BGR5A1_UNORM: return A1B5G5R5U pixelformat

kernel/thread: Include thread-related enums within the kernel namespace

.gitmodules

@@ -31,3 +31,6 @@
 [submodule "opus"]
     path = externals/opus
     url = https://github.com/ogniK5377/opus.git
+[submodule "soundtouch"]
+	path = externals/soundtouch
+	url = https://github.com/citra-emu/ext-soundtouch.git

CMakeLists.txt

@@ -431,6 +431,9 @@ enable_testing()
 add_subdirectory(externals)
 add_subdirectory(src)
 
+# Set yuzu project as default StartUp Project in Visual Studio
+set_property(DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR} PROPERTY VS_STARTUP_PROJECT yuzu)
+
 
 # Installation instructions
 # =========================

externals/CMakeLists.txt

@@ -50,6 +50,9 @@ add_subdirectory(open_source_archives EXCLUDE_FROM_ALL)
 add_library(unicorn-headers INTERFACE)
 target_include_directories(unicorn-headers INTERFACE ./unicorn/include)
 
+# SoundTouch
+add_subdirectory(soundtouch)
+
 # Xbyak
 if (ARCHITECTURE_x86_64)
     # Defined before "dynarmic" above

src/audio_core/CMakeLists.txt

@@ -17,6 +17,8 @@ add_library(audio_core STATIC
     sink_stream.h
     stream.cpp
     stream.h
+    time_stretch.cpp
+    time_stretch.h
 
     $<$<BOOL:${ENABLE_CUBEB}>:cubeb_sink.cpp cubeb_sink.h>
 )
@@ -24,6 +26,7 @@ add_library(audio_core STATIC
 create_target_directory_groups(audio_core)
 
 target_link_libraries(audio_core PUBLIC common core)
+target_link_libraries(audio_core PRIVATE SoundTouch)
 
 if(ENABLE_CUBEB)
     target_link_libraries(audio_core PRIVATE cubeb)

src/audio_core/cubeb_sink.cpp

@@ -3,27 +3,23 @@
 // Refer to the license.txt file included.
 
 #include <algorithm>
+#include <atomic>
 #include <cstring>
-#include <mutex>
-
 #include "audio_core/cubeb_sink.h"
 #include "audio_core/stream.h"
+#include "audio_core/time_stretch.h"
 #include "common/logging/log.h"
+#include "common/ring_buffer.h"
+#include "core/settings.h"
 
 namespace AudioCore {
 
-class SinkStreamImpl final : public SinkStream {
+class CubebSinkStream final : public SinkStream {
 public:
-    SinkStreamImpl(cubeb* ctx, u32 sample_rate, u32 num_channels_, cubeb_devid output_device,
-                   const std::string& name)
-        : ctx{ctx}, num_channels{num_channels_} {
-
-        if (num_channels == 6) {
-            // 6-channel audio does not seem to work with cubeb + SDL, so we downsample this to 2
-            // channel for now
-            is_6_channel = true;
-            num_channels = 2;
-        }
+    CubebSinkStream(cubeb* ctx, u32 sample_rate, u32 num_channels_, cubeb_devid output_device,
+                    const std::string& name)
+        : ctx{ctx}, num_channels{std::min(num_channels_, 2u)}, time_stretch{sample_rate,
+                                                                            num_channels} {
 
         cubeb_stream_params params{};
         params.rate = sample_rate;
@@ -38,7 +34,7 @@ public:
 
         if (cubeb_stream_init(ctx, &stream_backend, name.c_str(), nullptr, nullptr, output_device,
                               &params, std::max(512u, minimum_latency),
-                              &SinkStreamImpl::DataCallback, &SinkStreamImpl::StateCallback,
+                              &CubebSinkStream::DataCallback, &CubebSinkStream::StateCallback,
                               this) != CUBEB_OK) {
             LOG_CRITICAL(Audio_Sink, "Error initializing cubeb stream");
             return;
@@ -50,7 +46,7 @@ public:
         }
     }
 
-    ~SinkStreamImpl() {
+    ~CubebSinkStream() {
         if (!ctx) {
             return;
         }
@@ -62,27 +58,32 @@ public:
         cubeb_stream_destroy(stream_backend);
     }
 
-    void EnqueueSamples(u32 num_channels, const std::vector<s16>& samples) override {
-        if (!ctx) {
+    void EnqueueSamples(u32 source_num_channels, const std::vector<s16>& samples) override {
+        if (source_num_channels > num_channels) {
+            // Downsample 6 channels to 2
+            std::vector<s16> buf;
+            buf.reserve(samples.size() * num_channels / source_num_channels);
+            for (size_t i = 0; i < samples.size(); i += source_num_channels) {
+                for (size_t ch = 0; ch < num_channels; ch++) {
+                    buf.push_back(samples[i + ch]);
+                }
+            }
+            queue.Push(buf);
             return;
         }
 
-        std::lock_guard lock{queue_mutex};
+        queue.Push(samples);
+    }
 
-        queue.reserve(queue.size() + samples.size() * GetNumChannels());
+    size_t SamplesInQueue(u32 num_channels) const override {
+        if (!ctx)
+            return 0;
 
-        if (is_6_channel) {
-            // Downsample 6 channels to 2
-            const size_t sample_count_copy_size = samples.size() * 2;
-            queue.reserve(sample_count_copy_size);
-            for (size_t i = 0; i < samples.size(); i += num_channels) {
-                queue.push_back(samples[i]);
-                queue.push_back(samples[i + 1]);
-            }
-        } else {
-            // Copy as-is
-            std::copy(samples.begin(), samples.end(), std::back_inserter(queue));
-        }
+        return queue.Size() / num_channels;
+    }
+
+    void Flush() override {
+        should_flush = true;
     }
 
     u32 GetNumChannels() const {
@@ -95,10 +96,11 @@ private:
     cubeb* ctx{};
     cubeb_stream* stream_backend{};
     u32 num_channels{};
-    bool is_6_channel{};
 
-    std::mutex queue_mutex;
-    std::vector<s16> queue;
+    Common::RingBuffer<s16, 0x10000> queue;
+    std::array<s16, 2> last_frame;
+    std::atomic<bool> should_flush{};
+    TimeStretcher time_stretch;
 
     static long DataCallback(cubeb_stream* stream, void* user_data, const void* input_buffer,
                              void* output_buffer, long num_frames);
@@ -144,38 +146,52 @@ CubebSink::~CubebSink() {
 SinkStream& CubebSink::AcquireSinkStream(u32 sample_rate, u32 num_channels,
                                          const std::string& name) {
     sink_streams.push_back(
-        std::make_unique<SinkStreamImpl>(ctx, sample_rate, num_channels, output_device, name));
+        std::make_unique<CubebSinkStream>(ctx, sample_rate, num_channels, output_device, name));
     return *sink_streams.back();
 }
 
-long SinkStreamImpl::DataCallback(cubeb_stream* stream, void* user_data, const void* input_buffer,
-                                  void* output_buffer, long num_frames) {
-    SinkStreamImpl* impl = static_cast<SinkStreamImpl*>(user_data);
+long CubebSinkStream::DataCallback(cubeb_stream* stream, void* user_data, const void* input_buffer,
+                                   void* output_buffer, long num_frames) {
+    CubebSinkStream* impl = static_cast<CubebSinkStream*>(user_data);
     u8* buffer = reinterpret_cast<u8*>(output_buffer);
 
     if (!impl) {
         return {};
     }
 
-    std::lock_guard lock{impl->queue_mutex};
+    const size_t num_channels = impl->GetNumChannels();
+    const size_t samples_to_write = num_channels * num_frames;
+    size_t samples_written;
 
-    const size_t frames_to_write{
-        std::min(impl->queue.size() / impl->GetNumChannels(), static_cast<size_t>(num_frames))};
+    if (Settings::values.enable_audio_stretching) {
+        const std::vector<s16> in{impl->queue.Pop()};
+        const size_t num_in{in.size() / num_channels};
+        s16* const out{reinterpret_cast<s16*>(buffer)};
+        const size_t out_frames = impl->time_stretch.Process(in.data(), num_in, out, num_frames);
+        samples_written = out_frames * num_channels;
 
-    memcpy(buffer, impl->queue.data(), frames_to_write * sizeof(s16) * impl->GetNumChannels());
-    impl->queue.erase(impl->queue.begin(),
-                      impl->queue.begin() + frames_to_write * impl->GetNumChannels());
+        if (impl->should_flush) {
+            impl->time_stretch.Flush();
+            impl->should_flush = false;
+        }
+    } else {
+        samples_written = impl->queue.Pop(buffer, samples_to_write);
+    }
 
-    if (frames_to_write < num_frames) {
-        // Fill the rest of the frames with silence
-        memset(buffer + frames_to_write * sizeof(s16) * impl->GetNumChannels(), 0,
-               (num_frames - frames_to_write) * sizeof(s16) * impl->GetNumChannels());
+    if (samples_written >= num_channels) {
+        std::memcpy(&impl->last_frame[0], buffer + (samples_written - num_channels) * sizeof(s16),
+                    num_channels * sizeof(s16));
+    }
+
+    // Fill the rest of the frames with last_frame
+    for (size_t i = samples_written; i < samples_to_write; i += num_channels) {
+        std::memcpy(buffer + i * sizeof(s16), &impl->last_frame[0], num_channels * sizeof(s16));
     }
 
     return num_frames;
 }
 
-void SinkStreamImpl::StateCallback(cubeb_stream* stream, void* user_data, cubeb_state state) {}
+void CubebSinkStream::StateCallback(cubeb_stream* stream, void* user_data, cubeb_state state) {}
 
 std::vector<std::string> ListCubebSinkDevices() {
     std::vector<std::string> device_list;

src/audio_core/null_sink.h

@@ -21,6 +21,12 @@ public:
 private:
     struct NullSinkStreamImpl final : SinkStream {
         void EnqueueSamples(u32 /*num_channels*/, const std::vector<s16>& /*samples*/) override {}
+
+        size_t SamplesInQueue(u32 /*num_channels*/) const override {
+            return 0;
+        }
+
+        void Flush() override {}
     } null_sink_stream;
 };
 

src/audio_core/sink_details.cpp

@@ -24,7 +24,7 @@ const std::vector<SinkDetails> g_sink_details = {
                 [] { return std::vector<std::string>{"null"}; }},
 };
 
-const SinkDetails& GetSinkDetails(std::string sink_id) {
+const SinkDetails& GetSinkDetails(std::string_view sink_id) {
     auto iter =
         std::find_if(g_sink_details.begin(), g_sink_details.end(),
                      [sink_id](const auto& sink_detail) { return sink_detail.id == sink_id; });

src/audio_core/sink_details.h

@@ -6,6 +6,8 @@
 
 #include <functional>
 #include <memory>
+#include <string>
+#include <string_view>
 #include <utility>
 #include <vector>
 
@@ -30,6 +32,6 @@ struct SinkDetails {
 
 extern const std::vector<SinkDetails> g_sink_details;
 
-const SinkDetails& GetSinkDetails(std::string sink_id);
+const SinkDetails& GetSinkDetails(std::string_view sink_id);
 
 } // namespace AudioCore

src/audio_core/sink_stream.h

@@ -25,6 +25,10 @@ public:
      * @param samples Samples in interleaved stereo PCM16 format.
      */
     virtual void EnqueueSamples(u32 num_channels, const std::vector<s16>& samples) = 0;
+
+    virtual std::size_t SamplesInQueue(u32 num_channels) const = 0;
+
+    virtual void Flush() = 0;
 };
 
 using SinkStreamPtr = std::unique_ptr<SinkStream>;

src/audio_core/stream.cpp

@@ -73,6 +73,7 @@ static void VolumeAdjustSamples(std::vector<s16>& samples) {
 void Stream::PlayNextBuffer() {
     if (!IsPlaying()) {
         // Ensure we are in playing state before playing the next buffer
+        sink_stream.Flush();
         return;
     }
 
@@ -83,6 +84,7 @@ void Stream::PlayNextBuffer() {
 
     if (queued_buffers.empty()) {
         // No queued buffers - we are effectively paused
+        sink_stream.Flush();
         return;
     }
 
@@ -90,6 +92,7 @@ void Stream::PlayNextBuffer() {
     queued_buffers.pop();
 
     VolumeAdjustSamples(active_buffer->Samples());
+
     sink_stream.EnqueueSamples(GetNumChannels(), active_buffer->GetSamples());
 
     CoreTiming::ScheduleEventThreadsafe(GetBufferReleaseCycles(*active_buffer), release_event, {});

src/audio_core/time_stretch.cpp

@@ -0,0 +1,68 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <algorithm>
+#include <cmath>
+#include <cstddef>
+#include "audio_core/time_stretch.h"
+#include "common/logging/log.h"
+
+namespace AudioCore {
+
+TimeStretcher::TimeStretcher(u32 sample_rate, u32 channel_count)
+    : m_sample_rate(sample_rate), m_channel_count(channel_count) {
+    m_sound_touch.setChannels(channel_count);
+    m_sound_touch.setSampleRate(sample_rate);
+    m_sound_touch.setPitch(1.0);
+    m_sound_touch.setTempo(1.0);
+}
+
+void TimeStretcher::Clear() {
+    m_sound_touch.clear();
+}
+
+void TimeStretcher::Flush() {
+    m_sound_touch.flush();
+}
+
+size_t TimeStretcher::Process(const s16* in, size_t num_in, s16* out, size_t num_out) {
+    const double time_delta = static_cast<double>(num_out) / m_sample_rate; // seconds
+
+    // We were given actual_samples number of samples, and num_samples were requested from us.
+    double current_ratio = static_cast<double>(num_in) / static_cast<double>(num_out);
+
+    const double max_latency = 1.0; // seconds
+    const double max_backlog = m_sample_rate * max_latency;
+    const double backlog_fullness = m_sound_touch.numSamples() / max_backlog;
+    if (backlog_fullness > 5.0) {
+        // Too many samples in backlog: Don't push anymore on
+        num_in = 0;
+    }
+
+    // We ideally want the backlog to be about 50% full.
+    // This gives some headroom both ways to prevent underflow and overflow.
+    // We tweak current_ratio to encourage this.
+    constexpr double tweak_time_scale = 0.05; // seconds
+    const double tweak_correction = (backlog_fullness - 0.5) * (time_delta / tweak_time_scale);
+    current_ratio *= std::pow(1.0 + 2.0 * tweak_correction, tweak_correction < 0 ? 3.0 : 1.0);
+
+    // This low-pass filter smoothes out variance in the calculated stretch ratio.
+    // The time-scale determines how responsive this filter is.
+    constexpr double lpf_time_scale = 2.0; // seconds
+    const double lpf_gain = 1.0 - std::exp(-time_delta / lpf_time_scale);
+    m_stretch_ratio += lpf_gain * (current_ratio - m_stretch_ratio);
+
+    // Place a lower limit of 5% speed.  When a game boots up, there will be
+    // many silence samples.  These do not need to be timestretched.
+    m_stretch_ratio = std::max(m_stretch_ratio, 0.05);
+    m_sound_touch.setTempo(m_stretch_ratio);
+
+    LOG_DEBUG(Audio, "{:5}/{:5} ratio:{:0.6f} backlog:{:0.6f}", num_in, num_out, m_stretch_ratio,
+              backlog_fullness);
+
+    m_sound_touch.putSamples(in, num_in);
+    return m_sound_touch.receiveSamples(out, num_out);
+}
+
+} // namespace AudioCore

src/audio_core/time_stretch.h

@@ -0,0 +1,36 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include <cstddef>
+#include <SoundTouch.h>
+#include "common/common_types.h"
+
+namespace AudioCore {
+
+class TimeStretcher {
+public:
+    TimeStretcher(u32 sample_rate, u32 channel_count);
+
+    /// @param in       Input sample buffer
+    /// @param num_in   Number of input frames in `in`
+    /// @param out      Output sample buffer
+    /// @param num_out  Desired number of output frames in `out`
+    /// @returns Actual number of frames written to `out`
+    size_t Process(const s16* in, size_t num_in, s16* out, size_t num_out);
+
+    void Clear();
+
+    void Flush();
+
+private:
+    u32 m_sample_rate;
+    u32 m_channel_count;
+    soundtouch::SoundTouch m_sound_touch;
+    double m_stretch_ratio = 1.0;
+};
+
+} // namespace AudioCore

src/common/CMakeLists.txt

@@ -71,6 +71,7 @@ add_library(common STATIC
     param_package.cpp
     param_package.h
     quaternion.h
+    ring_buffer.h
     scm_rev.cpp
     scm_rev.h
     scope_exit.h

src/common/ring_buffer.h

@@ -0,0 +1,111 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <algorithm>
+#include <array>
+#include <atomic>
+#include <cstddef>
+#include <cstring>
+#include <type_traits>
+#include <vector>
+#include "common/common_types.h"
+
+namespace Common {
+
+/// SPSC ring buffer
+/// @tparam T            Element type
+/// @tparam capacity     Number of slots in ring buffer
+/// @tparam granularity  Slot size in terms of number of elements
+template <typename T, size_t capacity, size_t granularity = 1>
+class RingBuffer {
+    /// A "slot" is made of `granularity` elements of `T`.
+    static constexpr size_t slot_size = granularity * sizeof(T);
+    // T must be safely memcpy-able and have a trivial default constructor.
+    static_assert(std::is_trivial_v<T>);
+    // Ensure capacity is sensible.
+    static_assert(capacity < std::numeric_limits<size_t>::max() / 2 / granularity);
+    static_assert((capacity & (capacity - 1)) == 0, "capacity must be a power of two");
+    // Ensure lock-free.
+    static_assert(std::atomic<size_t>::is_always_lock_free);
+
+public:
+    /// Pushes slots into the ring buffer
+    /// @param new_slots   Pointer to the slots to push
+    /// @param slot_count  Number of slots to push
+    /// @returns The number of slots actually pushed
+    size_t Push(const void* new_slots, size_t slot_count) {
+        const size_t write_index = m_write_index.load();
+        const size_t slots_free = capacity + m_read_index.load() - write_index;
+        const size_t push_count = std::min(slot_count, slots_free);
+
+        const size_t pos = write_index % capacity;
+        const size_t first_copy = std::min(capacity - pos, push_count);
+        const size_t second_copy = push_count - first_copy;
+
+        const char* in = static_cast<const char*>(new_slots);
+        std::memcpy(m_data.data() + pos * granularity, in, first_copy * slot_size);
+        in += first_copy * slot_size;
+        std::memcpy(m_data.data(), in, second_copy * slot_size);
+
+        m_write_index.store(write_index + push_count);
+
+        return push_count;
+    }
+
+    size_t Push(const std::vector<T>& input) {
+        return Push(input.data(), input.size());
+    }
+
+    /// Pops slots from the ring buffer
+    /// @param output     Where to store the popped slots
+    /// @param max_slots  Maximum number of slots to pop
+    /// @returns The number of slots actually popped
+    size_t Pop(void* output, size_t max_slots = ~size_t(0)) {
+        const size_t read_index = m_read_index.load();
+        const size_t slots_filled = m_write_index.load() - read_index;
+        const size_t pop_count = std::min(slots_filled, max_slots);
+
+        const size_t pos = read_index % capacity;
+        const size_t first_copy = std::min(capacity - pos, pop_count);
+        const size_t second_copy = pop_count - first_copy;
+
+        char* out = static_cast<char*>(output);
+        std::memcpy(out, m_data.data() + pos * granularity, first_copy * slot_size);
+        out += first_copy * slot_size;
+        std::memcpy(out, m_data.data(), second_copy * slot_size);
+
+        m_read_index.store(read_index + pop_count);
+
+        return pop_count;
+    }
+
+    std::vector<T> Pop(size_t max_slots = ~size_t(0)) {
+        std::vector<T> out(std::min(max_slots, capacity) * granularity);
+        const size_t count = Pop(out.data(), out.size() / granularity);
+        out.resize(count * granularity);
+        return out;
+    }
+
+    /// @returns Number of slots used
+    size_t Size() const {
+        return m_write_index.load() - m_read_index.load();
+    }
+
+    /// @returns Maximum size of ring buffer
+    constexpr size_t Capacity() const {
+        return capacity;
+    }
+
+private:
+    // It is important to align the below variables for performance reasons:
+    // Having them on the same cache-line would result in false-sharing between them.
+    alignas(128) std::atomic<size_t> m_read_index{0};
+    alignas(128) std::atomic<size_t> m_write_index{0};
+
+    std::array<T, granularity * capacity> m_data;
+};
+
+} // namespace Common

src/core/hle/ipc.h

@@ -153,7 +153,7 @@ struct DataPayloadHeader {
     u32_le magic;
     INSERT_PADDING_WORDS(1);
 };
-static_assert(sizeof(DataPayloadHeader) == 8, "DataPayloadRequest size is incorrect");
+static_assert(sizeof(DataPayloadHeader) == 8, "DataPayloadHeader size is incorrect");
 
 struct DomainMessageHeader {
     enum class CommandType : u32_le {

src/core/hle/kernel/errors.h

@@ -21,6 +21,7 @@ enum {
     HandleTableFull = 105,
     InvalidMemoryState = 106,
     InvalidMemoryPermissions = 108,
+    InvalidThreadPriority = 112,
     InvalidProcessorId = 113,
     InvalidHandle = 114,
     InvalidCombination = 116,
@@ -36,7 +37,7 @@ enum {
 // WARNING: The kernel is quite inconsistent in it's usage of errors code. Make sure to always
 // double check that the code matches before re-using the constant.
 
-// TODO(bunnei): Replace these with correct errors for Switch OS
+// TODO(bunnei): Replace -1 with correct errors for Switch OS
 constexpr ResultCode ERR_HANDLE_TABLE_FULL(ErrorModule::Kernel, ErrCodes::HandleTableFull);
 constexpr ResultCode ERR_SESSION_CLOSED_BY_REMOTE(-1);
 constexpr ResultCode ERR_PORT_NAME_TOO_LONG(ErrorModule::Kernel, ErrCodes::TooLarge);
@@ -53,15 +54,16 @@ constexpr ResultCode ERR_INVALID_ADDRESS_STATE(ErrorModule::Kernel, ErrCodes::In
 constexpr ResultCode ERR_INVALID_MEMORY_PERMISSIONS(ErrorModule::Kernel,
                                                     ErrCodes::InvalidMemoryPermissions);
 constexpr ResultCode ERR_INVALID_HANDLE(ErrorModule::Kernel, ErrCodes::InvalidHandle);
+constexpr ResultCode ERR_INVALID_PROCESSOR_ID(ErrorModule::Kernel, ErrCodes::InvalidProcessorId);
 constexpr ResultCode ERR_INVALID_STATE(ErrorModule::Kernel, ErrCodes::InvalidState);
+constexpr ResultCode ERR_INVALID_THREAD_PRIORITY(ErrorModule::Kernel,
+                                                 ErrCodes::InvalidThreadPriority);
 constexpr ResultCode ERR_INVALID_POINTER(-1);
 constexpr ResultCode ERR_INVALID_OBJECT_ADDR(-1);
 constexpr ResultCode ERR_NOT_AUTHORIZED(-1);
 /// Alternate code returned instead of ERR_INVALID_HANDLE in some code paths.
 constexpr ResultCode ERR_INVALID_HANDLE_OS(-1);
 constexpr ResultCode ERR_NOT_FOUND(-1);
-constexpr ResultCode ERR_OUT_OF_RANGE(-1);
-constexpr ResultCode ERR_OUT_OF_RANGE_KERNEL(-1);
 constexpr ResultCode RESULT_TIMEOUT(ErrorModule::Kernel, ErrCodes::Timeout);
 /// Returned when Accept() is called on a port with no sessions to be accepted.
 constexpr ResultCode ERR_NO_PENDING_SESSIONS(-1);

src/core/hle/kernel/svc.cpp

@@ -273,7 +273,11 @@ static void Break(u64 reason, u64 info1, u64 info2) {
 }
 
 /// Used to output a message on a debug hardware unit - does nothing on a retail unit
-static void OutputDebugString(VAddr address, s32 len) {
+static void OutputDebugString(VAddr address, u64 len) {
+    if (len == 0) {
+        return;
+    }
+
     std::string str(len, '\0');
     Memory::ReadBlock(address, str.data(), str.size());
     LOG_DEBUG(Debug_Emulated, "{}", str);
@@ -378,7 +382,7 @@ static ResultCode GetThreadPriority(u32* priority, Handle handle) {
 /// Sets the priority for the specified thread
 static ResultCode SetThreadPriority(Handle handle, u32 priority) {
     if (priority > THREADPRIO_LOWEST) {
-        return ERR_OUT_OF_RANGE;
+        return ERR_INVALID_THREAD_PRIORITY;
     }
 
     auto& kernel = Core::System::GetInstance().Kernel();
@@ -523,7 +527,7 @@ static ResultCode CreateThread(Handle* out_handle, VAddr entry_point, u64 arg, V
     std::string name = fmt::format("unknown-{:X}", entry_point);
 
     if (priority > THREADPRIO_LOWEST) {
-        return ERR_OUT_OF_RANGE;
+        return ERR_INVALID_THREAD_PRIORITY;
     }
 
     SharedPtr<ResourceLimit>& resource_limit = Core::CurrentProcess()->resource_limit;
@@ -544,8 +548,8 @@ static ResultCode CreateThread(Handle* out_handle, VAddr entry_point, u64 arg, V
     case THREADPROCESSORID_3:
         break;
     default:
-        ASSERT_MSG(false, "Unsupported thread processor ID: {}", processor_id);
-        break;
+        LOG_ERROR(Kernel_SVC, "Invalid thread processor ID: {}", processor_id);
+        return ERR_INVALID_PROCESSOR_ID;
     }
 
     auto& kernel = Core::System::GetInstance().Kernel();

src/core/hle/kernel/svc_wrap.h

@@ -222,9 +222,9 @@ void SvcWrap() {
     func((s64)PARAM(0));
 }
 
-template <void func(u64, s32 len)>
+template <void func(u64, u64 len)>
 void SvcWrap() {
-    func(PARAM(0), (s32)(PARAM(1) & 0xFFFFFFFF));
+    func(PARAM(0), PARAM(1));
 }
 
 template <void func(u64, u64, u64)>

src/core/hle/kernel/thread.cpp

@@ -227,12 +227,12 @@ ResultVal<SharedPtr<Thread>> Thread::Create(KernelCore& kernel, std::string name
     // Check if priority is in ranged. Lowest priority -> highest priority id.
     if (priority > THREADPRIO_LOWEST) {
         LOG_ERROR(Kernel_SVC, "Invalid thread priority: {}", priority);
-        return ERR_OUT_OF_RANGE;
+        return ERR_INVALID_THREAD_PRIORITY;
     }
 
     if (processor_id > THREADPROCESSORID_MAX) {
         LOG_ERROR(Kernel_SVC, "Invalid processor id: {}", processor_id);
-        return ERR_OUT_OF_RANGE_KERNEL;
+        return ERR_INVALID_PROCESSOR_ID;
     }
 
     // TODO(yuriks): Other checks, returning 0xD9001BEA

src/core/hle/kernel/thread.h

@@ -15,6 +15,12 @@
 #include "core/hle/kernel/wait_object.h"
 #include "core/hle/result.h"
 
+namespace Kernel {
+
+class KernelCore;
+class Process;
+class Scheduler;
+
 enum ThreadPriority : u32 {
     THREADPRIO_HIGHEST = 0,       ///< Highest thread priority
     THREADPRIO_USERLAND_MAX = 24, ///< Highest thread priority for userland apps
@@ -54,12 +60,6 @@ enum class ThreadWakeupReason {
     Timeout // The thread was woken up due to a wait timeout.
 };
 
-namespace Kernel {
-
-class KernelCore;
-class Process;
-class Scheduler;
-
 class Thread final : public WaitObject {
 public:
     /**

src/core/hle/service/acc/acc_su.h

@@ -6,8 +6,7 @@
 
 #include "core/hle/service/acc/acc.h"
 
-namespace Service {
-namespace Account {
+namespace Service::Account {
 
 class ACC_SU final : public Module::Interface {
 public:
@@ -16,5 +15,4 @@ public:
     ~ACC_SU() override;
 };
 
-} // namespace Account
-} // namespace Service
+} // namespace Service::Account

src/core/hle/service/audio/audio.cpp

@@ -15,6 +15,7 @@
 #include "core/hle/service/audio/audren_u.h"
 #include "core/hle/service/audio/codecctl.h"
 #include "core/hle/service/audio/hwopus.h"
+#include "core/hle/service/service.h"
 
 namespace Service::Audio {
 

src/core/hle/service/audio/audio.h

@@ -4,7 +4,9 @@
 
 #pragma once
 
-#include "core/hle/service/service.h"
+namespace Service::SM {
+class ServiceManager;
+}
 
 namespace Service::Audio {
 

src/core/hle/service/audio/audout_u.cpp

@@ -3,15 +3,20 @@
 // Refer to the license.txt file included.
 
 #include <array>
+#include <cstring>
 #include <vector>
 
+#include "audio_core/audio_out.h"
 #include "audio_core/codec.h"
+#include "common/common_funcs.h"
 #include "common/logging/log.h"
+#include "common/swap.h"
 #include "core/core.h"
 #include "core/hle/ipc_helpers.h"
 #include "core/hle/kernel/event.h"
 #include "core/hle/kernel/hle_ipc.h"
 #include "core/hle/service/audio/audout_u.h"
+#include "core/memory.h"
 
 namespace Service::Audio {
 
@@ -25,6 +30,18 @@ enum {
 constexpr std::array<char, 10> DefaultDevice{{"DeviceOut"}};
 constexpr int DefaultSampleRate{48000};
 
+struct AudoutParams {
+    s32_le sample_rate;
+    u16_le channel_count;
+    INSERT_PADDING_BYTES(2);
+};
+static_assert(sizeof(AudoutParams) == 0x8, "AudoutParams is an invalid size");
+
+enum class AudioState : u32 {
+    Started,
+    Stopped,
+};
+
 class IAudioOut final : public ServiceFramework<IAudioOut> {
 public:
     IAudioOut(AudoutParams audio_params, AudioCore::AudioOut& audio_core)

src/core/hle/service/audio/audout_u.h

@@ -4,27 +4,18 @@
 
 #pragma once
 
-#include "audio_core/audio_out.h"
 #include "core/hle/service/service.h"
 
+namespace AudioCore {
+class AudioOut;
+}
+
 namespace Kernel {
 class HLERequestContext;
 }
 
 namespace Service::Audio {
 
-struct AudoutParams {
-    s32_le sample_rate;
-    u16_le channel_count;
-    INSERT_PADDING_BYTES(2);
-};
-static_assert(sizeof(AudoutParams) == 0x8, "AudoutParams is an invalid size");
-
-enum class AudioState : u32 {
-    Started,
-    Stopped,
-};
-
 class IAudioOut;
 
 class AudOutU final : public ServiceFramework<AudOutU> {

src/core/hle/service/audio/audren_u.cpp

@@ -2,12 +2,14 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <algorithm>
 #include <array>
+#include <memory>
 
+#include "audio_core/audio_renderer.h"
 #include "common/alignment.h"
+#include "common/common_funcs.h"
 #include "common/logging/log.h"
-#include "core/core_timing.h"
-#include "core/core_timing_util.h"
 #include "core/hle/ipc_helpers.h"
 #include "core/hle/kernel/event.h"
 #include "core/hle/kernel/hle_ipc.h"

src/core/hle/service/audio/audren_u.h

@@ -4,7 +4,6 @@
 
 #pragma once
 
-#include "audio_core/audio_renderer.h"
 #include "core/hle/service/service.h"
 
 namespace Kernel {

src/core/hle/service/audio/hwopus.cpp

@@ -3,7 +3,12 @@
 // Refer to the license.txt file included.
 
 #include <cstring>
+#include <memory>
+#include <vector>
+
 #include <opus.h>
+
+#include "common/common_funcs.h"
 #include "common/logging/log.h"
 #include "core/hle/ipc_helpers.h"
 #include "core/hle/kernel/hle_ipc.h"

src/core/hle/service/ns/pl_u.cpp

@@ -2,6 +2,10 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <algorithm>
+#include <cstring>
+#include <vector>
+
 #include <FontChineseSimplified.h>
 #include <FontChineseTraditional.h>
 #include <FontExtendedChineseSimplified.h>
@@ -9,14 +13,19 @@
 #include <FontNintendoExtended.h>
 #include <FontStandard.h>
 
+#include "common/assert.h"
 #include "common/common_paths.h"
+#include "common/common_types.h"
 #include "common/file_util.h"
+#include "common/logging/log.h"
+#include "common/swap.h"
 #include "core/core.h"
 #include "core/file_sys/content_archive.h"
 #include "core/file_sys/nca_metadata.h"
 #include "core/file_sys/registered_cache.h"
 #include "core/file_sys/romfs.h"
 #include "core/hle/ipc_helpers.h"
+#include "core/hle/kernel/shared_memory.h"
 #include "core/hle/service/filesystem/filesystem.h"
 #include "core/hle/service/ns/pl_u.h"
 
@@ -35,49 +44,41 @@ struct FontRegion {
     u32 size;
 };
 
-static constexpr std::array<std::pair<FontArchives, const char*>, 7> SHARED_FONTS{
+constexpr std::array<std::pair<FontArchives, const char*>, 7> SHARED_FONTS{
     std::make_pair(FontArchives::Standard, "nintendo_udsg-r_std_003.bfttf"),
     std::make_pair(FontArchives::ChineseSimple, "nintendo_udsg-r_org_zh-cn_003.bfttf"),
     std::make_pair(FontArchives::ChineseSimple, "nintendo_udsg-r_ext_zh-cn_003.bfttf"),
     std::make_pair(FontArchives::ChineseTraditional, "nintendo_udjxh-db_zh-tw_003.bfttf"),
     std::make_pair(FontArchives::Korean, "nintendo_udsg-r_ko_003.bfttf"),
     std::make_pair(FontArchives::Extension, "nintendo_ext_003.bfttf"),
-    std::make_pair(FontArchives::Extension, "nintendo_ext2_003.bfttf")};
+    std::make_pair(FontArchives::Extension, "nintendo_ext2_003.bfttf"),
+};
 
-static constexpr std::array<const char*, 7> SHARED_FONTS_TTF{"FontStandard.ttf",
-                                                             "FontChineseSimplified.ttf",
-                                                             "FontExtendedChineseSimplified.ttf",
-                                                             "FontChineseTraditional.ttf",
-                                                             "FontKorean.ttf",
-                                                             "FontNintendoExtended.ttf",
-                                                             "FontNintendoExtended2.ttf"};
+constexpr std::array<const char*, 7> SHARED_FONTS_TTF{
+    "FontStandard.ttf",
+    "FontChineseSimplified.ttf",
+    "FontExtendedChineseSimplified.ttf",
+    "FontChineseTraditional.ttf",
+    "FontKorean.ttf",
+    "FontNintendoExtended.ttf",
+    "FontNintendoExtended2.ttf",
+};
 
 // The below data is specific to shared font data dumped from Switch on f/w 2.2
 // Virtual address and offsets/sizes likely will vary by dump
-static constexpr VAddr SHARED_FONT_MEM_VADDR{0x00000009d3016000ULL};
-static constexpr u32 EXPECTED_RESULT{
-    0x7f9a0218}; // What we expect the decrypted bfttf first 4 bytes to be
-static constexpr u32 EXPECTED_MAGIC{
-    0x36f81a1e}; // What we expect the encrypted bfttf first 4 bytes to be
-static constexpr u64 SHARED_FONT_MEM_SIZE{0x1100000};
-static constexpr FontRegion EMPTY_REGION{0, 0};
-std::vector<FontRegion>
-    SHARED_FONT_REGIONS{}; // Automatically populated based on shared_fonts dump or system archives
-
-const FontRegion& GetSharedFontRegion(size_t index) {
-    if (index >= SHARED_FONT_REGIONS.size() || SHARED_FONT_REGIONS.empty()) {
-        // No font fallback
-        return EMPTY_REGION;
-    }
-    return SHARED_FONT_REGIONS.at(index);
-}
+constexpr VAddr SHARED_FONT_MEM_VADDR{0x00000009d3016000ULL};
+constexpr u32 EXPECTED_RESULT{0x7f9a0218}; // What we expect the decrypted bfttf first 4 bytes to be
+constexpr u32 EXPECTED_MAGIC{0x36f81a1e};  // What we expect the encrypted bfttf first 4 bytes to be
+constexpr u64 SHARED_FONT_MEM_SIZE{0x1100000};
+constexpr FontRegion EMPTY_REGION{0, 0};
 
 enum class LoadState : u32 {
     Loading = 0,
     Done = 1,
 };
 
-void DecryptSharedFont(const std::vector<u32>& input, std::vector<u8>& output, size_t& offset) {
+static void DecryptSharedFont(const std::vector<u32>& input, std::vector<u8>& output,
+                              size_t& offset) {
     ASSERT_MSG(offset + (input.size() * sizeof(u32)) < SHARED_FONT_MEM_SIZE,
                "Shared fonts exceeds 17mb!");
     ASSERT_MSG(input[0] == EXPECTED_MAGIC, "Failed to derive key, unexpected magic number");
@@ -104,28 +105,52 @@ static void EncryptSharedFont(const std::vector<u8>& input, std::vector<u8>& out
     offset += input.size() + (sizeof(u32) * 2);
 }
 
+// Helper function to make BuildSharedFontsRawRegions a bit nicer
 static u32 GetU32Swapped(const u8* data) {
     u32 value;
     std::memcpy(&value, data, sizeof(value));
-    return Common::swap32(value); // Helper function to make BuildSharedFontsRawRegions a bit nicer
+    return Common::swap32(value);
 }
 
-void BuildSharedFontsRawRegions(const std::vector<u8>& input) {
-    unsigned cur_offset = 0; // As we can derive the xor key we can just populate the offsets based
-                             // on the shared memory dump
-    for (size_t i = 0; i < SHARED_FONTS.size(); i++) {
-        // Out of shared fonts/Invalid font
-        if (GetU32Swapped(input.data() + cur_offset) != EXPECTED_RESULT)
-            break;
-        const u32 KEY = GetU32Swapped(input.data() + cur_offset) ^
-                        EXPECTED_MAGIC; // Derive key withing inverse xor
-        const u32 SIZE = GetU32Swapped(input.data() + cur_offset + 4) ^ KEY;
-        SHARED_FONT_REGIONS.push_back(FontRegion{cur_offset + 8, SIZE});
-        cur_offset += SIZE + 8;
+struct PL_U::Impl {
+    const FontRegion& GetSharedFontRegion(size_t index) const {
+        if (index >= shared_font_regions.size() || shared_font_regions.empty()) {
+            // No font fallback
+            return EMPTY_REGION;
+        }
+        return shared_font_regions.at(index);
     }
-}
 
-PL_U::PL_U() : ServiceFramework("pl:u") {
+    void BuildSharedFontsRawRegions(const std::vector<u8>& input) {
+        // As we can derive the xor key we can just populate the offsets
+        // based on the shared memory dump
+        unsigned cur_offset = 0;
+
+        for (size_t i = 0; i < SHARED_FONTS.size(); i++) {
+            // Out of shared fonts/invalid font
+            if (GetU32Swapped(input.data() + cur_offset) != EXPECTED_RESULT) {
+                break;
+            }
+
+            // Derive key withing inverse xor
+            const u32 KEY = GetU32Swapped(input.data() + cur_offset) ^ EXPECTED_MAGIC;
+            const u32 SIZE = GetU32Swapped(input.data() + cur_offset + 4) ^ KEY;
+            shared_font_regions.push_back(FontRegion{cur_offset + 8, SIZE});
+            cur_offset += SIZE + 8;
+        }
+    }
+
+    /// Handle to shared memory region designated for a shared font
+    Kernel::SharedPtr<Kernel::SharedMemory> shared_font_mem;
+
+    /// Backing memory for the shared font data
+    std::shared_ptr<std::vector<u8>> shared_font;
+
+    // Automatically populated based on shared_fonts dump or system archives.
+    std::vector<FontRegion> shared_font_regions;
+};
+
+PL_U::PL_U() : ServiceFramework("pl:u"), impl{std::make_unique<Impl>()} {
     static const FunctionInfo functions[] = {
         {0, &PL_U::RequestLoad, "RequestLoad"},
         {1, &PL_U::GetLoadState, "GetLoadState"},
@@ -141,7 +166,7 @@ PL_U::PL_U() : ServiceFramework("pl:u") {
     // Rebuild shared fonts from data ncas
     if (nand->HasEntry(static_cast<u64>(FontArchives::Standard),
                        FileSys::ContentRecordType::Data)) {
-        shared_font = std::make_shared<std::vector<u8>>(SHARED_FONT_MEM_SIZE);
+        impl->shared_font = std::make_shared<std::vector<u8>>(SHARED_FONT_MEM_SIZE);
         for (auto font : SHARED_FONTS) {
             const auto nca =
                 nand->GetEntry(static_cast<u64>(font.first), FileSys::ContentRecordType::Data);
@@ -177,12 +202,12 @@ PL_U::PL_U() : ServiceFramework("pl:u") {
                 static_cast<u32>(offset + 8),
                 static_cast<u32>((font_data_u32.size() * sizeof(u32)) -
                                  8)}; // Font offset and size do not account for the header
-            DecryptSharedFont(font_data_u32, *shared_font, offset);
-            SHARED_FONT_REGIONS.push_back(region);
+            DecryptSharedFont(font_data_u32, *impl->shared_font, offset);
+            impl->shared_font_regions.push_back(region);
         }
 
     } else {
-        shared_font = std::make_shared<std::vector<u8>>(
+        impl->shared_font = std::make_shared<std::vector<u8>>(
             SHARED_FONT_MEM_SIZE); // Shared memory needs to always be allocated and a fixed size
 
         const std::string user_path = FileUtil::GetUserPath(FileUtil::UserPath::SysDataDir);
@@ -206,8 +231,8 @@ PL_U::PL_U() : ServiceFramework("pl:u") {
                         static_cast<u32>(offset + 8),
                         static_cast<u32>(ttf_bytes.size())}; // Font offset and size do not account
                                                              // for the header
-                    EncryptSharedFont(ttf_bytes, *shared_font, offset);
-                    SHARED_FONT_REGIONS.push_back(region);
+                    EncryptSharedFont(ttf_bytes, *impl->shared_font, offset);
+                    impl->shared_font_regions.push_back(region);
                 } else {
                     LOG_WARNING(Service_NS, "Unable to load font: {}", font_ttf);
                 }
@@ -222,26 +247,28 @@ PL_U::PL_U() : ServiceFramework("pl:u") {
         if (file.IsOpen()) {
             // Read shared font data
             ASSERT(file.GetSize() == SHARED_FONT_MEM_SIZE);
-            file.ReadBytes(shared_font->data(), shared_font->size());
-            BuildSharedFontsRawRegions(*shared_font);
+            file.ReadBytes(impl->shared_font->data(), impl->shared_font->size());
+            impl->BuildSharedFontsRawRegions(*impl->shared_font);
         } else {
             LOG_WARNING(Service_NS,
                         "Shared Font file missing. Loading open source replacement from memory");
 
+            // clang-format off
             const std::vector<std::vector<u8>> open_source_shared_fonts_ttf = {
                 {std::begin(FontChineseSimplified), std::end(FontChineseSimplified)},
                 {std::begin(FontChineseTraditional), std::end(FontChineseTraditional)},
-                {std::begin(FontExtendedChineseSimplified),
-                 std::end(FontExtendedChineseSimplified)},
+                {std::begin(FontExtendedChineseSimplified), std::end(FontExtendedChineseSimplified)},
+                {std::begin(FontKorean), std::end(FontKorean)},
                 {std::begin(FontNintendoExtended), std::end(FontNintendoExtended)},
                 {std::begin(FontStandard), std::end(FontStandard)},
             };
+            // clang-format on
 
             for (const std::vector<u8>& font_ttf : open_source_shared_fonts_ttf) {
                 const FontRegion region{static_cast<u32>(offset + 8),
                                         static_cast<u32>(font_ttf.size())};
-                EncryptSharedFont(font_ttf, *shared_font, offset);
-                SHARED_FONT_REGIONS.push_back(region);
+                EncryptSharedFont(font_ttf, *impl->shared_font, offset);
+                impl->shared_font_regions.push_back(region);
             }
         }
     }
@@ -275,7 +302,7 @@ void PL_U::GetSize(Kernel::HLERequestContext& ctx) {
     LOG_DEBUG(Service_NS, "called, font_id={}", font_id);
     IPC::ResponseBuilder rb{ctx, 3};
     rb.Push(RESULT_SUCCESS);
-    rb.Push<u32>(GetSharedFontRegion(font_id).size);
+    rb.Push<u32>(impl->GetSharedFontRegion(font_id).size);
 }
 
 void PL_U::GetSharedMemoryAddressOffset(Kernel::HLERequestContext& ctx) {
@@ -285,17 +312,18 @@ void PL_U::GetSharedMemoryAddressOffset(Kernel::HLERequestContext& ctx) {
     LOG_DEBUG(Service_NS, "called, font_id={}", font_id);
     IPC::ResponseBuilder rb{ctx, 3};
     rb.Push(RESULT_SUCCESS);
-    rb.Push<u32>(GetSharedFontRegion(font_id).offset);
+    rb.Push<u32>(impl->GetSharedFontRegion(font_id).offset);
 }
 
 void PL_U::GetSharedMemoryNativeHandle(Kernel::HLERequestContext& ctx) {
     // Map backing memory for the font data
-    Core::CurrentProcess()->vm_manager.MapMemoryBlock(
-        SHARED_FONT_MEM_VADDR, shared_font, 0, SHARED_FONT_MEM_SIZE, Kernel::MemoryState::Shared);
+    Core::CurrentProcess()->vm_manager.MapMemoryBlock(SHARED_FONT_MEM_VADDR, impl->shared_font, 0,
+                                                      SHARED_FONT_MEM_SIZE,
+                                                      Kernel::MemoryState::Shared);
 
     // Create shared font memory object
     auto& kernel = Core::System::GetInstance().Kernel();
-    shared_font_mem = Kernel::SharedMemory::Create(
+    impl->shared_font_mem = Kernel::SharedMemory::Create(
         kernel, Core::CurrentProcess(), SHARED_FONT_MEM_SIZE, Kernel::MemoryPermission::ReadWrite,
         Kernel::MemoryPermission::Read, SHARED_FONT_MEM_VADDR, Kernel::MemoryRegion::BASE,
         "PL_U:shared_font_mem");
@@ -303,7 +331,7 @@ void PL_U::GetSharedMemoryNativeHandle(Kernel::HLERequestContext& ctx) {
     LOG_DEBUG(Service_NS, "called");
     IPC::ResponseBuilder rb{ctx, 2, 1};
     rb.Push(RESULT_SUCCESS);
-    rb.PushCopyObjects(shared_font_mem);
+    rb.PushCopyObjects(impl->shared_font_mem);
 }
 
 void PL_U::GetSharedFontInOrderOfPriority(Kernel::HLERequestContext& ctx) {
@@ -316,9 +344,9 @@ void PL_U::GetSharedFontInOrderOfPriority(Kernel::HLERequestContext& ctx) {
     std::vector<u32> font_sizes;
 
     // TODO(ogniK): Have actual priority order
-    for (size_t i = 0; i < SHARED_FONT_REGIONS.size(); i++) {
+    for (size_t i = 0; i < impl->shared_font_regions.size(); i++) {
         font_codes.push_back(static_cast<u32>(i));
-        auto region = GetSharedFontRegion(i);
+        auto region = impl->GetSharedFontRegion(i);
         font_offsets.push_back(region.offset);
         font_sizes.push_back(region.size);
     }

src/core/hle/service/ns/pl_u.h

@@ -5,7 +5,6 @@
 #pragma once
 
 #include <memory>
-#include "core/hle/kernel/shared_memory.h"
 #include "core/hle/service/service.h"
 
 namespace Service::NS {
@@ -23,11 +22,8 @@ private:
     void GetSharedMemoryNativeHandle(Kernel::HLERequestContext& ctx);
     void GetSharedFontInOrderOfPriority(Kernel::HLERequestContext& ctx);
 
-    /// Handle to shared memory region designated for a shared font
-    Kernel::SharedPtr<Kernel::SharedMemory> shared_font_mem;
-
-    /// Backing memory for the shared font data
-    std::shared_ptr<std::vector<u8>> shared_font;
+    struct Impl;
+    std::unique_ptr<Impl> impl;
 };
 
 } // namespace Service::NS

src/core/hle/service/nvflinger/buffer_queue.cpp

@@ -9,8 +9,7 @@
 #include "core/core.h"
 #include "core/hle/service/nvflinger/buffer_queue.h"
 
-namespace Service {
-namespace NVFlinger {
+namespace Service::NVFlinger {
 
 BufferQueue::BufferQueue(u32 id, u64 layer_id) : id(id), layer_id(layer_id) {
     auto& kernel = Core::System::GetInstance().Kernel();
@@ -104,5 +103,4 @@ u32 BufferQueue::Query(QueryType type) {
     return 0;
 }
 
-} // namespace NVFlinger
-} // namespace Service
+} // namespace Service::NVFlinger

src/core/hle/service/nvflinger/buffer_queue.h

@@ -15,8 +15,7 @@ namespace CoreTiming {
 struct EventType;
 }
 
-namespace Service {
-namespace NVFlinger {
+namespace Service::NVFlinger {
 
 struct IGBPBuffer {
     u32_le magic;
@@ -98,5 +97,4 @@ private:
     Kernel::SharedPtr<Kernel::Event> buffer_wait_event;
 };
 
-} // namespace NVFlinger
-} // namespace Service
+} // namespace Service::NVFlinger

src/core/hle/service/prepo/prepo.cpp

@@ -1,36 +1,47 @@
-#include <cinttypes>
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
 #include "common/logging/log.h"
 #include "core/hle/ipc_helpers.h"
-#include "core/hle/kernel/event.h"
 #include "core/hle/service/prepo/prepo.h"
+#include "core/hle/service/service.h"
 
 namespace Service::PlayReport {
-PlayReport::PlayReport(const char* name) : ServiceFramework(name) {
-    static const FunctionInfo functions[] = {
-        {10100, nullptr, "SaveReport"},
-        {10101, &PlayReport::SaveReportWithUser, "SaveReportWithUser"},
-        {10200, nullptr, "RequestImmediateTransmission"},
-        {10300, nullptr, "GetTransmissionStatus"},
-        {20100, nullptr, "SaveSystemReport"},
-        {20200, nullptr, "SetOperationMode"},
-        {20101, nullptr, "SaveSystemReportWithUser"},
-        {30100, nullptr, "ClearStorage"},
-        {40100, nullptr, "IsUserAgreementCheckEnabled"},
-        {40101, nullptr, "SetUserAgreementCheckEnabled"},
-        {90100, nullptr, "GetStorageUsage"},
-        {90200, nullptr, "GetStatistics"},
-        {90201, nullptr, "GetThroughputHistory"},
-        {90300, nullptr, "GetLastUploadError"},
-    };
-    RegisterHandlers(functions);
-};
 
-void PlayReport::SaveReportWithUser(Kernel::HLERequestContext& ctx) {
-    // TODO(ogniK): Do we want to add play report?
-    LOG_WARNING(Service_PREPO, "(STUBBED) called");
+class PlayReport final : public ServiceFramework<PlayReport> {
+public:
+    explicit PlayReport(const char* name) : ServiceFramework{name} {
+        // clang-format off
+        static const FunctionInfo functions[] = {
+            {10100, nullptr, "SaveReport"},
+            {10101, &PlayReport::SaveReportWithUser, "SaveReportWithUser"},
+            {10200, nullptr, "RequestImmediateTransmission"},
+            {10300, nullptr, "GetTransmissionStatus"},
+            {20100, nullptr, "SaveSystemReport"},
+            {20200, nullptr, "SetOperationMode"},
+            {20101, nullptr, "SaveSystemReportWithUser"},
+            {30100, nullptr, "ClearStorage"},
+            {40100, nullptr, "IsUserAgreementCheckEnabled"},
+            {40101, nullptr, "SetUserAgreementCheckEnabled"},
+            {90100, nullptr, "GetStorageUsage"},
+            {90200, nullptr, "GetStatistics"},
+            {90201, nullptr, "GetThroughputHistory"},
+            {90300, nullptr, "GetLastUploadError"},
+        };
+        // clang-format on
 
-    IPC::ResponseBuilder rb{ctx, 2};
-    rb.Push(RESULT_SUCCESS);
+        RegisterHandlers(functions);
+    }
+
+private:
+    void SaveReportWithUser(Kernel::HLERequestContext& ctx) {
+        // TODO(ogniK): Do we want to add play report?
+        LOG_WARNING(Service_PREPO, "(STUBBED) called");
+
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+    }
 };
 
 void InstallInterfaces(SM::ServiceManager& service_manager) {

src/core/hle/service/prepo/prepo.h

@@ -4,22 +4,12 @@
 
 #pragma once
 
-#include <memory>
-#include <string>
-#include "core/hle/kernel/event.h"
-#include "core/hle/service/service.h"
+namespace Service::SM {
+class ServiceManager;
+}
 
 namespace Service::PlayReport {
 
-class PlayReport final : public ServiceFramework<PlayReport> {
-public:
-    explicit PlayReport(const char* name);
-    ~PlayReport() = default;
-
-private:
-    void SaveReportWithUser(Kernel::HLERequestContext& ctx);
-};
-
 void InstallInterfaces(SM::ServiceManager& service_manager);
 
 } // namespace Service::PlayReport

src/core/hle/service/vi/vi.cpp

@@ -514,7 +514,7 @@ private:
                 ctx.SleepClientThread(
                     Kernel::GetCurrentThread(), "IHOSBinderDriver::DequeueBuffer", -1,
                     [=](Kernel::SharedPtr<Kernel::Thread> thread, Kernel::HLERequestContext& ctx,
-                        ThreadWakeupReason reason) {
+                        Kernel::ThreadWakeupReason reason) {
                         // Repeat TransactParcel DequeueBuffer when a buffer is available
                         auto buffer_queue = nv_flinger->GetBufferQueue(id);
                         boost::optional<u32> slot = buffer_queue->DequeueBuffer(width, height);

src/core/loader/nro.cpp

@@ -191,7 +191,7 @@ ResultStatus AppLoader_NRO::Load(Kernel::SharedPtr<Kernel::Process>& process) {
     process->svc_access_mask.set();
     process->resource_limit =
         kernel.ResourceLimitForCategory(Kernel::ResourceLimitCategory::APPLICATION);
-    process->Run(base_addr, THREADPRIO_DEFAULT, Memory::DEFAULT_STACK_SIZE);
+    process->Run(base_addr, Kernel::THREADPRIO_DEFAULT, Memory::DEFAULT_STACK_SIZE);
 
     is_loaded = true;
     return ResultStatus::Success;

src/core/loader/nso.cpp

@@ -157,7 +157,8 @@ ResultStatus AppLoader_NSO::Load(Kernel::SharedPtr<Kernel::Process>& process) {
     process->svc_access_mask.set();
     process->resource_limit =
         kernel.ResourceLimitForCategory(Kernel::ResourceLimitCategory::APPLICATION);
-    process->Run(Memory::PROCESS_IMAGE_VADDR, THREADPRIO_DEFAULT, Memory::DEFAULT_STACK_SIZE);
+    process->Run(Memory::PROCESS_IMAGE_VADDR, Kernel::THREADPRIO_DEFAULT,
+                 Memory::DEFAULT_STACK_SIZE);
 
     is_loaded = true;
     return ResultStatus::Success;

src/core/settings.h

@@ -148,6 +148,7 @@ struct Values {
 
     // Audio
     std::string sink_id;
+    bool enable_audio_stretching;
     std::string audio_device_id;
     float volume;
 

src/core/telemetry_session.cpp

@@ -120,6 +120,9 @@ TelemetrySession::TelemetrySession() {
     Telemetry::AppendOSInfo(field_collection);
 
     // Log user configuration information
+    AddField(Telemetry::FieldType::UserConfig, "Audio_SinkId", Settings::values.sink_id);
+    AddField(Telemetry::FieldType::UserConfig, "Audio_EnableAudioStretching",
+             Settings::values.enable_audio_stretching);
     AddField(Telemetry::FieldType::UserConfig, "Core_UseCpuJit", Settings::values.use_cpu_jit);
     AddField(Telemetry::FieldType::UserConfig, "Core_UseMultiCore",
              Settings::values.use_multi_core);

src/tests/CMakeLists.txt

@@ -1,5 +1,6 @@
 add_executable(tests
     common/param_package.cpp
+    common/ring_buffer.cpp
     core/arm/arm_test_common.cpp
     core/arm/arm_test_common.h
     core/core_timing.cpp

src/tests/common/ring_buffer.cpp

@@ -0,0 +1,130 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <algorithm>
+#include <array>
+#include <cstddef>
+#include <numeric>
+#include <thread>
+#include <vector>
+#include <catch2/catch.hpp>
+#include "common/ring_buffer.h"
+
+namespace Common {
+
+TEST_CASE("RingBuffer: Basic Tests", "[common]") {
+    RingBuffer<char, 4, 1> buf;
+
+    // Pushing values into a ring buffer with space should succeed.
+    for (size_t i = 0; i < 4; i++) {
+        const char elem = static_cast<char>(i);
+        const size_t count = buf.Push(&elem, 1);
+        REQUIRE(count == 1);
+    }
+
+    REQUIRE(buf.Size() == 4);
+
+    // Pushing values into a full ring buffer should fail.
+    {
+        const char elem = static_cast<char>(42);
+        const size_t count = buf.Push(&elem, 1);
+        REQUIRE(count == 0);
+    }
+
+    REQUIRE(buf.Size() == 4);
+
+    // Popping multiple values from a ring buffer with values should succeed.
+    {
+        const std::vector<char> popped = buf.Pop(2);
+        REQUIRE(popped.size() == 2);
+        REQUIRE(popped[0] == 0);
+        REQUIRE(popped[1] == 1);
+    }
+
+    REQUIRE(buf.Size() == 2);
+
+    // Popping a single value from a ring buffer with values should succeed.
+    {
+        const std::vector<char> popped = buf.Pop(1);
+        REQUIRE(popped.size() == 1);
+        REQUIRE(popped[0] == 2);
+    }
+
+    REQUIRE(buf.Size() == 1);
+
+    // Pushing more values than space available should partially suceed.
+    {
+        std::vector<char> to_push(6);
+        std::iota(to_push.begin(), to_push.end(), 88);
+        const size_t count = buf.Push(to_push);
+        REQUIRE(count == 3);
+    }
+
+    REQUIRE(buf.Size() == 4);
+
+    // Doing an unlimited pop should pop all values.
+    {
+        const std::vector<char> popped = buf.Pop();
+        REQUIRE(popped.size() == 4);
+        REQUIRE(popped[0] == 3);
+        REQUIRE(popped[1] == 88);
+        REQUIRE(popped[2] == 89);
+        REQUIRE(popped[3] == 90);
+    }
+
+    REQUIRE(buf.Size() == 0);
+}
+
+TEST_CASE("RingBuffer: Threaded Test", "[common]") {
+    RingBuffer<char, 4, 2> buf;
+    const char seed = 42;
+    const size_t count = 1000000;
+    size_t full = 0;
+    size_t empty = 0;
+
+    const auto next_value = [](std::array<char, 2>& value) {
+        value[0] += 1;
+        value[1] += 2;
+    };
+
+    std::thread producer{[&] {
+        std::array<char, 2> value = {seed, seed};
+        size_t i = 0;
+        while (i < count) {
+            if (const size_t c = buf.Push(&value[0], 1); c > 0) {
+                REQUIRE(c == 1);
+                i++;
+                next_value(value);
+            } else {
+                full++;
+                std::this_thread::yield();
+            }
+        }
+    }};
+
+    std::thread consumer{[&] {
+        std::array<char, 2> value = {seed, seed};
+        size_t i = 0;
+        while (i < count) {
+            if (const std::vector<char> v = buf.Pop(1); v.size() > 0) {
+                REQUIRE(v.size() == 2);
+                REQUIRE(v[0] == value[0]);
+                REQUIRE(v[1] == value[1]);
+                i++;
+                next_value(value);
+            } else {
+                empty++;
+                std::this_thread::yield();
+            }
+        }
+    }};
+
+    producer.join();
+    consumer.join();
+
+    REQUIRE(buf.Size() == 0);
+    printf("RingBuffer: Threaded Test: full: %zu, empty: %zu\n", full, empty);
+}
+
+} // namespace Common

src/video_core/CMakeLists.txt

@@ -12,7 +12,6 @@ add_library(video_core STATIC
     engines/maxwell_dma.cpp
     engines/maxwell_dma.h
     engines/shader_bytecode.h
-    engines/shader_header.h
     gpu.cpp
     gpu.h
     macro_interpreter.cpp

src/video_core/engines/shader_bytecode.h

@@ -254,6 +254,15 @@ enum class TextureQueryType : u64 {
     BorderColor = 22,
 };
 
+enum class TextureProcessMode : u64 {
+    None = 0,
+    LZ = 1,  // Unknown, appears to be the same as none.
+    LB = 2,  // Load Bias.
+    LL = 3,  // Load LOD (LevelOfDetail)
+    LBA = 6, // Load Bias. The A is unknown, does not appear to differ with LB
+    LLA = 7  // Load LOD. The A is unknown, does not appear to differ with LL
+};
+
 enum class IpaInterpMode : u64 { Linear = 0, Perspective = 1, Flat = 2, Sc = 3 };
 enum class IpaSampleMode : u64 { Default = 0, Centroid = 1, Offset = 2 };
 
@@ -423,6 +432,45 @@ union Instruction {
         }
     } bfe;
 
+    union {
+        BitField<48, 3, u64> pred48;
+
+        union {
+            BitField<20, 20, u64> entry_a;
+            BitField<39, 5, u64> entry_b;
+            BitField<45, 1, u64> neg;
+            BitField<46, 1, u64> uses_cc;
+        } imm;
+
+        union {
+            BitField<20, 14, u64> cb_index;
+            BitField<34, 5, u64> cb_offset;
+            BitField<56, 1, u64> neg;
+            BitField<57, 1, u64> uses_cc;
+        } hi;
+
+        union {
+            BitField<20, 14, u64> cb_index;
+            BitField<34, 5, u64> cb_offset;
+            BitField<39, 5, u64> entry_a;
+            BitField<45, 1, u64> neg;
+            BitField<46, 1, u64> uses_cc;
+        } rz;
+
+        union {
+            BitField<39, 5, u64> entry_a;
+            BitField<45, 1, u64> neg;
+            BitField<46, 1, u64> uses_cc;
+        } r1;
+
+        union {
+            BitField<28, 8, u64> entry_a;
+            BitField<37, 1, u64> neg;
+            BitField<38, 1, u64> uses_cc;
+        } r2;
+
+    } lea;
+
     union {
         BitField<0, 5, FlowCondition> cond;
     } flow;
@@ -477,6 +525,18 @@ union Instruction {
         BitField<45, 2, PredOperation> op;
     } psetp;
 
+    union {
+        BitField<12, 3, u64> pred12;
+        BitField<15, 1, u64> neg_pred12;
+        BitField<24, 2, PredOperation> cond;
+        BitField<29, 3, u64> pred29;
+        BitField<32, 1, u64> neg_pred29;
+        BitField<39, 3, u64> pred39;
+        BitField<42, 1, u64> neg_pred39;
+        BitField<44, 1, u64> bf;
+        BitField<45, 2, PredOperation> op;
+    } pset;
+
     union {
         BitField<39, 3, u64> pred39;
         BitField<42, 1, u64> neg_pred;
@@ -522,6 +582,7 @@ union Instruction {
         BitField<28, 1, u64> array;
         BitField<29, 2, TextureType> texture_type;
         BitField<31, 4, u64> component_mask;
+        BitField<55, 3, TextureProcessMode> process_mode;
 
         bool IsComponentEnabled(size_t component) const {
             return ((1ull << component) & component_mask) != 0;
@@ -696,6 +757,7 @@ public:
         STG, // Store in global memory
         TEX,
         TXQ,    // Texture Query
+        TXB,    // Texture Bias
         TEXS,   // Texture Fetch with scalar/non-vec4 source/destinations
         TLDS,   // Texture Load with scalar/non-vec4 source/destinations
         TLD4,   // Texture Load 4
@@ -726,6 +788,11 @@ public:
         ISCADD_C, // Scale and Add
         ISCADD_R,
         ISCADD_IMM,
+        LEA_R1,
+        LEA_R2,
+        LEA_RZ,
+        LEA_IMM,
+        LEA_HI,
         POPC_C,
         POPC_R,
         POPC_IMM,
@@ -784,6 +851,7 @@ public:
         ISET_C,
         ISET_IMM,
         PSETP,
+        PSET,
         XMAD_IMM,
         XMAD_CR,
         XMAD_RC,
@@ -807,6 +875,7 @@ public:
         IntegerSet,
         IntegerSetPredicate,
         PredicateSetPredicate,
+        PredicateSetRegister,
         Conversion,
         Xmad,
         Unknown,
@@ -922,6 +991,7 @@ private:
             INST("1110111011011---", Id::STG, Type::Memory, "STG"),
             INST("110000----111---", Id::TEX, Type::Memory, "TEX"),
             INST("1101111101001---", Id::TXQ, Type::Memory, "TXQ"),
+            INST("1101111010111---", Id::TXB, Type::Memory, "TXB"),
             INST("1101100---------", Id::TEXS, Type::Memory, "TEXS"),
             INST("1101101---------", Id::TLDS, Type::Memory, "TLDS"),
             INST("110010----111---", Id::TLD4, Type::Memory, "TLD4"),
@@ -958,6 +1028,11 @@ private:
             INST("0100110010100---", Id::SEL_C, Type::ArithmeticInteger, "SEL_C"),
             INST("0101110010100---", Id::SEL_R, Type::ArithmeticInteger, "SEL_R"),
             INST("0011100-10100---", Id::SEL_IMM, Type::ArithmeticInteger, "SEL_IMM"),
+            INST("0101101111011---", Id::LEA_R2, Type::ArithmeticInteger, "LEA_R2"),
+            INST("0101101111010---", Id::LEA_R1, Type::ArithmeticInteger, "LEA_R1"),
+            INST("001101101101----", Id::LEA_IMM, Type::ArithmeticInteger, "LEA_IMM"),
+            INST("010010111101----", Id::LEA_RZ, Type::ArithmeticInteger, "LEA_RZ"),
+            INST("00011000--------", Id::LEA_HI, Type::ArithmeticInteger, "LEA_HI"),
             INST("0101000010000---", Id::MUFU, Type::Arithmetic, "MUFU"),
             INST("0100110010010---", Id::RRO_C, Type::Arithmetic, "RRO_C"),
             INST("0101110010010---", Id::RRO_R, Type::Arithmetic, "RRO_R"),
@@ -1012,6 +1087,7 @@ private:
             INST("010110110101----", Id::ISET_R, Type::IntegerSet, "ISET_R"),
             INST("010010110101----", Id::ISET_C, Type::IntegerSet, "ISET_C"),
             INST("0011011-0101----", Id::ISET_IMM, Type::IntegerSet, "ISET_IMM"),
+            INST("0101000010001---", Id::PSET, Type::PredicateSetRegister, "PSET"),
             INST("0101000010010---", Id::PSETP, Type::PredicateSetPredicate, "PSETP"),
             INST("0011011-00------", Id::XMAD_IMM, Type::Xmad, "XMAD_IMM"),
             INST("0100111---------", Id::XMAD_CR, Type::Xmad, "XMAD_CR"),

src/video_core/engines/shader_header.h

@@ -1,103 +0,0 @@
-// Copyright 2018 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include "common/bit_field.h"
-#include "common/common_funcs.h"
-#include "common/common_types.h"
-
-namespace Tegra::Shader {
-
-enum class OutputTopology : u32 {
-    PointList = 1,
-    LineStrip = 6,
-    TriangleStrip = 7,
-};
-
-// Documentation in:
-// http://download.nvidia.com/open-gpu-doc/Shader-Program-Header/1/Shader-Program-Header.html#ImapTexture
-struct Header {
-    union {
-        BitField<0, 5, u32> sph_type;
-        BitField<5, 5, u32> version;
-        BitField<10, 4, u32> shader_type;
-        BitField<14, 1, u32> mrt_enable;
-        BitField<15, 1, u32> kills_pixels;
-        BitField<16, 1, u32> does_global_store;
-        BitField<17, 4, u32> sass_version;
-        BitField<21, 5, u32> reserved;
-        BitField<26, 1, u32> does_load_or_store;
-        BitField<27, 1, u32> does_fp64;
-        BitField<28, 4, u32> stream_out_mask;
-    } common0;
-
-    union {
-        BitField<0, 24, u32> shader_local_memory_low_size;
-        BitField<24, 8, u32> per_patch_attribute_count;
-    } common1;
-
-    union {
-        BitField<0, 24, u32> shader_local_memory_high_size;
-        BitField<24, 8, u32> threads_per_input_primitive;
-    } common2;
-
-    union {
-        BitField<0, 24, u32> shader_local_memory_crs_size;
-        BitField<24, 4, OutputTopology> output_topology;
-        BitField<28, 4, u32> reserved;
-    } common3;
-
-    union {
-        BitField<0, 12, u32> max_output_vertices;
-        BitField<12, 8, u32> store_req_start; // NOTE: not used by geometry shaders.
-        BitField<24, 4, u32> reserved;
-        BitField<12, 8, u32> store_req_end; // NOTE: not used by geometry shaders.
-    } common4;
-
-    union {
-        struct {
-            INSERT_PADDING_BYTES(3);  // ImapSystemValuesA
-            INSERT_PADDING_BYTES(1);  // ImapSystemValuesB
-            INSERT_PADDING_BYTES(16); // ImapGenericVector[32]
-            INSERT_PADDING_BYTES(2);  // ImapColor
-            INSERT_PADDING_BYTES(2);  // ImapSystemValuesC
-            INSERT_PADDING_BYTES(5);  // ImapFixedFncTexture[10]
-            INSERT_PADDING_BYTES(1);  // ImapReserved
-            INSERT_PADDING_BYTES(3);  // OmapSystemValuesA
-            INSERT_PADDING_BYTES(1);  // OmapSystemValuesB
-            INSERT_PADDING_BYTES(16); // OmapGenericVector[32]
-            INSERT_PADDING_BYTES(2);  // OmapColor
-            INSERT_PADDING_BYTES(2);  // OmapSystemValuesC
-            INSERT_PADDING_BYTES(5);  // OmapFixedFncTexture[10]
-            INSERT_PADDING_BYTES(1);  // OmapReserved
-        } vtg;
-
-        struct {
-            INSERT_PADDING_BYTES(3);  // ImapSystemValuesA
-            INSERT_PADDING_BYTES(1);  // ImapSystemValuesB
-            INSERT_PADDING_BYTES(32); // ImapGenericVector[32]
-            INSERT_PADDING_BYTES(2);  // ImapColor
-            INSERT_PADDING_BYTES(2);  // ImapSystemValuesC
-            INSERT_PADDING_BYTES(10); // ImapFixedFncTexture[10]
-            INSERT_PADDING_BYTES(2);  // ImapReserved
-            struct {
-                u32 target;
-                union {
-                    BitField<0, 1, u32> sample_mask;
-                    BitField<1, 1, u32> depth;
-                    BitField<2, 30, u32> reserved;
-                };
-            } omap;
-            bool IsColorComponentOutputEnabled(u32 render_target, u32 component) const {
-                const u32 bit = render_target * 4 + component;
-                return omap.target & (1 << bit);
-            }
-        } ps;
-    };
-};
-
-static_assert(sizeof(Header) == 0x50, "Incorrect structure size");
-
-} // namespace Tegra::Shader

src/video_core/renderer_base.cpp

@@ -19,6 +19,7 @@ void RendererBase::RefreshBaseSettings() {
     UpdateCurrentFramebufferLayout();
 
     renderer_settings.use_framelimiter = Settings::values.use_frame_limit;
+    renderer_settings.set_background_color = true;
 }
 
 void RendererBase::UpdateCurrentFramebufferLayout() {

src/video_core/renderer_base.h

@@ -19,6 +19,7 @@ namespace VideoCore {
 
 struct RendererSettings {
     std::atomic_bool use_framelimiter{false};
+    std::atomic_bool set_background_color{false};
 };
 
 class RendererBase : NonCopyable {

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -3,6 +3,7 @@
 // Refer to the license.txt file included.
 
 #include <algorithm>
+#include <array>
 #include <memory>
 #include <string>
 #include <string_view>
@@ -58,6 +59,8 @@ RasterizerOpenGL::RasterizerOpenGL(Core::Frontend::EmuWindow& window, ScreenInfo
 
         if (extension == "GL_ARB_direct_state_access") {
             has_ARB_direct_state_access = true;
+        } else if (extension == "GL_ARB_multi_bind") {
+            has_ARB_multi_bind = true;
         } else if (extension == "GL_ARB_separate_shader_objects") {
             has_ARB_separate_shader_objects = true;
         } else if (extension == "GL_ARB_vertex_attrib_binding") {
@@ -644,12 +647,23 @@ u32 RasterizerOpenGL::SetupConstBuffers(Maxwell::ShaderStage stage, Shader& shad
     const auto& shader_stage = maxwell3d.state.shader_stages[static_cast<size_t>(stage)];
     const auto& entries = shader->GetShaderEntries().const_buffer_entries;
 
+    constexpr u64 max_binds = Tegra::Engines::Maxwell3D::Regs::MaxConstBuffers;
+    std::array<GLuint, max_binds> bind_buffers;
+    std::array<GLintptr, max_binds> bind_offsets;
+    std::array<GLsizeiptr, max_binds> bind_sizes;
+
+    ASSERT_MSG(entries.size() <= max_binds, "Exceeded expected number of binding points.");
+
     // Upload only the enabled buffers from the 16 constbuffers of each shader stage
     for (u32 bindpoint = 0; bindpoint < entries.size(); ++bindpoint) {
         const auto& used_buffer = entries[bindpoint];
         const auto& buffer = shader_stage.const_buffers[used_buffer.GetIndex()];
 
         if (!buffer.enabled) {
+            // With disabled buffers set values as zero to unbind them
+            bind_buffers[bindpoint] = 0;
+            bind_offsets[bindpoint] = 0;
+            bind_sizes[bindpoint] = 0;
             continue;
         }
 
@@ -677,15 +691,20 @@ u32 RasterizerOpenGL::SetupConstBuffers(Maxwell::ShaderStage stage, Shader& shad
         GLintptr const_buffer_offset = buffer_cache.UploadMemory(
             buffer.address, size, static_cast<size_t>(uniform_buffer_alignment));
 
-        glBindBufferRange(GL_UNIFORM_BUFFER, current_bindpoint + bindpoint,
-                          buffer_cache.GetHandle(), const_buffer_offset, size);
-
         // Now configure the bindpoint of the buffer inside the shader
         glUniformBlockBinding(shader->GetProgramHandle(),
                               shader->GetProgramResourceIndex(used_buffer),
                               current_bindpoint + bindpoint);
+
+        // Prepare values for multibind
+        bind_buffers[bindpoint] = buffer_cache.GetHandle();
+        bind_offsets[bindpoint] = const_buffer_offset;
+        bind_sizes[bindpoint] = size;
     }
 
+    glBindBuffersRange(GL_UNIFORM_BUFFER, current_bindpoint, static_cast<GLsizei>(entries.size()),
+                       bind_buffers.data(), bind_offsets.data(), bind_sizes.data());
+
     return current_bindpoint + static_cast<u32>(entries.size());
 }
 

src/video_core/renderer_opengl/gl_rasterizer.h

@@ -159,6 +159,7 @@ private:
     void SyncLogicOpState();
 
     bool has_ARB_direct_state_access = false;
+    bool has_ARB_multi_bind = false;
     bool has_ARB_separate_shader_objects = false;
     bool has_ARB_vertex_attrib_binding = false;
 

src/video_core/renderer_opengl/gl_rasterizer_cache.cpp

@@ -53,8 +53,6 @@ static VAddr TryGetCpuAddr(Tegra::GPUVAddr gpu_addr) {
     params.width = Common::AlignUp(config.tic.Width(), GetCompressionFactor(params.pixel_format));
     params.height = Common::AlignUp(config.tic.Height(), GetCompressionFactor(params.pixel_format));
     params.unaligned_height = config.tic.Height();
-    params.cache_width = Common::AlignUp(params.width, 8);
-    params.cache_height = Common::AlignUp(params.height, 8);
     params.target = SurfaceTargetFromTextureType(config.tic.texture_type);
 
     switch (params.target) {
@@ -89,8 +87,6 @@ static VAddr TryGetCpuAddr(Tegra::GPUVAddr gpu_addr) {
     params.width = config.width;
     params.height = config.height;
     params.unaligned_height = config.height;
-    params.cache_width = Common::AlignUp(params.width, 8);
-    params.cache_height = Common::AlignUp(params.height, 8);
     params.target = SurfaceTarget::Texture2D;
     params.depth = 1;
     params.size_in_bytes = params.SizeInBytes();
@@ -110,8 +106,6 @@ static VAddr TryGetCpuAddr(Tegra::GPUVAddr gpu_addr) {
     params.width = zeta_width;
     params.height = zeta_height;
     params.unaligned_height = zeta_height;
-    params.cache_width = Common::AlignUp(params.width, 8);
-    params.cache_height = Common::AlignUp(params.height, 8);
     params.target = SurfaceTarget::Texture2D;
     params.depth = 1;
     params.size_in_bytes = params.SizeInBytes();
@@ -122,7 +116,7 @@ static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_form
     {GL_RGBA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8_REV, ComponentType::UNorm, false}, // ABGR8U
     {GL_RGBA8, GL_RGBA, GL_BYTE, ComponentType::SNorm, false},                     // ABGR8S
     {GL_RGBA8UI, GL_RGBA_INTEGER, GL_UNSIGNED_BYTE, ComponentType::UInt, false},   // ABGR8UI
-    {GL_RGB, GL_RGB, GL_UNSIGNED_SHORT_5_6_5_REV, ComponentType::UNorm, false},    // B5G6R5U
+    {GL_RGB8, GL_RGB, GL_UNSIGNED_SHORT_5_6_5_REV, ComponentType::UNorm, false},   // B5G6R5U
     {GL_RGB10_A2, GL_RGBA, GL_UNSIGNED_INT_2_10_10_10_REV, ComponentType::UNorm,
      false}, // A2B10G10R10U
     {GL_RGB5_A1, GL_RGBA, GL_UNSIGNED_SHORT_1_5_5_5_REV, ComponentType::UNorm, false}, // A1B5G5R5U
@@ -477,30 +471,27 @@ CachedSurface::CachedSurface(const SurfaceParams& params)
         // Only pre-create the texture for non-compressed textures.
         switch (params.target) {
         case SurfaceParams::SurfaceTarget::Texture1D:
-            glTexImage1D(SurfaceTargetToGL(params.target), 0, format_tuple.internal_format,
-                         rect.GetWidth(), 0, format_tuple.format, format_tuple.type, nullptr);
+            glTexStorage1D(SurfaceTargetToGL(params.target), 1, format_tuple.internal_format,
+                           rect.GetWidth());
             break;
         case SurfaceParams::SurfaceTarget::Texture2D:
-            glTexImage2D(SurfaceTargetToGL(params.target), 0, format_tuple.internal_format,
-                         rect.GetWidth(), rect.GetHeight(), 0, format_tuple.format,
-                         format_tuple.type, nullptr);
+            glTexStorage2D(SurfaceTargetToGL(params.target), 1, format_tuple.internal_format,
+                           rect.GetWidth(), rect.GetHeight());
             break;
         case SurfaceParams::SurfaceTarget::Texture3D:
         case SurfaceParams::SurfaceTarget::Texture2DArray:
-            glTexImage3D(SurfaceTargetToGL(params.target), 0, format_tuple.internal_format,
-                         rect.GetWidth(), rect.GetHeight(), params.depth, 0, format_tuple.format,
-                         format_tuple.type, nullptr);
+            glTexStorage3D(SurfaceTargetToGL(params.target), 1, format_tuple.internal_format,
+                           rect.GetWidth(), rect.GetHeight(), params.depth);
             break;
         default:
             LOG_CRITICAL(Render_OpenGL, "Unimplemented surface target={}",
                          static_cast<u32>(params.target));
             UNREACHABLE();
-            glTexImage2D(GL_TEXTURE_2D, 0, format_tuple.internal_format, rect.GetWidth(),
-                         rect.GetHeight(), 0, format_tuple.format, format_tuple.type, nullptr);
+            glTexStorage2D(GL_TEXTURE_2D, 1, format_tuple.internal_format, rect.GetWidth(),
+                           rect.GetHeight());
         }
     }
 
-    glTexParameteri(SurfaceTargetToGL(params.target), GL_TEXTURE_MAX_LEVEL, 0);
     glTexParameteri(SurfaceTargetToGL(params.target), GL_TEXTURE_MIN_FILTER, GL_LINEAR);
     glTexParameteri(SurfaceTargetToGL(params.target), GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
     glTexParameteri(SurfaceTargetToGL(params.target), GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
@@ -817,16 +808,20 @@ Surface RasterizerCacheOpenGL::RecreateSurface(const Surface& surface,
     // Get a new surface with the new parameters, and blit the previous surface to it
     Surface new_surface{GetUncachedSurface(new_params)};
 
-    // If format is unchanged, we can do a faster blit without reinterpreting pixel data
-    if (params.pixel_format == new_params.pixel_format) {
+    if (params.pixel_format == new_params.pixel_format ||
+        !Settings::values.use_accurate_framebuffers) {
+        // If the format is the same, just do a framebuffer blit. This is significantly faster than
+        // using PBOs. The is also likely less accurate, as textures will be converted rather than
+        // reinterpreted.
+
         BlitTextures(surface->Texture().handle, params.GetRect(), new_surface->Texture().handle,
                      params.GetRect(), params.type, read_framebuffer.handle,
                      draw_framebuffer.handle);
-        return new_surface;
-    }
+    } else {
+        // When use_accurate_framebuffers setting is enabled, perform a more accurate surface copy,
+        // where pixels are reinterpreted as a new format (without conversion). This code path uses
+        // OpenGL PBOs and is quite slow.
 
-    // When using accurate framebuffers, always copy old data to new surface, regardless of format
-    if (Settings::values.use_accurate_framebuffers) {
         auto source_format = GetFormatTuple(params.pixel_format, params.component_type);
         auto dest_format = GetFormatTuple(new_params.pixel_format, new_params.component_type);
 

src/video_core/renderer_opengl/gl_rasterizer_cache.h

@@ -316,6 +316,8 @@ struct SurfaceParams {
             return PixelFormat::R11FG11FB10F;
         case Tegra::RenderTargetFormat::B5G6R5_UNORM:
             return PixelFormat::B5G6R5U;
+        case Tegra::RenderTargetFormat::BGR5A1_UNORM:
+            return PixelFormat::A1B5G5R5U;
         case Tegra::RenderTargetFormat::RGBA32_UINT:
             return PixelFormat::RGBA32UI;
         case Tegra::RenderTargetFormat::R8_UNORM:
@@ -576,6 +578,7 @@ struct SurfaceParams {
         case Tegra::RenderTargetFormat::RG16_UNORM:
         case Tegra::RenderTargetFormat::R16_UNORM:
         case Tegra::RenderTargetFormat::B5G6R5_UNORM:
+        case Tegra::RenderTargetFormat::BGR5A1_UNORM:
         case Tegra::RenderTargetFormat::RG8_UNORM:
         case Tegra::RenderTargetFormat::RGBA16_UNORM:
             return ComponentType::UNorm;
@@ -680,8 +683,8 @@ struct SurfaceParams {
 
     /// Checks if surfaces are compatible for caching
     bool IsCompatibleSurface(const SurfaceParams& other) const {
-        return std::tie(pixel_format, type, cache_width, cache_height) ==
-               std::tie(other.pixel_format, other.type, other.cache_width, other.cache_height);
+        return std::tie(pixel_format, type, width, height) ==
+               std::tie(other.pixel_format, other.type, other.width, other.height);
     }
 
     VAddr addr;
@@ -696,10 +699,6 @@ struct SurfaceParams {
     u32 unaligned_height;
     size_t size_in_bytes;
     SurfaceTarget target;
-
-    // Parameters used for caching only
-    u32 cache_width;
-    u32 cache_height;
 };
 
 }; // namespace OpenGL

src/video_core/renderer_opengl/gl_shader_decompiler.cpp

@@ -12,7 +12,6 @@
 #include "common/assert.h"
 #include "common/common_types.h"
 #include "video_core/engines/shader_bytecode.h"
-#include "video_core/engines/shader_header.h"
 #include "video_core/renderer_opengl/gl_rasterizer.h"
 #include "video_core/renderer_opengl/gl_shader_decompiler.h"
 
@@ -27,7 +26,7 @@ using Tegra::Shader::Sampler;
 using Tegra::Shader::SubOp;
 
 constexpr u32 PROGRAM_END = MAX_PROGRAM_CODE_LENGTH;
-constexpr u32 PROGRAM_HEADER_SIZE = sizeof(Tegra::Shader::Header);
+constexpr u32 PROGRAM_HEADER_SIZE = 0x50;
 
 class DecompileFail : public std::runtime_error {
 public:
@@ -675,7 +674,7 @@ public:
                   u32 main_offset, Maxwell3D::Regs::ShaderStage stage, const std::string& suffix)
         : subroutines(subroutines), program_code(program_code), main_offset(main_offset),
           stage(stage), suffix(suffix) {
-        std::memcpy(&header, program_code.data(), sizeof(Tegra::Shader::Header));
+
         Generate(suffix);
     }
 
@@ -689,6 +688,23 @@ public:
     }
 
 private:
+    // Shader program header for a Fragment Shader.
+    struct FragmentHeader {
+        INSERT_PADDING_WORDS(5);
+        INSERT_PADDING_WORDS(13);
+        u32 enabled_color_outputs;
+        union {
+            BitField<0, 1, u32> writes_samplemask;
+            BitField<1, 1, u32> writes_depth;
+        };
+
+        bool IsColorComponentOutputEnabled(u32 render_target, u32 component) const {
+            const u32 bit = render_target * 4 + component;
+            return enabled_color_outputs & (1 << bit);
+        }
+    };
+    static_assert(sizeof(FragmentHeader) == PROGRAM_HEADER_SIZE, "FragmentHeader size is wrong");
+
     /// Gets the Subroutine object corresponding to the specified address.
     const Subroutine& GetSubroutine(u32 begin, u32 end) const {
         const auto iter = subroutines.find(Subroutine{begin, end, suffix});
@@ -994,8 +1010,10 @@ private:
     /// Writes the output values from a fragment shader to the corresponding GLSL output variables.
     void EmitFragmentOutputsWrite() {
         ASSERT(stage == Maxwell3D::Regs::ShaderStage::Fragment);
+        FragmentHeader header;
+        std::memcpy(&header, program_code.data(), PROGRAM_HEADER_SIZE);
 
-        ASSERT_MSG(header.ps.omap.sample_mask == 0, "Samplemask write is unimplemented");
+        ASSERT_MSG(header.writes_samplemask == 0, "Samplemask write is unimplemented");
 
         // Write the color outputs using the data in the shader registers, disabled
         // rendertargets/components are skipped in the register assignment.
@@ -1004,7 +1022,7 @@ private:
              ++render_target) {
             // TODO(Subv): Figure out how dual-source blending is configured in the Switch.
             for (u32 component = 0; component < 4; ++component) {
-                if (header.ps.IsColorComponentOutputEnabled(render_target, component)) {
+                if (header.IsColorComponentOutputEnabled(render_target, component)) {
                     shader.AddLine(fmt::format("FragColor{}[{}] = {};", render_target, component,
                                                regs.GetRegisterAsFloat(current_reg)));
                     ++current_reg;
@@ -1012,7 +1030,7 @@ private:
             }
         }
 
-        if (header.ps.omap.depth) {
+        if (header.writes_depth) {
             // The depth output is always 2 registers after the last color output, and current_reg
             // already contains one past the last color register.
 
@@ -1487,6 +1505,73 @@ private:
                                           1, 1);
                 break;
             }
+            case OpCode::Id::LEA_R2:
+            case OpCode::Id::LEA_R1:
+            case OpCode::Id::LEA_IMM:
+            case OpCode::Id::LEA_RZ:
+            case OpCode::Id::LEA_HI: {
+                std::string op_a;
+                std::string op_b;
+                std::string op_c;
+
+                switch (opcode->GetId()) {
+                case OpCode::Id::LEA_R2: {
+                    op_a = regs.GetRegisterAsInteger(instr.gpr20);
+                    op_b = regs.GetRegisterAsInteger(instr.gpr39);
+                    op_c = std::to_string(instr.lea.r2.entry_a);
+                    break;
+                }
+
+                case OpCode::Id::LEA_R1: {
+                    const bool neg = instr.lea.r1.neg != 0;
+                    op_a = regs.GetRegisterAsInteger(instr.gpr8);
+                    if (neg)
+                        op_a = "-(" + op_a + ')';
+                    op_b = regs.GetRegisterAsInteger(instr.gpr20);
+                    op_c = std::to_string(instr.lea.r1.entry_a);
+                    break;
+                }
+
+                case OpCode::Id::LEA_IMM: {
+                    const bool neg = instr.lea.imm.neg != 0;
+                    op_b = regs.GetRegisterAsInteger(instr.gpr8);
+                    if (neg)
+                        op_b = "-(" + op_b + ')';
+                    op_a = std::to_string(instr.lea.imm.entry_a);
+                    op_c = std::to_string(instr.lea.imm.entry_b);
+                    break;
+                }
+
+                case OpCode::Id::LEA_RZ: {
+                    const bool neg = instr.lea.rz.neg != 0;
+                    op_b = regs.GetRegisterAsInteger(instr.gpr8);
+                    if (neg)
+                        op_b = "-(" + op_b + ')';
+                    op_a = regs.GetUniform(instr.lea.rz.cb_index, instr.lea.rz.cb_offset,
+                                           GLSLRegister::Type::Integer);
+                    op_c = std::to_string(instr.lea.rz.entry_a);
+
+                    break;
+                }
+
+                case OpCode::Id::LEA_HI:
+                default: {
+                    op_b = regs.GetRegisterAsInteger(instr.gpr8);
+                    op_a = std::to_string(instr.lea.imm.entry_a);
+                    op_c = std::to_string(instr.lea.imm.entry_b);
+                    LOG_CRITICAL(HW_GPU, "Unhandled LEA subinstruction: {}", opcode->GetName());
+                    UNREACHABLE();
+                }
+                }
+                if (instr.lea.pred48 != static_cast<u64>(Pred::UnusedIndex)) {
+                    LOG_ERROR(HW_GPU, "Unhandled LEA Predicate");
+                    UNREACHABLE();
+                }
+                const std::string value = '(' + op_a + " + (" + op_b + "*(1 << " + op_c + ")))";
+                regs.SetRegisterToInteger(instr.gpr0, true, 0, value, 1, 1);
+
+                break;
+            }
             default: {
                 LOG_CRITICAL(HW_GPU, "Unhandled ArithmeticInteger instruction: {}",
                              opcode->GetName());
@@ -1768,15 +1853,47 @@ private:
                     coord = "vec2 coords = vec2(" + x + ", " + y + ");";
                     texture_type = Tegra::Shader::TextureType::Texture2D;
                 }
+                // TODO: make sure coordinates are always indexed to gpr8 and gpr20 is always bias
+                // or lod.
+                const std::string op_c = regs.GetRegisterAsFloat(instr.gpr20);
 
                 const std::string sampler = GetSampler(instr.sampler, texture_type, false);
                 // Add an extra scope and declare the texture coords inside to prevent
                 // overwriting them in case they are used as outputs of the texs instruction.
+
                 shader.AddLine("{");
                 ++shader.scope;
                 shader.AddLine(coord);
-                const std::string texture = "texture(" + sampler + ", coords)";
+                std::string texture;
 
+                switch (instr.tex.process_mode) {
+                case Tegra::Shader::TextureProcessMode::None: {
+                    texture = "texture(" + sampler + ", coords)";
+                    break;
+                }
+                case Tegra::Shader::TextureProcessMode::LZ: {
+                    texture = "textureLod(" + sampler + ", coords, 0.0)";
+                    break;
+                }
+                case Tegra::Shader::TextureProcessMode::LB:
+                case Tegra::Shader::TextureProcessMode::LBA: {
+                    // TODO: Figure if A suffix changes the equation at all.
+                    texture = "texture(" + sampler + ", coords, " + op_c + ')';
+                    break;
+                }
+                case Tegra::Shader::TextureProcessMode::LL:
+                case Tegra::Shader::TextureProcessMode::LLA: {
+                    // TODO: Figure if A suffix changes the equation at all.
+                    texture = "textureLod(" + sampler + ", coords, " + op_c + ')';
+                    break;
+                }
+                default: {
+                    texture = "texture(" + sampler + ", coords)";
+                    LOG_CRITICAL(HW_GPU, "Unhandled texture process mode {}",
+                                 static_cast<u32>(instr.tex.process_mode.Value()));
+                    UNREACHABLE();
+                }
+                }
                 size_t dest_elem{};
                 for (size_t elem = 0; elem < 4; ++elem) {
                     if (!instr.tex.IsComponentEnabled(elem)) {
@@ -2069,6 +2186,30 @@ private:
             }
             break;
         }
+        case OpCode::Type::PredicateSetRegister: {
+            const std::string op_a =
+                GetPredicateCondition(instr.pset.pred12, instr.pset.neg_pred12 != 0);
+            const std::string op_b =
+                GetPredicateCondition(instr.pset.pred29, instr.pset.neg_pred29 != 0);
+
+            const std::string second_pred =
+                GetPredicateCondition(instr.pset.pred39, instr.pset.neg_pred39 != 0);
+
+            const std::string combiner = GetPredicateCombiner(instr.pset.op);
+
+            const std::string predicate =
+                '(' + op_a + ") " + GetPredicateCombiner(instr.pset.cond) + " (" + op_b + ')';
+            const std::string result = '(' + predicate + ") " + combiner + " (" + second_pred + ')';
+            if (instr.pset.bf == 0) {
+                const std::string value = '(' + result + ") ? 0xFFFFFFFF : 0";
+                regs.SetRegisterToInteger(instr.gpr0, false, 0, value, 1, 1);
+            } else {
+                const std::string value = '(' + result + ") ? 1.0 : 0.0";
+                regs.SetRegisterToFloat(instr.gpr0, 0, value, 1, 1);
+            }
+
+            break;
+        }
         case OpCode::Type::PredicateSetPredicate: {
             const std::string op_a =
                 GetPredicateCondition(instr.psetp.pred12, instr.psetp.neg_pred12 != 0);
@@ -2484,7 +2625,6 @@ private:
 private:
     const std::set<Subroutine>& subroutines;
     const ProgramCode& program_code;
-    Tegra::Shader::Header header;
     const u32 main_offset;
     Maxwell3D::Regs::ShaderStage stage;
     const std::string& suffix;

src/video_core/renderer_opengl/renderer_opengl.cpp

@@ -369,6 +369,12 @@ void RendererOpenGL::DrawScreenTriangles(const ScreenInfo& screen_info, float x,
  * Draws the emulated screens to the emulator window.
  */
 void RendererOpenGL::DrawScreen() {
+    if (renderer_settings.set_background_color) {
+        // Update background color before drawing
+        glClearColor(Settings::values.bg_red, Settings::values.bg_green, Settings::values.bg_blue,
+                     0.0f);
+    }
+
     const auto& layout = render_window.GetFramebufferLayout();
     const auto& screen = layout.screen;
 

src/yuzu/configuration/config.cpp

@@ -95,6 +95,8 @@ void Config::ReadValues() {
 
     qt_config->beginGroup("Audio");
     Settings::values.sink_id = qt_config->value("output_engine", "auto").toString().toStdString();
+    Settings::values.enable_audio_stretching =
+        qt_config->value("enable_audio_stretching", true).toBool();
     Settings::values.audio_device_id =
         qt_config->value("output_device", "auto").toString().toStdString();
     Settings::values.volume = qt_config->value("volume", 1).toFloat();
@@ -230,6 +232,7 @@ void Config::SaveValues() {
 
     qt_config->beginGroup("Audio");
     qt_config->setValue("output_engine", QString::fromStdString(Settings::values.sink_id));
+    qt_config->setValue("enable_audio_stretching", Settings::values.enable_audio_stretching);
     qt_config->setValue("output_device", QString::fromStdString(Settings::values.audio_device_id));
     qt_config->setValue("volume", Settings::values.volume);
     qt_config->endGroup();

src/yuzu/configuration/configure_audio.cpp

@@ -46,6 +46,8 @@ void ConfigureAudio::setConfiguration() {
     }
     ui->output_sink_combo_box->setCurrentIndex(new_sink_index);
 
+    ui->toggle_audio_stretching->setChecked(Settings::values.enable_audio_stretching);
+
     // The device list cannot be pre-populated (nor listed) until the output sink is known.
     updateAudioDevices(new_sink_index);
 
@@ -67,6 +69,7 @@ void ConfigureAudio::applyConfiguration() {
     Settings::values.sink_id =
         ui->output_sink_combo_box->itemText(ui->output_sink_combo_box->currentIndex())
             .toStdString();
+    Settings::values.enable_audio_stretching = ui->toggle_audio_stretching->isChecked();
     Settings::values.audio_device_id =
         ui->audio_device_combo_box->itemText(ui->audio_device_combo_box->currentIndex())
             .toStdString();

src/yuzu/configuration/configure_audio.ui

@@ -31,6 +31,16 @@
         </item>
        </layout>
       </item>
+       <item>
+         <widget class="QCheckBox" name="toggle_audio_stretching">
+           <property name="toolTip">
+             <string>This post-processing effect adjusts audio speed to match emulation speed and helps prevent audio stutter. This however increases audio latency.</string>
+           </property>
+           <property name="text">
+             <string>Enable audio stretching</string>
+           </property>
+         </widget>
+       </item>
       <item>
        <layout class="QHBoxLayout">
         <item>

src/yuzu/configuration/configure_gamelist.cpp

@@ -2,47 +2,51 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#include "core/core.h"
+#include <array>
+#include <utility>
+
+#include "common/common_types.h"
 #include "core/settings.h"
 #include "ui_configure_gamelist.h"
-#include "ui_settings.h"
 #include "yuzu/configuration/configure_gamelist.h"
+#include "yuzu/ui_settings.h"
+
+namespace {
+constexpr std::array<std::pair<u32, const char*>, 5> default_icon_sizes{{
+    std::make_pair(0, QT_TR_NOOP("None")),
+    std::make_pair(32, QT_TR_NOOP("Small (32x32)")),
+    std::make_pair(64, QT_TR_NOOP("Standard (64x64)")),
+    std::make_pair(128, QT_TR_NOOP("Large (128x128)")),
+    std::make_pair(256, QT_TR_NOOP("Full Size (256x256)")),
+}};
+
+constexpr std::array<const char*, 4> row_text_names{{
+    QT_TR_NOOP("Filename"),
+    QT_TR_NOOP("Filetype"),
+    QT_TR_NOOP("Title ID"),
+    QT_TR_NOOP("Title Name"),
+}};
+} // Anonymous namespace
 
 ConfigureGameList::ConfigureGameList(QWidget* parent)
     : QWidget(parent), ui(new Ui::ConfigureGameList) {
     ui->setupUi(this);
 
-    static const std::vector<std::pair<u32, std::string>> default_icon_sizes{
-        std::make_pair(0, "None"),        std::make_pair(32, "Small"),
-        std::make_pair(64, "Standard"),   std::make_pair(128, "Large"),
-        std::make_pair(256, "Full Size"),
-    };
-
-    for (const auto& size : default_icon_sizes) {
-        ui->icon_size_combobox->addItem(QString::fromStdString(size.second + " (" +
-                                                               std::to_string(size.first) + "x" +
-                                                               std::to_string(size.first) + ")"),
-                                        size.first);
-    }
-
-    static const std::vector<std::string> row_text_names{
-        "Filename",
-        "Filetype",
-        "Title ID",
-        "Title Name",
-    };
-
-    for (size_t i = 0; i < row_text_names.size(); ++i) {
-        ui->row_1_text_combobox->addItem(QString::fromStdString(row_text_names[i]),
-                                         QVariant::fromValue(i));
-        ui->row_2_text_combobox->addItem(QString::fromStdString(row_text_names[i]),
-                                         QVariant::fromValue(i));
-    }
+    InitializeIconSizeComboBox();
+    InitializeRowComboBoxes();
 
     this->setConfiguration();
 }
 
-ConfigureGameList::~ConfigureGameList() {}
+ConfigureGameList::~ConfigureGameList() = default;
+
+void ConfigureGameList::applyConfiguration() {
+    UISettings::values.show_unknown = ui->show_unknown->isChecked();
+    UISettings::values.icon_size = ui->icon_size_combobox->currentData().toUInt();
+    UISettings::values.row_1_text_id = ui->row_1_text_combobox->currentData().toUInt();
+    UISettings::values.row_2_text_id = ui->row_2_text_combobox->currentData().toUInt();
+    Settings::Apply();
+}
 
 void ConfigureGameList::setConfiguration() {
     ui->show_unknown->setChecked(UISettings::values.show_unknown);
@@ -54,10 +58,39 @@ void ConfigureGameList::setConfiguration() {
         ui->row_2_text_combobox->findData(UISettings::values.row_2_text_id));
 }
 
-void ConfigureGameList::applyConfiguration() {
-    UISettings::values.show_unknown = ui->show_unknown->isChecked();
-    UISettings::values.icon_size = ui->icon_size_combobox->currentData().toUInt();
-    UISettings::values.row_1_text_id = ui->row_1_text_combobox->currentData().toUInt();
-    UISettings::values.row_2_text_id = ui->row_2_text_combobox->currentData().toUInt();
-    Settings::Apply();
+void ConfigureGameList::changeEvent(QEvent* event) {
+    if (event->type() == QEvent::LanguageChange) {
+        RetranslateUI();
+        return;
+    }
+
+    QWidget::changeEvent(event);
+}
+
+void ConfigureGameList::RetranslateUI() {
+    ui->retranslateUi(this);
+
+    for (int i = 0; i < ui->icon_size_combobox->count(); i++) {
+        ui->icon_size_combobox->setItemText(i, tr(default_icon_sizes[i].second));
+    }
+
+    for (int i = 0; i < ui->row_1_text_combobox->count(); i++) {
+        const QString name = tr(row_text_names[i]);
+
+        ui->row_1_text_combobox->setItemText(i, name);
+        ui->row_2_text_combobox->setItemText(i, name);
+    }
+}
+
+void ConfigureGameList::InitializeIconSizeComboBox() {
+    for (const auto& size : default_icon_sizes) {
+        ui->icon_size_combobox->addItem(size.second, size.first);
+    }
+}
+
+void ConfigureGameList::InitializeRowComboBoxes() {
+    for (size_t i = 0; i < row_text_names.size(); ++i) {
+        ui->row_1_text_combobox->addItem(row_text_names[i], QVariant::fromValue(i));
+        ui->row_2_text_combobox->addItem(row_text_names[i], QVariant::fromValue(i));
+    }
 }

src/yuzu/configuration/configure_gamelist.h

@@ -23,6 +23,11 @@ public:
 private:
     void setConfiguration();
 
-private:
+    void changeEvent(QEvent*) override;
+    void RetranslateUI();
+
+    void InitializeIconSizeComboBox();
+    void InitializeRowComboBoxes();
+
     std::unique_ptr<Ui::ConfigureGameList> ui;
 };

src/yuzu/configuration/configure_graphics.cpp

@@ -2,6 +2,7 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <QColorDialog>
 #include "core/core.h"
 #include "core/settings.h"
 #include "ui_configure_graphics.h"
@@ -16,6 +17,14 @@ ConfigureGraphics::ConfigureGraphics(QWidget* parent)
     ui->frame_limit->setEnabled(Settings::values.use_frame_limit);
     connect(ui->toggle_frame_limit, &QCheckBox::stateChanged, ui->frame_limit,
             &QSpinBox::setEnabled);
+    connect(ui->bg_button, &QPushButton::clicked, this, [this] {
+        const QColor new_bg_color = QColorDialog::getColor(bg_color);
+        if (!new_bg_color.isValid())
+            return;
+        bg_color = new_bg_color;
+        ui->bg_button->setStyleSheet(
+            QString("QPushButton { background-color: %1 }").arg(bg_color.name()));
+    });
 }
 
 ConfigureGraphics::~ConfigureGraphics() = default;
@@ -65,6 +74,10 @@ void ConfigureGraphics::setConfiguration() {
     ui->toggle_frame_limit->setChecked(Settings::values.use_frame_limit);
     ui->frame_limit->setValue(Settings::values.frame_limit);
     ui->use_accurate_framebuffers->setChecked(Settings::values.use_accurate_framebuffers);
+    bg_color = QColor::fromRgbF(Settings::values.bg_red, Settings::values.bg_green,
+                                Settings::values.bg_blue);
+    ui->bg_button->setStyleSheet(
+        QString("QPushButton { background-color: %1 }").arg(bg_color.name()));
 }
 
 void ConfigureGraphics::applyConfiguration() {
@@ -73,4 +86,7 @@ void ConfigureGraphics::applyConfiguration() {
     Settings::values.use_frame_limit = ui->toggle_frame_limit->isChecked();
     Settings::values.frame_limit = ui->frame_limit->value();
     Settings::values.use_accurate_framebuffers = ui->use_accurate_framebuffers->isChecked();
+    Settings::values.bg_red = static_cast<float>(bg_color.redF());
+    Settings::values.bg_green = static_cast<float>(bg_color.greenF());
+    Settings::values.bg_blue = static_cast<float>(bg_color.blueF());
 }

src/yuzu/configuration/configure_graphics.h

@@ -25,4 +25,5 @@ private:
 
 private:
     std::unique_ptr<Ui::ConfigureGraphics> ui;
+    QColor bg_color;
 };

src/yuzu/configuration/configure_graphics.ui

@@ -96,6 +96,27 @@
           </item>
          </layout>
         </item>
+         <item>
+           <layout class="QHBoxLayout" name="horizontalLayout_6">
+             <item>
+               <widget class="QLabel" name="bg_label">
+                 <property name="text">
+                   <string>Background Color:</string>
+                 </property>
+               </widget>
+             </item>
+             <item>
+               <widget class="QPushButton" name="bg_button">
+                 <property name="maximumSize">
+                   <size>
+                     <width>40</width>
+                     <height>16777215</height>
+                   </size>
+                 </property>
+               </widget>
+             </item>
+           </layout>
+         </item>
        </layout>
       </widget>
      </item>

src/yuzu/debugger/wait_tree.cpp

@@ -213,35 +213,35 @@ QString WaitTreeThread::GetText() const {
     const auto& thread = static_cast<const Kernel::Thread&>(object);
     QString status;
     switch (thread.status) {
-    case ThreadStatus::Running:
+    case Kernel::ThreadStatus::Running:
         status = tr("running");
         break;
-    case ThreadStatus::Ready:
+    case Kernel::ThreadStatus::Ready:
         status = tr("ready");
         break;
-    case ThreadStatus::WaitHLEEvent:
+    case Kernel::ThreadStatus::WaitHLEEvent:
         status = tr("waiting for HLE return");
         break;
-    case ThreadStatus::WaitSleep:
+    case Kernel::ThreadStatus::WaitSleep:
         status = tr("sleeping");
         break;
-    case ThreadStatus::WaitIPC:
+    case Kernel::ThreadStatus::WaitIPC:
         status = tr("waiting for IPC reply");
         break;
-    case ThreadStatus::WaitSynchAll:
-    case ThreadStatus::WaitSynchAny:
+    case Kernel::ThreadStatus::WaitSynchAll:
+    case Kernel::ThreadStatus::WaitSynchAny:
         status = tr("waiting for objects");
         break;
-    case ThreadStatus::WaitMutex:
+    case Kernel::ThreadStatus::WaitMutex:
         status = tr("waiting for mutex");
         break;
-    case ThreadStatus::WaitArb:
+    case Kernel::ThreadStatus::WaitArb:
         status = tr("waiting for address arbiter");
         break;
-    case ThreadStatus::Dormant:
+    case Kernel::ThreadStatus::Dormant:
         status = tr("dormant");
         break;
-    case ThreadStatus::Dead:
+    case Kernel::ThreadStatus::Dead:
         status = tr("dead");
         break;
     }
@@ -254,23 +254,23 @@ QString WaitTreeThread::GetText() const {
 QColor WaitTreeThread::GetColor() const {
     const auto& thread = static_cast<const Kernel::Thread&>(object);
     switch (thread.status) {
-    case ThreadStatus::Running:
+    case Kernel::ThreadStatus::Running:
         return QColor(Qt::GlobalColor::darkGreen);
-    case ThreadStatus::Ready:
+    case Kernel::ThreadStatus::Ready:
         return QColor(Qt::GlobalColor::darkBlue);
-    case ThreadStatus::WaitHLEEvent:
-    case ThreadStatus::WaitIPC:
+    case Kernel::ThreadStatus::WaitHLEEvent:
+    case Kernel::ThreadStatus::WaitIPC:
         return QColor(Qt::GlobalColor::darkRed);
-    case ThreadStatus::WaitSleep:
+    case Kernel::ThreadStatus::WaitSleep:
         return QColor(Qt::GlobalColor::darkYellow);
-    case ThreadStatus::WaitSynchAll:
-    case ThreadStatus::WaitSynchAny:
-    case ThreadStatus::WaitMutex:
-    case ThreadStatus::WaitArb:
+    case Kernel::ThreadStatus::WaitSynchAll:
+    case Kernel::ThreadStatus::WaitSynchAny:
+    case Kernel::ThreadStatus::WaitMutex:
+    case Kernel::ThreadStatus::WaitArb:
         return QColor(Qt::GlobalColor::red);
-    case ThreadStatus::Dormant:
+    case Kernel::ThreadStatus::Dormant:
         return QColor(Qt::GlobalColor::darkCyan);
-    case ThreadStatus::Dead:
+    case Kernel::ThreadStatus::Dead:
         return QColor(Qt::GlobalColor::gray);
     default:
         return WaitTreeItem::GetColor();
@@ -284,13 +284,13 @@ std::vector<std::unique_ptr<WaitTreeItem>> WaitTreeThread::GetChildren() const {
 
     QString processor;
     switch (thread.processor_id) {
-    case ThreadProcessorId::THREADPROCESSORID_DEFAULT:
+    case Kernel::ThreadProcessorId::THREADPROCESSORID_DEFAULT:
         processor = tr("default");
         break;
-    case ThreadProcessorId::THREADPROCESSORID_0:
-    case ThreadProcessorId::THREADPROCESSORID_1:
-    case ThreadProcessorId::THREADPROCESSORID_2:
-    case ThreadProcessorId::THREADPROCESSORID_3:
+    case Kernel::ThreadProcessorId::THREADPROCESSORID_0:
+    case Kernel::ThreadProcessorId::THREADPROCESSORID_1:
+    case Kernel::ThreadProcessorId::THREADPROCESSORID_2:
+    case Kernel::ThreadProcessorId::THREADPROCESSORID_3:
         processor = tr("core %1").arg(thread.processor_id);
         break;
     default:
@@ -314,8 +314,8 @@ std::vector<std::unique_ptr<WaitTreeItem>> WaitTreeThread::GetChildren() const {
     else
         list.push_back(std::make_unique<WaitTreeText>(tr("not waiting for mutex")));
 
-    if (thread.status == ThreadStatus::WaitSynchAny ||
-        thread.status == ThreadStatus::WaitSynchAll) {
+    if (thread.status == Kernel::ThreadStatus::WaitSynchAny ||
+        thread.status == Kernel::ThreadStatus::WaitSynchAll) {
         list.push_back(std::make_unique<WaitTreeObjectList>(thread.wait_objects,
                                                             thread.IsSleepingOnWaitAll()));
     }

src/yuzu/game_list.cpp

@@ -366,7 +366,7 @@ void GameList::LoadCompatibilityList() {
     QJsonDocument json = QJsonDocument::fromJson(string_content.toUtf8());
     QJsonArray arr = json.array();
 
-    for (const QJsonValue& value : arr) {
+    for (const QJsonValueRef& value : arr) {
         QJsonObject game = value.toObject();
 
         if (game.contains("compatibility") && game["compatibility"].isDouble()) {
@@ -374,9 +374,9 @@ void GameList::LoadCompatibilityList() {
             QString directory = game["directory"].toString();
             QJsonArray ids = game["releases"].toArray();
 
-            for (const QJsonValue& value : ids) {
-                QJsonObject object = value.toObject();
-                QString id = object["id"].toString();
+            for (const QJsonValueRef& id_ref : ids) {
+                QJsonObject id_object = id_ref.toObject();
+                QString id = id_object["id"].toString();
                 compatibility_list.emplace(
                     id.toUpper().toStdString(),
                     std::make_pair(QString::number(compatibility), directory));

src/yuzu/main.cpp

@@ -447,6 +447,10 @@ QStringList GMainWindow::GetUnsupportedGLExtensions() {
         unsupported_ext.append("ARB_texture_mirror_clamp_to_edge");
     if (!GLAD_GL_ARB_base_instance)
         unsupported_ext.append("ARB_base_instance");
+    if (!GLAD_GL_ARB_texture_storage)
+        unsupported_ext.append("ARB_texture_storage");
+    if (!GLAD_GL_ARB_multi_bind)
+        unsupported_ext.append("ARB_multi_bind");
 
     // Extensions required to support some texture formats.
     if (!GLAD_GL_EXT_texture_compression_s3tc)

src/yuzu_cmd/config.cpp

@@ -108,6 +108,8 @@ void Config::ReadValues() {
 
     // Audio
     Settings::values.sink_id = sdl2_config->Get("Audio", "output_engine", "auto");
+    Settings::values.enable_audio_stretching =
+        sdl2_config->GetBoolean("Audio", "enable_audio_stretching", true);
     Settings::values.audio_device_id = sdl2_config->Get("Audio", "output_device", "auto");
     Settings::values.volume = sdl2_config->GetReal("Audio", "volume", 1);
 

src/yuzu_cmd/default_ini.h

@@ -150,6 +150,12 @@ swap_screen =
 # auto (default): Auto-select, null: No audio output, cubeb: Cubeb audio engine (if available)
 output_engine =
 
+# Whether or not to enable the audio-stretching post-processing effect.
+# This effect adjusts audio speed to match emulation speed and helps prevent audio stutter,
+# at the cost of increasing audio latency.
+# 0: No, 1 (default): Yes
+enable_audio_stretching =
+
 # Which audio device to use.
 # auto (default): Auto-select
 output_device =

src/yuzu_cmd/emu_window/emu_window_sdl2.cpp

@@ -94,6 +94,10 @@ bool EmuWindow_SDL2::SupportsRequiredGLExtensions() {
         unsupported_ext.push_back("ARB_texture_mirror_clamp_to_edge");
     if (!GLAD_GL_ARB_base_instance)
         unsupported_ext.push_back("ARB_base_instance");
+    if (!GLAD_GL_ARB_texture_storage)
+        unsupported_ext.push_back("ARB_texture_storage");
+    if (!GLAD_GL_ARB_multi_bind)
+        unsupported_ext.push_back("ARB_multi_bind");
 
     // Extensions required to support some texture formats.
     if (!GLAD_GL_EXT_texture_compression_s3tc)