RasterizerCache: Swizzle to a staging buffer before copying it to memory when flushing a texture.

This gives us slightly increased memory safety, writing to out of bounds memory will be caught at the point of write instead of several calls down the line as memory corruption

.travis.yml

@@ -29,6 +29,19 @@ matrix:
       script: "./.travis/macos/build.sh"
       after_success: "./.travis/macos/upload.sh"
       cache: ccache
+    - os: linux
+      env: NAME="MinGW build"
+      sudo: required
+      dist: trusty
+      services: docker
+      addons:
+        apt:
+          packages:
+            - p7zip-full
+      install: "./.travis/linux-mingw/deps.sh"
+      script: "./.travis/linux-mingw/build.sh"
+      after_success: "./.travis/linux-mingw/upload.sh"
+      cache: ccache
 
 deploy:
   provider: releases

.travis/common/post-upload.sh

@@ -11,6 +11,9 @@ if [ -z $TRAVIS_TAG ]; then
     RELEASE_NAME=head
 else
     RELEASE_NAME=$(echo $TRAVIS_TAG | cut -d- -f1)
+    if [ "$NAME" = "MinGW build" ]; then
+        RELEASE_NAME="${RELEASE_NAME}-mingw"
+    fi
 fi
 
 mv "$REV_NAME" $RELEASE_NAME

.travis/linux-mingw/build.sh

@@ -0,0 +1,3 @@
+#!/bin/bash -ex
+mkdir "$HOME/.ccache" || true
+docker run --env-file .travis/common/travis-ci.env -v $(pwd):/yuzu -v "$HOME/.ccache":/root/.ccache ubuntu:18.04 /bin/bash -ex /yuzu/.travis/linux-mingw/docker.sh

.travis/linux-mingw/deps.sh

@@ -0,0 +1,3 @@
+#!/bin/sh -ex
+
+docker pull ubuntu:18.04

.travis/linux-mingw/docker.sh

@@ -0,0 +1,59 @@
+#!/bin/bash -ex
+
+cd /yuzu
+MINGW_PACKAGES="sdl2-mingw-w64 qt5base-mingw-w64 qt5tools-mingw-w64 libsamplerate-mingw-w64 qt5multimedia-mingw-w64"
+apt-get update
+apt-get install -y gpg wget git python3-pip python ccache g++-mingw-w64-x86-64 gcc-mingw-w64-x86-64 mingw-w64-tools cmake
+echo 'deb http://ppa.launchpad.net/tobydox/mingw-w64/ubuntu bionic main ' > /etc/apt/sources.list.d/extras.list
+apt-key adv --keyserver keyserver.ubuntu.com --recv '72931B477E22FEFD47F8DECE02FE5F12ADDE29B2'
+apt-get update
+apt-get install -y ${MINGW_PACKAGES}
+
+# fix a problem in current MinGW headers
+wget -q https://raw.githubusercontent.com/Alexpux/mingw-w64/d0d7f784833bbb0b2d279310ddc6afb52fe47a46/mingw-w64-headers/crt/errno.h -O /usr/x86_64-w64-mingw32/include/errno.h
+# override Travis CI unreasonable ccache size
+echo 'max_size = 3.0G' > "$HOME/.ccache/ccache.conf"
+
+# Dirty hack to trick unicorn makefile into believing we are in a MINGW system
+mv /bin/uname /bin/uname1 && echo -e '#!/bin/sh\necho MINGW64' >> /bin/uname
+chmod +x /bin/uname
+
+# Dirty hack to trick unicorn makefile into believing we have cmd
+echo '' >> /bin/cmd
+chmod +x /bin/cmd
+
+mkdir build && cd build
+cmake .. -DCMAKE_TOOLCHAIN_FILE="$(pwd)/../CMakeModules/MinGWCross.cmake" -DUSE_CCACHE=ON -DYUZU_USE_BUNDLED_UNICORN=ON -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON -DCMAKE_BUILD_TYPE=Release
+make -j4
+
+# Clean up the dirty hacks
+rm /bin/uname && mv /bin/uname1 /bin/uname
+rm /bin/cmd
+
+ccache -s
+
+echo "Tests skipped"
+#ctest -VV -C Release
+
+echo 'Prepare binaries...'
+cd ..
+mkdir package
+
+QT_PLATFORM_DLL_PATH='/usr/x86_64-w64-mingw32/lib/qt5/plugins/platforms/'
+find build/ -name "yuzu*.exe" -exec cp {} 'package' \;
+
+# copy Qt plugins
+mkdir package/platforms
+cp "${QT_PLATFORM_DLL_PATH}/qwindows.dll" package/platforms/
+cp -rv "${QT_PLATFORM_DLL_PATH}/../mediaservice/" package/
+cp -rv "${QT_PLATFORM_DLL_PATH}/../imageformats/" package/
+rm -f package/mediaservice/*d.dll
+
+for i in package/*.exe; do
+  # we need to process pdb here, however, cv2pdb
+  # does not work here, so we just simply strip all the debug symbols
+  x86_64-w64-mingw32-strip "${i}"
+done
+
+pip3 install pefile
+python3 .travis/linux-mingw/scan_dll.py package/*.exe "package/"

.travis/linux-mingw/scan_dll.py

@@ -0,0 +1,106 @@
+import pefile
+import sys
+import re
+import os
+import queue
+import shutil
+
+# constant definitions
+KNOWN_SYS_DLLS = ['WINMM.DLL', 'MSVCRT.DLL', 'VERSION.DLL', 'MPR.DLL',
+                  'DWMAPI.DLL', 'UXTHEME.DLL', 'DNSAPI.DLL', 'IPHLPAPI.DLL']
+# below is for Ubuntu 18.04 with specified PPA enabled, if you are using
+# other distro or different repositories, change the following accordingly
+DLL_PATH = [
+    '/usr/x86_64-w64-mingw32/bin/',
+    '/usr/x86_64-w64-mingw32/lib/',
+    '/usr/lib/gcc/x86_64-w64-mingw32/7.3-posix/'
+]
+
+missing = []
+
+
+def parse_imports(file_name):
+    results = []
+    pe = pefile.PE(file_name, fast_load=True)
+    pe.parse_data_directories()
+
+    for entry in pe.DIRECTORY_ENTRY_IMPORT:
+        current = entry.dll.decode()
+        current_u = current.upper()  # b/c Windows is often case insensitive
+        # here we filter out system dlls
+        # dll w/ names like *32.dll are likely to be system dlls
+        if current_u.upper() not in KNOWN_SYS_DLLS and not re.match(string=current_u, pattern=r'.*32\.DLL'):
+            results.append(current)
+
+    return results
+
+
+def parse_imports_recursive(file_name, path_list=[]):
+    q = queue.Queue()  # create a FIFO queue
+    # file_name can be a string or a list for the convience
+    if isinstance(file_name, str):
+        q.put(file_name)
+    elif isinstance(file_name, list):
+        for i in file_name:
+            q.put(i)
+    full_list = []
+    while q.qsize():
+        current = q.get_nowait()
+        print('> %s' % current)
+        deps = parse_imports(current)
+        # if this dll does not have any import, ignore it
+        if not deps:
+            continue
+        for dep in deps:
+            # the dependency already included in the list, skip
+            if dep in full_list:
+                continue
+            # find the requested dll in the provided paths
+            full_path = find_dll(dep)
+            if not full_path:
+                missing.append(dep)
+                continue
+            full_list.append(dep)
+            q.put(full_path)
+            path_list.append(full_path)
+    return full_list
+
+
+def find_dll(name):
+    for path in DLL_PATH:
+        for root, _, files in os.walk(path):
+            for f in files:
+                if name.lower() == f.lower():
+                    return os.path.join(root, f)
+
+
+def deploy(name, dst, dry_run=False):
+    dlls_path = []
+    parse_imports_recursive(name, dlls_path)
+    for dll_entry in dlls_path:
+        if not dry_run:
+            shutil.copy(dll_entry, dst)
+        else:
+            print('[Dry-Run] Copy %s to %s' % (dll_entry, dst))
+    print('Deploy completed.')
+    return dlls_path
+
+
+def main():
+    if len(sys.argv) < 3:
+        print('Usage: %s [files to examine ...] [target deploy directory]')
+        return 1
+    to_deploy = sys.argv[1:-1]
+    tgt_dir = sys.argv[-1]
+    if not os.path.isdir(tgt_dir):
+        print('%s is not a directory.' % tgt_dir)
+        return 1
+    print('Scanning dependencies...')
+    deploy(to_deploy, tgt_dir)
+    if missing:
+        print('Following DLLs are not found: %s' % ('\n'.join(missing)))
+    return 0
+
+
+if __name__ == '__main__':
+    main()

.travis/linux-mingw/upload.sh

@@ -0,0 +1,13 @@
+#!/bin/bash -ex
+
+. .travis/common/pre-upload.sh
+
+REV_NAME="yuzu-windows-mingw-${GITDATE}-${GITREV}"
+ARCHIVE_NAME="${REV_NAME}.tar.gz"
+COMPRESSION_FLAGS="-czvf"
+
+mkdir "$REV_NAME"
+# get around the permission issues
+cp -r package/* "$REV_NAME"
+
+. .travis/common/post-upload.sh

.travis/linux/docker.sh

@@ -6,7 +6,9 @@ apt-get install --no-install-recommends -y build-essential git libqt5opengl5-dev
 cd /yuzu
 
 mkdir build && cd build
-cmake .. -DYUZU_BUILD_UNICORN=ON -DCMAKE_BUILD_TYPE=Release -DCMAKE_C_COMPILER=/usr/lib/ccache/gcc -DCMAKE_CXX_COMPILER=/usr/lib/ccache/g++ -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON -G Ninja
+cmake .. -DYUZU_USE_BUNDLED_UNICORN=ON -DCMAKE_BUILD_TYPE=Release -DCMAKE_C_COMPILER=/usr/lib/ccache/gcc -DCMAKE_CXX_COMPILER=/usr/lib/ccache/g++ -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON -G Ninja
 ninja
 
+ccache -s
+
 ctest -VV -C Release

.travis/macos/build.sh

@@ -9,7 +9,9 @@ export PATH="/usr/local/opt/ccache/libexec:$PATH"
 
 mkdir build && cd build
 cmake --version
-cmake .. -DYUZU_BUILD_UNICORN=ON -DCMAKE_BUILD_TYPE=Release -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON
+cmake .. -DYUZU_USE_BUNDLED_UNICORN=ON -DCMAKE_BUILD_TYPE=Release -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON
 make -j4
 
+ccache -s
+
 ctest -VV -C Release

CMakeLists.txt

@@ -269,10 +269,18 @@ if (YUZU_USE_BUNDLED_UNICORN)
 
         find_package(PythonInterp 2.7 REQUIRED)
 
-        add_custom_command(OUTPUT ${LIBUNICORN_LIBRARY}
-            COMMAND ${CMAKE_COMMAND} -E env UNICORN_ARCHS="aarch64" PYTHON="${PYTHON_EXECUTABLE}" /bin/sh make.sh macos-universal-no
-            WORKING_DIRECTORY ${UNICORN_PREFIX}
-        )
+        if (MINGW)
+            add_custom_command(OUTPUT ${LIBUNICORN_LIBRARY}
+                COMMAND ${CMAKE_COMMAND} -E env UNICORN_ARCHS="aarch64" PYTHON="${PYTHON_EXECUTABLE}" /bin/sh make.sh cross-win64
+                WORKING_DIRECTORY ${UNICORN_PREFIX}
+            )
+        else()
+            add_custom_command(OUTPUT ${LIBUNICORN_LIBRARY}
+                COMMAND ${CMAKE_COMMAND} -E env UNICORN_ARCHS="aarch64" PYTHON="${PYTHON_EXECUTABLE}" /bin/sh make.sh macos-universal-no
+                WORKING_DIRECTORY ${UNICORN_PREFIX}
+            )
+        endif()
+
         # ALL makes this custom target build every time
         # but it won't actually build if LIBUNICORN_LIBRARY is up to date
         add_custom_target(unicorn-build ALL
@@ -286,6 +294,7 @@ endif()
 
 if (UNICORN_FOUND)
     add_library(unicorn INTERFACE)
+    add_dependencies(unicorn unicorn-build)
     target_link_libraries(unicorn INTERFACE "${LIBUNICORN_LIBRARY}")
     target_include_directories(unicorn INTERFACE "${LIBUNICORN_INCLUDE_DIR}")
 else()

CMakeModules/MinGWCross.cmake

@@ -0,0 +1,54 @@
+set(MINGW_PREFIX   /usr/x86_64-w64-mingw32/)
+set(CMAKE_SYSTEM_NAME               Windows)
+set(CMAKE_SYSTEM_PROCESSOR           x86_64)
+# Actually a hack, w/o this will cause some strange errors
+set(CMAKE_HOST_WIN32                 TRUE)
+
+
+set(CMAKE_FIND_ROOT_PATH            ${MINGW_PREFIX})
+set(SDL2_PATH                       ${MINGW_PREFIX})
+set(MINGW_TOOL_PREFIX               ${CMAKE_SYSTEM_PROCESSOR}-w64-mingw32-)
+
+# Specify the cross compiler
+set(CMAKE_C_COMPILER            ${MINGW_TOOL_PREFIX}gcc-posix)
+set(CMAKE_CXX_COMPILER          ${MINGW_TOOL_PREFIX}g++-posix)
+set(CMAKE_RC_COMPILER           ${MINGW_TOOL_PREFIX}windres)
+
+# Mingw tools
+set(STRIP                       ${MINGW_TOOL_PREFIX}strip)
+set(WINDRES                     ${MINGW_TOOL_PREFIX}windres)
+set(ENV{PKG_CONFIG}             ${MINGW_TOOL_PREFIX}pkg-config)
+
+# ccache wrapper
+option(USE_CCACHE "Use ccache for compilation" OFF)
+if(USE_CCACHE)
+    find_program(CCACHE ccache)
+    if(CCACHE)
+        message(STATUS "Using ccache found in PATH")
+        set_property(GLOBAL PROPERTY RULE_LAUNCH_COMPILE ${CCACHE})
+        set_property(GLOBAL PROPERTY RULE_LAUNCH_LINK ${CCACHE})
+    else(CCACHE)
+        message(WARNING "USE_CCACHE enabled, but no ccache found")
+    endif(CCACHE)
+endif(USE_CCACHE)
+
+# Search for programs in the build host directories
+set(CMAKE_FIND_ROOT_PATH_MODE_PROGRAM NEVER)
+
+
+# Echo modified cmake vars to screen for debugging purposes
+if(NOT DEFINED ENV{MINGW_DEBUG_INFO})
+        message("")
+        message("Custom cmake vars: (blank = system default)")
+        message("-----------------------------------------")
+        message("* CMAKE_C_COMPILER                     : ${CMAKE_C_COMPILER}")
+        message("* CMAKE_CXX_COMPILER                   : ${CMAKE_CXX_COMPILER}")
+        message("* CMAKE_RC_COMPILER                    : ${CMAKE_RC_COMPILER}")
+        message("* WINDRES                              : ${WINDRES}")
+        message("* ENV{PKG_CONFIG}                      : $ENV{PKG_CONFIG}")
+        message("* STRIP                                : ${STRIP}")
+        message("* USE_CCACHE                           : ${USE_CCACHE}")
+        message("")
+        # So that the debug info only appears once
+        set(ENV{MINGW_DEBUG_INFO} SHOWN)
+endif()

appveyor.yml

@@ -163,13 +163,3 @@ artifacts:
     name: build
     type: zip
 
-deploy:
-  provider: GitHub
-  release: $(appveyor_repo_tag_name)
-  auth_token:
-    secure: QqePPnXbkzmXct5c8hZ2X5AbsthbI6cS1Sr+VBzcD8oUOIjfWJJKXVAQGUbQAbb0
-  artifact: update,build
-  draft: false
-  prerelease: false
-  on:
-    appveyor_repo_tag: true

src/audio_core/audio_renderer.cpp

@@ -3,9 +3,12 @@
 // Refer to the license.txt file included.
 
 #include "audio_core/algorithm/interpolate.h"
+#include "audio_core/audio_out.h"
 #include "audio_core/audio_renderer.h"
+#include "audio_core/codec.h"
 #include "common/assert.h"
 #include "common/logging/log.h"
+#include "core/hle/kernel/event.h"
 #include "core/memory.h"
 
 namespace AudioCore {
@@ -13,6 +16,41 @@ namespace AudioCore {
 constexpr u32 STREAM_SAMPLE_RATE{48000};
 constexpr u32 STREAM_NUM_CHANNELS{2};
 
+class AudioRenderer::VoiceState {
+public:
+    bool IsPlaying() const {
+        return is_in_use && info.play_state == PlayState::Started;
+    }
+
+    const VoiceOutStatus& GetOutStatus() const {
+        return out_status;
+    }
+
+    const VoiceInfo& GetInfo() const {
+        return info;
+    }
+
+    VoiceInfo& Info() {
+        return info;
+    }
+
+    void SetWaveIndex(std::size_t index);
+    std::vector<s16> DequeueSamples(std::size_t sample_count);
+    void UpdateState();
+    void RefreshBuffer();
+
+private:
+    bool is_in_use{};
+    bool is_refresh_pending{};
+    std::size_t wave_index{};
+    std::size_t offset{};
+    Codec::ADPCMState adpcm_state{};
+    InterpolationState interp_state{};
+    std::vector<s16> samples;
+    VoiceOutStatus out_status{};
+    VoiceInfo info{};
+};
+
 AudioRenderer::AudioRenderer(AudioRendererParameter params,
                              Kernel::SharedPtr<Kernel::Event> buffer_event)
     : worker_params{params}, buffer_event{buffer_event}, voices(params.voice_count) {
@@ -27,6 +65,8 @@ AudioRenderer::AudioRenderer(AudioRendererParameter params,
     QueueMixedBuffer(2);
 }
 
+AudioRenderer::~AudioRenderer() = default;
+
 u32 AudioRenderer::GetSampleRate() const {
     return worker_params.sample_rate;
 }

src/audio_core/audio_renderer.h

@@ -8,16 +8,20 @@
 #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"
+#include "common/common_funcs.h"
 #include "common/common_types.h"
 #include "common/swap.h"
-#include "core/hle/kernel/event.h"
+#include "core/hle/kernel/object.h"
+
+namespace Kernel {
+class Event;
+}
 
 namespace AudioCore {
 
+class AudioOut;
+
 enum class PlayState : u8 {
     Started = 0,
     Stopped = 1,
@@ -158,6 +162,8 @@ static_assert(sizeof(UpdateDataHeader) == 0x40, "UpdateDataHeader has wrong size
 class AudioRenderer {
 public:
     AudioRenderer(AudioRendererParameter params, Kernel::SharedPtr<Kernel::Event> buffer_event);
+    ~AudioRenderer();
+
     std::vector<u8> UpdateAudioRenderer(const std::vector<u8>& input_params);
     void QueueMixedBuffer(Buffer::Tag tag);
     void ReleaseAndQueueBuffers();
@@ -166,45 +172,12 @@ public:
     u32 GetMixBufferCount() const;
 
 private:
-    class VoiceState {
-    public:
-        bool IsPlaying() const {
-            return is_in_use && info.play_state == PlayState::Started;
-        }
-
-        const VoiceOutStatus& GetOutStatus() const {
-            return out_status;
-        }
-
-        const VoiceInfo& GetInfo() const {
-            return info;
-        }
-
-        VoiceInfo& Info() {
-            return info;
-        }
-
-        void SetWaveIndex(std::size_t index);
-        std::vector<s16> DequeueSamples(std::size_t sample_count);
-        void UpdateState();
-        void RefreshBuffer();
-
-    private:
-        bool is_in_use{};
-        bool is_refresh_pending{};
-        std::size_t wave_index{};
-        std::size_t offset{};
-        Codec::ADPCMState adpcm_state{};
-        InterpolationState interp_state{};
-        std::vector<s16> samples;
-        VoiceOutStatus out_status{};
-        VoiceInfo info{};
-    };
+    class VoiceState;
 
     AudioRendererParameter worker_params;
     Kernel::SharedPtr<Kernel::Event> buffer_event;
     std::vector<VoiceState> voices;
-    std::unique_ptr<AudioCore::AudioOut> audio_out;
+    std::unique_ptr<AudioOut> audio_out;
     AudioCore::StreamPtr stream;
 };
 

src/audio_core/stream.cpp

@@ -7,6 +7,7 @@
 
 #include "audio_core/sink.h"
 #include "audio_core/sink_details.h"
+#include "audio_core/sink_stream.h"
 #include "audio_core/stream.h"
 #include "common/assert.h"
 #include "common/logging/log.h"

src/audio_core/stream.h

@@ -11,13 +11,16 @@
 #include <queue>
 
 #include "audio_core/buffer.h"
-#include "audio_core/sink_stream.h"
-#include "common/assert.h"
 #include "common/common_types.h"
-#include "core/core_timing.h"
+
+namespace CoreTiming {
+struct EventType;
+}
 
 namespace AudioCore {
 
+class SinkStream;
+
 /**
  * Represents an audio stream, which is a sequence of queued buffers, to be outputed by AudioOut
  */

src/audio_core/time_stretch.h

@@ -4,7 +4,6 @@
 
 #pragma once
 
-#include <array>
 #include <cstddef>
 #include <SoundTouch.h>
 #include "common/common_types.h"

src/common/ring_buffer.h

@@ -9,6 +9,7 @@
 #include <atomic>
 #include <cstddef>
 #include <cstring>
+#include <new>
 #include <type_traits>
 #include <vector>
 #include "common/common_types.h"
@@ -29,7 +30,7 @@ class RingBuffer {
     static_assert(capacity < std::numeric_limits<std::size_t>::max() / 2 / granularity);
     static_assert((capacity & (capacity - 1)) == 0, "capacity must be a power of two");
     // Ensure lock-free.
-    static_assert(std::atomic<std::size_t>::is_always_lock_free);
+    static_assert(std::atomic_size_t::is_always_lock_free);
 
 public:
     /// Pushes slots into the ring buffer
@@ -102,8 +103,15 @@ public:
 private:
     // It is important to align the below variables for performance reasons:
     // Having them on the same cache-line would result in false-sharing between them.
-    alignas(128) std::atomic<std::size_t> m_read_index{0};
-    alignas(128) std::atomic<std::size_t> m_write_index{0};
+    // TODO: Remove this ifdef whenever clang and GCC support
+    //       std::hardware_destructive_interference_size.
+#if defined(_MSC_VER) && _MSC_VER >= 1911
+    alignas(std::hardware_destructive_interference_size) std::atomic_size_t m_read_index{0};
+    alignas(std::hardware_destructive_interference_size) std::atomic_size_t m_write_index{0};
+#else
+    alignas(128) std::atomic_size_t m_read_index{0};
+    alignas(128) std::atomic_size_t m_write_index{0};
+#endif
 
     std::array<T, granularity * capacity> m_data;
 };

src/core/arm/arm_interface.h

@@ -6,11 +6,14 @@
 
 #include <array>
 #include "common/common_types.h"
-#include "core/hle/kernel/vm_manager.h"
+
+namespace Kernel {
+enum class VMAPermission : u8;
+}
 
 namespace Core {
 
-/// Generic ARM11 CPU interface
+/// Generic ARMv8 CPU interface
 class ARM_Interface : NonCopyable {
 public:
     virtual ~ARM_Interface() {}
@@ -19,9 +22,9 @@ public:
         std::array<u64, 31> cpu_registers;
         u64 sp;
         u64 pc;
-        u64 cpsr;
-        std::array<u128, 32> fpu_registers;
-        u64 fpscr;
+        u64 pstate;
+        std::array<u128, 32> vector_registers;
+        u64 fpcr;
     };
 
     /// Runs the CPU until an event happens
@@ -69,42 +72,50 @@ public:
      */
     virtual void SetReg(int index, u64 value) = 0;
 
-    virtual u128 GetExtReg(int index) const = 0;
-
-    virtual void SetExtReg(int index, u128 value) = 0;
+    /**
+     * Gets the value of a specified vector register.
+     *
+     * @param index The index of the vector register.
+     * @return the value within the vector register.
+     */
+    virtual u128 GetVectorReg(int index) const = 0;
 
     /**
-     * Gets the value of a VFP register
-     * @param index Register index (0-31)
-     * @return Returns the value in the register
+     * Sets a given value into a vector register.
+     *
+     * @param index The index of the vector register.
+     * @param value The new value to place in the register.
      */
-    virtual u32 GetVFPReg(int index) const = 0;
+    virtual void SetVectorReg(int index, u128 value) = 0;
 
     /**
-     * Sets a VFP register to the given value
-     * @param index Register index (0-31)
-     * @param value Value to set register to
+     * Get the current PSTATE register
+     * @return Returns the value of the PSTATE register
      */
-    virtual void SetVFPReg(int index, u32 value) = 0;
+    virtual u32 GetPSTATE() const = 0;
 
     /**
-     * Get the current CPSR register
-     * @return Returns the value of the CPSR register
+     * Set the current PSTATE register
+     * @param pstate Value to set PSTATE to
      */
-    virtual u32 GetCPSR() const = 0;
-
-    /**
-     * Set the current CPSR register
-     * @param cpsr Value to set CPSR to
-     */
-    virtual void SetCPSR(u32 cpsr) = 0;
+    virtual void SetPSTATE(u32 pstate) = 0;
 
     virtual VAddr GetTlsAddress() const = 0;
 
     virtual void SetTlsAddress(VAddr address) = 0;
 
+    /**
+     * Gets the value within the TPIDR_EL0 (read/write software thread ID) register.
+     *
+     * @return the value within the register.
+     */
     virtual u64 GetTPIDR_EL0() const = 0;
 
+    /**
+     * Sets a new value within the TPIDR_EL0 (read/write software thread ID) register.
+     *
+     * @param value The new value to place in the register.
+     */
     virtual void SetTPIDR_EL0(u64 value) = 0;
 
     /**
@@ -119,6 +130,7 @@ public:
      */
     virtual void LoadContext(const ThreadContext& ctx) = 0;
 
+    /// Clears the exclusive monitor's state.
     virtual void ClearExclusiveState() = 0;
 
     /// Prepare core for thread reschedule (if needed to correctly handle state)

src/core/arm/dynarmic/arm_dynarmic.cpp

@@ -12,8 +12,10 @@
 #include "core/core.h"
 #include "core/core_cpu.h"
 #include "core/core_timing.h"
+#include "core/gdbstub/gdbstub.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/svc.h"
+#include "core/hle/kernel/vm_manager.h"
 #include "core/memory.h"
 
 namespace Core {
@@ -79,6 +81,17 @@ public:
         case Dynarmic::A64::Exception::SendEventLocal:
         case Dynarmic::A64::Exception::Yield:
             return;
+        case Dynarmic::A64::Exception::Breakpoint:
+            if (GDBStub::IsServerEnabled()) {
+                parent.jit->HaltExecution();
+                parent.SetPC(pc);
+                Kernel::Thread* thread = Kernel::GetCurrentThread();
+                parent.SaveContext(thread->context);
+                GDBStub::Break();
+                GDBStub::SendTrap(thread, 5);
+                return;
+            }
+            [[fallthrough]];
         default:
             ASSERT_MSG(false, "ExceptionRaised(exception = {}, pc = {:X})",
                        static_cast<std::size_t>(exception), pc);
@@ -194,29 +207,20 @@ void ARM_Dynarmic::SetReg(int index, u64 value) {
     jit->SetRegister(index, value);
 }
 
-u128 ARM_Dynarmic::GetExtReg(int index) const {
+u128 ARM_Dynarmic::GetVectorReg(int index) const {
     return jit->GetVector(index);
 }
 
-void ARM_Dynarmic::SetExtReg(int index, u128 value) {
+void ARM_Dynarmic::SetVectorReg(int index, u128 value) {
     jit->SetVector(index, value);
 }
 
-u32 ARM_Dynarmic::GetVFPReg(int /*index*/) const {
-    UNIMPLEMENTED();
-    return {};
-}
-
-void ARM_Dynarmic::SetVFPReg(int /*index*/, u32 /*value*/) {
-    UNIMPLEMENTED();
-}
-
-u32 ARM_Dynarmic::GetCPSR() const {
+u32 ARM_Dynarmic::GetPSTATE() const {
     return jit->GetPstate();
 }
 
-void ARM_Dynarmic::SetCPSR(u32 cpsr) {
-    jit->SetPstate(cpsr);
+void ARM_Dynarmic::SetPSTATE(u32 pstate) {
+    jit->SetPstate(pstate);
 }
 
 u64 ARM_Dynarmic::GetTlsAddress() const {
@@ -239,18 +243,18 @@ void ARM_Dynarmic::SaveContext(ThreadContext& ctx) {
     ctx.cpu_registers = jit->GetRegisters();
     ctx.sp = jit->GetSP();
     ctx.pc = jit->GetPC();
-    ctx.cpsr = jit->GetPstate();
-    ctx.fpu_registers = jit->GetVectors();
-    ctx.fpscr = jit->GetFpcr();
+    ctx.pstate = jit->GetPstate();
+    ctx.vector_registers = jit->GetVectors();
+    ctx.fpcr = jit->GetFpcr();
 }
 
 void ARM_Dynarmic::LoadContext(const ThreadContext& ctx) {
     jit->SetRegisters(ctx.cpu_registers);
     jit->SetSP(ctx.sp);
     jit->SetPC(ctx.pc);
-    jit->SetPstate(static_cast<u32>(ctx.cpsr));
-    jit->SetVectors(ctx.fpu_registers);
-    jit->SetFpcr(static_cast<u32>(ctx.fpscr));
+    jit->SetPstate(static_cast<u32>(ctx.pstate));
+    jit->SetVectors(ctx.vector_registers);
+    jit->SetFpcr(static_cast<u32>(ctx.fpcr));
 }
 
 void ARM_Dynarmic::PrepareReschedule() {
@@ -304,8 +308,8 @@ bool DynarmicExclusiveMonitor::ExclusiveWrite64(std::size_t core_index, VAddr va
 
 bool DynarmicExclusiveMonitor::ExclusiveWrite128(std::size_t core_index, VAddr vaddr, u128 value) {
     return monitor.DoExclusiveOperation(core_index, vaddr, 16, [&] {
-        Memory::Write64(vaddr, value[0]);
-        Memory::Write64(vaddr, value[1]);
+        Memory::Write64(vaddr + 0, value[0]);
+        Memory::Write64(vaddr + 8, value[1]);
     });
 }
 

src/core/arm/dynarmic/arm_dynarmic.h

@@ -12,6 +12,10 @@
 #include "core/arm/exclusive_monitor.h"
 #include "core/arm/unicorn/arm_unicorn.h"
 
+namespace Memory {
+struct PageTable;
+}
+
 namespace Core {
 
 class ARM_Dynarmic_Callbacks;
@@ -29,14 +33,12 @@ public:
     u64 GetPC() const override;
     u64 GetReg(int index) const override;
     void SetReg(int index, u64 value) override;
-    u128 GetExtReg(int index) const override;
-    void SetExtReg(int index, u128 value) override;
-    u32 GetVFPReg(int index) const override;
-    void SetVFPReg(int index, u32 value) override;
-    u32 GetCPSR() const override;
+    u128 GetVectorReg(int index) const override;
+    void SetVectorReg(int index, u128 value) override;
+    u32 GetPSTATE() const override;
+    void SetPSTATE(u32 pstate) override;
     void Run() override;
     void Step() override;
-    void SetCPSR(u32 cpsr) override;
     VAddr GetTlsAddress() const override;
     void SetTlsAddress(VAddr address) override;
     void SetTPIDR_EL0(u64 value) override;

src/core/arm/unicorn/arm_unicorn.cpp

@@ -131,33 +131,24 @@ void ARM_Unicorn::SetReg(int regn, u64 val) {
     CHECKED(uc_reg_write(uc, treg, &val));
 }
 
-u128 ARM_Unicorn::GetExtReg(int /*index*/) const {
+u128 ARM_Unicorn::GetVectorReg(int /*index*/) const {
     UNIMPLEMENTED();
     static constexpr u128 res{};
     return res;
 }
 
-void ARM_Unicorn::SetExtReg(int /*index*/, u128 /*value*/) {
+void ARM_Unicorn::SetVectorReg(int /*index*/, u128 /*value*/) {
     UNIMPLEMENTED();
 }
 
-u32 ARM_Unicorn::GetVFPReg(int /*index*/) const {
-    UNIMPLEMENTED();
-    return {};
-}
-
-void ARM_Unicorn::SetVFPReg(int /*index*/, u32 /*value*/) {
-    UNIMPLEMENTED();
-}
-
-u32 ARM_Unicorn::GetCPSR() const {
+u32 ARM_Unicorn::GetPSTATE() const {
     u64 nzcv{};
     CHECKED(uc_reg_read(uc, UC_ARM64_REG_NZCV, &nzcv));
     return static_cast<u32>(nzcv);
 }
 
-void ARM_Unicorn::SetCPSR(u32 cpsr) {
-    u64 nzcv = cpsr;
+void ARM_Unicorn::SetPSTATE(u32 pstate) {
+    u64 nzcv = pstate;
     CHECKED(uc_reg_write(uc, UC_ARM64_REG_NZCV, &nzcv));
 }
 
@@ -219,7 +210,7 @@ void ARM_Unicorn::SaveContext(ThreadContext& ctx) {
 
     CHECKED(uc_reg_read(uc, UC_ARM64_REG_SP, &ctx.sp));
     CHECKED(uc_reg_read(uc, UC_ARM64_REG_PC, &ctx.pc));
-    CHECKED(uc_reg_read(uc, UC_ARM64_REG_NZCV, &ctx.cpsr));
+    CHECKED(uc_reg_read(uc, UC_ARM64_REG_NZCV, &ctx.pstate));
 
     for (auto i = 0; i < 29; ++i) {
         uregs[i] = UC_ARM64_REG_X0 + i;
@@ -234,7 +225,7 @@ void ARM_Unicorn::SaveContext(ThreadContext& ctx) {
 
     for (int i = 0; i < 32; ++i) {
         uregs[i] = UC_ARM64_REG_Q0 + i;
-        tregs[i] = &ctx.fpu_registers[i];
+        tregs[i] = &ctx.vector_registers[i];
     }
 
     CHECKED(uc_reg_read_batch(uc, uregs, tregs, 32));
@@ -246,7 +237,7 @@ void ARM_Unicorn::LoadContext(const ThreadContext& ctx) {
 
     CHECKED(uc_reg_write(uc, UC_ARM64_REG_SP, &ctx.sp));
     CHECKED(uc_reg_write(uc, UC_ARM64_REG_PC, &ctx.pc));
-    CHECKED(uc_reg_write(uc, UC_ARM64_REG_NZCV, &ctx.cpsr));
+    CHECKED(uc_reg_write(uc, UC_ARM64_REG_NZCV, &ctx.pstate));
 
     for (int i = 0; i < 29; ++i) {
         uregs[i] = UC_ARM64_REG_X0 + i;
@@ -261,7 +252,7 @@ void ARM_Unicorn::LoadContext(const ThreadContext& ctx) {
 
     for (auto i = 0; i < 32; ++i) {
         uregs[i] = UC_ARM64_REG_Q0 + i;
-        tregs[i] = (void*)&ctx.fpu_registers[i];
+        tregs[i] = (void*)&ctx.vector_registers[i];
     }
 
     CHECKED(uc_reg_write_batch(uc, uregs, tregs, 32));

src/core/arm/unicorn/arm_unicorn.h

@@ -22,12 +22,10 @@ public:
     u64 GetPC() const override;
     u64 GetReg(int index) const override;
     void SetReg(int index, u64 value) override;
-    u128 GetExtReg(int index) const override;
-    void SetExtReg(int index, u128 value) override;
-    u32 GetVFPReg(int index) const override;
-    void SetVFPReg(int index, u32 value) override;
-    u32 GetCPSR() const override;
-    void SetCPSR(u32 cpsr) override;
+    u128 GetVectorReg(int index) const override;
+    void SetVectorReg(int index, u128 value) override;
+    u32 GetPSTATE() const override;
+    void SetPSTATE(u32 pstate) override;
     VAddr GetTlsAddress() const override;
     void SetTlsAddress(VAddr address) override;
     void SetTPIDR_EL0(u64 value) override;

src/core/core_cpu.cpp

@@ -9,6 +9,7 @@
 #ifdef ARCHITECTURE_x86_64
 #include "core/arm/dynarmic/arm_dynarmic.h"
 #endif
+#include "core/arm/exclusive_monitor.h"
 #include "core/arm/unicorn/arm_unicorn.h"
 #include "core/core_cpu.h"
 #include "core/core_timing.h"
@@ -66,6 +67,8 @@ Cpu::Cpu(std::shared_ptr<ExclusiveMonitor> exclusive_monitor,
     scheduler = std::make_shared<Kernel::Scheduler>(arm_interface.get());
 }
 
+Cpu::~Cpu() = default;
+
 std::shared_ptr<ExclusiveMonitor> Cpu::MakeExclusiveMonitor(std::size_t num_cores) {
     if (Settings::values.use_cpu_jit) {
 #ifdef ARCHITECTURE_x86_64

src/core/core_cpu.h

@@ -6,11 +6,10 @@
 
 #include <atomic>
 #include <condition_variable>
+#include <cstddef>
 #include <memory>
 #include <mutex>
-#include <string>
 #include "common/common_types.h"
-#include "core/arm/exclusive_monitor.h"
 
 namespace Kernel {
 class Scheduler;
@@ -19,6 +18,7 @@ class Scheduler;
 namespace Core {
 
 class ARM_Interface;
+class ExclusiveMonitor;
 
 constexpr unsigned NUM_CPU_CORES{4};
 
@@ -43,6 +43,7 @@ class Cpu {
 public:
     Cpu(std::shared_ptr<ExclusiveMonitor> exclusive_monitor,
         std::shared_ptr<CpuBarrier> cpu_barrier, std::size_t core_index);
+    ~Cpu();
 
     void RunLoop(bool tight_loop = true);
 

src/core/file_sys/content_archive.cpp

@@ -463,6 +463,8 @@ NCA::NCA(VirtualFile file_, VirtualFile bktr_base_romfs_, u64 bktr_base_ivfc_off
     status = Loader::ResultStatus::Success;
 }
 
+NCA::~NCA() = default;
+
 Loader::ResultStatus NCA::GetStatus() const {
     return status;
 }

src/core/file_sys/content_archive.h

@@ -81,6 +81,8 @@ class NCA : public ReadOnlyVfsDirectory {
 public:
     explicit NCA(VirtualFile file, VirtualFile bktr_base_romfs = nullptr,
                  u64 bktr_base_ivfc_offset = 0);
+    ~NCA() override;
+
     Loader::ResultStatus GetStatus() const;
 
     std::vector<std::shared_ptr<VfsFile>> GetFiles() const override;

src/core/file_sys/control_metadata.cpp

@@ -8,6 +8,14 @@
 
 namespace FileSys {
 
+const std::array<const char*, 15> LANGUAGE_NAMES = {
+    "AmericanEnglish", "BritishEnglish", "Japanese",
+    "French",          "German",         "LatinAmericanSpanish",
+    "Spanish",         "Italian",        "Dutch",
+    "CanadianFrench",  "Portugese",      "Russian",
+    "Korean",          "Taiwanese",      "Chinese",
+};
+
 std::string LanguageEntry::GetApplicationName() const {
     return Common::StringFromFixedZeroTerminatedBuffer(application_name.data(), 0x200);
 }
@@ -20,18 +28,20 @@ NACP::NACP(VirtualFile file) : raw(std::make_unique<RawNACP>()) {
     file->ReadObject(raw.get());
 }
 
+NACP::~NACP() = default;
+
 const LanguageEntry& NACP::GetLanguageEntry(Language language) const {
     if (language != Language::Default) {
         return raw->language_entries.at(static_cast<u8>(language));
-    } else {
-        for (const auto& language_entry : raw->language_entries) {
-            if (!language_entry.GetApplicationName().empty())
-                return language_entry;
-        }
-
-        // Fallback to English
-        return GetLanguageEntry(Language::AmericanEnglish);
     }
+
+    for (const auto& language_entry : raw->language_entries) {
+        if (!language_entry.GetApplicationName().empty())
+            return language_entry;
+    }
+
+    // Fallback to English
+    return GetLanguageEntry(Language::AmericanEnglish);
 }
 
 std::string NACP::GetApplicationName(Language language) const {

src/core/file_sys/control_metadata.h

@@ -66,18 +66,15 @@ enum class Language : u8 {
     Default = 255,
 };
 
-static constexpr std::array<const char*, 15> LANGUAGE_NAMES = {
-    "AmericanEnglish", "BritishEnglish", "Japanese",
-    "French",          "German",         "LatinAmericanSpanish",
-    "Spanish",         "Italian",        "Dutch",
-    "CanadianFrench",  "Portugese",      "Russian",
-    "Korean",          "Taiwanese",      "Chinese"};
+extern const std::array<const char*, 15> LANGUAGE_NAMES;
 
 // A class representing the format used by NX metadata files, typically named Control.nacp.
 // These store application name, dev name, title id, and other miscellaneous data.
 class NACP {
 public:
     explicit NACP(VirtualFile file);
+    ~NACP();
+
     const LanguageEntry& GetLanguageEntry(Language language = Language::Default) const;
     std::string GetApplicationName(Language language = Language::Default) const;
     std::string GetDeveloperName(Language language = Language::Default) const;

src/core/file_sys/nca_metadata.cpp

@@ -51,6 +51,8 @@ CNMT::CNMT(CNMTHeader header, OptionalHeader opt_header, std::vector<ContentReco
     : header(std::move(header)), opt_header(std::move(opt_header)),
       content_records(std::move(content_records)), meta_records(std::move(meta_records)) {}
 
+CNMT::~CNMT() = default;
+
 u64 CNMT::GetTitleID() const {
     return header.title_id;
 }

src/core/file_sys/nca_metadata.h

@@ -87,6 +87,7 @@ public:
     explicit CNMT(VirtualFile file);
     CNMT(CNMTHeader header, OptionalHeader opt_header, std::vector<ContentRecord> content_records,
          std::vector<MetaRecord> meta_records);
+    ~CNMT();
 
     u64 GetTitleID() const;
     u32 GetTitleVersion() const;

src/core/file_sys/partition_filesystem.cpp

@@ -72,6 +72,8 @@ PartitionFilesystem::PartitionFilesystem(std::shared_ptr<VfsFile> file) {
     status = Loader::ResultStatus::Success;
 }
 
+PartitionFilesystem::~PartitionFilesystem() = default;
+
 Loader::ResultStatus PartitionFilesystem::GetStatus() const {
     return status;
 }

src/core/file_sys/partition_filesystem.h

@@ -25,6 +25,8 @@ namespace FileSys {
 class PartitionFilesystem : public ReadOnlyVfsDirectory {
 public:
     explicit PartitionFilesystem(std::shared_ptr<VfsFile> file);
+    ~PartitionFilesystem() override;
+
     Loader::ResultStatus GetStatus() const;
 
     std::vector<std::shared_ptr<VfsFile>> GetFiles() const override;

src/core/file_sys/patch_manager.cpp

@@ -41,6 +41,8 @@ std::string FormatPatchTypeName(PatchType type) {
 
 PatchManager::PatchManager(u64 title_id) : title_id(title_id) {}
 
+PatchManager::~PatchManager() = default;
+
 VirtualDir PatchManager::PatchExeFS(VirtualDir exefs) const {
     LOG_INFO(Loader, "Patching ExeFS for title_id={:016X}", title_id);
 

src/core/file_sys/patch_manager.h

@@ -34,6 +34,7 @@ std::string FormatPatchTypeName(PatchType type);
 class PatchManager {
 public:
     explicit PatchManager(u64 title_id);
+    ~PatchManager();
 
     // Currently tracked ExeFS patches:
     // - Game Updates

src/core/file_sys/program_metadata.cpp

@@ -12,6 +12,10 @@
 
 namespace FileSys {
 
+ProgramMetadata::ProgramMetadata() = default;
+
+ProgramMetadata::~ProgramMetadata() = default;
+
 Loader::ResultStatus ProgramMetadata::Load(VirtualFile file) {
     std::size_t total_size = static_cast<std::size_t>(file->GetSize());
     if (total_size < sizeof(Header))

src/core/file_sys/program_metadata.h

@@ -36,6 +36,9 @@ enum class ProgramFilePermission : u64 {
  */
 class ProgramMetadata {
 public:
+    ProgramMetadata();
+    ~ProgramMetadata();
+
     Loader::ResultStatus Load(VirtualFile file);
 
     bool Is64BitProgram() const;

src/core/file_sys/romfs_factory.cpp

@@ -28,6 +28,8 @@ RomFSFactory::RomFSFactory(Loader::AppLoader& app_loader) {
     ivfc_offset = app_loader.ReadRomFSIVFCOffset();
 }
 
+RomFSFactory::~RomFSFactory() = default;
+
 ResultVal<VirtualFile> RomFSFactory::OpenCurrentProcess() {
     if (!updatable)
         return MakeResult<VirtualFile>(file);

src/core/file_sys/romfs_factory.h

@@ -30,6 +30,7 @@ enum class StorageId : u8 {
 class RomFSFactory {
 public:
     explicit RomFSFactory(Loader::AppLoader& app_loader);
+    ~RomFSFactory();
 
     ResultVal<VirtualFile> OpenCurrentProcess();
     ResultVal<VirtualFile> Open(u64 title_id, StorageId storage, ContentRecordType type);

src/core/file_sys/savedata_factory.cpp

@@ -20,6 +20,8 @@ std::string SaveDataDescriptor::DebugInfo() const {
 
 SaveDataFactory::SaveDataFactory(VirtualDir save_directory) : dir(std::move(save_directory)) {}
 
+SaveDataFactory::~SaveDataFactory() = default;
+
 ResultVal<VirtualDir> SaveDataFactory::Open(SaveDataSpaceId space, SaveDataDescriptor meta) {
     if (meta.type == SaveDataType::SystemSaveData || meta.type == SaveDataType::SaveData) {
         if (meta.zero_1 != 0) {
@@ -85,10 +87,10 @@ std::string SaveDataFactory::GetFullPath(SaveDataSpaceId space, SaveDataType typ
 
     switch (space) {
     case SaveDataSpaceId::NandSystem:
-        out = "/system/save/";
+        out = "/system/";
         break;
     case SaveDataSpaceId::NandUser:
-        out = "/user/save/";
+        out = "/user/";
         break;
     default:
         ASSERT_MSG(false, "Unrecognized SaveDataSpaceId: {:02X}", static_cast<u8>(space));
@@ -96,9 +98,12 @@ std::string SaveDataFactory::GetFullPath(SaveDataSpaceId space, SaveDataType typ
 
     switch (type) {
     case SaveDataType::SystemSaveData:
-        return fmt::format("{}{:016X}/{:016X}{:016X}", out, save_id, user_id[1], user_id[0]);
+        return fmt::format("{}save/{:016X}/{:016X}{:016X}", out, save_id, user_id[1], user_id[0]);
     case SaveDataType::SaveData:
-        return fmt::format("{}{:016X}/{:016X}{:016X}/{:016X}", out, 0, user_id[1], user_id[0],
+        return fmt::format("{}save/{:016X}/{:016X}{:016X}/{:016X}", out, 0, user_id[1], user_id[0],
+                           title_id);
+    case SaveDataType::TemporaryStorage:
+        return fmt::format("{}temp/{:016X}/{:016X}{:016X}/{:016X}", out, 0, user_id[1], user_id[0],
                            title_id);
     default:
         ASSERT_MSG(false, "Unrecognized SaveDataType: {:02X}", static_cast<u8>(type));

src/core/file_sys/savedata_factory.h

@@ -48,6 +48,7 @@ static_assert(sizeof(SaveDataDescriptor) == 0x40, "SaveDataDescriptor has incorr
 class SaveDataFactory {
 public:
     explicit SaveDataFactory(VirtualDir dir);
+    ~SaveDataFactory();
 
     ResultVal<VirtualDir> Open(SaveDataSpaceId space, SaveDataDescriptor meta);
 

src/core/file_sys/submission_package.h

@@ -24,7 +24,7 @@ enum class ContentRecordType : u8;
 class NSP : public ReadOnlyVfsDirectory {
 public:
     explicit NSP(VirtualFile file);
-    ~NSP();
+    ~NSP() override;
 
     Loader::ResultStatus GetStatus() const;
     Loader::ResultStatus GetProgramStatus(u64 title_id) const;

src/core/file_sys/vfs_concat.cpp

@@ -27,6 +27,8 @@ ConcatenatedVfsFile::ConcatenatedVfsFile(std::vector<VirtualFile> files_, std::s
     }
 }
 
+ConcatenatedVfsFile::~ConcatenatedVfsFile() = default;
+
 std::string ConcatenatedVfsFile::GetName() const {
     if (files.empty())
         return "";

src/core/file_sys/vfs_concat.h

@@ -22,6 +22,8 @@ class ConcatenatedVfsFile : public VfsFile {
     ConcatenatedVfsFile(std::vector<VirtualFile> files, std::string name);
 
 public:
+    ~ConcatenatedVfsFile() override;
+
     std::string GetName() const override;
     std::size_t GetSize() const override;
     bool Resize(std::size_t new_size) override;

src/core/file_sys/vfs_offset.cpp

@@ -14,6 +14,8 @@ OffsetVfsFile::OffsetVfsFile(std::shared_ptr<VfsFile> file_, std::size_t size_,
     : file(file_), offset(offset_), size(size_), name(std::move(name_)),
       parent(parent_ == nullptr ? file->GetContainingDirectory() : std::move(parent_)) {}
 
+OffsetVfsFile::~OffsetVfsFile() = default;
+
 std::string OffsetVfsFile::GetName() const {
     return name.empty() ? file->GetName() : name;
 }

src/core/file_sys/vfs_offset.h

@@ -19,6 +19,7 @@ class OffsetVfsFile : public VfsFile {
 public:
     OffsetVfsFile(std::shared_ptr<VfsFile> file, std::size_t size, std::size_t offset = 0,
                   std::string new_name = "", VirtualDir new_parent = nullptr);
+    ~OffsetVfsFile() override;
 
     std::string GetName() const override;
     std::size_t GetSize() const override;

src/core/file_sys/vfs_vector.cpp

@@ -13,6 +13,8 @@ VectorVfsDirectory::VectorVfsDirectory(std::vector<VirtualFile> files_,
     : files(std::move(files_)), dirs(std::move(dirs_)), parent(std::move(parent_)),
       name(std::move(name_)) {}
 
+VectorVfsDirectory::~VectorVfsDirectory() = default;
+
 std::vector<std::shared_ptr<VfsFile>> VectorVfsDirectory::GetFiles() const {
     return files;
 }

src/core/file_sys/vfs_vector.h

@@ -15,6 +15,7 @@ public:
     explicit VectorVfsDirectory(std::vector<VirtualFile> files = {},
                                 std::vector<VirtualDir> dirs = {}, std::string name = "",
                                 VirtualDir parent = nullptr);
+    ~VectorVfsDirectory() override;
 
     std::vector<std::shared_ptr<VfsFile>> GetFiles() const override;
     std::vector<std::shared_ptr<VfsDirectory>> GetSubdirectories() const override;

src/core/file_sys/xts_archive.cpp

@@ -30,9 +30,6 @@ static bool CalculateHMAC256(Destination* out, const SourceKey* key, std::size_t
     mbedtls_md_context_t context;
     mbedtls_md_init(&context);
 
-    const auto key_f = reinterpret_cast<const u8*>(key);
-    const std::vector<u8> key_v(key_f, key_f + key_length);
-
     if (mbedtls_md_setup(&context, mbedtls_md_info_from_type(MBEDTLS_MD_SHA256), 1) ||
         mbedtls_md_hmac_starts(&context, reinterpret_cast<const u8*>(key), key_length) ||
         mbedtls_md_hmac_update(&context, reinterpret_cast<const u8*>(data), data_length) ||
@@ -45,7 +42,7 @@ static bool CalculateHMAC256(Destination* out, const SourceKey* key, std::size_t
     return true;
 }
 
-NAX::NAX(VirtualFile file_) : file(std::move(file_)), header(std::make_unique<NAXHeader>()) {
+NAX::NAX(VirtualFile file_) : header(std::make_unique<NAXHeader>()), file(std::move(file_)) {
     std::string path = FileUtil::SanitizePath(file->GetFullPath());
     static const std::regex nax_path_regex("/registered/(000000[0-9A-F]{2})/([0-9A-F]{32})\\.nca",
                                            std::regex_constants::ECMAScript |
@@ -65,13 +62,15 @@ NAX::NAX(VirtualFile file_) : file(std::move(file_)), header(std::make_unique<NA
 }
 
 NAX::NAX(VirtualFile file_, std::array<u8, 0x10> nca_id)
-    : file(std::move(file_)), header(std::make_unique<NAXHeader>()) {
+    : header(std::make_unique<NAXHeader>()), file(std::move(file_)) {
     Core::Crypto::SHA256Hash hash{};
     mbedtls_sha256(nca_id.data(), nca_id.size(), hash.data(), 0);
     status = Parse(fmt::format("/registered/000000{:02X}/{}.nca", hash[0],
                                Common::HexArrayToString(nca_id, false)));
 }
 
+NAX::~NAX() = default;
+
 Loader::ResultStatus NAX::Parse(std::string_view path) {
     if (file->ReadObject(header.get()) != sizeof(NAXHeader))
         return Loader::ResultStatus::ErrorBadNAXHeader;
@@ -138,9 +137,9 @@ VirtualFile NAX::GetDecrypted() const {
     return dec_file;
 }
 
-std::shared_ptr<NCA> NAX::AsNCA() const {
+std::unique_ptr<NCA> NAX::AsNCA() const {
     if (type == NAXContentType::NCA)
-        return std::make_shared<NCA>(GetDecrypted());
+        return std::make_unique<NCA>(GetDecrypted());
     return nullptr;
 }
 

src/core/file_sys/xts_archive.h

@@ -33,12 +33,13 @@ class NAX : public ReadOnlyVfsDirectory {
 public:
     explicit NAX(VirtualFile file);
     explicit NAX(VirtualFile file, std::array<u8, 0x10> nca_id);
+    ~NAX() override;
 
     Loader::ResultStatus GetStatus() const;
 
     VirtualFile GetDecrypted() const;
 
-    std::shared_ptr<NCA> AsNCA() const;
+    std::unique_ptr<NCA> AsNCA() const;
 
     NAXContentType GetContentType() const;
 
@@ -60,7 +61,7 @@ private:
 
     VirtualFile file;
     Loader::ResultStatus status;
-    NAXContentType type;
+    NAXContentType type{};
 
     VirtualFile dec_file;
 

src/core/gdbstub/gdbstub.cpp

@@ -65,9 +65,9 @@ constexpr u32 MSG_WAITALL = 8;
 constexpr u32 LR_REGISTER = 30;
 constexpr u32 SP_REGISTER = 31;
 constexpr u32 PC_REGISTER = 32;
-constexpr u32 CPSR_REGISTER = 33;
+constexpr u32 PSTATE_REGISTER = 33;
 constexpr u32 UC_ARM64_REG_Q0 = 34;
-constexpr u32 FPSCR_REGISTER = 66;
+constexpr u32 FPCR_REGISTER = 66;
 
 // TODO/WiP - Used while working on support for FPU
 constexpr u32 TODO_DUMMY_REG_997 = 997;
@@ -116,7 +116,7 @@ constexpr char target_xml[] =
 
     <reg name="pc" bitsize="64" type="code_ptr"/>
 
-    <flags id="cpsr_flags" size="4">
+    <flags id="pstate_flags" size="4">
       <field name="SP" start="0" end="0"/>
       <field name="" start="1" end="1"/>
       <field name="EL" start="2" end="3"/>
@@ -135,7 +135,7 @@ constexpr char target_xml[] =
       <field name="Z" start="30" end="30"/>
       <field name="N" start="31" end="31"/>
     </flags>
-    <reg name="cpsr" bitsize="32" type="cpsr_flags"/>
+    <reg name="pstate" bitsize="32" type="pstate_flags"/>
   </feature>
   <feature name="org.gnu.gdb.aarch64.fpu">
   </feature>
@@ -227,10 +227,10 @@ static u64 RegRead(std::size_t id, Kernel::Thread* thread = nullptr) {
         return thread->context.sp;
     } else if (id == PC_REGISTER) {
         return thread->context.pc;
-    } else if (id == CPSR_REGISTER) {
-        return thread->context.cpsr;
-    } else if (id > CPSR_REGISTER && id < FPSCR_REGISTER) {
-        return thread->context.fpu_registers[id - UC_ARM64_REG_Q0][0];
+    } else if (id == PSTATE_REGISTER) {
+        return thread->context.pstate;
+    } else if (id > PSTATE_REGISTER && id < FPCR_REGISTER) {
+        return thread->context.vector_registers[id - UC_ARM64_REG_Q0][0];
     } else {
         return 0;
     }
@@ -247,10 +247,10 @@ static void RegWrite(std::size_t id, u64 val, Kernel::Thread* thread = nullptr)
         thread->context.sp = val;
     } else if (id == PC_REGISTER) {
         thread->context.pc = val;
-    } else if (id == CPSR_REGISTER) {
-        thread->context.cpsr = val;
-    } else if (id > CPSR_REGISTER && id < FPSCR_REGISTER) {
-        thread->context.fpu_registers[id - (CPSR_REGISTER + 1)][0] = val;
+    } else if (id == PSTATE_REGISTER) {
+        thread->context.pstate = val;
+    } else if (id > PSTATE_REGISTER && id < FPCR_REGISTER) {
+        thread->context.vector_registers[id - (PSTATE_REGISTER + 1)][0] = val;
     }
 }
 
@@ -781,11 +781,11 @@ static void ReadRegister() {
         LongToGdbHex(reply, RegRead(id, current_thread));
     } else if (id == PC_REGISTER) {
         LongToGdbHex(reply, RegRead(id, current_thread));
-    } else if (id == CPSR_REGISTER) {
-        IntToGdbHex(reply, (u32)RegRead(id, current_thread));
-    } else if (id >= UC_ARM64_REG_Q0 && id < FPSCR_REGISTER) {
+    } else if (id == PSTATE_REGISTER) {
+        IntToGdbHex(reply, static_cast<u32>(RegRead(id, current_thread)));
+    } else if (id >= UC_ARM64_REG_Q0 && id < FPCR_REGISTER) {
         LongToGdbHex(reply, RegRead(id, current_thread));
-    } else if (id == FPSCR_REGISTER) {
+    } else if (id == FPCR_REGISTER) {
         LongToGdbHex(reply, RegRead(TODO_DUMMY_REG_998, current_thread));
     } else {
         LongToGdbHex(reply, RegRead(TODO_DUMMY_REG_997, current_thread));
@@ -811,7 +811,7 @@ static void ReadRegisters() {
 
     bufptr += 16;
 
-    IntToGdbHex(bufptr, (u32)RegRead(CPSR_REGISTER, current_thread));
+    IntToGdbHex(bufptr, static_cast<u32>(RegRead(PSTATE_REGISTER, current_thread)));
 
     bufptr += 8;
 
@@ -843,11 +843,11 @@ static void WriteRegister() {
         RegWrite(id, GdbHexToLong(buffer_ptr), current_thread);
     } else if (id == PC_REGISTER) {
         RegWrite(id, GdbHexToLong(buffer_ptr), current_thread);
-    } else if (id == CPSR_REGISTER) {
+    } else if (id == PSTATE_REGISTER) {
         RegWrite(id, GdbHexToInt(buffer_ptr), current_thread);
-    } else if (id >= UC_ARM64_REG_Q0 && id < FPSCR_REGISTER) {
+    } else if (id >= UC_ARM64_REG_Q0 && id < FPCR_REGISTER) {
         RegWrite(id, GdbHexToLong(buffer_ptr), current_thread);
-    } else if (id == FPSCR_REGISTER) {
+    } else if (id == FPCR_REGISTER) {
         RegWrite(TODO_DUMMY_REG_998, GdbHexToLong(buffer_ptr), current_thread);
     } else {
         RegWrite(TODO_DUMMY_REG_997, GdbHexToLong(buffer_ptr), current_thread);
@@ -866,16 +866,16 @@ static void WriteRegisters() {
     if (command_buffer[0] != 'G')
         return SendReply("E01");
 
-    for (u32 i = 0, reg = 0; reg <= FPSCR_REGISTER; i++, reg++) {
+    for (u32 i = 0, reg = 0; reg <= FPCR_REGISTER; i++, reg++) {
         if (reg <= SP_REGISTER) {
             RegWrite(reg, GdbHexToLong(buffer_ptr + i * 16), current_thread);
         } else if (reg == PC_REGISTER) {
             RegWrite(PC_REGISTER, GdbHexToLong(buffer_ptr + i * 16), current_thread);
-        } else if (reg == CPSR_REGISTER) {
-            RegWrite(CPSR_REGISTER, GdbHexToInt(buffer_ptr + i * 16), current_thread);
-        } else if (reg >= UC_ARM64_REG_Q0 && reg < FPSCR_REGISTER) {
+        } else if (reg == PSTATE_REGISTER) {
+            RegWrite(PSTATE_REGISTER, GdbHexToInt(buffer_ptr + i * 16), current_thread);
+        } else if (reg >= UC_ARM64_REG_Q0 && reg < FPCR_REGISTER) {
             RegWrite(reg, GdbHexToLong(buffer_ptr + i * 16), current_thread);
-        } else if (reg == FPSCR_REGISTER) {
+        } else if (reg == FPCR_REGISTER) {
             RegWrite(TODO_DUMMY_REG_998, GdbHexToLong(buffer_ptr + i * 16), current_thread);
         } else {
             UNIMPLEMENTED();
@@ -995,7 +995,7 @@ static bool CommitBreakpoint(BreakpointType type, VAddr addr, u64 len) {
     breakpoint.addr = addr;
     breakpoint.len = len;
     Memory::ReadBlock(addr, breakpoint.inst.data(), breakpoint.inst.size());
-    static constexpr std::array<u8, 4> btrap{{0xd4, 0x20, 0x7d, 0x0}};
+    static constexpr std::array<u8, 4> btrap{{0x00, 0x7d, 0x20, 0xd4}};
     Memory::WriteBlock(addr, btrap.data(), btrap.size());
     Core::System::GetInstance().InvalidateCpuInstructionCaches();
     p.insert({addr, breakpoint});

src/core/hle/ipc_helpers.h

@@ -290,13 +290,6 @@ public:
         Skip(CommandIdSize, false);
     }
 
-    ResponseBuilder MakeBuilder(u32 normal_params_size, u32 num_handles_to_copy,
-                                u32 num_handles_to_move,
-                                ResponseBuilder::Flags flags = ResponseBuilder::Flags::None) const {
-        return ResponseBuilder{*context, normal_params_size, num_handles_to_copy,
-                               num_handles_to_move, flags};
-    }
-
     template <typename T>
     T Pop();
 

src/core/hle/kernel/mutex.cpp

@@ -16,6 +16,7 @@
 #include "core/hle/kernel/object.h"
 #include "core/hle/kernel/thread.h"
 #include "core/hle/result.h"
+#include "core/memory.h"
 
 namespace Kernel {
 

src/core/hle/kernel/process.cpp

@@ -125,7 +125,7 @@ void Process::Run(VAddr entry_point, s32 main_thread_priority, u32 stack_size) {
     vm_manager.LogLayout();
     status = ProcessStatus::Running;
 
-    Kernel::SetupMainThread(kernel, entry_point, main_thread_priority, this);
+    Kernel::SetupMainThread(kernel, entry_point, main_thread_priority, *this);
 }
 
 void Process::LoadModule(SharedPtr<CodeSet> module_, VAddr base_addr) {

src/core/hle/kernel/svc_wrap.h

@@ -13,7 +13,9 @@
 
 namespace Kernel {
 
-#define PARAM(n) Core::CurrentArmInterface().GetReg(n)
+static inline u64 Param(int n) {
+    return Core::CurrentArmInterface().GetReg(n);
+}
 
 /**
  * HLE a function return from the current ARM userland process
@@ -28,23 +30,23 @@ static inline void FuncReturn(u64 res) {
 
 template <ResultCode func(u64)>
 void SvcWrap() {
-    FuncReturn(func(PARAM(0)).raw);
+    FuncReturn(func(Param(0)).raw);
 }
 
 template <ResultCode func(u32)>
 void SvcWrap() {
-    FuncReturn(func((u32)PARAM(0)).raw);
+    FuncReturn(func((u32)Param(0)).raw);
 }
 
 template <ResultCode func(u32, u32)>
 void SvcWrap() {
-    FuncReturn(func((u32)PARAM(0), (u32)PARAM(1)).raw);
+    FuncReturn(func((u32)Param(0), (u32)Param(1)).raw);
 }
 
 template <ResultCode func(u32*, u32)>
 void SvcWrap() {
     u32 param_1 = 0;
-    u32 retval = func(&param_1, (u32)PARAM(1)).raw;
+    u32 retval = func(&param_1, (u32)Param(1)).raw;
     Core::CurrentArmInterface().SetReg(1, param_1);
     FuncReturn(retval);
 }
@@ -52,39 +54,39 @@ void SvcWrap() {
 template <ResultCode func(u32*, u64)>
 void SvcWrap() {
     u32 param_1 = 0;
-    u32 retval = func(&param_1, PARAM(1)).raw;
+    u32 retval = func(&param_1, Param(1)).raw;
     Core::CurrentArmInterface().SetReg(1, param_1);
     FuncReturn(retval);
 }
 
 template <ResultCode func(u64, s32)>
 void SvcWrap() {
-    FuncReturn(func(PARAM(0), (s32)PARAM(1)).raw);
+    FuncReturn(func(Param(0), (s32)Param(1)).raw);
 }
 
 template <ResultCode func(u64*, u64)>
 void SvcWrap() {
     u64 param_1 = 0;
-    u32 retval = func(&param_1, PARAM(1)).raw;
+    u32 retval = func(&param_1, Param(1)).raw;
     Core::CurrentArmInterface().SetReg(1, param_1);
     FuncReturn(retval);
 }
 
 template <ResultCode func(u32, u64)>
 void SvcWrap() {
-    FuncReturn(func((u32)(PARAM(0) & 0xFFFFFFFF), PARAM(1)).raw);
+    FuncReturn(func((u32)(Param(0) & 0xFFFFFFFF), Param(1)).raw);
 }
 
 template <ResultCode func(u32, u32, u64)>
 void SvcWrap() {
-    FuncReturn(func((u32)(PARAM(0) & 0xFFFFFFFF), (u32)(PARAM(1) & 0xFFFFFFFF), PARAM(2)).raw);
+    FuncReturn(func((u32)(Param(0) & 0xFFFFFFFF), (u32)(Param(1) & 0xFFFFFFFF), Param(2)).raw);
 }
 
 template <ResultCode func(u32, u32*, u64*)>
 void SvcWrap() {
     u32 param_1 = 0;
     u64 param_2 = 0;
-    ResultCode retval = func((u32)(PARAM(2) & 0xFFFFFFFF), &param_1, &param_2);
+    ResultCode retval = func((u32)(Param(2) & 0xFFFFFFFF), &param_1, &param_2);
     Core::CurrentArmInterface().SetReg(1, param_1);
     Core::CurrentArmInterface().SetReg(2, param_2);
     FuncReturn(retval.raw);
@@ -93,46 +95,46 @@ void SvcWrap() {
 template <ResultCode func(u64, u64, u32, u32)>
 void SvcWrap() {
     FuncReturn(
-        func(PARAM(0), PARAM(1), (u32)(PARAM(3) & 0xFFFFFFFF), (u32)(PARAM(3) & 0xFFFFFFFF)).raw);
+        func(Param(0), Param(1), (u32)(Param(3) & 0xFFFFFFFF), (u32)(Param(3) & 0xFFFFFFFF)).raw);
 }
 
 template <ResultCode func(u32, u64, u32)>
 void SvcWrap() {
-    FuncReturn(func((u32)PARAM(0), PARAM(1), (u32)PARAM(2)).raw);
+    FuncReturn(func((u32)Param(0), Param(1), (u32)Param(2)).raw);
 }
 
 template <ResultCode func(u64, u64, u64)>
 void SvcWrap() {
-    FuncReturn(func(PARAM(0), PARAM(1), PARAM(2)).raw);
+    FuncReturn(func(Param(0), Param(1), Param(2)).raw);
 }
 
 template <ResultCode func(u32, u64, u64, u32)>
 void SvcWrap() {
-    FuncReturn(func((u32)PARAM(0), PARAM(1), PARAM(2), (u32)PARAM(3)).raw);
+    FuncReturn(func((u32)Param(0), Param(1), Param(2), (u32)Param(3)).raw);
 }
 
 template <ResultCode func(u32, u64, u64)>
 void SvcWrap() {
-    FuncReturn(func((u32)PARAM(0), PARAM(1), PARAM(2)).raw);
+    FuncReturn(func((u32)Param(0), Param(1), Param(2)).raw);
 }
 
 template <ResultCode func(u32*, u64, u64, s64)>
 void SvcWrap() {
     u32 param_1 = 0;
-    ResultCode retval = func(&param_1, PARAM(1), (u32)(PARAM(2) & 0xFFFFFFFF), (s64)PARAM(3));
+    ResultCode retval = func(&param_1, Param(1), (u32)(Param(2) & 0xFFFFFFFF), (s64)Param(3));
     Core::CurrentArmInterface().SetReg(1, param_1);
     FuncReturn(retval.raw);
 }
 
 template <ResultCode func(u64, u64, u32, s64)>
 void SvcWrap() {
-    FuncReturn(func(PARAM(0), PARAM(1), (u32)PARAM(2), (s64)PARAM(3)).raw);
+    FuncReturn(func(Param(0), Param(1), (u32)Param(2), (s64)Param(3)).raw);
 }
 
 template <ResultCode func(u64*, u64, u64, u64)>
 void SvcWrap() {
     u64 param_1 = 0;
-    u32 retval = func(&param_1, PARAM(1), PARAM(2), PARAM(3)).raw;
+    u32 retval = func(&param_1, Param(1), Param(2), Param(3)).raw;
     Core::CurrentArmInterface().SetReg(1, param_1);
     FuncReturn(retval);
 }
@@ -141,7 +143,7 @@ template <ResultCode func(u32*, u64, u64, u64, u32, s32)>
 void SvcWrap() {
     u32 param_1 = 0;
     u32 retval =
-        func(&param_1, PARAM(1), PARAM(2), PARAM(3), (u32)PARAM(4), (s32)(PARAM(5) & 0xFFFFFFFF))
+        func(&param_1, Param(1), Param(2), Param(3), (u32)Param(4), (s32)(Param(5) & 0xFFFFFFFF))
             .raw;
     Core::CurrentArmInterface().SetReg(1, param_1);
     FuncReturn(retval);
@@ -151,13 +153,13 @@ template <ResultCode func(MemoryInfo*, PageInfo*, u64)>
 void SvcWrap() {
     MemoryInfo memory_info = {};
     PageInfo page_info = {};
-    u32 retval = func(&memory_info, &page_info, PARAM(2)).raw;
+    u32 retval = func(&memory_info, &page_info, Param(2)).raw;
 
-    Memory::Write64(PARAM(0), memory_info.base_address);
-    Memory::Write64(PARAM(0) + 8, memory_info.size);
-    Memory::Write32(PARAM(0) + 16, memory_info.type);
-    Memory::Write32(PARAM(0) + 20, memory_info.attributes);
-    Memory::Write32(PARAM(0) + 24, memory_info.permission);
+    Memory::Write64(Param(0), memory_info.base_address);
+    Memory::Write64(Param(0) + 8, memory_info.size);
+    Memory::Write32(Param(0) + 16, memory_info.type);
+    Memory::Write32(Param(0) + 20, memory_info.attributes);
+    Memory::Write32(Param(0) + 24, memory_info.permission);
 
     FuncReturn(retval);
 }
@@ -165,7 +167,7 @@ void SvcWrap() {
 template <ResultCode func(u32*, u64, u64, u32)>
 void SvcWrap() {
     u32 param_1 = 0;
-    u32 retval = func(&param_1, PARAM(1), PARAM(2), (u32)(PARAM(3) & 0xFFFFFFFF)).raw;
+    u32 retval = func(&param_1, Param(1), Param(2), (u32)(Param(3) & 0xFFFFFFFF)).raw;
     Core::CurrentArmInterface().SetReg(1, param_1);
     FuncReturn(retval);
 }
@@ -174,7 +176,7 @@ template <ResultCode func(Handle*, u64, u32, u32)>
 void SvcWrap() {
     u32 param_1 = 0;
     u32 retval =
-        func(&param_1, PARAM(1), (u32)(PARAM(2) & 0xFFFFFFFF), (u32)(PARAM(3) & 0xFFFFFFFF)).raw;
+        func(&param_1, Param(1), (u32)(Param(2) & 0xFFFFFFFF), (u32)(Param(3) & 0xFFFFFFFF)).raw;
     Core::CurrentArmInterface().SetReg(1, param_1);
     FuncReturn(retval);
 }
@@ -182,14 +184,14 @@ void SvcWrap() {
 template <ResultCode func(u64, u32, s32, s64)>
 void SvcWrap() {
     FuncReturn(
-        func(PARAM(0), (u32)(PARAM(1) & 0xFFFFFFFF), (s32)(PARAM(2) & 0xFFFFFFFF), (s64)PARAM(3))
+        func(Param(0), (u32)(Param(1) & 0xFFFFFFFF), (s32)(Param(2) & 0xFFFFFFFF), (s64)Param(3))
             .raw);
 }
 
 template <ResultCode func(u64, u32, s32, s32)>
 void SvcWrap() {
-    FuncReturn(func(PARAM(0), (u32)(PARAM(1) & 0xFFFFFFFF), (s32)(PARAM(2) & 0xFFFFFFFF),
-                    (s32)(PARAM(3) & 0xFFFFFFFF))
+    FuncReturn(func(Param(0), (u32)(Param(1) & 0xFFFFFFFF), (s32)(Param(2) & 0xFFFFFFFF),
+                    (s32)(Param(3) & 0xFFFFFFFF))
                    .raw);
 }
 
@@ -219,20 +221,17 @@ void SvcWrap() {
 
 template <void func(s64)>
 void SvcWrap() {
-    func((s64)PARAM(0));
+    func((s64)Param(0));
 }
 
 template <void func(u64, u64 len)>
 void SvcWrap() {
-    func(PARAM(0), PARAM(1));
+    func(Param(0), Param(1));
 }
 
 template <void func(u64, u64, u64)>
 void SvcWrap() {
-    func(PARAM(0), PARAM(1), PARAM(2));
+    func(Param(0), Param(1), Param(2));
 }
 
-#undef PARAM
-#undef FuncReturn
-
 } // namespace Kernel

src/core/hle/kernel/thread.cpp

@@ -217,8 +217,8 @@ static void ResetThreadContext(Core::ARM_Interface::ThreadContext& context, VAdd
     context.cpu_registers[0] = arg;
     context.pc = entry_point;
     context.sp = stack_top;
-    context.cpsr = 0;
-    context.fpscr = 0;
+    context.pstate = 0;
+    context.fpcr = 0;
 }
 
 ResultVal<SharedPtr<Thread>> Thread::Create(KernelCore& kernel, std::string name, VAddr entry_point,
@@ -311,13 +311,13 @@ void Thread::BoostPriority(u32 priority) {
 }
 
 SharedPtr<Thread> SetupMainThread(KernelCore& kernel, VAddr entry_point, u32 priority,
-                                  SharedPtr<Process> owner_process) {
+                                  Process& owner_process) {
     // Setup page table so we can write to memory
-    SetCurrentPageTable(&Core::CurrentProcess()->vm_manager.page_table);
+    SetCurrentPageTable(&owner_process.vm_manager.page_table);
 
     // Initialize new "main" thread
     auto thread_res = Thread::Create(kernel, "main", entry_point, priority, 0, THREADPROCESSORID_0,
-                                     Memory::STACK_AREA_VADDR_END, std::move(owner_process));
+                                     Memory::STACK_AREA_VADDR_END, &owner_process);
 
     SharedPtr<Thread> thread = std::move(thread_res).Unwrap();
 

src/core/hle/kernel/thread.h

@@ -281,7 +281,7 @@ private:
  * @return A shared pointer to the main thread
  */
 SharedPtr<Thread> SetupMainThread(KernelCore& kernel, VAddr entry_point, u32 priority,
-                                  SharedPtr<Process> owner_process);
+                                  Process& owner_process);
 
 /**
  * Gets the current thread

src/core/hle/service/acc/acc.cpp

@@ -34,7 +34,7 @@ public:
         static const FunctionInfo functions[] = {
             {0, &IProfile::Get, "Get"},
             {1, &IProfile::GetBase, "GetBase"},
-            {10, nullptr, "GetImageSize"},
+            {10, &IProfile::GetImageSize, "GetImageSize"},
             {11, &IProfile::LoadImage, "LoadImage"},
         };
         RegisterHandlers(functions);
@@ -93,6 +93,14 @@ private:
         rb.Push<u32>(jpeg_size);
     }
 
+    void GetImageSize(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_ACC, "(STUBBED) called");
+        constexpr u32 jpeg_size = 107;
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(RESULT_SUCCESS);
+        rb.Push<u32>(jpeg_size);
+    }
+
     const ProfileManager& profile_manager;
     UUID user_id; ///< The user id this profile refers to.
 };
@@ -122,11 +130,10 @@ private:
 
     void GetAccountId(Kernel::HLERequestContext& ctx) {
         LOG_WARNING(Service_ACC, "(STUBBED) called");
-        // TODO(Subv): Find out what this actually does and implement it. Stub it as an error for
-        // now since we do not implement NNID. Returning a bogus id here will cause games to send
-        // invalid IPC requests after ListOpenUsers is called.
-        IPC::ResponseBuilder rb{ctx, 2};
-        rb.Push(ResultCode(-1));
+        // Should return a nintendo account ID
+        IPC::ResponseBuilder rb{ctx, 4};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushRaw<u64>(1);
     }
 };
 

src/core/hle/service/acc/profile_manager.cpp

@@ -25,7 +25,7 @@ const UUID& UUID::Generate() {
 ProfileManager::ProfileManager() {
     // TODO(ogniK): Create the default user we have for now until loading/saving users is added
     auto user_uuid = UUID{1, 0};
-    CreateNewUser(user_uuid, Settings::values.username);
+    ASSERT(CreateNewUser(user_uuid, Settings::values.username).IsSuccess());
     OpenUser(user_uuid);
 }
 
@@ -91,7 +91,8 @@ ResultCode ProfileManager::CreateNewUser(UUID uuid, const ProfileUsername& usern
 /// specifically by allowing an std::string for the username. This is required specifically since
 /// we're loading a string straight from the config
 ResultCode ProfileManager::CreateNewUser(UUID uuid, const std::string& username) {
-    ProfileUsername username_output;
+    ProfileUsername username_output{};
+
     if (username.size() > username_output.size()) {
         std::copy_n(username.begin(), username_output.size(), username_output.begin());
     } else {

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

@@ -20,6 +20,7 @@
 #include "core/hle/service/nvflinger/nvflinger.h"
 #include "core/hle/service/pm/pm.h"
 #include "core/hle/service/set/set.h"
+#include "core/hle/service/vi/vi.h"
 #include "core/settings.h"
 
 namespace Service::AM {
@@ -334,7 +335,7 @@ ICommonStateGetter::ICommonStateGetter() : ServiceFramework("ICommonStateGetter"
         {51, nullptr, "SetVrModeEnabled"},
         {52, nullptr, "SwitchLcdBacklight"},
         {55, nullptr, "IsInControllerFirmwareUpdateSection"},
-        {60, nullptr, "GetDefaultDisplayResolution"},
+        {60, &ICommonStateGetter::GetDefaultDisplayResolution, "GetDefaultDisplayResolution"},
         {61, &ICommonStateGetter::GetDefaultDisplayResolutionChangeEvent,
          "GetDefaultDisplayResolutionChangeEvent"},
         {62, nullptr, "GetHdcpAuthenticationState"},
@@ -393,6 +394,21 @@ void ICommonStateGetter::GetDefaultDisplayResolutionChangeEvent(Kernel::HLEReque
     LOG_WARNING(Service_AM, "(STUBBED) called");
 }
 
+void ICommonStateGetter::GetDefaultDisplayResolution(Kernel::HLERequestContext& ctx) {
+    IPC::ResponseBuilder rb{ctx, 4};
+    rb.Push(RESULT_SUCCESS);
+
+    if (Settings::values.use_docked_mode) {
+        rb.Push(static_cast<u32>(Service::VI::DisplayResolution::DockedWidth));
+        rb.Push(static_cast<u32>(Service::VI::DisplayResolution::DockedHeight));
+    } else {
+        rb.Push(static_cast<u32>(Service::VI::DisplayResolution::UndockedWidth));
+        rb.Push(static_cast<u32>(Service::VI::DisplayResolution::UndockedHeight));
+    }
+
+    LOG_DEBUG(Service_AM, "called");
+}
+
 void ICommonStateGetter::GetOperationMode(Kernel::HLERequestContext& ctx) {
     const bool use_docked_mode{Settings::values.use_docked_mode};
     IPC::ResponseBuilder rb{ctx, 3};
@@ -446,7 +462,7 @@ private:
 
         std::memcpy(&buffer[offset], data.data(), data.size());
 
-        IPC::ResponseBuilder rb{rp.MakeBuilder(2, 0, 0)};
+        IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
 
         LOG_DEBUG(Service_AM, "called, offset={}", offset);
@@ -462,7 +478,7 @@ private:
 
         ctx.WriteBuffer(buffer.data() + offset, size);
 
-        IPC::ResponseBuilder rb{rp.MakeBuilder(2, 0, 0)};
+        IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
 
         LOG_DEBUG(Service_AM, "called, offset={}", offset);
@@ -552,7 +568,7 @@ private:
         IPC::RequestParser rp{ctx};
         storage_stack.push(rp.PopIpcInterface<AM::IStorage>());
 
-        IPC::ResponseBuilder rb{rp.MakeBuilder(2, 0, 0)};
+        IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
 
         LOG_DEBUG(Service_AM, "called");
@@ -600,7 +616,7 @@ void ILibraryAppletCreator::CreateStorage(Kernel::HLERequestContext& ctx) {
     const u64 size{rp.Pop<u64>()};
     std::vector<u8> buffer(size);
 
-    IPC::ResponseBuilder rb{rp.MakeBuilder(2, 0, 1)};
+    IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);
     rb.PushIpcInterface<AM::IStorage>(std::move(buffer));
 

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

@@ -123,6 +123,7 @@ private:
     void GetOperationMode(Kernel::HLERequestContext& ctx);
     void GetPerformanceMode(Kernel::HLERequestContext& ctx);
     void GetBootMode(Kernel::HLERequestContext& ctx);
+    void GetDefaultDisplayResolution(Kernel::HLERequestContext& ctx);
 
     Kernel::SharedPtr<Kernel::Event> event;
 };

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

@@ -190,7 +190,7 @@ void AudOutU::ListAudioOutsImpl(Kernel::HLERequestContext& ctx) {
 
     ctx.WriteBuffer(DefaultDevice);
 
-    IPC::ResponseBuilder rb = rp.MakeBuilder(3, 0, 0);
+    IPC::ResponseBuilder rb{ctx, 3};
 
     rb.Push(RESULT_SUCCESS);
     rb.Push<u32>(1); // Amount of audio devices

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

@@ -137,7 +137,7 @@ private:
         constexpr std::array<char, 15> audio_interface{{"AudioInterface"}};
         ctx.WriteBuffer(audio_interface);
 
-        IPC::ResponseBuilder rb = rp.MakeBuilder(3, 0, 0);
+        IPC::ResponseBuilder rb{ctx, 3};
         rb.Push(RESULT_SUCCESS);
         rb.Push<u32>(1);
     }
@@ -151,7 +151,7 @@ private:
         auto file_buffer = ctx.ReadBuffer();
         auto end = std::find(file_buffer.begin(), file_buffer.end(), '\0');
 
-        IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0);
+        IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
     }
 
@@ -162,7 +162,7 @@ private:
         constexpr std::array<char, 12> audio_interface{{"AudioDevice"}};
         ctx.WriteBuffer(audio_interface);
 
-        IPC::ResponseBuilder rb = rp.MakeBuilder(3, 0, 0);
+        IPC::ResponseBuilder rb{ctx, 3};
         rb.Push(RESULT_SUCCESS);
         rb.Push<u32>(1);
     }

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

@@ -197,7 +197,7 @@ ResultVal<FileSys::VirtualDir> VfsDirectoryServiceWrapper::OpenDirectory(const s
     auto dir = GetDirectoryRelativeWrapped(backing, path);
     if (dir == nullptr) {
         // TODO(DarkLordZach): Find a better error code for this
-        return ResultCode(-1);
+        return FileSys::ERROR_PATH_NOT_FOUND;
     }
     return MakeResult(dir);
 }

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

@@ -313,7 +313,7 @@ public:
             {64, nullptr, "DeactivateJoySixAxisSensor"},
             {65, nullptr, "GetJoySixAxisSensorLifoHandle"},
             {66, &Hid::StartSixAxisSensor, "StartSixAxisSensor"},
-            {67, nullptr, "StopSixAxisSensor"},
+            {67, &Hid::StopSixAxisSensor, "StopSixAxisSensor"},
             {68, nullptr, "IsSixAxisSensorFusionEnabled"},
             {69, nullptr, "EnableSixAxisSensorFusion"},
             {70, nullptr, "SetSixAxisSensorFusionParameters"},
@@ -329,7 +329,7 @@ public:
             {80, nullptr, "GetGyroscopeZeroDriftMode"},
             {81, nullptr, "ResetGyroscopeZeroDriftMode"},
             {82, &Hid::IsSixAxisSensorAtRest, "IsSixAxisSensorAtRest"},
-            {91, nullptr, "ActivateGesture"},
+            {91, &Hid::ActivateGesture, "ActivateGesture"},
             {100, &Hid::SetSupportedNpadStyleSet, "SetSupportedNpadStyleSet"},
             {101, &Hid::GetSupportedNpadStyleSet, "GetSupportedNpadStyleSet"},
             {102, &Hid::SetSupportedNpadIdType, "SetSupportedNpadIdType"},
@@ -338,7 +338,7 @@ public:
             {106, &Hid::AcquireNpadStyleSetUpdateEventHandle, "AcquireNpadStyleSetUpdateEventHandle"},
             {107, &Hid::DisconnectNpad, "DisconnectNpad"},
             {108, &Hid::GetPlayerLedPattern, "GetPlayerLedPattern"},
-            {109, nullptr, "ActivateNpadWithRevision"},
+            {109, &Hid::ActivateNpadWithRevision, "ActivateNpadWithRevision"},
             {120, &Hid::SetNpadJoyHoldType, "SetNpadJoyHoldType"},
             {121, &Hid::GetNpadJoyHoldType, "GetNpadJoyHoldType"},
             {122, &Hid::SetNpadJoyAssignmentModeSingleByDefault, "SetNpadJoyAssignmentModeSingleByDefault"},
@@ -364,8 +364,8 @@ public:
             {208, nullptr, "GetActualVibrationGcErmCommand"},
             {209, nullptr, "BeginPermitVibrationSession"},
             {210, nullptr, "EndPermitVibrationSession"},
-            {300, nullptr, "ActivateConsoleSixAxisSensor"},
-            {301, nullptr, "StartConsoleSixAxisSensor"},
+            {300, &Hid::ActivateConsoleSixAxisSensor, "ActivateConsoleSixAxisSensor"},
+            {301, &Hid::StartConsoleSixAxisSensor, "StartConsoleSixAxisSensor"},
             {302, nullptr, "StopConsoleSixAxisSensor"},
             {303, nullptr, "ActivateSevenSixAxisSensor"},
             {304, nullptr, "StartSevenSixAxisSensor"},
@@ -579,6 +579,36 @@ private:
         rb.Push(RESULT_SUCCESS);
         LOG_WARNING(Service_HID, "(STUBBED) called");
     }
+
+    void ActivateConsoleSixAxisSensor(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+        LOG_WARNING(Service_HID, "(STUBBED) called");
+    }
+
+    void StartConsoleSixAxisSensor(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+        LOG_WARNING(Service_HID, "(STUBBED) called");
+    }
+
+    void StopSixAxisSensor(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+        LOG_WARNING(Service_HID, "(STUBBED) called");
+    }
+
+    void ActivateGesture(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+        LOG_WARNING(Service_HID, "(STUBBED) called");
+    }
+
+    void ActivateNpadWithRevision(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+        LOG_WARNING(Service_HID, "(STUBBED) called");
+    }
 };
 
 class HidDbg final : public ServiceFramework<HidDbg> {

src/core/hle/service/nifm/nifm.cpp

@@ -31,7 +31,7 @@ public:
             {1, &IRequest::GetResult, "GetResult"},
             {2, &IRequest::GetSystemEventReadableHandles, "GetSystemEventReadableHandles"},
             {3, &IRequest::Cancel, "Cancel"},
-            {4, nullptr, "Submit"},
+            {4, &IRequest::Submit, "Submit"},
             {5, nullptr, "SetRequirement"},
             {6, nullptr, "SetRequirementPreset"},
             {8, nullptr, "SetPriority"},
@@ -61,6 +61,12 @@ public:
     }
 
 private:
+    void Submit(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_NIFM, "(STUBBED) called");
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+    }
+
     void GetRequestState(Kernel::HLERequestContext& ctx) {
         LOG_WARNING(Service_NIFM, "(STUBBED) called");
         IPC::ResponseBuilder rb{ctx, 3};
@@ -114,10 +120,11 @@ public:
 
 private:
     void GetClientId(Kernel::HLERequestContext& ctx) {
+        static constexpr u32 client_id = 1;
         LOG_WARNING(Service_NIFM, "(STUBBED) called");
         IPC::ResponseBuilder rb{ctx, 4};
         rb.Push(RESULT_SUCCESS);
-        rb.Push<u64>(0);
+        rb.Push<u64>(client_id); // Client ID needs to be non zero otherwise it's considered invalid
     }
     void CreateScanRequest(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
@@ -141,10 +148,16 @@ private:
         rb.Push(RESULT_SUCCESS);
     }
     void CreateTemporaryNetworkProfile(Kernel::HLERequestContext& ctx) {
-        IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+        ASSERT_MSG(ctx.GetReadBufferSize() == 0x17c, "NetworkProfileData is not the correct size");
+        u128 uuid{};
+        auto buffer = ctx.ReadBuffer();
+        std::memcpy(&uuid, buffer.data() + 8, sizeof(u128));
+
+        IPC::ResponseBuilder rb{ctx, 6, 0, 1};
 
         rb.Push(RESULT_SUCCESS);
         rb.PushIpcInterface<INetworkProfile>();
+        rb.PushRaw<u128>(uuid);
 
         LOG_DEBUG(Service_NIFM, "called");
     }

src/core/hle/service/nim/nim.cpp

@@ -2,6 +2,10 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <chrono>
+#include <ctime>
+#include "core/hle/ipc_helpers.h"
+#include "core/hle/kernel/event.h"
 #include "core/hle/service/nim/nim.h"
 #include "core/hle/service/service.h"
 #include "core/hle/service/sm/sm.h"
@@ -100,19 +104,111 @@ public:
     }
 };
 
+class IEnsureNetworkClockAvailabilityService final
+    : public ServiceFramework<IEnsureNetworkClockAvailabilityService> {
+public:
+    IEnsureNetworkClockAvailabilityService()
+        : ServiceFramework("IEnsureNetworkClockAvailabilityService") {
+        static const FunctionInfo functions[] = {
+            {0, &IEnsureNetworkClockAvailabilityService::StartTask, "StartTask"},
+            {1, &IEnsureNetworkClockAvailabilityService::GetFinishNotificationEvent,
+             "GetFinishNotificationEvent"},
+            {2, &IEnsureNetworkClockAvailabilityService::GetResult, "GetResult"},
+            {3, &IEnsureNetworkClockAvailabilityService::Cancel, "Cancel"},
+            {4, &IEnsureNetworkClockAvailabilityService::IsProcessing, "IsProcessing"},
+            {5, &IEnsureNetworkClockAvailabilityService::GetServerTime, "GetServerTime"},
+        };
+        RegisterHandlers(functions);
+
+        auto& kernel = Core::System::GetInstance().Kernel();
+        finished_event =
+            Kernel::Event::Create(kernel, Kernel::ResetType::OneShot,
+                                  "IEnsureNetworkClockAvailabilityService:FinishEvent");
+    }
+
+private:
+    Kernel::SharedPtr<Kernel::Event> finished_event;
+
+    void StartTask(Kernel::HLERequestContext& ctx) {
+        // No need to connect to the internet, just finish the task straight away.
+        finished_event->Signal();
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+        LOG_DEBUG(Service_NIM, "called");
+    }
+
+    void GetFinishNotificationEvent(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushCopyObjects(finished_event);
+        LOG_DEBUG(Service_NIM, "called");
+    }
+
+    void GetResult(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+        LOG_DEBUG(Service_NIM, "called");
+    }
+
+    void Cancel(Kernel::HLERequestContext& ctx) {
+        finished_event->Clear();
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+        LOG_DEBUG(Service_NIM, "called");
+    }
+
+    void IsProcessing(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushRaw<u32>(0); // We instantly process the request
+        LOG_DEBUG(Service_NIM, "called");
+    }
+
+    void GetServerTime(Kernel::HLERequestContext& ctx) {
+        const s64 server_time{std::chrono::duration_cast<std::chrono::seconds>(
+                                  std::chrono::system_clock::now().time_since_epoch())
+                                  .count()};
+        IPC::ResponseBuilder rb{ctx, 4};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushRaw<s64>(server_time);
+        LOG_DEBUG(Service_NIM, "called");
+    }
+};
+
 class NTC final : public ServiceFramework<NTC> {
 public:
     explicit NTC() : ServiceFramework{"ntc"} {
         // clang-format off
         static const FunctionInfo functions[] = {
-            {0, nullptr, "OpenEnsureNetworkClockAvailabilityService"},
-            {100, nullptr, "SuspendAutonomicTimeCorrection"},
-            {101, nullptr, "ResumeAutonomicTimeCorrection"},
+            {0, &NTC::OpenEnsureNetworkClockAvailabilityService, "OpenEnsureNetworkClockAvailabilityService"},
+            {100, &NTC::SuspendAutonomicTimeCorrection, "SuspendAutonomicTimeCorrection"},
+            {101, &NTC::ResumeAutonomicTimeCorrection, "ResumeAutonomicTimeCorrection"},
         };
         // clang-format on
 
         RegisterHandlers(functions);
     }
+
+private:
+    void OpenEnsureNetworkClockAvailabilityService(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushIpcInterface<IEnsureNetworkClockAvailabilityService>();
+        LOG_DEBUG(Service_NIM, "called");
+    }
+
+    // TODO(ogniK): Do we need these?
+    void SuspendAutonomicTimeCorrection(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+        LOG_WARNING(Service_NIM, "(STUBBED) called");
+    }
+
+    void ResumeAutonomicTimeCorrection(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+        LOG_WARNING(Service_NIM, "(STUBBED) called");
+    }
 };
 
 void InstallInterfaces(SM::ServiceManager& sm) {

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

@@ -108,7 +108,7 @@ void SM::GetService(Kernel::HLERequestContext& ctx) {
 
     auto client_port = service_manager->GetServicePort(name);
     if (client_port.Failed()) {
-        IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0);
+        IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(client_port.Code());
         LOG_ERROR(Service_SM, "called service={} -> error 0x{:08X}", name, client_port.Code().raw);
         if (name.length() == 0)
@@ -121,8 +121,7 @@ void SM::GetService(Kernel::HLERequestContext& ctx) {
     ASSERT(session.Succeeded());
     if (session.Succeeded()) {
         LOG_DEBUG(Service_SM, "called service={} -> session={}", name, (*session)->GetObjectId());
-        IPC::ResponseBuilder rb =
-            rp.MakeBuilder(2, 0, 1, IPC::ResponseBuilder::Flags::AlwaysMoveHandles);
+        IPC::ResponseBuilder rb{ctx, 2, 0, 1, IPC::ResponseBuilder::Flags::AlwaysMoveHandles};
         rb.Push(session.Code());
         rb.PushMoveObjects(std::move(session).Unwrap());
     }

src/core/hle/service/ssl/ssl.cpp

@@ -71,7 +71,7 @@ private:
         LOG_WARNING(Service_SSL, "(STUBBED) called");
         IPC::RequestParser rp{ctx};
 
-        IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0);
+        IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
     }
 

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

@@ -650,7 +650,7 @@ private:
         u64 layer_id = rp.Pop<u64>();
         u64 z_value = rp.Pop<u64>();
 
-        IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0);
+        IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
     }
 
@@ -658,7 +658,7 @@ private:
         IPC::RequestParser rp{ctx};
         u64 layer_id = rp.Pop<u64>();
         bool visibility = rp.Pop<bool>();
-        IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0);
+        IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
         LOG_WARNING(Service_VI, "(STUBBED) called, layer_id=0x{:08X}, visibility={}", layer_id,
                     visibility);
@@ -747,7 +747,7 @@ private:
         IPC::RequestParser rp{ctx};
         u64 display = rp.Pop<u64>();
 
-        IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0);
+        IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
     }
 
@@ -761,7 +761,7 @@ private:
 
         u64 layer_id = nv_flinger->CreateLayer(display);
 
-        IPC::ResponseBuilder rb = rp.MakeBuilder(4, 0, 0);
+        IPC::ResponseBuilder rb{ctx, 4};
         rb.Push(RESULT_SUCCESS);
         rb.Push(layer_id);
     }
@@ -772,7 +772,7 @@ private:
         u32 stack = rp.Pop<u32>();
         u64 layer_id = rp.Pop<u64>();
 
-        IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0);
+        IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
     }
 
@@ -780,7 +780,7 @@ private:
         IPC::RequestParser rp{ctx};
         u64 layer_id = rp.Pop<u64>();
         bool visibility = rp.Pop<bool>();
-        IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0);
+        IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
         LOG_WARNING(Service_VI, "(STUBBED) called, layer_id=0x{:X}, visibility={}", layer_id,
                     visibility);
@@ -837,7 +837,7 @@ private:
 
         ASSERT_MSG(name == "Default", "Non-default displays aren't supported yet");
 
-        IPC::ResponseBuilder rb = rp.MakeBuilder(4, 0, 0);
+        IPC::ResponseBuilder rb{ctx, 4};
         rb.Push(RESULT_SUCCESS);
         rb.Push<u64>(nv_flinger->OpenDisplay(name));
     }
@@ -847,7 +847,7 @@ private:
         IPC::RequestParser rp{ctx};
         u64 display_id = rp.Pop<u64>();
 
-        IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0);
+        IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
     }
 
@@ -856,7 +856,7 @@ private:
         IPC::RequestParser rp{ctx};
         u64 display_id = rp.Pop<u64>();
 
-        IPC::ResponseBuilder rb = rp.MakeBuilder(6, 0, 0);
+        IPC::ResponseBuilder rb{ctx, 6};
         rb.Push(RESULT_SUCCESS);
 
         if (Settings::values.use_docked_mode) {
@@ -874,7 +874,7 @@ private:
         u32 scaling_mode = rp.Pop<u32>();
         u64 unknown = rp.Pop<u64>();
 
-        IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0);
+        IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
     }
 
@@ -882,7 +882,7 @@ private:
         IPC::RequestParser rp{ctx};
         DisplayInfo display_info;
         ctx.WriteBuffer(&display_info, sizeof(DisplayInfo));
-        IPC::ResponseBuilder rb = rp.MakeBuilder(4, 0, 0);
+        IPC::ResponseBuilder rb{ctx, 4};
         rb.Push(RESULT_SUCCESS);
         rb.Push<u64>(1);
         LOG_WARNING(Service_VI, "(STUBBED) called");
@@ -903,7 +903,7 @@ private:
         u32 buffer_queue_id = nv_flinger->GetBufferQueueId(display_id, layer_id);
 
         NativeWindow native_window{buffer_queue_id};
-        IPC::ResponseBuilder rb = rp.MakeBuilder(4, 0, 0);
+        IPC::ResponseBuilder rb{ctx, 4};
         rb.Push(RESULT_SUCCESS);
         rb.Push<u64>(ctx.WriteBuffer(native_window.Serialize()));
     }
@@ -922,7 +922,7 @@ private:
         u32 buffer_queue_id = nv_flinger->GetBufferQueueId(display_id, layer_id);
 
         NativeWindow native_window{buffer_queue_id};
-        IPC::ResponseBuilder rb = rp.MakeBuilder(6, 0, 0);
+        IPC::ResponseBuilder rb{ctx, 6};
         rb.Push(RESULT_SUCCESS);
         rb.Push(layer_id);
         rb.Push<u64>(ctx.WriteBuffer(native_window.Serialize()));
@@ -934,7 +934,7 @@ private:
         IPC::RequestParser rp{ctx};
         u64 layer_id = rp.Pop<u64>();
 
-        IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0);
+        IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
     }
 
@@ -945,7 +945,7 @@ private:
 
         auto vsync_event = nv_flinger->GetVsyncEvent(display_id);
 
-        IPC::ResponseBuilder rb = rp.MakeBuilder(2, 1, 0);
+        IPC::ResponseBuilder rb{ctx, 2, 1};
         rb.Push(RESULT_SUCCESS);
         rb.PushCopyObjects(vsync_event);
     }

src/core/loader/nax.cpp

@@ -11,6 +11,20 @@
 #include "core/loader/nca.h"
 
 namespace Loader {
+namespace {
+FileType IdentifyTypeImpl(const FileSys::NAX& nax) {
+    if (nax.GetStatus() != ResultStatus::Success) {
+        return FileType::Error;
+    }
+
+    const auto nca = nax.AsNCA();
+    if (nca == nullptr || nca->GetStatus() != ResultStatus::Success) {
+        return FileType::Error;
+    }
+
+    return FileType::NAX;
+}
+} // Anonymous namespace
 
 AppLoader_NAX::AppLoader_NAX(FileSys::VirtualFile file)
     : AppLoader(file), nax(std::make_unique<FileSys::NAX>(file)),
@@ -19,14 +33,12 @@ AppLoader_NAX::AppLoader_NAX(FileSys::VirtualFile file)
 AppLoader_NAX::~AppLoader_NAX() = default;
 
 FileType AppLoader_NAX::IdentifyType(const FileSys::VirtualFile& file) {
-    FileSys::NAX nax(file);
+    const FileSys::NAX nax(file);
+    return IdentifyTypeImpl(nax);
+}
 
-    if (nax.GetStatus() == ResultStatus::Success && nax.AsNCA() != nullptr &&
-        nax.AsNCA()->GetStatus() == ResultStatus::Success) {
-        return FileType::NAX;
-    }
-
-    return FileType::Error;
+FileType AppLoader_NAX::GetFileType() {
+    return IdentifyTypeImpl(*nax);
 }
 
 ResultStatus AppLoader_NAX::Load(Kernel::SharedPtr<Kernel::Process>& process) {

src/core/loader/nax.h

@@ -31,9 +31,7 @@ public:
      */
     static FileType IdentifyType(const FileSys::VirtualFile& file);
 
-    FileType GetFileType() override {
-        return IdentifyType(file);
-    }
+    FileType GetFileType() override;
 
     ResultStatus Load(Kernel::SharedPtr<Kernel::Process>& process) override;
 

src/core/loader/nso.cpp

@@ -32,11 +32,18 @@ static_assert(sizeof(NsoSegmentHeader) == 0x10, "NsoSegmentHeader has incorrect
 
 struct NsoHeader {
     u32_le magic;
-    INSERT_PADDING_BYTES(0xc);
+    u32_le version;
+    INSERT_PADDING_WORDS(1);
+    u8 flags;
     std::array<NsoSegmentHeader, 3> segments; // Text, RoData, Data (in that order)
     u32_le bss_size;
     INSERT_PADDING_BYTES(0x1c);
     std::array<u32_le, 3> segments_compressed_size;
+
+    bool IsSegmentCompressed(size_t segment_num) const {
+        ASSERT_MSG(segment_num < 3, "Invalid segment {}", segment_num);
+        return ((flags >> segment_num) & 1);
+    }
 };
 static_assert(sizeof(NsoHeader) == 0x6c, "NsoHeader has incorrect size.");
 static_assert(std::is_trivially_copyable_v<NsoHeader>, "NsoHeader isn't trivially copyable.");
@@ -105,9 +112,11 @@ VAddr AppLoader_NSO::LoadModule(FileSys::VirtualFile file, VAddr load_base) {
     Kernel::SharedPtr<Kernel::CodeSet> codeset = Kernel::CodeSet::Create(kernel, "");
     std::vector<u8> program_image;
     for (std::size_t i = 0; i < nso_header.segments.size(); ++i) {
-        const std::vector<u8> compressed_data =
+        std::vector<u8> data =
             file->ReadBytes(nso_header.segments_compressed_size[i], nso_header.segments[i].offset);
-        std::vector<u8> data = DecompressSegment(compressed_data, nso_header.segments[i]);
+        if (nso_header.IsSegmentCompressed(i)) {
+            data = DecompressSegment(data, nso_header.segments[i]);
+        }
         program_image.resize(nso_header.segments[i].location);
         program_image.insert(program_image.end(), data.begin(), data.end());
         codeset->segments[i].addr = nso_header.segments[i].location;

src/video_core/CMakeLists.txt

@@ -14,6 +14,7 @@ add_library(video_core STATIC
     engines/maxwell_dma.cpp
     engines/maxwell_dma.h
     engines/shader_bytecode.h
+    engines/shader_header.h
     gpu.cpp
     gpu.h
     macro_interpreter.cpp

src/video_core/engines/fermi_2d.cpp

@@ -4,11 +4,13 @@
 
 #include "core/memory.h"
 #include "video_core/engines/fermi_2d.h"
+#include "video_core/rasterizer_interface.h"
 #include "video_core/textures/decoders.h"
 
 namespace Tegra::Engines {
 
-Fermi2D::Fermi2D(MemoryManager& memory_manager) : memory_manager(memory_manager) {}
+Fermi2D::Fermi2D(VideoCore::RasterizerInterface& rasterizer, MemoryManager& memory_manager)
+    : memory_manager(memory_manager), rasterizer{rasterizer} {}
 
 void Fermi2D::WriteReg(u32 method, u32 value) {
     ASSERT_MSG(method < Regs::NUM_REGS,
@@ -52,6 +54,12 @@ void Fermi2D::HandleSurfaceCopy() {
         return;
     }
 
+    rasterizer.FlushRegion(source_cpu, src_bytes_per_pixel * regs.src.width * regs.src.height);
+    // We have to invalidate the destination region to evict any outdated surfaces from the cache.
+    // We do this before actually writing the new data because the destination address might contain
+    // a dirty surface that will have to be written back to memory.
+    rasterizer.InvalidateRegion(dest_cpu, dst_bytes_per_pixel * regs.dst.width * regs.dst.height);
+
     u8* src_buffer = Memory::GetPointer(source_cpu);
     u8* dst_buffer = Memory::GetPointer(dest_cpu);
 

src/video_core/engines/fermi_2d.h

@@ -12,6 +12,10 @@
 #include "video_core/gpu.h"
 #include "video_core/memory_manager.h"
 
+namespace VideoCore {
+class RasterizerInterface;
+}
+
 namespace Tegra::Engines {
 
 #define FERMI2D_REG_INDEX(field_name)                                                              \
@@ -19,7 +23,7 @@ namespace Tegra::Engines {
 
 class Fermi2D final {
 public:
-    explicit Fermi2D(MemoryManager& memory_manager);
+    explicit Fermi2D(VideoCore::RasterizerInterface& rasterizer, MemoryManager& memory_manager);
     ~Fermi2D() = default;
 
     /// Write the value to the register identified by method.
@@ -94,6 +98,8 @@ public:
     MemoryManager& memory_manager;
 
 private:
+    VideoCore::RasterizerInterface& rasterizer;
+
     /// Performs the copy from the source surface to the destination surface as configured in the
     /// registers.
     void HandleSurfaceCopy();

src/video_core/engines/kepler_memory.cpp

@@ -5,10 +5,14 @@
 #include "common/logging/log.h"
 #include "core/memory.h"
 #include "video_core/engines/kepler_memory.h"
+#include "video_core/rasterizer_interface.h"
 
 namespace Tegra::Engines {
 
-KeplerMemory::KeplerMemory(MemoryManager& memory_manager) : memory_manager(memory_manager) {}
+KeplerMemory::KeplerMemory(VideoCore::RasterizerInterface& rasterizer,
+                           MemoryManager& memory_manager)
+    : memory_manager(memory_manager), rasterizer{rasterizer} {}
+
 KeplerMemory::~KeplerMemory() = default;
 
 void KeplerMemory::WriteReg(u32 method, u32 value) {
@@ -37,6 +41,11 @@ void KeplerMemory::ProcessData(u32 data) {
     VAddr dest_address =
         *memory_manager.GpuToCpuAddress(address + state.write_offset * sizeof(u32));
 
+    // We have to invalidate the destination region to evict any outdated surfaces from the cache.
+    // We do this before actually writing the new data because the destination address might contain
+    // a dirty surface that will have to be written back to memory.
+    rasterizer.InvalidateRegion(dest_address, sizeof(u32));
+
     Memory::Write32(dest_address, data);
 
     state.write_offset++;

src/video_core/engines/kepler_memory.h

@@ -11,6 +11,10 @@
 #include "common/common_types.h"
 #include "video_core/memory_manager.h"
 
+namespace VideoCore {
+class RasterizerInterface;
+}
+
 namespace Tegra::Engines {
 
 #define KEPLERMEMORY_REG_INDEX(field_name)                                                         \
@@ -18,7 +22,7 @@ namespace Tegra::Engines {
 
 class KeplerMemory final {
 public:
-    KeplerMemory(MemoryManager& memory_manager);
+    KeplerMemory(VideoCore::RasterizerInterface& rasterizer, MemoryManager& memory_manager);
     ~KeplerMemory();
 
     /// Write the value to the register identified by method.
@@ -72,6 +76,7 @@ public:
 
 private:
     MemoryManager& memory_manager;
+    VideoCore::RasterizerInterface& rasterizer;
 
     void ProcessData(u32 data);
 };

src/video_core/engines/maxwell_dma.cpp

@@ -4,12 +4,14 @@
 
 #include "core/memory.h"
 #include "video_core/engines/maxwell_dma.h"
+#include "video_core/rasterizer_interface.h"
 #include "video_core/textures/decoders.h"
 
 namespace Tegra {
 namespace Engines {
 
-MaxwellDMA::MaxwellDMA(MemoryManager& memory_manager) : memory_manager(memory_manager) {}
+MaxwellDMA::MaxwellDMA(VideoCore::RasterizerInterface& rasterizer, MemoryManager& memory_manager)
+    : memory_manager(memory_manager), rasterizer{rasterizer} {}
 
 void MaxwellDMA::WriteReg(u32 method, u32 value) {
     ASSERT_MSG(method < Regs::NUM_REGS,
@@ -44,36 +46,79 @@ void MaxwellDMA::HandleCopy() {
     ASSERT(regs.exec.query_mode == Regs::QueryMode::None);
     ASSERT(regs.exec.query_intr == Regs::QueryIntr::None);
     ASSERT(regs.exec.copy_mode == Regs::CopyMode::Unk2);
-    ASSERT(regs.src_params.pos_x == 0);
-    ASSERT(regs.src_params.pos_y == 0);
     ASSERT(regs.dst_params.pos_x == 0);
     ASSERT(regs.dst_params.pos_y == 0);
 
-    if (regs.exec.is_dst_linear == regs.exec.is_src_linear) {
-        std::size_t copy_size = regs.x_count;
+    if (!regs.exec.is_dst_linear && !regs.exec.is_src_linear) {
+        // If both the source and the destination are in block layout, assert.
+        UNREACHABLE_MSG("Tiled->Tiled DMA transfers are not yet implemented");
+        return;
+    }
 
+    if (regs.exec.is_dst_linear && regs.exec.is_src_linear) {
         // When the enable_2d bit is disabled, the copy is performed as if we were copying a 1D
-        // buffer of length `x_count`, otherwise we copy a 2D buffer of size (x_count, y_count).
-        if (regs.exec.enable_2d) {
-            copy_size = copy_size * regs.y_count;
+        // buffer of length `x_count`, otherwise we copy a 2D image of dimensions (x_count,
+        // y_count).
+        if (!regs.exec.enable_2d) {
+            Memory::CopyBlock(dest_cpu, source_cpu, regs.x_count);
+            return;
         }
 
-        Memory::CopyBlock(dest_cpu, source_cpu, copy_size);
+        // If both the source and the destination are in linear layout, perform a line-by-line
+        // copy. We're going to take a subrect of size (x_count, y_count) from the source
+        // rectangle. There is no need to manually flush/invalidate the regions because
+        // CopyBlock does that for us.
+        for (u32 line = 0; line < regs.y_count; ++line) {
+            const VAddr source_line = source_cpu + line * regs.src_pitch;
+            const VAddr dest_line = dest_cpu + line * regs.dst_pitch;
+            Memory::CopyBlock(dest_line, source_line, regs.x_count);
+        }
         return;
     }
 
     ASSERT(regs.exec.enable_2d == 1);
+
+    size_t copy_size = regs.x_count * regs.y_count;
+
+    const auto FlushAndInvalidate = [&](u32 src_size, u32 dst_size) {
+        // TODO(Subv): For now, manually flush the regions until we implement GPU-accelerated
+        // copying.
+        rasterizer.FlushRegion(source_cpu, src_size);
+
+        // We have to invalidate the destination region to evict any outdated surfaces from the
+        // cache. We do this before actually writing the new data because the destination address
+        // might contain a dirty surface that will have to be written back to memory.
+        rasterizer.InvalidateRegion(dest_cpu, dst_size);
+    };
+
     u8* src_buffer = Memory::GetPointer(source_cpu);
     u8* dst_buffer = Memory::GetPointer(dest_cpu);
 
     if (regs.exec.is_dst_linear && !regs.exec.is_src_linear) {
+        ASSERT(regs.src_params.size_z == 1);
         // If the input is tiled and the output is linear, deswizzle the input and copy it over.
-        Texture::CopySwizzledData(regs.src_params.size_x, regs.src_params.size_y, 1, 1, src_buffer,
-                                  dst_buffer, true, regs.src_params.BlockHeight());
+
+        u32 src_bytes_per_pixel = regs.src_pitch / regs.src_params.size_x;
+
+        FlushAndInvalidate(regs.src_pitch * regs.src_params.size_y,
+                           copy_size * src_bytes_per_pixel);
+
+        Texture::UnswizzleSubrect(regs.x_count, regs.y_count, regs.dst_pitch,
+                                  regs.src_params.size_x, src_bytes_per_pixel, source_cpu, dest_cpu,
+                                  regs.src_params.BlockHeight(), regs.src_params.pos_x,
+                                  regs.src_params.pos_y);
     } else {
+        ASSERT(regs.dst_params.size_z == 1);
+        ASSERT(regs.src_pitch == regs.x_count);
+
+        u32 src_bpp = regs.src_pitch / regs.x_count;
+
+        FlushAndInvalidate(regs.src_pitch * regs.y_count,
+                           regs.dst_params.size_x * regs.dst_params.size_y * src_bpp);
+
         // If the input is linear and the output is tiled, swizzle the input and copy it over.
-        Texture::CopySwizzledData(regs.dst_params.size_x, regs.dst_params.size_y, 1, 1, dst_buffer,
-                                  src_buffer, false, regs.dst_params.BlockHeight());
+        Texture::SwizzleSubrect(regs.x_count, regs.y_count, regs.src_pitch, regs.dst_params.size_x,
+                                src_bpp, dest_cpu, source_cpu, regs.dst_params.BlockHeight());
     }
 }
 

src/video_core/engines/maxwell_dma.h

@@ -12,11 +12,15 @@
 #include "video_core/gpu.h"
 #include "video_core/memory_manager.h"
 
+namespace VideoCore {
+class RasterizerInterface;
+}
+
 namespace Tegra::Engines {
 
 class MaxwellDMA final {
 public:
-    explicit MaxwellDMA(MemoryManager& memory_manager);
+    explicit MaxwellDMA(VideoCore::RasterizerInterface& rasterizer, MemoryManager& memory_manager);
     ~MaxwellDMA() = default;
 
     /// Write the value to the register identified by method.
@@ -129,6 +133,8 @@ public:
     MemoryManager& memory_manager;
 
 private:
+    VideoCore::RasterizerInterface& rasterizer;
+
     /// Performs the copy from the source buffer to the destination buffer as configured in the
     /// registers.
     void HandleCopy();

src/video_core/engines/shader_bytecode.h

@@ -240,6 +240,41 @@ enum class FlowCondition : u64 {
     Fcsm_Tr = 0x1C, // TODO(bunnei): What is this used for?
 };
 
+enum class ControlCode : u64 {
+    F = 0,
+    LT = 1,
+    EQ = 2,
+    LE = 3,
+    GT = 4,
+    NE = 5,
+    GE = 6,
+    Num = 7,
+    Nan = 8,
+    LTU = 9,
+    EQU = 10,
+    LEU = 11,
+    GTU = 12,
+    NEU = 13,
+    GEU = 14,
+    //
+    OFF = 16,
+    LO = 17,
+    SFF = 18,
+    LS = 19,
+    HI = 20,
+    SFT = 21,
+    HS = 22,
+    OFT = 23,
+    CSM_TA = 24,
+    CSM_TR = 25,
+    CSM_MX = 26,
+    FCSM_TA = 27,
+    FCSM_TR = 28,
+    FCSM_MX = 29,
+    RLE = 30,
+    RGT = 31,
+};
+
 enum class PredicateResultMode : u64 {
     None = 0x0,
     NotZero = 0x3,
@@ -271,6 +306,15 @@ enum class TextureProcessMode : u64 {
     LLA = 7  // Load LOD. The A is unknown, does not appear to differ with LL
 };
 
+enum class TextureMiscMode : u64 {
+    DC,
+    AOFFI, // Uses Offset
+    NDV,
+    NODEP,
+    MZ,
+    PTP,
+};
+
 enum class IpaInterpMode : u64 { Linear = 0, Perspective = 1, Flat = 2, Sc = 3 };
 enum class IpaSampleMode : u64 { Default = 0, Centroid = 1, Offset = 2 };
 
@@ -545,6 +589,15 @@ union Instruction {
         BitField<45, 2, PredOperation> op;
     } pset;
 
+    union {
+        BitField<0, 3, u64> pred0;
+        BitField<3, 3, u64> pred3;
+        BitField<8, 5, ControlCode> cc; // flag in cc
+        BitField<39, 3, u64> pred39;
+        BitField<42, 1, u64> neg_pred39;
+        BitField<45, 4, PredOperation> op; // op with pred39
+    } csetp;
+
     union {
         BitField<39, 3, u64> pred39;
         BitField<42, 1, u64> neg_pred;
@@ -590,42 +643,127 @@ union Instruction {
         BitField<28, 1, u64> array;
         BitField<29, 2, TextureType> texture_type;
         BitField<31, 4, u64> component_mask;
+        BitField<49, 1, u64> nodep_flag;
+        BitField<50, 1, u64> dc_flag;
+        BitField<54, 1, u64> aoffi_flag;
         BitField<55, 3, TextureProcessMode> process_mode;
 
         bool IsComponentEnabled(std::size_t component) const {
             return ((1ull << component) & component_mask) != 0;
         }
+
+        TextureProcessMode GetTextureProcessMode() const {
+            return process_mode;
+        }
+
+        bool UsesMiscMode(TextureMiscMode mode) const {
+            switch (mode) {
+            case TextureMiscMode::DC:
+                return dc_flag != 0;
+            case TextureMiscMode::NODEP:
+                return nodep_flag != 0;
+            case TextureMiscMode::AOFFI:
+                return aoffi_flag != 0;
+            default:
+                break;
+            }
+            return false;
+        }
     } tex;
 
     union {
         BitField<22, 6, TextureQueryType> query_type;
         BitField<31, 4, u64> component_mask;
+        BitField<49, 1, u64> nodep_flag;
+
+        bool UsesMiscMode(TextureMiscMode mode) const {
+            switch (mode) {
+            case TextureMiscMode::NODEP:
+                return nodep_flag != 0;
+            default:
+                break;
+            }
+            return false;
+        }
     } txq;
 
     union {
         BitField<28, 1, u64> array;
         BitField<29, 2, TextureType> texture_type;
         BitField<31, 4, u64> component_mask;
+        BitField<35, 1, u64> ndv_flag;
+        BitField<49, 1, u64> nodep_flag;
 
         bool IsComponentEnabled(std::size_t component) const {
             return ((1ull << component) & component_mask) != 0;
         }
+
+        bool UsesMiscMode(TextureMiscMode mode) const {
+            switch (mode) {
+            case TextureMiscMode::NDV:
+                return (ndv_flag != 0);
+            case TextureMiscMode::NODEP:
+                return (nodep_flag != 0);
+            default:
+                break;
+            }
+            return false;
+        }
     } tmml;
 
     union {
         BitField<28, 1, u64> array;
         BitField<29, 2, TextureType> texture_type;
+        BitField<35, 1, u64> ndv_flag;
+        BitField<49, 1, u64> nodep_flag;
+        BitField<50, 1, u64> dc_flag;
+        BitField<54, 2, u64> info;
         BitField<56, 2, u64> component;
+
+        bool UsesMiscMode(TextureMiscMode mode) const {
+            switch (mode) {
+            case TextureMiscMode::NDV:
+                return ndv_flag != 0;
+            case TextureMiscMode::NODEP:
+                return nodep_flag != 0;
+            case TextureMiscMode::DC:
+                return dc_flag != 0;
+            case TextureMiscMode::AOFFI:
+                return info == 1;
+            case TextureMiscMode::PTP:
+                return info == 2;
+            default:
+                break;
+            }
+            return false;
+        }
     } tld4;
 
     union {
+        BitField<49, 1, u64> nodep_flag;
+        BitField<50, 1, u64> dc_flag;
+        BitField<51, 1, u64> aoffi_flag;
         BitField<52, 2, u64> component;
+
+        bool UsesMiscMode(TextureMiscMode mode) const {
+            switch (mode) {
+            case TextureMiscMode::DC:
+                return dc_flag != 0;
+            case TextureMiscMode::NODEP:
+                return nodep_flag != 0;
+            case TextureMiscMode::AOFFI:
+                return aoffi_flag != 0;
+            default:
+                break;
+            }
+            return false;
+        }
     } tld4s;
 
     union {
         BitField<0, 8, Register> gpr0;
         BitField<28, 8, Register> gpr28;
-        BitField<49, 1, u64> nodep;
+        BitField<49, 1, u64> nodep_flag;
         BitField<50, 3, u64> component_mask_selector;
         BitField<53, 4, u64> texture_info;
 
@@ -645,6 +783,37 @@ union Instruction {
             UNREACHABLE();
         }
 
+        TextureProcessMode GetTextureProcessMode() const {
+            switch (texture_info) {
+            case 0:
+            case 2:
+            case 6:
+            case 8:
+            case 9:
+            case 11:
+                return TextureProcessMode::LZ;
+            case 3:
+            case 5:
+            case 13:
+                return TextureProcessMode::LL;
+            default:
+                break;
+            }
+            return TextureProcessMode::None;
+        }
+
+        bool UsesMiscMode(TextureMiscMode mode) const {
+            switch (mode) {
+            case TextureMiscMode::DC:
+                return (texture_info >= 4 && texture_info <= 6) || texture_info == 9;
+            case TextureMiscMode::NODEP:
+                return nodep_flag != 0;
+            default:
+                break;
+            }
+            return false;
+        }
+
         bool IsArrayTexture() const {
             // TEXS only supports Texture2D arrays.
             return texture_info >= 7 && texture_info <= 9;
@@ -673,6 +842,7 @@ union Instruction {
     } texs;
 
     union {
+        BitField<49, 1, u64> nodep_flag;
         BitField<53, 4, u64> texture_info;
 
         TextureType GetTextureType() const {
@@ -693,6 +863,26 @@ union Instruction {
             UNREACHABLE();
         }
 
+        TextureProcessMode GetTextureProcessMode() const {
+            if (texture_info == 1 || texture_info == 5 || texture_info == 12)
+                return TextureProcessMode::LL;
+            return TextureProcessMode::LZ;
+        }
+
+        bool UsesMiscMode(TextureMiscMode mode) const {
+            switch (mode) {
+            case TextureMiscMode::AOFFI:
+                return texture_info == 12 || texture_info == 4;
+            case TextureMiscMode::MZ:
+                return texture_info == 5;
+            case TextureMiscMode::NODEP:
+                return nodep_flag != 0;
+            default:
+                break;
+            }
+            return false;
+        }
+
         bool IsArrayTexture() const {
             // TEXS only supports Texture2D arrays.
             return texture_info == 8;
@@ -735,6 +925,7 @@ union Instruction {
         BitField<36, 5, u64> index;
     } cbuf36;
 
+    BitField<47, 1, u64> generates_cc;
     BitField<61, 1, u64> is_b_imm;
     BitField<60, 1, u64> is_b_gpr;
     BitField<59, 1, u64> is_c_gpr;
@@ -859,6 +1050,7 @@ public:
         ISET_IMM,
         PSETP,
         PSET,
+        CSETP,
         XMAD_IMM,
         XMAD_CR,
         XMAD_RC,
@@ -1095,6 +1287,7 @@ private:
             INST("0011011-0101----", Id::ISET_IMM, Type::IntegerSet, "ISET_IMM"),
             INST("0101000010001---", Id::PSET, Type::PredicateSetRegister, "PSET"),
             INST("0101000010010---", Id::PSETP, Type::PredicateSetPredicate, "PSETP"),
+            INST("010100001010----", Id::CSETP, Type::PredicateSetPredicate, "CSETP"),
             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"),

src/video_core/engines/shader_header.h

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

src/video_core/gpu.cpp

@@ -25,10 +25,10 @@ u32 FramebufferConfig::BytesPerPixel(PixelFormat format) {
 GPU::GPU(VideoCore::RasterizerInterface& rasterizer) {
     memory_manager = std::make_unique<Tegra::MemoryManager>();
     maxwell_3d = std::make_unique<Engines::Maxwell3D>(rasterizer, *memory_manager);
-    fermi_2d = std::make_unique<Engines::Fermi2D>(*memory_manager);
+    fermi_2d = std::make_unique<Engines::Fermi2D>(rasterizer, *memory_manager);
     maxwell_compute = std::make_unique<Engines::MaxwellCompute>();
-    maxwell_dma = std::make_unique<Engines::MaxwellDMA>(*memory_manager);
-    kepler_memory = std::make_unique<Engines::KeplerMemory>(*memory_manager);
+    maxwell_dma = std::make_unique<Engines::MaxwellDMA>(rasterizer, *memory_manager);
+    kepler_memory = std::make_unique<Engines::KeplerMemory>(rasterizer, *memory_manager);
 }
 
 GPU::~GPU() = default;

src/video_core/rasterizer_cache.h

@@ -17,6 +17,22 @@
 template <class T>
 class RasterizerCache : NonCopyable {
 public:
+    /// Write any cached resources overlapping the region back to memory (if dirty)
+    void FlushRegion(Tegra::GPUVAddr addr, size_t size) {
+        if (size == 0)
+            return;
+
+        const ObjectInterval interval{addr, addr + size};
+        for (auto& pair : boost::make_iterator_range(object_cache.equal_range(interval))) {
+            for (auto& cached_object : pair.second) {
+                if (!cached_object)
+                    continue;
+
+                cached_object->Flush();
+            }
+        }
+    }
+
     /// Mark the specified region as being invalidated
     void InvalidateRegion(VAddr addr, u64 size) {
         if (size == 0)
@@ -71,6 +87,7 @@ protected:
     void Unregister(const T& object) {
         auto& rasterizer = Core::System::GetInstance().Renderer().Rasterizer();
         rasterizer.UpdatePagesCachedCount(object->GetAddr(), object->GetSizeInBytes(), -1);
+        object->Flush();
         object_cache.subtract({GetInterval(object), ObjectSet{object}});
     }
 

src/video_core/renderer_opengl/gl_buffer_cache.h

@@ -23,6 +23,9 @@ struct CachedBufferEntry final {
         return size;
     }
 
+    // We do not have to flush this cache as things in it are never modified by us.
+    void Flush() {}
+
     VAddr addr;
     std::size_t size;
     GLintptr offset;

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -322,6 +322,13 @@ void RasterizerOpenGL::ConfigureFramebuffers(bool using_color_fb, bool using_dep
             // Used when just a single color attachment is enabled, e.g. for clearing a color buffer
             Surface color_surface =
                 res_cache.GetColorBufferSurface(*single_color_target, preserve_contents);
+
+            if (color_surface) {
+                // Assume that a surface will be written to if it is used as a framebuffer, even if
+                // the shader doesn't actually write to it.
+                color_surface->MarkAsDirty();
+            }
+
             glFramebufferTexture2D(
                 GL_DRAW_FRAMEBUFFER,
                 GL_COLOR_ATTACHMENT0 + static_cast<GLenum>(*single_color_target), GL_TEXTURE_2D,
@@ -332,6 +339,11 @@ void RasterizerOpenGL::ConfigureFramebuffers(bool using_color_fb, bool using_dep
             std::array<GLenum, Maxwell::NumRenderTargets> buffers;
             for (std::size_t index = 0; index < Maxwell::NumRenderTargets; ++index) {
                 Surface color_surface = res_cache.GetColorBufferSurface(index, preserve_contents);
+                if (color_surface) {
+                    // Assume that a surface will be written to if it is used as a framebuffer, even
+                    // if the shader doesn't actually write to it.
+                    color_surface->MarkAsDirty();
+                }
                 buffers[index] = GL_COLOR_ATTACHMENT0 + regs.rt_control.GetMap(index);
                 glFramebufferTexture2D(
                     GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0 + static_cast<GLenum>(index),
@@ -351,6 +363,10 @@ void RasterizerOpenGL::ConfigureFramebuffers(bool using_color_fb, bool using_dep
     }
 
     if (depth_surface) {
+        // Assume that a surface will be written to if it is used as a framebuffer, even if
+        // the shader doesn't actually write to it.
+        depth_surface->MarkAsDirty();
+
         if (regs.stencil_enable) {
             // Attach both depth and stencil
             glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D,
@@ -383,7 +399,7 @@ void RasterizerOpenGL::Clear() {
     bool use_stencil{};
 
     OpenGLState clear_state;
-    clear_state.draw.draw_framebuffer = state.draw.draw_framebuffer;
+    clear_state.draw.draw_framebuffer = framebuffer.handle;
     clear_state.color_mask.red_enabled = regs.clear_buffers.R ? GL_TRUE : GL_FALSE;
     clear_state.color_mask.green_enabled = regs.clear_buffers.G ? GL_TRUE : GL_FALSE;
     clear_state.color_mask.blue_enabled = regs.clear_buffers.B ? GL_TRUE : GL_FALSE;
@@ -484,8 +500,13 @@ void RasterizerOpenGL::DrawArrays() {
     GLintptr index_buffer_offset = 0;
     if (is_indexed) {
         MICROPROFILE_SCOPE(OpenGL_Index);
-        index_buffer_offset =
-            buffer_cache.UploadMemory(regs.index_array.StartAddress(), index_buffer_size);
+
+        // Adjust the index buffer offset so it points to the first desired index.
+        auto index_start = regs.index_array.StartAddress();
+        index_start += static_cast<size_t>(regs.index_array.first) *
+                       static_cast<size_t>(regs.index_array.FormatSizeInBytes());
+
+        index_buffer_offset = buffer_cache.UploadMemory(index_start, index_buffer_size);
     }
 
     SetupShaders();
@@ -499,10 +520,6 @@ void RasterizerOpenGL::DrawArrays() {
     if (is_indexed) {
         const GLint base_vertex{static_cast<GLint>(regs.vb_element_base)};
 
-        // Adjust the index buffer offset so it points to the first desired index.
-        index_buffer_offset += static_cast<GLintptr>(regs.index_array.first) *
-                               static_cast<GLintptr>(regs.index_array.FormatSizeInBytes());
-
         if (gpu.state.current_instance > 0) {
             glDrawElementsInstancedBaseVertexBaseInstance(
                 primitive_mode, regs.index_array.count,
@@ -537,9 +554,19 @@ void RasterizerOpenGL::DrawArrays() {
     state.Apply();
 }
 
-void RasterizerOpenGL::FlushAll() {}
+void RasterizerOpenGL::FlushAll() {
+    MICROPROFILE_SCOPE(OpenGL_CacheManagement);
+    res_cache.FlushRegion(0, Kernel::VMManager::MAX_ADDRESS);
+    shader_cache.FlushRegion(0, Kernel::VMManager::MAX_ADDRESS);
+    buffer_cache.FlushRegion(0, Kernel::VMManager::MAX_ADDRESS);
+}
 
-void RasterizerOpenGL::FlushRegion(VAddr addr, u64 size) {}
+void RasterizerOpenGL::FlushRegion(VAddr addr, u64 size) {
+    MICROPROFILE_SCOPE(OpenGL_CacheManagement);
+    res_cache.FlushRegion(addr, size);
+    shader_cache.FlushRegion(addr, size);
+    buffer_cache.FlushRegion(addr, size);
+}
 
 void RasterizerOpenGL::InvalidateRegion(VAddr addr, u64 size) {
     MICROPROFILE_SCOPE(OpenGL_CacheManagement);
@@ -549,6 +576,7 @@ void RasterizerOpenGL::InvalidateRegion(VAddr addr, u64 size) {
 }
 
 void RasterizerOpenGL::FlushAndInvalidateRegion(VAddr addr, u64 size) {
+    FlushRegion(addr, size);
     InvalidateRegion(addr, size);
 }
 

src/video_core/renderer_opengl/gl_rasterizer_cache.cpp

@@ -265,20 +265,22 @@ void MortonCopy(u32 stride, u32 block_height, u32 height, u8* gl_buffer, std::si
     constexpr u32 bytes_per_pixel = SurfaceParams::GetFormatBpp(format) / CHAR_BIT;
     constexpr u32 gl_bytes_per_pixel = CachedSurface::GetGLBytesPerPixel(format);
 
+    // With the BCn formats (DXT and DXN), each 4x4 tile is swizzled instead of just individual
+    // pixel values.
+    const u32 tile_size{IsFormatBCn(format) ? 4U : 1U};
+
     if (morton_to_gl) {
-        // With the BCn formats (DXT and DXN), each 4x4 tile is swizzled instead of just individual
-        // pixel values.
-        const u32 tile_size{IsFormatBCn(format) ? 4U : 1U};
         const std::vector<u8> data = Tegra::Texture::UnswizzleTexture(
             addr, tile_size, bytes_per_pixel, stride, height, block_height);
         const std::size_t size_to_copy{std::min(gl_buffer_size, data.size())};
         memcpy(gl_buffer, data.data(), size_to_copy);
     } else {
-        // TODO(bunnei): Assumes the default rendering GOB size of 16 (128 lines). We should
-        // check the configuration for this and perform more generic un/swizzle
-        LOG_WARNING(Render_OpenGL, "need to use correct swizzle/GOB parameters!");
-        VideoCore::MortonCopyPixels128(stride, height, bytes_per_pixel, gl_bytes_per_pixel,
-                                       Memory::GetPointer(addr), gl_buffer, morton_to_gl);
+        std::vector<u8> data(height * stride * bytes_per_pixel);
+        Tegra::Texture::CopySwizzledData(stride / tile_size, height / tile_size, bytes_per_pixel,
+                                         bytes_per_pixel, data.data(), gl_buffer, false,
+                                         block_height);
+        const std::size_t size_to_copy{std::min(gl_buffer_size, data.size())};
+        memcpy(Memory::GetPointer(addr), data.data(), size_to_copy);
     }
 }
 
@@ -357,17 +359,16 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, std::size_t, VAddr),
         MortonCopy<false, PixelFormat::RGBA16UI>,
         MortonCopy<false, PixelFormat::R11FG11FB10F>,
         MortonCopy<false, PixelFormat::RGBA32UI>,
-        // TODO(Subv): Swizzling DXT1/DXT23/DXT45/DXN1/DXN2/BC7U/BC6H_UF16/BC6H_SF16/ASTC_2D_4X4
-        // formats are not supported
-        nullptr,
-        nullptr,
-        nullptr,
-        nullptr,
-        nullptr,
-        nullptr,
-        nullptr,
-        nullptr,
-        nullptr,
+        MortonCopy<false, PixelFormat::DXT1>,
+        MortonCopy<false, PixelFormat::DXT23>,
+        MortonCopy<false, PixelFormat::DXT45>,
+        MortonCopy<false, PixelFormat::DXN1>,
+        MortonCopy<false, PixelFormat::DXN2UNORM>,
+        MortonCopy<false, PixelFormat::DXN2SNORM>,
+        MortonCopy<false, PixelFormat::BC7U>,
+        MortonCopy<false, PixelFormat::BC6H_UF16>,
+        MortonCopy<false, PixelFormat::BC6H_SF16>,
+        // TODO(Subv): Swizzling ASTC formats are not supported
         nullptr,
         MortonCopy<false, PixelFormat::G8R8U>,
         MortonCopy<false, PixelFormat::G8R8S>,
@@ -503,7 +504,7 @@ CachedSurface::CachedSurface(const SurfaceParams& params)
     glTexParameteri(SurfaceTargetToGL(params.target), GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
 }
 
-static void ConvertS8Z24ToZ24S8(std::vector<u8>& data, u32 width, u32 height) {
+static void ConvertS8Z24ToZ24S8(std::vector<u8>& data, u32 width, u32 height, bool reverse) {
     union S8Z24 {
         BitField<0, 24, u32> z24;
         BitField<24, 8, u32> s8;
@@ -516,16 +517,23 @@ static void ConvertS8Z24ToZ24S8(std::vector<u8>& data, u32 width, u32 height) {
     };
     static_assert(sizeof(Z24S8) == 4, "Z24S8 is incorrect size");
 
-    S8Z24 input_pixel{};
-    Z24S8 output_pixel{};
+    S8Z24 s8z24_pixel{};
+    Z24S8 z24s8_pixel{};
     constexpr auto bpp{CachedSurface::GetGLBytesPerPixel(PixelFormat::S8Z24)};
     for (std::size_t y = 0; y < height; ++y) {
         for (std::size_t x = 0; x < width; ++x) {
             const std::size_t offset{bpp * (y * width + x)};
-            std::memcpy(&input_pixel, &data[offset], sizeof(S8Z24));
-            output_pixel.s8.Assign(input_pixel.s8);
-            output_pixel.z24.Assign(input_pixel.z24);
-            std::memcpy(&data[offset], &output_pixel, sizeof(Z24S8));
+            if (reverse) {
+                std::memcpy(&z24s8_pixel, &data[offset], sizeof(Z24S8));
+                s8z24_pixel.s8.Assign(z24s8_pixel.s8);
+                s8z24_pixel.z24.Assign(z24s8_pixel.z24);
+                std::memcpy(&data[offset], &s8z24_pixel, sizeof(S8Z24));
+            } else {
+                std::memcpy(&s8z24_pixel, &data[offset], sizeof(S8Z24));
+                z24s8_pixel.s8.Assign(s8z24_pixel.s8);
+                z24s8_pixel.z24.Assign(s8z24_pixel.z24);
+                std::memcpy(&data[offset], &z24s8_pixel, sizeof(Z24S8));
+            }
         }
     }
 }
@@ -561,7 +569,7 @@ static void ConvertFormatAsNeeded_LoadGLBuffer(std::vector<u8>& data, PixelForma
     }
     case PixelFormat::S8Z24:
         // Convert the S8Z24 depth format to Z24S8, as OpenGL does not support S8Z24.
-        ConvertS8Z24ToZ24S8(data, width, height);
+        ConvertS8Z24ToZ24S8(data, width, height, false);
         break;
 
     case PixelFormat::G8R8U:
@@ -572,6 +580,30 @@ static void ConvertFormatAsNeeded_LoadGLBuffer(std::vector<u8>& data, PixelForma
     }
 }
 
+/**
+ * Helper function to perform software conversion (as needed) when flushing a buffer from OpenGL to
+ * Switch memory. This is for Maxwell pixel formats that cannot be represented as-is in OpenGL or
+ * with typical desktop GPUs.
+ */
+static void ConvertFormatAsNeeded_FlushGLBuffer(std::vector<u8>& data, PixelFormat pixel_format,
+                                                u32 width, u32 height) {
+    switch (pixel_format) {
+    case PixelFormat::G8R8U:
+    case PixelFormat::G8R8S:
+    case PixelFormat::ASTC_2D_4X4:
+    case PixelFormat::ASTC_2D_8X8: {
+        LOG_CRITICAL(HW_GPU, "Conversion of format {} after texture flushing is not implemented",
+                     static_cast<u32>(pixel_format));
+        UNREACHABLE();
+        break;
+    }
+    case PixelFormat::S8Z24:
+        // Convert the Z24S8 depth format to S8Z24, as OpenGL does not support S8Z24.
+        ConvertS8Z24ToZ24S8(data, width, height, true);
+        break;
+    }
+}
+
 MICROPROFILE_DEFINE(OpenGL_SurfaceLoad, "OpenGL", "Surface Load", MP_RGB(128, 64, 192));
 void CachedSurface::LoadGLBuffer() {
     ASSERT(params.type != SurfaceType::Fill);
@@ -609,11 +641,70 @@ void CachedSurface::LoadGLBuffer() {
     }
 
     ConvertFormatAsNeeded_LoadGLBuffer(gl_buffer, params.pixel_format, params.width, params.height);
+
+    dirty = false;
 }
 
 MICROPROFILE_DEFINE(OpenGL_SurfaceFlush, "OpenGL", "Surface Flush", MP_RGB(128, 192, 64));
 void CachedSurface::FlushGLBuffer() {
-    ASSERT_MSG(false, "Unimplemented");
+    MICROPROFILE_SCOPE(OpenGL_SurfaceFlush);
+
+    const auto& rect{params.GetRect()};
+
+    // Load data from memory to the surface
+    const GLint x0 = static_cast<GLint>(rect.left);
+    const GLint y0 = static_cast<GLint>(rect.bottom);
+    const size_t buffer_offset =
+        static_cast<size_t>(static_cast<size_t>(y0) * params.width + static_cast<size_t>(x0)) *
+        GetGLBytesPerPixel(params.pixel_format);
+
+    const u32 bytes_per_pixel = GetGLBytesPerPixel(params.pixel_format);
+    const u32 copy_size = params.width * params.height * bytes_per_pixel;
+    gl_buffer.resize(static_cast<size_t>(params.depth) * copy_size);
+
+    const FormatTuple& tuple = GetFormatTuple(params.pixel_format, params.component_type);
+
+    // Ensure no bad interactions with GL_UNPACK_ALIGNMENT
+    ASSERT(params.width * GetGLBytesPerPixel(params.pixel_format) % 4 == 0);
+    glPixelStorei(GL_PACK_ROW_LENGTH, static_cast<GLint>(params.width));
+
+    ASSERT(!tuple.compressed);
+    ASSERT(x0 == 0 && y0 == 0);
+
+    glBindBuffer(GL_PIXEL_PACK_BUFFER, 0);
+    glGetTextureImage(texture.handle, 0, tuple.format, tuple.type, gl_buffer.size(),
+                      gl_buffer.data());
+    glPixelStorei(GL_PACK_ROW_LENGTH, 0);
+
+    ConvertFormatAsNeeded_FlushGLBuffer(gl_buffer, params.pixel_format, params.width,
+                                        params.height);
+
+    ASSERT(params.type != SurfaceType::Fill);
+
+    const u8* const texture_src_data = Memory::GetPointer(params.addr);
+
+    ASSERT(texture_src_data);
+
+    if (params.is_tiled) {
+        // TODO(bunnei): This only swizzles and copies a 2D texture - we do not yet know how to do
+        // this for 3D textures, etc.
+        switch (params.target) {
+        case SurfaceParams::SurfaceTarget::Texture2D:
+            // Pass impl. to the fallback code below
+            break;
+        default:
+            LOG_CRITICAL(HW_GPU, "Unimplemented tiled unload for target={}",
+                         static_cast<u32>(params.target));
+            UNREACHABLE();
+        }
+
+        gl_to_morton_fns[static_cast<size_t>(params.pixel_format)](
+            params.width, params.block_height, params.height, &gl_buffer[buffer_offset], copy_size,
+            params.addr + buffer_offset);
+    } else {
+        Memory::WriteBlock(params.addr + buffer_offset, &gl_buffer[buffer_offset],
+                           gl_buffer.size() - buffer_offset);
+    }
 }
 
 MICROPROFILE_DEFINE(OpenGL_TextureUL, "OpenGL", "Texture Upload", MP_RGB(128, 64, 192));

src/video_core/renderer_opengl/gl_rasterizer_cache.h

@@ -741,6 +741,19 @@ public:
         return params.size_in_bytes;
     }
 
+    void Flush() {
+        // There is no need to flush the surface if it hasn't been modified by us.
+        if (!dirty)
+            return;
+
+        FlushGLBuffer();
+        dirty = false;
+    }
+
+    void MarkAsDirty() {
+        dirty = true;
+    }
+
     const OGLTexture& Texture() const {
         return texture;
     }
@@ -772,6 +785,7 @@ private:
     std::vector<u8> gl_buffer;
     SurfaceParams params;
     GLenum gl_target;
+    bool dirty = false;
 };
 
 class RasterizerCacheOpenGL final : public RasterizerCache<Surface> {

src/video_core/renderer_opengl/gl_shader_cache.h

@@ -32,6 +32,9 @@ public:
         return GLShader::MAX_PROGRAM_CODE_LENGTH * sizeof(u64);
     }
 
+    // We do not have to flush this cache as things in it are never modified by us.
+    void Flush() {}
+
     /// Gets the shader entries for the shader
     const GLShader::ShaderEntries& GetShaderEntries() const {
         return entries;

src/video_core/renderer_opengl/gl_shader_decompiler.cpp

@@ -12,6 +12,7 @@
 #include "common/assert.h"
 #include "common/common_types.h"
 #include "video_core/engines/shader_bytecode.h"
+#include "video_core/engines/shader_header.h"
 #include "video_core/renderer_opengl/gl_rasterizer.h"
 #include "video_core/renderer_opengl/gl_shader_decompiler.h"
 
@@ -26,7 +27,7 @@ using Tegra::Shader::Sampler;
 using Tegra::Shader::SubOp;
 
 constexpr u32 PROGRAM_END = MAX_PROGRAM_CODE_LENGTH;
-constexpr u32 PROGRAM_HEADER_SIZE = 0x50;
+constexpr u32 PROGRAM_HEADER_SIZE = sizeof(Tegra::Shader::Header);
 
 class DecompileFail : public std::runtime_error {
 public:
@@ -235,6 +236,14 @@ private:
     const std::string& suffix;
 };
 
+enum class InternalFlag : u64 {
+    ZeroFlag = 0,
+    CarryFlag = 1,
+    OverflowFlag = 2,
+    NaNFlag = 3,
+    Amount
+};
+
 /**
  * Used to manage shader registers that are emulated with GLSL. This class keeps track of the state
  * of all registers (e.g. whether they are currently being used as Floats or Integers), and
@@ -328,13 +337,19 @@ public:
     void SetRegisterToInteger(const Register& reg, bool is_signed, u64 elem,
                               const std::string& value, u64 dest_num_components,
                               u64 value_num_components, bool is_saturated = false,
-                              u64 dest_elem = 0, Register::Size size = Register::Size::Word) {
+                              u64 dest_elem = 0, Register::Size size = Register::Size::Word,
+                              bool sets_cc = false) {
         ASSERT_MSG(!is_saturated, "Unimplemented");
 
         const std::string func{is_signed ? "intBitsToFloat" : "uintBitsToFloat"};
 
         SetRegister(reg, elem, func + '(' + ConvertIntegerSize(value, size) + ')',
                     dest_num_components, value_num_components, dest_elem);
+
+        if (sets_cc) {
+            const std::string zero_condition = "( " + ConvertIntegerSize(value, size) + " == 0 )";
+            SetInternalFlag(InternalFlag::ZeroFlag, zero_condition);
+        }
     }
 
     /**
@@ -351,6 +366,26 @@ public:
         shader.AddLine(dest + " = " + src + ';');
     }
 
+    std::string GetControlCode(const Tegra::Shader::ControlCode cc) const {
+        switch (cc) {
+        case Tegra::Shader::ControlCode::NEU:
+            return "!(" + GetInternalFlag(InternalFlag::ZeroFlag) + ')';
+        default:
+            LOG_CRITICAL(HW_GPU, "Unimplemented Control Code {}", static_cast<u32>(cc));
+            UNREACHABLE();
+            return "false";
+        }
+    }
+
+    std::string GetInternalFlag(const InternalFlag ii) const {
+        const u32 code = static_cast<u32>(ii);
+        return "internalFlag_" + std::to_string(code) + suffix;
+    }
+
+    void SetInternalFlag(const InternalFlag ii, const std::string& value) const {
+        shader.AddLine(GetInternalFlag(ii) + " = " + value + ';');
+    }
+
     /**
      * Writes code that does a output attribute assignment to register operation. Output attributes
      * are stored as floats, so this may require conversion.
@@ -414,6 +449,12 @@ public:
         }
         declarations.AddNewLine();
 
+        for (u32 ii = 0; ii < static_cast<u64>(InternalFlag::Amount); ii++) {
+            const InternalFlag code = static_cast<InternalFlag>(ii);
+            declarations.AddLine("bool " + GetInternalFlag(code) + " = false;");
+        }
+        declarations.AddNewLine();
+
         for (const auto element : declr_input_attribute) {
             // TODO(bunnei): Use proper number of elements for these
             u32 idx =
@@ -674,7 +715,7 @@ public:
                   u32 main_offset, Maxwell3D::Regs::ShaderStage stage, const std::string& suffix)
         : subroutines(subroutines), program_code(program_code), main_offset(main_offset),
           stage(stage), suffix(suffix) {
-
+        std::memcpy(&header, program_code.data(), sizeof(Tegra::Shader::Header));
         Generate(suffix);
     }
 
@@ -688,23 +729,6 @@ public:
     }
 
 private:
-    // Shader program header for a Fragment Shader.
-    struct FragmentHeader {
-        INSERT_PADDING_WORDS(5);
-        INSERT_PADDING_WORDS(13);
-        u32 enabled_color_outputs;
-        union {
-            BitField<0, 1, u32> writes_samplemask;
-            BitField<1, 1, u32> writes_depth;
-        };
-
-        bool IsColorComponentOutputEnabled(u32 render_target, u32 component) const {
-            const u32 bit = render_target * 4 + component;
-            return enabled_color_outputs & (1 << bit);
-        }
-    };
-    static_assert(sizeof(FragmentHeader) == PROGRAM_HEADER_SIZE, "FragmentHeader size is wrong");
-
     /// Gets the Subroutine object corresponding to the specified address.
     const Subroutine& GetSubroutine(u32 begin, u32 end) const {
         const auto iter = subroutines.find(Subroutine{begin, end, suffix});
@@ -954,8 +978,6 @@ private:
         // TEXS has two destination registers and a swizzle. The first two elements in the swizzle
         // go into gpr0+0 and gpr0+1, and the rest goes into gpr28+0 and gpr28+1
 
-        ASSERT_MSG(instr.texs.nodep == 0, "TEXS nodep not implemented");
-
         std::size_t written_components = 0;
         for (u32 component = 0; component < 4; ++component) {
             if (!instr.texs.IsComponentEnabled(component)) {
@@ -1010,10 +1032,8 @@ private:
     /// Writes the output values from a fragment shader to the corresponding GLSL output variables.
     void EmitFragmentOutputsWrite() {
         ASSERT(stage == Maxwell3D::Regs::ShaderStage::Fragment);
-        FragmentHeader header;
-        std::memcpy(&header, program_code.data(), PROGRAM_HEADER_SIZE);
 
-        ASSERT_MSG(header.writes_samplemask == 0, "Samplemask write is unimplemented");
+        ASSERT_MSG(header.ps.omap.sample_mask == 0, "Samplemask write is unimplemented");
 
         // Write the color outputs using the data in the shader registers, disabled
         // rendertargets/components are skipped in the register assignment.
@@ -1022,7 +1042,7 @@ private:
              ++render_target) {
             // TODO(Subv): Figure out how dual-source blending is configured in the Switch.
             for (u32 component = 0; component < 4; ++component) {
-                if (header.IsColorComponentOutputEnabled(render_target, component)) {
+                if (header.ps.IsColorComponentOutputEnabled(render_target, component)) {
                     shader.AddLine(fmt::format("FragColor{}[{}] = {};", render_target, component,
                                                regs.GetRegisterAsFloat(current_reg)));
                     ++current_reg;
@@ -1030,7 +1050,7 @@ private:
             }
         }
 
-        if (header.writes_depth) {
+        if (header.ps.omap.depth) {
             // The depth output is always 2 registers after the last color output, and current_reg
             // already contains one past the last color register.
 
@@ -1640,7 +1660,8 @@ private:
                 }
 
                 regs.SetRegisterToInteger(instr.gpr0, instr.conversion.is_output_signed, 0, op_a, 1,
-                                          1, instr.alu.saturate_d, 0, instr.conversion.dest_size);
+                                          1, instr.alu.saturate_d, 0, instr.conversion.dest_size,
+                                          instr.generates_cc.Value() != 0);
                 break;
             }
             case OpCode::Id::I2F_R:
@@ -1779,8 +1800,8 @@ private:
                 Tegra::Shader::IpaMode input_mode{Tegra::Shader::IpaInterpMode::Perspective,
                                                   Tegra::Shader::IpaSampleMode::Default};
 
-                u32 next_element = instr.attribute.fmt20.element;
-                u32 next_index = static_cast<u32>(instr.attribute.fmt20.index.Value());
+                u64 next_element = instr.attribute.fmt20.element;
+                u64 next_index = static_cast<u64>(instr.attribute.fmt20.index.Value());
 
                 const auto LoadNextElement = [&](u32 reg_offset) {
                     regs.SetRegisterToInputAttibute(instr.gpr0.Value() + reg_offset, next_element,
@@ -1844,8 +1865,8 @@ private:
                 ASSERT_MSG((instr.attribute.fmt20.immediate.Value() % sizeof(u32)) == 0,
                            "Unaligned attribute loads are not supported");
 
-                u32 next_element = instr.attribute.fmt20.element;
-                u32 next_index = static_cast<u32>(instr.attribute.fmt20.index.Value());
+                u64 next_element = instr.attribute.fmt20.element;
+                u64 next_index = static_cast<u64>(instr.attribute.fmt20.index.Value());
 
                 const auto StoreNextElement = [&](u32 reg_offset) {
                     regs.SetOutputAttributeToRegister(static_cast<Attribute::Index>(next_index),
@@ -1871,6 +1892,13 @@ private:
                 Tegra::Shader::TextureType texture_type{instr.tex.texture_type};
                 std::string coord;
 
+                ASSERT_MSG(!instr.tex.UsesMiscMode(Tegra::Shader::TextureMiscMode::NODEP),
+                           "NODEP is not implemented");
+                ASSERT_MSG(!instr.tex.UsesMiscMode(Tegra::Shader::TextureMiscMode::AOFFI),
+                           "AOFFI is not implemented");
+                ASSERT_MSG(!instr.tex.UsesMiscMode(Tegra::Shader::TextureMiscMode::DC),
+                           "DC is not implemented");
+
                 switch (texture_type) {
                 case Tegra::Shader::TextureType::Texture1D: {
                     const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
@@ -1953,6 +1981,11 @@ private:
                 Tegra::Shader::TextureType texture_type{instr.texs.GetTextureType()};
                 bool is_array{instr.texs.IsArrayTexture()};
 
+                ASSERT_MSG(!instr.texs.UsesMiscMode(Tegra::Shader::TextureMiscMode::NODEP),
+                           "NODEP is not implemented");
+                ASSERT_MSG(!instr.texs.UsesMiscMode(Tegra::Shader::TextureMiscMode::DC),
+                           "DC is not implemented");
+
                 switch (texture_type) {
                 case Tegra::Shader::TextureType::Texture2D: {
                     if (is_array) {
@@ -1989,6 +2022,13 @@ private:
                 ASSERT(instr.tlds.IsArrayTexture() == false);
                 std::string coord;
 
+                ASSERT_MSG(!instr.tlds.UsesMiscMode(Tegra::Shader::TextureMiscMode::NODEP),
+                           "NODEP is not implemented");
+                ASSERT_MSG(!instr.tlds.UsesMiscMode(Tegra::Shader::TextureMiscMode::AOFFI),
+                           "AOFFI is not implemented");
+                ASSERT_MSG(!instr.tlds.UsesMiscMode(Tegra::Shader::TextureMiscMode::MZ),
+                           "MZ is not implemented");
+
                 switch (instr.tlds.GetTextureType()) {
                 case Tegra::Shader::TextureType::Texture2D: {
                     if (instr.tlds.IsArrayTexture()) {
@@ -2017,6 +2057,17 @@ private:
                 ASSERT(instr.tld4.array == 0);
                 std::string coord;
 
+                ASSERT_MSG(!instr.tld4.UsesMiscMode(Tegra::Shader::TextureMiscMode::NODEP),
+                           "NODEP is not implemented");
+                ASSERT_MSG(!instr.tld4.UsesMiscMode(Tegra::Shader::TextureMiscMode::AOFFI),
+                           "AOFFI is not implemented");
+                ASSERT_MSG(!instr.tld4.UsesMiscMode(Tegra::Shader::TextureMiscMode::DC),
+                           "DC is not implemented");
+                ASSERT_MSG(!instr.tld4.UsesMiscMode(Tegra::Shader::TextureMiscMode::NDV),
+                           "NDV is not implemented");
+                ASSERT_MSG(!instr.tld4.UsesMiscMode(Tegra::Shader::TextureMiscMode::PTP),
+                           "PTP is not implemented");
+
                 switch (instr.tld4.texture_type) {
                 case Tegra::Shader::TextureType::Texture2D: {
                     const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
@@ -2054,6 +2105,13 @@ private:
                 break;
             }
             case OpCode::Id::TLD4S: {
+                ASSERT_MSG(!instr.tld4s.UsesMiscMode(Tegra::Shader::TextureMiscMode::NODEP),
+                           "NODEP is not implemented");
+                ASSERT_MSG(!instr.tld4s.UsesMiscMode(Tegra::Shader::TextureMiscMode::AOFFI),
+                           "AOFFI is not implemented");
+                ASSERT_MSG(!instr.tld4s.UsesMiscMode(Tegra::Shader::TextureMiscMode::DC),
+                           "DC is not implemented");
+
                 const std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
                 const std::string op_b = regs.GetRegisterAsFloat(instr.gpr20);
                 // TODO(Subv): Figure out how the sampler type is encoded in the TLD4S instruction.
@@ -2066,6 +2124,9 @@ private:
                 break;
             }
             case OpCode::Id::TXQ: {
+                ASSERT_MSG(!instr.txq.UsesMiscMode(Tegra::Shader::TextureMiscMode::NODEP),
+                           "NODEP is not implemented");
+
                 // TODO: the new commits on the texture refactor, change the way samplers work.
                 // Sadly, not all texture instructions specify the type of texture their sampler
                 // uses. This must be fixed at a later instance.
@@ -2086,6 +2147,11 @@ private:
                 break;
             }
             case OpCode::Id::TMML: {
+                ASSERT_MSG(!instr.tmml.UsesMiscMode(Tegra::Shader::TextureMiscMode::NODEP),
+                           "NODEP is not implemented");
+                ASSERT_MSG(!instr.tmml.UsesMiscMode(Tegra::Shader::TextureMiscMode::NDV),
+                           "NDV is not implemented");
+
                 const std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
                 const std::string op_b = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
                 const bool is_array = instr.tmml.array != 0;
@@ -2252,31 +2318,55 @@ private:
             break;
         }
         case OpCode::Type::PredicateSetPredicate: {
-            const std::string op_a =
-                GetPredicateCondition(instr.psetp.pred12, instr.psetp.neg_pred12 != 0);
-            const std::string op_b =
-                GetPredicateCondition(instr.psetp.pred29, instr.psetp.neg_pred29 != 0);
+            switch (opcode->GetId()) {
+            case OpCode::Id::PSETP: {
+                const std::string op_a =
+                    GetPredicateCondition(instr.psetp.pred12, instr.psetp.neg_pred12 != 0);
+                const std::string op_b =
+                    GetPredicateCondition(instr.psetp.pred29, instr.psetp.neg_pred29 != 0);
 
-            // We can't use the constant predicate as destination.
-            ASSERT(instr.psetp.pred3 != static_cast<u64>(Pred::UnusedIndex));
+                // We can't use the constant predicate as destination.
+                ASSERT(instr.psetp.pred3 != static_cast<u64>(Pred::UnusedIndex));
 
-            const std::string second_pred =
-                GetPredicateCondition(instr.psetp.pred39, instr.psetp.neg_pred39 != 0);
+                const std::string second_pred =
+                    GetPredicateCondition(instr.psetp.pred39, instr.psetp.neg_pred39 != 0);
 
-            const std::string combiner = GetPredicateCombiner(instr.psetp.op);
+                const std::string combiner = GetPredicateCombiner(instr.psetp.op);
 
-            const std::string predicate =
-                '(' + op_a + ") " + GetPredicateCombiner(instr.psetp.cond) + " (" + op_b + ')';
+                const std::string predicate =
+                    '(' + op_a + ") " + GetPredicateCombiner(instr.psetp.cond) + " (" + op_b + ')';
 
-            // Set the primary predicate to the result of Predicate OP SecondPredicate
-            SetPredicate(instr.psetp.pred3,
-                         '(' + predicate + ") " + combiner + " (" + second_pred + ')');
+                // Set the primary predicate to the result of Predicate OP SecondPredicate
+                SetPredicate(instr.psetp.pred3,
+                             '(' + predicate + ") " + combiner + " (" + second_pred + ')');
 
-            if (instr.psetp.pred0 != static_cast<u64>(Pred::UnusedIndex)) {
-                // Set the secondary predicate to the result of !Predicate OP SecondPredicate,
-                // if enabled
-                SetPredicate(instr.psetp.pred0,
-                             "!(" + predicate + ") " + combiner + " (" + second_pred + ')');
+                if (instr.psetp.pred0 != static_cast<u64>(Pred::UnusedIndex)) {
+                    // Set the secondary predicate to the result of !Predicate OP SecondPredicate,
+                    // if enabled
+                    SetPredicate(instr.psetp.pred0,
+                                 "!(" + predicate + ") " + combiner + " (" + second_pred + ')');
+                }
+                break;
+            }
+            case OpCode::Id::CSETP: {
+                const std::string pred =
+                    GetPredicateCondition(instr.csetp.pred39, instr.csetp.neg_pred39 != 0);
+                const std::string combiner = GetPredicateCombiner(instr.csetp.op);
+                const std::string controlCode = regs.GetControlCode(instr.csetp.cc);
+                if (instr.csetp.pred3 != static_cast<u64>(Pred::UnusedIndex)) {
+                    SetPredicate(instr.csetp.pred3,
+                                 '(' + controlCode + ") " + combiner + " (" + pred + ')');
+                }
+                if (instr.csetp.pred0 != static_cast<u64>(Pred::UnusedIndex)) {
+                    SetPredicate(instr.csetp.pred0,
+                                 "!(" + controlCode + ") " + combiner + " (" + pred + ')');
+                }
+                break;
+            }
+            default: {
+                LOG_CRITICAL(HW_GPU, "Unhandled predicate instruction: {}", opcode->GetName());
+                UNREACHABLE();
+            }
             }
             break;
         }
@@ -2666,6 +2756,7 @@ private:
 private:
     const std::set<Subroutine>& subroutines;
     const ProgramCode& program_code;
+    Tegra::Shader::Header header;
     const u32 main_offset;
     Maxwell3D::Regs::ShaderStage stage;
     const std::string& suffix;

src/video_core/renderer_opengl/gl_stream_buffer.cpp

@@ -74,7 +74,7 @@ std::tuple<u8*, GLintptr, bool> OGLStreamBuffer::Map(GLsizeiptr size, GLintptr a
         }
     }
 
-    if (invalidate | !persistent) {
+    if (invalidate || !persistent) {
         GLbitfield flags = GL_MAP_WRITE_BIT | (persistent ? GL_MAP_PERSISTENT_BIT : 0) |
                            (coherent ? GL_MAP_COHERENT_BIT : GL_MAP_FLUSH_EXPLICIT_BIT) |
                            (invalidate ? GL_MAP_INVALIDATE_BUFFER_BIT : GL_MAP_UNSYNCHRONIZED_BIT);

src/video_core/textures/decoders.cpp

@@ -88,6 +88,38 @@ void FastSwizzleData(u32 width, u32 height, u32 bytes_per_pixel, u8* swizzled_da
     }
 }
 
+void SwizzleSubrect(u32 subrect_width, u32 subrect_height, u32 source_pitch, u32 swizzled_width,
+                    u32 bytes_per_pixel, VAddr swizzled_data, VAddr unswizzled_data,
+                    u32 block_height) {
+    for (u32 line = 0; line < subrect_height; ++line) {
+        for (u32 x = 0; x < subrect_width; ++x) {
+            u32 swizzled_offset =
+                GetSwizzleOffset(x, line, swizzled_width, bytes_per_pixel, block_height);
+
+            const VAddr source_line = unswizzled_data + line * source_pitch + x * bytes_per_pixel;
+            const VAddr dest_addr = swizzled_data + swizzled_offset;
+
+            Memory::CopyBlock(dest_addr, source_line, bytes_per_pixel);
+        }
+    }
+}
+
+void UnswizzleSubrect(u32 subrect_width, u32 subrect_height, u32 dest_pitch, u32 swizzled_width,
+                      u32 bytes_per_pixel, VAddr swizzled_data, VAddr unswizzled_data,
+                      u32 block_height, u32 offset_x, u32 offset_y) {
+    for (u32 line = 0; line < subrect_height; ++line) {
+        for (u32 x = 0; x < subrect_width; ++x) {
+            u32 swizzled_offset = GetSwizzleOffset(offset_x, line + offset_y, swizzled_width,
+                                                   bytes_per_pixel, block_height);
+
+            const VAddr dest_line = unswizzled_data + line * dest_pitch + x;
+            const VAddr source_addr = swizzled_data + swizzled_offset;
+
+            Memory::CopyBlock(dest_line, source_addr, bytes_per_pixel);
+        }
+    }
+}
+
 u32 BytesPerPixel(TextureFormat format) {
     switch (format) {
     case TextureFormat::DXT1:

src/video_core/textures/decoders.h

@@ -26,6 +26,16 @@ std::vector<u8> UnswizzleDepthTexture(VAddr address, DepthFormat format, u32 wid
 void CopySwizzledData(u32 width, u32 height, u32 bytes_per_pixel, u32 out_bytes_per_pixel,
                       u8* swizzled_data, u8* unswizzled_data, bool unswizzle, u32 block_height);
 
+/// Copies an untiled subrectangle into a tiled surface.
+void SwizzleSubrect(u32 subrect_width, u32 subrect_height, u32 source_pitch, u32 swizzled_width,
+                    u32 bytes_per_pixel, VAddr swizzled_data, VAddr unswizzled_data,
+                    u32 block_height);
+
+/// Copies a tiled subrectangle into a linear surface.
+void UnswizzleSubrect(u32 subrect_width, u32 subrect_height, u32 dest_pitch, u32 swizzled_width,
+                      u32 bytes_per_pixel, VAddr swizzled_data, VAddr unswizzled_data,
+                      u32 block_height, u32 offset_x, u32 offset_y);
+
 /**
  * Decodes an unswizzled texture into a A8R8G8B8 texture.
  */

src/yuzu/debugger/wait_tree.cpp

@@ -15,6 +15,7 @@
 #include "core/hle/kernel/thread.h"
 #include "core/hle/kernel/timer.h"
 #include "core/hle/kernel/wait_object.h"
+#include "core/memory.h"
 
 WaitTreeItem::WaitTreeItem() = default;
 WaitTreeItem::~WaitTreeItem() = default;

src/yuzu/game_list.cpp

@@ -89,15 +89,7 @@ bool GameList::SearchField::KeyReleaseEater::eventFilter(QObject* obj, QEvent* e
 }
 
 void GameList::SearchField::setFilterResult(int visible, int total) {
-    QString result_of_text = tr("of");
-    QString result_text;
-    if (total == 1) {
-        result_text = tr("result");
-    } else {
-        result_text = tr("results");
-    }
-    label_filter_result->setText(
-        QString("%1 %2 %3 %4").arg(visible).arg(result_of_text).arg(total).arg(result_text));
+    label_filter_result->setText(tr("%1 of %n result(s)", "", total).arg(visible));
 }
 
 void GameList::SearchField::clear() {

src/yuzu_cmd/config.cpp

@@ -125,6 +125,10 @@ void Config::ReadValues() {
 
     // System
     Settings::values.use_docked_mode = sdl2_config->GetBoolean("System", "use_docked_mode", false);
+    Settings::values.username = sdl2_config->Get("System", "username", "yuzu");
+    if (Settings::values.username.empty()) {
+        Settings::values.username = "yuzu";
+    }
 
     // Miscellaneous
     Settings::values.log_filter = sdl2_config->Get("Miscellaneous", "log_filter", "*:Trace");

src/yuzu_cmd/default_ini.h

@@ -176,7 +176,7 @@ use_docked_mode =
 
 # Sets the account username, max length is 32 characters
 # yuzu (default)
-username =
+username = yuzu
 
 # Sets the systems language index
 # 0: Japanese, 1: English (default), 2: French, 3: German, 4: Italian, 5: Spanish, 6: Chinese,

src/yuzu_cmd/yuzu.cpp

@@ -20,8 +20,10 @@
 #include "common/string_util.h"
 #include "common/telemetry.h"
 #include "core/core.h"
+#include "core/crypto/key_manager.h"
 #include "core/file_sys/vfs_real.h"
 #include "core/gdbstub/gdbstub.h"
+#include "core/hle/service/filesystem/filesystem.h"
 #include "core/loader/loader.h"
 #include "core/settings.h"
 #include "core/telemetry_session.h"
@@ -29,7 +31,6 @@
 #include "yuzu_cmd/emu_window/emu_window_sdl2.h"
 
 #include <getopt.h>
-#include "core/crypto/key_manager.h"
 #ifndef _MSC_VER
 #include <unistd.h>
 #endif
@@ -169,6 +170,7 @@ int main(int argc, char** argv) {
 
     Core::System& system{Core::System::GetInstance()};
     system.SetFilesystem(std::make_shared<FileSys::RealVfsFilesystem>());
+    Service::FileSystem::CreateFactories(system.GetFilesystem());
 
     SCOPE_EXIT({ system.Shutdown(); });