value& index, Id
                    const IR::Value& offset, const IR::Value& offset2);
 Id EmitImageGatherDref(EmitContext& ctx, IR::Inst* inst, const IR::Value& index, Id coords,
                        const IR::Value& offset, const IR::Value& offset2, Id dref);
-Id EmitImageFetch(EmitContext& ctx, IR::Inst* inst, const IR::Value& index, Id coords,
-                  const IR::Value& offset, Id lod, Id ms);
+Id EmitImageFetch(EmitContext& ctx, IR::Inst* inst, const IR::Value& index, Id coords, Id offset,
+                  Id lod, Id ms);
 Id EmitImageQueryDimensions(EmitContext& ctx, IR::Inst* inst, const IR::Value& index, Id lod,
                             const IR::Value& skip_mips);
 Id EmitImageQueryLod(EmitContext& ctx, IR::Inst* inst, const IR::Value& index, Id coords);

src/video_core/buffer_cache/buffer_cache.h

@@ -1488,7 +1488,10 @@ void BufferCache<P>::ImmediateUploadMemory([[maybe_unused]] Buffer& buffer,
             std::span<const u8> upload_span;
             const DAddr device_addr = buffer.CpuAddr() + copy.dst_offset;
             if (IsRangeGranular(device_addr, copy.size)) {
-                upload_span = std::span(device_memory.GetPointer<u8>(device_addr), copy.size);
+                auto* const ptr = device_memory.GetPointer<u8>(device_addr);
+                if (ptr != nullptr) {
+                    upload_span = std::span(ptr, copy.size);
+                }
             } else {
                 if (immediate_buffer.empty()) {
                     immediate_buffer = ImmediateBuffer(largest_copy);

src/video_core/renderer_vulkan/vk_query_cache.cpp

@@ -1064,8 +1064,6 @@ public:
                 }
             });
         }
-        auto* ptr = device_memory.GetPointer<u8>(new_query->dependant_address);
-        ASSERT(ptr != nullptr);
 
         new_query->dependant_manage = must_manage_dependance;
         pending_flush_queries.push_back(index);
@@ -1104,9 +1102,11 @@ public:
                 tfb_streamer.Free(query->dependant_index);
             } else {
                 u8* pointer = device_memory.GetPointer<u8>(query->dependant_address);
-                u32 result;
-                std::memcpy(&result, pointer, sizeof(u32));
-                num_vertices = static_cast<u64>(result) / query->stride;
+                if (pointer != nullptr) {
+                    u32 result;
+                    std::memcpy(&result, pointer, sizeof(u32));
+                    num_vertices = static_cast<u64>(result) / query->stride;
+                }
             }
             query->value = [&]() -> u64 {
                 switch (query->topology) {
@@ -1360,7 +1360,9 @@ bool QueryCacheRuntime::HostConditionalRenderingCompareValues(VideoCommon::Looku
     const auto check_value = [&](DAddr address) {
         u8* ptr = impl->device_memory.GetPointer<u8>(address);
         u64 value{};
-        std::memcpy(&value, ptr, sizeof(value));
+        if (ptr != nullptr) {
+            std::memcpy(&value, ptr, sizeof(value));
+        }
         return value == 0;
     };
     std::array<VideoCommon::LookupData*, 2> objects{&object_1, &object_2};

src/video_core/renderer_vulkan/vk_rasterizer.cpp

@@ -1050,37 +1050,16 @@ void RasterizerVulkan::UpdateDepthBias(Tegra::Engines::Maxwell3D::Regs& regs) {
                         regs.zeta.format == Tegra::DepthFormat::X8Z24_UNORM ||
                         regs.zeta.format == Tegra::DepthFormat::S8Z24_UNORM ||
                         regs.zeta.format == Tegra::DepthFormat::V8Z24_UNORM;
-    bool force_unorm = ([&] {
-        if (!is_d24 || device.SupportsD24DepthBuffer()) {
-            return false;
-        }
-        if (device.IsExtDepthBiasControlSupported()) {
-            return true;
-        }
-        if (!Settings::values.renderer_amdvlk_depth_bias_workaround) {
-            return false;
-        }
+    if (is_d24 && !device.SupportsD24DepthBuffer() &&
+        Settings::values.renderer_amdvlk_depth_bias_workaround) {
         // the base formulas can be obtained from here:
         //   https://docs.microsoft.com/en-us/windows/win32/direct3d11/d3d10-graphics-programming-guide-output-merger-stage-depth-bias
         const double rescale_factor =
             static_cast<double>(1ULL << (32 - 24)) / (static_cast<double>(0x1.ep+127));
         units = static_cast<float>(static_cast<double>(units) * rescale_factor);
-        return false;
-    })();
+    }
     scheduler.Record([constant = units, clamp = regs.depth_bias_clamp,
-                      factor = regs.slope_scale_depth_bias, force_unorm,
-                      precise = device.HasExactDepthBiasControl()](vk::CommandBuffer cmdbuf) {
-        if (force_unorm) {
-            VkDepthBiasRepresentationInfoEXT info{
-                .sType = VK_STRUCTURE_TYPE_DEPTH_BIAS_REPRESENTATION_INFO_EXT,
-                .pNext = nullptr,
-                .depthBiasRepresentation =
-                    VK_DEPTH_BIAS_REPRESENTATION_LEAST_REPRESENTABLE_VALUE_FORCE_UNORM_EXT,
-                .depthBiasExact = precise ? VK_TRUE : VK_FALSE,
-            };
-            cmdbuf.SetDepthBias(constant, clamp, factor, &info);
-            return;
-        }
+                      factor = regs.slope_scale_depth_bias](vk::CommandBuffer cmdbuf) {
         cmdbuf.SetDepthBias(constant, clamp, factor);
     });
 }

src/video_core/vulkan_common/vulkan_device.cpp

@@ -1137,13 +1137,6 @@ void Device::RemoveUnsuitableExtensions() {
     RemoveExtensionFeatureIfUnsuitable(extensions.custom_border_color, features.custom_border_color,
                                        VK_EXT_CUSTOM_BORDER_COLOR_EXTENSION_NAME);
 
-    // VK_EXT_depth_bias_control
-    extensions.depth_bias_control =
-        features.depth_bias_control.depthBiasControl &&
-        features.depth_bias_control.leastRepresentableValueForceUnormRepresentation;
-    RemoveExtensionFeatureIfUnsuitable(extensions.depth_bias_control, features.depth_bias_control,
-                                       VK_EXT_DEPTH_BIAS_CONTROL_EXTENSION_NAME);
-
     // VK_EXT_depth_clip_control
     extensions.depth_clip_control = features.depth_clip_control.depthClipControl;
     RemoveExtensionFeatureIfUnsuitable(extensions.depth_clip_control, features.depth_clip_control,

src/video_core/vulkan_common/vulkan_device.h

@@ -41,7 +41,6 @@ VK_DEFINE_HANDLE(VmaAllocator)
 // Define all features which may be used by the implementation and require an extension here.
 #define FOR_EACH_VK_FEATURE_EXT(FEATURE)                                                           \
     FEATURE(EXT, CustomBorderColor, CUSTOM_BORDER_COLOR, custom_border_color)                      \
-    FEATURE(EXT, DepthBiasControl, DEPTH_BIAS_CONTROL, depth_bias_control)                         \
     FEATURE(EXT, DepthClipControl, DEPTH_CLIP_CONTROL, depth_clip_control)                         \
     FEATURE(EXT, ExtendedDynamicState, EXTENDED_DYNAMIC_STATE, extended_dynamic_state)             \
     FEATURE(EXT, ExtendedDynamicState2, EXTENDED_DYNAMIC_STATE_2, extended_dynamic_state2)         \
@@ -97,7 +96,6 @@ VK_DEFINE_HANDLE(VmaAllocator)
 #define FOR_EACH_VK_RECOMMENDED_EXTENSION(EXTENSION_NAME)                                          \
     EXTENSION_NAME(VK_EXT_CONDITIONAL_RENDERING_EXTENSION_NAME)                                    \
     EXTENSION_NAME(VK_EXT_CONSERVATIVE_RASTERIZATION_EXTENSION_NAME)                               \
-    EXTENSION_NAME(VK_EXT_DEPTH_BIAS_CONTROL_EXTENSION_NAME)                                       \
     EXTENSION_NAME(VK_EXT_DEPTH_RANGE_UNRESTRICTED_EXTENSION_NAME)                                 \
     EXTENSION_NAME(VK_EXT_EXTENDED_DYNAMIC_STATE_EXTENSION_NAME)                                   \
     EXTENSION_NAME(VK_EXT_EXTENDED_DYNAMIC_STATE_2_EXTENSION_NAME)                                 \
@@ -150,9 +148,6 @@ VK_DEFINE_HANDLE(VmaAllocator)
 // Define features where the absence of the feature may result in a degraded experience.
 #define FOR_EACH_VK_RECOMMENDED_FEATURE(FEATURE_NAME)                                              \
     FEATURE_NAME(custom_border_color, customBorderColors)                                          \
-    FEATURE_NAME(depth_bias_control, depthBiasControl)                                             \
-    FEATURE_NAME(depth_bias_control, leastRepresentableValueForceUnormRepresentation)              \
-    FEATURE_NAME(depth_bias_control, depthBiasExact)                                               \
     FEATURE_NAME(extended_dynamic_state, extendedDynamicState)                                     \
     FEATURE_NAME(format_a4b4g4r4, formatA4B4G4R4)                                                  \
     FEATURE_NAME(index_type_uint8, indexTypeUint8)                                                 \
@@ -479,11 +474,6 @@ public:
         return extensions.depth_clip_control;
     }
 
-    /// Returns true if the device supports VK_EXT_depth_bias_control.
-    bool IsExtDepthBiasControlSupported() const {
-        return extensions.depth_bias_control;
-    }
-
     /// Returns true if the device supports VK_EXT_shader_viewport_index_layer.
     bool IsExtShaderViewportIndexLayerSupported() const {
         return extensions.shader_viewport_index_layer;
@@ -649,10 +639,6 @@ public:
         return features.robustness2.nullDescriptor;
     }
 
-    bool HasExactDepthBiasControl() const {
-        return features.depth_bias_control.depthBiasExact;
-    }
-
     u32 GetMaxVertexInputAttributes() const {
         return properties.properties.limits.maxVertexInputAttributes;
     }