Android #94

vk_present_manager: recreate surface on any surface loss

README.md

@@ -1,3 +1,11 @@
+| Pull Request | Commit | Title | Author | Merged? |
+|----|----|----|----|----|
+
+
+End of merge log. You can find the original README.md below the break.
+
+-----
+
 <!--
 SPDX-FileCopyrightText: 2018 yuzu Emulator Project
 SPDX-License-Identifier: GPL-2.0-or-later

src/android/app/src/main/jni/native.cpp

@@ -218,7 +218,6 @@ public:
             return;
         }
         m_window->OnSurfaceChanged(m_native_window);
-        m_system.Renderer().NotifySurfaceChanged();
     }
 
     void ConfigureFilesystemProvider(const std::string& filepath) {

src/common/string_util.cpp

@@ -135,6 +135,11 @@ std::u16string UTF8ToUTF16(std::string_view input) {
     return convert.from_bytes(input.data(), input.data() + input.size());
 }
 
+std::u32string UTF8ToUTF32(std::string_view input) {
+    std::wstring_convert<std::codecvt_utf8<char32_t>, char32_t> convert;
+    return convert.from_bytes(input.data(), input.data() + input.size());
+}
+
 #ifdef _WIN32
 static std::wstring CPToUTF16(u32 code_page, std::string_view input) {
     const auto size =

src/common/string_util.h

@@ -38,6 +38,7 @@ bool SplitPath(const std::string& full_path, std::string* _pPath, std::string* _
 
 [[nodiscard]] std::string UTF16ToUTF8(std::u16string_view input);
 [[nodiscard]] std::u16string UTF8ToUTF16(std::string_view input);
+[[nodiscard]] std::u32string UTF8ToUTF32(std::string_view input);
 
 #ifdef _WIN32
 [[nodiscard]] std::string UTF16ToUTF8(std::wstring_view input);

src/core/core.cpp

@@ -1078,6 +1078,10 @@ void System::ApplySettings() {
     impl->RefreshTime();
 
     if (IsPoweredOn()) {
+        if (Settings::values.custom_rtc_enabled) {
+            const s64 posix_time{Settings::values.custom_rtc.GetValue()};
+            GetTimeManager().UpdateLocalSystemClockTime(posix_time);
+        }
         Renderer().RefreshBaseSettings();
     }
 }

src/core/debugger/gdbstub.cpp

@@ -2,6 +2,8 @@
 // SPDX-License-Identifier: GPL-2.0-or-later
 
 #include <atomic>
+#include <codecvt>
+#include <locale>
 #include <numeric>
 #include <optional>
 #include <thread>
@@ -12,6 +14,7 @@
 #include "common/logging/log.h"
 #include "common/scope_exit.h"
 #include "common/settings.h"
+#include "common/string_util.h"
 #include "core/arm/arm_interface.h"
 #include "core/core.h"
 #include "core/debugger/gdbstub.h"
@@ -68,10 +71,16 @@ static std::string EscapeGDB(std::string_view data) {
 }
 
 static std::string EscapeXML(std::string_view data) {
+    std::u32string converted = U"[Encoding error]";
+    try {
+        converted = Common::UTF8ToUTF32(data);
+    } catch (std::range_error&) {
+    }
+
     std::string escaped;
     escaped.reserve(data.size());
 
-    for (char c : data) {
+    for (char32_t c : converted) {
         switch (c) {
         case '&':
             escaped += "&amp;";
@@ -86,7 +95,11 @@ static std::string EscapeXML(std::string_view data) {
             escaped += "&gt;";
             break;
         default:
-            escaped += c;
+            if (c > 0x7f) {
+                escaped += fmt::format("&#{};", static_cast<u32>(c));
+            } else {
+                escaped += static_cast<char>(c);
+            }
             break;
         }
     }

src/core/file_sys/program_metadata.cpp

@@ -5,6 +5,7 @@
 #include <vector>
 
 #include "common/logging/log.h"
+#include "common/scope_exit.h"
 #include "core/file_sys/program_metadata.h"
 #include "core/file_sys/vfs.h"
 #include "core/loader/loader.h"
@@ -95,6 +96,13 @@ Loader::ResultStatus ProgramMetadata::Load(VirtualFile file) {
     return Loader::ResultStatus::Success;
 }
 
+Loader::ResultStatus ProgramMetadata::Reload(VirtualFile file) {
+    const u64 original_program_id = aci_header.title_id;
+    SCOPE_EXIT({ aci_header.title_id = original_program_id; });
+
+    return this->Load(file);
+}
+
 /*static*/ ProgramMetadata ProgramMetadata::GetDefault() {
     // Allow use of cores 0~3 and thread priorities 1~63.
     constexpr u32 default_thread_info_capability = 0x30007F7;

src/core/file_sys/program_metadata.h

@@ -56,6 +56,7 @@ public:
     static ProgramMetadata GetDefault();
 
     Loader::ResultStatus Load(VirtualFile file);
+    Loader::ResultStatus Reload(VirtualFile file);
 
     /// Load from parameters instead of NPDM file, used for KIP
     void LoadManual(bool is_64_bit, ProgramAddressSpaceType address_space, s32 main_thread_prio,

src/core/loader/deconstructed_rom_directory.cpp

@@ -118,7 +118,7 @@ AppLoader_DeconstructedRomDirectory::LoadResult AppLoader_DeconstructedRomDirect
         return {ResultStatus::ErrorMissingNPDM, {}};
     }
 
-    const ResultStatus result2 = metadata.Load(npdm);
+    const ResultStatus result2 = metadata.Reload(npdm);
     if (result2 != ResultStatus::Success) {
         return {result2, {}};
     }

src/video_core/host_shaders/CMakeLists.txt

@@ -19,6 +19,7 @@ set(SHADER_FILES
     block_linear_unswizzle_2d.comp
     block_linear_unswizzle_3d.comp
     convert_abgr8_to_d24s8.frag
+    convert_d32f_to_abgr8.frag
     convert_d24s8_to_abgr8.frag
     convert_depth_to_float.frag
     convert_float_to_depth.frag

src/video_core/host_shaders/convert_d32f_to_abgr8.frag

@@ -0,0 +1,14 @@
+// SPDX-FileCopyrightText: Copyright 2021 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#version 450
+
+layout(binding = 0) uniform sampler2D depth_tex;
+
+layout(location = 0) out vec4 color;
+
+void main() {
+    ivec2 coord = ivec2(gl_FragCoord.xy);
+    float depth = textureLod(depth_tex, coord, 0).r;
+    color = vec4(depth, depth, depth, 1.0);
+}

src/video_core/renderer_base.h

@@ -89,9 +89,6 @@ public:
     void RequestScreenshot(void* data, std::function<void(bool)> callback,
                            const Layout::FramebufferLayout& layout);
 
-    /// This is called to notify the rendering backend of a surface change
-    virtual void NotifySurfaceChanged() {}
-
 protected:
     Core::Frontend::EmuWindow& render_window; ///< Reference to the render window handle.
     std::unique_ptr<Core::Frontend::GraphicsContext> context;

src/video_core/renderer_opengl/maxwell_to_gl.h

@@ -116,6 +116,7 @@ constexpr std::array<FormatTuple, VideoCore::Surface::MaxPixelFormat> FORMAT_TAB
     {GL_RGB9_E5, GL_RGB, GL_UNSIGNED_INT_5_9_9_9_REV},                // E5B9G9R9_FLOAT
     {GL_DEPTH_COMPONENT32F, GL_DEPTH_COMPONENT, GL_FLOAT},            // D32_FLOAT
     {GL_DEPTH_COMPONENT16, GL_DEPTH_COMPONENT, GL_UNSIGNED_SHORT},    // D16_UNORM
+    {GL_DEPTH_COMPONENT24, GL_DEPTH_COMPONENT, GL_UNSIGNED_INT_24_8}, // X8_D24_UNORM
     {GL_STENCIL_INDEX8, GL_STENCIL, GL_UNSIGNED_BYTE},                // S8_UINT
     {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8},    // D24_UNORM_S8_UINT
     {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8},    // S8_UINT_D24_UNORM

src/video_core/renderer_vulkan/blit_image.cpp

@@ -9,6 +9,7 @@
 #include "video_core/host_shaders/blit_color_float_frag_spv.h"
 #include "video_core/host_shaders/convert_abgr8_to_d24s8_frag_spv.h"
 #include "video_core/host_shaders/convert_d24s8_to_abgr8_frag_spv.h"
+#include "video_core/host_shaders/convert_d32f_to_abgr8_frag_spv.h"
 #include "video_core/host_shaders/convert_depth_to_float_frag_spv.h"
 #include "video_core/host_shaders/convert_float_to_depth_frag_spv.h"
 #include "video_core/host_shaders/convert_s8d24_to_abgr8_frag_spv.h"
@@ -433,6 +434,7 @@ BlitImageHelper::BlitImageHelper(const Device& device_, Scheduler& scheduler_,
       convert_depth_to_float_frag(BuildShader(device, CONVERT_DEPTH_TO_FLOAT_FRAG_SPV)),
       convert_float_to_depth_frag(BuildShader(device, CONVERT_FLOAT_TO_DEPTH_FRAG_SPV)),
       convert_abgr8_to_d24s8_frag(BuildShader(device, CONVERT_ABGR8_TO_D24S8_FRAG_SPV)),
+      convert_d32f_to_abgr8_frag(BuildShader(device, CONVERT_D32F_TO_ABGR8_FRAG_SPV)),
       convert_d24s8_to_abgr8_frag(BuildShader(device, CONVERT_D24S8_TO_ABGR8_FRAG_SPV)),
       convert_s8d24_to_abgr8_frag(BuildShader(device, CONVERT_S8D24_TO_ABGR8_FRAG_SPV)),
       linear_sampler(device.GetLogical().CreateSampler(SAMPLER_CREATE_INFO<VK_FILTER_LINEAR>)),
@@ -557,6 +559,13 @@ void BlitImageHelper::ConvertABGR8ToD24S8(const Framebuffer* dst_framebuffer,
     Convert(*convert_abgr8_to_d24s8_pipeline, dst_framebuffer, src_image_view);
 }
 
+void BlitImageHelper::ConvertD32FToABGR8(const Framebuffer* dst_framebuffer,
+                                         ImageView& src_image_view) {
+    ConvertPipelineColorTargetEx(convert_d32f_to_abgr8_pipeline, dst_framebuffer->RenderPass(),
+                                 convert_d32f_to_abgr8_frag);
+    ConvertDepthStencil(*convert_d32f_to_abgr8_pipeline, dst_framebuffer, src_image_view);
+}
+
 void BlitImageHelper::ConvertD24S8ToABGR8(const Framebuffer* dst_framebuffer,
                                           ImageView& src_image_view) {
     ConvertPipelineColorTargetEx(convert_d24s8_to_abgr8_pipeline, dst_framebuffer->RenderPass(),
@@ -609,6 +618,8 @@ void BlitImageHelper::ClearDepthStencil(const Framebuffer* dst_framebuffer, bool
     const VkPipelineLayout layout = *clear_color_pipeline_layout;
     scheduler.RequestRenderpass(dst_framebuffer);
     scheduler.Record([pipeline, layout, clear_depth, dst_region](vk::CommandBuffer cmdbuf) {
+        constexpr std::array blend_constants{0.0f, 0.0f, 0.0f, 0.0f};
+        cmdbuf.SetBlendConstants(blend_constants.data());
         cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, pipeline);
         BindBlitState(cmdbuf, dst_region);
         cmdbuf.PushConstants(layout, VK_SHADER_STAGE_FRAGMENT_BIT, clear_depth);
@@ -865,7 +876,7 @@ VkPipeline BlitImageHelper::FindOrEmplaceClearStencilPipeline(
         .sType = VK_STRUCTURE_TYPE_PIPELINE_DEPTH_STENCIL_STATE_CREATE_INFO,
         .pNext = nullptr,
         .flags = 0,
-        .depthTestEnable = VK_FALSE,
+        .depthTestEnable = key.depth_clear,
         .depthWriteEnable = key.depth_clear,
         .depthCompareOp = VK_COMPARE_OP_ALWAYS,
         .depthBoundsTestEnable = VK_FALSE,

src/video_core/renderer_vulkan/blit_image.h

@@ -67,6 +67,8 @@ public:
 
     void ConvertABGR8ToD24S8(const Framebuffer* dst_framebuffer, const ImageView& src_image_view);
 
+    void ConvertD32FToABGR8(const Framebuffer* dst_framebuffer, ImageView& src_image_view);
+
     void ConvertD24S8ToABGR8(const Framebuffer* dst_framebuffer, ImageView& src_image_view);
 
     void ConvertS8D24ToABGR8(const Framebuffer* dst_framebuffer, ImageView& src_image_view);
@@ -128,6 +130,7 @@ private:
     vk::ShaderModule convert_depth_to_float_frag;
     vk::ShaderModule convert_float_to_depth_frag;
     vk::ShaderModule convert_abgr8_to_d24s8_frag;
+    vk::ShaderModule convert_d32f_to_abgr8_frag;
     vk::ShaderModule convert_d24s8_to_abgr8_frag;
     vk::ShaderModule convert_s8d24_to_abgr8_frag;
     vk::Sampler linear_sampler;
@@ -146,6 +149,7 @@ private:
     vk::Pipeline convert_d16_to_r16_pipeline;
     vk::Pipeline convert_r16_to_d16_pipeline;
     vk::Pipeline convert_abgr8_to_d24s8_pipeline;
+    vk::Pipeline convert_d32f_to_abgr8_pipeline;
     vk::Pipeline convert_d24s8_to_abgr8_pipeline;
     vk::Pipeline convert_s8d24_to_abgr8_pipeline;
 };

src/video_core/renderer_vulkan/maxwell_to_vk.cpp

@@ -216,6 +216,7 @@ struct FormatTuple {
     // Depth formats
     {VK_FORMAT_D32_SFLOAT, Attachable},          // D32_FLOAT
     {VK_FORMAT_D16_UNORM, Attachable},           // D16_UNORM
+    {VK_FORMAT_X8_D24_UNORM_PACK32, Attachable}, // X8_D24_UNORM
 
     // Stencil formats
     {VK_FORMAT_S8_UINT, Attachable}, // S8_UINT

src/video_core/renderer_vulkan/renderer_vulkan.h

@@ -56,10 +56,6 @@ public:
         return device.GetDriverName();
     }
 
-    void NotifySurfaceChanged() override {
-        present_manager.NotifySurfaceChanged();
-    }
-
 private:
     void Report() const;
 

src/video_core/renderer_vulkan/vk_blit_screen.cpp

@@ -96,6 +96,7 @@ std::size_t GetSizeInBytes(const Tegra::FramebufferConfig& framebuffer) {
 VkFormat GetFormat(const Tegra::FramebufferConfig& framebuffer) {
     switch (framebuffer.pixel_format) {
     case Service::android::PixelFormat::Rgba8888:
+    case Service::android::PixelFormat::Rgbx8888:
         return VK_FORMAT_A8B8G8R8_UNORM_PACK32;
     case Service::android::PixelFormat::Rgb565:
         return VK_FORMAT_R5G6B5_UNORM_PACK16;

src/video_core/renderer_vulkan/vk_present_manager.cpp

@@ -103,8 +103,7 @@ PresentManager::PresentManager(const vk::Instance& instance_,
       surface{surface_}, blit_supported{CanBlitToSwapchain(device.GetPhysical(),
                                                            swapchain.GetImageViewFormat())},
       use_present_thread{Settings::values.async_presentation.GetValue()},
-      image_count{swapchain.GetImageCount()}, last_render_surface{
+      image_count{swapchain.GetImageCount()} {
-                                                  render_window_.GetWindowInfo().render_surface} {
 
     auto& dld = device.GetLogical();
     cmdpool = dld.CreateCommandPool({
@@ -289,44 +288,36 @@ void PresentManager::PresentThread(std::stop_token token) {
     }
 }
 
-void PresentManager::NotifySurfaceChanged() {
+void PresentManager::RecreateSwapchain(Frame* frame) {
-#ifdef ANDROID
+    swapchain.Create(*surface, frame->width, frame->height, frame->is_srgb);
-    std::scoped_lock lock{recreate_surface_mutex};
+    image_count = swapchain.GetImageCount();
-    recreate_surface_cv.notify_one();
-#endif
 }
 
 void PresentManager::CopyToSwapchain(Frame* frame) {
-    MICROPROFILE_SCOPE(Vulkan_CopyToSwapchain);
+    bool requires_recreation = false;
 
-    const auto recreate_swapchain = [&] {
+    while (true) {
-        swapchain.Create(*surface, frame->width, frame->height, frame->is_srgb);
+        try {
-        image_count = swapchain.GetImageCount();
+            // Recreate surface and swapchain if needed.
-    };
+            if (requires_recreation) {
-
-#ifdef ANDROID
-    std::unique_lock lock{recreate_surface_mutex};
-
-    const auto needs_recreation = [&] {
-        if (last_render_surface != render_window.GetWindowInfo().render_surface) {
-            return true;
-        }
-        if (swapchain.NeedsRecreation(frame->is_srgb)) {
-            return true;
-        }
-        return false;
-    };
-
-    recreate_surface_cv.wait_for(lock, std::chrono::milliseconds(400),
-                                 [&]() { return !needs_recreation(); });
-
-    // If the frontend recreated the surface, recreate the renderer surface and swapchain.
-    if (last_render_surface != render_window.GetWindowInfo().render_surface) {
-        last_render_surface = render_window.GetWindowInfo().render_surface;
                 surface = CreateSurface(instance, render_window.GetWindowInfo());
-        recreate_swapchain();
+                RecreateSwapchain(frame);
             }
-#endif
+
+            // Draw to swapchain.
+            return CopyToSwapchainImpl(frame);
+        } catch (const vk::Exception& except) {
+            if (except.GetResult() != VK_ERROR_SURFACE_LOST_KHR) {
+                throw;
+            }
+
+            requires_recreation = true;
+        }
+    }
+}
+
+void PresentManager::CopyToSwapchainImpl(Frame* frame) {
+    MICROPROFILE_SCOPE(Vulkan_CopyToSwapchain);
 
     // If the size or colorspace of the incoming frames has changed, recreate the swapchain
     // to account for that.
@@ -334,11 +325,11 @@ void PresentManager::CopyToSwapchain(Frame* frame) {
     const bool size_changed =
         swapchain.GetWidth() != frame->width || swapchain.GetHeight() != frame->height;
     if (srgb_changed || size_changed) {
-        recreate_swapchain();
+        RecreateSwapchain(frame);
     }
 
     while (swapchain.AcquireNextImage()) {
-        recreate_swapchain();
+        RecreateSwapchain(frame);
     }
 
     const vk::CommandBuffer cmdbuf{frame->cmdbuf};

src/video_core/renderer_vulkan/vk_present_manager.h

@@ -54,14 +54,15 @@ public:
     /// Waits for the present thread to finish presenting all queued frames.
     void WaitPresent();
 
-    /// This is called to notify the rendering backend of a surface change
-    void NotifySurfaceChanged();
-
 private:
     void PresentThread(std::stop_token token);
 
     void CopyToSwapchain(Frame* frame);
 
+    void CopyToSwapchainImpl(Frame* frame);
+
+    void RecreateSwapchain(Frame* frame);
+
 private:
     const vk::Instance& instance;
     Core::Frontend::EmuWindow& render_window;
@@ -76,16 +77,13 @@ private:
     std::queue<Frame*> free_queue;
     std::condition_variable_any frame_cv;
     std::condition_variable free_cv;
-    std::condition_variable recreate_surface_cv;
     std::mutex swapchain_mutex;
-    std::mutex recreate_surface_mutex;
     std::mutex queue_mutex;
     std::mutex free_mutex;
     std::jthread present_thread;
     bool blit_supported;
     bool use_present_thread;
     std::size_t image_count{};
-    void* last_render_surface{};
 };
 
 } // namespace Vulkan

src/video_core/renderer_vulkan/vk_rasterizer.cpp

@@ -422,7 +422,8 @@ void RasterizerVulkan::Clear(u32 layer_count) {
         return;
     }
 
-    if (use_stencil && regs.stencil_front_mask != 0xFF && regs.stencil_front_mask != 0) {
+    if (use_stencil && framebuffer->HasAspectStencilBit() && regs.stencil_front_mask != 0xFF &&
+        regs.stencil_front_mask != 0) {
         Region2D dst_region = {
             Offset2D{.x = clear_rect.rect.offset.x, .y = clear_rect.rect.offset.y},
             Offset2D{.x = clear_rect.rect.offset.x + static_cast<s32>(clear_rect.rect.extent.width),

src/video_core/renderer_vulkan/vk_staging_buffer_pool.cpp

@@ -24,25 +24,38 @@ using namespace Common::Literals;
 
 // Maximum potential alignment of a Vulkan buffer
 constexpr VkDeviceSize MAX_ALIGNMENT = 256;
-// Maximum size to put elements in the stream buffer
-constexpr VkDeviceSize MAX_STREAM_BUFFER_REQUEST_SIZE = 8_MiB;
 // Stream buffer size in bytes
-constexpr VkDeviceSize STREAM_BUFFER_SIZE = 128_MiB;
+constexpr VkDeviceSize MAX_STREAM_BUFFER_SIZE = 128_MiB;
-constexpr VkDeviceSize REGION_SIZE = STREAM_BUFFER_SIZE / StagingBufferPool::NUM_SYNCS;
 
-size_t Region(size_t iterator) noexcept {
+size_t GetStreamBufferSize(const Device& device) {
-    return iterator / REGION_SIZE;
+    VkDeviceSize size{0};
+    if (device.HasDebuggingToolAttached()) {
+        ForEachDeviceLocalHostVisibleHeap(device, [&size](size_t index, VkMemoryHeap& heap) {
+            size = std::max(size, heap.size);
+        });
+        // If rebar is not supported, cut the max heap size to 40%. This will allow 2 captures to be
+        // loaded at the same time in RenderDoc. If rebar is supported, this shouldn't be an issue
+        // as the heap will be much larger.
+        if (size <= 256_MiB) {
+            size = size * 40 / 100;
+        }
+    } else {
+        size = MAX_STREAM_BUFFER_SIZE;
+    }
+    return std::min(Common::AlignUp(size, MAX_ALIGNMENT), MAX_STREAM_BUFFER_SIZE);
 }
 } // Anonymous namespace
 
 StagingBufferPool::StagingBufferPool(const Device& device_, MemoryAllocator& memory_allocator_,
                                      Scheduler& scheduler_)
-    : device{device_}, memory_allocator{memory_allocator_}, scheduler{scheduler_} {
+    : device{device_}, memory_allocator{memory_allocator_}, scheduler{scheduler_},
+      stream_buffer_size{GetStreamBufferSize(device)}, region_size{stream_buffer_size /
+                                                                   StagingBufferPool::NUM_SYNCS} {
     VkBufferCreateInfo stream_ci = {
         .sType = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO,
         .pNext = nullptr,
         .flags = 0,
-        .size = STREAM_BUFFER_SIZE,
+        .size = stream_buffer_size,
         .usage = VK_BUFFER_USAGE_TRANSFER_SRC_BIT | VK_BUFFER_USAGE_UNIFORM_BUFFER_BIT |
                  VK_BUFFER_USAGE_INDEX_BUFFER_BIT | VK_BUFFER_USAGE_STORAGE_BUFFER_BIT,
         .sharingMode = VK_SHARING_MODE_EXCLUSIVE,
@@ -63,7 +76,7 @@ StagingBufferPool::StagingBufferPool(const Device& device_, MemoryAllocator& mem
 StagingBufferPool::~StagingBufferPool() = default;
 
 StagingBufferRef StagingBufferPool::Request(size_t size, MemoryUsage usage, bool deferred) {
-    if (!deferred && usage == MemoryUsage::Upload && size <= MAX_STREAM_BUFFER_REQUEST_SIZE) {
+    if (!deferred && usage == MemoryUsage::Upload && size <= region_size) {
         return GetStreamBuffer(size);
     }
     return GetStagingBuffer(size, usage, deferred);
@@ -101,7 +114,7 @@ StagingBufferRef StagingBufferPool::GetStreamBuffer(size_t size) {
     used_iterator = iterator;
     free_iterator = std::max(free_iterator, iterator + size);
 
-    if (iterator + size >= STREAM_BUFFER_SIZE) {
+    if (iterator + size >= stream_buffer_size) {
         std::fill(sync_ticks.begin() + Region(used_iterator), sync_ticks.begin() + NUM_SYNCS,
                   current_tick);
         used_iterator = 0;

src/video_core/renderer_vulkan/vk_staging_buffer_pool.h

@@ -90,6 +90,9 @@ private:
     void ReleaseCache(MemoryUsage usage);
 
     void ReleaseLevel(StagingBuffersCache& cache, size_t log2);
+    size_t Region(size_t iter) const noexcept {
+        return iter / region_size;
+    }
 
     const Device& device;
     MemoryAllocator& memory_allocator;
@@ -97,6 +100,8 @@ private:
 
     vk::Buffer stream_buffer;
     std::span<u8> stream_pointer;
+    VkDeviceSize stream_buffer_size;
+    VkDeviceSize region_size;
 
     size_t iterator = 0;
     size_t used_iterator = 0;

src/video_core/renderer_vulkan/vk_texture_cache.cpp

@@ -238,6 +238,7 @@ constexpr VkBorderColor ConvertBorderColor(const std::array<float, 4>& color) {
         return any_r ? VK_IMAGE_ASPECT_STENCIL_BIT : VK_IMAGE_ASPECT_DEPTH_BIT;
     case PixelFormat::D16_UNORM:
     case PixelFormat::D32_FLOAT:
+    case PixelFormat::X8_D24_UNORM:
         return VK_IMAGE_ASPECT_DEPTH_BIT;
     case PixelFormat::S8_UINT:
         return VK_IMAGE_ASPECT_STENCIL_BIT;
@@ -1200,6 +1201,9 @@ void TextureCacheRuntime::ConvertImage(Framebuffer* dst, ImageView& dst_view, Im
         if (src_view.format == PixelFormat::D24_UNORM_S8_UINT) {
             return blit_image_helper.ConvertS8D24ToABGR8(dst, src_view);
         }
+        if (src_view.format == PixelFormat::D32_FLOAT) {
+            return blit_image_helper.ConvertD32FToABGR8(dst, src_view);
+        }
         break;
     case PixelFormat::R32_FLOAT:
         if (src_view.format == PixelFormat::D32_FLOAT) {

src/video_core/surface.cpp

@@ -85,6 +85,8 @@ PixelFormat PixelFormatFromDepthFormat(Tegra::DepthFormat format) {
         return PixelFormat::S8_UINT;
     case Tegra::DepthFormat::Z32_FLOAT_X24S8_UINT:
         return PixelFormat::D32_FLOAT_S8_UINT;
+    case Tegra::DepthFormat::X8Z24_UNORM:
+        return PixelFormat::X8_D24_UNORM;
     default:
         UNIMPLEMENTED_MSG("Unimplemented format={}", format);
         return PixelFormat::S8_UINT_D24_UNORM;
@@ -202,6 +204,7 @@ PixelFormat PixelFormatFromRenderTargetFormat(Tegra::RenderTargetFormat format)
 PixelFormat PixelFormatFromGPUPixelFormat(Service::android::PixelFormat format) {
     switch (format) {
     case Service::android::PixelFormat::Rgba8888:
+    case Service::android::PixelFormat::Rgbx8888:
         return PixelFormat::A8B8G8R8_UNORM;
     case Service::android::PixelFormat::Rgb565:
         return PixelFormat::R5G6B5_UNORM;

src/video_core/surface.h

@@ -115,6 +115,7 @@ enum class PixelFormat {
     // Depth formats
     D32_FLOAT = MaxColorFormat,
     D16_UNORM,
+    X8_D24_UNORM,
 
     MaxDepthFormat,
 
@@ -251,6 +252,7 @@ constexpr std::array<u8, MaxPixelFormat> BLOCK_WIDTH_TABLE = {{
     1,  // E5B9G9R9_FLOAT
     1,  // D32_FLOAT
     1,  // D16_UNORM
+    1,  // X8_D24_UNORM
     1,  // S8_UINT
     1,  // D24_UNORM_S8_UINT
     1,  // S8_UINT_D24_UNORM
@@ -360,6 +362,7 @@ constexpr std::array<u8, MaxPixelFormat> BLOCK_HEIGHT_TABLE = {{
     1,  // E5B9G9R9_FLOAT
     1,  // D32_FLOAT
     1,  // D16_UNORM
+    1,  // X8_D24_UNORM
     1,  // S8_UINT
     1,  // D24_UNORM_S8_UINT
     1,  // S8_UINT_D24_UNORM
@@ -469,6 +472,7 @@ constexpr std::array<u8, MaxPixelFormat> BITS_PER_BLOCK_TABLE = {{
     32,  // E5B9G9R9_FLOAT
     32,  // D32_FLOAT
     16,  // D16_UNORM
+    32,  // X8_D24_UNORM
     8,   // S8_UINT
     32,  // D24_UNORM_S8_UINT
     32,  // S8_UINT_D24_UNORM

src/video_core/texture_cache/format_lookup_table.cpp

@@ -142,6 +142,10 @@ PixelFormat PixelFormatFromTextureInfo(TextureFormat format, ComponentType red,
         return PixelFormat::D16_UNORM;
     case Hash(TextureFormat::Z16, UNORM, UINT, UINT, UINT, LINEAR):
         return PixelFormat::D16_UNORM;
+    case Hash(TextureFormat::X8Z24, UNORM):
+        return PixelFormat::X8_D24_UNORM;
+    case Hash(TextureFormat::X8Z24, UNORM, UINT, UINT, UINT, LINEAR):
+        return PixelFormat::X8_D24_UNORM;
     case Hash(TextureFormat::Z24S8, UINT, UNORM, UNORM, UNORM, LINEAR):
         return PixelFormat::S8_UINT_D24_UNORM;
     case Hash(TextureFormat::Z24S8, UINT, UNORM, UINT, UINT, LINEAR):

src/video_core/texture_cache/formatter.h

@@ -211,6 +211,8 @@ struct fmt::formatter<VideoCore::Surface::PixelFormat> : fmt::formatter<fmt::str
                 return "D32_FLOAT";
             case PixelFormat::D16_UNORM:
                 return "D16_UNORM";
+            case PixelFormat::X8_D24_UNORM:
+                return "X8_D24_UNORM";
             case PixelFormat::S8_UINT:
                 return "S8_UINT";
             case PixelFormat::D24_UNORM_S8_UINT:

src/video_core/texture_cache/image_view_base.cpp

@@ -85,6 +85,7 @@ bool ImageViewBase::SupportsAnisotropy() const noexcept {
     // Depth formats
     case PixelFormat::D32_FLOAT:
     case PixelFormat::D16_UNORM:
+    case PixelFormat::X8_D24_UNORM:
     // Stencil formats
     case PixelFormat::S8_UINT:
     // DepthStencil formats

src/video_core/vulkan_common/vulkan_device.cpp

@@ -84,9 +84,12 @@ constexpr std::array VK_FORMAT_A4B4G4R4_UNORM_PACK16{
 } // namespace Alternatives
 
 enum class NvidiaArchitecture {
-    AmpereOrNewer,
+    KeplerOrOlder,
+    Maxwell,
+    Pascal,
+    Volta,
     Turing,
-    VoltaOrOlder,
+    AmpereOrNewer,
 };
 
 template <typename T>
@@ -200,6 +203,7 @@ std::unordered_map<VkFormat, VkFormatProperties> GetFormatProperties(vk::Physica
         VK_FORMAT_BC7_UNORM_BLOCK,
         VK_FORMAT_D16_UNORM,
         VK_FORMAT_D16_UNORM_S8_UINT,
+        VK_FORMAT_X8_D24_UNORM_PACK32,
         VK_FORMAT_D24_UNORM_S8_UINT,
         VK_FORMAT_D32_SFLOAT,
         VK_FORMAT_D32_SFLOAT_S8_UINT,
@@ -321,13 +325,38 @@ NvidiaArchitecture GetNvidiaArchitecture(vk::PhysicalDevice physical,
         physical.GetProperties2(physical_properties);
         if (shading_rate_props.primitiveFragmentShadingRateWithMultipleViewports) {
             // Only Ampere and newer support this feature
+            // TODO: Find a way to differentiate Ampere and Ada
             return NvidiaArchitecture::AmpereOrNewer;
         }
-    }
-    if (exts.contains(VK_NV_SHADING_RATE_IMAGE_EXTENSION_NAME)) {
         return NvidiaArchitecture::Turing;
     }
-    return NvidiaArchitecture::VoltaOrOlder;
+
+    if (exts.contains(VK_EXT_BLEND_OPERATION_ADVANCED_EXTENSION_NAME)) {
+        VkPhysicalDeviceBlendOperationAdvancedPropertiesEXT advanced_blending_props{};
+        advanced_blending_props.sType =
+            VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_BLEND_OPERATION_ADVANCED_PROPERTIES_EXT;
+        VkPhysicalDeviceProperties2 physical_properties{};
+        physical_properties.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_PROPERTIES_2;
+        physical_properties.pNext = &advanced_blending_props;
+        physical.GetProperties2(physical_properties);
+        if (advanced_blending_props.advancedBlendMaxColorAttachments == 1) {
+            return NvidiaArchitecture::Maxwell;
+        }
+
+        if (exts.contains(VK_EXT_CONSERVATIVE_RASTERIZATION_EXTENSION_NAME)) {
+            VkPhysicalDeviceConservativeRasterizationPropertiesEXT conservative_raster_props{};
+            conservative_raster_props.sType =
+                VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_CONSERVATIVE_RASTERIZATION_PROPERTIES_EXT;
+            physical_properties.pNext = &conservative_raster_props;
+            physical.GetProperties2(physical_properties);
+            if (conservative_raster_props.degenerateLinesRasterized) {
+                return NvidiaArchitecture::Volta;
+            }
+            return NvidiaArchitecture::Pascal;
+        }
+    }
+
+    return NvidiaArchitecture::KeplerOrOlder;
 }
 
 std::vector<const char*> ExtensionListForVulkan(
@@ -504,19 +533,14 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR
     if (is_nvidia) {
         const u32 nv_major_version = (properties.properties.driverVersion >> 22) & 0x3ff;
         const auto arch = GetNvidiaArchitecture(physical, supported_extensions);
-        switch (arch) {
+        if (arch >= NvidiaArchitecture::AmpereOrNewer) {
-        case NvidiaArchitecture::AmpereOrNewer:
             LOG_WARNING(Render_Vulkan, "Ampere and newer have broken float16 math");
             features.shader_float16_int8.shaderFloat16 = false;
-            break;
+        } else if (arch <= NvidiaArchitecture::Volta) {
-        case NvidiaArchitecture::Turing:
-            break;
-        case NvidiaArchitecture::VoltaOrOlder:
             if (nv_major_version < 527) {
                 LOG_WARNING(Render_Vulkan, "Volta and older have broken VK_KHR_push_descriptor");
                 RemoveExtension(extensions.push_descriptor, VK_KHR_PUSH_DESCRIPTOR_EXTENSION_NAME);
             }
-            break;
         }
         if (nv_major_version >= 510) {
             LOG_WARNING(Render_Vulkan, "NVIDIA Drivers >= 510 do not support MSAA image blits");
@@ -661,7 +685,15 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR
                         "ANV drivers 22.3.0 to 23.1.0 have broken VK_KHR_push_descriptor");
             RemoveExtension(extensions.push_descriptor, VK_KHR_PUSH_DESCRIPTOR_EXTENSION_NAME);
         }
+    } else if (extensions.push_descriptor && is_nvidia) {
+        const auto arch = GetNvidiaArchitecture(physical, supported_extensions);
+        if (arch <= NvidiaArchitecture::Pascal) {
+            LOG_WARNING(Render_Vulkan,
+                        "Pascal and older architectures have broken VK_KHR_push_descriptor");
+            RemoveExtension(extensions.push_descriptor, VK_KHR_PUSH_DESCRIPTOR_EXTENSION_NAME);
         }
+    }
+
     if (is_mvk) {
         LOG_WARNING(Render_Vulkan,
                     "MVK driver breaks when using more than 16 vertex attributes/bindings");

src/video_core/vulkan_common/vulkan_memory_allocator.cpp

@@ -9,6 +9,7 @@
 #include "common/alignment.h"
 #include "common/assert.h"
 #include "common/common_types.h"
+#include "common/literals.h"
 #include "common/logging/log.h"
 #include "common/polyfill_ranges.h"
 #include "video_core/vulkan_common/vma.h"
@@ -69,8 +70,7 @@ struct Range {
     case MemoryUsage::Download:
         return VMA_ALLOCATION_CREATE_HOST_ACCESS_RANDOM_BIT;
     case MemoryUsage::DeviceLocal:
-        return VMA_ALLOCATION_CREATE_HOST_ACCESS_SEQUENTIAL_WRITE_BIT |
+        return {};
-               VMA_ALLOCATION_CREATE_HOST_ACCESS_ALLOW_TRANSFER_INSTEAD_BIT;
     }
     return {};
 }
@@ -212,7 +212,20 @@ MemoryAllocator::MemoryAllocator(const Device& device_)
     : device{device_}, allocator{device.GetAllocator()},
       properties{device_.GetPhysical().GetMemoryProperties().memoryProperties},
       buffer_image_granularity{
-          device_.GetPhysical().GetProperties().limits.bufferImageGranularity} {}
+          device_.GetPhysical().GetProperties().limits.bufferImageGranularity} {
+    // GPUs not supporting rebar may only have a region with less than 256MB host visible/device
+    // local memory. In that case, opening 2 RenderDoc captures side-by-side is not possible due to
+    // the heap running out of memory. With RenderDoc attached and only a small host/device region,
+    // only allow the stream buffer in this memory heap.
+    if (device.HasDebuggingToolAttached()) {
+        using namespace Common::Literals;
+        ForEachDeviceLocalHostVisibleHeap(device, [this](size_t index, VkMemoryHeap& heap) {
+            if (heap.size <= 256_MiB) {
+                valid_memory_types &= ~(1u << index);
+            }
+        });
+    }
+}
 
 MemoryAllocator::~MemoryAllocator() = default;
 
@@ -244,7 +257,7 @@ vk::Buffer MemoryAllocator::CreateBuffer(const VkBufferCreateInfo& ci, MemoryUsa
         .usage = MemoryUsageVma(usage),
         .requiredFlags = 0,
         .preferredFlags = MemoryUsagePreferedVmaFlags(usage),
-        .memoryTypeBits = 0,
+        .memoryTypeBits = usage == MemoryUsage::Stream ? 0u : valid_memory_types,
         .pool = VK_NULL_HANDLE,
         .pUserData = nullptr,
         .priority = 0.f,

src/video_core/vulkan_common/vulkan_memory_allocator.h

@@ -7,6 +7,7 @@
 #include <span>
 #include <vector>
 #include "common/common_types.h"
+#include "video_core/vulkan_common/vulkan_device.h"
 #include "video_core/vulkan_common/vulkan_wrapper.h"
 
 VK_DEFINE_HANDLE(VmaAllocator)
@@ -26,6 +27,18 @@ enum class MemoryUsage {
     Stream,      ///< Requests device local host visible buffer, falling back host memory.
 };
 
+template <typename F>
+void ForEachDeviceLocalHostVisibleHeap(const Device& device, F&& f) {
+    auto memory_props = device.GetPhysical().GetMemoryProperties().memoryProperties;
+    for (size_t i = 0; i < memory_props.memoryTypeCount; i++) {
+        auto& memory_type = memory_props.memoryTypes[i];
+        if ((memory_type.propertyFlags & VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT) &&
+            (memory_type.propertyFlags & VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT)) {
+            f(memory_type.heapIndex, memory_props.memoryHeaps[memory_type.heapIndex]);
+        }
+    }
+}
+
 /// Ownership handle of a memory commitment.
 /// Points to a subregion of a memory allocation.
 class MemoryCommit {
@@ -124,6 +137,7 @@ private:
     std::vector<std::unique_ptr<MemoryAllocation>> allocations; ///< Current allocations.
     VkDeviceSize buffer_image_granularity; // The granularity for adjacent offsets between buffers
                                            // and optimal images
+    u32 valid_memory_types{~0u};
 };
 
 } // namespace Vulkan

src/video_core/vulkan_common/vulkan_wrapper.h

@@ -117,6 +117,9 @@ public:
     virtual ~Exception() = default;
 
     const char* what() const noexcept override;
+    VkResult GetResult() const noexcept {
+        return result;
+    }
 
 private:
     VkResult result;

src/yuzu/configuration/configure_audio.cpp

@@ -42,6 +42,9 @@ void ConfigureAudio::Setup(const ConfigurationShared::Builder& builder) {
         for (auto* setting : Settings::values.linkage.by_category[category]) {
             settings.push_back(setting);
         }
+        for (auto* setting : UISettings::values.linkage.by_category[category]) {
+            settings.push_back(setting);
+        }
     };
 
     push(Settings::Category::Audio);

src/yuzu/configuration/shared_translation.cpp

@@ -29,9 +29,10 @@ std::unique_ptr<TranslationMap> InitializeTranslations(QWidget* parent) {
     INSERT(Settings, sink_id, "Output Engine:", "");
     INSERT(Settings, audio_output_device_id, "Output Device:", "");
     INSERT(Settings, audio_input_device_id, "Input Device:", "");
-    INSERT(Settings, audio_muted, "Mute audio when in background", "");
+    INSERT(Settings, audio_muted, "Mute audio", "");
     INSERT(Settings, volume, "Volume:", "");
     INSERT(Settings, dump_audio_commands, "", "");
+    INSERT(UISettings, mute_when_in_background, "Mute audio when in background", "");
 
     // Core
     INSERT(Settings, use_multi_core, "Multicore CPU Emulation", "");

src/yuzu/main.cpp

@@ -1447,6 +1447,7 @@ void GMainWindow::OnAppFocusStateChanged(Qt::ApplicationState state) {
             Settings::values.audio_muted = false;
             auto_muted = false;
         }
+        UpdateVolumeUI();
     }
 }
 

src/yuzu/uisettings.h

@@ -103,7 +103,7 @@ struct Values {
                                            true,
                                            true};
     Setting<bool> mute_when_in_background{
-        linkage, false, "muteWhenInBackground", Category::Ui, Settings::Specialization::Default,
+        linkage, false, "muteWhenInBackground", Category::Audio, Settings::Specialization::Default,
         true,    true};
     Setting<bool> hide_mouse{
         linkage, true, "hideInactiveMouse", Category::UiGeneral, Settings::Specialization::Default,