HID: Update controllers less often

gl_rasterizer_cache: Implement texture format G8R8.

src/core/hle/kernel/hle_ipc.cpp

@@ -214,8 +214,8 @@ ResultCode HLERequestContext::WriteToOutgoingCommandBuffer(Thread& thread) {
             (sizeof(IPC::CommandHeader) + sizeof(IPC::HandleDescriptorHeader)) / sizeof(u32);
         ASSERT_MSG(!handle_descriptor_header->send_current_pid, "Sending PID is not implemented");
 
-        ASSERT_MSG(copy_objects.size() == handle_descriptor_header->num_handles_to_copy);
-        ASSERT_MSG(move_objects.size() == handle_descriptor_header->num_handles_to_move);
+        ASSERT(copy_objects.size() == handle_descriptor_header->num_handles_to_copy);
+        ASSERT(move_objects.size() == handle_descriptor_header->num_handles_to_move);
 
         // We don't make a distinction between copy and move handles when translating since HLE
         // services don't deal with handles directly. However, the guest applications might check

src/core/hle/service/hid/hid.cpp

@@ -18,9 +18,9 @@ namespace Service::HID {
 
 // Updating period for each HID device.
 // TODO(shinyquagsire23): These need better values.
-constexpr u64 pad_update_ticks = CoreTiming::BASE_CLOCK_RATE / 10000;
-constexpr u64 accelerometer_update_ticks = CoreTiming::BASE_CLOCK_RATE / 10000;
-constexpr u64 gyroscope_update_ticks = CoreTiming::BASE_CLOCK_RATE / 10000;
+constexpr u64 pad_update_ticks = CoreTiming::BASE_CLOCK_RATE / 100;
+constexpr u64 accelerometer_update_ticks = CoreTiming::BASE_CLOCK_RATE / 100;
+constexpr u64 gyroscope_update_ticks = CoreTiming::BASE_CLOCK_RATE / 100;
 
 class IAppletResource final : public ServiceFramework<IAppletResource> {
 public:

src/core/hle/service/sockets/bsd.cpp

@@ -19,10 +19,9 @@ void BSD::RegisterClient(Kernel::HLERequestContext& ctx) {
 void BSD::StartMonitoring(Kernel::HLERequestContext& ctx) {
     LOG_WARNING(Service, "(STUBBED) called");
 
-    IPC::ResponseBuilder rb{ctx, 3};
+    IPC::ResponseBuilder rb{ctx, 2};
 
     rb.Push(RESULT_SUCCESS);
-    rb.Push<u32>(0); // bsd errno
 }
 
 void BSD::Socket(Kernel::HLERequestContext& ctx) {

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -15,6 +15,7 @@
 #include "common/microprofile.h"
 #include "common/scope_exit.h"
 #include "core/core.h"
+#include "core/frontend/emu_window.h"
 #include "core/hle/kernel/process.h"
 #include "core/settings.h"
 #include "video_core/engines/maxwell_3d.h"
@@ -22,6 +23,7 @@
 #include "video_core/renderer_opengl/gl_shader_gen.h"
 #include "video_core/renderer_opengl/maxwell_to_gl.h"
 #include "video_core/renderer_opengl/renderer_opengl.h"
+#include "video_core/video_core.h"
 
 using Maxwell = Tegra::Engines::Maxwell3D::Regs;
 using PixelFormat = SurfaceParams::PixelFormat;
@@ -398,6 +400,8 @@ void RasterizerOpenGL::Clear() {
     if (clear_mask == 0)
         return;
 
+    ScopeAcquireGLContext acquire_context;
+
     auto [dirty_color_surface, dirty_depth_surface] =
         ConfigureFramebuffers(use_color_fb, use_depth_fb);
 
@@ -424,6 +428,8 @@ void RasterizerOpenGL::DrawArrays() {
     MICROPROFILE_SCOPE(OpenGL_Drawing);
     const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
 
+    ScopeAcquireGLContext acquire_context;
+
     auto [dirty_color_surface, dirty_depth_surface] =
         ConfigureFramebuffers(true, regs.zeta.Address() != 0);
 

src/video_core/renderer_opengl/gl_rasterizer_cache.cpp

@@ -105,6 +105,7 @@ static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_form
     {GL_COMPRESSED_RGBA_BPTC_UNORM_ARB, GL_RGB, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
      true},                                                             // BC7U
     {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_4X4
+    {GL_RG8, GL_RG, GL_UNSIGNED_BYTE, ComponentType::UNorm, false},     // G8R8
 
     // DepthStencil formats
     {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8, ComponentType::UNorm,
@@ -112,6 +113,8 @@ static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_form
     {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8, ComponentType::UNorm,
      false},                                                                            // S8Z24
     {GL_DEPTH_COMPONENT32F, GL_DEPTH_COMPONENT, GL_FLOAT, ComponentType::Float, false}, // Z32F
+    {GL_DEPTH_COMPONENT16, GL_DEPTH_COMPONENT, GL_UNSIGNED_SHORT, ComponentType::UNorm,
+     false}, // Z16
 }};
 
 static const FormatTuple& GetFormatTuple(PixelFormat pixel_format, ComponentType component_type) {
@@ -194,8 +197,9 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr),
         MortonCopy<true, PixelFormat::DXT1>,         MortonCopy<true, PixelFormat::DXT23>,
         MortonCopy<true, PixelFormat::DXT45>,        MortonCopy<true, PixelFormat::DXN1>,
         MortonCopy<true, PixelFormat::BC7U>,         MortonCopy<true, PixelFormat::ASTC_2D_4X4>,
-        MortonCopy<true, PixelFormat::Z24S8>,        MortonCopy<true, PixelFormat::S8Z24>,
-        MortonCopy<true, PixelFormat::Z32F>,
+        MortonCopy<true, PixelFormat::G8R8>,         MortonCopy<true, PixelFormat::Z24S8>,
+        MortonCopy<true, PixelFormat::S8Z24>,        MortonCopy<true, PixelFormat::Z32F>,
+        MortonCopy<true, PixelFormat::Z16>,
 };
 
 static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr),
@@ -215,10 +219,12 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr),
         nullptr,
         nullptr,
         nullptr,
-        MortonCopy<false, PixelFormat::ABGR8>,
+        nullptr,
+        MortonCopy<false, PixelFormat::G8R8>,
         MortonCopy<false, PixelFormat::Z24S8>,
         MortonCopy<false, PixelFormat::S8Z24>,
         MortonCopy<false, PixelFormat::Z32F>,
+        MortonCopy<false, PixelFormat::Z16>,
 };
 
 // Allocate an uninitialized texture of appropriate size and format for the surface
@@ -271,9 +277,10 @@ static void ConvertS8Z24ToZ24S8(std::vector<u8>& data, u32 width, u32 height) {
 
     S8Z24 input_pixel{};
     Z24S8 output_pixel{};
+    const auto bpp{CachedSurface::GetGLBytesPerPixel(PixelFormat::S8Z24)};
     for (size_t y = 0; y < height; ++y) {
         for (size_t x = 0; x < width; ++x) {
-            const size_t offset{y * width + x};
+            const size_t offset{bpp * (y * width + x)};
             std::memcpy(&input_pixel, &data[offset], sizeof(S8Z24));
             output_pixel.s8.Assign(input_pixel.s8);
             output_pixel.z24.Assign(input_pixel.z24);
@@ -281,6 +288,19 @@ static void ConvertS8Z24ToZ24S8(std::vector<u8>& data, u32 width, u32 height) {
         }
     }
 }
+
+static void ConvertG8R8ToR8G8(std::vector<u8>& data, u32 width, u32 height) {
+    const auto bpp{CachedSurface::GetGLBytesPerPixel(PixelFormat::G8R8)};
+    for (size_t y = 0; y < height; ++y) {
+        for (size_t x = 0; x < width; ++x) {
+            const size_t offset{bpp * (y * width + x)};
+            const u8 temp{data[offset]};
+            data[offset] = data[offset + 1];
+            data[offset + 1] = temp;
+        }
+    }
+}
+
 /**
  * Helper function to perform software conversion (as needed) when loading a buffer from Switch
  * memory. This is for Maxwell pixel formats that cannot be represented as-is in OpenGL or with
@@ -301,6 +321,11 @@ static void ConvertFormatAsNeeded_LoadGLBuffer(std::vector<u8>& data, PixelForma
         // Convert the S8Z24 depth format to Z24S8, as OpenGL does not support S8Z24.
         ConvertS8Z24ToZ24S8(data, width, height);
         break;
+
+    case PixelFormat::G8R8:
+        // Convert the G8R8 color format to R8G8, as OpenGL does not support G8R8.
+        ConvertG8R8ToR8G8(data, width, height);
+        break;
     }
 }
 

src/video_core/renderer_opengl/gl_rasterizer_cache.h

@@ -37,13 +37,15 @@ struct SurfaceParams {
         DXN1 = 11, // This is also known as BC4
         BC7U = 12,
         ASTC_2D_4X4 = 13,
+        G8R8 = 14,
 
         MaxColorFormat,
 
         // DepthStencil formats
-        Z24S8 = 14,
-        S8Z24 = 15,
-        Z32F = 16,
+        Z24S8 = 15,
+        S8Z24 = 16,
+        Z32F = 17,
+        Z16 = 18,
 
         MaxDepthStencilFormat,
 
@@ -95,9 +97,11 @@ struct SurfaceParams {
             4, // DXN1
             4, // BC7U
             4, // ASTC_2D_4X4
+            1, // G8R8
             1, // Z24S8
             1, // S8Z24
             1, // Z32F
+            1, // Z16
         }};
 
         ASSERT(static_cast<size_t>(format) < compression_factor_table.size());
@@ -123,9 +127,11 @@ struct SurfaceParams {
             64,  // DXN1
             128, // BC7U
             32,  // ASTC_2D_4X4
+            16,  // G8R8
             32,  // Z24S8
             32,  // S8Z24
             32,  // Z32F
+            16,  // Z16
         }};
 
         ASSERT(static_cast<size_t>(format) < bpp_table.size());
@@ -143,6 +149,8 @@ struct SurfaceParams {
             return PixelFormat::Z24S8;
         case Tegra::DepthFormat::Z32_FLOAT:
             return PixelFormat::Z32F;
+        case Tegra::DepthFormat::Z16_UNORM:
+            return PixelFormat::Z16;
         default:
             LOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
             UNREACHABLE();
@@ -181,6 +189,8 @@ struct SurfaceParams {
             return PixelFormat::A1B5G5R5;
         case Tegra::Texture::TextureFormat::R8:
             return PixelFormat::R8;
+        case Tegra::Texture::TextureFormat::G8R8:
+            return PixelFormat::G8R8;
         case Tegra::Texture::TextureFormat::R16_G16_B16_A16:
             return PixelFormat::RGBA16F;
         case Tegra::Texture::TextureFormat::BF10GF11RF11:
@@ -218,6 +228,8 @@ struct SurfaceParams {
             return Tegra::Texture::TextureFormat::A1B5G5R5;
         case PixelFormat::R8:
             return Tegra::Texture::TextureFormat::R8;
+        case PixelFormat::G8R8:
+            return Tegra::Texture::TextureFormat::G8R8;
         case PixelFormat::RGBA16F:
             return Tegra::Texture::TextureFormat::R16_G16_B16_A16;
         case PixelFormat::R11FG11FB10F:
@@ -249,6 +261,8 @@ struct SurfaceParams {
             return Tegra::DepthFormat::Z24_S8_UNORM;
         case PixelFormat::Z32F:
             return Tegra::DepthFormat::Z32_FLOAT;
+        case PixelFormat::Z16:
+            return Tegra::DepthFormat::Z16_UNORM;
         default:
             UNREACHABLE();
         }
@@ -295,6 +309,7 @@ struct SurfaceParams {
 
     static ComponentType ComponentTypeFromDepthFormat(Tegra::DepthFormat format) {
         switch (format) {
+        case Tegra::DepthFormat::Z16_UNORM:
         case Tegra::DepthFormat::S8_Z24_UNORM:
         case Tegra::DepthFormat::Z24_S8_UNORM:
             return ComponentType::UNorm;

src/video_core/renderer_opengl/renderer_opengl.cpp

@@ -92,11 +92,24 @@ static std::array<GLfloat, 3 * 2> MakeOrthographicMatrix(const float width, cons
     return matrix;
 }
 
+ScopeAcquireGLContext::ScopeAcquireGLContext() {
+    if (Settings::values.use_multi_core) {
+        VideoCore::g_emu_window->MakeCurrent();
+    }
+}
+ScopeAcquireGLContext::~ScopeAcquireGLContext() {
+    if (Settings::values.use_multi_core) {
+        VideoCore::g_emu_window->DoneCurrent();
+    }
+}
+
 RendererOpenGL::RendererOpenGL() = default;
 RendererOpenGL::~RendererOpenGL() = default;
 
 /// Swap buffers (render frame)
 void RendererOpenGL::SwapBuffers(boost::optional<const Tegra::FramebufferConfig&> framebuffer) {
+    ScopeAcquireGLContext acquire_context;
+
     Core::System::GetInstance().perf_stats.EndSystemFrame();
 
     // Maintain the rasterizer's state as a priority
@@ -418,7 +431,7 @@ static void APIENTRY DebugHandler(GLenum source, GLenum type, GLuint id, GLenum
 
 /// Initialize the renderer
 bool RendererOpenGL::Init() {
-    render_window->MakeCurrent();
+    ScopeAcquireGLContext acquire_context;
 
     if (GLAD_GL_KHR_debug) {
         glEnable(GL_DEBUG_OUTPUT);

src/video_core/renderer_opengl/renderer_opengl.h

@@ -31,6 +31,13 @@ struct ScreenInfo {
     TextureInfo texture;
 };
 
+/// Helper class to acquire/release OpenGL context within a given scope
+class ScopeAcquireGLContext : NonCopyable {
+public:
+    ScopeAcquireGLContext();
+    ~ScopeAcquireGLContext();
+};
+
 class RendererOpenGL : public RendererBase {
 public:
     RendererOpenGL();

src/video_core/textures/decoders.cpp

@@ -62,6 +62,7 @@ u32 BytesPerPixel(TextureFormat format) {
         return 4;
     case TextureFormat::A1B5G5R5:
     case TextureFormat::B5G6R5:
+    case TextureFormat::G8R8:
         return 2;
     case TextureFormat::R8:
         return 1;
@@ -77,6 +78,8 @@ u32 BytesPerPixel(TextureFormat format) {
 
 static u32 DepthBytesPerPixel(DepthFormat format) {
     switch (format) {
+    case DepthFormat::Z16_UNORM:
+        return 2;
     case DepthFormat::S8_Z24_UNORM:
     case DepthFormat::Z24_S8_UNORM:
     case DepthFormat::Z32_FLOAT:
@@ -110,6 +113,7 @@ std::vector<u8> UnswizzleTexture(VAddr address, TextureFormat format, u32 width,
     case TextureFormat::A1B5G5R5:
     case TextureFormat::B5G6R5:
     case TextureFormat::R8:
+    case TextureFormat::G8R8:
     case TextureFormat::R16_G16_B16_A16:
     case TextureFormat::R32_G32_B32_A32:
     case TextureFormat::BF10GF11RF11:
@@ -133,6 +137,7 @@ std::vector<u8> UnswizzleDepthTexture(VAddr address, DepthFormat format, u32 wid
     std::vector<u8> unswizzled_data(width * height * bytes_per_pixel);
 
     switch (format) {
+    case DepthFormat::Z16_UNORM:
     case DepthFormat::S8_Z24_UNORM:
     case DepthFormat::Z24_S8_UNORM:
     case DepthFormat::Z32_FLOAT:
@@ -164,6 +169,7 @@ std::vector<u8> DecodeTexture(const std::vector<u8>& texture_data, TextureFormat
     case TextureFormat::A1B5G5R5:
     case TextureFormat::B5G6R5:
     case TextureFormat::R8:
+    case TextureFormat::G8R8:
     case TextureFormat::BF10GF11RF11:
     case TextureFormat::R32_G32_B32_A32:
         // TODO(Subv): For the time being just forward the same data without any decoding.

src/yuzu/bootmanager.cpp

@@ -20,7 +20,10 @@
 EmuThread::EmuThread(GRenderWindow* render_window) : render_window(render_window) {}
 
 void EmuThread::run() {
-    render_window->MakeCurrent();
+    if (!Settings::values.use_multi_core) {
+        // Single core mode must acquire OpenGL context for entire emulation session
+        render_window->MakeCurrent();
+    }
 
     MicroProfileOnThreadCreate("EmuThread");
 

src/yuzu/main.cpp

@@ -374,6 +374,8 @@ bool GMainWindow::LoadROM(const QString& filename) {
 
     const Core::System::ResultStatus result{system.Load(render_window, filename.toStdString())};
 
+    render_window->DoneCurrent();
+
     if (result != Core::System::ResultStatus::Success) {
         switch (result) {
         case Core::System::ResultStatus::ErrorGetLoader:
@@ -916,6 +918,7 @@ int main(int argc, char* argv[]) {
     QCoreApplication::setApplicationName("yuzu");
 
     QApplication::setAttribute(Qt::AA_X11InitThreads);
+    QApplication::setAttribute(Qt::AA_DontCheckOpenGLContextThreadAffinity);
     QApplication app(argc, argv);
 
     // Qt changes the locale and causes issues in float conversion using std::to_string() when

src/yuzu_cmd/yuzu.cpp

@@ -148,6 +148,11 @@ int main(int argc, char** argv) {
 
     std::unique_ptr<EmuWindow_SDL2> emu_window{std::make_unique<EmuWindow_SDL2>(fullscreen)};
 
+    if (!Settings::values.use_multi_core) {
+        // Single core mode must acquire OpenGL context for entire emulation session
+        emu_window->MakeCurrent();
+    }
+
     Core::System& system{Core::System::GetInstance()};
 
     SCOPE_EXIT({ system.Shutdown(); });