gl_shader_decompiler: Implement VMAD

Fixed block height settings for RenderTargets and Depth Buffers

.gitmodules

@@ -34,3 +34,9 @@
 [submodule "soundtouch"]
 	path = externals/soundtouch
 	url = https://github.com/citra-emu/ext-soundtouch.git
+[submodule "libressl"]
+    path = externals/libressl
+    url = https://github.com/citra-emu/ext-libressl-portable.git
+[submodule "discord-rpc"]
+    path = externals/discord-rpc
+    url = https://github.com/discordapp/discord-rpc.git

.travis/common/travis-ci.env

@@ -10,3 +10,7 @@ TRAVIS_JOB_ID
 TRAVIS_JOB_NUMBER
 TRAVIS_REPO_SLUG
 TRAVIS_TAG
+
+# yuzu specific flags
+ENABLE_COMPATIBILITY_REPORTING
+USE_DISCORD_PRESENCE

.travis/linux/build.sh

@@ -1,4 +1,4 @@
 #!/bin/bash -ex
 
 mkdir -p "$HOME/.ccache"
-docker run --env-file .travis/common/travis-ci.env -v $(pwd):/yuzu -v "$HOME/.ccache":/root/.ccache ubuntu:18.04 /bin/bash /yuzu/.travis/linux/docker.sh
+docker run -e ENABLE_COMPATIBILITY_REPORTING --env-file .travis/common/travis-ci.env -v $(pwd):/yuzu -v "$HOME/.ccache":/root/.ccache ubuntu:18.04 /bin/bash /yuzu/.travis/linux/docker.sh

.travis/linux/docker.sh

@@ -6,7 +6,7 @@ apt-get install --no-install-recommends -y build-essential git libqt5opengl5-dev
 cd /yuzu
 
 mkdir build && cd build
-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
+cmake .. -DYUZU_USE_BUNDLED_UNICORN=ON -DCMAKE_BUILD_TYPE=Release -DCMAKE_C_COMPILER=/usr/lib/ccache/gcc -DCMAKE_CXX_COMPILER=/usr/lib/ccache/g++ -DYUZU_ENABLE_COMPATIBILITY_REPORTING=${ENABLE_COMPATIBILITY_REPORTING:-"OFF"} -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON -DUSE_DISCORD_PRESENCE=ON -G Ninja
 ninja
 
 ccache -s

.travis/macos/build.sh

@@ -9,7 +9,7 @@ export PATH="/usr/local/opt/ccache/libexec:$PATH"
 
 mkdir build && cd build
 cmake --version
-cmake .. -DYUZU_USE_BUNDLED_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 -DYUZU_ENABLE_COMPATIBILITY_REPORTING=${ENABLE_COMPATIBILITY_REPORTING:-"OFF"} -DUSE_DISCORD_PRESENCE=ON
 make -j4
 
 ccache -s

CMakeLists.txt

@@ -15,10 +15,14 @@ CMAKE_DEPENDENT_OPTION(YUZU_USE_BUNDLED_SDL2 "Download bundled SDL2 binaries" ON
 option(ENABLE_QT "Enable the Qt frontend" ON)
 CMAKE_DEPENDENT_OPTION(YUZU_USE_BUNDLED_QT "Download bundled Qt binaries" ON "ENABLE_QT;MSVC" OFF)
 
+option(ENABLE_WEB_SERVICE "Enable web services (telemetry, etc.)" ON)
+
 option(YUZU_USE_BUNDLED_UNICORN "Build/Download bundled Unicorn" ON)
 
 option(ENABLE_CUBEB "Enables the cubeb audio backend" ON)
 
+option(USE_DISCORD_PRESENCE "Enables Discord Rich Presence" OFF)
+
 if(NOT EXISTS ${CMAKE_SOURCE_DIR}/.git/hooks/pre-commit)
     message(STATUS "Copying pre-commit hook")
     file(COPY hooks/pre-commit

CONTRIBUTING.md

@@ -22,7 +22,7 @@ If clang format is found, then cmake will add a custom build target that can be
 * Don't ever introduce new external dependencies into Core
 * Don't use any platform specific code in Core
 * Use namespaces often
-* Avoid the use of C-style casts and instead prefer C++-style `static_cast` and `reinterpret_cast`. Try to avoid using `dynamic_cast`. Never use `const_cast`. The only exception to this rule is for casting between two numeric types, where C-style casts are encouraged for brevity and readability.
+* Avoid the use of C-style casts and instead prefer C++-style `static_cast` and `reinterpret_cast`. Try to avoid using `dynamic_cast`. Never use `const_cast`.
 
 ### Naming Rules
 * Functions: `PascalCase`

appveyor.yml

@@ -39,11 +39,12 @@ before_build:
   - mkdir %BUILD_TYPE%_build
   - cd %BUILD_TYPE%_build
   - ps: |
+        $COMPAT = if ($env:ENABLE_COMPATIBILITY_REPORTING -eq $null) {0} else {$env:ENABLE_COMPATIBILITY_REPORTING}
         if ($env:BUILD_TYPE -eq 'msvc') {
           # redirect stderr and change the exit code to prevent powershell from cancelling the build if cmake prints a warning
-          cmd /C 'cmake -G "Visual Studio 15 2017 Win64" -DYUZU_USE_BUNDLED_QT=1 -DYUZU_USE_BUNDLED_SDL2=1 -DYUZU_USE_BUNDLED_UNICORN=1 -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON .. 2>&1 && exit 0'
+          cmd /C 'cmake -G "Visual Studio 15 2017 Win64" -DYUZU_USE_BUNDLED_QT=1 -DYUZU_USE_BUNDLED_SDL2=1 -DYUZU_USE_BUNDLED_UNICORN=1 -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON -DYUZU_ENABLE_COMPATIBILITY_REPORTING=${COMPAT} -DUSE_DISCORD_PRESENCE=ON .. 2>&1 && exit 0'
         } else {
-          C:\msys64\usr\bin\bash.exe -lc "cmake -G 'MSYS Makefiles' -DYUZU_BUILD_UNICORN=1 -DCMAKE_BUILD_TYPE=Release -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON .. 2>&1"
+          C:\msys64\usr\bin\bash.exe -lc "cmake -G 'MSYS Makefiles' -DYUZU_BUILD_UNICORN=1 -DCMAKE_BUILD_TYPE=Release -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON -DYUZU_ENABLE_COMPATIBILITY_REPORTING=${COMPAT} -DUSE_DISCORD_PRESENCE=ON .. 2>&1"
         }
   - cd ..
 

externals/CMakeLists.txt

@@ -70,3 +70,28 @@ if(ENABLE_CUBEB)
     set(BUILD_TESTS OFF CACHE BOOL "")
     add_subdirectory(cubeb EXCLUDE_FROM_ALL)
 endif()
+
+# DiscordRPC
+if (USE_DISCORD_PRESENCE)
+    add_subdirectory(discord-rpc EXCLUDE_FROM_ALL)
+    target_include_directories(discord-rpc INTERFACE ./discord-rpc/include)
+endif()
+
+if (ENABLE_WEB_SERVICE)
+    # LibreSSL
+    set(LIBRESSL_SKIP_INSTALL ON CACHE BOOL "")
+    add_subdirectory(libressl EXCLUDE_FROM_ALL)
+    target_include_directories(ssl INTERFACE ./libressl/include)
+	target_compile_definitions(ssl PRIVATE -DHAVE_INET_NTOP)
+
+    # lurlparser
+    add_subdirectory(lurlparser EXCLUDE_FROM_ALL)
+
+    # httplib
+    add_library(httplib INTERFACE)
+    target_include_directories(httplib INTERFACE ./httplib)
+
+    # JSON
+    add_library(json-headers INTERFACE)
+    target_include_directories(json-headers INTERFACE ./json)
+endif()

externals/httplib/README.md

@@ -0,0 +1,15 @@
+From https://github.com/yhirose/cpp-httplib/commit/d9479bc0b12e8a1e8bce2d34da4feeef488581f3
+
+MIT License
+
+===
+
+cpp-httplib
+
+A C++11 header-only HTTP library.
+
+It's extremely easy to setup. Just include httplib.h file in your code!
+
+Inspired by Sinatra and express.
+
+© 2017 Yuji Hirose

externals/json/README.md

@@ -0,0 +1,9 @@
+JSON for Modern C++
+===================
+
+v3.1.2
+
+This is a mirror providing the single required header file.
+
+The original repository can be found at:
+https://github.com/nlohmann/json/commit/d2dd27dc3b8472dbaa7d66f83619b3ebcd9185fe

externals/lurlparser/CMakeLists.txt

@@ -0,0 +1,8 @@
+add_library(lurlparser
+        LUrlParser.cpp
+        LUrlParser.h
+)
+
+create_target_directory_groups(lurlparser)
+
+target_include_directories(lurlparser INTERFACE ${CMAKE_CURRENT_SOURCE_DIR})

externals/lurlparser/LUrlParser.cpp

@@ -0,0 +1,265 @@
+/*
+ * Lightweight URL & URI parser (RFC 1738, RFC 3986)
+ * https://github.com/corporateshark/LUrlParser
+ *
+ * The MIT License (MIT)
+ *
+ * Copyright (C) 2015 Sergey Kosarevsky (sk@linderdaum.com)
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "LUrlParser.h"
+
+#include <algorithm>
+#include <cstring>
+#include <stdlib.h>
+
+// check if the scheme name is valid
+static bool IsSchemeValid( const std::string& SchemeName )
+{
+    for ( auto c : SchemeName  )
+    {
+        if ( !isalpha( c ) && c != '+' && c != '-' && c != '.' ) return false;
+    }
+
+    return true;
+}
+
+bool LUrlParser::clParseURL::GetPort( int* OutPort ) const
+{
+    if ( !IsValid() ) { return false; }
+
+    int Port = atoi( m_Port.c_str() );
+
+    if ( Port <= 0 || Port > 65535 ) { return false; }
+
+    if ( OutPort ) { *OutPort = Port; }
+
+    return true;
+}
+
+// based on RFC 1738 and RFC 3986
+LUrlParser::clParseURL LUrlParser::clParseURL::ParseURL( const std::string& URL )
+{
+    LUrlParser::clParseURL Result;
+
+    const char* CurrentString = URL.c_str();
+
+    /*
+     *	<scheme>:<scheme-specific-part>
+     *	<scheme> := [a-z\+\-\.]+
+     *	For resiliency, programs interpreting URLs should treat upper case letters as equivalent to lower case in scheme names
+     */
+
+    // try to read scheme
+    {
+        const char* LocalString = strchr( CurrentString, ':' );
+
+        if ( !LocalString )
+        {
+            return clParseURL( LUrlParserError_NoUrlCharacter );
+        }
+
+        // save the scheme name
+        Result.m_Scheme = std::string( CurrentString, LocalString - CurrentString );
+
+        if ( !IsSchemeValid( Result.m_Scheme ) )
+        {
+            return clParseURL( LUrlParserError_InvalidSchemeName );
+        }
+
+        // scheme should be lowercase
+        std::transform( Result.m_Scheme.begin(), Result.m_Scheme.end(), Result.m_Scheme.begin(), ::tolower );
+
+        // skip ':'
+        CurrentString = LocalString+1;
+    }
+
+    /*
+     *	//<user>:<password>@<host>:<port>/<url-path>
+     *	any ":", "@" and "/" must be normalized
+     */
+
+    // skip "//"
+    if ( *CurrentString++ != '/' ) return clParseURL( LUrlParserError_NoDoubleSlash );
+    if ( *CurrentString++ != '/' ) return clParseURL( LUrlParserError_NoDoubleSlash );
+
+    // check if the user name and password are specified
+    bool bHasUserName = false;
+
+    const char* LocalString = CurrentString;
+
+    while ( *LocalString )
+    {
+        if ( *LocalString == '@' )
+        {
+            // user name and password are specified
+            bHasUserName = true;
+            break;
+        }
+        else if ( *LocalString == '/' )
+        {
+            // end of <host>:<port> specification
+            bHasUserName = false;
+            break;
+        }
+
+        LocalString++;
+    }
+
+    // user name and password
+    LocalString = CurrentString;
+
+    if ( bHasUserName )
+    {
+        // read user name
+        while ( *LocalString && *LocalString != ':' && *LocalString != '@' ) LocalString++;
+
+        Result.m_UserName = std::string( CurrentString, LocalString - CurrentString );
+
+        // proceed with the current pointer
+        CurrentString = LocalString;
+
+        if ( *CurrentString == ':' )
+        {
+            // skip ':'
+            CurrentString++;
+
+            // read password
+            LocalString = CurrentString;
+
+            while ( *LocalString && *LocalString != '@' ) LocalString++;
+
+            Result.m_Password = std::string( CurrentString, LocalString - CurrentString );
+
+            CurrentString = LocalString;
+        }
+
+        // skip '@'
+        if ( *CurrentString != '@' )
+        {
+            return clParseURL( LUrlParserError_NoAtSign );
+        }
+
+        CurrentString++;
+    }
+
+    bool bHasBracket = ( *CurrentString == '[' );
+
+    // go ahead, read the host name
+    LocalString = CurrentString;
+
+    while ( *LocalString )
+    {
+        if ( bHasBracket && *LocalString == ']' )
+        {
+            // end of IPv6 address
+            LocalString++;
+            break;
+        }
+        else if ( !bHasBracket && ( *LocalString == ':' || *LocalString == '/' ) )
+        {
+            // port number is specified
+            break;
+        }
+
+        LocalString++;
+    }
+
+    Result.m_Host = std::string( CurrentString, LocalString - CurrentString );
+
+    CurrentString = LocalString;
+
+    // is port number specified?
+    if ( *CurrentString == ':' )
+    {
+        CurrentString++;
+
+        // read port number
+        LocalString = CurrentString;
+
+        while ( *LocalString && *LocalString != '/' ) LocalString++;
+
+        Result.m_Port = std::string( CurrentString, LocalString - CurrentString );
+
+        CurrentString = LocalString;
+    }
+
+    // end of string
+    if ( !*CurrentString )
+    {
+        Result.m_ErrorCode = LUrlParserError_Ok;
+
+        return Result;
+    }
+
+    // skip '/'
+    if ( *CurrentString != '/' )
+    {
+        return clParseURL( LUrlParserError_NoSlash );
+    }
+
+    CurrentString++;
+
+    // parse the path
+    LocalString = CurrentString;
+
+    while ( *LocalString && *LocalString != '#' && *LocalString != '?' ) LocalString++;
+
+    Result.m_Path = std::string( CurrentString, LocalString - CurrentString );
+
+    CurrentString = LocalString;
+
+    // check for query
+    if ( *CurrentString == '?' )
+    {
+        // skip '?'
+        CurrentString++;
+
+        // read query
+        LocalString = CurrentString;
+
+        while ( *LocalString && *LocalString != '#' ) LocalString++;
+
+        Result.m_Query = std::string( CurrentString, LocalString - CurrentString );
+
+        CurrentString = LocalString;
+    }
+
+    // check for fragment
+    if ( *CurrentString == '#' )
+    {
+        // skip '#'
+        CurrentString++;
+
+        // read fragment
+        LocalString = CurrentString;
+
+        while ( *LocalString ) LocalString++;
+
+        Result.m_Fragment = std::string( CurrentString, LocalString - CurrentString );
+
+        CurrentString = LocalString;
+    }
+
+    Result.m_ErrorCode = LUrlParserError_Ok;
+
+    return Result;
+}

externals/lurlparser/LUrlParser.h

@@ -0,0 +1,78 @@
+/*
+ * Lightweight URL & URI parser (RFC 1738, RFC 3986)
+ * https://github.com/corporateshark/LUrlParser
+ *
+ * The MIT License (MIT)
+ *
+ * Copyright (C) 2015 Sergey Kosarevsky (sk@linderdaum.com)
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#pragma once
+
+#include <string>
+
+namespace LUrlParser
+{
+enum LUrlParserError
+{
+    LUrlParserError_Ok = 0,
+    LUrlParserError_Uninitialized = 1,
+    LUrlParserError_NoUrlCharacter = 2,
+    LUrlParserError_InvalidSchemeName = 3,
+    LUrlParserError_NoDoubleSlash = 4,
+    LUrlParserError_NoAtSign = 5,
+    LUrlParserError_UnexpectedEndOfLine = 6,
+    LUrlParserError_NoSlash = 7,
+};
+
+class clParseURL
+{
+public:
+    LUrlParserError m_ErrorCode;
+    std::string m_Scheme;
+    std::string m_Host;
+    std::string m_Port;
+    std::string m_Path;
+    std::string m_Query;
+    std::string m_Fragment;
+    std::string m_UserName;
+    std::string m_Password;
+
+    clParseURL()
+            : m_ErrorCode( LUrlParserError_Uninitialized )
+    {}
+
+    /// return 'true' if the parsing was successful
+    bool IsValid() const { return m_ErrorCode == LUrlParserError_Ok; }
+
+    /// helper to convert the port number to int, return 'true' if the port is valid (within the 0..65535 range)
+    bool GetPort( int* OutPort ) const;
+
+    /// parse the URL
+    static clParseURL ParseURL( const std::string& URL );
+
+private:
+    explicit clParseURL( LUrlParserError ErrorCode )
+            : m_ErrorCode( ErrorCode )
+    {}
+};
+
+} // namespace LUrlParser

externals/lurlparser/README.md

@@ -0,0 +1,19 @@
+From https://github.com/corporateshark/LUrlParser/commit/455d5e2d27e3946f11ad0328fee9ee2628e6a8e2
+
+MIT License
+
+===
+
+Lightweight URL & URI parser (RFC 1738, RFC 3986)
+
+(C) Sergey Kosarevsky, 2015
+
+@corporateshark sk@linderdaum.com
+
+http://www.linderdaum.com
+
+http://blog.linderdaum.com
+
+=============================
+
+A tiny and lightweight URL & URI parser (RFC 1738, RFC 3986) written in C++.

src/CMakeLists.txt

@@ -13,3 +13,6 @@ endif()
 if (ENABLE_QT)
     add_subdirectory(yuzu)
 endif()
+if (ENABLE_WEB_SERVICE)
+    add_subdirectory(web_service)
+endif()

src/audio_core/audio_renderer.cpp

@@ -30,7 +30,7 @@ public:
         return info;
     }
 
-    VoiceInfo& Info() {
+    VoiceInfo& GetInfo() {
         return info;
     }
 
@@ -51,9 +51,30 @@ private:
     VoiceInfo info{};
 };
 
+class AudioRenderer::EffectState {
+public:
+    const EffectOutStatus& GetOutStatus() const {
+        return out_status;
+    }
+
+    const EffectInStatus& GetInfo() const {
+        return info;
+    }
+
+    EffectInStatus& GetInfo() {
+        return info;
+    }
+
+    void UpdateState();
+
+private:
+    EffectOutStatus out_status{};
+    EffectInStatus info{};
+};
 AudioRenderer::AudioRenderer(AudioRendererParameter params,
                              Kernel::SharedPtr<Kernel::Event> buffer_event)
-    : worker_params{params}, buffer_event{buffer_event}, voices(params.voice_count) {
+    : worker_params{params}, buffer_event{buffer_event}, voices(params.voice_count),
+      effects(params.effect_count) {
 
     audio_out = std::make_unique<AudioCore::AudioOut>();
     stream = audio_out->OpenStream(STREAM_SAMPLE_RATE, STREAM_NUM_CHANNELS, "AudioRenderer",
@@ -96,11 +117,29 @@ std::vector<u8> AudioRenderer::UpdateAudioRenderer(const std::vector<u8>& input_
                 memory_pool_count * sizeof(MemoryPoolInfo));
 
     // Copy VoiceInfo structs
-    std::size_t offset{sizeof(UpdateDataHeader) + config.behavior_size + config.memory_pools_size +
-                       config.voice_resource_size};
+    std::size_t voice_offset{sizeof(UpdateDataHeader) + config.behavior_size +
+                             config.memory_pools_size + config.voice_resource_size};
     for (auto& voice : voices) {
-        std::memcpy(&voice.Info(), input_params.data() + offset, sizeof(VoiceInfo));
-        offset += sizeof(VoiceInfo);
+        std::memcpy(&voice.GetInfo(), input_params.data() + voice_offset, sizeof(VoiceInfo));
+        voice_offset += sizeof(VoiceInfo);
+    }
+
+    std::size_t effect_offset{sizeof(UpdateDataHeader) + config.behavior_size +
+                              config.memory_pools_size + config.voice_resource_size +
+                              config.voices_size};
+    for (auto& effect : effects) {
+        std::memcpy(&effect.GetInfo(), input_params.data() + effect_offset, sizeof(EffectInStatus));
+        effect_offset += sizeof(EffectInStatus);
+    }
+
+    // Update memory pool state
+    std::vector<MemoryPoolEntry> memory_pool(memory_pool_count);
+    for (std::size_t index = 0; index < memory_pool.size(); ++index) {
+        if (mem_pool_info[index].pool_state == MemoryPoolStates::RequestAttach) {
+            memory_pool[index].state = MemoryPoolStates::Attached;
+        } else if (mem_pool_info[index].pool_state == MemoryPoolStates::RequestDetach) {
+            memory_pool[index].state = MemoryPoolStates::Detached;
+        }
     }
 
     // Update voices
@@ -114,14 +153,8 @@ std::vector<u8> AudioRenderer::UpdateAudioRenderer(const std::vector<u8>& input_
         }
     }
 
-    // Update memory pool state
-    std::vector<MemoryPoolEntry> memory_pool(memory_pool_count);
-    for (std::size_t index = 0; index < memory_pool.size(); ++index) {
-        if (mem_pool_info[index].pool_state == MemoryPoolStates::RequestAttach) {
-            memory_pool[index].state = MemoryPoolStates::Attached;
-        } else if (mem_pool_info[index].pool_state == MemoryPoolStates::RequestDetach) {
-            memory_pool[index].state = MemoryPoolStates::Detached;
-        }
+    for (auto& effect : effects) {
+        effect.UpdateState();
     }
 
     // Release previous buffers and queue next ones for playback
@@ -144,6 +177,14 @@ std::vector<u8> AudioRenderer::UpdateAudioRenderer(const std::vector<u8>& input_
         voice_out_status_offset += sizeof(VoiceOutStatus);
     }
 
+    std::size_t effect_out_status_offset{
+        sizeof(UpdateDataHeader) + response_data.memory_pools_size + response_data.voices_size +
+        response_data.voice_resource_size};
+    for (const auto& effect : effects) {
+        std::memcpy(output_params.data() + effect_out_status_offset, &effect.GetOutStatus(),
+                    sizeof(EffectOutStatus));
+        effect_out_status_offset += sizeof(EffectOutStatus);
+    }
     return output_params;
 }
 
@@ -244,11 +285,29 @@ void AudioRenderer::VoiceState::RefreshBuffer() {
         break;
     }
 
-    samples = Interpolate(interp_state, std::move(samples), Info().sample_rate, STREAM_SAMPLE_RATE);
+    samples =
+        Interpolate(interp_state, std::move(samples), GetInfo().sample_rate, STREAM_SAMPLE_RATE);
 
     is_refresh_pending = false;
 }
 
+void AudioRenderer::EffectState::UpdateState() {
+    if (info.is_new) {
+        out_status.state = EffectStatus::New;
+    } else {
+        if (info.type == Effect::Aux) {
+            ASSERT_MSG(Memory::Read32(info.aux_info.return_buffer_info) == 0,
+                       "Aux buffers tried to update");
+            ASSERT_MSG(Memory::Read32(info.aux_info.send_buffer_info) == 0,
+                       "Aux buffers tried to update");
+            ASSERT_MSG(Memory::Read32(info.aux_info.return_buffer_base) == 0,
+                       "Aux buffers tried to update");
+            ASSERT_MSG(Memory::Read32(info.aux_info.send_buffer_base) == 0,
+                       "Aux buffers tried to update");
+        }
+    }
+}
+
 static constexpr s16 ClampToS16(s32 value) {
     return static_cast<s16>(std::clamp(value, -32768, 32767));
 }

src/audio_core/audio_renderer.h

@@ -28,6 +28,16 @@ enum class PlayState : u8 {
     Paused = 2,
 };
 
+enum class Effect : u8 {
+    None = 0,
+    Aux = 2,
+};
+
+enum class EffectStatus : u8 {
+    None = 0,
+    New = 1,
+};
+
 struct AudioRendererParameter {
     u32_le sample_rate;
     u32_le sample_count;
@@ -128,6 +138,43 @@ struct VoiceOutStatus {
 };
 static_assert(sizeof(VoiceOutStatus) == 0x10, "VoiceOutStatus has wrong size");
 
+struct AuxInfo {
+    std::array<u8, 24> input_mix_buffers;
+    std::array<u8, 24> output_mix_buffers;
+    u32_le mix_buffer_count;
+    u32_le sample_rate; // Stored in the aux buffer currently
+    u32_le sampe_count;
+    u64_le send_buffer_info;
+    u64_le send_buffer_base;
+
+    u64_le return_buffer_info;
+    u64_le return_buffer_base;
+};
+static_assert(sizeof(AuxInfo) == 0x60, "AuxInfo is an invalid size");
+
+struct EffectInStatus {
+    Effect type;
+    u8 is_new;
+    u8 is_enabled;
+    INSERT_PADDING_BYTES(1);
+    u32_le mix_id;
+    u64_le buffer_base;
+    u64_le buffer_sz;
+    s32_le priority;
+    INSERT_PADDING_BYTES(4);
+    union {
+        std::array<u8, 0xa0> raw;
+        AuxInfo aux_info;
+    };
+};
+static_assert(sizeof(EffectInStatus) == 0xc0, "EffectInStatus is an invalid size");
+
+struct EffectOutStatus {
+    EffectStatus state;
+    INSERT_PADDING_BYTES(0xf);
+};
+static_assert(sizeof(EffectOutStatus) == 0x10, "EffectOutStatus is an invalid size");
+
 struct UpdateDataHeader {
     UpdateDataHeader() {}
 
@@ -173,11 +220,13 @@ public:
     Stream::State GetStreamState() const;
 
 private:
+    class EffectState;
     class VoiceState;
 
     AudioRendererParameter worker_params;
     Kernel::SharedPtr<Kernel::Event> buffer_event;
     std::vector<VoiceState> voices;
+    std::vector<EffectState> effects;
     std::unique_ptr<AudioOut> audio_out;
     AudioCore::StreamPtr stream;
 };

src/common/CMakeLists.txt

@@ -29,7 +29,7 @@ if ($ENV{CI})
       if (BUILD_VERSION)
         # This leaves a trailing space on the last word, but we actually want that
         # because of how it's styled in the title bar.
-        set(BUILD_FULLNAME "${REPO_NAME} #${BUILD_VERSION} ")
+        set(BUILD_FULLNAME "${REPO_NAME} ${BUILD_VERSION} ")
       else()
         set(BUILD_FULLNAME "")
       endif()
@@ -41,6 +41,8 @@ configure_file("${CMAKE_CURRENT_SOURCE_DIR}/scm_rev.cpp.in" "${CMAKE_CURRENT_SOU
 add_library(common STATIC
     alignment.h
     assert.h
+    detached_tasks.cpp
+    detached_tasks.h
     bit_field.h
     bit_set.h
     cityhash.cpp
@@ -87,6 +89,7 @@ add_library(common STATIC
     timer.cpp
     timer.h
     vector_math.h
+    web_result.h
 )
 
 if(ARCHITECTURE_x86_64)

src/common/detached_tasks.cpp

@@ -0,0 +1,41 @@
+// Copyright 2018 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <thread>
+#include "common/assert.h"
+#include "common/detached_tasks.h"
+
+namespace Common {
+
+DetachedTasks* DetachedTasks::instance = nullptr;
+
+DetachedTasks::DetachedTasks() {
+    ASSERT(instance == nullptr);
+    instance = this;
+}
+
+void DetachedTasks::WaitForAllTasks() {
+    std::unique_lock<std::mutex> lock(mutex);
+    cv.wait(lock, [this]() { return count == 0; });
+}
+
+DetachedTasks::~DetachedTasks() {
+    std::unique_lock<std::mutex> lock(mutex);
+    ASSERT(count == 0);
+    instance = nullptr;
+}
+
+void DetachedTasks::AddTask(std::function<void()> task) {
+    std::unique_lock<std::mutex> lock(instance->mutex);
+    ++instance->count;
+    std::thread([task{std::move(task)}]() {
+        task();
+        std::unique_lock<std::mutex> lock(instance->mutex);
+        --instance->count;
+        std::notify_all_at_thread_exit(instance->cv, std::move(lock));
+    })
+        .detach();
+}
+
+} // namespace Common

src/common/detached_tasks.h

@@ -0,0 +1,40 @@
+// Copyright 2018 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <condition_variable>
+#include <functional>
+
+namespace Common {
+
+/**
+ * A background manager which ensures that all detached task is finished before program exits.
+ *
+ * Some tasks, telemetry submission for example, prefer executing asynchronously and don't care
+ * about the result. These tasks are suitable for std::thread::detach(). However, this is unsafe if
+ * the task is launched just before the program exits (which is a common case for telemetry), so we
+ * need to block on these tasks on program exit.
+ *
+ * To make detached task safe, a single DetachedTasks object should be placed in the main(), and
+ * call WaitForAllTasks() after all program execution but before global/static variable destruction.
+ * Any potentially unsafe detached task should be executed via DetachedTasks::AddTask.
+ */
+class DetachedTasks {
+public:
+    DetachedTasks();
+    ~DetachedTasks();
+    void WaitForAllTasks();
+
+    static void AddTask(std::function<void()> task);
+
+private:
+    static DetachedTasks* instance;
+
+    std::condition_variable cv;
+    std::mutex mutex;
+    int count = 0;
+};
+
+} // namespace Common

src/common/logging/text_formatter.cpp

@@ -31,7 +31,7 @@ std::string FormatLogMessage(const Entry& entry) {
 }
 
 void PrintMessage(const Entry& entry) {
-    auto str = FormatLogMessage(entry) + '\n';
+    const auto str = FormatLogMessage(entry).append(1, '\n');
     fputs(str.c_str(), stderr);
 }
 

src/common/param_package.cpp

@@ -20,7 +20,15 @@ constexpr char KEY_VALUE_SEPARATOR_ESCAPE[] = "$0";
 constexpr char PARAM_SEPARATOR_ESCAPE[] = "$1";
 constexpr char ESCAPE_CHARACTER_ESCAPE[] = "$2";
 
+/// A placeholder for empty param packages to avoid empty strings
+/// (they may be recognized as "not set" by some frontend libraries like qt)
+constexpr char EMPTY_PLACEHOLDER[] = "[empty]";
+
 ParamPackage::ParamPackage(const std::string& serialized) {
+    if (serialized == EMPTY_PLACEHOLDER) {
+        return;
+    }
+
     std::vector<std::string> pairs;
     Common::SplitString(serialized, PARAM_SEPARATOR, pairs);
 
@@ -46,7 +54,7 @@ ParamPackage::ParamPackage(std::initializer_list<DataType::value_type> list) : d
 
 std::string ParamPackage::Serialize() const {
     if (data.empty())
-        return "";
+        return EMPTY_PLACEHOLDER;
 
     std::string result;
 
@@ -120,4 +128,12 @@ bool ParamPackage::Has(const std::string& key) const {
     return data.find(key) != data.end();
 }
 
+void ParamPackage::Erase(const std::string& key) {
+    data.erase(key);
+}
+
+void ParamPackage::Clear() {
+    data.clear();
+}
+
 } // namespace Common

src/common/param_package.h

@@ -32,6 +32,8 @@ public:
     void Set(const std::string& key, int value);
     void Set(const std::string& key, float value);
     bool Has(const std::string& key) const;
+    void Erase(const std::string& key);
+    void Clear();
 
 private:
     DataType data;

src/common/web_result.h

@@ -0,0 +1,24 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <string>
+
+namespace Common {
+struct WebResult {
+    enum class Code : u32 {
+        Success,
+        InvalidURL,
+        CredentialsMissing,
+        LibError,
+        HttpError,
+        WrongContent,
+        NoWebservice,
+    };
+    Code result_code;
+    std::string result_string;
+    std::string returned_data;
+};
+} // namespace Common

src/core/CMakeLists.txt

@@ -396,6 +396,10 @@ create_target_directory_groups(core)
 
 target_link_libraries(core PUBLIC common PRIVATE audio_core video_core)
 target_link_libraries(core PUBLIC Boost::boost PRIVATE fmt lz4_static mbedtls opus unicorn open_source_archives)
+if (ENABLE_WEB_SERVICE)
+    add_definitions(-DENABLE_WEB_SERVICE)
+    target_link_libraries(core PUBLIC json-headers web_service)
+endif()
 
 if (ARCHITECTURE_x86_64)
     target_sources(core PRIVATE

src/core/arm/dynarmic/arm_dynarmic.cpp

@@ -86,7 +86,7 @@ public:
                 parent.jit->HaltExecution();
                 parent.SetPC(pc);
                 Kernel::Thread* thread = Kernel::GetCurrentThread();
-                parent.SaveContext(thread->context);
+                parent.SaveContext(thread->GetContext());
                 GDBStub::Break();
                 GDBStub::SendTrap(thread, 5);
                 return;
@@ -129,7 +129,7 @@ public:
 };
 
 std::unique_ptr<Dynarmic::A64::Jit> ARM_Dynarmic::MakeJit() const {
-    auto& current_process = Core::CurrentProcess();
+    auto* current_process = Core::CurrentProcess();
     auto** const page_table = current_process->VMManager().page_table.pointers.data();
 
     Dynarmic::A64::UserConfig config;

src/core/arm/unicorn/arm_unicorn.cpp

@@ -195,7 +195,7 @@ void ARM_Unicorn::ExecuteInstructions(int num_instructions) {
             uc_reg_write(uc, UC_ARM64_REG_PC, &last_bkpt.address);
         }
         Kernel::Thread* thread = Kernel::GetCurrentThread();
-        SaveContext(thread->context);
+        SaveContext(thread->GetContext());
         if (last_bkpt_hit || GDBStub::GetCpuStepFlag()) {
             last_bkpt_hit = false;
             GDBStub::Break();

src/core/core.cpp

@@ -136,7 +136,8 @@ struct System::Impl {
         if (virtual_filesystem == nullptr)
             virtual_filesystem = std::make_shared<FileSys::RealVfsFilesystem>();
 
-        kernel.MakeCurrentProcess(Kernel::Process::Create(kernel, "main"));
+        auto main_process = Kernel::Process::Create(kernel, "main");
+        kernel.MakeCurrentProcess(main_process.get());
 
         cpu_barrier = std::make_shared<CpuBarrier>();
         cpu_exclusive_monitor = Cpu::MakeExclusiveMonitor(cpu_cores.size());
@@ -361,11 +362,11 @@ const std::shared_ptr<Kernel::Scheduler>& System::Scheduler(std::size_t core_ind
     return impl->cpu_cores[core_index]->Scheduler();
 }
 
-Kernel::SharedPtr<Kernel::Process>& System::CurrentProcess() {
+Kernel::Process* System::CurrentProcess() {
     return impl->kernel.CurrentProcess();
 }
 
-const Kernel::SharedPtr<Kernel::Process>& System::CurrentProcess() const {
+const Kernel::Process* System::CurrentProcess() const {
     return impl->kernel.CurrentProcess();
 }
 

src/core/core.h

@@ -174,11 +174,11 @@ public:
     /// Gets the scheduler for the CPU core with the specified index
     const std::shared_ptr<Kernel::Scheduler>& Scheduler(std::size_t core_index);
 
-    /// Provides a reference to the current process
-    Kernel::SharedPtr<Kernel::Process>& CurrentProcess();
+    /// Provides a pointer to the current process
+    Kernel::Process* CurrentProcess();
 
-    /// Provides a constant reference to the current process.
-    const Kernel::SharedPtr<Kernel::Process>& CurrentProcess() const;
+    /// Provides a constant pointer to the current process.
+    const Kernel::Process* CurrentProcess() const;
 
     /// Provides a reference to the kernel instance.
     Kernel::KernelCore& Kernel();
@@ -246,7 +246,7 @@ inline TelemetrySession& Telemetry() {
     return System::GetInstance().TelemetrySession();
 }
 
-inline Kernel::SharedPtr<Kernel::Process>& CurrentProcess() {
+inline Kernel::Process* CurrentProcess() {
     return System::GetInstance().CurrentProcess();
 }
 

src/core/file_sys/fsmitm_romfsbuild.cpp

@@ -26,6 +26,7 @@
 #include "common/alignment.h"
 #include "common/assert.h"
 #include "core/file_sys/fsmitm_romfsbuild.h"
+#include "core/file_sys/ips_layer.h"
 #include "core/file_sys/vfs.h"
 #include "core/file_sys/vfs_vector.h"
 
@@ -123,7 +124,7 @@ static u64 romfs_get_hash_table_count(u64 num_entries) {
     return count;
 }
 
-void RomFSBuildContext::VisitDirectory(VirtualDir root_romfs,
+void RomFSBuildContext::VisitDirectory(VirtualDir root_romfs, VirtualDir ext,
                                        std::shared_ptr<RomFSBuildDirectoryContext> parent) {
     std::vector<std::shared_ptr<RomFSBuildDirectoryContext>> child_dirs;
 
@@ -144,6 +145,9 @@ void RomFSBuildContext::VisitDirectory(VirtualDir root_romfs,
             child->path_len = child->cur_path_ofs + static_cast<u32>(kv.first.size());
             child->path = parent->path + "/" + kv.first;
 
+            if (ext != nullptr && ext->GetFileRelative(child->path + ".stub") != nullptr)
+                continue;
+
             // Sanity check on path_len
             ASSERT(child->path_len < FS_MAX_PATH);
 
@@ -157,11 +161,24 @@ void RomFSBuildContext::VisitDirectory(VirtualDir root_romfs,
             child->path_len = child->cur_path_ofs + static_cast<u32>(kv.first.size());
             child->path = parent->path + "/" + kv.first;
 
+            if (ext != nullptr && ext->GetFileRelative(child->path + ".stub") != nullptr)
+                continue;
+
             // Sanity check on path_len
             ASSERT(child->path_len < FS_MAX_PATH);
 
             child->source = root_romfs->GetFileRelative(child->path);
 
+            if (ext != nullptr) {
+                const auto ips = ext->GetFileRelative(child->path + ".ips");
+
+                if (ips != nullptr) {
+                    auto patched = PatchIPS(child->source, ips);
+                    if (patched != nullptr)
+                        child->source = std::move(patched);
+                }
+            }
+
             child->size = child->source->GetSize();
 
             AddFile(parent, child);
@@ -169,7 +186,7 @@ void RomFSBuildContext::VisitDirectory(VirtualDir root_romfs,
     }
 
     for (auto& child : child_dirs) {
-        this->VisitDirectory(root_romfs, child);
+        this->VisitDirectory(root_romfs, ext, child);
     }
 }
 
@@ -208,14 +225,15 @@ bool RomFSBuildContext::AddFile(std::shared_ptr<RomFSBuildDirectoryContext> pare
     return true;
 }
 
-RomFSBuildContext::RomFSBuildContext(VirtualDir base_) : base(std::move(base_)) {
+RomFSBuildContext::RomFSBuildContext(VirtualDir base_, VirtualDir ext_)
+    : base(std::move(base_)), ext(std::move(ext_)) {
     root = std::make_shared<RomFSBuildDirectoryContext>();
     root->path = "\0";
     directories.emplace(root->path, root);
     num_dirs = 1;
     dir_table_size = 0x18;
 
-    VisitDirectory(base, root);
+    VisitDirectory(base, ext, root);
 }
 
 RomFSBuildContext::~RomFSBuildContext() = default;

src/core/file_sys/fsmitm_romfsbuild.h

@@ -40,7 +40,7 @@ struct RomFSFileEntry;
 
 class RomFSBuildContext {
 public:
-    explicit RomFSBuildContext(VirtualDir base);
+    explicit RomFSBuildContext(VirtualDir base, VirtualDir ext = nullptr);
     ~RomFSBuildContext();
 
     // This finalizes the context.
@@ -48,6 +48,7 @@ public:
 
 private:
     VirtualDir base;
+    VirtualDir ext;
     std::shared_ptr<RomFSBuildDirectoryContext> root;
     std::map<std::string, std::shared_ptr<RomFSBuildDirectoryContext>, std::less<>> directories;
     std::map<std::string, std::shared_ptr<RomFSBuildFileContext>, std::less<>> files;
@@ -59,7 +60,8 @@ private:
     u64 file_hash_table_size = 0;
     u64 file_partition_size = 0;
 
-    void VisitDirectory(VirtualDir filesys, std::shared_ptr<RomFSBuildDirectoryContext> parent);
+    void VisitDirectory(VirtualDir filesys, VirtualDir ext,
+                        std::shared_ptr<RomFSBuildDirectoryContext> parent);
 
     bool AddDirectory(std::shared_ptr<RomFSBuildDirectoryContext> parent_dir_ctx,
                       std::shared_ptr<RomFSBuildDirectoryContext> dir_ctx);

src/core/file_sys/ips_layer.cpp

@@ -2,9 +2,15 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <algorithm>
+#include <cstring>
+#include <map>
 #include <sstream>
-#include "common/assert.h"
+#include <string>
+#include <utility>
+
 #include "common/hex_util.h"
+#include "common/logging/log.h"
 #include "common/swap.h"
 #include "core/file_sys/ips_layer.h"
 #include "core/file_sys/vfs_vector.h"
@@ -17,22 +23,48 @@ enum class IPSFileType {
     Error,
 };
 
-constexpr std::array<std::pair<const char*, const char*>, 11> ESCAPE_CHARACTER_MAP{
-    std::pair{"\\a", "\a"}, {"\\b", "\b"},  {"\\f", "\f"},  {"\\n", "\n"},
-    {"\\r", "\r"},          {"\\t", "\t"},  {"\\v", "\v"},  {"\\\\", "\\"},
-    {"\\\'", "\'"},         {"\\\"", "\""}, {"\\\?", "\?"},
-};
+constexpr std::array<std::pair<const char*, const char*>, 11> ESCAPE_CHARACTER_MAP{{
+    {"\\a", "\a"},
+    {"\\b", "\b"},
+    {"\\f", "\f"},
+    {"\\n", "\n"},
+    {"\\r", "\r"},
+    {"\\t", "\t"},
+    {"\\v", "\v"},
+    {"\\\\", "\\"},
+    {"\\\'", "\'"},
+    {"\\\"", "\""},
+    {"\\\?", "\?"},
+}};
 
 static IPSFileType IdentifyMagic(const std::vector<u8>& magic) {
-    if (magic.size() != 5)
+    if (magic.size() != 5) {
         return IPSFileType::Error;
-    if (magic == std::vector<u8>{'P', 'A', 'T', 'C', 'H'})
+    }
+
+    constexpr std::array<u8, 5> patch_magic{{'P', 'A', 'T', 'C', 'H'}};
+    if (std::equal(magic.begin(), magic.end(), patch_magic.begin())) {
         return IPSFileType::IPS;
-    if (magic == std::vector<u8>{'I', 'P', 'S', '3', '2'})
+    }
+
+    constexpr std::array<u8, 5> ips32_magic{{'I', 'P', 'S', '3', '2'}};
+    if (std::equal(magic.begin(), magic.end(), ips32_magic.begin())) {
         return IPSFileType::IPS32;
+    }
+
     return IPSFileType::Error;
 }
 
+static bool IsEOF(IPSFileType type, const std::vector<u8>& data) {
+    constexpr std::array<u8, 3> eof{{'E', 'O', 'F'}};
+    if (type == IPSFileType::IPS && std::equal(data.begin(), data.end(), eof.begin())) {
+        return true;
+    }
+
+    constexpr std::array<u8, 4> eeof{{'E', 'E', 'O', 'F'}};
+    return type == IPSFileType::IPS32 && std::equal(data.begin(), data.end(), eeof.begin());
+}
+
 VirtualFile PatchIPS(const VirtualFile& in, const VirtualFile& ips) {
     if (in == nullptr || ips == nullptr)
         return nullptr;
@@ -47,8 +79,7 @@ VirtualFile PatchIPS(const VirtualFile& in, const VirtualFile& ips) {
     u64 offset = 5; // After header
     while (ips->Read(temp.data(), temp.size(), offset) == temp.size()) {
         offset += temp.size();
-        if (type == IPSFileType::IPS32 && temp == std::vector<u8>{'E', 'E', 'O', 'F'} ||
-            type == IPSFileType::IPS && temp == std::vector<u8>{'E', 'O', 'F'}) {
+        if (IsEOF(type, temp)) {
             break;
         }
 
@@ -76,23 +107,32 @@ VirtualFile PatchIPS(const VirtualFile& in, const VirtualFile& ips) {
                 return nullptr;
 
             if (real_offset + rle_size > in_data.size())
-                rle_size = in_data.size() - real_offset;
+                rle_size = static_cast<u16>(in_data.size() - real_offset);
             std::memset(in_data.data() + real_offset, data.get(), rle_size);
         } else { // Standard Patch
             auto read = data_size;
             if (real_offset + read > in_data.size())
-                read = in_data.size() - real_offset;
+                read = static_cast<u16>(in_data.size() - real_offset);
             if (ips->Read(in_data.data() + real_offset, read, offset) != data_size)
                 return nullptr;
             offset += data_size;
         }
     }
 
-    if (temp != std::vector<u8>{'E', 'E', 'O', 'F'} && temp != std::vector<u8>{'E', 'O', 'F'})
+    if (!IsEOF(type, temp)) {
         return nullptr;
-    return std::make_shared<VectorVfsFile>(in_data, in->GetName(), in->GetContainingDirectory());
+    }
+
+    return std::make_shared<VectorVfsFile>(std::move(in_data), in->GetName(),
+                                           in->GetContainingDirectory());
 }
 
+struct IPSwitchCompiler::IPSwitchPatch {
+    std::string name;
+    bool enabled;
+    std::map<u32, std::vector<u8>> records;
+};
+
 IPSwitchCompiler::IPSwitchCompiler(VirtualFile patch_text_) : patch_text(std::move(patch_text_)) {
     Parse();
 }
@@ -225,7 +265,7 @@ void IPSwitchCompiler::Parse() {
                 if (patch_line.length() < 11)
                     break;
                 auto offset = std::stoul(patch_line.substr(0, 8), nullptr, 16);
-                offset += offset_shift;
+                offset += static_cast<unsigned long>(offset_shift);
 
                 std::vector<u8> replace;
                 // 9 - first char of replacement val
@@ -291,7 +331,8 @@ VirtualFile IPSwitchCompiler::Apply(const VirtualFile& in) const {
         }
     }
 
-    return std::make_shared<VectorVfsFile>(in_data, in->GetName(), in->GetContainingDirectory());
+    return std::make_shared<VectorVfsFile>(std::move(in_data), in->GetName(),
+                                           in->GetContainingDirectory());
 }
 
 } // namespace FileSys

src/core/file_sys/ips_layer.h

@@ -4,8 +4,11 @@
 
 #pragma once
 
+#include <array>
 #include <memory>
+#include <vector>
 
+#include "common/common_types.h"
 #include "core/file_sys/vfs.h"
 
 namespace FileSys {
@@ -22,17 +25,13 @@ public:
     VirtualFile Apply(const VirtualFile& in) const;
 
 private:
+    struct IPSwitchPatch;
+
     void ParseFlag(const std::string& flag);
     void Parse();
 
     bool valid = false;
 
-    struct IPSwitchPatch {
-        std::string name;
-        bool enabled;
-        std::map<u32, std::vector<u8>> records;
-    };
-
     VirtualFile patch_text;
     std::vector<IPSwitchPatch> patches;
     std::array<u8, 0x20> nso_build_id{};

src/core/file_sys/patch_manager.cpp

@@ -182,11 +182,17 @@ static void ApplyLayeredFS(VirtualFile& romfs, u64 title_id, ContentRecordType t
               [](const VirtualDir& l, const VirtualDir& r) { return l->GetName() < r->GetName(); });
 
     std::vector<VirtualDir> layers;
+    std::vector<VirtualDir> layers_ext;
     layers.reserve(patch_dirs.size() + 1);
+    layers_ext.reserve(patch_dirs.size() + 1);
     for (const auto& subdir : patch_dirs) {
         auto romfs_dir = subdir->GetSubdirectory("romfs");
         if (romfs_dir != nullptr)
             layers.push_back(std::move(romfs_dir));
+
+        auto ext_dir = subdir->GetSubdirectory("romfs_ext");
+        if (ext_dir != nullptr)
+            layers_ext.push_back(std::move(ext_dir));
     }
     layers.push_back(std::move(extracted));
 
@@ -195,7 +201,9 @@ static void ApplyLayeredFS(VirtualFile& romfs, u64 title_id, ContentRecordType t
         return;
     }
 
-    auto packed = CreateRomFS(std::move(layered));
+    auto layered_ext = LayeredVfsDirectory::MakeLayeredDirectory(std::move(layers_ext));
+
+    auto packed = CreateRomFS(std::move(layered), std::move(layered_ext));
     if (packed == nullptr) {
         return;
     }
@@ -204,8 +212,8 @@ static void ApplyLayeredFS(VirtualFile& romfs, u64 title_id, ContentRecordType t
     romfs = std::move(packed);
 }
 
-VirtualFile PatchManager::PatchRomFS(VirtualFile romfs, u64 ivfc_offset,
-                                     ContentRecordType type) const {
+VirtualFile PatchManager::PatchRomFS(VirtualFile romfs, u64 ivfc_offset, ContentRecordType type,
+                                     VirtualFile update_raw) const {
     LOG_INFO(Loader, "Patching RomFS for title_id={:016X}, type={:02X}", title_id,
              static_cast<u8>(type));
 
@@ -225,6 +233,13 @@ VirtualFile PatchManager::PatchRomFS(VirtualFile romfs, u64 ivfc_offset,
                      FormatTitleVersion(installed->GetEntryVersion(update_tid).get_value_or(0)));
             romfs = new_nca->GetRomFS();
         }
+    } else if (update_raw != nullptr) {
+        const auto new_nca = std::make_shared<NCA>(update_raw, romfs, ivfc_offset);
+        if (new_nca->GetStatus() == Loader::ResultStatus::Success &&
+            new_nca->GetRomFS() != nullptr) {
+            LOG_INFO(Loader, "    RomFS: Update (PACKED) applied successfully");
+            romfs = new_nca->GetRomFS();
+        }
     }
 
     // LayeredFS
@@ -244,7 +259,8 @@ static bool IsDirValidAndNonEmpty(const VirtualDir& dir) {
     return dir != nullptr && (!dir->GetFiles().empty() || !dir->GetSubdirectories().empty());
 }
 
-std::map<std::string, std::string, std::less<>> PatchManager::GetPatchVersionNames() const {
+std::map<std::string, std::string, std::less<>> PatchManager::GetPatchVersionNames(
+    VirtualFile update_raw) const {
     std::map<std::string, std::string, std::less<>> out;
     const auto installed = Service::FileSystem::GetUnionContents();
 
@@ -265,6 +281,8 @@ std::map<std::string, std::string, std::less<>> PatchManager::GetPatchVersionNam
                     "Update",
                     FormatTitleVersion(meta_ver.get(), TitleVersionFormat::ThreeElements));
             }
+        } else if (update_raw != nullptr) {
+            out.insert_or_assign("Update", "PACKED");
         }
     }
 
@@ -327,23 +345,22 @@ std::map<std::string, std::string, std::less<>> PatchManager::GetPatchVersionNam
     return out;
 }
 
-std::pair<std::shared_ptr<NACP>, VirtualFile> PatchManager::GetControlMetadata() const {
+std::pair<std::unique_ptr<NACP>, VirtualFile> PatchManager::GetControlMetadata() const {
     const auto& installed{Service::FileSystem::GetUnionContents()};
 
     const auto base_control_nca = installed->GetEntry(title_id, ContentRecordType::Control);
     if (base_control_nca == nullptr)
         return {};
 
-    return ParseControlNCA(base_control_nca);
+    return ParseControlNCA(*base_control_nca);
 }
 
-std::pair<std::shared_ptr<NACP>, VirtualFile> PatchManager::ParseControlNCA(
-    const std::shared_ptr<NCA>& nca) const {
-    const auto base_romfs = nca->GetRomFS();
+std::pair<std::unique_ptr<NACP>, VirtualFile> PatchManager::ParseControlNCA(const NCA& nca) const {
+    const auto base_romfs = nca.GetRomFS();
     if (base_romfs == nullptr)
         return {};
 
-    const auto romfs = PatchRomFS(base_romfs, nca->GetBaseIVFCOffset(), ContentRecordType::Control);
+    const auto romfs = PatchRomFS(base_romfs, nca.GetBaseIVFCOffset(), ContentRecordType::Control);
     if (romfs == nullptr)
         return {};
 
@@ -355,7 +372,7 @@ std::pair<std::shared_ptr<NACP>, VirtualFile> PatchManager::ParseControlNCA(
     if (nacp_file == nullptr)
         nacp_file = extracted->GetFile("Control.nacp");
 
-    const auto nacp = nacp_file == nullptr ? nullptr : std::make_shared<NACP>(nacp_file);
+    auto nacp = nacp_file == nullptr ? nullptr : std::make_unique<NACP>(nacp_file);
 
     VirtualFile icon_file;
     for (const auto& language : FileSys::LANGUAGE_NAMES) {
@@ -364,6 +381,6 @@ std::pair<std::shared_ptr<NACP>, VirtualFile> PatchManager::ParseControlNCA(
             break;
     }
 
-    return {nacp, icon_file};
+    return {std::move(nacp), icon_file};
 }
 } // namespace FileSys

src/core/file_sys/patch_manager.h

@@ -47,19 +47,20 @@ public:
     // - Game Updates
     // - LayeredFS
     VirtualFile PatchRomFS(VirtualFile base, u64 ivfc_offset,
-                           ContentRecordType type = ContentRecordType::Program) const;
+                           ContentRecordType type = ContentRecordType::Program,
+                           VirtualFile update_raw = nullptr) const;
 
     // Returns a vector of pairs between patch names and patch versions.
     // i.e. Update 3.2.2 will return {"Update", "3.2.2"}
-    std::map<std::string, std::string, std::less<>> GetPatchVersionNames() const;
+    std::map<std::string, std::string, std::less<>> GetPatchVersionNames(
+        VirtualFile update_raw = nullptr) const;
 
     // Given title_id of the program, attempts to get the control data of the update and parse it,
     // falling back to the base control data.
-    std::pair<std::shared_ptr<NACP>, VirtualFile> GetControlMetadata() const;
+    std::pair<std::unique_ptr<NACP>, VirtualFile> GetControlMetadata() const;
 
     // Version of GetControlMetadata that takes an arbitrary NCA
-    std::pair<std::shared_ptr<NACP>, VirtualFile> ParseControlNCA(
-        const std::shared_ptr<NCA>& nca) const;
+    std::pair<std::unique_ptr<NACP>, VirtualFile> ParseControlNCA(const NCA& nca) const;
 
 private:
     u64 title_id;

src/core/file_sys/romfs.cpp

@@ -129,11 +129,11 @@ VirtualDir ExtractRomFS(VirtualFile file, RomFSExtractionType type) {
     return out;
 }
 
-VirtualFile CreateRomFS(VirtualDir dir) {
+VirtualFile CreateRomFS(VirtualDir dir, VirtualDir ext) {
     if (dir == nullptr)
         return nullptr;
 
-    RomFSBuildContext ctx{dir};
+    RomFSBuildContext ctx{dir, ext};
     return ConcatenatedVfsFile::MakeConcatenatedFile(0, ctx.Build(), dir->GetName());
 }
 

src/core/file_sys/romfs.h

@@ -44,6 +44,6 @@ VirtualDir ExtractRomFS(VirtualFile file,
 
 // Converts a VFS filesystem into a RomFS binary
 // Returns nullptr on failure
-VirtualFile CreateRomFS(VirtualDir dir);
+VirtualFile CreateRomFS(VirtualDir dir, VirtualDir ext = nullptr);
 
 } // namespace FileSys

src/core/file_sys/romfs_factory.cpp

@@ -30,12 +30,17 @@ RomFSFactory::RomFSFactory(Loader::AppLoader& app_loader) {
 
 RomFSFactory::~RomFSFactory() = default;
 
+void RomFSFactory::SetPackedUpdate(VirtualFile update_raw) {
+    this->update_raw = std::move(update_raw);
+}
+
 ResultVal<VirtualFile> RomFSFactory::OpenCurrentProcess() {
     if (!updatable)
         return MakeResult<VirtualFile>(file);
 
     const PatchManager patch_manager(Core::CurrentProcess()->GetTitleID());
-    return MakeResult<VirtualFile>(patch_manager.PatchRomFS(file, ivfc_offset));
+    return MakeResult<VirtualFile>(
+        patch_manager.PatchRomFS(file, ivfc_offset, ContentRecordType::Program, update_raw));
 }
 
 ResultVal<VirtualFile> RomFSFactory::Open(u64 title_id, StorageId storage, ContentRecordType type) {

src/core/file_sys/romfs_factory.h

@@ -32,11 +32,13 @@ public:
     explicit RomFSFactory(Loader::AppLoader& app_loader);
     ~RomFSFactory();
 
+    void SetPackedUpdate(VirtualFile update_raw);
     ResultVal<VirtualFile> OpenCurrentProcess();
     ResultVal<VirtualFile> Open(u64 title_id, StorageId storage, ContentRecordType type);
 
 private:
     VirtualFile file;
+    VirtualFile update_raw;
     bool updatable;
     u64 ivfc_offset;
 };

src/core/file_sys/submission_package.cpp

@@ -259,8 +259,11 @@ void NSP::ReadNCAs(const std::vector<VirtualFile>& files) {
                 auto next_nca = std::make_shared<NCA>(next_file);
                 if (next_nca->GetType() == NCAContentType::Program)
                     program_status[cnmt.GetTitleID()] = next_nca->GetStatus();
-                if (next_nca->GetStatus() == Loader::ResultStatus::Success)
+                if (next_nca->GetStatus() == Loader::ResultStatus::Success ||
+                    (next_nca->GetStatus() == Loader::ResultStatus::ErrorMissingBKTRBaseRomFS &&
+                     (cnmt.GetTitleID() & 0x800) != 0)) {
                     ncas_title[rec.type] = std::move(next_nca);
+                }
             }
 
             break;

src/core/gdbstub/gdbstub.cpp

@@ -209,7 +209,7 @@ static Kernel::Thread* FindThreadById(int id) {
     for (u32 core = 0; core < Core::NUM_CPU_CORES; core++) {
         const auto& threads = Core::System::GetInstance().Scheduler(core)->GetThreadList();
         for (auto& thread : threads) {
-            if (thread->GetThreadId() == static_cast<u32>(id)) {
+            if (thread->GetThreadID() == static_cast<u32>(id)) {
                 current_core = core;
                 return thread.get();
             }
@@ -223,16 +223,18 @@ static u64 RegRead(std::size_t id, Kernel::Thread* thread = nullptr) {
         return 0;
     }
 
+    const auto& thread_context = thread->GetContext();
+
     if (id < SP_REGISTER) {
-        return thread->context.cpu_registers[id];
+        return thread_context.cpu_registers[id];
     } else if (id == SP_REGISTER) {
-        return thread->context.sp;
+        return thread_context.sp;
     } else if (id == PC_REGISTER) {
-        return thread->context.pc;
+        return thread_context.pc;
     } else if (id == PSTATE_REGISTER) {
-        return thread->context.pstate;
+        return thread_context.pstate;
     } else if (id > PSTATE_REGISTER && id < FPCR_REGISTER) {
-        return thread->context.vector_registers[id - UC_ARM64_REG_Q0][0];
+        return thread_context.vector_registers[id - UC_ARM64_REG_Q0][0];
     } else {
         return 0;
     }
@@ -243,16 +245,18 @@ static void RegWrite(std::size_t id, u64 val, Kernel::Thread* thread = nullptr)
         return;
     }
 
+    auto& thread_context = thread->GetContext();
+
     if (id < SP_REGISTER) {
-        thread->context.cpu_registers[id] = val;
+        thread_context.cpu_registers[id] = val;
     } else if (id == SP_REGISTER) {
-        thread->context.sp = val;
+        thread_context.sp = val;
     } else if (id == PC_REGISTER) {
-        thread->context.pc = val;
+        thread_context.pc = val;
     } else if (id == PSTATE_REGISTER) {
-        thread->context.pstate = static_cast<u32>(val);
+        thread_context.pstate = static_cast<u32>(val);
     } else if (id > PSTATE_REGISTER && id < FPCR_REGISTER) {
-        thread->context.vector_registers[id - (PSTATE_REGISTER + 1)][0] = val;
+        thread_context.vector_registers[id - (PSTATE_REGISTER + 1)][0] = val;
     }
 }
 
@@ -595,7 +599,7 @@ static void HandleQuery() {
         for (u32 core = 0; core < Core::NUM_CPU_CORES; core++) {
             const auto& threads = Core::System::GetInstance().Scheduler(core)->GetThreadList();
             for (const auto& thread : threads) {
-                val += fmt::format("{:x}", thread->GetThreadId());
+                val += fmt::format("{:x}", thread->GetThreadID());
                 val += ",";
             }
         }
@@ -612,7 +616,7 @@ static void HandleQuery() {
             for (const auto& thread : threads) {
                 buffer +=
                     fmt::format(R"*(<thread id="{:x}" core="{:d}" name="Thread {:x}"></thread>)*",
-                                thread->GetThreadId(), core, thread->GetThreadId());
+                                thread->GetThreadID(), core, thread->GetThreadID());
             }
         }
         buffer += "</threads>";
@@ -693,7 +697,7 @@ static void SendSignal(Kernel::Thread* thread, u32 signal, bool full = true) {
     }
 
     if (thread) {
-        buffer += fmt::format(";thread:{:x};", thread->GetThreadId());
+        buffer += fmt::format(";thread:{:x};", thread->GetThreadID());
     }
 
     SendReply(buffer.c_str());
@@ -857,7 +861,9 @@ static void WriteRegister() {
     }
 
     // Update Unicorn context skipping scheduler, no running threads at this point
-    Core::System::GetInstance().ArmInterface(current_core).LoadContext(current_thread->context);
+    Core::System::GetInstance()
+        .ArmInterface(current_core)
+        .LoadContext(current_thread->GetContext());
 
     SendReply("OK");
 }
@@ -886,7 +892,9 @@ static void WriteRegisters() {
     }
 
     // Update Unicorn context skipping scheduler, no running threads at this point
-    Core::System::GetInstance().ArmInterface(current_core).LoadContext(current_thread->context);
+    Core::System::GetInstance()
+        .ArmInterface(current_core)
+        .LoadContext(current_thread->GetContext());
 
     SendReply("OK");
 }
@@ -960,7 +968,9 @@ static void Step() {
     if (command_length > 1) {
         RegWrite(PC_REGISTER, GdbHexToLong(command_buffer + 1), current_thread);
         // Update Unicorn context skipping scheduler, no running threads at this point
-        Core::System::GetInstance().ArmInterface(current_core).LoadContext(current_thread->context);
+        Core::System::GetInstance()
+            .ArmInterface(current_core)
+            .LoadContext(current_thread->GetContext());
     }
     step_loop = true;
     halt_loop = true;

src/core/hle/kernel/address_arbiter.cpp

@@ -23,13 +23,13 @@ namespace AddressArbiter {
 // Performs actual address waiting logic.
 static ResultCode WaitForAddress(VAddr address, s64 timeout) {
     SharedPtr<Thread> current_thread = GetCurrentThread();
-    current_thread->arb_wait_address = address;
-    current_thread->status = ThreadStatus::WaitArb;
-    current_thread->wakeup_callback = nullptr;
+    current_thread->SetArbiterWaitAddress(address);
+    current_thread->SetStatus(ThreadStatus::WaitArb);
+    current_thread->InvalidateWakeupCallback();
 
     current_thread->WakeAfterDelay(timeout);
 
-    Core::System::GetInstance().CpuCore(current_thread->processor_id).PrepareReschedule();
+    Core::System::GetInstance().CpuCore(current_thread->GetProcessorID()).PrepareReschedule();
     return RESULT_TIMEOUT;
 }
 
@@ -39,10 +39,10 @@ static std::vector<SharedPtr<Thread>> GetThreadsWaitingOnAddress(VAddr address)
                                            std::vector<SharedPtr<Thread>>& waiting_threads,
                                            VAddr arb_addr) {
         const auto& scheduler = Core::System::GetInstance().Scheduler(core_index);
-        auto& thread_list = scheduler->GetThreadList();
+        const auto& thread_list = scheduler->GetThreadList();
 
-        for (auto& thread : thread_list) {
-            if (thread->arb_wait_address == arb_addr)
+        for (const auto& thread : thread_list) {
+            if (thread->GetArbiterWaitAddress() == arb_addr)
                 waiting_threads.push_back(thread);
         }
     };
@@ -57,7 +57,7 @@ static std::vector<SharedPtr<Thread>> GetThreadsWaitingOnAddress(VAddr address)
     // Sort them by priority, such that the highest priority ones come first.
     std::sort(threads.begin(), threads.end(),
               [](const SharedPtr<Thread>& lhs, const SharedPtr<Thread>& rhs) {
-                  return lhs->current_priority < rhs->current_priority;
+                  return lhs->GetPriority() < rhs->GetPriority();
               });
 
     return threads;
@@ -73,9 +73,9 @@ static void WakeThreads(std::vector<SharedPtr<Thread>>& waiting_threads, s32 num
 
     // Signal the waiting threads.
     for (std::size_t i = 0; i < last; i++) {
-        ASSERT(waiting_threads[i]->status == ThreadStatus::WaitArb);
+        ASSERT(waiting_threads[i]->GetStatus() == ThreadStatus::WaitArb);
         waiting_threads[i]->SetWaitSynchronizationResult(RESULT_SUCCESS);
-        waiting_threads[i]->arb_wait_address = 0;
+        waiting_threads[i]->SetArbiterWaitAddress(0);
         waiting_threads[i]->ResumeFromWait();
     }
 }

src/core/hle/kernel/client_port.cpp

@@ -17,6 +17,10 @@ namespace Kernel {
 ClientPort::ClientPort(KernelCore& kernel) : Object{kernel} {}
 ClientPort::~ClientPort() = default;
 
+SharedPtr<ServerPort> ClientPort::GetServerPort() const {
+    return server_port;
+}
+
 ResultVal<SharedPtr<ClientSession>> ClientPort::Connect() {
     // Note: Threads do not wait for the server endpoint to call
     // AcceptSession before returning from this call.

src/core/hle/kernel/client_port.h

@@ -30,6 +30,8 @@ public:
         return HANDLE_TYPE;
     }
 
+    SharedPtr<ServerPort> GetServerPort() const;
+
     /**
      * Creates a new Session pair, adds the created ServerSession to the associated ServerPort's
      * list of pending sessions, and signals the ServerPort, causing any threads

src/core/hle/kernel/hle_ipc.cpp

@@ -42,14 +42,14 @@ SharedPtr<Event> HLERequestContext::SleepClientThread(SharedPtr<Thread> thread,
                                                       Kernel::SharedPtr<Kernel::Event> event) {
 
     // Put the client thread to sleep until the wait event is signaled or the timeout expires.
-    thread->wakeup_callback = [context = *this, callback](
+    thread->SetWakeupCallback([context = *this, callback](
                                   ThreadWakeupReason reason, SharedPtr<Thread> thread,
                                   SharedPtr<WaitObject> object, std::size_t index) mutable -> bool {
-        ASSERT(thread->status == ThreadStatus::WaitHLEEvent);
+        ASSERT(thread->GetStatus() == ThreadStatus::WaitHLEEvent);
         callback(thread, context, reason);
         context.WriteToOutgoingCommandBuffer(*thread);
         return true;
-    };
+    });
 
     if (!event) {
         // Create event if not provided
@@ -59,8 +59,8 @@ SharedPtr<Event> HLERequestContext::SleepClientThread(SharedPtr<Thread> thread,
     }
 
     event->Clear();
-    thread->status = ThreadStatus::WaitHLEEvent;
-    thread->wait_objects = {event};
+    thread->SetStatus(ThreadStatus::WaitHLEEvent);
+    thread->SetWaitObjects({event});
     event->AddWaitingThread(thread);
 
     if (timeout > 0) {
@@ -209,7 +209,7 @@ ResultCode HLERequestContext::PopulateFromIncomingCommandBuffer(u32_le* src_cmdb
 
 ResultCode HLERequestContext::WriteToOutgoingCommandBuffer(const Thread& thread) {
     std::array<u32, IPC::COMMAND_BUFFER_LENGTH> dst_cmdbuf;
-    Memory::ReadBlock(*thread.owner_process, thread.GetTLSAddress(), dst_cmdbuf.data(),
+    Memory::ReadBlock(*thread.GetOwnerProcess(), thread.GetTLSAddress(), dst_cmdbuf.data(),
                       dst_cmdbuf.size() * sizeof(u32));
 
     // The header was already built in the internal command buffer. Attempt to parse it to verify
@@ -268,7 +268,7 @@ ResultCode HLERequestContext::WriteToOutgoingCommandBuffer(const Thread& thread)
     }
 
     // Copy the translated command buffer back into the thread's command buffer area.
-    Memory::WriteBlock(*thread.owner_process, thread.GetTLSAddress(), dst_cmdbuf.data(),
+    Memory::WriteBlock(*thread.GetOwnerProcess(), thread.GetTLSAddress(), dst_cmdbuf.data(),
                        dst_cmdbuf.size() * sizeof(u32));
 
     return RESULT_SUCCESS;

src/core/hle/kernel/kernel.cpp

@@ -46,40 +46,40 @@ static void ThreadWakeupCallback(u64 thread_handle, [[maybe_unused]] int cycles_
 
     bool resume = true;
 
-    if (thread->status == ThreadStatus::WaitSynchAny ||
-        thread->status == ThreadStatus::WaitSynchAll ||
-        thread->status == ThreadStatus::WaitHLEEvent) {
+    if (thread->GetStatus() == ThreadStatus::WaitSynchAny ||
+        thread->GetStatus() == ThreadStatus::WaitSynchAll ||
+        thread->GetStatus() == ThreadStatus::WaitHLEEvent) {
         // Remove the thread from each of its waiting objects' waitlists
-        for (auto& object : thread->wait_objects) {
+        for (const auto& object : thread->GetWaitObjects()) {
             object->RemoveWaitingThread(thread.get());
         }
-        thread->wait_objects.clear();
+        thread->ClearWaitObjects();
 
         // Invoke the wakeup callback before clearing the wait objects
-        if (thread->wakeup_callback) {
-            resume = thread->wakeup_callback(ThreadWakeupReason::Timeout, thread, nullptr, 0);
+        if (thread->HasWakeupCallback()) {
+            resume = thread->InvokeWakeupCallback(ThreadWakeupReason::Timeout, thread, nullptr, 0);
         }
     }
 
-    if (thread->mutex_wait_address != 0 || thread->condvar_wait_address != 0 ||
-        thread->wait_handle) {
-        ASSERT(thread->status == ThreadStatus::WaitMutex);
-        thread->mutex_wait_address = 0;
-        thread->condvar_wait_address = 0;
-        thread->wait_handle = 0;
+    if (thread->GetMutexWaitAddress() != 0 || thread->GetCondVarWaitAddress() != 0 ||
+        thread->GetWaitHandle() != 0) {
+        ASSERT(thread->GetStatus() == ThreadStatus::WaitMutex);
+        thread->SetMutexWaitAddress(0);
+        thread->SetCondVarWaitAddress(0);
+        thread->SetWaitHandle(0);
 
-        auto lock_owner = thread->lock_owner;
+        auto* const lock_owner = thread->GetLockOwner();
         // Threads waking up by timeout from WaitProcessWideKey do not perform priority inheritance
         // and don't have a lock owner unless SignalProcessWideKey was called first and the thread
         // wasn't awakened due to the mutex already being acquired.
-        if (lock_owner) {
+        if (lock_owner != nullptr) {
             lock_owner->RemoveMutexWaiter(thread);
         }
     }
 
-    if (thread->arb_wait_address != 0) {
-        ASSERT(thread->status == ThreadStatus::WaitArb);
-        thread->arb_wait_address = 0;
+    if (thread->GetArbiterWaitAddress() != 0) {
+        ASSERT(thread->GetStatus() == ThreadStatus::WaitArb);
+        thread->SetArbiterWaitAddress(0);
     }
 
     if (resume) {
@@ -116,7 +116,7 @@ struct KernelCore::Impl {
         next_thread_id = 1;
 
         process_list.clear();
-        current_process.reset();
+        current_process = nullptr;
 
         handle_table.Clear();
         resource_limits.fill(nullptr);
@@ -207,7 +207,7 @@ struct KernelCore::Impl {
 
     // Lists all processes that exist in the current session.
     std::vector<SharedPtr<Process>> process_list;
-    SharedPtr<Process> current_process;
+    Process* current_process = nullptr;
 
     Kernel::HandleTable handle_table;
     std::array<SharedPtr<ResourceLimit>, 4> resource_limits;
@@ -266,15 +266,15 @@ void KernelCore::AppendNewProcess(SharedPtr<Process> process) {
     impl->process_list.push_back(std::move(process));
 }
 
-void KernelCore::MakeCurrentProcess(SharedPtr<Process> process) {
-    impl->current_process = std::move(process);
+void KernelCore::MakeCurrentProcess(Process* process) {
+    impl->current_process = process;
 }
 
-SharedPtr<Process>& KernelCore::CurrentProcess() {
+Process* KernelCore::CurrentProcess() {
     return impl->current_process;
 }
 
-const SharedPtr<Process>& KernelCore::CurrentProcess() const {
+const Process* KernelCore::CurrentProcess() const {
     return impl->current_process;
 }
 

src/core/hle/kernel/kernel.h

@@ -66,13 +66,13 @@ public:
     void AppendNewProcess(SharedPtr<Process> process);
 
     /// Makes the given process the new current process.
-    void MakeCurrentProcess(SharedPtr<Process> process);
+    void MakeCurrentProcess(Process* process);
 
-    /// Retrieves a reference to the current process.
-    SharedPtr<Process>& CurrentProcess();
+    /// Retrieves a pointer to the current process.
+    Process* CurrentProcess();
 
-    /// Retrieves a const reference to the current process.
-    const SharedPtr<Process>& CurrentProcess() const;
+    /// Retrieves a const pointer to the current process.
+    const Process* CurrentProcess() const;
 
     /// Adds a port to the named port table
     void AddNamedPort(std::string name, SharedPtr<ClientPort> port);

src/core/hle/kernel/mutex.cpp

@@ -28,11 +28,11 @@ static std::pair<SharedPtr<Thread>, u32> GetHighestPriorityMutexWaitingThread(
     SharedPtr<Thread> highest_priority_thread;
     u32 num_waiters = 0;
 
-    for (auto& thread : current_thread->wait_mutex_threads) {
-        if (thread->mutex_wait_address != mutex_addr)
+    for (const auto& thread : current_thread->GetMutexWaitingThreads()) {
+        if (thread->GetMutexWaitAddress() != mutex_addr)
             continue;
 
-        ASSERT(thread->status == ThreadStatus::WaitMutex);
+        ASSERT(thread->GetStatus() == ThreadStatus::WaitMutex);
 
         ++num_waiters;
         if (highest_priority_thread == nullptr ||
@@ -47,12 +47,12 @@ static std::pair<SharedPtr<Thread>, u32> GetHighestPriorityMutexWaitingThread(
 /// Update the mutex owner field of all threads waiting on the mutex to point to the new owner.
 static void TransferMutexOwnership(VAddr mutex_addr, SharedPtr<Thread> current_thread,
                                    SharedPtr<Thread> new_owner) {
-    auto threads = current_thread->wait_mutex_threads;
-    for (auto& thread : threads) {
-        if (thread->mutex_wait_address != mutex_addr)
+    const auto threads = current_thread->GetMutexWaitingThreads();
+    for (const auto& thread : threads) {
+        if (thread->GetMutexWaitAddress() != mutex_addr)
             continue;
 
-        ASSERT(thread->lock_owner == current_thread);
+        ASSERT(thread->GetLockOwner() == current_thread);
         current_thread->RemoveMutexWaiter(thread);
         if (new_owner != thread)
             new_owner->AddMutexWaiter(thread);
@@ -84,11 +84,11 @@ ResultCode Mutex::TryAcquire(HandleTable& handle_table, VAddr address, Handle ho
         return ERR_INVALID_HANDLE;
 
     // Wait until the mutex is released
-    GetCurrentThread()->mutex_wait_address = address;
-    GetCurrentThread()->wait_handle = requesting_thread_handle;
+    GetCurrentThread()->SetMutexWaitAddress(address);
+    GetCurrentThread()->SetWaitHandle(requesting_thread_handle);
 
-    GetCurrentThread()->status = ThreadStatus::WaitMutex;
-    GetCurrentThread()->wakeup_callback = nullptr;
+    GetCurrentThread()->SetStatus(ThreadStatus::WaitMutex);
+    GetCurrentThread()->InvalidateWakeupCallback();
 
     // Update the lock holder thread's priority to prevent priority inversion.
     holding_thread->AddMutexWaiter(GetCurrentThread());
@@ -115,7 +115,7 @@ ResultCode Mutex::Release(VAddr address) {
     // Transfer the ownership of the mutex from the previous owner to the new one.
     TransferMutexOwnership(address, GetCurrentThread(), thread);
 
-    u32 mutex_value = thread->wait_handle;
+    u32 mutex_value = thread->GetWaitHandle();
 
     if (num_waiters >= 2) {
         // Notify the guest that there are still some threads waiting for the mutex
@@ -125,13 +125,13 @@ ResultCode Mutex::Release(VAddr address) {
     // Grant the mutex to the next waiting thread and resume it.
     Memory::Write32(address, mutex_value);
 
-    ASSERT(thread->status == ThreadStatus::WaitMutex);
+    ASSERT(thread->GetStatus() == ThreadStatus::WaitMutex);
     thread->ResumeFromWait();
 
-    thread->lock_owner = nullptr;
-    thread->condvar_wait_address = 0;
-    thread->mutex_wait_address = 0;
-    thread->wait_handle = 0;
+    thread->SetLockOwner(nullptr);
+    thread->SetCondVarWaitAddress(0);
+    thread->SetMutexWaitAddress(0);
+    thread->SetWaitHandle(0);
 
     return RESULT_SUCCESS;
 }

src/core/hle/kernel/process.cpp

@@ -144,15 +144,15 @@ void Process::PrepareForTermination() {
 
     const auto stop_threads = [this](const std::vector<SharedPtr<Thread>>& thread_list) {
         for (auto& thread : thread_list) {
-            if (thread->owner_process != this)
+            if (thread->GetOwnerProcess() != this)
                 continue;
 
             if (thread == GetCurrentThread())
                 continue;
 
             // TODO(Subv): When are the other running/ready threads terminated?
-            ASSERT_MSG(thread->status == ThreadStatus::WaitSynchAny ||
-                           thread->status == ThreadStatus::WaitSynchAll,
+            ASSERT_MSG(thread->GetStatus() == ThreadStatus::WaitSynchAny ||
+                           thread->GetStatus() == ThreadStatus::WaitSynchAll,
                        "Exiting processes with non-waiting threads is currently unimplemented");
 
             thread->Stop();

src/core/hle/kernel/scheduler.cpp

@@ -9,7 +9,7 @@
 #include "common/logging/log.h"
 #include "core/arm/arm_interface.h"
 #include "core/core.h"
-#include "core/core_timing.h"
+#include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/scheduler.h"
 
@@ -38,10 +38,10 @@ Thread* Scheduler::PopNextReadyThread() {
     Thread* next = nullptr;
     Thread* thread = GetCurrentThread();
 
-    if (thread && thread->status == ThreadStatus::Running) {
+    if (thread && thread->GetStatus() == ThreadStatus::Running) {
         // We have to do better than the current thread.
         // This call returns null when that's not possible.
-        next = ready_queue.pop_first_better(thread->current_priority);
+        next = ready_queue.pop_first_better(thread->GetPriority());
         if (!next) {
             // Otherwise just keep going with the current thread
             next = thread;
@@ -58,40 +58,40 @@ void Scheduler::SwitchContext(Thread* new_thread) {
 
     // Save context for previous thread
     if (previous_thread) {
-        previous_thread->last_running_ticks = CoreTiming::GetTicks();
-        cpu_core.SaveContext(previous_thread->context);
+        cpu_core.SaveContext(previous_thread->GetContext());
         // Save the TPIDR_EL0 system register in case it was modified.
-        previous_thread->tpidr_el0 = cpu_core.GetTPIDR_EL0();
+        previous_thread->SetTPIDR_EL0(cpu_core.GetTPIDR_EL0());
 
-        if (previous_thread->status == ThreadStatus::Running) {
+        if (previous_thread->GetStatus() == ThreadStatus::Running) {
             // This is only the case when a reschedule is triggered without the current thread
             // yielding execution (i.e. an event triggered, system core time-sliced, etc)
-            ready_queue.push_front(previous_thread->current_priority, previous_thread);
-            previous_thread->status = ThreadStatus::Ready;
+            ready_queue.push_front(previous_thread->GetPriority(), previous_thread);
+            previous_thread->SetStatus(ThreadStatus::Ready);
         }
     }
 
     // Load context of new thread
     if (new_thread) {
-        ASSERT_MSG(new_thread->status == ThreadStatus::Ready,
+        ASSERT_MSG(new_thread->GetStatus() == ThreadStatus::Ready,
                    "Thread must be ready to become running.");
 
         // Cancel any outstanding wakeup events for this thread
         new_thread->CancelWakeupTimer();
 
-        auto previous_process = Core::CurrentProcess();
+        auto* const previous_process = Core::CurrentProcess();
 
         current_thread = new_thread;
 
-        ready_queue.remove(new_thread->current_priority, new_thread);
-        new_thread->status = ThreadStatus::Running;
+        ready_queue.remove(new_thread->GetPriority(), new_thread);
+        new_thread->SetStatus(ThreadStatus::Running);
 
-        if (previous_process != current_thread->owner_process) {
-            Core::CurrentProcess() = current_thread->owner_process;
+        auto* const thread_owner_process = current_thread->GetOwnerProcess();
+        if (previous_process != thread_owner_process) {
+            Core::System::GetInstance().Kernel().MakeCurrentProcess(thread_owner_process);
             SetCurrentPageTable(&Core::CurrentProcess()->VMManager().page_table);
         }
 
-        cpu_core.LoadContext(new_thread->context);
+        cpu_core.LoadContext(new_thread->GetContext());
         cpu_core.SetTlsAddress(new_thread->GetTLSAddress());
         cpu_core.SetTPIDR_EL0(new_thread->GetTPIDR_EL0());
         cpu_core.ClearExclusiveState();
@@ -136,14 +136,14 @@ void Scheduler::RemoveThread(Thread* thread) {
 void Scheduler::ScheduleThread(Thread* thread, u32 priority) {
     std::lock_guard<std::mutex> lock(scheduler_mutex);
 
-    ASSERT(thread->status == ThreadStatus::Ready);
+    ASSERT(thread->GetStatus() == ThreadStatus::Ready);
     ready_queue.push_back(priority, thread);
 }
 
 void Scheduler::UnscheduleThread(Thread* thread, u32 priority) {
     std::lock_guard<std::mutex> lock(scheduler_mutex);
 
-    ASSERT(thread->status == ThreadStatus::Ready);
+    ASSERT(thread->GetStatus() == ThreadStatus::Ready);
     ready_queue.remove(priority, thread);
 }
 
@@ -151,8 +151,8 @@ void Scheduler::SetThreadPriority(Thread* thread, u32 priority) {
     std::lock_guard<std::mutex> lock(scheduler_mutex);
 
     // If thread was ready, adjust queues
-    if (thread->status == ThreadStatus::Ready)
-        ready_queue.move(thread, thread->current_priority, priority);
+    if (thread->GetStatus() == ThreadStatus::Ready)
+        ready_queue.move(thread, thread->GetPriority(), priority);
     else
         ready_queue.prepare(priority);
 }

src/core/hle/kernel/server_port.h

@@ -11,6 +11,7 @@
 #include "common/common_types.h"
 #include "core/hle/kernel/object.h"
 #include "core/hle/kernel/wait_object.h"
+#include "core/hle/result.h"
 
 namespace Kernel {
 

src/core/hle/kernel/server_session.cpp

@@ -120,10 +120,10 @@ ResultCode ServerSession::HandleSyncRequest(SharedPtr<Thread> thread) {
         result = hle_handler->HandleSyncRequest(context);
     }
 
-    if (thread->status == ThreadStatus::Running) {
+    if (thread->GetStatus() == ThreadStatus::Running) {
         // Put the thread to sleep until the server replies, it will be awoken in
         // svcReplyAndReceive for LLE servers.
-        thread->status = ThreadStatus::WaitIPC;
+        thread->SetStatus(ThreadStatus::WaitIPC);
 
         if (hle_handler != nullptr) {
             // For HLE services, we put the request threads to sleep for a short duration to

src/core/hle/kernel/svc.cpp

@@ -156,7 +156,7 @@ static ResultCode GetThreadId(u32* thread_id, Handle thread_handle) {
         return ERR_INVALID_HANDLE;
     }
 
-    *thread_id = thread->GetThreadId();
+    *thread_id = thread->GetThreadID();
     return RESULT_SUCCESS;
 }
 
@@ -177,7 +177,7 @@ static ResultCode GetProcessId(u32* process_id, Handle process_handle) {
 /// Default thread wakeup callback for WaitSynchronization
 static bool DefaultThreadWakeupCallback(ThreadWakeupReason reason, SharedPtr<Thread> thread,
                                         SharedPtr<WaitObject> object, std::size_t index) {
-    ASSERT(thread->status == ThreadStatus::WaitSynchAny);
+    ASSERT(thread->GetStatus() == ThreadStatus::WaitSynchAny);
 
     if (reason == ThreadWakeupReason::Timeout) {
         thread->SetWaitSynchronizationResult(RESULT_TIMEOUT);
@@ -204,10 +204,10 @@ static ResultCode WaitSynchronization(Handle* index, VAddr handles_address, u64
     if (handle_count > MaxHandles)
         return ResultCode(ErrorModule::Kernel, ErrCodes::TooLarge);
 
-    auto thread = GetCurrentThread();
+    auto* const thread = GetCurrentThread();
 
-    using ObjectPtr = SharedPtr<WaitObject>;
-    std::vector<ObjectPtr> objects(handle_count);
+    using ObjectPtr = Thread::ThreadWaitObjects::value_type;
+    Thread::ThreadWaitObjects objects(handle_count);
     auto& kernel = Core::System::GetInstance().Kernel();
 
     for (u64 i = 0; i < handle_count; ++i) {
@@ -244,14 +244,14 @@ static ResultCode WaitSynchronization(Handle* index, VAddr handles_address, u64
     for (auto& object : objects)
         object->AddWaitingThread(thread);
 
-    thread->wait_objects = std::move(objects);
-    thread->status = ThreadStatus::WaitSynchAny;
+    thread->SetWaitObjects(std::move(objects));
+    thread->SetStatus(ThreadStatus::WaitSynchAny);
 
     // Create an event to wake the thread up after the specified nanosecond delay has passed
     thread->WakeAfterDelay(nano_seconds);
-    thread->wakeup_callback = DefaultThreadWakeupCallback;
+    thread->SetWakeupCallback(DefaultThreadWakeupCallback);
 
-    Core::System::GetInstance().CpuCore(thread->processor_id).PrepareReschedule();
+    Core::System::GetInstance().CpuCore(thread->GetProcessorID()).PrepareReschedule();
 
     return RESULT_TIMEOUT;
 }
@@ -266,7 +266,7 @@ static ResultCode CancelSynchronization(Handle thread_handle) {
         return ERR_INVALID_HANDLE;
     }
 
-    ASSERT(thread->status == ThreadStatus::WaitSynchAny);
+    ASSERT(thread->GetStatus() == ThreadStatus::WaitSynchAny);
     thread->SetWaitSynchronizationResult(
         ResultCode(ErrorModule::Kernel, ErrCodes::SynchronizationCanceled));
     thread->ResumeFromWait();
@@ -301,13 +301,28 @@ static ResultCode ArbitrateUnlock(VAddr mutex_addr) {
     return Mutex::Release(mutex_addr);
 }
 
+struct BreakReason {
+    union {
+        u64 raw;
+        BitField<31, 1, u64> dont_kill_application;
+    };
+};
+
 /// Break program execution
 static void Break(u64 reason, u64 info1, u64 info2) {
-    LOG_CRITICAL(
-        Debug_Emulated,
-        "Emulated program broke execution! reason=0x{:016X}, info1=0x{:016X}, info2=0x{:016X}",
-        reason, info1, info2);
-    ASSERT(false);
+    BreakReason break_reason{reason};
+    if (break_reason.dont_kill_application) {
+        LOG_ERROR(
+            Debug_Emulated,
+            "Emulated program broke execution! reason=0x{:016X}, info1=0x{:016X}, info2=0x{:016X}",
+            reason, info1, info2);
+    } else {
+        LOG_CRITICAL(
+            Debug_Emulated,
+            "Emulated program broke execution! reason=0x{:016X}, info1=0x{:016X}, info2=0x{:016X}",
+            reason, info1, info2);
+        ASSERT(false);
+    }
 }
 
 /// Used to output a message on a debug hardware unit - does nothing on a retail unit
@@ -326,7 +341,7 @@ static ResultCode GetInfo(u64* result, u64 info_id, u64 handle, u64 info_sub_id)
     LOG_TRACE(Kernel_SVC, "called info_id=0x{:X}, info_sub_id=0x{:X}, handle=0x{:08X}", info_id,
               info_sub_id, handle);
 
-    const auto& current_process = Core::CurrentProcess();
+    const auto* current_process = Core::CurrentProcess();
     const auto& vm_manager = current_process->VMManager();
 
     switch (static_cast<GetInfoType>(info_id)) {
@@ -424,8 +439,8 @@ static ResultCode GetThreadContext(VAddr thread_context, Handle handle) {
         return ERR_INVALID_HANDLE;
     }
 
-    const auto current_process = Core::CurrentProcess();
-    if (thread->owner_process != current_process) {
+    const auto* current_process = Core::CurrentProcess();
+    if (thread->GetOwnerProcess() != current_process) {
         return ERR_INVALID_HANDLE;
     }
 
@@ -433,7 +448,7 @@ static ResultCode GetThreadContext(VAddr thread_context, Handle handle) {
         return ERR_ALREADY_REGISTERED;
     }
 
-    Core::ARM_Interface::ThreadContext ctx = thread->context;
+    Core::ARM_Interface::ThreadContext ctx = thread->GetContext();
     // Mask away mode bits, interrupt bits, IL bit, and other reserved bits.
     ctx.pstate &= 0xFF0FFE20;
 
@@ -479,14 +494,14 @@ static ResultCode SetThreadPriority(Handle handle, u32 priority) {
 
     thread->SetPriority(priority);
 
-    Core::System::GetInstance().CpuCore(thread->processor_id).PrepareReschedule();
+    Core::System::GetInstance().CpuCore(thread->GetProcessorID()).PrepareReschedule();
     return RESULT_SUCCESS;
 }
 
 /// Get which CPU core is executing the current thread
 static u32 GetCurrentProcessorNumber() {
     LOG_TRACE(Kernel_SVC, "called");
-    return GetCurrentThread()->processor_id;
+    return GetCurrentThread()->GetProcessorID();
 }
 
 static ResultCode MapSharedMemory(Handle shared_memory_handle, VAddr addr, u64 size,
@@ -516,7 +531,7 @@ static ResultCode MapSharedMemory(Handle shared_memory_handle, VAddr addr, u64 s
         return ERR_INVALID_HANDLE;
     }
 
-    return shared_memory->Map(Core::CurrentProcess().get(), addr, permissions_type,
+    return shared_memory->Map(Core::CurrentProcess(), addr, permissions_type,
                               MemoryPermission::DontCare);
 }
 
@@ -535,7 +550,7 @@ static ResultCode UnmapSharedMemory(Handle shared_memory_handle, VAddr addr, u64
     auto& kernel = Core::System::GetInstance().Kernel();
     auto shared_memory = kernel.HandleTable().Get<SharedMemory>(shared_memory_handle);
 
-    return shared_memory->Unmap(Core::CurrentProcess().get(), addr);
+    return shared_memory->Unmap(Core::CurrentProcess(), addr);
 }
 
 /// Query process memory
@@ -573,7 +588,7 @@ static ResultCode QueryMemory(MemoryInfo* memory_info, PageInfo* page_info, VAdd
 
 /// Exits the current process
 static void ExitProcess() {
-    auto& current_process = Core::CurrentProcess();
+    auto* current_process = Core::CurrentProcess();
 
     LOG_INFO(Kernel_SVC, "Process {} exiting", current_process->GetProcessID());
     ASSERT_MSG(current_process->GetStatus() == ProcessStatus::Running,
@@ -621,11 +636,15 @@ static ResultCode CreateThread(Handle* out_handle, VAddr entry_point, u64 arg, V
     auto& kernel = Core::System::GetInstance().Kernel();
     CASCADE_RESULT(SharedPtr<Thread> thread,
                    Thread::Create(kernel, name, entry_point, priority, arg, processor_id, stack_top,
-                                  Core::CurrentProcess()));
-    CASCADE_RESULT(thread->guest_handle, kernel.HandleTable().Create(thread));
-    *out_handle = thread->guest_handle;
+                                  *Core::CurrentProcess()));
+    const auto new_guest_handle = kernel.HandleTable().Create(thread);
+    if (new_guest_handle.Failed()) {
+        return new_guest_handle.Code();
+    }
+    thread->SetGuestHandle(*new_guest_handle);
+    *out_handle = *new_guest_handle;
 
-    Core::System::GetInstance().CpuCore(thread->processor_id).PrepareReschedule();
+    Core::System::GetInstance().CpuCore(thread->GetProcessorID()).PrepareReschedule();
 
     LOG_TRACE(Kernel_SVC,
               "called entrypoint=0x{:08X} ({}), arg=0x{:08X}, stacktop=0x{:08X}, "
@@ -645,10 +664,10 @@ static ResultCode StartThread(Handle thread_handle) {
         return ERR_INVALID_HANDLE;
     }
 
-    ASSERT(thread->status == ThreadStatus::Dormant);
+    ASSERT(thread->GetStatus() == ThreadStatus::Dormant);
 
     thread->ResumeFromWait();
-    Core::System::GetInstance().CpuCore(thread->processor_id).PrepareReschedule();
+    Core::System::GetInstance().CpuCore(thread->GetProcessorID()).PrepareReschedule();
 
     return RESULT_SUCCESS;
 }
@@ -694,17 +713,17 @@ static ResultCode WaitProcessWideKeyAtomic(VAddr mutex_addr, VAddr condition_var
     CASCADE_CODE(Mutex::Release(mutex_addr));
 
     SharedPtr<Thread> current_thread = GetCurrentThread();
-    current_thread->condvar_wait_address = condition_variable_addr;
-    current_thread->mutex_wait_address = mutex_addr;
-    current_thread->wait_handle = thread_handle;
-    current_thread->status = ThreadStatus::WaitMutex;
-    current_thread->wakeup_callback = nullptr;
+    current_thread->SetCondVarWaitAddress(condition_variable_addr);
+    current_thread->SetMutexWaitAddress(mutex_addr);
+    current_thread->SetWaitHandle(thread_handle);
+    current_thread->SetStatus(ThreadStatus::WaitMutex);
+    current_thread->InvalidateWakeupCallback();
 
     current_thread->WakeAfterDelay(nano_seconds);
 
     // Note: Deliberately don't attempt to inherit the lock owner's priority.
 
-    Core::System::GetInstance().CpuCore(current_thread->processor_id).PrepareReschedule();
+    Core::System::GetInstance().CpuCore(current_thread->GetProcessorID()).PrepareReschedule();
     return RESULT_SUCCESS;
 }
 
@@ -713,14 +732,14 @@ static ResultCode SignalProcessWideKey(VAddr condition_variable_addr, s32 target
     LOG_TRACE(Kernel_SVC, "called, condition_variable_addr=0x{:X}, target=0x{:08X}",
               condition_variable_addr, target);
 
-    auto RetrieveWaitingThreads = [](std::size_t core_index,
-                                     std::vector<SharedPtr<Thread>>& waiting_threads,
-                                     VAddr condvar_addr) {
+    const auto RetrieveWaitingThreads = [](std::size_t core_index,
+                                           std::vector<SharedPtr<Thread>>& waiting_threads,
+                                           VAddr condvar_addr) {
         const auto& scheduler = Core::System::GetInstance().Scheduler(core_index);
-        auto& thread_list = scheduler->GetThreadList();
+        const auto& thread_list = scheduler->GetThreadList();
 
-        for (auto& thread : thread_list) {
-            if (thread->condvar_wait_address == condvar_addr)
+        for (const auto& thread : thread_list) {
+            if (thread->GetCondVarWaitAddress() == condvar_addr)
                 waiting_threads.push_back(thread);
         }
     };
@@ -734,7 +753,7 @@ static ResultCode SignalProcessWideKey(VAddr condition_variable_addr, s32 target
     // Sort them by priority, such that the highest priority ones come first.
     std::sort(waiting_threads.begin(), waiting_threads.end(),
               [](const SharedPtr<Thread>& lhs, const SharedPtr<Thread>& rhs) {
-                  return lhs->current_priority < rhs->current_priority;
+                  return lhs->GetPriority() < rhs->GetPriority();
               });
 
     // Only process up to 'target' threads, unless 'target' is -1, in which case process
@@ -750,7 +769,7 @@ static ResultCode SignalProcessWideKey(VAddr condition_variable_addr, s32 target
     for (std::size_t index = 0; index < last; ++index) {
         auto& thread = waiting_threads[index];
 
-        ASSERT(thread->condvar_wait_address == condition_variable_addr);
+        ASSERT(thread->GetCondVarWaitAddress() == condition_variable_addr);
 
         std::size_t current_core = Core::System::GetInstance().CurrentCoreIndex();
 
@@ -759,42 +778,43 @@ static ResultCode SignalProcessWideKey(VAddr condition_variable_addr, s32 target
         // Atomically read the value of the mutex.
         u32 mutex_val = 0;
         do {
-            monitor.SetExclusive(current_core, thread->mutex_wait_address);
+            monitor.SetExclusive(current_core, thread->GetMutexWaitAddress());
 
             // If the mutex is not yet acquired, acquire it.
-            mutex_val = Memory::Read32(thread->mutex_wait_address);
+            mutex_val = Memory::Read32(thread->GetMutexWaitAddress());
 
             if (mutex_val != 0) {
                 monitor.ClearExclusive();
                 break;
             }
-        } while (!monitor.ExclusiveWrite32(current_core, thread->mutex_wait_address,
-                                           thread->wait_handle));
+        } while (!monitor.ExclusiveWrite32(current_core, thread->GetMutexWaitAddress(),
+                                           thread->GetWaitHandle()));
 
         if (mutex_val == 0) {
             // We were able to acquire the mutex, resume this thread.
-            ASSERT(thread->status == ThreadStatus::WaitMutex);
+            ASSERT(thread->GetStatus() == ThreadStatus::WaitMutex);
             thread->ResumeFromWait();
 
-            auto lock_owner = thread->lock_owner;
-            if (lock_owner)
+            auto* const lock_owner = thread->GetLockOwner();
+            if (lock_owner != nullptr) {
                 lock_owner->RemoveMutexWaiter(thread);
+            }
 
-            thread->lock_owner = nullptr;
-            thread->mutex_wait_address = 0;
-            thread->condvar_wait_address = 0;
-            thread->wait_handle = 0;
+            thread->SetLockOwner(nullptr);
+            thread->SetMutexWaitAddress(0);
+            thread->SetCondVarWaitAddress(0);
+            thread->SetWaitHandle(0);
         } else {
             // Atomically signal that the mutex now has a waiting thread.
             do {
-                monitor.SetExclusive(current_core, thread->mutex_wait_address);
+                monitor.SetExclusive(current_core, thread->GetMutexWaitAddress());
 
                 // Ensure that the mutex value is still what we expect.
-                u32 value = Memory::Read32(thread->mutex_wait_address);
+                u32 value = Memory::Read32(thread->GetMutexWaitAddress());
                 // TODO(Subv): When this happens, the kernel just clears the exclusive state and
                 // retries the initial read for this thread.
                 ASSERT_MSG(mutex_val == value, "Unhandled synchronization primitive case");
-            } while (!monitor.ExclusiveWrite32(current_core, thread->mutex_wait_address,
+            } while (!monitor.ExclusiveWrite32(current_core, thread->GetMutexWaitAddress(),
                                                mutex_val | Mutex::MutexHasWaitersFlag));
 
             // The mutex is already owned by some other thread, make this thread wait on it.
@@ -802,12 +822,12 @@ static ResultCode SignalProcessWideKey(VAddr condition_variable_addr, s32 target
             Handle owner_handle = static_cast<Handle>(mutex_val & Mutex::MutexOwnerMask);
             auto owner = kernel.HandleTable().Get<Thread>(owner_handle);
             ASSERT(owner);
-            ASSERT(thread->status == ThreadStatus::WaitMutex);
-            thread->wakeup_callback = nullptr;
+            ASSERT(thread->GetStatus() == ThreadStatus::WaitMutex);
+            thread->InvalidateWakeupCallback();
 
             owner->AddMutexWaiter(thread);
 
-            Core::System::GetInstance().CpuCore(thread->processor_id).PrepareReschedule();
+            Core::System::GetInstance().CpuCore(thread->GetProcessorID()).PrepareReschedule();
         }
     }
 
@@ -913,8 +933,8 @@ static ResultCode GetThreadCoreMask(Handle thread_handle, u32* core, u64* mask)
         return ERR_INVALID_HANDLE;
     }
 
-    *core = thread->ideal_core;
-    *mask = thread->affinity_mask;
+    *core = thread->GetIdealCore();
+    *mask = thread->GetAffinityMask();
 
     return RESULT_SUCCESS;
 }
@@ -930,11 +950,13 @@ static ResultCode SetThreadCoreMask(Handle thread_handle, u32 core, u64 mask) {
     }
 
     if (core == static_cast<u32>(THREADPROCESSORID_DEFAULT)) {
-        ASSERT(thread->owner_process->GetDefaultProcessorID() !=
-               static_cast<u8>(THREADPROCESSORID_DEFAULT));
+        const u8 default_processor_id = thread->GetOwnerProcess()->GetDefaultProcessorID();
+
+        ASSERT(default_processor_id != static_cast<u8>(THREADPROCESSORID_DEFAULT));
+
         // Set the target CPU to the one specified in the process' exheader.
-        core = thread->owner_process->GetDefaultProcessorID();
-        mask = 1ull << core;
+        core = default_processor_id;
+        mask = 1ULL << core;
     }
 
     if (mask == 0) {
@@ -945,7 +967,7 @@ static ResultCode SetThreadCoreMask(Handle thread_handle, u32 core, u64 mask) {
     static constexpr u32 OnlyChangeMask = static_cast<u32>(-3);
 
     if (core == OnlyChangeMask) {
-        core = thread->ideal_core;
+        core = thread->GetIdealCore();
     } else if (core >= Core::NUM_CPU_CORES && core != static_cast<u32>(-1)) {
         return ResultCode(ErrorModule::Kernel, ErrCodes::InvalidProcessorId);
     }

src/core/hle/kernel/thread.cpp

@@ -70,7 +70,7 @@ void Thread::Stop() {
 
 void WaitCurrentThread_Sleep() {
     Thread* thread = GetCurrentThread();
-    thread->status = ThreadStatus::WaitSleep;
+    thread->SetStatus(ThreadStatus::WaitSleep);
 }
 
 void ExitCurrentThread() {
@@ -169,7 +169,7 @@ void Thread::ResumeFromWait() {
     next_scheduler->ScheduleThread(this, current_priority);
 
     // Change thread's scheduler
-    scheduler = next_scheduler;
+    scheduler = next_scheduler.get();
 
     Core::System::GetInstance().CpuCore(processor_id).PrepareReschedule();
 }
@@ -194,7 +194,7 @@ static void ResetThreadContext(Core::ARM_Interface::ThreadContext& context, VAdd
 
 ResultVal<SharedPtr<Thread>> Thread::Create(KernelCore& kernel, std::string name, VAddr entry_point,
                                             u32 priority, u64 arg, s32 processor_id,
-                                            VAddr stack_top, SharedPtr<Process> owner_process) {
+                                            VAddr stack_top, Process& owner_process) {
     // Check if priority is in ranged. Lowest priority -> highest priority id.
     if (priority > THREADPRIO_LOWEST) {
         LOG_ERROR(Kernel_SVC, "Invalid thread priority: {}", priority);
@@ -208,7 +208,7 @@ ResultVal<SharedPtr<Thread>> Thread::Create(KernelCore& kernel, std::string name
 
     // TODO(yuriks): Other checks, returning 0xD9001BEA
 
-    if (!Memory::IsValidVirtualAddress(*owner_process, entry_point)) {
+    if (!Memory::IsValidVirtualAddress(owner_process, entry_point)) {
         LOG_ERROR(Kernel_SVC, "(name={}): invalid entry {:016X}", name, entry_point);
         // TODO (bunnei): Find the correct error code to use here
         return ResultCode(-1);
@@ -232,8 +232,8 @@ ResultVal<SharedPtr<Thread>> Thread::Create(KernelCore& kernel, std::string name
     thread->wait_handle = 0;
     thread->name = std::move(name);
     thread->callback_handle = kernel.ThreadWakeupCallbackHandleTable().Create(thread).Unwrap();
-    thread->owner_process = owner_process;
-    thread->scheduler = Core::System::GetInstance().Scheduler(processor_id);
+    thread->owner_process = &owner_process;
+    thread->scheduler = Core::System::GetInstance().Scheduler(processor_id).get();
     thread->scheduler->AddThread(thread, priority);
     thread->tls_address = thread->owner_process->MarkNextAvailableTLSSlotAsUsed(*thread);
 
@@ -264,14 +264,14 @@ SharedPtr<Thread> SetupMainThread(KernelCore& kernel, VAddr entry_point, u32 pri
     // Initialize new "main" thread
     const VAddr stack_top = owner_process.VMManager().GetTLSIORegionEndAddress();
     auto thread_res = Thread::Create(kernel, "main", entry_point, priority, 0, THREADPROCESSORID_0,
-                                     stack_top, &owner_process);
+                                     stack_top, owner_process);
 
     SharedPtr<Thread> thread = std::move(thread_res).Unwrap();
 
     // Register 1 must be a handle to the main thread
-    thread->guest_handle = kernel.HandleTable().Create(thread).Unwrap();
-
-    thread->context.cpu_registers[1] = thread->guest_handle;
+    const Handle guest_handle = kernel.HandleTable().Create(thread).Unwrap();
+    thread->SetGuestHandle(guest_handle);
+    thread->GetContext().cpu_registers[1] = guest_handle;
 
     // Threads by default are dormant, wake up the main thread so it runs when the scheduler fires
     thread->ResumeFromWait();
@@ -299,6 +299,18 @@ VAddr Thread::GetCommandBufferAddress() const {
     return GetTLSAddress() + CommandHeaderOffset;
 }
 
+void Thread::SetStatus(ThreadStatus new_status) {
+    if (new_status == status) {
+        return;
+    }
+
+    if (status == ThreadStatus::Running) {
+        last_running_ticks = CoreTiming::GetTicks();
+    }
+
+    status = new_status;
+}
+
 void Thread::AddMutexWaiter(SharedPtr<Thread> thread) {
     if (thread->lock_owner == this) {
         // If the thread is already waiting for this thread to release the mutex, ensure that the
@@ -388,11 +400,23 @@ void Thread::ChangeCore(u32 core, u64 mask) {
     next_scheduler->ScheduleThread(this, current_priority);
 
     // Change thread's scheduler
-    scheduler = next_scheduler;
+    scheduler = next_scheduler.get();
 
     Core::System::GetInstance().CpuCore(processor_id).PrepareReschedule();
 }
 
+bool Thread::AllWaitObjectsReady() {
+    return std::none_of(
+        wait_objects.begin(), wait_objects.end(),
+        [this](const SharedPtr<WaitObject>& object) { return object->ShouldWait(this); });
+}
+
+bool Thread::InvokeWakeupCallback(ThreadWakeupReason reason, SharedPtr<Thread> thread,
+                                  SharedPtr<WaitObject> object, std::size_t index) {
+    ASSERT(wakeup_callback);
+    return wakeup_callback(reason, std::move(thread), std::move(object), index);
+}
+
 ////////////////////////////////////////////////////////////////////////////////////////////////////
 
 /**

src/core/hle/kernel/thread.h

@@ -65,6 +65,15 @@ public:
     using TLSMemory = std::vector<u8>;
     using TLSMemoryPtr = std::shared_ptr<TLSMemory>;
 
+    using MutexWaitingThreads = std::vector<SharedPtr<Thread>>;
+
+    using ThreadContext = Core::ARM_Interface::ThreadContext;
+
+    using ThreadWaitObjects = std::vector<SharedPtr<WaitObject>>;
+
+    using WakeupCallback = std::function<bool(ThreadWakeupReason reason, SharedPtr<Thread> thread,
+                                              SharedPtr<WaitObject> object, std::size_t index)>;
+
     /**
      * Creates and returns a new thread. The new thread is immediately scheduled
      * @param kernel The kernel instance this thread will be created under.
@@ -80,7 +89,7 @@ public:
     static ResultVal<SharedPtr<Thread>> Create(KernelCore& kernel, std::string name,
                                                VAddr entry_point, u32 priority, u64 arg,
                                                s32 processor_id, VAddr stack_top,
-                                               SharedPtr<Process> owner_process);
+                                               Process& owner_process);
 
     std::string GetName() const override {
         return name;
@@ -105,6 +114,14 @@ public:
         return current_priority;
     }
 
+    /**
+     * Gets the thread's nominal priority.
+     * @return The current thread's nominal priority.
+     */
+    u32 GetNominalPriority() const {
+        return nominal_priority;
+    }
+
     /**
      * Sets the thread's current priority
      * @param priority The new priority
@@ -133,7 +150,7 @@ public:
      * Gets the thread's thread ID
      * @return The thread's ID
      */
-    u32 GetThreadId() const {
+    u32 GetThreadID() const {
         return thread_id;
     }
 
@@ -203,6 +220,11 @@ public:
         return tpidr_el0;
     }
 
+    /// Sets the value of the TPIDR_EL0 Read/Write system register for this thread.
+    void SetTPIDR_EL0(u64 value) {
+        tpidr_el0 = value;
+    }
+
     /*
      * Returns the address of the current thread's command buffer, located in the TLS.
      * @returns VAddr of the thread's command buffer.
@@ -218,69 +240,193 @@ public:
         return status == ThreadStatus::WaitSynchAll;
     }
 
-    Core::ARM_Interface::ThreadContext context;
+    ThreadContext& GetContext() {
+        return context;
+    }
 
-    u32 thread_id;
+    const ThreadContext& GetContext() const {
+        return context;
+    }
 
-    ThreadStatus status;
-    VAddr entry_point;
-    VAddr stack_top;
+    ThreadStatus GetStatus() const {
+        return status;
+    }
 
-    u32 nominal_priority; ///< Nominal thread priority, as set by the emulated application
-    u32 current_priority; ///< Current thread priority, can be temporarily changed
+    void SetStatus(ThreadStatus new_status);
 
-    u64 last_running_ticks; ///< CPU tick when thread was last running
+    u64 GetLastRunningTicks() const {
+        return last_running_ticks;
+    }
 
-    s32 processor_id;
+    s32 GetProcessorID() const {
+        return processor_id;
+    }
 
-    VAddr tls_address; ///< Virtual address of the Thread Local Storage of the thread
-    u64 tpidr_el0;     ///< TPIDR_EL0 read/write system register.
+    Process* GetOwnerProcess() {
+        return owner_process;
+    }
 
-    SharedPtr<Process> owner_process; ///< Process that owns this thread
+    const Process* GetOwnerProcess() const {
+        return owner_process;
+    }
 
-    /// Objects that the thread is waiting on, in the same order as they were
-    // passed to WaitSynchronization1/N.
-    std::vector<SharedPtr<WaitObject>> wait_objects;
+    const ThreadWaitObjects& GetWaitObjects() const {
+        return wait_objects;
+    }
 
-    /// List of threads that are waiting for a mutex that is held by this thread.
-    std::vector<SharedPtr<Thread>> wait_mutex_threads;
+    void SetWaitObjects(ThreadWaitObjects objects) {
+        wait_objects = std::move(objects);
+    }
 
-    /// Thread that owns the lock that this thread is waiting for.
-    SharedPtr<Thread> lock_owner;
+    void ClearWaitObjects() {
+        wait_objects.clear();
+    }
 
-    // If waiting on a ConditionVariable, this is the ConditionVariable  address
-    VAddr condvar_wait_address;
-    VAddr mutex_wait_address; ///< If waiting on a Mutex, this is the mutex address
-    Handle wait_handle;       ///< The handle used to wait for the mutex.
+    /// Determines whether all the objects this thread is waiting on are ready.
+    bool AllWaitObjectsReady();
 
-    // If waiting for an AddressArbiter, this is the address being waited on.
-    VAddr arb_wait_address{0};
+    const MutexWaitingThreads& GetMutexWaitingThreads() const {
+        return wait_mutex_threads;
+    }
 
-    std::string name;
+    Thread* GetLockOwner() const {
+        return lock_owner.get();
+    }
 
-    /// Handle used by guest emulated application to access this thread
-    Handle guest_handle;
+    void SetLockOwner(SharedPtr<Thread> owner) {
+        lock_owner = std::move(owner);
+    }
 
-    /// Handle used as userdata to reference this object when inserting into the CoreTiming queue.
-    Handle callback_handle;
+    VAddr GetCondVarWaitAddress() const {
+        return condvar_wait_address;
+    }
 
-    using WakeupCallback = bool(ThreadWakeupReason reason, SharedPtr<Thread> thread,
-                                SharedPtr<WaitObject> object, std::size_t index);
-    // Callback that will be invoked when the thread is resumed from a waiting state. If the thread
-    // was waiting via WaitSynchronizationN then the object will be the last object that became
-    // available. In case of a timeout, the object will be nullptr.
-    std::function<WakeupCallback> wakeup_callback;
+    void SetCondVarWaitAddress(VAddr address) {
+        condvar_wait_address = address;
+    }
 
-    std::shared_ptr<Scheduler> scheduler;
+    VAddr GetMutexWaitAddress() const {
+        return mutex_wait_address;
+    }
 
-    u32 ideal_core{0xFFFFFFFF};
-    u64 affinity_mask{0x1};
+    void SetMutexWaitAddress(VAddr address) {
+        mutex_wait_address = address;
+    }
+
+    Handle GetWaitHandle() const {
+        return wait_handle;
+    }
+
+    void SetWaitHandle(Handle handle) {
+        wait_handle = handle;
+    }
+
+    VAddr GetArbiterWaitAddress() const {
+        return arb_wait_address;
+    }
+
+    void SetArbiterWaitAddress(VAddr address) {
+        arb_wait_address = address;
+    }
+
+    void SetGuestHandle(Handle handle) {
+        guest_handle = handle;
+    }
+
+    bool HasWakeupCallback() const {
+        return wakeup_callback != nullptr;
+    }
+
+    void SetWakeupCallback(WakeupCallback callback) {
+        wakeup_callback = std::move(callback);
+    }
+
+    void InvalidateWakeupCallback() {
+        SetWakeupCallback(nullptr);
+    }
+
+    /**
+     * Invokes the thread's wakeup callback.
+     *
+     * @pre A valid wakeup callback has been set. Violating this precondition
+     *      will cause an assertion to trigger.
+     */
+    bool InvokeWakeupCallback(ThreadWakeupReason reason, SharedPtr<Thread> thread,
+                              SharedPtr<WaitObject> object, std::size_t index);
+
+    u32 GetIdealCore() const {
+        return ideal_core;
+    }
+
+    u64 GetAffinityMask() const {
+        return affinity_mask;
+    }
 
 private:
     explicit Thread(KernelCore& kernel);
     ~Thread() override;
 
+    Core::ARM_Interface::ThreadContext context{};
+
+    u32 thread_id = 0;
+
+    ThreadStatus status = ThreadStatus::Dormant;
+
+    VAddr entry_point = 0;
+    VAddr stack_top = 0;
+
+    u32 nominal_priority = 0; ///< Nominal thread priority, as set by the emulated application
+    u32 current_priority = 0; ///< Current thread priority, can be temporarily changed
+
+    u64 last_running_ticks = 0; ///< CPU tick when thread was last running
+
+    s32 processor_id = 0;
+
+    VAddr tls_address = 0; ///< Virtual address of the Thread Local Storage of the thread
+    u64 tpidr_el0 = 0;     ///< TPIDR_EL0 read/write system register.
+
+    /// Process that owns this thread
+    Process* owner_process;
+
+    /// Objects that the thread is waiting on, in the same order as they were
+    /// passed to WaitSynchronization1/N.
+    ThreadWaitObjects wait_objects;
+
+    /// List of threads that are waiting for a mutex that is held by this thread.
+    MutexWaitingThreads wait_mutex_threads;
+
+    /// Thread that owns the lock that this thread is waiting for.
+    SharedPtr<Thread> lock_owner;
+
+    /// If waiting on a ConditionVariable, this is the ConditionVariable address
+    VAddr condvar_wait_address = 0;
+    /// If waiting on a Mutex, this is the mutex address
+    VAddr mutex_wait_address = 0;
+    /// The handle used to wait for the mutex.
+    Handle wait_handle = 0;
+
+    /// If waiting for an AddressArbiter, this is the address being waited on.
+    VAddr arb_wait_address{0};
+
+    /// Handle used by guest emulated application to access this thread
+    Handle guest_handle = 0;
+
+    /// Handle used as userdata to reference this object when inserting into the CoreTiming queue.
+    Handle callback_handle = 0;
+
+    /// Callback that will be invoked when the thread is resumed from a waiting state. If the thread
+    /// was waiting via WaitSynchronizationN then the object will be the last object that became
+    /// available. In case of a timeout, the object will be nullptr.
+    WakeupCallback wakeup_callback;
+
+    Scheduler* scheduler = nullptr;
+
+    u32 ideal_core{0xFFFFFFFF};
+    u64 affinity_mask{0x1};
+
     TLSMemoryPtr tls_memory = std::make_shared<TLSMemory>();
+
+    std::string name;
 };
 
 /**

src/core/hle/kernel/wait_object.cpp

@@ -35,13 +35,15 @@ SharedPtr<Thread> WaitObject::GetHighestPriorityReadyThread() {
     u32 candidate_priority = THREADPRIO_LOWEST + 1;
 
     for (const auto& thread : waiting_threads) {
+        const ThreadStatus thread_status = thread->GetStatus();
+
         // The list of waiting threads must not contain threads that are not waiting to be awakened.
-        ASSERT_MSG(thread->status == ThreadStatus::WaitSynchAny ||
-                       thread->status == ThreadStatus::WaitSynchAll ||
-                       thread->status == ThreadStatus::WaitHLEEvent,
+        ASSERT_MSG(thread_status == ThreadStatus::WaitSynchAny ||
+                       thread_status == ThreadStatus::WaitSynchAll ||
+                       thread_status == ThreadStatus::WaitHLEEvent,
                    "Inconsistent thread statuses in waiting_threads");
 
-        if (thread->current_priority >= candidate_priority)
+        if (thread->GetPriority() >= candidate_priority)
             continue;
 
         if (ShouldWait(thread.get()))
@@ -50,16 +52,13 @@ SharedPtr<Thread> WaitObject::GetHighestPriorityReadyThread() {
         // A thread is ready to run if it's either in ThreadStatus::WaitSynchAny or
         // in ThreadStatus::WaitSynchAll and the rest of the objects it is waiting on are ready.
         bool ready_to_run = true;
-        if (thread->status == ThreadStatus::WaitSynchAll) {
-            ready_to_run = std::none_of(thread->wait_objects.begin(), thread->wait_objects.end(),
-                                        [&thread](const SharedPtr<WaitObject>& object) {
-                                            return object->ShouldWait(thread.get());
-                                        });
+        if (thread_status == ThreadStatus::WaitSynchAll) {
+            ready_to_run = thread->AllWaitObjectsReady();
         }
 
         if (ready_to_run) {
             candidate = thread.get();
-            candidate_priority = thread->current_priority;
+            candidate_priority = thread->GetPriority();
         }
     }
 
@@ -75,24 +74,24 @@ void WaitObject::WakeupWaitingThread(SharedPtr<Thread> thread) {
     if (!thread->IsSleepingOnWaitAll()) {
         Acquire(thread.get());
     } else {
-        for (auto& object : thread->wait_objects) {
+        for (const auto& object : thread->GetWaitObjects()) {
             ASSERT(!object->ShouldWait(thread.get()));
             object->Acquire(thread.get());
         }
     }
 
-    std::size_t index = thread->GetWaitObjectIndex(this);
+    const std::size_t index = thread->GetWaitObjectIndex(this);
 
-    for (auto& object : thread->wait_objects)
+    for (const auto& object : thread->GetWaitObjects())
         object->RemoveWaitingThread(thread.get());
-    thread->wait_objects.clear();
+    thread->ClearWaitObjects();
 
     thread->CancelWakeupTimer();
 
     bool resume = true;
 
-    if (thread->wakeup_callback)
-        resume = thread->wakeup_callback(ThreadWakeupReason::Signal, thread, this, index);
+    if (thread->HasWakeupCallback())
+        resume = thread->InvokeWakeupCallback(ThreadWakeupReason::Signal, thread, this, index);
 
     if (resume)
         thread->ResumeFromWait();

src/core/hle/service/aoc/aoc_u.cpp

@@ -61,13 +61,13 @@ AOC_U::AOC_U() : ServiceFramework("aoc:u"), add_on_content(AccumulateAOCTitleIDs
 AOC_U::~AOC_U() = default;
 
 void AOC_U::CountAddOnContent(Kernel::HLERequestContext& ctx) {
-    IPC::ResponseBuilder rb{ctx, 4};
+    IPC::ResponseBuilder rb{ctx, 3};
     rb.Push(RESULT_SUCCESS);
 
     const auto current = Core::System::GetInstance().CurrentProcess()->GetTitleID();
-    rb.Push<u32>(std::count_if(add_on_content.begin(), add_on_content.end(), [&current](u64 tid) {
-        return (tid & DLC_BASE_TITLE_ID_MASK) == current;
-    }));
+    rb.Push<u32>(static_cast<u32>(
+        std::count_if(add_on_content.begin(), add_on_content.end(),
+                      [&current](u64 tid) { return (tid & DLC_BASE_TITLE_ID_MASK) == current; })));
 }
 
 void AOC_U::ListAddOnContent(Kernel::HLERequestContext& ctx) {
@@ -91,7 +91,7 @@ void AOC_U::ListAddOnContent(Kernel::HLERequestContext& ctx) {
         return;
     }
 
-    count = std::min<size_t>(out.size() - offset, count);
+    count = static_cast<u32>(std::min<size_t>(out.size() - offset, count));
     std::rotate(out.begin(), out.begin() + offset, out.end());
     out.resize(count);
 

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

@@ -264,6 +264,15 @@ ResultCode RegisterBIS(std::unique_ptr<FileSys::BISFactory>&& factory) {
     return RESULT_SUCCESS;
 }
 
+void SetPackedUpdate(FileSys::VirtualFile update_raw) {
+    LOG_TRACE(Service_FS, "Setting packed update for romfs");
+
+    if (romfs_factory == nullptr)
+        return;
+
+    romfs_factory->SetPackedUpdate(std::move(update_raw));
+}
+
 ResultVal<FileSys::VirtualFile> OpenRomFSCurrentProcess() {
     LOG_TRACE(Service_FS, "Opening RomFS for current process");
 

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

@@ -39,6 +39,7 @@ ResultCode RegisterSaveData(std::unique_ptr<FileSys::SaveDataFactory>&& factory)
 ResultCode RegisterSDMC(std::unique_ptr<FileSys::SDMCFactory>&& factory);
 ResultCode RegisterBIS(std::unique_ptr<FileSys::BISFactory>&& factory);
 
+void SetPackedUpdate(FileSys::VirtualFile update_raw);
 ResultVal<FileSys::VirtualFile> OpenRomFSCurrentProcess();
 ResultVal<FileSys::VirtualFile> OpenRomFS(u64 title_id, FileSys::StorageId storage_id,
                                           FileSys::ContentRecordType type);

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

@@ -157,7 +157,12 @@ u32 nvhost_as_gpu::UnmapBuffer(const std::vector<u8>& input, std::vector<u8>& ou
     LOG_DEBUG(Service_NVDRV, "called, offset=0x{:X}", params.offset);
 
     const auto itr = buffer_mappings.find(params.offset);
-    ASSERT_MSG(itr != buffer_mappings.end(), "Tried to unmap invalid mapping");
+    if (itr == buffer_mappings.end()) {
+        LOG_WARNING(Service_NVDRV, "Tried to unmap an invalid offset 0x{:X}", params.offset);
+        // Hardware tests shows that unmapping an already unmapped buffer always returns successful
+        // and doesn't fail.
+        return 0;
+    }
 
     auto& system_instance = Core::System::GetInstance();
 

src/core/hle/service/sm/sm.h

@@ -6,9 +6,12 @@
 
 #include <memory>
 #include <string>
+#include <type_traits>
 #include <unordered_map>
 
+#include "core/hle/kernel/client_port.h"
 #include "core/hle/kernel/object.h"
+#include "core/hle/kernel/server_port.h"
 #include "core/hle/result.h"
 #include "core/hle/service/service.h"
 
@@ -48,6 +51,22 @@ public:
     ResultVal<Kernel::SharedPtr<Kernel::ClientPort>> GetServicePort(const std::string& name);
     ResultVal<Kernel::SharedPtr<Kernel::ClientSession>> ConnectToService(const std::string& name);
 
+    template <typename T>
+    std::shared_ptr<T> GetService(const std::string& service_name) const {
+        static_assert(std::is_base_of_v<Kernel::SessionRequestHandler, T>,
+                      "Not a base of ServiceFrameworkBase");
+        auto service = registered_services.find(service_name);
+        if (service == registered_services.end()) {
+            LOG_DEBUG(Service, "Can't find service: {}", service_name);
+            return nullptr;
+        }
+        auto port = service->second->GetServerPort();
+        if (port == nullptr) {
+            return nullptr;
+        }
+        return std::static_pointer_cast<T>(port->hle_handler);
+    }
+
     void InvokeControlRequest(Kernel::HLERequestContext& context);
 
 private:

src/core/loader/deconstructed_rom_directory.cpp

@@ -3,6 +3,7 @@
 // Refer to the license.txt file included.
 
 #include <cinttypes>
+#include <cstring>
 #include "common/common_funcs.h"
 #include "common/file_util.h"
 #include "common/logging/log.h"
@@ -140,7 +141,8 @@ ResultStatus AppLoader_DeconstructedRomDirectory::Load(Kernel::Process& process)
         const FileSys::VirtualFile module_file = dir->GetFile(module);
         if (module_file != nullptr) {
             const VAddr load_addr = next_load_addr;
-            next_load_addr = AppLoader_NSO::LoadModule(module_file, load_addr, pm);
+            next_load_addr = AppLoader_NSO::LoadModule(module_file, load_addr,
+                                                       std::strcmp(module, "rtld") == 0, pm);
             LOG_DEBUG(Loader, "loaded module {} @ 0x{:X}", module, load_addr);
             // Register module with GDBStub
             GDBStub::RegisterModule(module, load_addr, next_load_addr - 1, false);

src/core/loader/loader.cpp

@@ -93,7 +93,7 @@ std::string GetFileTypeString(FileType type) {
     return "unknown";
 }
 
-constexpr std::array<const char*, 58> RESULT_MESSAGES{
+constexpr std::array<const char*, 59> RESULT_MESSAGES{
     "The operation completed successfully.",
     "The loader requested to load is already loaded.",
     "The operation is not implemented.",
@@ -152,6 +152,7 @@ constexpr std::array<const char*, 58> RESULT_MESSAGES{
     "The BKTR-type NCA has a bad Relocation bucket.",
     "The BKTR-type NCA has a bad Subsection bucket.",
     "The BKTR-type NCA is missing the base RomFS.",
+    "The NSP or XCI does not contain an update in addition to the base game.",
 };
 
 std::ostream& operator<<(std::ostream& os, ResultStatus status) {

src/core/loader/loader.h

@@ -114,6 +114,7 @@ enum class ResultStatus : u16 {
     ErrorBadRelocationBuckets,
     ErrorBadSubsectionBuckets,
     ErrorMissingBKTRBaseRomFS,
+    ErrorNoPackedUpdate,
 };
 
 std::ostream& operator<<(std::ostream& os, ResultStatus status);
@@ -196,10 +197,19 @@ public:
     /**
      * Get the RomFS of the application
      * Since the RomFS can be huge, we return a file reference instead of copying to a buffer
-     * @param dir The directory containing the RomFS
+     * @param file The directory containing the RomFS
      * @return ResultStatus result of function
      */
-    virtual ResultStatus ReadRomFS(FileSys::VirtualFile& dir) {
+    virtual ResultStatus ReadRomFS(FileSys::VirtualFile& file) {
+        return ResultStatus::ErrorNotImplemented;
+    }
+
+    /**
+     * Get the raw update of the application, should it come packed with one
+     * @param file The raw update NCA file (Program-type
+     * @return ResultStatus result of function
+     */
+    virtual ResultStatus ReadUpdateRaw(FileSys::VirtualFile& file) {
         return ResultStatus::ErrorNotImplemented;
     }
 

src/core/loader/nax.cpp

@@ -72,6 +72,10 @@ ResultStatus AppLoader_NAX::ReadRomFS(FileSys::VirtualFile& dir) {
     return nca_loader->ReadRomFS(dir);
 }
 
+u64 AppLoader_NAX::ReadRomFSIVFCOffset() const {
+    return nca_loader->ReadRomFSIVFCOffset();
+}
+
 ResultStatus AppLoader_NAX::ReadProgramId(u64& out_program_id) {
     return nca_loader->ReadProgramId(out_program_id);
 }

src/core/loader/nax.h

@@ -36,6 +36,7 @@ public:
     ResultStatus Load(Kernel::Process& process) override;
 
     ResultStatus ReadRomFS(FileSys::VirtualFile& dir) override;
+    u64 ReadRomFSIVFCOffset() const override;
     ResultStatus ReadProgramId(u64& out_program_id) override;
 
 private:

src/core/loader/nro.cpp

@@ -18,7 +18,9 @@
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/vm_manager.h"
 #include "core/loader/nro.h"
+#include "core/loader/nso.h"
 #include "core/memory.h"
+#include "core/settings.h"
 
 namespace Loader {
 
@@ -150,6 +152,19 @@ bool AppLoader_NRO::LoadNro(FileSys::VirtualFile file, VAddr load_base) {
         codeset->segments[i].size = PageAlignSize(nro_header.segments[i].size);
     }
 
+    if (!Settings::values.program_args.empty()) {
+        const auto arg_data = Settings::values.program_args;
+        codeset->DataSegment().size += NSO_ARGUMENT_DATA_ALLOCATION_SIZE;
+        NSOArgumentHeader args_header{
+            NSO_ARGUMENT_DATA_ALLOCATION_SIZE, static_cast<u32_le>(arg_data.size()), {}};
+        const auto end_offset = program_image.size();
+        program_image.resize(static_cast<u32>(program_image.size()) +
+                             NSO_ARGUMENT_DATA_ALLOCATION_SIZE);
+        std::memcpy(program_image.data() + end_offset, &args_header, sizeof(NSOArgumentHeader));
+        std::memcpy(program_image.data() + end_offset + sizeof(NSOArgumentHeader), arg_data.data(),
+                    arg_data.size());
+    }
+
     // Read MOD header
     ModHeader mod_header{};
     // Default .bss to NRO header bss size if MOD0 section doesn't exist

src/core/loader/nso.cpp

@@ -17,6 +17,7 @@
 #include "core/hle/kernel/vm_manager.h"
 #include "core/loader/nso.h"
 #include "core/memory.h"
+#include "core/settings.h"
 
 namespace Loader {
 
@@ -94,6 +95,7 @@ static constexpr u32 PageAlignSize(u32 size) {
 }
 
 VAddr AppLoader_NSO::LoadModule(FileSys::VirtualFile file, VAddr load_base,
+                                bool should_pass_arguments,
                                 boost::optional<FileSys::PatchManager> pm) {
     if (file == nullptr)
         return {};
@@ -125,6 +127,19 @@ VAddr AppLoader_NSO::LoadModule(FileSys::VirtualFile file, VAddr load_base,
         codeset->segments[i].size = PageAlignSize(static_cast<u32>(data.size()));
     }
 
+    if (should_pass_arguments && !Settings::values.program_args.empty()) {
+        const auto arg_data = Settings::values.program_args;
+        codeset->DataSegment().size += NSO_ARGUMENT_DATA_ALLOCATION_SIZE;
+        NSOArgumentHeader args_header{
+            NSO_ARGUMENT_DATA_ALLOCATION_SIZE, static_cast<u32_le>(arg_data.size()), {}};
+        const auto end_offset = program_image.size();
+        program_image.resize(static_cast<u32>(program_image.size()) +
+                             NSO_ARGUMENT_DATA_ALLOCATION_SIZE);
+        std::memcpy(program_image.data() + end_offset, &args_header, sizeof(NSOArgumentHeader));
+        std::memcpy(program_image.data() + end_offset + sizeof(NSOArgumentHeader), arg_data.data(),
+                    arg_data.size());
+    }
+
     // MOD header pointer is at .text offset + 4
     u32 module_offset;
     std::memcpy(&module_offset, program_image.data() + 4, sizeof(u32));
@@ -172,7 +187,7 @@ ResultStatus AppLoader_NSO::Load(Kernel::Process& process) {
 
     // Load module
     const VAddr base_address = process.VMManager().GetCodeRegionBaseAddress();
-    LoadModule(file, base_address);
+    LoadModule(file, base_address, true);
     LOG_DEBUG(Loader, "loaded module {} @ 0x{:X}", file->GetName(), base_address);
 
     process.Run(base_address, Kernel::THREADPRIO_DEFAULT, Memory::DEFAULT_STACK_SIZE);

src/core/loader/nso.h

@@ -11,6 +11,15 @@
 
 namespace Loader {
 
+constexpr u64 NSO_ARGUMENT_DATA_ALLOCATION_SIZE = 0x9000;
+
+struct NSOArgumentHeader {
+    u32_le allocated_size;
+    u32_le actual_size;
+    INSERT_PADDING_BYTES(0x18);
+};
+static_assert(sizeof(NSOArgumentHeader) == 0x20, "NSOArgumentHeader has incorrect size.");
+
 /// Loads an NSO file
 class AppLoader_NSO final : public AppLoader, Linker {
 public:
@@ -27,7 +36,7 @@ public:
         return IdentifyType(file);
     }
 
-    static VAddr LoadModule(FileSys::VirtualFile file, VAddr load_base,
+    static VAddr LoadModule(FileSys::VirtualFile file, VAddr load_base, bool should_pass_arguments,
                             boost::optional<FileSys::PatchManager> pm = boost::none);
 
     ResultStatus Load(Kernel::Process& process) override;

src/core/loader/nsp.cpp

@@ -10,8 +10,10 @@
 #include "core/file_sys/control_metadata.h"
 #include "core/file_sys/nca_metadata.h"
 #include "core/file_sys/patch_manager.h"
+#include "core/file_sys/registered_cache.h"
 #include "core/file_sys/submission_package.h"
 #include "core/hle/kernel/process.h"
+#include "core/hle/service/filesystem/filesystem.h"
 #include "core/loader/deconstructed_rom_directory.h"
 #include "core/loader/nca.h"
 #include "core/loader/nsp.h"
@@ -33,7 +35,7 @@ AppLoader_NSP::AppLoader_NSP(FileSys::VirtualFile file)
         return;
 
     std::tie(nacp_file, icon_file) =
-        FileSys::PatchManager(nsp->GetProgramTitleID()).ParseControlNCA(control_nca);
+        FileSys::PatchManager(nsp->GetProgramTitleID()).ParseControlNCA(*control_nca);
 }
 
 AppLoader_NSP::~AppLoader_NSP() = default;
@@ -91,13 +93,39 @@ ResultStatus AppLoader_NSP::Load(Kernel::Process& process) {
     if (result != ResultStatus::Success)
         return result;
 
+    FileSys::VirtualFile update_raw;
+    if (ReadUpdateRaw(update_raw) == ResultStatus::Success && update_raw != nullptr)
+        Service::FileSystem::SetPackedUpdate(std::move(update_raw));
+
     is_loaded = true;
 
     return ResultStatus::Success;
 }
 
-ResultStatus AppLoader_NSP::ReadRomFS(FileSys::VirtualFile& dir) {
-    return secondary_loader->ReadRomFS(dir);
+ResultStatus AppLoader_NSP::ReadRomFS(FileSys::VirtualFile& file) {
+    return secondary_loader->ReadRomFS(file);
+}
+
+u64 AppLoader_NSP::ReadRomFSIVFCOffset() const {
+    return secondary_loader->ReadRomFSIVFCOffset();
+}
+
+ResultStatus AppLoader_NSP::ReadUpdateRaw(FileSys::VirtualFile& file) {
+    if (nsp->IsExtractedType())
+        return ResultStatus::ErrorNoPackedUpdate;
+
+    const auto read =
+        nsp->GetNCAFile(FileSys::GetUpdateTitleID(title_id), FileSys::ContentRecordType::Program);
+
+    if (read == nullptr)
+        return ResultStatus::ErrorNoPackedUpdate;
+    const auto nca_test = std::make_shared<FileSys::NCA>(read);
+
+    if (nca_test->GetStatus() != ResultStatus::ErrorMissingBKTRBaseRomFS)
+        return nca_test->GetStatus();
+
+    file = read;
+    return ResultStatus::Success;
 }
 
 ResultStatus AppLoader_NSP::ReadProgramId(u64& out_program_id) {

src/core/loader/nsp.h

@@ -37,7 +37,9 @@ public:
 
     ResultStatus Load(Kernel::Process& process) override;
 
-    ResultStatus ReadRomFS(FileSys::VirtualFile& dir) override;
+    ResultStatus ReadRomFS(FileSys::VirtualFile& file) override;
+    u64 ReadRomFSIVFCOffset() const override;
+    ResultStatus ReadUpdateRaw(FileSys::VirtualFile& file) override;
     ResultStatus ReadProgramId(u64& out_program_id) override;
     ResultStatus ReadIcon(std::vector<u8>& buffer) override;
     ResultStatus ReadTitle(std::string& title) override;
@@ -47,7 +49,7 @@ private:
     std::unique_ptr<AppLoader> secondary_loader;
 
     FileSys::VirtualFile icon_file;
-    std::shared_ptr<FileSys::NACP> nacp_file;
+    std::unique_ptr<FileSys::NACP> nacp_file;
     u64 title_id;
 };
 

src/core/loader/xci.cpp

@@ -9,7 +9,11 @@
 #include "core/file_sys/content_archive.h"
 #include "core/file_sys/control_metadata.h"
 #include "core/file_sys/patch_manager.h"
+#include "core/file_sys/registered_cache.h"
+#include "core/file_sys/romfs.h"
+#include "core/file_sys/submission_package.h"
 #include "core/hle/kernel/process.h"
+#include "core/hle/service/filesystem/filesystem.h"
 #include "core/loader/nca.h"
 #include "core/loader/xci.h"
 
@@ -26,7 +30,7 @@ AppLoader_XCI::AppLoader_XCI(FileSys::VirtualFile file)
         return;
 
     std::tie(nacp_file, icon_file) =
-        FileSys::PatchManager(xci->GetProgramTitleID()).ParseControlNCA(control_nca);
+        FileSys::PatchManager(xci->GetProgramTitleID()).ParseControlNCA(*control_nca);
 }
 
 AppLoader_XCI::~AppLoader_XCI() = default;
@@ -63,13 +67,41 @@ ResultStatus AppLoader_XCI::Load(Kernel::Process& process) {
     if (result != ResultStatus::Success)
         return result;
 
+    FileSys::VirtualFile update_raw;
+    if (ReadUpdateRaw(update_raw) == ResultStatus::Success && update_raw != nullptr)
+        Service::FileSystem::SetPackedUpdate(std::move(update_raw));
+
     is_loaded = true;
 
     return ResultStatus::Success;
 }
 
-ResultStatus AppLoader_XCI::ReadRomFS(FileSys::VirtualFile& dir) {
-    return nca_loader->ReadRomFS(dir);
+ResultStatus AppLoader_XCI::ReadRomFS(FileSys::VirtualFile& file) {
+    return nca_loader->ReadRomFS(file);
+}
+
+u64 AppLoader_XCI::ReadRomFSIVFCOffset() const {
+    return nca_loader->ReadRomFSIVFCOffset();
+}
+
+ResultStatus AppLoader_XCI::ReadUpdateRaw(FileSys::VirtualFile& file) {
+    u64 program_id{};
+    nca_loader->ReadProgramId(program_id);
+    if (program_id == 0)
+        return ResultStatus::ErrorXCIMissingProgramNCA;
+
+    const auto read = xci->GetSecurePartitionNSP()->GetNCAFile(
+        FileSys::GetUpdateTitleID(program_id), FileSys::ContentRecordType::Program);
+
+    if (read == nullptr)
+        return ResultStatus::ErrorNoPackedUpdate;
+    const auto nca_test = std::make_shared<FileSys::NCA>(read);
+
+    if (nca_test->GetStatus() != ResultStatus::ErrorMissingBKTRBaseRomFS)
+        return nca_test->GetStatus();
+
+    file = read;
+    return ResultStatus::Success;
 }
 
 ResultStatus AppLoader_XCI::ReadProgramId(u64& out_program_id) {

src/core/loader/xci.h

@@ -37,7 +37,9 @@ public:
 
     ResultStatus Load(Kernel::Process& process) override;
 
-    ResultStatus ReadRomFS(FileSys::VirtualFile& dir) override;
+    ResultStatus ReadRomFS(FileSys::VirtualFile& file) override;
+    u64 ReadRomFSIVFCOffset() const override;
+    ResultStatus ReadUpdateRaw(FileSys::VirtualFile& file) override;
     ResultStatus ReadProgramId(u64& out_program_id) override;
     ResultStatus ReadIcon(std::vector<u8>& buffer) override;
     ResultStatus ReadTitle(std::string& title) override;
@@ -47,7 +49,7 @@ private:
     std::unique_ptr<AppLoader_NCA> nca_loader;
 
     FileSys::VirtualFile icon_file;
-    std::shared_ptr<FileSys::NACP> nacp_file;
+    std::unique_ptr<FileSys::NACP> nacp_file;
 };
 
 } // namespace Loader

src/core/settings.h

@@ -155,6 +155,13 @@ struct Values {
     // Debugging
     bool use_gdbstub;
     u16 gdbstub_port;
+    std::string program_args;
+
+    // WebService
+    bool enable_telemetry;
+    std::string web_api_url;
+    std::string yuzu_username;
+    std::string yuzu_token;
 } extern values;
 
 void Apply();

src/core/telemetry_session.cpp

@@ -2,9 +2,15 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <array>
+
+#include <mbedtls/ctr_drbg.h>
+#include <mbedtls/entropy.h>
+
 #include "common/assert.h"
 #include "common/common_types.h"
 #include "common/file_util.h"
+#include "common/logging/log.h"
 
 #include "core/core.h"
 #include "core/file_sys/control_metadata.h"
@@ -13,10 +19,31 @@
 #include "core/settings.h"
 #include "core/telemetry_session.h"
 
+#ifdef ENABLE_WEB_SERVICE
+#include "web_service/telemetry_json.h"
+#include "web_service/verify_login.h"
+#endif
+
 namespace Core {
 
 static u64 GenerateTelemetryId() {
     u64 telemetry_id{};
+
+    mbedtls_entropy_context entropy;
+    mbedtls_entropy_init(&entropy);
+    mbedtls_ctr_drbg_context ctr_drbg;
+    constexpr std::array<char, 18> personalization{{"yuzu Telemetry ID"}};
+
+    mbedtls_ctr_drbg_init(&ctr_drbg);
+    ASSERT(mbedtls_ctr_drbg_seed(&ctr_drbg, mbedtls_entropy_func, &entropy,
+                                 reinterpret_cast<const unsigned char*>(personalization.data()),
+                                 personalization.size()) == 0);
+    ASSERT(mbedtls_ctr_drbg_random(&ctr_drbg, reinterpret_cast<unsigned char*>(&telemetry_id),
+                                   sizeof(u64)) == 0);
+
+    mbedtls_ctr_drbg_free(&ctr_drbg);
+    mbedtls_entropy_free(&entropy);
+
     return telemetry_id;
 }
 
@@ -25,14 +52,21 @@ u64 GetTelemetryId() {
     const std::string filename{FileUtil::GetUserPath(FileUtil::UserPath::ConfigDir) +
                                "telemetry_id"};
 
-    if (FileUtil::Exists(filename)) {
+    bool generate_new_id = !FileUtil::Exists(filename);
+    if (!generate_new_id) {
         FileUtil::IOFile file(filename, "rb");
         if (!file.IsOpen()) {
             LOG_ERROR(Core, "failed to open telemetry_id: {}", filename);
             return {};
         }
         file.ReadBytes(&telemetry_id, sizeof(u64));
-    } else {
+        if (telemetry_id == 0) {
+            LOG_ERROR(Frontend, "telemetry_id is 0. Generating a new one.", telemetry_id);
+            generate_new_id = true;
+        }
+    }
+
+    if (generate_new_id) {
         FileUtil::IOFile file(filename, "wb");
         if (!file.IsOpen()) {
             LOG_ERROR(Core, "failed to open telemetry_id: {}", filename);
@@ -59,23 +93,20 @@ u64 RegenerateTelemetryId() {
     return new_telemetry_id;
 }
 
-std::future<bool> VerifyLogin(std::string username, std::string token, std::function<void()> func) {
+bool VerifyLogin(const std::string& username, const std::string& token) {
 #ifdef ENABLE_WEB_SERVICE
-    return WebService::VerifyLogin(username, token, Settings::values.verify_endpoint_url, func);
+    return WebService::VerifyLogin(Settings::values.web_api_url, username, token);
 #else
-    return std::async(std::launch::async, [func{std::move(func)}]() {
-        func();
-        return false;
-    });
+    return false;
 #endif
 }
 
 TelemetrySession::TelemetrySession() {
 #ifdef ENABLE_WEB_SERVICE
     if (Settings::values.enable_telemetry) {
-        backend = std::make_unique<WebService::TelemetryJson>(
-            Settings::values.telemetry_endpoint_url, Settings::values.yuzu_username,
-            Settings::values.yuzu_token);
+        backend = std::make_unique<WebService::TelemetryJson>(Settings::values.web_api_url,
+                                                              Settings::values.yuzu_username,
+                                                              Settings::values.yuzu_token);
     } else {
         backend = std::make_unique<Telemetry::NullVisitor>();
     }
@@ -94,7 +125,8 @@ TelemetrySession::TelemetrySession() {
     u64 program_id{};
     const Loader::ResultStatus res{System::GetInstance().GetAppLoader().ReadProgramId(program_id)};
     if (res == Loader::ResultStatus::Success) {
-        AddField(Telemetry::FieldType::Session, "ProgramId", program_id);
+        const std::string formatted_program_id{fmt::format("{:016X}", program_id)};
+        AddField(Telemetry::FieldType::Session, "ProgramId", formatted_program_id);
 
         std::string name;
         System::GetInstance().GetAppLoader().ReadTitle(name);

src/core/telemetry_session.h

@@ -4,8 +4,8 @@
 
 #pragma once
 
-#include <future>
 #include <memory>
+#include <string>
 #include "common/telemetry.h"
 
 namespace Core {
@@ -52,9 +52,8 @@ u64 RegenerateTelemetryId();
  * Verifies the username and token.
  * @param username yuzu username to use for authentication.
  * @param token yuzu token to use for authentication.
- * @param func A function that gets exectued when the verification is finished
  * @returns Future with bool indicating whether the verification succeeded
  */
-std::future<bool> VerifyLogin(std::string username, std::string token, std::function<void()> func);
+bool VerifyLogin(const std::string& username, const std::string& token);
 
 } // namespace Core

src/tests/core/arm/arm_test_common.cpp

@@ -15,7 +15,8 @@ namespace ArmTests {
 TestEnvironment::TestEnvironment(bool mutable_memory_)
     : mutable_memory(mutable_memory_), test_memory(std::make_shared<TestMemory>(this)) {
 
-    Core::CurrentProcess() = Kernel::Process::Create(kernel, "");
+    auto process = Kernel::Process::Create(kernel, "");
+    kernel.MakeCurrentProcess(process.get());
     page_table = &Core::CurrentProcess()->VMManager().page_table;
 
     std::fill(page_table->pointers.begin(), page_table->pointers.end(), nullptr);

src/video_core/CMakeLists.txt

@@ -27,6 +27,8 @@ add_library(video_core STATIC
     renderer_base.h
     renderer_opengl/gl_buffer_cache.cpp
     renderer_opengl/gl_buffer_cache.h
+    renderer_opengl/gl_primitive_assembler.cpp
+    renderer_opengl/gl_primitive_assembler.h
     renderer_opengl/gl_rasterizer.cpp
     renderer_opengl/gl_rasterizer.h
     renderer_opengl/gl_rasterizer_cache.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,
@@ -44,27 +46,31 @@ void Fermi2D::HandleSurfaceCopy() {
     u32 src_bytes_per_pixel = RenderTargetBytesPerPixel(regs.src.format);
     u32 dst_bytes_per_pixel = RenderTargetBytesPerPixel(regs.dst.format);
 
-    if (regs.src.linear == regs.dst.linear) {
-        // If the input layout and the output layout are the same, just perform a raw copy.
-        ASSERT(regs.src.BlockHeight() == regs.dst.BlockHeight());
-        Memory::CopyBlock(dest_cpu, source_cpu,
-                          src_bytes_per_pixel * regs.dst.width * regs.dst.height);
-        return;
-    }
+    if (!rasterizer.AccelerateSurfaceCopy(regs.src, regs.dst)) {
+        // TODO(bunnei): The below implementation currently will not get hit, as
+        // AccelerateSurfaceCopy tries to always copy and will always return success. This should be
+        // changed once we properly support flushing.
 
-    u8* src_buffer = Memory::GetPointer(source_cpu);
-    u8* dst_buffer = Memory::GetPointer(dest_cpu);
-
-    if (!regs.src.linear && regs.dst.linear) {
-        // If the input is tiled and the output is linear, deswizzle the input and copy it over.
-        Texture::CopySwizzledData(regs.src.width, regs.src.height, src_bytes_per_pixel,
-                                  dst_bytes_per_pixel, src_buffer, dst_buffer, true,
-                                  regs.src.BlockHeight());
-    } else {
-        // If the input is linear and the output is tiled, swizzle the input and copy it over.
-        Texture::CopySwizzledData(regs.src.width, regs.src.height, src_bytes_per_pixel,
-                                  dst_bytes_per_pixel, dst_buffer, src_buffer, false,
-                                  regs.dst.BlockHeight());
+        if (regs.src.linear == regs.dst.linear) {
+            // If the input layout and the output layout are the same, just perform a raw copy.
+            ASSERT(regs.src.BlockHeight() == regs.dst.BlockHeight());
+            Memory::CopyBlock(dest_cpu, source_cpu,
+                              src_bytes_per_pixel * regs.dst.width * regs.dst.height);
+            return;
+        }
+        u8* src_buffer = Memory::GetPointer(source_cpu);
+        u8* dst_buffer = Memory::GetPointer(dest_cpu);
+        if (!regs.src.linear && regs.dst.linear) {
+            // If the input is tiled and the output is linear, deswizzle the input and copy it over.
+            Texture::CopySwizzledData(regs.src.width, regs.src.height, src_bytes_per_pixel,
+                                      dst_bytes_per_pixel, src_buffer, dst_buffer, true,
+                                      regs.src.BlockHeight());
+        } else {
+            // If the input is linear and the output is tiled, swizzle the input and copy it over.
+            Texture::CopySwizzledData(regs.src.width, regs.src.height, src_bytes_per_pixel,
+                                      dst_bytes_per_pixel, dst_buffer, src_buffer, false,
+                                      regs.dst.BlockHeight());
+        }
     }
 }
 

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.
@@ -32,9 +36,9 @@ public:
             RenderTargetFormat format;
             BitField<0, 1, u32> linear;
             union {
-                BitField<0, 4, u32> block_depth;
+                BitField<0, 4, u32> block_width;
                 BitField<4, 4, u32> block_height;
-                BitField<8, 4, u32> block_width;
+                BitField<8, 4, u32> block_depth;
             };
             u32 depth;
             u32 layer;
@@ -49,10 +53,20 @@ public:
                                              address_low);
             }
 
+            u32 BlockWidth() const {
+                // The block width is stored in log2 format.
+                return 1 << block_width;
+            }
+
             u32 BlockHeight() const {
                 // The block height is stored in log2 format.
                 return 1 << block_height;
             }
+
+            u32 BlockDepth() const {
+                // The block depth is stored in log2 format.
+                return 1 << block_depth;
+            }
         };
         static_assert(sizeof(Surface) == 0x28, "Surface has incorrect size");
 
@@ -94,6 +108,8 @@ 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/maxwell_3d.h

@@ -347,6 +347,16 @@ public:
             DecrWrap = 8,
         };
 
+        enum class MemoryLayout : u32 {
+            Linear = 0,
+            BlockLinear = 1,
+        };
+
+        enum class InvMemoryLayout : u32 {
+            BlockLinear = 0,
+            Linear = 1,
+        };
+
         struct Cull {
             enum class FrontFace : u32 {
                 ClockWise = 0x0900,
@@ -432,7 +442,12 @@ public:
             u32 width;
             u32 height;
             Tegra::RenderTargetFormat format;
-            u32 block_dimensions;
+            union {
+                BitField<0, 3, u32> block_width;
+                BitField<4, 3, u32> block_height;
+                BitField<8, 3, u32> block_depth;
+                BitField<12, 1, InvMemoryLayout> type;
+            } memory_layout;
             u32 array_mode;
             u32 layer_stride;
             u32 base_layer;
@@ -532,7 +547,21 @@ public:
                 INSERT_PADDING_WORDS(0x3);
                 s32 clear_stencil;
 
-                INSERT_PADDING_WORDS(0x6C);
+                INSERT_PADDING_WORDS(0x17);
+
+                struct {
+                    u32 enable;
+                    union {
+                        BitField<0, 16, u32> min_x;
+                        BitField<16, 16, u32> max_x;
+                    };
+                    union {
+                        BitField<0, 16, u32> min_y;
+                        BitField<16, 16, u32> max_y;
+                    };
+                } scissor_test;
+
+                INSERT_PADDING_WORDS(0x52);
 
                 s32 stencil_back_func_ref;
                 u32 stencil_back_mask;
@@ -548,7 +577,12 @@ public:
                     u32 address_high;
                     u32 address_low;
                     Tegra::DepthFormat format;
-                    u32 block_dimensions;
+                    union {
+                        BitField<0, 4, u32> block_width;
+                        BitField<4, 4, u32> block_height;
+                        BitField<8, 4, u32> block_depth;
+                        BitField<20, 1, InvMemoryLayout> type;
+                    } memory_layout;
                     u32 layer_stride;
 
                     GPUVAddr Address() const {
@@ -744,6 +778,12 @@ public:
                         return static_cast<GPUVAddr>((static_cast<GPUVAddr>(end_addr_high) << 32) |
                                                      end_addr_low);
                     }
+
+                    /// Adjust the index buffer offset so it points to the first desired index.
+                    GPUVAddr IndexStart() const {
+                        return StartAddress() + static_cast<size_t>(first) *
+                                                    static_cast<size_t>(FormatSizeInBytes());
+                    }
                 } index_array;
 
                 INSERT_PADDING_WORDS(0x7);
@@ -996,6 +1036,7 @@ ASSERT_REG_POSITION(vertex_buffer, 0x35D);
 ASSERT_REG_POSITION(clear_color[0], 0x360);
 ASSERT_REG_POSITION(clear_depth, 0x364);
 ASSERT_REG_POSITION(clear_stencil, 0x368);
+ASSERT_REG_POSITION(scissor_test, 0x380);
 ASSERT_REG_POSITION(stencil_back_func_ref, 0x3D5);
 ASSERT_REG_POSITION(stencil_back_mask, 0x3D6);
 ASSERT_REG_POSITION(stencil_back_func_mask, 0x3D7);

src/video_core/engines/shader_bytecode.h

@@ -214,6 +214,18 @@ enum class IMinMaxExchange : u64 {
     XHi = 3,
 };
 
+enum class VmadType : u64 {
+    Size16_Low = 0,
+    Size16_High = 1,
+    Size32 = 2,
+    Invalid = 3,
+};
+
+enum class VmadShr : u64 {
+    Shr7 = 1,
+    Shr15 = 2,
+};
+
 enum class XmadMode : u64 {
     None = 0,
     CLo = 1,
@@ -314,6 +326,15 @@ enum class TextureMiscMode : u64 {
     PTP,
 };
 
+enum class IsberdMode : u64 {
+    None = 0,
+    Patch = 1,
+    Prim = 2,
+    Attr = 3,
+};
+
+enum class IsberdShift : u64 { None = 0, U16 = 1, B32 = 2 };
+
 enum class IpaInterpMode : u64 {
     Linear = 0,
     Perspective = 1,
@@ -340,6 +361,87 @@ struct IpaMode {
     }
 };
 
+enum class SystemVariable : u64 {
+    LaneId = 0x00,
+    VirtCfg = 0x02,
+    VirtId = 0x03,
+    Pm0 = 0x04,
+    Pm1 = 0x05,
+    Pm2 = 0x06,
+    Pm3 = 0x07,
+    Pm4 = 0x08,
+    Pm5 = 0x09,
+    Pm6 = 0x0a,
+    Pm7 = 0x0b,
+    OrderingTicket = 0x0f,
+    PrimType = 0x10,
+    InvocationId = 0x11,
+    Ydirection = 0x12,
+    ThreadKill = 0x13,
+    ShaderType = 0x14,
+    DirectBeWriteAddressLow = 0x15,
+    DirectBeWriteAddressHigh = 0x16,
+    DirectBeWriteEnabled = 0x17,
+    MachineId0 = 0x18,
+    MachineId1 = 0x19,
+    MachineId2 = 0x1a,
+    MachineId3 = 0x1b,
+    Affinity = 0x1c,
+    InvocationInfo = 0x1d,
+    WscaleFactorXY = 0x1e,
+    WscaleFactorZ = 0x1f,
+    Tid = 0x20,
+    TidX = 0x21,
+    TidY = 0x22,
+    TidZ = 0x23,
+    CtaParam = 0x24,
+    CtaIdX = 0x25,
+    CtaIdY = 0x26,
+    CtaIdZ = 0x27,
+    NtId = 0x28,
+    CirQueueIncrMinusOne = 0x29,
+    Nlatc = 0x2a,
+    SmSpaVersion = 0x2c,
+    MultiPassShaderInfo = 0x2d,
+    LwinHi = 0x2e,
+    SwinHi = 0x2f,
+    SwinLo = 0x30,
+    SwinSz = 0x31,
+    SmemSz = 0x32,
+    SmemBanks = 0x33,
+    LwinLo = 0x34,
+    LwinSz = 0x35,
+    LmemLosz = 0x36,
+    LmemHioff = 0x37,
+    EqMask = 0x38,
+    LtMask = 0x39,
+    LeMask = 0x3a,
+    GtMask = 0x3b,
+    GeMask = 0x3c,
+    RegAlloc = 0x3d,
+    CtxAddr = 0x3e,      // .fmask = F_SM50
+    BarrierAlloc = 0x3e, // .fmask = F_SM60
+    GlobalErrorStatus = 0x40,
+    WarpErrorStatus = 0x42,
+    WarpErrorStatusClear = 0x43,
+    PmHi0 = 0x48,
+    PmHi1 = 0x49,
+    PmHi2 = 0x4a,
+    PmHi3 = 0x4b,
+    PmHi4 = 0x4c,
+    PmHi5 = 0x4d,
+    PmHi6 = 0x4e,
+    PmHi7 = 0x4f,
+    ClockLo = 0x50,
+    ClockHi = 0x51,
+    GlobalTimerLo = 0x52,
+    GlobalTimerHi = 0x53,
+    HwTaskId = 0x60,
+    CircularQueueEntryIndex = 0x61,
+    CircularQueueEntryAddressLow = 0x62,
+    CircularQueueEntryAddressHigh = 0x63,
+};
+
 union Instruction {
     Instruction& operator=(const Instruction& instr) {
         value = instr.value;
@@ -362,6 +464,7 @@ union Instruction {
     BitField<48, 16, u64> opcode;
 
     union {
+        BitField<20, 16, u64> imm20_16;
         BitField<20, 19, u64> imm20_19;
         BitField<20, 32, s64> imm20_32;
         BitField<45, 1, u64> negate_b;
@@ -403,6 +506,10 @@ union Instruction {
             }
         } lop3;
 
+        u16 GetImm20_16() const {
+            return static_cast<u16>(imm20_16);
+        }
+
         u32 GetImm20_19() const {
             u32 imm{static_cast<u32>(imm20_19)};
             imm <<= 12;
@@ -914,6 +1021,35 @@ union Instruction {
         }
     } bra;
 
+    union {
+        BitField<39, 1, u64> emit; // EmitVertex
+        BitField<40, 1, u64> cut;  // EndPrimitive
+    } out;
+
+    union {
+        BitField<31, 1, u64> skew;
+        BitField<32, 1, u64> o;
+        BitField<33, 2, IsberdMode> mode;
+        BitField<47, 2, IsberdShift> shift;
+    } isberd;
+
+    union {
+        BitField<48, 1, u64> signed_a;
+        BitField<38, 1, u64> is_byte_chunk_a;
+        BitField<36, 2, VmadType> type_a;
+        BitField<36, 2, u64> byte_height_a;
+
+        BitField<49, 1, u64> signed_b;
+        BitField<50, 1, u64> use_register_b;
+        BitField<30, 1, u64> is_byte_chunk_b;
+        BitField<28, 2, VmadType> type_b;
+        BitField<28, 2, u64> byte_height_b;
+
+        BitField<51, 2, VmadShr> shr;
+        BitField<55, 1, u64> saturate; // Saturates the result (a * b + c)
+        BitField<47, 1, u64> cc;
+    } vmad;
+
     union {
         BitField<20, 16, u64> imm20_16;
         BitField<36, 1, u64> product_shift_left;
@@ -936,6 +1072,10 @@ union Instruction {
         BitField<36, 5, u64> index;
     } cbuf36;
 
+    // Unsure about the size of this one.
+    // It's always used with a gpr0, so any size should be fine.
+    BitField<20, 8, SystemVariable> sys20;
+
     BitField<47, 1, u64> generates_cc;
     BitField<61, 1, u64> is_b_imm;
     BitField<60, 1, u64> is_b_gpr;
@@ -975,6 +1115,9 @@ public:
         TMML,   // Texture Mip Map Level
         EXIT,
         IPA,
+        OUT_R, // Emit vertex/primitive
+        ISBERD,
+        VMAD,
         FFMA_IMM, // Fused Multiply and Add
         FFMA_CR,
         FFMA_RC,
@@ -1034,6 +1177,7 @@ public:
         MOV_C,
         MOV_R,
         MOV_IMM,
+        MOV_SYS,
         MOV32_IMM,
         SHL_C,
         SHL_R,
@@ -1209,6 +1353,9 @@ private:
             INST("1101111101011---", Id::TMML, Type::Memory, "TMML"),
             INST("111000110000----", Id::EXIT, Type::Trivial, "EXIT"),
             INST("11100000--------", Id::IPA, Type::Trivial, "IPA"),
+            INST("1111101111100---", Id::OUT_R, Type::Trivial, "OUT_R"),
+            INST("1110111111010---", Id::ISBERD, Type::Trivial, "ISBERD"),
+            INST("01011111--------", Id::VMAD, Type::Trivial, "VMAD"),
             INST("0011001-1-------", Id::FFMA_IMM, Type::Ffma, "FFMA_IMM"),
             INST("010010011-------", Id::FFMA_CR, Type::Ffma, "FFMA_CR"),
             INST("010100011-------", Id::FFMA_RC, Type::Ffma, "FFMA_RC"),
@@ -1255,6 +1402,7 @@ private:
             INST("0100110010011---", Id::MOV_C, Type::Arithmetic, "MOV_C"),
             INST("0101110010011---", Id::MOV_R, Type::Arithmetic, "MOV_R"),
             INST("0011100-10011---", Id::MOV_IMM, Type::Arithmetic, "MOV_IMM"),
+            INST("1111000011001---", Id::MOV_SYS, Type::Trivial, "MOV_SYS"),
             INST("000000010000----", Id::MOV32_IMM, Type::ArithmeticImmediate, "MOV32_IMM"),
             INST("0100110001100---", Id::FMNMX_C, Type::Arithmetic, "FMNMX_C"),
             INST("0101110001100---", Id::FMNMX_R, Type::Arithmetic, "FMNMX_R"),

src/video_core/gpu.cpp

@@ -25,7 +25,7 @@ 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);

src/video_core/rasterizer_interface.h

@@ -5,6 +5,7 @@
 #pragma once
 
 #include "common/common_types.h"
+#include "video_core/engines/fermi_2d.h"
 #include "video_core/gpu.h"
 #include "video_core/memory_manager.h"
 
@@ -33,13 +34,9 @@ public:
     /// and invalidated
     virtual void FlushAndInvalidateRegion(VAddr addr, u64 size) = 0;
 
-    /// Attempt to use a faster method to perform a display transfer with is_texture_copy = 0
-    virtual bool AccelerateDisplayTransfer(const void* config) {
-        return false;
-    }
-
-    /// Attempt to use a faster method to perform a display transfer with is_texture_copy = 1
-    virtual bool AccelerateTextureCopy(const void* config) {
+    /// Attempt to use a faster method to perform a surface copy
+    virtual bool AccelerateSurfaceCopy(const Tegra::Engines::Fermi2D::Regs::Surface& src,
+                                       const Tegra::Engines::Fermi2D::Regs::Surface& dst) {
         return false;
     }
 

src/video_core/renderer_opengl/gl_buffer_cache.cpp

@@ -34,7 +34,7 @@ GLintptr OGLBufferCache::UploadMemory(Tegra::GPUVAddr gpu_addr, std::size_t size
     }
 
     AlignBuffer(alignment);
-    GLintptr uploaded_offset = buffer_offset;
+    const GLintptr uploaded_offset = buffer_offset;
 
     Memory::ReadBlock(*cpu_addr, buffer_ptr, size);
 
@@ -57,13 +57,23 @@ GLintptr OGLBufferCache::UploadHostMemory(const void* raw_pointer, std::size_t s
                                           std::size_t alignment) {
     AlignBuffer(alignment);
     std::memcpy(buffer_ptr, raw_pointer, size);
-    GLintptr uploaded_offset = buffer_offset;
+    const GLintptr uploaded_offset = buffer_offset;
 
     buffer_ptr += size;
     buffer_offset += size;
     return uploaded_offset;
 }
 
+std::tuple<u8*, GLintptr> OGLBufferCache::ReserveMemory(std::size_t size, std::size_t alignment) {
+    AlignBuffer(alignment);
+    u8* const uploaded_ptr = buffer_ptr;
+    const GLintptr uploaded_offset = buffer_offset;
+
+    buffer_ptr += size;
+    buffer_offset += size;
+    return std::make_tuple(uploaded_ptr, uploaded_offset);
+}
+
 void OGLBufferCache::Map(std::size_t max_size) {
     bool invalidate;
     std::tie(buffer_ptr, buffer_offset_base, invalidate) =
@@ -74,6 +84,7 @@ void OGLBufferCache::Map(std::size_t max_size) {
         InvalidateAll();
     }
 }
+
 void OGLBufferCache::Unmap() {
     stream_buffer.Unmap(buffer_offset - buffer_offset_base);
 }
@@ -84,7 +95,7 @@ GLuint OGLBufferCache::GetHandle() const {
 
 void OGLBufferCache::AlignBuffer(std::size_t alignment) {
     // Align the offset, not the mapped pointer
-    GLintptr offset_aligned =
+    const GLintptr offset_aligned =
         static_cast<GLintptr>(Common::AlignUp(static_cast<std::size_t>(buffer_offset), alignment));
     buffer_ptr += offset_aligned - buffer_offset;
     buffer_offset = offset_aligned;

src/video_core/renderer_opengl/gl_buffer_cache.h

@@ -6,6 +6,7 @@
 
 #include <cstddef>
 #include <memory>
+#include <tuple>
 
 #include "common/common_types.h"
 #include "video_core/rasterizer_cache.h"
@@ -33,11 +34,17 @@ class OGLBufferCache final : public RasterizerCache<std::shared_ptr<CachedBuffer
 public:
     explicit OGLBufferCache(std::size_t size);
 
+    /// Uploads data from a guest GPU address. Returns host's buffer offset where it's been
+    /// allocated.
     GLintptr UploadMemory(Tegra::GPUVAddr gpu_addr, std::size_t size, std::size_t alignment = 4,
                           bool cache = true);
 
+    /// Uploads from a host memory. Returns host's buffer offset where it's been allocated.
     GLintptr UploadHostMemory(const void* raw_pointer, std::size_t size, std::size_t alignment = 4);
 
+    /// Reserves memory to be used by host's CPU. Returns mapped address and offset.
+    std::tuple<u8*, GLintptr> ReserveMemory(std::size_t size, std::size_t alignment = 4);
+
     void Map(std::size_t max_size);
     void Unmap();
 

src/video_core/renderer_opengl/gl_primitive_assembler.cpp

@@ -0,0 +1,64 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <algorithm>
+#include <array>
+#include "common/assert.h"
+#include "common/common_types.h"
+#include "core/memory.h"
+#include "video_core/renderer_opengl/gl_buffer_cache.h"
+#include "video_core/renderer_opengl/gl_primitive_assembler.h"
+
+namespace OpenGL {
+
+constexpr u32 TRIANGLES_PER_QUAD = 6;
+constexpr std::array<u32, TRIANGLES_PER_QUAD> QUAD_MAP = {0, 1, 2, 0, 2, 3};
+
+PrimitiveAssembler::PrimitiveAssembler(OGLBufferCache& buffer_cache) : buffer_cache(buffer_cache) {}
+
+PrimitiveAssembler::~PrimitiveAssembler() = default;
+
+std::size_t PrimitiveAssembler::CalculateQuadSize(u32 count) const {
+    ASSERT_MSG(count % 4 == 0, "Quad count is expected to be a multiple of 4");
+    return (count / 4) * TRIANGLES_PER_QUAD * sizeof(GLuint);
+}
+
+GLintptr PrimitiveAssembler::MakeQuadArray(u32 first, u32 count) {
+    const std::size_t size{CalculateQuadSize(count)};
+    auto [dst_pointer, index_offset] = buffer_cache.ReserveMemory(size);
+
+    for (u32 primitive = 0; primitive < count / 4; ++primitive) {
+        for (u32 i = 0; i < TRIANGLES_PER_QUAD; ++i) {
+            const u32 index = first + primitive * 4 + QUAD_MAP[i];
+            std::memcpy(dst_pointer, &index, sizeof(index));
+            dst_pointer += sizeof(index);
+        }
+    }
+
+    return index_offset;
+}
+
+GLintptr PrimitiveAssembler::MakeQuadIndexed(Tegra::GPUVAddr gpu_addr, std::size_t index_size,
+                                             u32 count) {
+    const std::size_t map_size{CalculateQuadSize(count)};
+    auto [dst_pointer, index_offset] = buffer_cache.ReserveMemory(map_size);
+
+    auto& memory_manager = Core::System::GetInstance().GPU().MemoryManager();
+    const boost::optional<VAddr> cpu_addr{memory_manager.GpuToCpuAddress(gpu_addr)};
+    const u8* source{Memory::GetPointer(*cpu_addr)};
+
+    for (u32 primitive = 0; primitive < count / 4; ++primitive) {
+        for (std::size_t i = 0; i < TRIANGLES_PER_QUAD; ++i) {
+            const u32 index = primitive * 4 + QUAD_MAP[i];
+            const u8* src_offset = source + (index * index_size);
+
+            std::memcpy(dst_pointer, src_offset, index_size);
+            dst_pointer += index_size;
+        }
+    }
+
+    return index_offset;
+}
+
+} // namespace OpenGL

src/video_core/renderer_opengl/gl_primitive_assembler.h

@@ -0,0 +1,33 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <vector>
+#include <glad/glad.h>
+
+#include "common/common_types.h"
+#include "video_core/memory_manager.h"
+
+namespace OpenGL {
+
+class OGLBufferCache;
+
+class PrimitiveAssembler {
+public:
+    explicit PrimitiveAssembler(OGLBufferCache& buffer_cache);
+    ~PrimitiveAssembler();
+
+    /// Calculates the size required by MakeQuadArray and MakeQuadIndexed.
+    std::size_t CalculateQuadSize(u32 count) const;
+
+    GLintptr MakeQuadArray(u32 first, u32 count);
+
+    GLintptr MakeQuadIndexed(Tegra::GPUVAddr gpu_addr, std::size_t index_size, u32 count);
+
+private:
+    OGLBufferCache& buffer_cache;
+};
+
+} // namespace OpenGL

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -42,6 +42,41 @@ MICROPROFILE_DEFINE(OpenGL_Framebuffer, "OpenGL", "Framebuffer Setup", MP_RGB(12
 MICROPROFILE_DEFINE(OpenGL_Drawing, "OpenGL", "Drawing", MP_RGB(128, 128, 192));
 MICROPROFILE_DEFINE(OpenGL_Blits, "OpenGL", "Blits", MP_RGB(128, 128, 192));
 MICROPROFILE_DEFINE(OpenGL_CacheManagement, "OpenGL", "Cache Mgmt", MP_RGB(100, 255, 100));
+MICROPROFILE_DEFINE(OpenGL_PrimitiveAssembly, "OpenGL", "Prim Asmbl", MP_RGB(255, 100, 100));
+
+struct DrawParameters {
+    GLenum primitive_mode;
+    GLsizei count;
+    GLint current_instance;
+    bool use_indexed;
+
+    GLint vertex_first;
+
+    GLenum index_format;
+    GLint base_vertex;
+    GLintptr index_buffer_offset;
+
+    void DispatchDraw() const {
+        if (use_indexed) {
+            const auto index_buffer_ptr = reinterpret_cast<const void*>(index_buffer_offset);
+            if (current_instance > 0) {
+                glDrawElementsInstancedBaseVertexBaseInstance(primitive_mode, count, index_format,
+                                                              index_buffer_ptr, 1, base_vertex,
+                                                              current_instance);
+            } else {
+                glDrawElementsBaseVertex(primitive_mode, count, index_format, index_buffer_ptr,
+                                         base_vertex);
+            }
+        } else {
+            if (current_instance > 0) {
+                glDrawArraysInstancedBaseInstance(primitive_mode, vertex_first, count, 1,
+                                                  current_instance);
+            } else {
+                glDrawArrays(primitive_mode, vertex_first, count);
+            }
+        }
+    }
+};
 
 RasterizerOpenGL::RasterizerOpenGL(Core::Frontend::EmuWindow& window, ScreenInfo& info)
     : emu_window{window}, screen_info{info}, buffer_cache(STREAM_BUFFER_SIZE) {
@@ -172,7 +207,55 @@ void RasterizerOpenGL::SetupVertexArrays() {
     }
 }
 
-void RasterizerOpenGL::SetupShaders() {
+DrawParameters RasterizerOpenGL::SetupDraw() {
+    const auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
+    const auto& regs = gpu.regs;
+    const bool is_indexed = accelerate_draw == AccelDraw::Indexed;
+
+    DrawParameters params{};
+    params.current_instance = gpu.state.current_instance;
+
+    if (regs.draw.topology == Maxwell::PrimitiveTopology::Quads) {
+        MICROPROFILE_SCOPE(OpenGL_PrimitiveAssembly);
+
+        params.use_indexed = true;
+        params.primitive_mode = GL_TRIANGLES;
+
+        if (is_indexed) {
+            params.index_format = MaxwellToGL::IndexFormat(regs.index_array.format);
+            params.count = (regs.index_array.count / 4) * 6;
+            params.index_buffer_offset = primitive_assembler.MakeQuadIndexed(
+                regs.index_array.IndexStart(), regs.index_array.FormatSizeInBytes(),
+                regs.index_array.count);
+            params.base_vertex = static_cast<GLint>(regs.vb_element_base);
+        } else {
+            // MakeQuadArray always generates u32 indexes
+            params.index_format = GL_UNSIGNED_INT;
+            params.count = (regs.vertex_buffer.count / 4) * 6;
+            params.index_buffer_offset =
+                primitive_assembler.MakeQuadArray(regs.vertex_buffer.first, params.count);
+        }
+        return params;
+    }
+
+    params.use_indexed = is_indexed;
+    params.primitive_mode = MaxwellToGL::PrimitiveTopology(regs.draw.topology);
+
+    if (is_indexed) {
+        MICROPROFILE_SCOPE(OpenGL_Index);
+        params.index_format = MaxwellToGL::IndexFormat(regs.index_array.format);
+        params.count = regs.index_array.count;
+        params.index_buffer_offset =
+            buffer_cache.UploadMemory(regs.index_array.IndexStart(), CalculateIndexBufferSize());
+        params.base_vertex = static_cast<GLint>(regs.vb_element_base);
+    } else {
+        params.count = regs.vertex_buffer.count;
+        params.vertex_first = regs.vertex_buffer.first;
+    }
+    return params;
+}
+
+void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) {
     MICROPROFILE_SCOPE(OpenGL_Shader);
     const auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
 
@@ -187,6 +270,11 @@ void RasterizerOpenGL::SetupShaders() {
 
         // Skip stages that are not enabled
         if (!gpu.regs.IsShaderConfigEnabled(index)) {
+            switch (program) {
+            case Maxwell::ShaderProgram::Geometry:
+                shader_program_manager->UseTrivialGeometryShader();
+                break;
+            }
             continue;
         }
 
@@ -205,11 +293,18 @@ void RasterizerOpenGL::SetupShaders() {
         switch (program) {
         case Maxwell::ShaderProgram::VertexA:
         case Maxwell::ShaderProgram::VertexB: {
-            shader_program_manager->UseProgrammableVertexShader(shader->GetProgramHandle());
+            shader_program_manager->UseProgrammableVertexShader(
+                shader->GetProgramHandle(primitive_mode));
+            break;
+        }
+        case Maxwell::ShaderProgram::Geometry: {
+            shader_program_manager->UseProgrammableGeometryShader(
+                shader->GetProgramHandle(primitive_mode));
             break;
         }
         case Maxwell::ShaderProgram::Fragment: {
-            shader_program_manager->UseProgrammableFragmentShader(shader->GetProgramHandle());
+            shader_program_manager->UseProgrammableFragmentShader(
+                shader->GetProgramHandle(primitive_mode));
             break;
         }
         default:
@@ -219,12 +314,13 @@ void RasterizerOpenGL::SetupShaders() {
         }
 
         // Configure the const buffers for this shader stage.
-        current_constbuffer_bindpoint = SetupConstBuffers(static_cast<Maxwell::ShaderStage>(stage),
-                                                          shader, current_constbuffer_bindpoint);
+        current_constbuffer_bindpoint =
+            SetupConstBuffers(static_cast<Maxwell::ShaderStage>(stage), shader, primitive_mode,
+                              current_constbuffer_bindpoint);
 
         // Configure the textures for this shader stage.
         current_texture_bindpoint = SetupTextures(static_cast<Maxwell::ShaderStage>(stage), shader,
-                                                  current_texture_bindpoint);
+                                                  primitive_mode, current_texture_bindpoint);
 
         // When VertexA is enabled, we have dual vertex shaders
         if (program == Maxwell::ShaderProgram::VertexA) {
@@ -234,8 +330,6 @@ void RasterizerOpenGL::SetupShaders() {
     }
 
     state.Apply();
-
-    shader_program_manager->UseTrivialGeometryShader();
 }
 
 std::size_t RasterizerOpenGL::CalculateVertexArraysSize() const {
@@ -256,6 +350,13 @@ std::size_t RasterizerOpenGL::CalculateVertexArraysSize() const {
     return size;
 }
 
+std::size_t RasterizerOpenGL::CalculateIndexBufferSize() const {
+    const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+
+    return static_cast<std::size_t>(regs.index_array.count) *
+           static_cast<std::size_t>(regs.index_array.FormatSizeInBytes());
+}
+
 bool RasterizerOpenGL::AccelerateDrawBatch(bool is_indexed) {
     accelerate_draw = is_indexed ? AccelDraw::Indexed : AccelDraw::Arrays;
     DrawArrays();
@@ -451,6 +552,7 @@ void RasterizerOpenGL::DrawArrays() {
     SyncLogicOpState();
     SyncCullMode();
     SyncAlphaTest();
+    SyncScissorTest();
     SyncTransformFeedback();
     SyncPointState();
 
@@ -459,16 +561,23 @@ void RasterizerOpenGL::DrawArrays() {
 
     // Draw the vertex batch
     const bool is_indexed = accelerate_draw == AccelDraw::Indexed;
-    const u64 index_buffer_size{static_cast<u64>(regs.index_array.count) *
-                                static_cast<u64>(regs.index_array.FormatSizeInBytes())};
 
     state.draw.vertex_buffer = buffer_cache.GetHandle();
     state.Apply();
 
     std::size_t buffer_size = CalculateVertexArraysSize();
 
-    if (is_indexed) {
-        buffer_size = Common::AlignUp<std::size_t>(buffer_size, 4) + index_buffer_size;
+    // Add space for index buffer (keeping in mind non-core primitives)
+    switch (regs.draw.topology) {
+    case Maxwell::PrimitiveTopology::Quads:
+        buffer_size = Common::AlignUp<std::size_t>(buffer_size, 4) +
+                      primitive_assembler.CalculateQuadSize(regs.vertex_buffer.count);
+        break;
+    default:
+        if (is_indexed) {
+            buffer_size = Common::AlignUp<std::size_t>(buffer_size, 4) + CalculateIndexBufferSize();
+        }
+        break;
     }
 
     // Uniform space for the 5 shader stages
@@ -482,52 +591,16 @@ void RasterizerOpenGL::DrawArrays() {
     buffer_cache.Map(buffer_size);
 
     SetupVertexArrays();
-
-    // If indexed mode, copy the index buffer
-    GLintptr index_buffer_offset = 0;
-    if (is_indexed) {
-        MICROPROFILE_SCOPE(OpenGL_Index);
-
-        // 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();
+    DrawParameters params = SetupDraw();
+    SetupShaders(params.primitive_mode);
 
     buffer_cache.Unmap();
 
     shader_program_manager->ApplyTo(state);
     state.Apply();
 
-    const GLenum primitive_mode{MaxwellToGL::PrimitiveTopology(regs.draw.topology)};
-    if (is_indexed) {
-        const GLint base_vertex{static_cast<GLint>(regs.vb_element_base)};
-
-        if (gpu.state.current_instance > 0) {
-            glDrawElementsInstancedBaseVertexBaseInstance(
-                primitive_mode, regs.index_array.count,
-                MaxwellToGL::IndexFormat(regs.index_array.format),
-                reinterpret_cast<const void*>(index_buffer_offset), 1, base_vertex,
-                gpu.state.current_instance);
-        } else {
-            glDrawElementsBaseVertex(primitive_mode, regs.index_array.count,
-                                     MaxwellToGL::IndexFormat(regs.index_array.format),
-                                     reinterpret_cast<const void*>(index_buffer_offset),
-                                     base_vertex);
-        }
-    } else {
-        if (gpu.state.current_instance > 0) {
-            glDrawArraysInstancedBaseInstance(primitive_mode, regs.vertex_buffer.first,
-                                              regs.vertex_buffer.count, 1,
-                                              gpu.state.current_instance);
-        } else {
-            glDrawArrays(primitive_mode, regs.vertex_buffer.first, regs.vertex_buffer.count);
-        }
-    }
+    // Execute draw call
+    params.DispatchDraw();
 
     // Disable scissor test
     state.scissor.enabled = false;
@@ -556,14 +629,10 @@ void RasterizerOpenGL::FlushAndInvalidateRegion(VAddr addr, u64 size) {
     InvalidateRegion(addr, size);
 }
 
-bool RasterizerOpenGL::AccelerateDisplayTransfer(const void* config) {
+bool RasterizerOpenGL::AccelerateSurfaceCopy(const Tegra::Engines::Fermi2D::Regs::Surface& src,
+                                             const Tegra::Engines::Fermi2D::Regs::Surface& dst) {
     MICROPROFILE_SCOPE(OpenGL_Blits);
-    UNREACHABLE();
-    return true;
-}
-
-bool RasterizerOpenGL::AccelerateTextureCopy(const void* config) {
-    UNREACHABLE();
+    res_cache.FermiCopySurface(src, dst);
     return true;
 }
 
@@ -601,10 +670,13 @@ void RasterizerOpenGL::SamplerInfo::Create() {
     sampler.Create();
     mag_filter = min_filter = Tegra::Texture::TextureFilter::Linear;
     wrap_u = wrap_v = wrap_p = Tegra::Texture::WrapMode::Wrap;
+    uses_depth_compare = false;
+    depth_compare_func = Tegra::Texture::DepthCompareFunc::Never;
 
     // default is GL_LINEAR_MIPMAP_LINEAR
     glSamplerParameteri(sampler.handle, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
     // Other attributes have correct defaults
+    glSamplerParameteri(sampler.handle, GL_TEXTURE_COMPARE_FUNC, GL_NEVER);
 }
 
 void RasterizerOpenGL::SamplerInfo::SyncWithConfig(const Tegra::Texture::TSCEntry& config) {
@@ -632,6 +704,21 @@ void RasterizerOpenGL::SamplerInfo::SyncWithConfig(const Tegra::Texture::TSCEntr
         glSamplerParameteri(s, GL_TEXTURE_WRAP_R, MaxwellToGL::WrapMode(wrap_p));
     }
 
+    if (uses_depth_compare != (config.depth_compare_enabled == 1)) {
+        uses_depth_compare = (config.depth_compare_enabled == 1);
+        if (uses_depth_compare) {
+            glSamplerParameteri(s, GL_TEXTURE_COMPARE_MODE, GL_COMPARE_REF_TO_TEXTURE);
+        } else {
+            glSamplerParameteri(s, GL_TEXTURE_COMPARE_MODE, GL_NONE);
+        }
+    }
+
+    if (depth_compare_func != config.depth_compare_func) {
+        depth_compare_func = config.depth_compare_func;
+        glSamplerParameteri(s, GL_TEXTURE_COMPARE_FUNC,
+                            MaxwellToGL::DepthCompareFunc(depth_compare_func));
+    }
+
     if (wrap_u == Tegra::Texture::WrapMode::Border || wrap_v == Tegra::Texture::WrapMode::Border ||
         wrap_p == Tegra::Texture::WrapMode::Border) {
         const GLvec4 new_border_color = {{config.border_color_r, config.border_color_g,
@@ -644,7 +731,7 @@ void RasterizerOpenGL::SamplerInfo::SyncWithConfig(const Tegra::Texture::TSCEntr
 }
 
 u32 RasterizerOpenGL::SetupConstBuffers(Maxwell::ShaderStage stage, Shader& shader,
-                                        u32 current_bindpoint) {
+                                        GLenum primitive_mode, u32 current_bindpoint) {
     MICROPROFILE_SCOPE(OpenGL_UBO);
     const auto& gpu = Core::System::GetInstance().GPU();
     const auto& maxwell3d = gpu.Maxwell3D();
@@ -696,7 +783,7 @@ u32 RasterizerOpenGL::SetupConstBuffers(Maxwell::ShaderStage stage, Shader& shad
             buffer.address, size, static_cast<std::size_t>(uniform_buffer_alignment));
 
         // Now configure the bindpoint of the buffer inside the shader
-        glUniformBlockBinding(shader->GetProgramHandle(),
+        glUniformBlockBinding(shader->GetProgramHandle(primitive_mode),
                               shader->GetProgramResourceIndex(used_buffer),
                               current_bindpoint + bindpoint);
 
@@ -712,7 +799,8 @@ u32 RasterizerOpenGL::SetupConstBuffers(Maxwell::ShaderStage stage, Shader& shad
     return current_bindpoint + static_cast<u32>(entries.size());
 }
 
-u32 RasterizerOpenGL::SetupTextures(Maxwell::ShaderStage stage, Shader& shader, u32 current_unit) {
+u32 RasterizerOpenGL::SetupTextures(Maxwell::ShaderStage stage, Shader& shader,
+                                    GLenum primitive_mode, u32 current_unit) {
     MICROPROFILE_SCOPE(OpenGL_Texture);
     const auto& gpu = Core::System::GetInstance().GPU();
     const auto& maxwell3d = gpu.Maxwell3D();
@@ -727,8 +815,8 @@ u32 RasterizerOpenGL::SetupTextures(Maxwell::ShaderStage stage, Shader& shader,
 
         // Bind the uniform to the sampler.
 
-        glProgramUniform1i(shader->GetProgramHandle(), shader->GetUniformLocation(entry),
-                           current_bindpoint);
+        glProgramUniform1i(shader->GetProgramHandle(primitive_mode),
+                           shader->GetUniformLocation(entry), current_bindpoint);
 
         const auto texture = maxwell3d.GetStageTexture(entry.GetStage(), entry.GetOffset());
 
@@ -897,6 +985,22 @@ void RasterizerOpenGL::SyncAlphaTest() {
     }
 }
 
+void RasterizerOpenGL::SyncScissorTest() {
+    const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+
+    state.scissor.enabled = (regs.scissor_test.enable != 0);
+    // TODO(Blinkhawk): Figure if the hardware supports scissor testing per viewport and how it's
+    // implemented.
+    if (regs.scissor_test.enable != 0) {
+        const u32 width = regs.scissor_test.max_x - regs.scissor_test.min_x;
+        const u32 height = regs.scissor_test.max_y - regs.scissor_test.min_y;
+        state.scissor.x = regs.scissor_test.min_x;
+        state.scissor.y = regs.scissor_test.min_y;
+        state.scissor.width = width;
+        state.scissor.height = height;
+    }
+}
+
 void RasterizerOpenGL::SyncTransformFeedback() {
     const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
 

src/video_core/renderer_opengl/gl_rasterizer.h

@@ -23,6 +23,7 @@
 #include "video_core/rasterizer_cache.h"
 #include "video_core/rasterizer_interface.h"
 #include "video_core/renderer_opengl/gl_buffer_cache.h"
+#include "video_core/renderer_opengl/gl_primitive_assembler.h"
 #include "video_core/renderer_opengl/gl_rasterizer_cache.h"
 #include "video_core/renderer_opengl/gl_resource_manager.h"
 #include "video_core/renderer_opengl/gl_shader_cache.h"
@@ -38,6 +39,7 @@ class EmuWindow;
 namespace OpenGL {
 
 struct ScreenInfo;
+struct DrawParameters;
 
 class RasterizerOpenGL : public VideoCore::RasterizerInterface {
 public:
@@ -50,8 +52,8 @@ public:
     void FlushRegion(VAddr addr, u64 size) override;
     void InvalidateRegion(VAddr addr, u64 size) override;
     void FlushAndInvalidateRegion(VAddr addr, u64 size) override;
-    bool AccelerateDisplayTransfer(const void* config) override;
-    bool AccelerateTextureCopy(const void* config) override;
+    bool AccelerateSurfaceCopy(const Tegra::Engines::Fermi2D::Regs::Surface& src,
+                               const Tegra::Engines::Fermi2D::Regs::Surface& dst) override;
     bool AccelerateFill(const void* config) override;
     bool AccelerateDisplay(const Tegra::FramebufferConfig& config, VAddr framebuffer_addr,
                            u32 pixel_stride) override;
@@ -94,6 +96,8 @@ private:
         Tegra::Texture::WrapMode wrap_u;
         Tegra::Texture::WrapMode wrap_v;
         Tegra::Texture::WrapMode wrap_p;
+        bool uses_depth_compare;
+        Tegra::Texture::DepthCompareFunc depth_compare_func;
         GLvec4 border_color;
     };
 
@@ -116,7 +120,7 @@ private:
      * @returns The next available bindpoint for use in the next shader stage.
      */
     u32 SetupConstBuffers(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage, Shader& shader,
-                          u32 current_bindpoint);
+                          GLenum primitive_mode, u32 current_bindpoint);
 
     /*
      * Configures the current textures to use for the draw command.
@@ -126,7 +130,7 @@ private:
      * @returns The next available bindpoint for use in the next shader stage.
      */
     u32 SetupTextures(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage, Shader& shader,
-                      u32 current_unit);
+                      GLenum primitive_mode, u32 current_unit);
 
     /// Syncs the viewport to match the guest state
     void SyncViewport();
@@ -161,6 +165,9 @@ private:
     /// Syncs the alpha test state to match the guest state
     void SyncAlphaTest();
 
+    /// Syncs the scissor test state to match the guest state
+    void SyncScissorTest();
+
     /// Syncs the transform feedback state to match the guest state
     void SyncTransformFeedback();
 
@@ -192,13 +199,18 @@ private:
     static constexpr std::size_t STREAM_BUFFER_SIZE = 128 * 1024 * 1024;
     OGLBufferCache buffer_cache;
     OGLFramebuffer framebuffer;
+    PrimitiveAssembler primitive_assembler{buffer_cache};
     GLint uniform_buffer_alignment;
 
     std::size_t CalculateVertexArraysSize() const;
 
+    std::size_t CalculateIndexBufferSize() const;
+
     void SetupVertexArrays();
 
-    void SetupShaders();
+    DrawParameters SetupDraw();
+
+    void SetupShaders(GLenum primitive_mode);
 
     enum class AccelDraw { Disabled, Arrays, Indexed };
     AccelDraw accelerate_draw = AccelDraw::Disabled;

src/video_core/renderer_opengl/gl_rasterizer_cache.cpp

@@ -45,7 +45,9 @@ static VAddr TryGetCpuAddr(Tegra::GPUVAddr gpu_addr) {
     SurfaceParams params{};
     params.addr = TryGetCpuAddr(config.tic.Address());
     params.is_tiled = config.tic.IsTiled();
+    params.block_width = params.is_tiled ? config.tic.BlockWidth() : 0,
     params.block_height = params.is_tiled ? config.tic.BlockHeight() : 0,
+    params.block_depth = params.is_tiled ? config.tic.BlockDepth() : 0,
     params.pixel_format =
         PixelFormatFromTextureFormat(config.tic.format, config.tic.r_type.Value());
     params.component_type = ComponentTypeFromTexture(config.tic.r_type.Value());
@@ -97,8 +99,11 @@ static VAddr TryGetCpuAddr(Tegra::GPUVAddr gpu_addr) {
     const auto& config{Core::System::GetInstance().GPU().Maxwell3D().regs.rt[index]};
     SurfaceParams params{};
     params.addr = TryGetCpuAddr(config.Address());
-    params.is_tiled = true;
-    params.block_height = Tegra::Texture::TICEntry::DefaultBlockHeight;
+    params.is_tiled =
+        config.memory_layout.type == Tegra::Engines::Maxwell3D::Regs::InvMemoryLayout::BlockLinear;
+    params.block_width = 1 << config.memory_layout.block_width;
+    params.block_height = 1 << config.memory_layout.block_height;
+    params.block_depth = 1 << config.memory_layout.block_depth;
     params.pixel_format = PixelFormatFromRenderTargetFormat(config.format);
     params.component_type = ComponentTypeFromRenderTarget(config.format);
     params.type = GetFormatType(params.pixel_format);
@@ -120,13 +125,16 @@ static VAddr TryGetCpuAddr(Tegra::GPUVAddr gpu_addr) {
     return params;
 }
 
-/*static*/ SurfaceParams SurfaceParams::CreateForDepthBuffer(u32 zeta_width, u32 zeta_height,
-                                                             Tegra::GPUVAddr zeta_address,
-                                                             Tegra::DepthFormat format) {
+/*static*/ SurfaceParams SurfaceParams::CreateForDepthBuffer(
+    u32 zeta_width, u32 zeta_height, Tegra::GPUVAddr zeta_address, Tegra::DepthFormat format,
+    u32 block_width, u32 block_height, u32 block_depth,
+    Tegra::Engines::Maxwell3D::Regs::InvMemoryLayout type) {
     SurfaceParams params{};
     params.addr = TryGetCpuAddr(zeta_address);
-    params.is_tiled = true;
-    params.block_height = Tegra::Texture::TICEntry::DefaultBlockHeight;
+    params.is_tiled = type == Tegra::Engines::Maxwell3D::Regs::InvMemoryLayout::BlockLinear;
+    params.block_width = 1 << std::min(block_width, 5U);
+    params.block_height = 1 << std::min(block_height, 5U);
+    params.block_depth = 1 << std::min(block_depth, 5U);
     params.pixel_format = PixelFormatFromDepthFormat(format);
     params.component_type = ComponentTypeFromDepthFormat(format);
     params.type = GetFormatType(params.pixel_format);
@@ -143,6 +151,30 @@ static VAddr TryGetCpuAddr(Tegra::GPUVAddr gpu_addr) {
     return params;
 }
 
+/*static*/ SurfaceParams SurfaceParams::CreateForFermiCopySurface(
+    const Tegra::Engines::Fermi2D::Regs::Surface& config) {
+    SurfaceParams params{};
+    params.addr = TryGetCpuAddr(config.Address());
+    params.is_tiled = !config.linear;
+    params.block_width = params.is_tiled ? std::min(config.BlockWidth(), 32U) : 0,
+    params.block_height = params.is_tiled ? std::min(config.BlockHeight(), 32U) : 0,
+    params.block_depth = params.is_tiled ? std::min(config.BlockDepth(), 32U) : 0,
+    params.pixel_format = PixelFormatFromRenderTargetFormat(config.format);
+    params.component_type = ComponentTypeFromRenderTarget(config.format);
+    params.type = GetFormatType(params.pixel_format);
+    params.width = config.width;
+    params.height = config.height;
+    params.unaligned_height = config.height;
+    params.target = SurfaceTarget::Texture2D;
+    params.depth = 1;
+    params.size_in_bytes_total = params.SizeInBytesTotal();
+    params.size_in_bytes_2d = params.SizeInBytes2D();
+    params.max_mip_level = 0;
+    params.rt = {};
+
+    return params;
+}
+
 static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_format_tuples = {{
     {GL_RGBA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8_REV, ComponentType::UNorm, false}, // ABGR8U
     {GL_RGBA8, GL_RGBA, GL_BYTE, ComponentType::SNorm, false},                     // ABGR8S
@@ -559,6 +591,18 @@ static bool BlitSurface(const Surface& src_surface, const Surface& dst_surface,
     return true;
 }
 
+static void FastCopySurface(const Surface& src_surface, const Surface& dst_surface) {
+    const auto& src_params{src_surface->GetSurfaceParams()};
+    const auto& dst_params{dst_surface->GetSurfaceParams()};
+
+    const u32 width{std::min(src_params.width, dst_params.width)};
+    const u32 height{std::min(src_params.height, dst_params.height)};
+
+    glCopyImageSubData(src_surface->Texture().handle, SurfaceTargetToGL(src_params.target), 0, 0, 0,
+                       0, dst_surface->Texture().handle, SurfaceTargetToGL(dst_params.target), 0, 0,
+                       0, 0, width, height, 1);
+}
+
 static void CopySurface(const Surface& src_surface, const Surface& dst_surface,
                         GLuint copy_pbo_handle, GLenum src_attachment = 0,
                         GLenum dst_attachment = 0, std::size_t cubemap_face = 0) {
@@ -784,6 +828,11 @@ void CachedSurface::LoadGLBuffer() {
     if (params.is_tiled) {
         gl_buffer.resize(total_size);
 
+        ASSERT_MSG(params.block_width == 1, "Block width is defined as {} on texture type {}",
+                   params.block_width, static_cast<u32>(params.target));
+        ASSERT_MSG(params.block_depth == 1, "Block depth is defined as {} on texture type {}",
+                   params.block_depth, static_cast<u32>(params.target));
+
         // TODO(bunnei): This only unswizzles and copies a 2D texture - we do not yet know how to do
         // this for 3D textures, etc.
         switch (params.target) {
@@ -955,7 +1004,9 @@ Surface RasterizerCacheOpenGL::GetDepthBufferSurface(bool preserve_contents) {
     }
 
     SurfaceParams depth_params{SurfaceParams::CreateForDepthBuffer(
-        regs.zeta_width, regs.zeta_height, regs.zeta.Address(), regs.zeta.format)};
+        regs.zeta_width, regs.zeta_height, regs.zeta.Address(), regs.zeta.format,
+        regs.zeta.memory_layout.block_width, regs.zeta.memory_layout.block_height,
+        regs.zeta.memory_layout.block_depth, regs.zeta.memory_layout.type)};
 
     return GetSurface(depth_params, preserve_contents);
 }
@@ -1033,6 +1084,26 @@ Surface RasterizerCacheOpenGL::GetUncachedSurface(const SurfaceParams& params) {
     return surface;
 }
 
+void RasterizerCacheOpenGL::FermiCopySurface(
+    const Tegra::Engines::Fermi2D::Regs::Surface& src_config,
+    const Tegra::Engines::Fermi2D::Regs::Surface& dst_config) {
+
+    const auto& src_params = SurfaceParams::CreateForFermiCopySurface(src_config);
+    const auto& dst_params = SurfaceParams::CreateForFermiCopySurface(dst_config);
+
+    ASSERT(src_params.width == dst_params.width);
+    ASSERT(src_params.height == dst_params.height);
+    ASSERT(src_params.pixel_format == dst_params.pixel_format);
+    ASSERT(src_params.block_height == dst_params.block_height);
+    ASSERT(src_params.is_tiled == dst_params.is_tiled);
+    ASSERT(src_params.depth == dst_params.depth);
+    ASSERT(src_params.depth == 1); // Currently, FastCopySurface only works with 2D surfaces
+    ASSERT(src_params.target == dst_params.target);
+    ASSERT(src_params.rt.index == dst_params.rt.index);
+
+    FastCopySurface(GetSurface(src_params, true), GetSurface(dst_params, false));
+}
+
 Surface RasterizerCacheOpenGL::RecreateSurface(const Surface& old_surface,
                                                const SurfaceParams& new_params) {
     // Verify surface is compatible for blitting
@@ -1041,6 +1112,15 @@ Surface RasterizerCacheOpenGL::RecreateSurface(const Surface& old_surface,
     // Get a new surface with the new parameters, and blit the previous surface to it
     Surface new_surface{GetUncachedSurface(new_params)};
 
+    // For compatible surfaces, we can just do fast glCopyImageSubData based copy
+    if (old_params.target == new_params.target && old_params.type == new_params.type &&
+        old_params.depth == new_params.depth && old_params.depth == 1 &&
+        SurfaceParams::GetFormatBpp(old_params.pixel_format) ==
+            SurfaceParams::GetFormatBpp(new_params.pixel_format)) {
+        FastCopySurface(old_surface, new_surface);
+        return new_surface;
+    }
+
     // If the format is the same, just do a framebuffer blit. This is significantly faster than
     // using PBOs. The is also likely less accurate, as textures will be converted rather than
     // reinterpreted. When use_accurate_framebuffers setting is enabled, perform a more accurate

src/video_core/renderer_opengl/gl_rasterizer_cache.h

@@ -13,6 +13,7 @@
 #include "common/common_types.h"
 #include "common/hash.h"
 #include "common/math_util.h"
+#include "video_core/engines/fermi_2d.h"
 #include "video_core/engines/maxwell_3d.h"
 #include "video_core/rasterizer_cache.h"
 #include "video_core/renderer_opengl/gl_resource_manager.h"
@@ -715,9 +716,14 @@ struct SurfaceParams {
     static SurfaceParams CreateForFramebuffer(std::size_t index);
 
     /// Creates SurfaceParams for a depth buffer configuration
-    static SurfaceParams CreateForDepthBuffer(u32 zeta_width, u32 zeta_height,
-                                              Tegra::GPUVAddr zeta_address,
-                                              Tegra::DepthFormat format);
+    static SurfaceParams CreateForDepthBuffer(
+        u32 zeta_width, u32 zeta_height, Tegra::GPUVAddr zeta_address, Tegra::DepthFormat format,
+        u32 block_width, u32 block_height, u32 block_depth,
+        Tegra::Engines::Maxwell3D::Regs::InvMemoryLayout type);
+
+    /// Creates SurfaceParams for a Fermi2D surface copy
+    static SurfaceParams CreateForFermiCopySurface(
+        const Tegra::Engines::Fermi2D::Regs::Surface& config);
 
     /// Checks if surfaces are compatible for caching
     bool IsCompatibleSurface(const SurfaceParams& other) const {
@@ -728,7 +734,9 @@ struct SurfaceParams {
 
     VAddr addr;
     bool is_tiled;
+    u32 block_width;
     u32 block_height;
+    u32 block_depth;
     PixelFormat pixel_format;
     ComponentType component_type;
     SurfaceType type;
@@ -837,6 +845,10 @@ public:
     /// Tries to find a framebuffer using on the provided CPU address
     Surface TryFindFramebufferSurface(VAddr addr) const;
 
+    /// Copies the contents of one surface to another
+    void FermiCopySurface(const Tegra::Engines::Fermi2D::Regs::Surface& src_config,
+                          const Tegra::Engines::Fermi2D::Regs::Surface& dst_config);
+
 private:
     void LoadSurface(const Surface& surface);
     Surface GetSurface(const SurfaceParams& params, bool preserve_contents = true);

src/video_core/renderer_opengl/gl_shader_cache.cpp

@@ -68,6 +68,10 @@ CachedShader::CachedShader(VAddr addr, Maxwell::ShaderProgram program_type)
         program_result = GLShader::GenerateVertexShader(setup);
         gl_type = GL_VERTEX_SHADER;
         break;
+    case Maxwell::ShaderProgram::Geometry:
+        program_result = GLShader::GenerateGeometryShader(setup);
+        gl_type = GL_GEOMETRY_SHADER;
+        break;
     case Maxwell::ShaderProgram::Fragment:
         program_result = GLShader::GenerateFragmentShader(setup);
         gl_type = GL_FRAGMENT_SHADER;
@@ -80,11 +84,16 @@ CachedShader::CachedShader(VAddr addr, Maxwell::ShaderProgram program_type)
 
     entries = program_result.second;
 
-    OGLShader shader;
-    shader.Create(program_result.first.c_str(), gl_type);
-    program.Create(true, shader.handle);
-    SetShaderUniformBlockBindings(program.handle);
-    VideoCore::LabelGLObject(GL_PROGRAM, program.handle, addr);
+    if (program_type != Maxwell::ShaderProgram::Geometry) {
+        OGLShader shader;
+        shader.Create(program_result.first.c_str(), gl_type);
+        program.Create(true, shader.handle);
+        SetShaderUniformBlockBindings(program.handle);
+        VideoCore::LabelGLObject(GL_PROGRAM, program.handle, addr);
+    } else {
+        // Store shader's code to lazily build it on draw
+        geometry_programs.code = program_result.first;
+    }
 }
 
 GLuint CachedShader::GetProgramResourceIndex(const GLShader::ConstBufferEntry& buffer) {
@@ -110,6 +119,21 @@ GLint CachedShader::GetUniformLocation(const GLShader::SamplerEntry& sampler) {
     return search->second;
 }
 
+GLuint CachedShader::LazyGeometryProgram(OGLProgram& target_program,
+                                         const std::string& glsl_topology,
+                                         const std::string& debug_name) {
+    if (target_program.handle != 0) {
+        return target_program.handle;
+    }
+    const std::string source{geometry_programs.code + "layout (" + glsl_topology + ") in;\n"};
+    OGLShader shader;
+    shader.Create(source.c_str(), GL_GEOMETRY_SHADER);
+    target_program.Create(true, shader.handle);
+    SetShaderUniformBlockBindings(target_program.handle);
+    VideoCore::LabelGLObject(GL_PROGRAM, target_program.handle, addr, debug_name);
+    return target_program.handle;
+};
+
 Shader ShaderCacheOpenGL::GetStageProgram(Maxwell::ShaderProgram program) {
     const VAddr program_addr{GetShaderAddress(program)};