correct coding style

Require by Toki Tori and Toki Tori 2+

CMakeLists.txt

@@ -1,5 +1,5 @@
-# CMake 3.6 required for FindBoost to define IMPORTED libs properly on unknown Boost versions
-cmake_minimum_required(VERSION 3.6)
+cmake_minimum_required(VERSION 3.7)
+
 list(APPEND CMAKE_MODULE_PATH "${CMAKE_CURRENT_SOURCE_DIR}/CMakeModules")
 list(APPEND CMAKE_MODULE_PATH "${CMAKE_CURRENT_SOURCE_DIR}/externals/cmake-modules")
 include(DownloadExternals)
@@ -66,10 +66,12 @@ if (NOT ENABLE_GENERIC)
         detect_architecture("_M_AMD64" x86_64)
         detect_architecture("_M_IX86" x86)
         detect_architecture("_M_ARM" ARM)
+        detect_architecture("_M_ARM64" ARM64)
     else()
         detect_architecture("__x86_64__" x86_64)
         detect_architecture("__i386__" x86)
         detect_architecture("__arm__" ARM)
+        detect_architecture("__aarch64__" ARM64)
     endif()
 endif()
 
@@ -187,8 +189,8 @@ find_package(Threads REQUIRED)
 if (ENABLE_SDL2)
     if (YUZU_USE_BUNDLED_SDL2)
         # Detect toolchain and platform
-        if (MSVC14 AND ARCHITECTURE_x86_64)
-            set(SDL2_VER "SDL2-2.0.5")
+        if ((MSVC_VERSION GREATER_EQUAL 1910 AND MSVC_VERSION LESS 1920) AND ARCHITECTURE_x86_64)
+            set(SDL2_VER "SDL2-2.0.8")
         else()
             message(FATAL_ERROR "No bundled SDL2 binaries for your toolchain. Disable YUZU_USE_BUNDLED_SDL2 and provide your own.")
         endif()
@@ -220,7 +222,7 @@ if (YUZU_USE_BUNDLED_UNICORN)
     if (MSVC)
         message(STATUS "unicorn not found, falling back to bundled")
         # Detect toolchain and platform
-        if (MSVC14 AND ARCHITECTURE_x86_64)
+        if ((MSVC_VERSION GREATER_EQUAL 1910 AND MSVC_VERSION LESS 1920) AND ARCHITECTURE_x86_64)
             set(UNICORN_VER "unicorn-yuzu")
         else()
             message(FATAL_ERROR "No bundled Unicorn binaries for your toolchain. Disable YUZU_USE_BUNDLED_UNICORN and provide your own.")
@@ -279,7 +281,7 @@ endif()
 
 if (ENABLE_QT)
     if (YUZU_USE_BUNDLED_QT)
-        if (MSVC14 AND ARCHITECTURE_x86_64)
+        if ((MSVC_VERSION GREATER_EQUAL 1910 AND MSVC_VERSION LESS 1920) AND ARCHITECTURE_x86_64)
             set(QT_VER qt-5.10.0-msvc2015_64)
         else()
             message(FATAL_ERROR "No bundled Qt binaries for your toolchain. Disable YUZU_USE_BUNDLED_QT and provide your own.")
@@ -303,7 +305,7 @@ endif()
 # ======================================
 
 IF (APPLE)
-    FIND_LIBRARY(COCOA_LIBRARY Cocoa)           # Umbrella framework for everything GUI-related
+    find_library(COCOA_LIBRARY Cocoa)           # Umbrella framework for everything GUI-related
     set(PLATFORM_LIBRARIES ${COCOA_LIBRARY} ${IOKIT_LIBRARY} ${COREVIDEO_LIBRARY})
 
     if (CMAKE_CXX_COMPILER_ID STREQUAL Clang)

src/audio_core/CMakeLists.txt

@@ -1,4 +1,8 @@
 add_library(audio_core STATIC
+    algorithm/filter.cpp
+    algorithm/filter.h
+    algorithm/interpolate.cpp
+    algorithm/interpolate.h
     audio_out.cpp
     audio_out.h
     audio_renderer.cpp
@@ -7,12 +11,12 @@ add_library(audio_core STATIC
     codec.cpp
     codec.h
     null_sink.h
-    stream.cpp
-    stream.h
     sink.h
     sink_details.cpp
     sink_details.h
     sink_stream.h
+    stream.cpp
+    stream.h
 
     $<$<BOOL:${ENABLE_CUBEB}>:cubeb_sink.cpp cubeb_sink.h>
 )

src/audio_core/algorithm/filter.cpp

@@ -0,0 +1,79 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#define _USE_MATH_DEFINES
+
+#include <algorithm>
+#include <array>
+#include <cmath>
+#include <vector>
+#include "audio_core/algorithm/filter.h"
+#include "common/common_types.h"
+
+namespace AudioCore {
+
+Filter Filter::LowPass(double cutoff, double Q) {
+    const double w0 = 2.0 * M_PI * cutoff;
+    const double sin_w0 = std::sin(w0);
+    const double cos_w0 = std::cos(w0);
+    const double alpha = sin_w0 / (2 * Q);
+
+    const double a0 = 1 + alpha;
+    const double a1 = -2.0 * cos_w0;
+    const double a2 = 1 - alpha;
+    const double b0 = 0.5 * (1 - cos_w0);
+    const double b1 = 1.0 * (1 - cos_w0);
+    const double b2 = 0.5 * (1 - cos_w0);
+
+    return {a0, a1, a2, b0, b1, b2};
+}
+
+Filter::Filter() : Filter(1.0, 0.0, 0.0, 1.0, 0.0, 0.0) {}
+
+Filter::Filter(double a0, double a1, double a2, double b0, double b1, double b2)
+    : a1(a1 / a0), a2(a2 / a0), b0(b0 / a0), b1(b1 / a0), b2(b2 / a0) {}
+
+void Filter::Process(std::vector<s16>& signal) {
+    const size_t num_frames = signal.size() / 2;
+    for (size_t i = 0; i < num_frames; i++) {
+        std::rotate(in.begin(), in.end() - 1, in.end());
+        std::rotate(out.begin(), out.end() - 1, out.end());
+
+        for (size_t ch = 0; ch < channel_count; ch++) {
+            in[0][ch] = signal[i * channel_count + ch];
+
+            out[0][ch] = b0 * in[0][ch] + b1 * in[1][ch] + b2 * in[2][ch] - a1 * out[1][ch] -
+                         a2 * out[2][ch];
+
+            signal[i * 2 + ch] = std::clamp(out[0][ch], -32768.0, 32767.0);
+        }
+    }
+}
+
+/// Calculates the appropriate Q for each biquad in a cascading filter.
+/// @param total_count The total number of biquads to be cascaded.
+/// @param index 0-index of the biquad to calculate the Q value for.
+static double CascadingBiquadQ(size_t total_count, size_t index) {
+    const double pole = M_PI * (2 * index + 1) / (4.0 * total_count);
+    return 1.0 / (2.0 * std::cos(pole));
+}
+
+CascadingFilter CascadingFilter::LowPass(double cutoff, size_t cascade_size) {
+    std::vector<Filter> cascade(cascade_size);
+    for (size_t i = 0; i < cascade_size; i++) {
+        cascade[i] = Filter::LowPass(cutoff, CascadingBiquadQ(cascade_size, i));
+    }
+    return CascadingFilter{std::move(cascade)};
+}
+
+CascadingFilter::CascadingFilter() = default;
+CascadingFilter::CascadingFilter(std::vector<Filter> filters) : filters(std::move(filters)) {}
+
+void CascadingFilter::Process(std::vector<s16>& signal) {
+    for (auto& filter : filters) {
+        filter.Process(signal);
+    }
+}
+
+} // namespace AudioCore

src/audio_core/algorithm/filter.h

@@ -0,0 +1,62 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include <vector>
+#include "common/common_types.h"
+
+namespace AudioCore {
+
+/// Digital biquad filter:
+///
+///          b0 + b1 z^-1 + b2 z^-2
+///  H(z) = ------------------------
+///          a0 + a1 z^-1 + b2 z^-2
+class Filter {
+public:
+    /// Creates a low-pass filter.
+    /// @param cutoff Determines the cutoff frequency. A value from 0.0 to 1.0.
+    /// @param Q Determines the quality factor of this filter.
+    static Filter LowPass(double cutoff, double Q = 0.7071);
+
+    /// Passthrough filter.
+    Filter();
+
+    Filter(double a0, double a1, double a2, double b0, double b1, double b2);
+
+    void Process(std::vector<s16>& signal);
+
+private:
+    static constexpr size_t channel_count = 2;
+
+    /// Coefficients are in normalized form (a0 = 1.0).
+    double a1, a2, b0, b1, b2;
+    /// Input History
+    std::array<std::array<double, channel_count>, 3> in;
+    /// Output History
+    std::array<std::array<double, channel_count>, 3> out;
+};
+
+/// Cascade filters to build up higher-order filters from lower-order ones.
+class CascadingFilter {
+public:
+    /// Creates a cascading low-pass filter.
+    /// @param cutoff Determines the cutoff frequency. A value from 0.0 to 1.0.
+    /// @param cascade_size Number of biquads in cascade.
+    static CascadingFilter LowPass(double cutoff, size_t cascade_size);
+
+    /// Passthrough.
+    CascadingFilter();
+
+    explicit CascadingFilter(std::vector<Filter> filters);
+
+    void Process(std::vector<s16>& signal);
+
+private:
+    std::vector<Filter> filters;
+};
+
+} // namespace AudioCore

src/audio_core/algorithm/interpolate.cpp

@@ -0,0 +1,71 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#define _USE_MATH_DEFINES
+
+#include <algorithm>
+#include <cmath>
+#include <vector>
+#include "audio_core/algorithm/interpolate.h"
+#include "common/common_types.h"
+#include "common/logging/log.h"
+
+namespace AudioCore {
+
+/// The Lanczos kernel
+static double Lanczos(size_t a, double x) {
+    if (x == 0.0)
+        return 1.0;
+    const double px = M_PI * x;
+    return a * std::sin(px) * std::sin(px / a) / (px * px);
+}
+
+std::vector<s16> Interpolate(InterpolationState& state, std::vector<s16> input, double ratio) {
+    if (input.size() < 2)
+        return {};
+
+    if (ratio <= 0) {
+        LOG_CRITICAL(Audio, "Nonsensical interpolation ratio {}", ratio);
+        ratio = 1.0;
+    }
+
+    if (ratio != state.current_ratio) {
+        const double cutoff_frequency = std::min(0.5 / ratio, 0.5 * ratio);
+        state.nyquist = CascadingFilter::LowPass(std::clamp(cutoff_frequency, 0.0, 0.4), 3);
+        state.current_ratio = ratio;
+    }
+    state.nyquist.Process(input);
+
+    constexpr size_t taps = InterpolationState::lanczos_taps;
+    const size_t num_frames = input.size() / 2;
+
+    std::vector<s16> output;
+    output.reserve(static_cast<size_t>(input.size() / ratio + 4));
+
+    double& pos = state.position;
+    auto& h = state.history;
+    for (size_t i = 0; i < num_frames; ++i) {
+        std::rotate(h.begin(), h.end() - 1, h.end());
+        h[0][0] = input[i * 2 + 0];
+        h[0][1] = input[i * 2 + 1];
+
+        while (pos <= 1.0) {
+            double l = 0.0;
+            double r = 0.0;
+            for (size_t j = 0; j < h.size(); j++) {
+                l += Lanczos(taps, pos + j - taps + 1) * h[j][0];
+                r += Lanczos(taps, pos + j - taps + 1) * h[j][1];
+            }
+            output.emplace_back(static_cast<s16>(std::clamp(l, -32768.0, 32767.0)));
+            output.emplace_back(static_cast<s16>(std::clamp(r, -32768.0, 32767.0)));
+
+            pos += ratio;
+        }
+        pos -= 1.0;
+    }
+
+    return output;
+}
+
+} // namespace AudioCore

src/audio_core/algorithm/interpolate.h

@@ -0,0 +1,43 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include <vector>
+#include "audio_core/algorithm/filter.h"
+#include "common/common_types.h"
+
+namespace AudioCore {
+
+struct InterpolationState {
+    static constexpr size_t lanczos_taps = 4;
+    static constexpr size_t history_size = lanczos_taps * 2 - 1;
+
+    double current_ratio = 0.0;
+    CascadingFilter nyquist;
+    std::array<std::array<s16, 2>, history_size> history = {};
+    double position = 0;
+};
+
+/// Interpolates input signal to produce output signal.
+/// @param input The signal to interpolate.
+/// @param ratio Interpolation ratio.
+///              ratio > 1.0 results in fewer output samples.
+///              ratio < 1.0 results in more output samples.
+/// @returns Output signal.
+std::vector<s16> Interpolate(InterpolationState& state, std::vector<s16> input, double ratio);
+
+/// Interpolates input signal to produce output signal.
+/// @param input The signal to interpolate.
+/// @param input_rate The sample rate of input.
+/// @param output_rate The desired sample rate of the output.
+/// @returns Output signal.
+inline std::vector<s16> Interpolate(InterpolationState& state, std::vector<s16> input,
+                                    u32 input_rate, u32 output_rate) {
+    const double ratio = static_cast<double>(input_rate) / static_cast<double>(output_rate);
+    return Interpolate(state, std::move(input), ratio);
+}
+
+} // namespace AudioCore

src/audio_core/audio_renderer.cpp

@@ -2,6 +2,7 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include "audio_core/algorithm/interpolate.h"
 #include "audio_core/audio_renderer.h"
 #include "common/assert.h"
 #include "common/logging/log.h"
@@ -26,6 +27,18 @@ AudioRenderer::AudioRenderer(AudioRendererParameter params,
     QueueMixedBuffer(2);
 }
 
+u32 AudioRenderer::GetSampleRate() const {
+    return worker_params.sample_rate;
+}
+
+u32 AudioRenderer::GetSampleCount() const {
+    return worker_params.sample_count;
+}
+
+u32 AudioRenderer::GetMixBufferCount() const {
+    return worker_params.mix_buffer_count;
+}
+
 std::vector<u8> AudioRenderer::UpdateAudioRenderer(const std::vector<u8>& input_params) {
     // Copy UpdateDataHeader struct
     UpdateDataHeader config{};
@@ -187,6 +200,8 @@ void AudioRenderer::VoiceState::RefreshBuffer() {
         break;
     }
 
+    samples = Interpolate(interp_state, std::move(samples), Info().sample_rate, STREAM_SAMPLE_RATE);
+
     is_refresh_pending = false;
 }
 
@@ -212,7 +227,7 @@ void AudioRenderer::QueueMixedBuffer(Buffer::Tag tag) {
                 break;
             }
 
-            samples_remaining -= samples.size();
+            samples_remaining -= samples.size() / stream->GetNumChannels();
 
             for (const auto& sample : samples) {
                 const s32 buffer_sample{buffer[offset]};

src/audio_core/audio_renderer.h

@@ -8,6 +8,7 @@
 #include <memory>
 #include <vector>
 
+#include "audio_core/algorithm/interpolate.h"
 #include "audio_core/audio_out.h"
 #include "audio_core/codec.h"
 #include "audio_core/stream.h"
@@ -26,7 +27,7 @@ enum class PlayState : u8 {
 struct AudioRendererParameter {
     u32_le sample_rate;
     u32_le sample_count;
-    u32_le unknown_8;
+    u32_le mix_buffer_count;
     u32_le unknown_c;
     u32_le voice_count;
     u32_le sink_count;
@@ -160,6 +161,9 @@ public:
     std::vector<u8> UpdateAudioRenderer(const std::vector<u8>& input_params);
     void QueueMixedBuffer(Buffer::Tag tag);
     void ReleaseAndQueueBuffers();
+    u32 GetSampleRate() const;
+    u32 GetSampleCount() const;
+    u32 GetMixBufferCount() const;
 
 private:
     class VoiceState {
@@ -191,6 +195,7 @@ private:
         size_t wave_index{};
         size_t offset{};
         Codec::ADPCMState adpcm_state{};
+        InterpolationState interp_state{};
         std::vector<s16> samples;
         VoiceOutStatus out_status{};
         VoiceInfo info{};

src/audio_core/cubeb_sink.cpp

@@ -4,6 +4,7 @@
 
 #include <algorithm>
 #include <cstring>
+#include <mutex>
 
 #include "audio_core/cubeb_sink.h"
 #include "audio_core/stream.h"
@@ -66,6 +67,8 @@ public:
             return;
         }
 
+        std::lock_guard lock{queue_mutex};
+
         queue.reserve(queue.size() + samples.size() * GetNumChannels());
 
         if (is_6_channel) {
@@ -94,6 +97,7 @@ private:
     u32 num_channels{};
     bool is_6_channel{};
 
+    std::mutex queue_mutex;
     std::vector<s16> queue;
 
     static long DataCallback(cubeb_stream* stream, void* user_data, const void* input_buffer,
@@ -153,6 +157,8 @@ long SinkStreamImpl::DataCallback(cubeb_stream* stream, void* user_data, const v
         return {};
     }
 
+    std::lock_guard lock{impl->queue_mutex};
+
     const size_t frames_to_write{
         std::min(impl->queue.size() / impl->GetNumChannels(), static_cast<size_t>(num_frames))};
 

src/common/CMakeLists.txt

@@ -29,8 +29,6 @@ add_library(common STATIC
     assert.h
     bit_field.h
     bit_set.h
-    break_points.cpp
-    break_points.h
     cityhash.cpp
     cityhash.h
     color.h
@@ -40,6 +38,8 @@ add_library(common STATIC
     file_util.cpp
     file_util.h
     hash.h
+    hex_util.cpp
+    hex_util.h
     logging/backend.cpp
     logging/backend.h
     logging/filter.cpp

src/common/break_points.cpp

@@ -1,90 +0,0 @@
-// Copyright 2013 Dolphin Emulator Project / 2014 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include <algorithm>
-#include <sstream>
-#include "common/break_points.h"
-
-bool BreakPoints::IsAddressBreakPoint(u32 iAddress) const {
-    auto cond = [&iAddress](const TBreakPoint& bp) { return bp.iAddress == iAddress; };
-    auto it = std::find_if(m_BreakPoints.begin(), m_BreakPoints.end(), cond);
-    return it != m_BreakPoints.end();
-}
-
-bool BreakPoints::IsTempBreakPoint(u32 iAddress) const {
-    auto cond = [&iAddress](const TBreakPoint& bp) {
-        return bp.iAddress == iAddress && bp.bTemporary;
-    };
-    auto it = std::find_if(m_BreakPoints.begin(), m_BreakPoints.end(), cond);
-    return it != m_BreakPoints.end();
-}
-
-BreakPoints::TBreakPointsStr BreakPoints::GetStrings() const {
-    TBreakPointsStr bps;
-    for (auto breakpoint : m_BreakPoints) {
-        if (!breakpoint.bTemporary) {
-            std::stringstream bp;
-            bp << std::hex << breakpoint.iAddress << " " << (breakpoint.bOn ? "n" : "");
-            bps.push_back(bp.str());
-        }
-    }
-
-    return bps;
-}
-
-void BreakPoints::AddFromStrings(const TBreakPointsStr& bps) {
-    for (auto bps_item : bps) {
-        TBreakPoint bp;
-        std::stringstream bpstr;
-        bpstr << std::hex << bps_item;
-        bpstr >> bp.iAddress;
-        bp.bOn = bps_item.find("n") != bps_item.npos;
-        bp.bTemporary = false;
-        Add(bp);
-    }
-}
-
-void BreakPoints::Add(const TBreakPoint& bp) {
-    if (!IsAddressBreakPoint(bp.iAddress)) {
-        m_BreakPoints.push_back(bp);
-        // if (jit)
-        //    jit->GetBlockCache()->InvalidateICache(bp.iAddress, 4);
-    }
-}
-
-void BreakPoints::Add(u32 em_address, bool temp) {
-    if (!IsAddressBreakPoint(em_address)) // only add new addresses
-    {
-        TBreakPoint pt; // breakpoint settings
-        pt.bOn = true;
-        pt.bTemporary = temp;
-        pt.iAddress = em_address;
-
-        m_BreakPoints.push_back(pt);
-
-        // if (jit)
-        //    jit->GetBlockCache()->InvalidateICache(em_address, 4);
-    }
-}
-
-void BreakPoints::Remove(u32 em_address) {
-    auto cond = [&em_address](const TBreakPoint& bp) { return bp.iAddress == em_address; };
-    auto it = std::find_if(m_BreakPoints.begin(), m_BreakPoints.end(), cond);
-    if (it != m_BreakPoints.end())
-        m_BreakPoints.erase(it);
-}
-
-void BreakPoints::Clear() {
-    // if (jit)
-    //{
-    //    std::for_each(m_BreakPoints.begin(), m_BreakPoints.end(),
-    //        [](const TBreakPoint& bp)
-    //        {
-    //            jit->GetBlockCache()->InvalidateICache(bp.iAddress, 4);
-    //        }
-    //    );
-    //}
-
-    m_BreakPoints.clear();
-}

src/common/break_points.h

@@ -1,49 +0,0 @@
-// Copyright 2013 Dolphin Emulator Project / 2014 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <string>
-#include <vector>
-#include "common/common_types.h"
-
-class DebugInterface;
-
-struct TBreakPoint {
-    u32 iAddress;
-    bool bOn;
-    bool bTemporary;
-};
-
-// Code breakpoints.
-class BreakPoints {
-public:
-    typedef std::vector<TBreakPoint> TBreakPoints;
-    typedef std::vector<std::string> TBreakPointsStr;
-
-    const TBreakPoints& GetBreakPoints() {
-        return m_BreakPoints;
-    }
-
-    TBreakPointsStr GetStrings() const;
-    void AddFromStrings(const TBreakPointsStr& bps);
-
-    // is address breakpoint
-    bool IsAddressBreakPoint(u32 iAddress) const;
-    bool IsTempBreakPoint(u32 iAddress) const;
-
-    // Add BreakPoint
-    void Add(u32 em_address, bool temp = false);
-    void Add(const TBreakPoint& bp);
-
-    // Remove Breakpoint
-    void Remove(u32 iAddress);
-    void Clear();
-
-    void DeleteByAddress(u32 Address);
-
-private:
-    TBreakPoints m_BreakPoints;
-    u32 m_iBreakOnCount;
-};

src/common/file_util.cpp

@@ -750,6 +750,12 @@ std::string GetHactoolConfigurationPath() {
 #endif
 }
 
+std::string GetNANDRegistrationDir(bool system) {
+    if (system)
+        return GetUserPath(UserPath::NANDDir) + "system/Contents/registered/";
+    return GetUserPath(UserPath::NANDDir) + "user/Contents/registered/";
+}
+
 size_t WriteStringToFile(bool text_file, const std::string& str, const char* filename) {
     return FileUtil::IOFile(filename, text_file ? "w" : "wb").WriteBytes(str.data(), str.size());
 }

src/common/file_util.h

@@ -129,6 +129,8 @@ const std::string& GetUserPath(UserPath path, const std::string& new_path = "");
 
 std::string GetHactoolConfigurationPath();
 
+std::string GetNANDRegistrationDir(bool system = false);
+
 // Returns the path to where the sys file are
 std::string GetSysDirectory();
 

src/common/hex_util.cpp

@@ -0,0 +1,31 @@
+// Copyright 2013 Dolphin Emulator Project / 2014 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "common/hex_util.h"
+
+namespace Common {
+
+u8 ToHexNibble(char c1) {
+    if (c1 >= 65 && c1 <= 70)
+        return c1 - 55;
+    if (c1 >= 97 && c1 <= 102)
+        return c1 - 87;
+    if (c1 >= 48 && c1 <= 57)
+        return c1 - 48;
+    throw std::logic_error("Invalid hex digit");
+}
+
+std::array<u8, 16> operator""_array16(const char* str, size_t len) {
+    if (len != 32)
+        throw std::logic_error("Not of correct size.");
+    return HexStringToArray<16>(str);
+}
+
+std::array<u8, 32> operator""_array32(const char* str, size_t len) {
+    if (len != 64)
+        throw std::logic_error("Not of correct size.");
+    return HexStringToArray<32>(str);
+}
+
+} // namespace Common

src/common/hex_util.h

@@ -0,0 +1,41 @@
+// Copyright 2013 Dolphin Emulator Project / 2014 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include <cstddef>
+#include <string>
+#include <fmt/format.h>
+#include "common/common_types.h"
+
+namespace Common {
+
+u8 ToHexNibble(char c1);
+
+template <size_t Size, bool le = false>
+std::array<u8, Size> HexStringToArray(std::string_view str) {
+    std::array<u8, Size> out{};
+    if constexpr (le) {
+        for (size_t i = 2 * Size - 2; i <= 2 * Size; i -= 2)
+            out[i / 2] = (ToHexNibble(str[i]) << 4) | ToHexNibble(str[i + 1]);
+    } else {
+        for (size_t i = 0; i < 2 * Size; i += 2)
+            out[i / 2] = (ToHexNibble(str[i]) << 4) | ToHexNibble(str[i + 1]);
+    }
+    return out;
+}
+
+template <size_t Size>
+std::string HexArrayToString(std::array<u8, Size> array, bool upper = true) {
+    std::string out;
+    for (u8 c : array)
+        out += fmt::format(upper ? "{:02X}" : "{:02x}", c);
+    return out;
+}
+
+std::array<u8, 0x10> operator"" _array16(const char* str, size_t len);
+std::array<u8, 0x20> operator"" _array32(const char* str, size_t len);
+
+} // namespace Common

src/common/logging/backend.cpp

@@ -302,13 +302,14 @@ Backend* GetBackend(std::string_view backend_name) {
 void FmtLogMessageImpl(Class log_class, Level log_level, const char* filename,
                        unsigned int line_num, const char* function, const char* format,
                        const fmt::format_args& args) {
-    auto filter = Impl::Instance().GetGlobalFilter();
+    auto& instance = Impl::Instance();
+    const auto& filter = instance.GetGlobalFilter();
     if (!filter.CheckMessage(log_class, log_level))
         return;
 
     Entry entry =
         CreateEntry(log_class, log_level, filename, line_num, function, fmt::vformat(format, args));
 
-    Impl::Instance().PushEntry(std::move(entry));
+    instance.PushEntry(std::move(entry));
 }
 } // namespace Log

src/common/misc.cpp

@@ -4,7 +4,7 @@
 
 #include <cstddef>
 #ifdef _WIN32
-#include <Windows.h>
+#include <windows.h>
 #else
 #include <cerrno>
 #include <cstring>

src/common/thread_queue_list.h

@@ -16,7 +16,7 @@ struct ThreadQueueList {
     //               (dynamically resizable) circular buffers to remove their overhead when
     //               inserting and popping.
 
-    typedef unsigned int Priority;
+    using Priority = unsigned int;
 
     // Number of priority levels. (Valid levels are [0..NUM_QUEUES).)
     static const Priority NUM_QUEUES = N;
@@ -26,9 +26,9 @@ struct ThreadQueueList {
     }
 
     // Only for debugging, returns priority level.
-    Priority contains(const T& uid) {
+    Priority contains(const T& uid) const {
         for (Priority i = 0; i < NUM_QUEUES; ++i) {
-            Queue& cur = queues[i];
+            const Queue& cur = queues[i];
             if (std::find(cur.data.cbegin(), cur.data.cend(), uid) != cur.data.cend()) {
                 return i;
             }
@@ -37,8 +37,8 @@ struct ThreadQueueList {
         return -1;
     }
 
-    T get_first() {
-        Queue* cur = first;
+    T get_first() const {
+        const Queue* cur = first;
         while (cur != nullptr) {
             if (!cur->data.empty()) {
                 return cur->data.front();

src/common/x64/xbyak_abi.h

@@ -9,10 +9,9 @@
 #include "common/assert.h"
 #include "common/bit_set.h"
 
-namespace Common {
-namespace X64 {
+namespace Common::X64 {
 
-int RegToIndex(const Xbyak::Reg& reg) {
+inline int RegToIndex(const Xbyak::Reg& reg) {
     using Kind = Xbyak::Reg::Kind;
     ASSERT_MSG((reg.getKind() & (Kind::REG | Kind::XMM)) != 0,
                "RegSet only support GPRs and XMM registers.");
@@ -152,8 +151,8 @@ constexpr size_t ABI_SHADOW_SPACE = 0;
 
 #endif
 
-void ABI_CalculateFrameSize(BitSet32 regs, size_t rsp_alignment, size_t needed_frame_size,
-                            s32* out_subtraction, s32* out_xmm_offset) {
+inline void ABI_CalculateFrameSize(BitSet32 regs, size_t rsp_alignment, size_t needed_frame_size,
+                                   s32* out_subtraction, s32* out_xmm_offset) {
     int count = (regs & ABI_ALL_GPRS).Count();
     rsp_alignment -= count * 8;
     size_t subtraction = 0;
@@ -174,8 +173,8 @@ void ABI_CalculateFrameSize(BitSet32 regs, size_t rsp_alignment, size_t needed_f
     *out_xmm_offset = (s32)(subtraction - xmm_base_subtraction);
 }
 
-size_t ABI_PushRegistersAndAdjustStack(Xbyak::CodeGenerator& code, BitSet32 regs,
-                                       size_t rsp_alignment, size_t needed_frame_size = 0) {
+inline size_t ABI_PushRegistersAndAdjustStack(Xbyak::CodeGenerator& code, BitSet32 regs,
+                                              size_t rsp_alignment, size_t needed_frame_size = 0) {
     s32 subtraction, xmm_offset;
     ABI_CalculateFrameSize(regs, rsp_alignment, needed_frame_size, &subtraction, &xmm_offset);
 
@@ -195,8 +194,8 @@ size_t ABI_PushRegistersAndAdjustStack(Xbyak::CodeGenerator& code, BitSet32 regs
     return ABI_SHADOW_SPACE;
 }
 
-void ABI_PopRegistersAndAdjustStack(Xbyak::CodeGenerator& code, BitSet32 regs, size_t rsp_alignment,
-                                    size_t needed_frame_size = 0) {
+inline void ABI_PopRegistersAndAdjustStack(Xbyak::CodeGenerator& code, BitSet32 regs,
+                                           size_t rsp_alignment, size_t needed_frame_size = 0) {
     s32 subtraction, xmm_offset;
     ABI_CalculateFrameSize(regs, rsp_alignment, needed_frame_size, &subtraction, &xmm_offset);
 
@@ -217,5 +216,4 @@ void ABI_PopRegistersAndAdjustStack(Xbyak::CodeGenerator& code, BitSet32 regs, s
     }
 }
 
-} // namespace X64
-} // namespace Common
+} // namespace Common::X64

src/common/x64/xbyak_util.h

@@ -8,8 +8,7 @@
 #include <xbyak.h>
 #include "common/x64/xbyak_abi.h"
 
-namespace Common {
-namespace X64 {
+namespace Common::X64 {
 
 // Constants for use with cmpps/cmpss
 enum {
@@ -45,5 +44,4 @@ inline void CallFarFunction(Xbyak::CodeGenerator& code, const T f) {
     }
 }
 
-} // namespace X64
-} // namespace Common
+} // namespace Common::X64

src/core/CMakeLists.txt

@@ -20,6 +20,8 @@ add_library(core STATIC
     crypto/key_manager.h
     crypto/ctr_encryption_layer.cpp
     crypto/ctr_encryption_layer.h
+    file_sys/bis_factory.cpp
+    file_sys/bis_factory.h
     file_sys/card_image.cpp
     file_sys/card_image.h
     file_sys/content_archive.cpp
@@ -29,10 +31,14 @@ add_library(core STATIC
     file_sys/directory.h
     file_sys/errors.h
     file_sys/mode.h
+    file_sys/nca_metadata.cpp
+    file_sys/nca_metadata.h
     file_sys/partition_filesystem.cpp
     file_sys/partition_filesystem.h
     file_sys/program_metadata.cpp
     file_sys/program_metadata.h
+    file_sys/registered_cache.cpp
+    file_sys/registered_cache.h
     file_sys/romfs.cpp
     file_sys/romfs.h
     file_sys/romfs_factory.cpp
@@ -43,6 +49,8 @@ add_library(core STATIC
     file_sys/sdmc_factory.h
     file_sys/vfs.cpp
     file_sys/vfs.h
+    file_sys/vfs_concat.cpp
+    file_sys/vfs_concat.h
     file_sys/vfs_offset.cpp
     file_sys/vfs_offset.h
     file_sys/vfs_real.cpp
@@ -249,6 +257,10 @@ add_library(core STATIC
     hle/service/nvdrv/devices/nvhost_gpu.h
     hle/service/nvdrv/devices/nvhost_nvdec.cpp
     hle/service/nvdrv/devices/nvhost_nvdec.h
+    hle/service/nvdrv/devices/nvhost_nvjpg.cpp
+    hle/service/nvdrv/devices/nvhost_nvjpg.h
+    hle/service/nvdrv/devices/nvhost_vic.cpp
+    hle/service/nvdrv/devices/nvhost_vic.h
     hle/service/nvdrv/devices/nvmap.cpp
     hle/service/nvdrv/devices/nvmap.h
     hle/service/nvdrv/interface.cpp

src/core/arm/dynarmic/arm_dynarmic.cpp

@@ -86,7 +86,16 @@ public:
     }
 
     void AddTicks(u64 ticks) override {
-        CoreTiming::AddTicks(ticks - num_interpreted_instructions);
+        // Divide the number of ticks by the amount of CPU cores. TODO(Subv): This yields only a
+        // rough approximation of the amount of executed ticks in the system, it may be thrown off
+        // if not all cores are doing a similar amount of work. Instead of doing this, we should
+        // device a way so that timing is consistent across all cores without increasing the ticks 4
+        // times.
+        u64 amortized_ticks = (ticks - num_interpreted_instructions) / Core::NUM_CPU_CORES;
+        // Always execute at least one tick.
+        amortized_ticks = std::max<u64>(amortized_ticks, 1);
+
+        CoreTiming::AddTicks(amortized_ticks);
         num_interpreted_instructions = 0;
     }
     u64 GetTicksRemaining() override {
@@ -234,9 +243,7 @@ void ARM_Dynarmic::LoadContext(const ThreadContext& ctx) {
 }
 
 void ARM_Dynarmic::PrepareReschedule() {
-    if (jit->IsExecuting()) {
-        jit->HaltExecution();
-    }
+    jit->HaltExecution();
 }
 
 void ARM_Dynarmic::ClearInstructionCache() {

src/core/core.cpp

@@ -5,6 +5,7 @@
 #include <memory>
 #include <utility>
 #include "common/logging/log.h"
+#include "common/string_util.h"
 #include "core/core.h"
 #include "core/core_timing.h"
 #include "core/gdbstub/gdbstub.h"
@@ -17,6 +18,7 @@
 #include "core/hle/service/sm/sm.h"
 #include "core/loader/loader.h"
 #include "core/settings.h"
+#include "file_sys/vfs_concat.h"
 #include "file_sys/vfs_real.h"
 #include "video_core/renderer_base.h"
 #include "video_core/video_core.h"
@@ -88,8 +90,39 @@ System::ResultStatus System::SingleStep() {
     return RunLoop(false);
 }
 
-System::ResultStatus System::Load(EmuWindow& emu_window, const std::string& filepath) {
-    app_loader = Loader::GetLoader(virtual_filesystem->OpenFile(filepath, FileSys::Mode::Read));
+static FileSys::VirtualFile GetGameFileFromPath(const FileSys::VirtualFilesystem& vfs,
+                                                const std::string& path) {
+    // To account for split 00+01+etc files.
+    std::string dir_name;
+    std::string filename;
+    Common::SplitPath(path, &dir_name, &filename, nullptr);
+    if (filename == "00") {
+        const auto dir = vfs->OpenDirectory(dir_name, FileSys::Mode::Read);
+        std::vector<FileSys::VirtualFile> concat;
+        for (u8 i = 0; i < 0x10; ++i) {
+            auto next = dir->GetFile(fmt::format("{:02X}", i));
+            if (next != nullptr)
+                concat.push_back(std::move(next));
+            else {
+                next = dir->GetFile(fmt::format("{:02x}", i));
+                if (next != nullptr)
+                    concat.push_back(std::move(next));
+                else
+                    break;
+            }
+        }
+
+        if (concat.empty())
+            return nullptr;
+
+        return FileSys::ConcatenateFiles(concat, dir->GetName());
+    }
+
+    return vfs->OpenFile(path, FileSys::Mode::Read);
+}
+
+System::ResultStatus System::Load(Frontend::EmuWindow& emu_window, const std::string& filepath) {
+    app_loader = Loader::GetLoader(GetGameFileFromPath(virtual_filesystem, filepath));
 
     if (!app_loader) {
         LOG_CRITICAL(Core, "Failed to obtain loader for {}!", filepath);
@@ -151,7 +184,7 @@ Cpu& System::CpuCore(size_t core_index) {
     return *cpu_cores[core_index];
 }
 
-System::ResultStatus System::Init(EmuWindow& emu_window) {
+System::ResultStatus System::Init(Frontend::EmuWindow& emu_window) {
     LOG_DEBUG(HW_Memory, "initialized OK");
 
     CoreTiming::Init();

src/core/core.h

@@ -22,9 +22,12 @@
 #include "video_core/debug_utils/debug_utils.h"
 #include "video_core/gpu.h"
 
-class EmuWindow;
 class ARM_Interface;
 
+namespace Core::Frontend {
+class EmuWindow;
+}
+
 namespace Service::SM {
 class ServiceManager;
 }
@@ -99,7 +102,7 @@ public:
      * @param filepath String path to the executable application to load on the host file system.
      * @returns ResultStatus code, indicating if the operation succeeded.
      */
-    ResultStatus Load(EmuWindow& emu_window, const std::string& filepath);
+    ResultStatus Load(Frontend::EmuWindow& emu_window, const std::string& filepath);
 
     /**
      * Indicates if the emulated system is powered on (all subsystems initialized and able to run an
@@ -227,7 +230,7 @@ private:
      *                   input.
      * @return ResultStatus code, indicating if the operation succeeded.
      */
-    ResultStatus Init(EmuWindow& emu_window);
+    ResultStatus Init(Frontend::EmuWindow& emu_window);
 
     /// RealVfsFilesystem instance
     FileSys::VirtualFilesystem virtual_filesystem;

src/core/core_cpu.cpp

@@ -14,6 +14,7 @@
 #include "core/core_timing.h"
 #include "core/hle/kernel/scheduler.h"
 #include "core/hle/kernel/thread.h"
+#include "core/hle/lock.h"
 #include "core/settings.h"
 
 namespace Core {
@@ -90,6 +91,7 @@ void Cpu::RunLoop(bool tight_loop) {
         LOG_TRACE(Core, "Core-{} idling", core_index);
 
         if (IsMainCore()) {
+            // TODO(Subv): Only let CoreTiming idle if all 4 cores are idling.
             CoreTiming::Idle();
             CoreTiming::Advance();
         }
@@ -125,6 +127,8 @@ void Cpu::Reschedule() {
     }
 
     reschedule_pending = false;
+    // Lock the global kernel mutex when we manipulate the HLE state
+    std::lock_guard<std::recursive_mutex> lock(HLE::g_hle_lock);
     scheduler->Reschedule();
 }
 

src/core/core_cpu.h

@@ -79,7 +79,7 @@ private:
     std::shared_ptr<CpuBarrier> cpu_barrier;
     std::shared_ptr<Kernel::Scheduler> scheduler;
 
-    bool reschedule_pending{};
+    std::atomic<bool> reschedule_pending = false;
     size_t core_index;
 };
 

src/core/core_timing.cpp

@@ -56,6 +56,9 @@ static u64 event_fifo_id;
 // to the event_queue by the emu thread
 static Common::MPSCQueue<Event, false> ts_queue;
 
+// the queue for unscheduling the events from other threads threadsafe
+static Common::MPSCQueue<std::pair<const EventType*, u64>, false> unschedule_queue;
+
 constexpr int MAX_SLICE_LENGTH = 20000;
 
 static s64 idled_cycles;
@@ -135,11 +138,9 @@ void ClearPendingEvents() {
 void ScheduleEvent(s64 cycles_into_future, const EventType* event_type, u64 userdata) {
     ASSERT(event_type != nullptr);
     s64 timeout = GetTicks() + cycles_into_future;
-
     // If this event needs to be scheduled before the next advance(), force one early
     if (!is_global_timer_sane)
         ForceExceptionCheck(cycles_into_future);
-
     event_queue.emplace_back(Event{timeout, event_fifo_id++, userdata, event_type});
     std::push_heap(event_queue.begin(), event_queue.end(), std::greater<>());
 }
@@ -160,6 +161,10 @@ void UnscheduleEvent(const EventType* event_type, u64 userdata) {
     }
 }
 
+void UnscheduleEventThreadsafe(const EventType* event_type, u64 userdata) {
+    unschedule_queue.Push(std::make_pair(event_type, userdata));
+}
+
 void RemoveEvent(const EventType* event_type) {
     auto itr = std::remove_if(event_queue.begin(), event_queue.end(),
                               [&](const Event& e) { return e.type == event_type; });
@@ -196,6 +201,9 @@ void MoveEvents() {
 
 void Advance() {
     MoveEvents();
+    for (std::pair<const EventType*, u64> ev; unschedule_queue.Pop(ev);) {
+        UnscheduleEvent(ev.first, ev.second);
+    }
 
     int cycles_executed = slice_length - downcount;
     global_timer += cycles_executed;

src/core/core_timing.h

@@ -65,6 +65,7 @@ void ScheduleEvent(s64 cycles_into_future, const EventType* event_type, u64 user
 void ScheduleEventThreadsafe(s64 cycles_into_future, const EventType* event_type, u64 userdata);
 
 void UnscheduleEvent(const EventType* event_type, u64 userdata);
+void UnscheduleEventThreadsafe(const EventType* event_type, u64 userdata);
 
 /// We only permit one event of each type in the queue at a time.
 void RemoveEvent(const EventType* event_type);

src/core/crypto/key_manager.cpp

@@ -10,44 +10,13 @@
 #include <string_view>
 #include "common/common_paths.h"
 #include "common/file_util.h"
+#include "common/hex_util.h"
+#include "common/logging/log.h"
 #include "core/crypto/key_manager.h"
 #include "core/settings.h"
 
 namespace Core::Crypto {
 
-static u8 ToHexNibble(char c1) {
-    if (c1 >= 65 && c1 <= 70)
-        return c1 - 55;
-    if (c1 >= 97 && c1 <= 102)
-        return c1 - 87;
-    if (c1 >= 48 && c1 <= 57)
-        return c1 - 48;
-    throw std::logic_error("Invalid hex digit");
-}
-
-template <size_t Size>
-static std::array<u8, Size> HexStringToArray(std::string_view str) {
-    std::array<u8, Size> out{};
-    for (size_t i = 0; i < 2 * Size; i += 2) {
-        auto d1 = str[i];
-        auto d2 = str[i + 1];
-        out[i / 2] = (ToHexNibble(d1) << 4) | ToHexNibble(d2);
-    }
-    return out;
-}
-
-std::array<u8, 16> operator""_array16(const char* str, size_t len) {
-    if (len != 32)
-        throw std::logic_error("Not of correct size.");
-    return HexStringToArray<16>(str);
-}
-
-std::array<u8, 32> operator""_array32(const char* str, size_t len) {
-    if (len != 64)
-        throw std::logic_error("Not of correct size.");
-    return HexStringToArray<32>(str);
-}
-
 KeyManager::KeyManager() {
     // Initialize keys
     const std::string hactool_keys_dir = FileUtil::GetHactoolConfigurationPath();
@@ -83,20 +52,20 @@ void KeyManager::LoadFromFile(const std::string& filename, bool is_title_keys) {
         out[1].erase(std::remove(out[1].begin(), out[1].end(), ' '), out[1].end());
 
         if (is_title_keys) {
-            auto rights_id_raw = HexStringToArray<16>(out[0]);
+            auto rights_id_raw = Common::HexStringToArray<16>(out[0]);
             u128 rights_id{};
             std::memcpy(rights_id.data(), rights_id_raw.data(), rights_id_raw.size());
-            Key128 key = HexStringToArray<16>(out[1]);
+            Key128 key = Common::HexStringToArray<16>(out[1]);
             SetKey(S128KeyType::Titlekey, key, rights_id[1], rights_id[0]);
         } else {
             std::transform(out[0].begin(), out[0].end(), out[0].begin(), ::tolower);
             if (s128_file_id.find(out[0]) != s128_file_id.end()) {
                 const auto index = s128_file_id.at(out[0]);
-                Key128 key = HexStringToArray<16>(out[1]);
+                Key128 key = Common::HexStringToArray<16>(out[1]);
                 SetKey(index.type, key, index.field1, index.field2);
             } else if (s256_file_id.find(out[0]) != s256_file_id.end()) {
                 const auto index = s256_file_id.at(out[0]);
-                Key256 key = HexStringToArray<32>(out[1]);
+                Key256 key = Common::HexStringToArray<32>(out[1]);
                 SetKey(index.type, key, index.field1, index.field2);
             }
         }

src/core/crypto/key_manager.h

@@ -87,9 +87,6 @@ struct hash<Core::Crypto::KeyIndex<KeyType>> {
 
 namespace Core::Crypto {
 
-std::array<u8, 0x10> operator"" _array16(const char* str, size_t len);
-std::array<u8, 0x20> operator"" _array32(const char* str, size_t len);
-
 class KeyManager {
 public:
     KeyManager();

src/core/file_sys/bis_factory.cpp

@@ -0,0 +1,31 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/file_sys/bis_factory.h"
+
+namespace FileSys {
+
+static VirtualDir GetOrCreateDirectory(const VirtualDir& dir, std::string_view path) {
+    const auto res = dir->GetDirectoryRelative(path);
+    if (res == nullptr)
+        return dir->CreateDirectoryRelative(path);
+    return res;
+}
+
+BISFactory::BISFactory(VirtualDir nand_root_)
+    : nand_root(std::move(nand_root_)),
+      sysnand_cache(std::make_shared<RegisteredCache>(
+          GetOrCreateDirectory(nand_root, "/system/Contents/registered"))),
+      usrnand_cache(std::make_shared<RegisteredCache>(
+          GetOrCreateDirectory(nand_root, "/user/Contents/registered"))) {}
+
+std::shared_ptr<RegisteredCache> BISFactory::GetSystemNANDContents() const {
+    return sysnand_cache;
+}
+
+std::shared_ptr<RegisteredCache> BISFactory::GetUserNANDContents() const {
+    return usrnand_cache;
+}
+
+} // namespace FileSys

src/core/file_sys/bis_factory.h

@@ -0,0 +1,30 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <memory>
+#include "core/loader/loader.h"
+#include "registered_cache.h"
+
+namespace FileSys {
+
+/// File system interface to the Built-In Storage
+/// This is currently missing accessors to BIS partitions, but seemed like a good place for the NAND
+/// registered caches.
+class BISFactory {
+public:
+    explicit BISFactory(VirtualDir nand_root);
+
+    std::shared_ptr<RegisteredCache> GetSystemNANDContents() const;
+    std::shared_ptr<RegisteredCache> GetUserNANDContents() const;
+
+private:
+    VirtualDir nand_root;
+
+    std::shared_ptr<RegisteredCache> sysnand_cache;
+    std::shared_ptr<RegisteredCache> usrnand_cache;
+};
+
+} // namespace FileSys

src/core/file_sys/card_image.cpp

@@ -4,11 +4,14 @@
 
 #include <array>
 #include <string>
-#include <core/loader/loader.h>
+
+#include <fmt/ostream.h>
+
 #include "common/logging/log.h"
 #include "core/file_sys/card_image.h"
 #include "core/file_sys/partition_filesystem.h"
 #include "core/file_sys/vfs_offset.h"
+#include "core/loader/loader.h"
 
 namespace FileSys {
 
@@ -93,6 +96,10 @@ VirtualDir XCI::GetLogoPartition() const {
     return GetPartition(XCIPartition::Logo);
 }
 
+const std::vector<std::shared_ptr<NCA>>& XCI::GetNCAs() const {
+    return ncas;
+}
+
 std::shared_ptr<NCA> XCI::GetNCAByType(NCAContentType type) const {
     const auto iter =
         std::find_if(ncas.begin(), ncas.end(),
@@ -107,19 +114,19 @@ VirtualFile XCI::GetNCAFileByType(NCAContentType type) const {
     return nullptr;
 }
 
-std::vector<std::shared_ptr<VfsFile>> XCI::GetFiles() const {
+std::vector<VirtualFile> XCI::GetFiles() const {
     return {};
 }
 
-std::vector<std::shared_ptr<VfsDirectory>> XCI::GetSubdirectories() const {
-    return std::vector<std::shared_ptr<VfsDirectory>>();
+std::vector<VirtualDir> XCI::GetSubdirectories() const {
+    return {};
 }
 
 std::string XCI::GetName() const {
     return file->GetName();
 }
 
-std::shared_ptr<VfsDirectory> XCI::GetParentDirectory() const {
+VirtualDir XCI::GetParentDirectory() const {
     return file->GetContainingDirectory();
 }
 
@@ -142,7 +149,7 @@ Loader::ResultStatus XCI::AddNCAFromPartition(XCIPartition part) {
             const u16 error_id = static_cast<u16>(nca->GetStatus());
             LOG_CRITICAL(Loader, "Could not load NCA {}/{}, failed with error code {:04X} ({})",
                          partition_names[static_cast<size_t>(part)], nca->GetName(), error_id,
-                         Loader::GetMessageForResultStatus(nca->GetStatus()));
+                         nca->GetStatus());
         }
     }
 

src/core/file_sys/card_image.h

@@ -68,16 +68,17 @@ public:
     VirtualDir GetUpdatePartition() const;
     VirtualDir GetLogoPartition() const;
 
+    const std::vector<std::shared_ptr<NCA>>& GetNCAs() const;
     std::shared_ptr<NCA> GetNCAByType(NCAContentType type) const;
     VirtualFile GetNCAFileByType(NCAContentType type) const;
 
-    std::vector<std::shared_ptr<VfsFile>> GetFiles() const override;
+    std::vector<VirtualFile> GetFiles() const override;
 
-    std::vector<std::shared_ptr<VfsDirectory>> GetSubdirectories() const override;
+    std::vector<VirtualDir> GetSubdirectories() const override;
 
     std::string GetName() const override;
 
-    std::shared_ptr<VfsDirectory> GetParentDirectory() const override;
+    VirtualDir GetParentDirectory() const override;
 
 protected:
     bool ReplaceFileWithSubdirectory(VirtualFile file, VirtualDir dir) override;

src/core/file_sys/control_metadata.cpp

@@ -16,7 +16,7 @@ std::string LanguageEntry::GetDeveloperName() const {
     return Common::StringFromFixedZeroTerminatedBuffer(developer_name.data(), 0x100);
 }
 
-NACP::NACP(VirtualFile file_) : file(std::move(file_)), raw(std::make_unique<RawNACP>()) {
+NACP::NACP(VirtualFile file) : raw(std::make_unique<RawNACP>()) {
     file->ReadObject(raw.get());
 }
 

src/core/file_sys/control_metadata.h

@@ -81,7 +81,6 @@ public:
     std::string GetVersionString() const;
 
 private:
-    VirtualFile file;
     std::unique_ptr<RawNACP> raw;
 };
 

src/core/file_sys/nca_metadata.cpp

@@ -0,0 +1,131 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <cstring>
+#include "common/common_funcs.h"
+#include "common/logging/log.h"
+#include "common/swap.h"
+#include "content_archive.h"
+#include "core/file_sys/nca_metadata.h"
+
+namespace FileSys {
+
+bool operator>=(TitleType lhs, TitleType rhs) {
+    return static_cast<size_t>(lhs) >= static_cast<size_t>(rhs);
+}
+
+bool operator<=(TitleType lhs, TitleType rhs) {
+    return static_cast<size_t>(lhs) <= static_cast<size_t>(rhs);
+}
+
+CNMT::CNMT(VirtualFile file) {
+    if (file->ReadObject(&header) != sizeof(CNMTHeader))
+        return;
+
+    // If type is {Application, Update, AOC} has opt-header.
+    if (header.type >= TitleType::Application && header.type <= TitleType::AOC) {
+        if (file->ReadObject(&opt_header, sizeof(CNMTHeader)) != sizeof(OptionalHeader)) {
+            LOG_WARNING(Loader, "Failed to read optional header.");
+        }
+    }
+
+    for (u16 i = 0; i < header.number_content_entries; ++i) {
+        auto& next = content_records.emplace_back(ContentRecord{});
+        if (file->ReadObject(&next, sizeof(CNMTHeader) + i * sizeof(ContentRecord) +
+                                        header.table_offset) != sizeof(ContentRecord)) {
+            content_records.erase(content_records.end() - 1);
+        }
+    }
+
+    for (u16 i = 0; i < header.number_meta_entries; ++i) {
+        auto& next = meta_records.emplace_back(MetaRecord{});
+        if (file->ReadObject(&next, sizeof(CNMTHeader) + i * sizeof(MetaRecord) +
+                                        header.table_offset) != sizeof(MetaRecord)) {
+            meta_records.erase(meta_records.end() - 1);
+        }
+    }
+}
+
+CNMT::CNMT(CNMTHeader header, OptionalHeader opt_header, std::vector<ContentRecord> content_records,
+           std::vector<MetaRecord> meta_records)
+    : header(std::move(header)), opt_header(std::move(opt_header)),
+      content_records(std::move(content_records)), meta_records(std::move(meta_records)) {}
+
+u64 CNMT::GetTitleID() const {
+    return header.title_id;
+}
+
+u32 CNMT::GetTitleVersion() const {
+    return header.title_version;
+}
+
+TitleType CNMT::GetType() const {
+    return header.type;
+}
+
+const std::vector<ContentRecord>& CNMT::GetContentRecords() const {
+    return content_records;
+}
+
+const std::vector<MetaRecord>& CNMT::GetMetaRecords() const {
+    return meta_records;
+}
+
+bool CNMT::UnionRecords(const CNMT& other) {
+    bool change = false;
+    for (const auto& rec : other.content_records) {
+        const auto iter = std::find_if(content_records.begin(), content_records.end(),
+                                       [&rec](const ContentRecord& r) {
+                                           return r.nca_id == rec.nca_id && r.type == rec.type;
+                                       });
+        if (iter == content_records.end()) {
+            content_records.emplace_back(rec);
+            ++header.number_content_entries;
+            change = true;
+        }
+    }
+    for (const auto& rec : other.meta_records) {
+        const auto iter =
+            std::find_if(meta_records.begin(), meta_records.end(), [&rec](const MetaRecord& r) {
+                return r.title_id == rec.title_id && r.title_version == rec.title_version &&
+                       r.type == rec.type;
+            });
+        if (iter == meta_records.end()) {
+            meta_records.emplace_back(rec);
+            ++header.number_meta_entries;
+            change = true;
+        }
+    }
+    return change;
+}
+
+std::vector<u8> CNMT::Serialize() const {
+    const bool has_opt_header =
+        header.type >= TitleType::Application && header.type <= TitleType::AOC;
+    const auto dead_zone = header.table_offset + sizeof(CNMTHeader);
+    std::vector<u8> out(
+        std::max(sizeof(CNMTHeader) + (has_opt_header ? sizeof(OptionalHeader) : 0), dead_zone) +
+        content_records.size() * sizeof(ContentRecord) + meta_records.size() * sizeof(MetaRecord));
+    memcpy(out.data(), &header, sizeof(CNMTHeader));
+
+    // Optional Header
+    if (has_opt_header) {
+        memcpy(out.data() + sizeof(CNMTHeader), &opt_header, sizeof(OptionalHeader));
+    }
+
+    auto offset = header.table_offset;
+
+    for (const auto& rec : content_records) {
+        memcpy(out.data() + offset + sizeof(CNMTHeader), &rec, sizeof(ContentRecord));
+        offset += sizeof(ContentRecord);
+    }
+
+    for (const auto& rec : meta_records) {
+        memcpy(out.data() + offset + sizeof(CNMTHeader), &rec, sizeof(MetaRecord));
+        offset += sizeof(MetaRecord);
+    }
+
+    return out;
+}
+} // namespace FileSys

src/core/file_sys/nca_metadata.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 <cstring>
+#include <memory>
+#include <vector>
+#include "common/common_types.h"
+#include "common/swap.h"
+#include "core/file_sys/vfs.h"
+
+namespace FileSys {
+class CNMT;
+
+struct CNMTHeader;
+struct OptionalHeader;
+
+enum class TitleType : u8 {
+    SystemProgram = 0x01,
+    SystemDataArchive = 0x02,
+    SystemUpdate = 0x03,
+    FirmwarePackageA = 0x04,
+    FirmwarePackageB = 0x05,
+    Application = 0x80,
+    Update = 0x81,
+    AOC = 0x82,
+    DeltaTitle = 0x83,
+};
+
+bool operator>=(TitleType lhs, TitleType rhs);
+bool operator<=(TitleType lhs, TitleType rhs);
+
+enum class ContentRecordType : u8 {
+    Meta = 0,
+    Program = 1,
+    Data = 2,
+    Control = 3,
+    Manual = 4,
+    Legal = 5,
+    Patch = 6,
+};
+
+struct ContentRecord {
+    std::array<u8, 0x20> hash;
+    std::array<u8, 0x10> nca_id;
+    std::array<u8, 0x6> size;
+    ContentRecordType type;
+    INSERT_PADDING_BYTES(1);
+};
+static_assert(sizeof(ContentRecord) == 0x38, "ContentRecord has incorrect size.");
+
+constexpr ContentRecord EMPTY_META_CONTENT_RECORD{{}, {}, {}, ContentRecordType::Meta, {}};
+
+struct MetaRecord {
+    u64_le title_id;
+    u32_le title_version;
+    TitleType type;
+    u8 install_byte;
+    INSERT_PADDING_BYTES(2);
+};
+static_assert(sizeof(MetaRecord) == 0x10, "MetaRecord has incorrect size.");
+
+struct OptionalHeader {
+    u64_le title_id;
+    u64_le minimum_version;
+};
+static_assert(sizeof(OptionalHeader) == 0x10, "OptionalHeader has incorrect size.");
+
+struct CNMTHeader {
+    u64_le title_id;
+    u32_le title_version;
+    TitleType type;
+    INSERT_PADDING_BYTES(1);
+    u16_le table_offset;
+    u16_le number_content_entries;
+    u16_le number_meta_entries;
+    INSERT_PADDING_BYTES(12);
+};
+static_assert(sizeof(CNMTHeader) == 0x20, "CNMTHeader has incorrect size.");
+
+// A class representing the format used by NCA metadata files, typically named {}.cnmt.nca or
+// meta0.ncd. These describe which NCA's belong with which titles in the registered cache.
+class CNMT {
+public:
+    explicit CNMT(VirtualFile file);
+    CNMT(CNMTHeader header, OptionalHeader opt_header, std::vector<ContentRecord> content_records,
+         std::vector<MetaRecord> meta_records);
+
+    u64 GetTitleID() const;
+    u32 GetTitleVersion() const;
+    TitleType GetType() const;
+
+    const std::vector<ContentRecord>& GetContentRecords() const;
+    const std::vector<MetaRecord>& GetMetaRecords() const;
+
+    bool UnionRecords(const CNMT& other);
+    std::vector<u8> Serialize() const;
+
+private:
+    CNMTHeader header;
+    OptionalHeader opt_header;
+    std::vector<ContentRecord> content_records;
+    std::vector<MetaRecord> meta_records;
+
+    // TODO(DarkLordZach): According to switchbrew, for Patch-type there is additional data
+    // after the table. This is not documented, unfortunately.
+};
+
+} // namespace FileSys

src/core/file_sys/partition_filesystem.h

@@ -13,7 +13,7 @@
 #include "core/file_sys/vfs.h"
 
 namespace Loader {
-enum class ResultStatus;
+enum class ResultStatus : u16;
 }
 
 namespace FileSys {

src/core/file_sys/program_metadata.h

@@ -13,7 +13,7 @@
 #include "partition_filesystem.h"
 
 namespace Loader {
-enum class ResultStatus;
+enum class ResultStatus : u16;
 }
 
 namespace FileSys {

src/core/file_sys/registered_cache.cpp

@@ -0,0 +1,478 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <regex>
+#include <mbedtls/sha256.h>
+#include "common/assert.h"
+#include "common/hex_util.h"
+#include "common/logging/log.h"
+#include "core/crypto/encryption_layer.h"
+#include "core/file_sys/card_image.h"
+#include "core/file_sys/nca_metadata.h"
+#include "core/file_sys/registered_cache.h"
+#include "core/file_sys/vfs_concat.h"
+
+namespace FileSys {
+std::string RegisteredCacheEntry::DebugInfo() const {
+    return fmt::format("title_id={:016X}, content_type={:02X}", title_id, static_cast<u8>(type));
+}
+
+bool operator<(const RegisteredCacheEntry& lhs, const RegisteredCacheEntry& rhs) {
+    return (lhs.title_id < rhs.title_id) || (lhs.title_id == rhs.title_id && lhs.type < rhs.type);
+}
+
+static bool FollowsTwoDigitDirFormat(std::string_view name) {
+    static const std::regex two_digit_regex("000000[0-9A-F]{2}", std::regex_constants::ECMAScript |
+                                                                     std::regex_constants::icase);
+    return std::regex_match(name.begin(), name.end(), two_digit_regex);
+}
+
+static bool FollowsNcaIdFormat(std::string_view name) {
+    static const std::regex nca_id_regex("[0-9A-F]{32}\\.nca", std::regex_constants::ECMAScript |
+                                                                   std::regex_constants::icase);
+    return name.size() == 36 && std::regex_match(name.begin(), name.end(), nca_id_regex);
+}
+
+static std::string GetRelativePathFromNcaID(const std::array<u8, 16>& nca_id, bool second_hex_upper,
+                                            bool within_two_digit) {
+    if (!within_two_digit)
+        return fmt::format("/{}.nca", Common::HexArrayToString(nca_id, second_hex_upper));
+
+    Core::Crypto::SHA256Hash hash{};
+    mbedtls_sha256(nca_id.data(), nca_id.size(), hash.data(), 0);
+    return fmt::format("/000000{:02X}/{}.nca", hash[0],
+                       Common::HexArrayToString(nca_id, second_hex_upper));
+}
+
+static std::string GetCNMTName(TitleType type, u64 title_id) {
+    constexpr std::array<const char*, 9> TITLE_TYPE_NAMES{
+        "SystemProgram",
+        "SystemData",
+        "SystemUpdate",
+        "BootImagePackage",
+        "BootImagePackageSafe",
+        "Application",
+        "Patch",
+        "AddOnContent",
+        "" ///< Currently unknown 'DeltaTitle'
+    };
+
+    auto index = static_cast<size_t>(type);
+    // If the index is after the jump in TitleType, subtract it out.
+    if (index >= static_cast<size_t>(TitleType::Application)) {
+        index -= static_cast<size_t>(TitleType::Application) -
+                 static_cast<size_t>(TitleType::FirmwarePackageB);
+    }
+    return fmt::format("{}_{:016x}.cnmt", TITLE_TYPE_NAMES[index], title_id);
+}
+
+static ContentRecordType GetCRTypeFromNCAType(NCAContentType type) {
+    switch (type) {
+    case NCAContentType::Program:
+        // TODO(DarkLordZach): Differentiate between Program and Patch
+        return ContentRecordType::Program;
+    case NCAContentType::Meta:
+        return ContentRecordType::Meta;
+    case NCAContentType::Control:
+        return ContentRecordType::Control;
+    case NCAContentType::Data:
+        return ContentRecordType::Data;
+    case NCAContentType::Manual:
+        // TODO(DarkLordZach): Peek at NCA contents to differentiate Manual and Legal.
+        return ContentRecordType::Manual;
+    default:
+        UNREACHABLE();
+    }
+}
+
+VirtualFile RegisteredCache::OpenFileOrDirectoryConcat(const VirtualDir& dir,
+                                                       std::string_view path) const {
+    if (dir->GetFileRelative(path) != nullptr)
+        return dir->GetFileRelative(path);
+    if (dir->GetDirectoryRelative(path) != nullptr) {
+        const auto nca_dir = dir->GetDirectoryRelative(path);
+        VirtualFile file = nullptr;
+
+        const auto files = nca_dir->GetFiles();
+        if (files.size() == 1 && files[0]->GetName() == "00") {
+            file = files[0];
+        } else {
+            std::vector<VirtualFile> concat;
+            // Since the files are a two-digit hex number, max is FF.
+            for (size_t i = 0; i < 0x100; ++i) {
+                auto next = nca_dir->GetFile(fmt::format("{:02X}", i));
+                if (next != nullptr) {
+                    concat.push_back(std::move(next));
+                } else {
+                    next = nca_dir->GetFile(fmt::format("{:02x}", i));
+                    if (next != nullptr)
+                        concat.push_back(std::move(next));
+                    else
+                        break;
+                }
+            }
+
+            if (concat.empty())
+                return nullptr;
+
+            file = FileSys::ConcatenateFiles(concat);
+        }
+
+        return file;
+    }
+    return nullptr;
+}
+
+VirtualFile RegisteredCache::GetFileAtID(NcaID id) const {
+    VirtualFile file;
+    // Try all four modes of file storage:
+    // (bit 1 = uppercase/lower, bit 0 = within a two-digit dir)
+    // 00: /000000**/{:032X}.nca
+    // 01: /{:032X}.nca
+    // 10: /000000**/{:032x}.nca
+    // 11: /{:032x}.nca
+    for (u8 i = 0; i < 4; ++i) {
+        const auto path = GetRelativePathFromNcaID(id, (i & 0b10) == 0, (i & 0b01) == 0);
+        file = OpenFileOrDirectoryConcat(dir, path);
+        if (file != nullptr)
+            return file;
+    }
+    return file;
+}
+
+static boost::optional<NcaID> CheckMapForContentRecord(
+    const boost::container::flat_map<u64, CNMT>& map, u64 title_id, ContentRecordType type) {
+    if (map.find(title_id) == map.end())
+        return boost::none;
+
+    const auto& cnmt = map.at(title_id);
+
+    const auto iter = std::find_if(cnmt.GetContentRecords().begin(), cnmt.GetContentRecords().end(),
+                                   [type](const ContentRecord& rec) { return rec.type == type; });
+    if (iter == cnmt.GetContentRecords().end())
+        return boost::none;
+
+    return boost::make_optional(iter->nca_id);
+}
+
+boost::optional<NcaID> RegisteredCache::GetNcaIDFromMetadata(u64 title_id,
+                                                             ContentRecordType type) const {
+    if (type == ContentRecordType::Meta && meta_id.find(title_id) != meta_id.end())
+        return meta_id.at(title_id);
+
+    const auto res1 = CheckMapForContentRecord(yuzu_meta, title_id, type);
+    if (res1 != boost::none)
+        return res1;
+    return CheckMapForContentRecord(meta, title_id, type);
+}
+
+std::vector<NcaID> RegisteredCache::AccumulateFiles() const {
+    std::vector<NcaID> ids;
+    for (const auto& d2_dir : dir->GetSubdirectories()) {
+        if (FollowsNcaIdFormat(d2_dir->GetName())) {
+            ids.push_back(Common::HexStringToArray<0x10, true>(d2_dir->GetName().substr(0, 0x20)));
+            continue;
+        }
+
+        if (!FollowsTwoDigitDirFormat(d2_dir->GetName()))
+            continue;
+
+        for (const auto& nca_dir : d2_dir->GetSubdirectories()) {
+            if (!FollowsNcaIdFormat(nca_dir->GetName()))
+                continue;
+
+            ids.push_back(Common::HexStringToArray<0x10, true>(nca_dir->GetName().substr(0, 0x20)));
+        }
+
+        for (const auto& nca_file : d2_dir->GetFiles()) {
+            if (!FollowsNcaIdFormat(nca_file->GetName()))
+                continue;
+
+            ids.push_back(
+                Common::HexStringToArray<0x10, true>(nca_file->GetName().substr(0, 0x20)));
+        }
+    }
+
+    for (const auto& d2_file : dir->GetFiles()) {
+        if (FollowsNcaIdFormat(d2_file->GetName()))
+            ids.push_back(Common::HexStringToArray<0x10, true>(d2_file->GetName().substr(0, 0x20)));
+    }
+    return ids;
+}
+
+void RegisteredCache::ProcessFiles(const std::vector<NcaID>& ids) {
+    for (const auto& id : ids) {
+        const auto file = GetFileAtID(id);
+
+        if (file == nullptr)
+            continue;
+        const auto nca = std::make_shared<NCA>(parser(file, id));
+        if (nca->GetStatus() != Loader::ResultStatus::Success ||
+            nca->GetType() != NCAContentType::Meta) {
+            continue;
+        }
+
+        const auto section0 = nca->GetSubdirectories()[0];
+
+        for (const auto& file : section0->GetFiles()) {
+            if (file->GetExtension() != "cnmt")
+                continue;
+
+            meta.insert_or_assign(nca->GetTitleId(), CNMT(file));
+            meta_id.insert_or_assign(nca->GetTitleId(), id);
+            break;
+        }
+    }
+}
+
+void RegisteredCache::AccumulateYuzuMeta() {
+    const auto dir = this->dir->GetSubdirectory("yuzu_meta");
+    if (dir == nullptr)
+        return;
+
+    for (const auto& file : dir->GetFiles()) {
+        if (file->GetExtension() != "cnmt")
+            continue;
+
+        CNMT cnmt(file);
+        yuzu_meta.insert_or_assign(cnmt.GetTitleID(), std::move(cnmt));
+    }
+}
+
+void RegisteredCache::Refresh() {
+    if (dir == nullptr)
+        return;
+    const auto ids = AccumulateFiles();
+    ProcessFiles(ids);
+    AccumulateYuzuMeta();
+}
+
+RegisteredCache::RegisteredCache(VirtualDir dir_, RegisteredCacheParsingFunction parsing_function)
+    : dir(std::move(dir_)), parser(std::move(parsing_function)) {
+    Refresh();
+}
+
+bool RegisteredCache::HasEntry(u64 title_id, ContentRecordType type) const {
+    return GetEntryRaw(title_id, type) != nullptr;
+}
+
+bool RegisteredCache::HasEntry(RegisteredCacheEntry entry) const {
+    return GetEntryRaw(entry) != nullptr;
+}
+
+VirtualFile RegisteredCache::GetEntryRaw(u64 title_id, ContentRecordType type) const {
+    const auto id = GetNcaIDFromMetadata(title_id, type);
+    if (id == boost::none)
+        return nullptr;
+
+    return parser(GetFileAtID(id.get()), id.get());
+}
+
+VirtualFile RegisteredCache::GetEntryRaw(RegisteredCacheEntry entry) const {
+    return GetEntryRaw(entry.title_id, entry.type);
+}
+
+std::shared_ptr<NCA> RegisteredCache::GetEntry(u64 title_id, ContentRecordType type) const {
+    const auto raw = GetEntryRaw(title_id, type);
+    if (raw == nullptr)
+        return nullptr;
+    return std::make_shared<NCA>(raw);
+}
+
+std::shared_ptr<NCA> RegisteredCache::GetEntry(RegisteredCacheEntry entry) const {
+    return GetEntry(entry.title_id, entry.type);
+}
+
+template <typename T>
+void RegisteredCache::IterateAllMetadata(
+    std::vector<T>& out, std::function<T(const CNMT&, const ContentRecord&)> proc,
+    std::function<bool(const CNMT&, const ContentRecord&)> filter) const {
+    for (const auto& kv : meta) {
+        const auto& cnmt = kv.second;
+        if (filter(cnmt, EMPTY_META_CONTENT_RECORD))
+            out.push_back(proc(cnmt, EMPTY_META_CONTENT_RECORD));
+        for (const auto& rec : cnmt.GetContentRecords()) {
+            if (GetFileAtID(rec.nca_id) != nullptr && filter(cnmt, rec)) {
+                out.push_back(proc(cnmt, rec));
+            }
+        }
+    }
+    for (const auto& kv : yuzu_meta) {
+        const auto& cnmt = kv.second;
+        for (const auto& rec : cnmt.GetContentRecords()) {
+            if (GetFileAtID(rec.nca_id) != nullptr && filter(cnmt, rec)) {
+                out.push_back(proc(cnmt, rec));
+            }
+        }
+    }
+}
+
+std::vector<RegisteredCacheEntry> RegisteredCache::ListEntries() const {
+    std::vector<RegisteredCacheEntry> out;
+    IterateAllMetadata<RegisteredCacheEntry>(
+        out,
+        [](const CNMT& c, const ContentRecord& r) {
+            return RegisteredCacheEntry{c.GetTitleID(), r.type};
+        },
+        [](const CNMT& c, const ContentRecord& r) { return true; });
+    return out;
+}
+
+std::vector<RegisteredCacheEntry> RegisteredCache::ListEntriesFilter(
+    boost::optional<TitleType> title_type, boost::optional<ContentRecordType> record_type,
+    boost::optional<u64> title_id) const {
+    std::vector<RegisteredCacheEntry> out;
+    IterateAllMetadata<RegisteredCacheEntry>(
+        out,
+        [](const CNMT& c, const ContentRecord& r) {
+            return RegisteredCacheEntry{c.GetTitleID(), r.type};
+        },
+        [&title_type, &record_type, &title_id](const CNMT& c, const ContentRecord& r) {
+            if (title_type != boost::none && title_type.get() != c.GetType())
+                return false;
+            if (record_type != boost::none && record_type.get() != r.type)
+                return false;
+            if (title_id != boost::none && title_id.get() != c.GetTitleID())
+                return false;
+            return true;
+        });
+    return out;
+}
+
+static std::shared_ptr<NCA> GetNCAFromXCIForID(std::shared_ptr<XCI> xci, const NcaID& id) {
+    const auto filename = fmt::format("{}.nca", Common::HexArrayToString(id, false));
+    const auto iter =
+        std::find_if(xci->GetNCAs().begin(), xci->GetNCAs().end(),
+                     [&filename](std::shared_ptr<NCA> nca) { return nca->GetName() == filename; });
+    return iter == xci->GetNCAs().end() ? nullptr : *iter;
+}
+
+InstallResult RegisteredCache::InstallEntry(std::shared_ptr<XCI> xci, bool overwrite_if_exists,
+                                            const VfsCopyFunction& copy) {
+    const auto& ncas = xci->GetNCAs();
+    const auto& meta_iter = std::find_if(ncas.begin(), ncas.end(), [](std::shared_ptr<NCA> nca) {
+        return nca->GetType() == NCAContentType::Meta;
+    });
+
+    if (meta_iter == ncas.end()) {
+        LOG_ERROR(Loader, "The XCI you are attempting to install does not have a metadata NCA and "
+                          "is therefore malformed. Double check your encryption keys.");
+        return InstallResult::ErrorMetaFailed;
+    }
+
+    // Install Metadata File
+    const auto meta_id_raw = (*meta_iter)->GetName().substr(0, 32);
+    const auto meta_id = Common::HexStringToArray<16>(meta_id_raw);
+
+    const auto res = RawInstallNCA(*meta_iter, copy, overwrite_if_exists, meta_id);
+    if (res != InstallResult::Success)
+        return res;
+
+    // Install all the other NCAs
+    const auto section0 = (*meta_iter)->GetSubdirectories()[0];
+    const auto cnmt_file = section0->GetFiles()[0];
+    const CNMT cnmt(cnmt_file);
+    for (const auto& record : cnmt.GetContentRecords()) {
+        const auto nca = GetNCAFromXCIForID(xci, record.nca_id);
+        if (nca == nullptr)
+            return InstallResult::ErrorCopyFailed;
+        const auto res2 = RawInstallNCA(nca, copy, overwrite_if_exists, record.nca_id);
+        if (res2 != InstallResult::Success)
+            return res2;
+    }
+
+    Refresh();
+    return InstallResult::Success;
+}
+
+InstallResult RegisteredCache::InstallEntry(std::shared_ptr<NCA> nca, TitleType type,
+                                            bool overwrite_if_exists, const VfsCopyFunction& copy) {
+    CNMTHeader header{
+        nca->GetTitleId(), ///< Title ID
+        0,                 ///< Ignore/Default title version
+        type,              ///< Type
+        {},                ///< Padding
+        0x10,              ///< Default table offset
+        1,                 ///< 1 Content Entry
+        0,                 ///< No Meta Entries
+        {},                ///< Padding
+    };
+    OptionalHeader opt_header{0, 0};
+    ContentRecord c_rec{{}, {}, {}, GetCRTypeFromNCAType(nca->GetType()), {}};
+    const auto& data = nca->GetBaseFile()->ReadBytes(0x100000);
+    mbedtls_sha256(data.data(), data.size(), c_rec.hash.data(), 0);
+    memcpy(&c_rec.nca_id, &c_rec.hash, 16);
+    const CNMT new_cnmt(header, opt_header, {c_rec}, {});
+    if (!RawInstallYuzuMeta(new_cnmt))
+        return InstallResult::ErrorMetaFailed;
+    return RawInstallNCA(nca, copy, overwrite_if_exists, c_rec.nca_id);
+}
+
+InstallResult RegisteredCache::RawInstallNCA(std::shared_ptr<NCA> nca, const VfsCopyFunction& copy,
+                                             bool overwrite_if_exists,
+                                             boost::optional<NcaID> override_id) {
+    const auto in = nca->GetBaseFile();
+    Core::Crypto::SHA256Hash hash{};
+
+    // Calculate NcaID
+    // NOTE: Because computing the SHA256 of an entire NCA is quite expensive (especially if the
+    // game is massive), we're going to cheat and only hash the first MB of the NCA.
+    // Also, for XCIs the NcaID matters, so if the override id isn't none, use that.
+    NcaID id{};
+    if (override_id == boost::none) {
+        const auto& data = in->ReadBytes(0x100000);
+        mbedtls_sha256(data.data(), data.size(), hash.data(), 0);
+        memcpy(id.data(), hash.data(), 16);
+    } else {
+        id = override_id.get();
+    }
+
+    std::string path = GetRelativePathFromNcaID(id, false, true);
+
+    if (GetFileAtID(id) != nullptr && !overwrite_if_exists) {
+        LOG_WARNING(Loader, "Attempting to overwrite existing NCA. Skipping...");
+        return InstallResult::ErrorAlreadyExists;
+    }
+
+    if (GetFileAtID(id) != nullptr) {
+        LOG_WARNING(Loader, "Overwriting existing NCA...");
+        VirtualDir c_dir;
+        { c_dir = dir->GetFileRelative(path)->GetContainingDirectory(); }
+        c_dir->DeleteFile(FileUtil::GetFilename(path));
+    }
+
+    auto out = dir->CreateFileRelative(path);
+    if (out == nullptr)
+        return InstallResult::ErrorCopyFailed;
+    return copy(in, out) ? InstallResult::Success : InstallResult::ErrorCopyFailed;
+}
+
+bool RegisteredCache::RawInstallYuzuMeta(const CNMT& cnmt) {
+    // Reasoning behind this method can be found in the comment for InstallEntry, NCA overload.
+    const auto dir = this->dir->CreateDirectoryRelative("yuzu_meta");
+    const auto filename = GetCNMTName(cnmt.GetType(), cnmt.GetTitleID());
+    if (dir->GetFile(filename) == nullptr) {
+        auto out = dir->CreateFile(filename);
+        const auto buffer = cnmt.Serialize();
+        out->Resize(buffer.size());
+        out->WriteBytes(buffer);
+    } else {
+        auto out = dir->GetFile(filename);
+        CNMT old_cnmt(out);
+        // Returns true on change
+        if (old_cnmt.UnionRecords(cnmt)) {
+            out->Resize(0);
+            const auto buffer = old_cnmt.Serialize();
+            out->Resize(buffer.size());
+            out->WriteBytes(buffer);
+        }
+    }
+    Refresh();
+    return std::find_if(yuzu_meta.begin(), yuzu_meta.end(),
+                        [&cnmt](const std::pair<u64, CNMT>& kv) {
+                            return kv.second.GetType() == cnmt.GetType() &&
+                                   kv.second.GetTitleID() == cnmt.GetTitleID();
+                        }) != yuzu_meta.end();
+}
+} // namespace FileSys

src/core/file_sys/registered_cache.h

@@ -0,0 +1,124 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include <functional>
+#include <map>
+#include <memory>
+#include <string>
+#include <vector>
+#include <boost/container/flat_map.hpp>
+#include "common/common_funcs.h"
+#include "common/common_types.h"
+#include "content_archive.h"
+#include "core/file_sys/nca_metadata.h"
+#include "core/file_sys/vfs.h"
+
+namespace FileSys {
+class XCI;
+class CNMT;
+
+using NcaID = std::array<u8, 0x10>;
+using RegisteredCacheParsingFunction = std::function<VirtualFile(const VirtualFile&, const NcaID&)>;
+using VfsCopyFunction = std::function<bool(VirtualFile, VirtualFile)>;
+
+enum class InstallResult {
+    Success,
+    ErrorAlreadyExists,
+    ErrorCopyFailed,
+    ErrorMetaFailed,
+};
+
+struct RegisteredCacheEntry {
+    u64 title_id;
+    ContentRecordType type;
+
+    std::string DebugInfo() const;
+};
+
+// boost flat_map requires operator< for O(log(n)) lookups.
+bool operator<(const RegisteredCacheEntry& lhs, const RegisteredCacheEntry& rhs);
+
+/*
+ * A class that catalogues NCAs in the registered directory structure.
+ * Nintendo's registered format follows this structure:
+ *
+ * Root
+ *   | 000000XX <- XX is the ____ two digits of the NcaID
+ *       | <hash>.nca <- hash is the NcaID (first half of SHA256 over entire file) (folder)
+ *         | 00
+ *         | 01 <- Actual content split along 4GB boundaries. (optional)
+ *
+ * (This impl also supports substituting the nca dir for an nca file, as that's more convenient when
+ * 4GB splitting can be ignored.)
+ */
+class RegisteredCache {
+public:
+    // Parsing function defines the conversion from raw file to NCA. If there are other steps
+    // besides creating the NCA from the file (e.g. NAX0 on SD Card), that should go in a custom
+    // parsing function.
+    explicit RegisteredCache(VirtualDir dir,
+                             RegisteredCacheParsingFunction parsing_function =
+                                 [](const VirtualFile& file, const NcaID& id) { return file; });
+
+    void Refresh();
+
+    bool HasEntry(u64 title_id, ContentRecordType type) const;
+    bool HasEntry(RegisteredCacheEntry entry) const;
+
+    VirtualFile GetEntryRaw(u64 title_id, ContentRecordType type) const;
+    VirtualFile GetEntryRaw(RegisteredCacheEntry entry) const;
+
+    std::shared_ptr<NCA> GetEntry(u64 title_id, ContentRecordType type) const;
+    std::shared_ptr<NCA> GetEntry(RegisteredCacheEntry entry) const;
+
+    std::vector<RegisteredCacheEntry> ListEntries() const;
+    // If a parameter is not boost::none, it will be filtered for from all entries.
+    std::vector<RegisteredCacheEntry> ListEntriesFilter(
+        boost::optional<TitleType> title_type = boost::none,
+        boost::optional<ContentRecordType> record_type = boost::none,
+        boost::optional<u64> title_id = boost::none) const;
+
+    // Raw copies all the ncas from the xci to the csache. Does some quick checks to make sure there
+    // is a meta NCA and all of them are accessible.
+    InstallResult InstallEntry(std::shared_ptr<XCI> xci, bool overwrite_if_exists = false,
+                               const VfsCopyFunction& copy = &VfsRawCopy);
+
+    // Due to the fact that we must use Meta-type NCAs to determine the existance of files, this
+    // poses quite a challenge. Instead of creating a new meta NCA for this file, yuzu will create a
+    // dir inside the NAND called 'yuzu_meta' and store the raw CNMT there.
+    // TODO(DarkLordZach): Author real meta-type NCAs and install those.
+    InstallResult InstallEntry(std::shared_ptr<NCA> nca, TitleType type,
+                               bool overwrite_if_exists = false,
+                               const VfsCopyFunction& copy = &VfsRawCopy);
+
+private:
+    template <typename T>
+    void IterateAllMetadata(std::vector<T>& out,
+                            std::function<T(const CNMT&, const ContentRecord&)> proc,
+                            std::function<bool(const CNMT&, const ContentRecord&)> filter) const;
+    std::vector<NcaID> AccumulateFiles() const;
+    void ProcessFiles(const std::vector<NcaID>& ids);
+    void AccumulateYuzuMeta();
+    boost::optional<NcaID> GetNcaIDFromMetadata(u64 title_id, ContentRecordType type) const;
+    VirtualFile GetFileAtID(NcaID id) const;
+    VirtualFile OpenFileOrDirectoryConcat(const VirtualDir& dir, std::string_view path) const;
+    InstallResult RawInstallNCA(std::shared_ptr<NCA> nca, const VfsCopyFunction& copy,
+                                bool overwrite_if_exists,
+                                boost::optional<NcaID> override_id = boost::none);
+    bool RawInstallYuzuMeta(const CNMT& cnmt);
+
+    VirtualDir dir;
+    RegisteredCacheParsingFunction parser;
+    // maps tid -> NcaID of meta
+    boost::container::flat_map<u64, NcaID> meta_id;
+    // maps tid -> meta
+    boost::container::flat_map<u64, CNMT> meta;
+    // maps tid -> meta for CNMT in yuzu_meta
+    boost::container::flat_map<u64, CNMT> yuzu_meta;
+};
+
+} // namespace FileSys

src/core/file_sys/romfs.cpp

@@ -65,7 +65,7 @@ void ProcessFile(VirtualFile file, size_t file_offset, size_t data_offset, u32 t
         auto entry = GetEntry<FileEntry>(file, file_offset + this_file_offset);
 
         parent->AddFile(std::make_shared<OffsetVfsFile>(
-            file, entry.first.size, entry.first.offset + data_offset, entry.second, parent));
+            file, entry.first.size, entry.first.offset + data_offset, entry.second));
 
         if (entry.first.sibling == ROMFS_ENTRY_EMPTY)
             break;
@@ -79,7 +79,7 @@ void ProcessDirectory(VirtualFile file, size_t dir_offset, size_t file_offset, s
     while (true) {
         auto entry = GetEntry<DirectoryEntry>(file, dir_offset + this_dir_offset);
         auto current = std::make_shared<VectorVfsDirectory>(
-            std::vector<VirtualFile>{}, std::vector<VirtualDir>{}, parent, entry.second);
+            std::vector<VirtualFile>{}, std::vector<VirtualDir>{}, entry.second);
 
         if (entry.first.child_file != ROMFS_ENTRY_EMPTY) {
             ProcessFile(file, file_offset, data_offset, entry.first.child_file, current);
@@ -108,9 +108,9 @@ VirtualDir ExtractRomFS(VirtualFile file) {
     const u64 file_offset = header.file_meta.offset;
     const u64 dir_offset = header.directory_meta.offset + 4;
 
-    const auto root =
+    auto root =
         std::make_shared<VectorVfsDirectory>(std::vector<VirtualFile>{}, std::vector<VirtualDir>{},
-                                             file->GetContainingDirectory(), file->GetName());
+                                             file->GetName(), file->GetContainingDirectory());
 
     ProcessDirectory(file, dir_offset, file_offset, header.data_offset, 0, root);
 

src/core/file_sys/vfs.cpp

@@ -74,15 +74,15 @@ VirtualFile VfsFilesystem::CopyFile(std::string_view old_path_, std::string_view
     return new_file;
 }
 
-VirtualFile VfsFilesystem::MoveFile(std::string_view old_path_, std::string_view new_path_) {
-    const auto old_path = FileUtil::SanitizePath(old_path_);
-    const auto new_path = FileUtil::SanitizePath(new_path_);
+VirtualFile VfsFilesystem::MoveFile(std::string_view old_path, std::string_view new_path) {
+    const auto sanitized_old_path = FileUtil::SanitizePath(old_path);
+    const auto sanitized_new_path = FileUtil::SanitizePath(new_path);
 
     // Again, non-default impls are highly encouraged to provide a more optimized version of this.
-    auto out = CopyFile(old_path_, new_path_);
+    auto out = CopyFile(sanitized_old_path, sanitized_new_path);
     if (out == nullptr)
         return nullptr;
-    if (DeleteFile(old_path))
+    if (DeleteFile(sanitized_old_path))
         return out;
     return nullptr;
 }
@@ -137,15 +137,15 @@ VirtualDir VfsFilesystem::CopyDirectory(std::string_view old_path_, std::string_
     return new_dir;
 }
 
-VirtualDir VfsFilesystem::MoveDirectory(std::string_view old_path_, std::string_view new_path_) {
-    const auto old_path = FileUtil::SanitizePath(old_path_);
-    const auto new_path = FileUtil::SanitizePath(new_path_);
+VirtualDir VfsFilesystem::MoveDirectory(std::string_view old_path, std::string_view new_path) {
+    const auto sanitized_old_path = FileUtil::SanitizePath(old_path);
+    const auto sanitized_new_path = FileUtil::SanitizePath(new_path);
 
     // Non-default impls are highly encouraged to provide a more optimized version of this.
-    auto out = CopyDirectory(old_path_, new_path_);
+    auto out = CopyDirectory(sanitized_old_path, sanitized_new_path);
     if (out == nullptr)
         return nullptr;
-    if (DeleteDirectory(old_path))
+    if (DeleteDirectory(sanitized_old_path))
         return out;
     return nullptr;
 }

src/core/file_sys/vfs.h

@@ -15,9 +15,9 @@
 
 namespace FileSys {
 
-struct VfsFilesystem;
-struct VfsFile;
-struct VfsDirectory;
+class VfsDirectory;
+class VfsFile;
+class VfsFilesystem;
 
 // Convenience typedefs to use Vfs* interfaces
 using VirtualFilesystem = std::shared_ptr<VfsFilesystem>;
@@ -34,8 +34,9 @@ enum class VfsEntryType {
 // A class representing an abstract filesystem. A default implementation given the root VirtualDir
 // is provided for convenience, but if the Vfs implementation has any additional state or
 // functionality, they will need to override.
-struct VfsFilesystem : NonCopyable {
-    VfsFilesystem(VirtualDir root);
+class VfsFilesystem : NonCopyable {
+public:
+    explicit VfsFilesystem(VirtualDir root);
     virtual ~VfsFilesystem();
 
     // Gets the friendly name for the filesystem.
@@ -81,7 +82,8 @@ protected:
 };
 
 // A class representing a file in an abstract filesystem.
-struct VfsFile : NonCopyable {
+class VfsFile : NonCopyable {
+public:
     virtual ~VfsFile();
 
     // Retrieves the file name.
@@ -179,7 +181,8 @@ struct VfsFile : NonCopyable {
 };
 
 // A class representing a directory in an abstract filesystem.
-struct VfsDirectory : NonCopyable {
+class VfsDirectory : NonCopyable {
+public:
     virtual ~VfsDirectory();
 
     // Retrives the file located at path as if the current directory was root. Returns nullptr if
@@ -295,7 +298,8 @@ protected:
 
 // A convenience partial-implementation of VfsDirectory that stubs out methods that should only work
 // if writable. This is to avoid redundant empty methods everywhere.
-struct ReadOnlyVfsDirectory : public VfsDirectory {
+class ReadOnlyVfsDirectory : public VfsDirectory {
+public:
     bool IsWritable() const override;
     bool IsReadable() const override;
     std::shared_ptr<VfsDirectory> CreateSubdirectory(std::string_view name) override;

src/core/file_sys/vfs_concat.cpp

@@ -0,0 +1,94 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <algorithm>
+#include <utility>
+
+#include "core/file_sys/vfs_concat.h"
+
+namespace FileSys {
+
+VirtualFile ConcatenateFiles(std::vector<VirtualFile> files, std::string name) {
+    if (files.empty())
+        return nullptr;
+    if (files.size() == 1)
+        return files[0];
+
+    return std::shared_ptr<VfsFile>(new ConcatenatedVfsFile(std::move(files), std::move(name)));
+}
+
+ConcatenatedVfsFile::ConcatenatedVfsFile(std::vector<VirtualFile> files_, std::string name)
+    : name(std::move(name)) {
+    size_t next_offset = 0;
+    for (const auto& file : files_) {
+        files[next_offset] = file;
+        next_offset += file->GetSize();
+    }
+}
+
+std::string ConcatenatedVfsFile::GetName() const {
+    if (files.empty())
+        return "";
+    if (!name.empty())
+        return name;
+    return files.begin()->second->GetName();
+}
+
+size_t ConcatenatedVfsFile::GetSize() const {
+    if (files.empty())
+        return 0;
+    return files.rbegin()->first + files.rbegin()->second->GetSize();
+}
+
+bool ConcatenatedVfsFile::Resize(size_t new_size) {
+    return false;
+}
+
+std::shared_ptr<VfsDirectory> ConcatenatedVfsFile::GetContainingDirectory() const {
+    if (files.empty())
+        return nullptr;
+    return files.begin()->second->GetContainingDirectory();
+}
+
+bool ConcatenatedVfsFile::IsWritable() const {
+    return false;
+}
+
+bool ConcatenatedVfsFile::IsReadable() const {
+    return true;
+}
+
+size_t ConcatenatedVfsFile::Read(u8* data, size_t length, size_t offset) const {
+    auto entry = files.end();
+    for (auto iter = files.begin(); iter != files.end(); ++iter) {
+        if (iter->first > offset) {
+            entry = --iter;
+            break;
+        }
+    }
+
+    // Check if the entry should be the last one. The loop above will make it end().
+    if (entry == files.end() && offset < files.rbegin()->first + files.rbegin()->second->GetSize())
+        --entry;
+
+    if (entry == files.end())
+        return 0;
+
+    const auto remaining = entry->second->GetSize() + offset - entry->first;
+    if (length > remaining) {
+        return entry->second->Read(data, remaining, offset - entry->first) +
+               Read(data + remaining, length - remaining, offset + remaining);
+    }
+
+    return entry->second->Read(data, length, offset - entry->first);
+}
+
+size_t ConcatenatedVfsFile::Write(const u8* data, size_t length, size_t offset) {
+    return 0;
+}
+
+bool ConcatenatedVfsFile::Rename(std::string_view name) {
+    return false;
+}
+} // namespace FileSys

src/core/file_sys/vfs_concat.h

@@ -0,0 +1,41 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <memory>
+#include <string_view>
+#include <boost/container/flat_map.hpp>
+#include "core/file_sys/vfs.h"
+
+namespace FileSys {
+
+// Wrapper function to allow for more efficient handling of files.size() == 0, 1 cases.
+VirtualFile ConcatenateFiles(std::vector<VirtualFile> files, std::string name = "");
+
+// Class that wraps multiple vfs files and concatenates them, making reads seamless. Currently
+// read-only.
+class ConcatenatedVfsFile : public VfsFile {
+    friend VirtualFile ConcatenateFiles(std::vector<VirtualFile> files, std::string name);
+
+    ConcatenatedVfsFile(std::vector<VirtualFile> files, std::string name);
+
+public:
+    std::string GetName() const override;
+    size_t GetSize() const override;
+    bool Resize(size_t new_size) override;
+    std::shared_ptr<VfsDirectory> GetContainingDirectory() const override;
+    bool IsWritable() const override;
+    bool IsReadable() const override;
+    size_t Read(u8* data, size_t length, size_t offset) const override;
+    size_t Write(const u8* data, size_t length, size_t offset) override;
+    bool Rename(std::string_view name) override;
+
+private:
+    // Maps starting offset to file -- more efficient.
+    boost::container::flat_map<u64, VirtualFile> files;
+    std::string name;
+};
+
+} // namespace FileSys

src/core/file_sys/vfs_offset.h

@@ -15,7 +15,8 @@ namespace FileSys {
 // Similar to seeking to an offset.
 // If the file is writable, operations that would write past the end of the offset file will expand
 // the size of this wrapper.
-struct OffsetVfsFile : public VfsFile {
+class OffsetVfsFile : public VfsFile {
+public:
     OffsetVfsFile(std::shared_ptr<VfsFile> file, size_t size, size_t offset = 0,
                   std::string new_name = "", VirtualDir new_parent = nullptr);
 

src/core/file_sys/vfs_real.cpp

@@ -83,8 +83,12 @@ VirtualFile RealVfsFilesystem::OpenFile(std::string_view path_, Mode perms) {
 
 VirtualFile RealVfsFilesystem::CreateFile(std::string_view path_, Mode perms) {
     const auto path = FileUtil::SanitizePath(path_, FileUtil::DirectorySeparator::PlatformDefault);
-    if (!FileUtil::Exists(path) && !FileUtil::CreateEmptyFile(path))
-        return nullptr;
+    const auto path_fwd = FileUtil::SanitizePath(path, FileUtil::DirectorySeparator::ForwardSlash);
+    if (!FileUtil::Exists(path)) {
+        FileUtil::CreateFullPath(path_fwd);
+        if (!FileUtil::CreateEmptyFile(path))
+            return nullptr;
+    }
     return OpenFile(path, perms);
 }
 
@@ -140,8 +144,12 @@ VirtualDir RealVfsFilesystem::OpenDirectory(std::string_view path_, Mode perms)
 
 VirtualDir RealVfsFilesystem::CreateDirectory(std::string_view path_, Mode perms) {
     const auto path = FileUtil::SanitizePath(path_, FileUtil::DirectorySeparator::PlatformDefault);
-    if (!FileUtil::Exists(path) && !FileUtil::CreateDir(path))
-        return nullptr;
+    const auto path_fwd = FileUtil::SanitizePath(path, FileUtil::DirectorySeparator::ForwardSlash);
+    if (!FileUtil::Exists(path)) {
+        FileUtil::CreateFullPath(path_fwd);
+        if (!FileUtil::CreateDir(path))
+            return nullptr;
+    }
     // Cannot use make_shared as RealVfsDirectory constructor is private
     return std::shared_ptr<RealVfsDirectory>(new RealVfsDirectory(*this, path, perms));
 }
@@ -306,14 +314,14 @@ RealVfsDirectory::RealVfsDirectory(RealVfsFilesystem& base_, const std::string&
 
 std::shared_ptr<VfsFile> RealVfsDirectory::GetFileRelative(std::string_view path) const {
     const auto full_path = FileUtil::SanitizePath(this->path + DIR_SEP + std::string(path));
-    if (!FileUtil::Exists(full_path))
+    if (!FileUtil::Exists(full_path) || FileUtil::IsDirectory(full_path))
         return nullptr;
     return base.OpenFile(full_path, perms);
 }
 
 std::shared_ptr<VfsDirectory> RealVfsDirectory::GetDirectoryRelative(std::string_view path) const {
     const auto full_path = FileUtil::SanitizePath(this->path + DIR_SEP + std::string(path));
-    if (!FileUtil::Exists(full_path))
+    if (!FileUtil::Exists(full_path) || !FileUtil::IsDirectory(full_path))
         return nullptr;
     return base.OpenDirectory(full_path, perms);
 }

src/core/file_sys/vfs_real.h

@@ -5,7 +5,6 @@
 #pragma once
 
 #include <string_view>
-
 #include <boost/container/flat_map.hpp>
 #include "common/file_util.h"
 #include "core/file_sys/mode.h"

src/core/file_sys/vfs_vector.cpp

@@ -8,8 +8,8 @@
 
 namespace FileSys {
 VectorVfsDirectory::VectorVfsDirectory(std::vector<VirtualFile> files_,
-                                       std::vector<VirtualDir> dirs_, VirtualDir parent_,
-                                       std::string name_)
+                                       std::vector<VirtualDir> dirs_, std::string name_,
+                                       VirtualDir parent_)
     : files(std::move(files_)), dirs(std::move(dirs_)), parent(std::move(parent_)),
       name(std::move(name_)) {}
 

src/core/file_sys/vfs_vector.h

@@ -10,10 +10,11 @@ namespace FileSys {
 
 // An implementation of VfsDirectory that maintains two vectors for subdirectories and files.
 // Vector data is supplied upon construction.
-struct VectorVfsDirectory : public VfsDirectory {
+class VectorVfsDirectory : public VfsDirectory {
+public:
     explicit VectorVfsDirectory(std::vector<VirtualFile> files = {},
-                                std::vector<VirtualDir> dirs = {}, VirtualDir parent = nullptr,
-                                std::string name = "");
+                                std::vector<VirtualDir> dirs = {}, std::string name = "",
+                                VirtualDir parent = nullptr);
 
     std::vector<std::shared_ptr<VfsFile>> GetFiles() const override;
     std::vector<std::shared_ptr<VfsDirectory>> GetSubdirectories() const override;

src/core/frontend/emu_window.cpp

@@ -8,6 +8,8 @@
 #include "core/frontend/input.h"
 #include "core/settings.h"
 
+namespace Core::Frontend {
+
 class EmuWindow::TouchState : public Input::Factory<Input::TouchDevice>,
                               public std::enable_shared_from_this<TouchState> {
 public:
@@ -108,3 +110,5 @@ void EmuWindow::TouchMoved(unsigned framebuffer_x, unsigned framebuffer_y) {
 void EmuWindow::UpdateCurrentFramebufferLayout(unsigned width, unsigned height) {
     NotifyFramebufferLayoutChanged(Layout::DefaultFrameLayout(width, height));
 }
+
+} // namespace Core::Frontend

src/core/frontend/emu_window.h

@@ -10,6 +10,8 @@
 #include "common/common_types.h"
 #include "core/frontend/framebuffer_layout.h"
 
+namespace Core::Frontend {
+
 /**
  * Abstraction class used to provide an interface between emulation code and the frontend
  * (e.g. SDL, QGLWidget, GLFW, etc...).
@@ -32,9 +34,9 @@ class EmuWindow {
 public:
     /// Data structure to store emuwindow configuration
     struct WindowConfig {
-        bool fullscreen;
-        int res_width;
-        int res_height;
+        bool fullscreen = false;
+        int res_width = 0;
+        int res_height = 0;
         std::pair<unsigned, unsigned> min_client_area_size;
     };
 
@@ -166,3 +168,5 @@ private:
      */
     std::tuple<unsigned, unsigned> ClipToTouchScreen(unsigned new_x, unsigned new_y);
 };
+
+} // namespace Core::Frontend

src/core/hle/kernel/kernel.cpp

@@ -11,7 +11,7 @@
 
 namespace Kernel {
 
-unsigned int Object::next_object_id;
+std::atomic<u32> Object::next_object_id{0};
 
 /// Initialize the kernel
 void Init() {

src/core/hle/kernel/object.h

@@ -4,6 +4,7 @@
 
 #pragma once
 
+#include <atomic>
 #include <string>
 #include <utility>
 
@@ -42,8 +43,8 @@ public:
     virtual ~Object();
 
     /// Returns a unique identifier for the object. For debugging purposes only.
-    unsigned int GetObjectId() const {
-        return object_id;
+    u32 GetObjectId() const {
+        return object_id.load(std::memory_order_relaxed);
     }
 
     virtual std::string GetTypeName() const {
@@ -61,23 +62,23 @@ public:
     bool IsWaitable() const;
 
 public:
-    static unsigned int next_object_id;
+    static std::atomic<u32> next_object_id;
 
 private:
     friend void intrusive_ptr_add_ref(Object*);
     friend void intrusive_ptr_release(Object*);
 
-    unsigned int ref_count = 0;
-    unsigned int object_id = next_object_id++;
+    std::atomic<u32> ref_count{0};
+    std::atomic<u32> object_id{next_object_id++};
 };
 
 // Special functions used by boost::instrusive_ptr to do automatic ref-counting
 inline void intrusive_ptr_add_ref(Object* object) {
-    ++object->ref_count;
+    object->ref_count.fetch_add(1, std::memory_order_relaxed);
 }
 
 inline void intrusive_ptr_release(Object* object) {
-    if (--object->ref_count == 0) {
+    if (object->ref_count.fetch_sub(1, std::memory_order_acq_rel) == 1) {
         delete object;
     }
 }

src/core/hle/kernel/scheduler.cpp

@@ -25,7 +25,7 @@ Scheduler::~Scheduler() {
     }
 }
 
-bool Scheduler::HaveReadyThreads() {
+bool Scheduler::HaveReadyThreads() const {
     std::lock_guard<std::mutex> lock(scheduler_mutex);
     return ready_queue.get_first() != nullptr;
 }

src/core/hle/kernel/scheduler.h

@@ -21,7 +21,7 @@ public:
     ~Scheduler();
 
     /// Returns whether there are any threads that are ready to run.
-    bool HaveReadyThreads();
+    bool HaveReadyThreads() const;
 
     /// Reschedules to the next available thread (call after current thread is suspended)
     void Reschedule();

src/core/hle/kernel/server_session.cpp

@@ -71,6 +71,14 @@ ResultCode ServerSession::HandleDomainSyncRequest(Kernel::HLERequestContext& con
         const u32 object_id{context.GetDomainMessageHeader()->object_id};
         switch (domain_message_header->command) {
         case IPC::DomainMessageHeader::CommandType::SendMessage:
+            if (object_id > domain_request_handlers.size()) {
+                LOG_CRITICAL(IPC,
+                             "object_id {} is too big! This probably means a recent service call "
+                             "to {} needed to return a new interface!",
+                             object_id, name);
+                UNREACHABLE();
+                return RESULT_SUCCESS; // Ignore error if asserts are off
+            }
             return domain_request_handlers[object_id - 1]->HandleSyncRequest(context);
 
         case IPC::DomainMessageHeader::CommandType::CloseVirtualHandle: {
@@ -144,7 +152,7 @@ ResultCode ServerSession::HandleSyncRequest(SharedPtr<Thread> thread) {
     // Handle scenario when ConvertToDomain command was issued, as we must do the conversion at the
     // end of the command such that only commands following this one are handled as domains
     if (convert_to_domain) {
-        ASSERT_MSG(domain_request_handlers.empty(), "already a domain");
+        ASSERT_MSG(IsSession(), "ServerSession is already a domain instance.");
         domain_request_handlers = {hle_handler};
         convert_to_domain = false;
     }

src/core/hle/kernel/server_session.h

@@ -97,7 +97,12 @@ public:
 
     /// Returns true if the session has been converted to a domain, otherwise False
     bool IsDomain() const {
-        return !domain_request_handlers.empty();
+        return !IsSession();
+    }
+
+    /// Returns true if this session has not been converted to a domain, otherwise false.
+    bool IsSession() const {
+        return domain_request_handlers.empty();
     }
 
     /// Converts the session to a domain at the end of the current command

src/core/hle/kernel/svc.cpp

@@ -250,8 +250,11 @@ static ResultCode ArbitrateUnlock(VAddr mutex_addr) {
 }
 
 /// Break program execution
-static void Break(u64 unk_0, u64 unk_1, u64 unk_2) {
-    LOG_CRITICAL(Debug_Emulated, "Emulated program broke execution!");
+static void Break(u64 reason, u64 info1, u64 info2) {
+    LOG_CRITICAL(
+        Debug_Emulated,
+        "Emulated program broke execution! reason=0x{:016X}, info1=0x{:016X}, info2=0x{:016X}",
+        reason, info1, info2);
     ASSERT(false);
 }
 
@@ -532,7 +535,6 @@ static ResultCode CreateThread(Handle* out_handle, VAddr entry_point, u64 arg, V
     CASCADE_RESULT(thread->guest_handle, g_handle_table.Create(thread));
     *out_handle = thread->guest_handle;
 
-    Core::System::GetInstance().PrepareReschedule();
     Core::System::GetInstance().CpuCore(thread->processor_id).PrepareReschedule();
 
     LOG_TRACE(Kernel_SVC,
@@ -706,8 +708,7 @@ static ResultCode SignalProcessWideKey(VAddr condition_variable_addr, s32 target
             Handle owner_handle = static_cast<Handle>(mutex_val & Mutex::MutexOwnerMask);
             auto owner = g_handle_table.Get<Thread>(owner_handle);
             ASSERT(owner);
-            ASSERT(thread->status != ThreadStatus::Running);
-            thread->status = ThreadStatus::WaitMutex;
+            ASSERT(thread->status == ThreadStatus::WaitMutex);
             thread->wakeup_callback = nullptr;
 
             owner->AddMutexWaiter(thread);

src/core/hle/kernel/thread.cpp

@@ -23,6 +23,7 @@
 #include "core/hle/kernel/object.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/thread.h"
+#include "core/hle/lock.h"
 #include "core/hle/result.h"
 #include "core/memory.h"
 
@@ -104,6 +105,10 @@ void ExitCurrentThread() {
  */
 static void ThreadWakeupCallback(u64 thread_handle, int cycles_late) {
     const auto proper_handle = static_cast<Handle>(thread_handle);
+
+    // Lock the global kernel mutex when we enter the kernel HLE.
+    std::lock_guard<std::recursive_mutex> lock(HLE::g_hle_lock);
+
     SharedPtr<Thread> thread = wakeup_callback_handle_table.Get<Thread>(proper_handle);
     if (thread == nullptr) {
         LOG_CRITICAL(Kernel, "Callback fired for invalid thread {:08X}", proper_handle);
@@ -155,12 +160,14 @@ void Thread::WakeAfterDelay(s64 nanoseconds) {
     if (nanoseconds == -1)
         return;
 
-    CoreTiming::ScheduleEvent(CoreTiming::nsToCycles(nanoseconds), ThreadWakeupEventType,
-                              callback_handle);
+    // This function might be called from any thread so we have to be cautious and use the
+    // thread-safe version of ScheduleEvent.
+    CoreTiming::ScheduleEventThreadsafe(CoreTiming::nsToCycles(nanoseconds), ThreadWakeupEventType,
+                                        callback_handle);
 }
 
 void Thread::CancelWakeupTimer() {
-    CoreTiming::UnscheduleEvent(ThreadWakeupEventType, callback_handle);
+    CoreTiming::UnscheduleEventThreadsafe(ThreadWakeupEventType, callback_handle);
 }
 
 static boost::optional<s32> GetNextProcessorId(u64 mask) {
@@ -419,12 +426,33 @@ VAddr Thread::GetCommandBufferAddress() const {
 }
 
 void Thread::AddMutexWaiter(SharedPtr<Thread> thread) {
+    if (thread->lock_owner == this) {
+        // If the thread is already waiting for this thread to release the mutex, ensure that the
+        // waiters list is consistent and return without doing anything.
+        auto itr = std::find(wait_mutex_threads.begin(), wait_mutex_threads.end(), thread);
+        ASSERT(itr != wait_mutex_threads.end());
+        return;
+    }
+
+    // A thread can't wait on two different mutexes at the same time.
+    ASSERT(thread->lock_owner == nullptr);
+
+    // Ensure that the thread is not already in the list of mutex waiters
+    auto itr = std::find(wait_mutex_threads.begin(), wait_mutex_threads.end(), thread);
+    ASSERT(itr == wait_mutex_threads.end());
+
     thread->lock_owner = this;
     wait_mutex_threads.emplace_back(std::move(thread));
     UpdatePriority();
 }
 
 void Thread::RemoveMutexWaiter(SharedPtr<Thread> thread) {
+    ASSERT(thread->lock_owner == this);
+
+    // Ensure that the thread is in the list of mutex waiters
+    auto itr = std::find(wait_mutex_threads.begin(), wait_mutex_threads.end(), thread);
+    ASSERT(itr != wait_mutex_threads.end());
+
     boost::remove_erase(wait_mutex_threads, thread);
     thread->lock_owner = nullptr;
     UpdatePriority();

src/core/hle/service/am/am.cpp

@@ -306,7 +306,8 @@ ICommonStateGetter::ICommonStateGetter() : ServiceFramework("ICommonStateGetter"
         {52, nullptr, "SwitchLcdBacklight"},
         {55, nullptr, "IsInControllerFirmwareUpdateSection"},
         {60, nullptr, "GetDefaultDisplayResolution"},
-        {61, nullptr, "GetDefaultDisplayResolutionChangeEvent"},
+        {61, &ICommonStateGetter::GetDefaultDisplayResolutionChangeEvent,
+         "GetDefaultDisplayResolutionChangeEvent"},
         {62, nullptr, "GetHdcpAuthenticationState"},
         {63, nullptr, "GetHdcpAuthenticationStateChangeEvent"},
     };
@@ -341,6 +342,16 @@ void ICommonStateGetter::GetCurrentFocusState(Kernel::HLERequestContext& ctx) {
     LOG_WARNING(Service_AM, "(STUBBED) called");
 }
 
+void ICommonStateGetter::GetDefaultDisplayResolutionChangeEvent(Kernel::HLERequestContext& ctx) {
+    event->Signal();
+
+    IPC::ResponseBuilder rb{ctx, 2, 1};
+    rb.Push(RESULT_SUCCESS);
+    rb.PushCopyObjects(event);
+
+    LOG_WARNING(Service_AM, "(STUBBED) called");
+}
+
 void ICommonStateGetter::GetOperationMode(Kernel::HLERequestContext& ctx) {
     const bool use_docked_mode{Settings::values.use_docked_mode};
     IPC::ResponseBuilder rb{ctx, 3};

src/core/hle/service/am/am.h

@@ -110,6 +110,7 @@ private:
     void GetEventHandle(Kernel::HLERequestContext& ctx);
     void ReceiveMessage(Kernel::HLERequestContext& ctx);
     void GetCurrentFocusState(Kernel::HLERequestContext& ctx);
+    void GetDefaultDisplayResolutionChangeEvent(Kernel::HLERequestContext& ctx);
     void GetOperationMode(Kernel::HLERequestContext& ctx);
     void GetPerformanceMode(Kernel::HLERequestContext& ctx);
 

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

@@ -28,7 +28,7 @@ constexpr int DefaultSampleRate{48000};
 class IAudioOut final : public ServiceFramework<IAudioOut> {
 public:
     IAudioOut(AudoutParams audio_params, AudioCore::AudioOut& audio_core)
-        : ServiceFramework("IAudioOut"), audio_params(audio_params), audio_core(audio_core) {
+        : ServiceFramework("IAudioOut"), audio_core(audio_core), audio_params(audio_params) {
 
         static const FunctionInfo functions[] = {
             {0, &IAudioOut::GetAudioOutState, "GetAudioOutState"},

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

@@ -20,9 +20,9 @@ public:
     explicit IAudioRenderer(AudioCore::AudioRendererParameter audren_params)
         : ServiceFramework("IAudioRenderer") {
         static const FunctionInfo functions[] = {
-            {0, nullptr, "GetAudioRendererSampleRate"},
-            {1, nullptr, "GetAudioRendererSampleCount"},
-            {2, nullptr, "GetAudioRendererMixBufferCount"},
+            {0, &IAudioRenderer::GetAudioRendererSampleRate, "GetAudioRendererSampleRate"},
+            {1, &IAudioRenderer::GetAudioRendererSampleCount, "GetAudioRendererSampleCount"},
+            {2, &IAudioRenderer::GetAudioRendererMixBufferCount, "GetAudioRendererMixBufferCount"},
             {3, nullptr, "GetAudioRendererState"},
             {4, &IAudioRenderer::RequestUpdateAudioRenderer, "RequestUpdateAudioRenderer"},
             {5, &IAudioRenderer::StartAudioRenderer, "StartAudioRenderer"},
@@ -45,6 +45,27 @@ private:
         system_event->Signal();
     }
 
+    void GetAudioRendererSampleRate(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(RESULT_SUCCESS);
+        rb.Push<u32>(renderer->GetSampleRate());
+        LOG_DEBUG(Service_Audio, "called");
+    }
+
+    void GetAudioRendererSampleCount(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(RESULT_SUCCESS);
+        rb.Push<u32>(renderer->GetSampleCount());
+        LOG_DEBUG(Service_Audio, "called");
+    }
+
+    void GetAudioRendererMixBufferCount(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(RESULT_SUCCESS);
+        rb.Push<u32>(renderer->GetMixBufferCount());
+        LOG_DEBUG(Service_Audio, "called");
+    }
+
     void RequestUpdateAudioRenderer(Kernel::HLERequestContext& ctx) {
         ctx.WriteBuffer(renderer->UpdateAudioRenderer(ctx.ReadBuffer()));
         IPC::ResponseBuilder rb{ctx, 2};
@@ -169,7 +190,8 @@ AudRenU::AudRenU() : ServiceFramework("audren:u") {
         {1, &AudRenU::GetAudioRendererWorkBufferSize, "GetAudioRendererWorkBufferSize"},
         {2, &AudRenU::GetAudioDevice, "GetAudioDevice"},
         {3, nullptr, "OpenAudioRendererAuto"},
-        {4, nullptr, "GetAudioDeviceServiceWithRevisionInfo"},
+        {4, &AudRenU::GetAudioDeviceServiceWithRevisionInfo,
+         "GetAudioDeviceServiceWithRevisionInfo"},
     };
     RegisterHandlers(functions);
 }
@@ -189,7 +211,7 @@ void AudRenU::GetAudioRendererWorkBufferSize(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
     auto params = rp.PopRaw<AudioCore::AudioRendererParameter>();
 
-    u64 buffer_sz = Common::AlignUp(4 * params.unknown_8, 0x40);
+    u64 buffer_sz = Common::AlignUp(4 * params.mix_buffer_count, 0x40);
     buffer_sz += params.unknown_c * 1024;
     buffer_sz += 0x940 * (params.unknown_c + 1);
     buffer_sz += 0x3F0 * params.voice_count;
@@ -197,7 +219,7 @@ void AudRenU::GetAudioRendererWorkBufferSize(Kernel::HLERequestContext& ctx) {
     buffer_sz += Common::AlignUp(8 * params.voice_count, 0x10);
     buffer_sz +=
         Common::AlignUp((0x3C0 * (params.sink_count + params.unknown_c) + 4 * params.sample_count) *
-                            (params.unknown_8 + 6),
+                            (params.mix_buffer_count + 6),
                         0x40);
 
     if (IsFeatureSupported(AudioFeatures::Splitter, params.revision)) {
@@ -253,6 +275,16 @@ void AudRenU::GetAudioDevice(Kernel::HLERequestContext& ctx) {
     LOG_DEBUG(Service_Audio, "called");
 }
 
+void AudRenU::GetAudioDeviceServiceWithRevisionInfo(Kernel::HLERequestContext& ctx) {
+    IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+
+    rb.Push(RESULT_SUCCESS);
+    rb.PushIpcInterface<Audio::IAudioDevice>();
+
+    LOG_WARNING(Service_Audio, "(STUBBED) called"); // TODO(ogniK): Figure out what is different
+                                                    // based on the current revision
+}
+
 bool AudRenU::IsFeatureSupported(AudioFeatures feature, u32_le revision) const {
     u32_be version_num = (revision - Common::MakeMagic('R', 'E', 'V', '0')); // Byte swap
     switch (feature) {

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

@@ -22,6 +22,7 @@ private:
     void OpenAudioRenderer(Kernel::HLERequestContext& ctx);
     void GetAudioRendererWorkBufferSize(Kernel::HLERequestContext& ctx);
     void GetAudioDevice(Kernel::HLERequestContext& ctx);
+    void GetAudioDeviceServiceWithRevisionInfo(Kernel::HLERequestContext& ctx);
 
     enum class AudioFeatures : u32 {
         Splitter,

src/core/hle/service/filesystem/filesystem.cpp

@@ -226,6 +226,7 @@ ResultVal<FileSys::EntryType> VfsDirectoryServiceWrapper::GetEntryType(
 static std::unique_ptr<FileSys::RomFSFactory> romfs_factory;
 static std::unique_ptr<FileSys::SaveDataFactory> save_data_factory;
 static std::unique_ptr<FileSys::SDMCFactory> sdmc_factory;
+static std::unique_ptr<FileSys::BISFactory> bis_factory;
 
 ResultCode RegisterRomFS(std::unique_ptr<FileSys::RomFSFactory>&& factory) {
     ASSERT_MSG(romfs_factory == nullptr, "Tried to register a second RomFS");
@@ -248,6 +249,13 @@ ResultCode RegisterSDMC(std::unique_ptr<FileSys::SDMCFactory>&& factory) {
     return RESULT_SUCCESS;
 }
 
+ResultCode RegisterBIS(std::unique_ptr<FileSys::BISFactory>&& factory) {
+    ASSERT_MSG(bis_factory == nullptr, "Tried to register a second BIS");
+    bis_factory = std::move(factory);
+    LOG_DEBUG(Service_FS, "Registred BIS");
+    return RESULT_SUCCESS;
+}
+
 ResultVal<FileSys::VirtualFile> OpenRomFS(u64 title_id) {
     LOG_TRACE(Service_FS, "Opening RomFS for title_id={:016X}", title_id);
 
@@ -281,6 +289,14 @@ ResultVal<FileSys::VirtualDir> OpenSDMC() {
     return sdmc_factory->Open();
 }
 
+std::shared_ptr<FileSys::RegisteredCache> GetSystemNANDContents() {
+    return bis_factory->GetSystemNANDContents();
+}
+
+std::shared_ptr<FileSys::RegisteredCache> GetUserNANDContents() {
+    return bis_factory->GetUserNANDContents();
+}
+
 void RegisterFileSystems(const FileSys::VirtualFilesystem& vfs) {
     romfs_factory = nullptr;
     save_data_factory = nullptr;
@@ -291,6 +307,9 @@ void RegisterFileSystems(const FileSys::VirtualFilesystem& vfs) {
     auto sd_directory = vfs->OpenDirectory(FileUtil::GetUserPath(FileUtil::UserPath::SDMCDir),
                                            FileSys::Mode::ReadWrite);
 
+    if (bis_factory == nullptr)
+        bis_factory = std::make_unique<FileSys::BISFactory>(nand_directory);
+
     auto savedata = std::make_unique<FileSys::SaveDataFactory>(std::move(nand_directory));
     save_data_factory = std::move(savedata);
 

src/core/hle/service/filesystem/filesystem.h

@@ -6,6 +6,7 @@
 
 #include <memory>
 #include "common/common_types.h"
+#include "core/file_sys/bis_factory.h"
 #include "core/file_sys/directory.h"
 #include "core/file_sys/mode.h"
 #include "core/file_sys/romfs_factory.h"
@@ -24,16 +25,15 @@ namespace FileSystem {
 ResultCode RegisterRomFS(std::unique_ptr<FileSys::RomFSFactory>&& factory);
 ResultCode RegisterSaveData(std::unique_ptr<FileSys::SaveDataFactory>&& factory);
 ResultCode RegisterSDMC(std::unique_ptr<FileSys::SDMCFactory>&& factory);
+ResultCode RegisterBIS(std::unique_ptr<FileSys::BISFactory>&& factory);
 
-// TODO(DarkLordZach): BIS Filesystem
-// ResultCode RegisterBIS(std::unique_ptr<FileSys::BISFactory>&& factory);
 ResultVal<FileSys::VirtualFile> OpenRomFS(u64 title_id);
 ResultVal<FileSys::VirtualDir> OpenSaveData(FileSys::SaveDataSpaceId space,
                                             FileSys::SaveDataDescriptor save_struct);
 ResultVal<FileSys::VirtualDir> OpenSDMC();
 
-// TODO(DarkLordZach): BIS Filesystem
-// ResultVal<std::unique_ptr<FileSys::FileSystemBackend>> OpenBIS();
+std::shared_ptr<FileSys::RegisteredCache> GetSystemNANDContents();
+std::shared_ptr<FileSys::RegisteredCache> GetUserNANDContents();
 
 /// Registers all Filesystem services with the specified service manager.
 void InstallInterfaces(SM::ServiceManager& service_manager, const FileSys::VirtualFilesystem& vfs);

src/core/hle/service/friend/friend.cpp

@@ -9,10 +9,110 @@
 
 namespace Service::Friend {
 
+class IFriendService final : public ServiceFramework<IFriendService> {
+public:
+    IFriendService() : ServiceFramework("IFriendService") {
+        static const FunctionInfo functions[] = {
+            {0, nullptr, "GetCompletionEvent"},
+            {1, nullptr, "Cancel"},
+            {10100, nullptr, "GetFriendListIds"},
+            {10101, nullptr, "GetFriendList"},
+            {10102, nullptr, "UpdateFriendInfo"},
+            {10110, nullptr, "GetFriendProfileImage"},
+            {10200, nullptr, "SendFriendRequestForApplication"},
+            {10211, nullptr, "AddFacedFriendRequestForApplication"},
+            {10400, nullptr, "GetBlockedUserListIds"},
+            {10500, nullptr, "GetProfileList"},
+            {10600, nullptr, "DeclareOpenOnlinePlaySession"},
+            {10601, &IFriendService::DeclareCloseOnlinePlaySession,
+             "DeclareCloseOnlinePlaySession"},
+            {10610, &IFriendService::UpdateUserPresence, "UpdateUserPresence"},
+            {10700, nullptr, "GetPlayHistoryRegistrationKey"},
+            {10701, nullptr, "GetPlayHistoryRegistrationKeyWithNetworkServiceAccountId"},
+            {10702, nullptr, "AddPlayHistory"},
+            {11000, nullptr, "GetProfileImageUrl"},
+            {20100, nullptr, "GetFriendCount"},
+            {20101, nullptr, "GetNewlyFriendCount"},
+            {20102, nullptr, "GetFriendDetailedInfo"},
+            {20103, nullptr, "SyncFriendList"},
+            {20104, nullptr, "RequestSyncFriendList"},
+            {20110, nullptr, "LoadFriendSetting"},
+            {20200, nullptr, "GetReceivedFriendRequestCount"},
+            {20201, nullptr, "GetFriendRequestList"},
+            {20300, nullptr, "GetFriendCandidateList"},
+            {20301, nullptr, "GetNintendoNetworkIdInfo"},
+            {20302, nullptr, "GetSnsAccountLinkage"},
+            {20303, nullptr, "GetSnsAccountProfile"},
+            {20304, nullptr, "GetSnsAccountFriendList"},
+            {20400, nullptr, "GetBlockedUserList"},
+            {20401, nullptr, "SyncBlockedUserList"},
+            {20500, nullptr, "GetProfileExtraList"},
+            {20501, nullptr, "GetRelationship"},
+            {20600, nullptr, "GetUserPresenceView"},
+            {20700, nullptr, "GetPlayHistoryList"},
+            {20701, nullptr, "GetPlayHistoryStatistics"},
+            {20800, nullptr, "LoadUserSetting"},
+            {20801, nullptr, "SyncUserSetting"},
+            {20900, nullptr, "RequestListSummaryOverlayNotification"},
+            {21000, nullptr, "GetExternalApplicationCatalog"},
+            {30100, nullptr, "DropFriendNewlyFlags"},
+            {30101, nullptr, "DeleteFriend"},
+            {30110, nullptr, "DropFriendNewlyFlag"},
+            {30120, nullptr, "ChangeFriendFavoriteFlag"},
+            {30121, nullptr, "ChangeFriendOnlineNotificationFlag"},
+            {30200, nullptr, "SendFriendRequest"},
+            {30201, nullptr, "SendFriendRequestWithApplicationInfo"},
+            {30202, nullptr, "CancelFriendRequest"},
+            {30203, nullptr, "AcceptFriendRequest"},
+            {30204, nullptr, "RejectFriendRequest"},
+            {30205, nullptr, "ReadFriendRequest"},
+            {30210, nullptr, "GetFacedFriendRequestRegistrationKey"},
+            {30211, nullptr, "AddFacedFriendRequest"},
+            {30212, nullptr, "CancelFacedFriendRequest"},
+            {30213, nullptr, "GetFacedFriendRequestProfileImage"},
+            {30214, nullptr, "GetFacedFriendRequestProfileImageFromPath"},
+            {30215, nullptr, "SendFriendRequestWithExternalApplicationCatalogId"},
+            {30216, nullptr, "ResendFacedFriendRequest"},
+            {30217, nullptr, "SendFriendRequestWithNintendoNetworkIdInfo"},
+            {30300, nullptr, "GetSnsAccountLinkPageUrl"},
+            {30301, nullptr, "UnlinkSnsAccount"},
+            {30400, nullptr, "BlockUser"},
+            {30401, nullptr, "BlockUserWithApplicationInfo"},
+            {30402, nullptr, "UnblockUser"},
+            {30500, nullptr, "GetProfileExtraFromFriendCode"},
+            {30700, nullptr, "DeletePlayHistory"},
+            {30810, nullptr, "ChangePresencePermission"},
+            {30811, nullptr, "ChangeFriendRequestReception"},
+            {30812, nullptr, "ChangePlayLogPermission"},
+            {30820, nullptr, "IssueFriendCode"},
+            {30830, nullptr, "ClearPlayLog"},
+            {49900, nullptr, "DeleteNetworkServiceAccountCache"},
+        };
+
+        RegisterHandlers(functions);
+    }
+
+private:
+    void DeclareCloseOnlinePlaySession(Kernel::HLERequestContext& ctx) {
+        // Stub used by Splatoon 2
+        LOG_WARNING(Service_ACC, "(STUBBED) called");
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+    }
+
+    void UpdateUserPresence(Kernel::HLERequestContext& ctx) {
+        // Stub used by Retro City Rampage
+        LOG_WARNING(Service_ACC, "(STUBBED) called");
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+    }
+};
+
 void Module::Interface::CreateFriendService(Kernel::HLERequestContext& ctx) {
-    IPC::ResponseBuilder rb{ctx, 2};
+    IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);
-    LOG_WARNING(Service_Friend, "(STUBBED) called");
+    rb.PushIpcInterface<IFriendService>();
+    LOG_DEBUG(Service_ACC, "called");
 }
 
 Module::Interface::Interface(std::shared_ptr<Module> module, const char* name)

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

@@ -291,6 +291,7 @@ private:
 class Hid final : public ServiceFramework<Hid> {
 public:
     Hid() : ServiceFramework("hid") {
+        // clang-format off
         static const FunctionInfo functions[] = {
             {0, &Hid::CreateAppletResource, "CreateAppletResource"},
             {1, &Hid::ActivateDebugPad, "ActivateDebugPad"},
@@ -333,15 +334,13 @@ public:
             {102, &Hid::SetSupportedNpadIdType, "SetSupportedNpadIdType"},
             {103, &Hid::ActivateNpad, "ActivateNpad"},
             {104, nullptr, "DeactivateNpad"},
-            {106, &Hid::AcquireNpadStyleSetUpdateEventHandle,
-             "AcquireNpadStyleSetUpdateEventHandle"},
-            {107, nullptr, "DisconnectNpad"},
+            {106, &Hid::AcquireNpadStyleSetUpdateEventHandle, "AcquireNpadStyleSetUpdateEventHandle"},
+            {107, &Hid::DisconnectNpad, "DisconnectNpad"},
             {108, &Hid::GetPlayerLedPattern, "GetPlayerLedPattern"},
             {109, nullptr, "ActivateNpadWithRevision"},
             {120, &Hid::SetNpadJoyHoldType, "SetNpadJoyHoldType"},
             {121, &Hid::GetNpadJoyHoldType, "GetNpadJoyHoldType"},
-            {122, &Hid::SetNpadJoyAssignmentModeSingleByDefault,
-             "SetNpadJoyAssignmentModeSingleByDefault"},
+            {122, &Hid::SetNpadJoyAssignmentModeSingleByDefault, "SetNpadJoyAssignmentModeSingleByDefault"},
             {123, nullptr, "SetNpadJoyAssignmentModeSingleByDefault"},
             {124, &Hid::SetNpadJoyAssignmentModeDual, "SetNpadJoyAssignmentModeDual"},
             {125, &Hid::MergeSingleJoyAsDualJoy, "MergeSingleJoyAsDualJoy"},
@@ -398,6 +397,8 @@ public:
             {1000, nullptr, "SetNpadCommunicationMode"},
             {1001, nullptr, "GetNpadCommunicationMode"},
         };
+        // clang-format on
+
         RegisterHandlers(functions);
 
         event = Kernel::Event::Create(Kernel::ResetType::OneShot, "hid:EventHandle");
@@ -496,6 +497,12 @@ private:
         LOG_WARNING(Service_HID, "(STUBBED) called");
     }
 
+    void DisconnectNpad(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+        LOG_WARNING(Service_HID, "(STUBBED) called");
+    }
+
     void GetPlayerLedPattern(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);

src/core/hle/service/lm/lm.cpp

@@ -92,7 +92,11 @@ private:
 
         // Parse out log metadata
         u32 line{};
-        std::string message, filename, function;
+        std::string module;
+        std::string message;
+        std::string filename;
+        std::string function;
+        std::string thread;
         while (addr < end_addr) {
             const Field field{static_cast<Field>(Memory::Read8(addr++))};
             const size_t length{Memory::Read8(addr++)};
@@ -102,6 +106,8 @@ private:
             }
 
             switch (field) {
+            case Field::Skip:
+                break;
             case Field::Message:
                 message = Memory::ReadCString(addr, length);
                 break;
@@ -114,6 +120,12 @@ private:
             case Field::Function:
                 function = Memory::ReadCString(addr, length);
                 break;
+            case Field::Module:
+                module = Memory::ReadCString(addr, length);
+                break;
+            case Field::Thread:
+                thread = Memory::ReadCString(addr, length);
+                break;
             }
 
             addr += length;
@@ -128,12 +140,18 @@ private:
         if (!filename.empty()) {
             log_stream << filename << ':';
         }
+        if (!module.empty()) {
+            log_stream << module << ':';
+        }
         if (!function.empty()) {
             log_stream << function << ':';
         }
         if (line) {
             log_stream << std::to_string(line) << ':';
         }
+        if (!thread.empty()) {
+            log_stream << thread << ':';
+        }
         if (log_stream.str().length() > 0 && log_stream.str().back() == ':') {
             log_stream << ' ';
         }
@@ -142,7 +160,7 @@ private:
         if (header.IsTailLog()) {
             switch (header.severity) {
             case MessageHeader::Severity::Trace:
-                LOG_TRACE(Debug_Emulated, "{}", log_stream.str());
+                LOG_DEBUG(Debug_Emulated, "{}", log_stream.str());
                 break;
             case MessageHeader::Severity::Info:
                 LOG_INFO(Debug_Emulated, "{}", log_stream.str());

src/core/hle/service/mm/mm_u.cpp

@@ -9,42 +9,63 @@
 
 namespace Service::MM {
 
+class MM_U final : public ServiceFramework<MM_U> {
+public:
+    explicit MM_U() : ServiceFramework{"mm:u"} {
+        // clang-format off
+        static const FunctionInfo functions[] = {
+            {0, &MM_U::Initialize, "InitializeOld"},
+            {1, &MM_U::Finalize, "FinalizeOld"},
+            {2, &MM_U::SetAndWait, "SetAndWaitOld"},
+            {3, &MM_U::Get, "GetOld"},
+            {4, &MM_U::Initialize, "Initialize"},
+            {5, &MM_U::Finalize, "Finalize"},
+            {6, &MM_U::SetAndWait, "SetAndWait"},
+            {7, &MM_U::Get, "Get"},
+        };
+        // clang-format on
+
+        RegisterHandlers(functions);
+    }
+
+private:
+    void Initialize(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_MM, "(STUBBED) called");
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+    }
+
+    void Finalize(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_MM, "(STUBBED) called");
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+    }
+
+    void SetAndWait(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        min = rp.Pop<u32>();
+        max = rp.Pop<u32>();
+        current = min;
+
+        LOG_WARNING(Service_MM, "(STUBBED) called, min=0x{:X}, max=0x{:X}", min, max);
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+    }
+
+    void Get(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_MM, "(STUBBED) called");
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(RESULT_SUCCESS);
+        rb.Push(current);
+    }
+
+    u32 min{0};
+    u32 max{0};
+    u32 current{0};
+};
+
 void InstallInterfaces(SM::ServiceManager& service_manager) {
     std::make_shared<MM_U>()->InstallAsService(service_manager);
 }
 
-void MM_U::Initialize(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_MM, "(STUBBED) called");
-    IPC::ResponseBuilder rb{ctx, 2};
-    rb.Push(RESULT_SUCCESS);
-}
-
-void MM_U::SetAndWait(Kernel::HLERequestContext& ctx) {
-    IPC::RequestParser rp{ctx};
-    min = rp.Pop<u32>();
-    max = rp.Pop<u32>();
-    current = min;
-
-    LOG_WARNING(Service_MM, "(STUBBED) called, min=0x{:X}, max=0x{:X}", min, max);
-    IPC::ResponseBuilder rb{ctx, 2};
-    rb.Push(RESULT_SUCCESS);
-}
-
-void MM_U::Get(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_MM, "(STUBBED) called");
-    IPC::ResponseBuilder rb{ctx, 3};
-    rb.Push(RESULT_SUCCESS);
-    rb.Push(current);
-}
-
-MM_U::MM_U() : ServiceFramework("mm:u") {
-    static const FunctionInfo functions[] = {
-        {0, nullptr, "InitializeOld"},        {1, nullptr, "FinalizeOld"},
-        {2, nullptr, "SetAndWaitOld"},        {3, nullptr, "GetOld"},
-        {4, &MM_U::Initialize, "Initialize"}, {5, nullptr, "Finalize"},
-        {6, &MM_U::SetAndWait, "SetAndWait"}, {7, &MM_U::Get, "Get"},
-    };
-    RegisterHandlers(functions);
-}
-
 } // namespace Service::MM

src/core/hle/service/mm/mm_u.h

@@ -8,21 +8,6 @@
 
 namespace Service::MM {
 
-class MM_U final : public ServiceFramework<MM_U> {
-public:
-    MM_U();
-    ~MM_U() = default;
-
-private:
-    void Initialize(Kernel::HLERequestContext& ctx);
-    void SetAndWait(Kernel::HLERequestContext& ctx);
-    void Get(Kernel::HLERequestContext& ctx);
-
-    u32 min{0};
-    u32 max{0};
-    u32 current{0};
-};
-
 /// Registers all MM services with the specified service manager.
 void InstallInterfaces(SM::ServiceManager& service_manager);
 

src/core/hle/service/nvdrv/devices/nvhost_nvjpg.cpp

@@ -0,0 +1,34 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <cstring>
+
+#include "common/assert.h"
+#include "common/logging/log.h"
+#include "core/hle/service/nvdrv/devices/nvhost_nvjpg.h"
+
+namespace Service::Nvidia::Devices {
+
+u32 nvhost_nvjpg::ioctl(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) {
+    LOG_DEBUG(Service_NVDRV, "called, command=0x{:08X}, input_size=0x{:X}, output_size=0x{:X}",
+              command.raw, input.size(), output.size());
+
+    switch (static_cast<IoctlCommand>(command.raw)) {
+    case IoctlCommand::IocSetNVMAPfdCommand:
+        return SetNVMAPfd(input, output);
+    }
+
+    UNIMPLEMENTED_MSG("Unimplemented ioctl");
+    return 0;
+}
+
+u32 nvhost_nvjpg::SetNVMAPfd(const std::vector<u8>& input, std::vector<u8>& output) {
+    IoctlSetNvmapFD params{};
+    std::memcpy(&params, input.data(), input.size());
+    LOG_DEBUG(Service_NVDRV, "called, fd={}", params.nvmap_fd);
+    nvmap_fd = params.nvmap_fd;
+    return 0;
+}
+
+} // namespace Service::Nvidia::Devices

src/core/hle/service/nvdrv/devices/nvhost_nvjpg.h

@@ -0,0 +1,36 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <vector>
+#include "common/common_types.h"
+#include "common/swap.h"
+#include "core/hle/service/nvdrv/devices/nvdevice.h"
+
+namespace Service::Nvidia::Devices {
+
+class nvhost_nvjpg final : public nvdevice {
+public:
+    nvhost_nvjpg() = default;
+    ~nvhost_nvjpg() override = default;
+
+    u32 ioctl(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override;
+
+private:
+    enum class IoctlCommand : u32_le {
+        IocSetNVMAPfdCommand = 0x40044801,
+    };
+
+    struct IoctlSetNvmapFD {
+        u32_le nvmap_fd;
+    };
+    static_assert(sizeof(IoctlSetNvmapFD) == 4, "IoctlSetNvmapFD is incorrect size");
+
+    u32_le nvmap_fd{};
+
+    u32 SetNVMAPfd(const std::vector<u8>& input, std::vector<u8>& output);
+};
+
+} // namespace Service::Nvidia::Devices

src/core/hle/service/nvdrv/devices/nvhost_vic.cpp

@@ -0,0 +1,34 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <cstring>
+
+#include "common/assert.h"
+#include "common/logging/log.h"
+#include "core/hle/service/nvdrv/devices/nvhost_vic.h"
+
+namespace Service::Nvidia::Devices {
+
+u32 nvhost_vic::ioctl(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) {
+    LOG_DEBUG(Service_NVDRV, "called, command=0x{:08X}, input_size=0x{:X}, output_size=0x{:X}",
+              command.raw, input.size(), output.size());
+
+    switch (static_cast<IoctlCommand>(command.raw)) {
+    case IoctlCommand::IocSetNVMAPfdCommand:
+        return SetNVMAPfd(input, output);
+    }
+
+    UNIMPLEMENTED_MSG("Unimplemented ioctl");
+    return 0;
+}
+
+u32 nvhost_vic::SetNVMAPfd(const std::vector<u8>& input, std::vector<u8>& output) {
+    IoctlSetNvmapFD params{};
+    std::memcpy(&params, input.data(), input.size());
+    LOG_DEBUG(Service_NVDRV, "called, fd={}", params.nvmap_fd);
+    nvmap_fd = params.nvmap_fd;
+    return 0;
+}
+
+} // namespace Service::Nvidia::Devices

src/core/hle/service/nvdrv/devices/nvhost_vic.h

@@ -0,0 +1,36 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <vector>
+#include "common/common_types.h"
+#include "common/swap.h"
+#include "core/hle/service/nvdrv/devices/nvdevice.h"
+
+namespace Service::Nvidia::Devices {
+
+class nvhost_vic final : public nvdevice {
+public:
+    nvhost_vic() = default;
+    ~nvhost_vic() override = default;
+
+    u32 ioctl(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override;
+
+private:
+    enum class IoctlCommand : u32_le {
+        IocSetNVMAPfdCommand = 0x40044801,
+    };
+
+    struct IoctlSetNvmapFD {
+        u32_le nvmap_fd;
+    };
+    static_assert(sizeof(IoctlSetNvmapFD) == 4, "IoctlSetNvmapFD is incorrect size");
+
+    u32_le nvmap_fd{};
+
+    u32 SetNVMAPfd(const std::vector<u8>& input, std::vector<u8>& output);
+};
+
+} // namespace Service::Nvidia::Devices

src/core/hle/service/nvdrv/nvdrv.cpp

@@ -12,6 +12,8 @@
 #include "core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.h"
 #include "core/hle/service/nvdrv/devices/nvhost_gpu.h"
 #include "core/hle/service/nvdrv/devices/nvhost_nvdec.h"
+#include "core/hle/service/nvdrv/devices/nvhost_nvjpg.h"
+#include "core/hle/service/nvdrv/devices/nvhost_vic.h"
 #include "core/hle/service/nvdrv/devices/nvmap.h"
 #include "core/hle/service/nvdrv/interface.h"
 #include "core/hle/service/nvdrv/nvdrv.h"
@@ -39,6 +41,8 @@ Module::Module() {
     devices["/dev/nvdisp_disp0"] = std::make_shared<Devices::nvdisp_disp0>(nvmap_dev);
     devices["/dev/nvhost-ctrl"] = std::make_shared<Devices::nvhost_ctrl>();
     devices["/dev/nvhost-nvdec"] = std::make_shared<Devices::nvhost_nvdec>();
+    devices["/dev/nvhost-nvjpg"] = std::make_shared<Devices::nvhost_nvjpg>();
+    devices["/dev/nvhost-vic"] = std::make_shared<Devices::nvhost_vic>();
 }
 
 u32 Module::Open(const std::string& device_name) {

src/core/hle/service/service.h

@@ -23,7 +23,7 @@ class HLERequestContext;
 } // namespace Kernel
 
 namespace FileSys {
-struct VfsFilesystem;
+class VfsFilesystem;
 }
 
 namespace Service {

src/core/hle/service/sm/controller.cpp

@@ -10,7 +10,7 @@
 namespace Service::SM {
 
 void Controller::ConvertSessionToDomain(Kernel::HLERequestContext& ctx) {
-    ASSERT_MSG(!ctx.Session()->IsDomain(), "session is alread a domain");
+    ASSERT_MSG(ctx.Session()->IsSession(), "Session is already a domain");
     ctx.Session()->ConvertToDomain();
 
     IPC::ResponseBuilder rb{ctx, 3};
@@ -41,7 +41,7 @@ void Controller::DuplicateSessionEx(Kernel::HLERequestContext& ctx) {
 void Controller::QueryPointerBufferSize(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 3};
     rb.Push(RESULT_SUCCESS);
-    rb.Push<u32>(0x500);
+    rb.Push<u16>(0x500);
 
     LOG_WARNING(Service, "(STUBBED) called");
 }

src/core/loader/deconstructed_rom_directory.cpp

@@ -118,7 +118,6 @@ ResultStatus AppLoader_DeconstructedRomDirectory::Load(
 
     process->program_id = metadata.GetTitleID();
     process->svc_access_mask.set();
-    process->address_mappings = default_address_mappings;
     process->resource_limit =
         Kernel::ResourceLimit::GetForCategory(Kernel::ResourceLimitCategory::APPLICATION);
     process->Run(Memory::PROCESS_IMAGE_VADDR, metadata.GetMainThreadPriority(),

src/core/loader/elf.cpp

@@ -398,7 +398,6 @@ ResultStatus AppLoader_ELF::Load(Kernel::SharedPtr<Kernel::Process>& process) {
 
     process->LoadModule(codeset, codeset->entrypoint);
     process->svc_access_mask.set();
-    process->address_mappings = default_address_mappings;
 
     // Attach the default resource limit (APPLICATION) to the process
     process->resource_limit =

src/core/loader/loader.cpp

@@ -3,6 +3,7 @@
 // Refer to the license.txt file included.
 
 #include <memory>
+#include <ostream>
 #include <string>
 #include "common/logging/log.h"
 #include "common/string_util.h"
@@ -17,12 +18,6 @@
 
 namespace Loader {
 
-const std::initializer_list<Kernel::AddressMapping> default_address_mappings = {
-    {0x1FF50000, 0x8000, true},    // part of DSP RAM
-    {0x1FF70000, 0x8000, true},    // part of DSP RAM
-    {0x1F000000, 0x600000, false}, // entire VRAM
-};
-
 FileType IdentifyFile(FileSys::VirtualFile file) {
     FileType type;
 
@@ -46,6 +41,8 @@ FileType IdentifyFile(FileSys::VirtualFile file) {
 FileType GuessFromFilename(const std::string& name) {
     if (name == "main")
         return FileType::DeconstructedRomDirectory;
+    if (name == "00")
+        return FileType::NCA;
 
     const std::string extension =
         Common::ToLower(std::string(FileUtil::GetExtensionFromFilename(name)));
@@ -125,14 +122,9 @@ constexpr std::array<const char*, 36> RESULT_MESSAGES{
     "There is no control data available.",
 };
 
-std::string GetMessageForResultStatus(ResultStatus status) {
-    return GetMessageForResultStatus(static_cast<size_t>(status));
-}
-
-std::string GetMessageForResultStatus(u16 status) {
-    if (status >= 36)
-        return "";
-    return RESULT_MESSAGES[status];
+std::ostream& operator<<(std::ostream& os, ResultStatus status) {
+    os << RESULT_MESSAGES.at(static_cast<size_t>(status));
+    return os;
 }
 
 /**

src/core/loader/loader.h

@@ -5,7 +5,7 @@
 #pragma once
 
 #include <algorithm>
-#include <initializer_list>
+#include <iosfwd>
 #include <memory>
 #include <string>
 #include <utility>
@@ -56,7 +56,7 @@ FileType GuessFromFilename(const std::string& name);
 std::string GetFileTypeString(FileType type);
 
 /// Return type for functions in Loader namespace
-enum class ResultStatus {
+enum class ResultStatus : u16 {
     Success,
     ErrorAlreadyLoaded,
     ErrorNotImplemented,
@@ -95,8 +95,7 @@ enum class ResultStatus {
     ErrorNoControl,
 };
 
-std::string GetMessageForResultStatus(ResultStatus status);
-std::string GetMessageForResultStatus(u16 status);
+std::ostream& operator<<(std::ostream& os, ResultStatus status);
 
 /// Interface for loading an application
 class AppLoader : NonCopyable {
@@ -207,12 +206,6 @@ protected:
     bool is_loaded = false;
 };
 
-/**
- * Common address mappings found in most games, used for binary formats that don't have this
- * information.
- */
-extern const std::initializer_list<Kernel::AddressMapping> default_address_mappings;
-
 /**
  * Identifies a bootable file and return a suitable loader
  * @param file The bootable file

src/core/loader/nca.cpp

@@ -3,28 +3,22 @@
 // Refer to the license.txt file included.
 
 #include <utility>
-#include <vector>
 
 #include "common/file_util.h"
 #include "common/logging/log.h"
-#include "common/string_util.h"
-#include "common/swap.h"
-#include "core/core.h"
 #include "core/file_sys/content_archive.h"
-#include "core/file_sys/program_metadata.h"
-#include "core/gdbstub/gdbstub.h"
 #include "core/hle/kernel/process.h"
-#include "core/hle/kernel/resource_limit.h"
 #include "core/hle/service/filesystem/filesystem.h"
+#include "core/loader/deconstructed_rom_directory.h"
 #include "core/loader/nca.h"
-#include "core/loader/nso.h"
-#include "core/memory.h"
 
 namespace Loader {
 
 AppLoader_NCA::AppLoader_NCA(FileSys::VirtualFile file_)
     : AppLoader(std::move(file_)), nca(std::make_unique<FileSys::NCA>(file)) {}
 
+AppLoader_NCA::~AppLoader_NCA() = default;
+
 FileType AppLoader_NCA::IdentifyType(const FileSys::VirtualFile& file) {
     FileSys::NCA nca(file);
 
@@ -83,6 +77,4 @@ ResultStatus AppLoader_NCA::ReadProgramId(u64& out_program_id) {
     return ResultStatus::Success;
 }
 
-AppLoader_NCA::~AppLoader_NCA() = default;
-
 } // namespace Loader

src/core/loader/nca.h

@@ -4,20 +4,24 @@
 
 #pragma once
 
-#include <string>
 #include "common/common_types.h"
-#include "core/file_sys/content_archive.h"
-#include "core/file_sys/program_metadata.h"
+#include "core/file_sys/vfs.h"
 #include "core/hle/kernel/object.h"
 #include "core/loader/loader.h"
-#include "deconstructed_rom_directory.h"
+
+namespace FileSys {
+class NCA;
+}
 
 namespace Loader {
 
+class AppLoader_DeconstructedRomDirectory;
+
 /// Loads an NCA file
 class AppLoader_NCA final : public AppLoader {
 public:
     explicit AppLoader_NCA(FileSys::VirtualFile file);
+    ~AppLoader_NCA() override;
 
     /**
      * Returns the type of the file
@@ -35,12 +39,7 @@ public:
     ResultStatus ReadRomFS(FileSys::VirtualFile& dir) override;
     ResultStatus ReadProgramId(u64& out_program_id) override;
 
-    ~AppLoader_NCA();
-
 private:
-    FileSys::ProgramMetadata metadata;
-
-    FileSys::NCAHeader header;
     std::unique_ptr<FileSys::NCA> nca;
     std::unique_ptr<AppLoader_DeconstructedRomDirectory> directory_loader;
 };

src/core/loader/nro.cpp

@@ -186,7 +186,6 @@ ResultStatus AppLoader_NRO::Load(Kernel::SharedPtr<Kernel::Process>& process) {
     }
 
     process->svc_access_mask.set();
-    process->address_mappings = default_address_mappings;
     process->resource_limit =
         Kernel::ResourceLimit::GetForCategory(Kernel::ResourceLimitCategory::APPLICATION);
     process->Run(base_addr, THREADPRIO_DEFAULT, Memory::DEFAULT_STACK_SIZE);

src/core/loader/nso.cpp

@@ -152,7 +152,6 @@ ResultStatus AppLoader_NSO::Load(Kernel::SharedPtr<Kernel::Process>& process) {
     LOG_DEBUG(Loader, "loaded module {} @ 0x{:X}", file->GetName(), Memory::PROCESS_IMAGE_VADDR);
 
     process->svc_access_mask.set();
-    process->address_mappings = default_address_mappings;
     process->resource_limit =
         Kernel::ResourceLimit::GetForCategory(Kernel::ResourceLimitCategory::APPLICATION);
     process->Run(Memory::PROCESS_IMAGE_VADDR, THREADPRIO_DEFAULT, Memory::DEFAULT_STACK_SIZE);

src/core/loader/xci.cpp

@@ -4,22 +4,14 @@
 
 #include <vector>
 
-#include "common/file_util.h"
-#include "common/logging/log.h"
-#include "common/string_util.h"
-#include "common/swap.h"
-#include "core/core.h"
+#include "common/common_types.h"
+#include "core/file_sys/card_image.h"
 #include "core/file_sys/content_archive.h"
 #include "core/file_sys/control_metadata.h"
-#include "core/file_sys/program_metadata.h"
 #include "core/file_sys/romfs.h"
-#include "core/gdbstub/gdbstub.h"
 #include "core/hle/kernel/process.h"
-#include "core/hle/kernel/resource_limit.h"
-#include "core/hle/service/filesystem/filesystem.h"
-#include "core/loader/nso.h"
+#include "core/loader/nca.h"
 #include "core/loader/xci.h"
-#include "core/memory.h"
 
 namespace Loader {
 

src/core/loader/xci.h

@@ -6,12 +6,18 @@
 
 #include <memory>
 #include "common/common_types.h"
-#include "core/file_sys/card_image.h"
+#include "core/file_sys/vfs.h"
 #include "core/loader/loader.h"
-#include "core/loader/nca.h"
+
+namespace FileSys {
+class NACP;
+class XCI;
+} // namespace FileSys
 
 namespace Loader {
 
+class AppLoader_NCA;
+
 /// Loads an XCI file
 class AppLoader_XCI final : public AppLoader {
 public:
@@ -37,8 +43,6 @@ public:
     ResultStatus ReadTitle(std::string& title) override;
 
 private:
-    FileSys::ProgramMetadata metadata;
-
     std::unique_ptr<FileSys::XCI> xci;
     std::unique_ptr<AppLoader_NCA> nca_loader;
 

src/video_core/engines/maxwell_3d.h

@@ -357,6 +357,27 @@ public:
                 OneMinusConstantColor = 0x62,
                 ConstantAlpha = 0x63,
                 OneMinusConstantAlpha = 0x64,
+
+                // These values are used by Nouveau and some games.
+                ZeroGL = 0x4000,
+                OneGL = 0x4001,
+                SourceColorGL = 0x4300,
+                OneMinusSourceColorGL = 0x4301,
+                SourceAlphaGL = 0x4302,
+                OneMinusSourceAlphaGL = 0x4303,
+                DestAlphaGL = 0x4304,
+                OneMinusDestAlphaGL = 0x4305,
+                DestColorGL = 0x4306,
+                OneMinusDestColorGL = 0x4307,
+                SourceAlphaSaturateGL = 0x4308,
+                ConstantColorGL = 0xc001,
+                OneMinusConstantColorGL = 0xc002,
+                ConstantAlphaGL = 0xc003,
+                OneMinusConstantAlphaGL = 0xc004,
+                Source1ColorGL = 0xc900,
+                OneMinusSource1ColorGL = 0xc901,
+                Source1AlphaGL = 0xc902,
+                OneMinusSource1AlphaGL = 0xc903,
             };
 
             u32 separate_alpha;

src/video_core/engines/shader_bytecode.h

@@ -200,6 +200,14 @@ enum class IMinMaxExchange : u64 {
     XHi = 3,
 };
 
+enum class XmadMode : u64 {
+    None = 0,
+    CLo = 1,
+    CHi = 2,
+    CSfu = 3,
+    CBcc = 4,
+};
+
 enum class FlowCondition : u64 {
     Always = 0xF,
     Fcsm_Tr = 0x1C, // TODO(bunnei): What is this used for?
@@ -456,6 +464,18 @@ union Instruction {
         }
     } bra;
 
+    union {
+        BitField<20, 16, u64> imm20_16;
+        BitField<36, 1, u64> product_shift_left;
+        BitField<37, 1, u64> merge_37;
+        BitField<48, 1, u64> sign_a;
+        BitField<49, 1, u64> sign_b;
+        BitField<50, 3, XmadMode> mode;
+        BitField<52, 1, u64> high_b;
+        BitField<53, 1, u64> high_a;
+        BitField<56, 1, u64> merge_56;
+    } xmad;
+
     union {
         BitField<20, 14, u64> offset;
         BitField<34, 5, u64> index;
@@ -593,6 +613,7 @@ public:
         IntegerSetPredicate,
         PredicateSetPredicate,
         Conversion,
+        Xmad,
         Unknown,
     };
 
@@ -782,10 +803,10 @@ private:
             INST("010010110101----", Id::ISET_C, Type::IntegerSet, "ISET_C"),
             INST("0011011-0101----", Id::ISET_IMM, Type::IntegerSet, "ISET_IMM"),
             INST("0101000010010---", Id::PSETP, Type::PredicateSetPredicate, "PSETP"),
-            INST("0011011-00------", Id::XMAD_IMM, Type::Arithmetic, "XMAD_IMM"),
-            INST("0100111---------", Id::XMAD_CR, Type::Arithmetic, "XMAD_CR"),
-            INST("010100010-------", Id::XMAD_RC, Type::Arithmetic, "XMAD_RC"),
-            INST("0101101100------", Id::XMAD_RR, Type::Arithmetic, "XMAD_RR"),
+            INST("0011011-00------", Id::XMAD_IMM, Type::Xmad, "XMAD_IMM"),
+            INST("0100111---------", Id::XMAD_CR, Type::Xmad, "XMAD_CR"),
+            INST("010100010-------", Id::XMAD_RC, Type::Xmad, "XMAD_RC"),
+            INST("0101101100------", Id::XMAD_RR, Type::Xmad, "XMAD_RR"),
         };
 #undef INST
         std::stable_sort(table.begin(), table.end(), [](const auto& a, const auto& b) {

src/video_core/gpu.cpp

@@ -46,8 +46,11 @@ u32 RenderTargetBytesPerPixel(RenderTargetFormat format) {
     case RenderTargetFormat::RGBA32_FLOAT:
     case RenderTargetFormat::RGBA32_UINT:
         return 16;
+    case RenderTargetFormat::RGBA16_UINT:
+    case RenderTargetFormat::RGBA16_UNORM:
     case RenderTargetFormat::RGBA16_FLOAT:
     case RenderTargetFormat::RG32_FLOAT:
+    case RenderTargetFormat::RG32_UINT:
         return 8;
     case RenderTargetFormat::RGBA8_UNORM:
     case RenderTargetFormat::RGBA8_SNORM:
@@ -61,15 +64,18 @@ u32 RenderTargetBytesPerPixel(RenderTargetFormat format) {
     case RenderTargetFormat::RG16_FLOAT:
     case RenderTargetFormat::R32_FLOAT:
     case RenderTargetFormat::R11G11B10_FLOAT:
+    case RenderTargetFormat::R32_UINT:
         return 4;
     case RenderTargetFormat::R16_UNORM:
     case RenderTargetFormat::R16_SNORM:
     case RenderTargetFormat::R16_UINT:
     case RenderTargetFormat::R16_SINT:
     case RenderTargetFormat::R16_FLOAT:
+    case RenderTargetFormat::RG8_UNORM:
     case RenderTargetFormat::RG8_SNORM:
         return 2;
     case RenderTargetFormat::R8_UNORM:
+    case RenderTargetFormat::R8_UINT:
         return 1;
     default:
         UNIMPLEMENTED_MSG("Unimplemented render target format {}", static_cast<u32>(format));

src/video_core/gpu.h

@@ -20,8 +20,11 @@ enum class RenderTargetFormat : u32 {
     NONE = 0x0,
     RGBA32_FLOAT = 0xC0,
     RGBA32_UINT = 0xC2,
+    RGBA16_UNORM = 0xC6,
+    RGBA16_UINT = 0xC9,
     RGBA16_FLOAT = 0xCA,
     RG32_FLOAT = 0xCB,
+    RG32_UINT = 0xCD,
     BGRA8_UNORM = 0xCF,
     RGB10_A2_UNORM = 0xD1,
     RGBA8_UNORM = 0xD5,
@@ -33,8 +36,10 @@ enum class RenderTargetFormat : u32 {
     RG16_UINT = 0xDD,
     RG16_FLOAT = 0xDE,
     R11G11B10_FLOAT = 0xE0,
+    R32_UINT = 0xE4,
     R32_FLOAT = 0xE5,
     B5G6R5_UNORM = 0xE8,
+    RG8_UNORM = 0xEA,
     RG8_SNORM = 0xEB,
     R16_UNORM = 0xEE,
     R16_SNORM = 0xEF,
@@ -42,6 +47,7 @@ enum class RenderTargetFormat : u32 {
     R16_UINT = 0xF1,
     R16_FLOAT = 0xF2,
     R8_UNORM = 0xF3,
+    R8_UINT = 0xF6,
 };
 
 enum class DepthFormat : u32 {

src/video_core/renderer_base.cpp

@@ -10,7 +10,7 @@
 
 namespace VideoCore {
 
-RendererBase::RendererBase(EmuWindow& window) : render_window{window} {
+RendererBase::RendererBase(Core::Frontend::EmuWindow& window) : render_window{window} {
     RefreshBaseSettings();
 }
 

src/video_core/renderer_base.h

@@ -11,7 +11,9 @@
 #include "video_core/gpu.h"
 #include "video_core/rasterizer_interface.h"
 
+namespace Core::Frontend {
 class EmuWindow;
+}
 
 namespace VideoCore {
 
@@ -21,7 +23,7 @@ struct RendererSettings {
 
 class RendererBase : NonCopyable {
 public:
-    explicit RendererBase(EmuWindow& window);
+    explicit RendererBase(Core::Frontend::EmuWindow& window);
     virtual ~RendererBase();
 
     /// Swap buffers (render frame)
@@ -59,7 +61,7 @@ protected:
     /// Refreshes settings specific to the rasterizer.
     void RefreshRasterizerSetting();
 
-    EmuWindow& render_window; ///< Reference to the render window handle.
+    Core::Frontend::EmuWindow& render_window; ///< Reference to the render window handle.
     std::unique_ptr<RasterizerInterface> rasterizer;
     f32 m_current_fps = 0.0f; ///< Current framerate, should be set by the renderer
     int m_current_frame = 0;  ///< Current frame, should be set by the renderer

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -36,30 +36,21 @@ MICROPROFILE_DEFINE(OpenGL_Drawing, "OpenGL", "Drawing", MP_RGB(128, 128, 192));
 MICROPROFILE_DEFINE(OpenGL_Blits, "OpenGL", "Blits", MP_RGB(100, 100, 255));
 MICROPROFILE_DEFINE(OpenGL_CacheManagement, "OpenGL", "Cache Mgmt", MP_RGB(100, 255, 100));
 
-RasterizerOpenGL::RasterizerOpenGL(EmuWindow& window) : emu_window{window} {
+RasterizerOpenGL::RasterizerOpenGL(Core::Frontend::EmuWindow& window)
+    : emu_window{window}, stream_buffer(GL_ARRAY_BUFFER, STREAM_BUFFER_SIZE) {
     // Create sampler objects
     for (size_t i = 0; i < texture_samplers.size(); ++i) {
         texture_samplers[i].Create();
         state.texture_units[i].sampler = texture_samplers[i].sampler.handle;
     }
 
-    // Create SSBOs
-    for (size_t stage = 0; stage < ssbos.size(); ++stage) {
-        for (size_t buffer = 0; buffer < ssbos[stage].size(); ++buffer) {
-            ssbos[stage][buffer].Create();
-            state.draw.const_buffers[stage][buffer].ssbo = ssbos[stage][buffer].handle;
-        }
-    }
-
     GLint ext_num;
     glGetIntegerv(GL_NUM_EXTENSIONS, &ext_num);
     for (GLint i = 0; i < ext_num; i++) {
         const std::string_view extension{
             reinterpret_cast<const char*>(glGetStringi(GL_EXTENSIONS, i))};
 
-        if (extension == "GL_ARB_buffer_storage") {
-            has_ARB_buffer_storage = true;
-        } else if (extension == "GL_ARB_direct_state_access") {
+        if (extension == "GL_ARB_direct_state_access") {
             has_ARB_direct_state_access = true;
         } else if (extension == "GL_ARB_separate_shader_objects") {
             has_ARB_separate_shader_objects = true;
@@ -86,47 +77,31 @@ RasterizerOpenGL::RasterizerOpenGL(EmuWindow& window) : emu_window{window} {
 
     hw_vao.Create();
 
-    stream_buffer = OGLStreamBuffer::MakeBuffer(has_ARB_buffer_storage, GL_ARRAY_BUFFER);
-    stream_buffer->Create(STREAM_BUFFER_SIZE, STREAM_BUFFER_SIZE / 2);
-    state.draw.vertex_buffer = stream_buffer->GetHandle();
+    state.draw.vertex_buffer = stream_buffer.GetHandle();
 
     shader_program_manager = std::make_unique<GLShader::ProgramManager>();
     state.draw.shader_program = 0;
     state.draw.vertex_array = hw_vao.handle;
     state.Apply();
 
-    glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, stream_buffer->GetHandle());
-
-    for (unsigned index = 0; index < uniform_buffers.size(); ++index) {
-        auto& buffer = uniform_buffers[index];
-        buffer.Create();
-        glBindBuffer(GL_UNIFORM_BUFFER, buffer.handle);
-        glBufferData(GL_UNIFORM_BUFFER, sizeof(GLShader::MaxwellUniformData), nullptr,
-                     GL_STREAM_COPY);
-        glBindBufferBase(GL_UNIFORM_BUFFER, index, buffer.handle);
-    }
+    glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, stream_buffer.GetHandle());
 
     glEnable(GL_BLEND);
 
+    glGetIntegerv(GL_UNIFORM_BUFFER_OFFSET_ALIGNMENT, &uniform_buffer_alignment);
+
     LOG_CRITICAL(Render_OpenGL, "Sync fixed function OpenGL state here!");
 }
 
-RasterizerOpenGL::~RasterizerOpenGL() {
-    if (stream_buffer != nullptr) {
-        state.draw.vertex_buffer = stream_buffer->GetHandle();
-        state.Apply();
-        stream_buffer->Release();
-    }
-}
+RasterizerOpenGL::~RasterizerOpenGL() {}
 
 std::pair<u8*, GLintptr> RasterizerOpenGL::SetupVertexArrays(u8* array_ptr,
                                                              GLintptr buffer_offset) {
     MICROPROFILE_SCOPE(OpenGL_VAO);
     const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
-    const auto& memory_manager = Core::System::GetInstance().GPU().memory_manager;
 
     state.draw.vertex_array = hw_vao.handle;
-    state.draw.vertex_buffer = stream_buffer->GetHandle();
+    state.draw.vertex_buffer = stream_buffer.GetHandle();
     state.Apply();
 
     // Upload all guest vertex arrays sequentially to our buffer
@@ -141,16 +116,15 @@ std::pair<u8*, GLintptr> RasterizerOpenGL::SetupVertexArrays(u8* array_ptr,
         ASSERT(end > start);
         u64 size = end - start + 1;
 
-        // Copy vertex array data
-        Memory::ReadBlock(*memory_manager->GpuToCpuAddress(start), array_ptr, size);
+        GLintptr vertex_buffer_offset;
+        std::tie(array_ptr, buffer_offset, vertex_buffer_offset) =
+            UploadMemory(array_ptr, buffer_offset, start, size);
 
         // Bind the vertex array to the buffer at the current offset.
-        glBindVertexBuffer(index, stream_buffer->GetHandle(), buffer_offset, vertex_array.stride);
+        glBindVertexBuffer(index, stream_buffer.GetHandle(), vertex_buffer_offset,
+                           vertex_array.stride);
 
         ASSERT_MSG(vertex_array.divisor == 0, "Vertex buffer divisor unimplemented");
-
-        array_ptr += size;
-        buffer_offset += size;
     }
 
     // Use the vertex array as-is, assumes that the data is formatted correctly for OpenGL.
@@ -201,22 +175,12 @@ static GLShader::ProgramCode GetShaderProgramCode(Maxwell::ShaderProgram program
     return program_code;
 }
 
-void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset) {
-    // Helper function for uploading uniform data
-    const auto copy_buffer = [&](GLuint handle, GLintptr offset, GLsizeiptr size) {
-        if (has_ARB_direct_state_access) {
-            glCopyNamedBufferSubData(stream_buffer->GetHandle(), handle, offset, 0, size);
-        } else {
-            glBindBuffer(GL_COPY_WRITE_BUFFER, handle);
-            glCopyBufferSubData(GL_ARRAY_BUFFER, GL_COPY_WRITE_BUFFER, offset, 0, size);
-        }
-    };
-
+std::pair<u8*, GLintptr> RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset) {
     auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
 
     // Next available bindpoints to use when uploading the const buffers and textures to the GLSL
     // shaders. The constbuffer bindpoint starts after the shader stage configuration bind points.
-    u32 current_constbuffer_bindpoint = uniform_buffers.size();
+    u32 current_constbuffer_bindpoint = Tegra::Engines::Maxwell3D::Regs::MaxShaderStage;
     u32 current_texture_bindpoint = 0;
 
     for (size_t index = 0; index < Maxwell::MaxShaderProgram; ++index) {
@@ -228,22 +192,21 @@ void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset) {
             continue;
         }
 
+        std::tie(buffer_ptr, buffer_offset) =
+            AlignBuffer(buffer_ptr, buffer_offset, static_cast<size_t>(uniform_buffer_alignment));
+
         const size_t stage{index == 0 ? 0 : index - 1}; // Stage indices are 0 - 5
 
         GLShader::MaxwellUniformData ubo{};
         ubo.SetFromRegs(gpu.state.shader_stages[stage]);
         std::memcpy(buffer_ptr, &ubo, sizeof(ubo));
 
-        // Flush the buffer so that the GPU can see the data we just wrote.
-        glFlushMappedBufferRange(GL_ARRAY_BUFFER, buffer_offset, sizeof(ubo));
+        // Bind the buffer
+        glBindBufferRange(GL_UNIFORM_BUFFER, stage, stream_buffer.GetHandle(), buffer_offset,
+                          sizeof(ubo));
 
-        // Upload uniform data as one UBO per stage
-        const GLintptr ubo_offset = buffer_offset;
-        copy_buffer(uniform_buffers[stage].handle, ubo_offset,
-                    sizeof(GLShader::MaxwellUniformData));
-
-        buffer_ptr += sizeof(GLShader::MaxwellUniformData);
-        buffer_offset += sizeof(GLShader::MaxwellUniformData);
+        buffer_ptr += sizeof(ubo);
+        buffer_offset += sizeof(ubo);
 
         GLShader::ShaderSetup setup{GetShaderProgramCode(program)};
         GLShader::ShaderEntries shader_resources;
@@ -282,9 +245,9 @@ void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset) {
             static_cast<Maxwell::ShaderStage>(stage));
 
         // Configure the const buffers for this shader stage.
-        current_constbuffer_bindpoint =
-            SetupConstBuffers(static_cast<Maxwell::ShaderStage>(stage), gl_stage_program,
-                              current_constbuffer_bindpoint, shader_resources.const_buffer_entries);
+        std::tie(buffer_ptr, buffer_offset, current_constbuffer_bindpoint) = SetupConstBuffers(
+            buffer_ptr, buffer_offset, static_cast<Maxwell::ShaderStage>(stage), gl_stage_program,
+            current_constbuffer_bindpoint, shader_resources.const_buffer_entries);
 
         // Configure the textures for this shader stage.
         current_texture_bindpoint =
@@ -299,6 +262,8 @@ void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset) {
     }
 
     shader_program_manager->UseTrivialGeometryShader();
+
+    return {buffer_ptr, buffer_offset};
 }
 
 size_t RasterizerOpenGL::CalculateVertexArraysSize() const {
@@ -432,6 +397,31 @@ void RasterizerOpenGL::Clear() {
     }
 }
 
+std::pair<u8*, GLintptr> RasterizerOpenGL::AlignBuffer(u8* buffer_ptr, GLintptr buffer_offset,
+                                                       size_t alignment) {
+    // Align the offset, not the mapped pointer
+    GLintptr offset_aligned =
+        static_cast<GLintptr>(Common::AlignUp(static_cast<size_t>(buffer_offset), alignment));
+    return {buffer_ptr + (offset_aligned - buffer_offset), offset_aligned};
+}
+
+std::tuple<u8*, GLintptr, GLintptr> RasterizerOpenGL::UploadMemory(u8* buffer_ptr,
+                                                                   GLintptr buffer_offset,
+                                                                   Tegra::GPUVAddr gpu_addr,
+                                                                   size_t size, size_t alignment) {
+    std::tie(buffer_ptr, buffer_offset) = AlignBuffer(buffer_ptr, buffer_offset, alignment);
+    GLintptr uploaded_offset = buffer_offset;
+
+    const auto& memory_manager = Core::System::GetInstance().GPU().memory_manager;
+    const boost::optional<VAddr> cpu_addr{memory_manager->GpuToCpuAddress(gpu_addr)};
+    Memory::ReadBlock(*cpu_addr, buffer_ptr, size);
+
+    buffer_ptr += size;
+    buffer_offset += size;
+
+    return {buffer_ptr, buffer_offset, uploaded_offset};
+}
+
 void RasterizerOpenGL::DrawArrays() {
     if (accelerate_draw == AccelDraw::Disabled)
         return;
@@ -456,7 +446,7 @@ void RasterizerOpenGL::DrawArrays() {
     const u64 index_buffer_size{regs.index_array.count * regs.index_array.FormatSizeInBytes()};
     const unsigned vertex_num{is_indexed ? regs.index_array.count : regs.vertex_buffer.count};
 
-    state.draw.vertex_buffer = stream_buffer->GetHandle();
+    state.draw.vertex_buffer = stream_buffer.GetHandle();
     state.Apply();
 
     size_t buffer_size = CalculateVertexArraysSize();
@@ -466,41 +456,31 @@ void RasterizerOpenGL::DrawArrays() {
     }
 
     // Uniform space for the 5 shader stages
-    buffer_size = Common::AlignUp<size_t>(buffer_size, 4) +
-                  sizeof(GLShader::MaxwellUniformData) * Maxwell::MaxShaderStage;
+    buffer_size =
+        Common::AlignUp<size_t>(buffer_size, 4) +
+        (sizeof(GLShader::MaxwellUniformData) + uniform_buffer_alignment) * Maxwell::MaxShaderStage;
+
+    // Add space for at least 18 constant buffers
+    buffer_size += Maxwell::MaxConstBuffers * (MaxConstbufferSize + uniform_buffer_alignment);
 
     u8* buffer_ptr;
     GLintptr buffer_offset;
-    std::tie(buffer_ptr, buffer_offset) =
-        stream_buffer->Map(static_cast<GLsizeiptr>(buffer_size), 4);
+    std::tie(buffer_ptr, buffer_offset, std::ignore) =
+        stream_buffer.Map(static_cast<GLsizeiptr>(buffer_size), 4);
+    u8* buffer_ptr_base = buffer_ptr;
 
-    u8* offseted_buffer;
-    std::tie(offseted_buffer, buffer_offset) = SetupVertexArrays(buffer_ptr, buffer_offset);
-
-    offseted_buffer =
-        reinterpret_cast<u8*>(Common::AlignUp(reinterpret_cast<size_t>(offseted_buffer), 4));
-    buffer_offset = Common::AlignUp<size_t>(buffer_offset, 4);
+    std::tie(buffer_ptr, buffer_offset) = SetupVertexArrays(buffer_ptr, buffer_offset);
 
     // If indexed mode, copy the index buffer
     GLintptr index_buffer_offset = 0;
     if (is_indexed) {
-        const auto& memory_manager = Core::System::GetInstance().GPU().memory_manager;
-        const boost::optional<VAddr> index_data_addr{
-            memory_manager->GpuToCpuAddress(regs.index_array.StartAddress())};
-        Memory::ReadBlock(*index_data_addr, offseted_buffer, index_buffer_size);
-
-        index_buffer_offset = buffer_offset;
-        offseted_buffer += index_buffer_size;
-        buffer_offset += index_buffer_size;
+        std::tie(buffer_ptr, buffer_offset, index_buffer_offset) = UploadMemory(
+            buffer_ptr, buffer_offset, regs.index_array.StartAddress(), index_buffer_size);
     }
 
-    offseted_buffer =
-        reinterpret_cast<u8*>(Common::AlignUp(reinterpret_cast<size_t>(offseted_buffer), 4));
-    buffer_offset = Common::AlignUp<size_t>(buffer_offset, 4);
+    std::tie(buffer_ptr, buffer_offset) = SetupShaders(buffer_ptr, buffer_offset);
 
-    SetupShaders(offseted_buffer, buffer_offset);
-
-    stream_buffer->Unmap();
+    stream_buffer.Unmap(buffer_ptr - buffer_ptr_base);
 
     shader_program_manager->ApplyTo(state);
     state.Apply();
@@ -647,45 +627,32 @@ void RasterizerOpenGL::SamplerInfo::SyncWithConfig(const Tegra::Texture::TSCEntr
     }
 }
 
-u32 RasterizerOpenGL::SetupConstBuffers(Maxwell::ShaderStage stage, GLuint program,
-                                        u32 current_bindpoint,
-                                        const std::vector<GLShader::ConstBufferEntry>& entries) {
+std::tuple<u8*, GLintptr, u32> RasterizerOpenGL::SetupConstBuffers(
+    u8* buffer_ptr, GLintptr buffer_offset, Maxwell::ShaderStage stage, GLuint program,
+    u32 current_bindpoint, const std::vector<GLShader::ConstBufferEntry>& entries) {
     const auto& gpu = Core::System::GetInstance().GPU();
     const auto& maxwell3d = gpu.Maxwell3D();
 
-    // Reset all buffer draw state for this stage.
-    for (auto& buffer : state.draw.const_buffers[static_cast<size_t>(stage)]) {
-        buffer.bindpoint = 0;
-        buffer.enabled = false;
-    }
-
     // Upload only the enabled buffers from the 16 constbuffers of each shader stage
     const auto& shader_stage = maxwell3d.state.shader_stages[static_cast<size_t>(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()];
-        auto& buffer_draw_state =
-            state.draw.const_buffers[static_cast<size_t>(stage)][used_buffer.GetIndex()];
 
         if (!buffer.enabled) {
             continue;
         }
 
-        buffer_draw_state.enabled = true;
-        buffer_draw_state.bindpoint = current_bindpoint + bindpoint;
-
-        boost::optional<VAddr> addr = gpu.memory_manager->GpuToCpuAddress(buffer.address);
-
         size_t size = 0;
 
         if (used_buffer.IsIndirect()) {
             // Buffer is accessed indirectly, so upload the entire thing
-            size = buffer.size * sizeof(float);
+            size = buffer.size;
 
             if (size > MaxConstbufferSize) {
-                LOG_ERROR(HW_GPU, "indirect constbuffer size {} exceeds maximum {}", size,
-                          MaxConstbufferSize);
+                LOG_CRITICAL(HW_GPU, "indirect constbuffer size {} exceeds maximum {}", size,
+                             MaxConstbufferSize);
                 size = MaxConstbufferSize;
             }
         } else {
@@ -698,25 +665,26 @@ u32 RasterizerOpenGL::SetupConstBuffers(Maxwell::ShaderStage stage, GLuint progr
         size = Common::AlignUp(size, sizeof(GLvec4));
         ASSERT_MSG(size <= MaxConstbufferSize, "Constbuffer too big");
 
-        std::vector<u8> data(size);
-        Memory::ReadBlock(*addr, data.data(), data.size());
+        GLintptr const_buffer_offset;
+        std::tie(buffer_ptr, buffer_offset, const_buffer_offset) =
+            UploadMemory(buffer_ptr, buffer_offset, buffer.address, size,
+                         static_cast<size_t>(uniform_buffer_alignment));
 
-        glBindBuffer(GL_UNIFORM_BUFFER, buffer_draw_state.ssbo);
-        glBufferData(GL_UNIFORM_BUFFER, data.size(), data.data(), GL_DYNAMIC_DRAW);
-        glBindBuffer(GL_UNIFORM_BUFFER, 0);
+        glBindBufferRange(GL_UNIFORM_BUFFER, current_bindpoint + bindpoint,
+                          stream_buffer.GetHandle(), const_buffer_offset, size);
 
         // Now configure the bindpoint of the buffer inside the shader
         const std::string buffer_name = used_buffer.GetName();
         const GLuint index =
             glGetProgramResourceIndex(program, GL_UNIFORM_BLOCK, buffer_name.c_str());
         if (index != GL_INVALID_INDEX) {
-            glUniformBlockBinding(program, index, buffer_draw_state.bindpoint);
+            glUniformBlockBinding(program, index, current_bindpoint + bindpoint);
         }
     }
 
     state.Apply();
 
-    return current_bindpoint + static_cast<u32>(entries.size());
+    return {buffer_ptr, buffer_offset, current_bindpoint + static_cast<u32>(entries.size())};
 }
 
 u32 RasterizerOpenGL::SetupTextures(Maxwell::ShaderStage stage, GLuint program, u32 current_unit,

src/video_core/renderer_opengl/gl_rasterizer.h

@@ -7,6 +7,7 @@
 #include <array>
 #include <cstddef>
 #include <memory>
+#include <tuple>
 #include <utility>
 #include <vector>
 #include <glad/glad.h>
@@ -21,12 +22,15 @@
 #include "video_core/renderer_opengl/gl_state.h"
 #include "video_core/renderer_opengl/gl_stream_buffer.h"
 
-class EmuWindow;
 struct ScreenInfo;
 
+namespace Core::Frontend {
+class EmuWindow;
+}
+
 class RasterizerOpenGL : public VideoCore::RasterizerInterface {
 public:
-    explicit RasterizerOpenGL(EmuWindow& renderer);
+    explicit RasterizerOpenGL(Core::Frontend::EmuWindow& renderer);
     ~RasterizerOpenGL() override;
 
     void DrawArrays() override;
@@ -97,9 +101,10 @@ private:
      * @param entries Vector describing the buffers that are actually used in the guest shader.
      * @returns The next available bindpoint for use in the next shader stage.
      */
-    u32 SetupConstBuffers(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage, GLuint program,
-                          u32 current_bindpoint,
-                          const std::vector<GLShader::ConstBufferEntry>& entries);
+    std::tuple<u8*, GLintptr, u32> SetupConstBuffers(
+        u8* buffer_ptr, GLintptr buffer_offset, Tegra::Engines::Maxwell3D::Regs::ShaderStage stage,
+        GLuint program, u32 current_bindpoint,
+        const std::vector<GLShader::ConstBufferEntry>& entries);
 
     /*
      * Configures the current textures to use for the draw command.
@@ -136,7 +141,6 @@ private:
     /// Syncs the blend state to match the guest state
     void SyncBlendState();
 
-    bool has_ARB_buffer_storage = false;
     bool has_ARB_direct_state_access = false;
     bool has_ARB_separate_shader_objects = false;
     bool has_ARB_vertex_attrib_binding = false;
@@ -145,29 +149,31 @@ private:
 
     RasterizerCacheOpenGL res_cache;
 
-    EmuWindow& emu_window;
+    Core::Frontend::EmuWindow& emu_window;
 
     std::unique_ptr<GLShader::ProgramManager> shader_program_manager;
     OGLVertexArray sw_vao;
     OGLVertexArray hw_vao;
 
     std::array<SamplerInfo, GLShader::NumTextureSamplers> texture_samplers;
-    std::array<std::array<OGLBuffer, Tegra::Engines::Maxwell3D::Regs::MaxConstBuffers>,
-               Tegra::Engines::Maxwell3D::Regs::MaxShaderStage>
-        ssbos;
 
     static constexpr size_t STREAM_BUFFER_SIZE = 128 * 1024 * 1024;
-    std::unique_ptr<OGLStreamBuffer> stream_buffer;
+    OGLStreamBuffer stream_buffer;
     OGLBuffer uniform_buffer;
     OGLFramebuffer framebuffer;
+    GLint uniform_buffer_alignment;
 
     size_t CalculateVertexArraysSize() const;
 
     std::pair<u8*, GLintptr> SetupVertexArrays(u8* array_ptr, GLintptr buffer_offset);
 
-    std::array<OGLBuffer, Tegra::Engines::Maxwell3D::Regs::MaxShaderStage> uniform_buffers;
+    std::pair<u8*, GLintptr> SetupShaders(u8* buffer_ptr, GLintptr buffer_offset);
 
-    void SetupShaders(u8* buffer_ptr, GLintptr buffer_offset);
+    std::pair<u8*, GLintptr> AlignBuffer(u8* buffer_ptr, GLintptr buffer_offset, size_t alignment);
+
+    std::tuple<u8*, GLintptr, GLintptr> UploadMemory(u8* buffer_ptr, GLintptr buffer_offset,
+                                                     Tegra::GPUVAddr gpu_addr, size_t size,
+                                                     size_t alignment = 4);
 
     enum class AccelDraw { Disabled, Arrays, Indexed };
     AccelDraw accelerate_draw = AccelDraw::Disabled;