externals: Update catch to v2.2.3

Keeps the unit-testing library up to date.

.gitignore

@@ -11,6 +11,7 @@ src/common/scm_rev.cpp
 .idea/
 .vs/
 .vscode/
+CMakeLists.txt.user
 
 # *nix related
 # Common convention for backup or temporary files

externals/cmake-modules/WindowsCopyFiles.cmake

@@ -1,4 +1,4 @@
-# Copyright 2016 Citra Emulator Project
+# Copyright 2018 Yuzu Emulator Project
 # Licensed under GPLv2 or any later version
 # Refer to the license.txt file included.
 
@@ -22,7 +22,7 @@ function(windows_copy_files TARGET SOURCE_DIR DEST_DIR)
     # cmake adds an extra check for command success which doesn't work too well with robocopy
     # so trick it into thinking the command was successful with the || cmd /c "exit /b 0"
     add_custom_command(TARGET ${TARGET} POST_BUILD
-        COMMAND if not exist ${DEST_DIR} mkdir ${DEST_DIR} 2> nul
+        COMMAND ${CMAKE_COMMAND} -E make_directory ${DEST_DIR}
         COMMAND robocopy ${SOURCE_DIR} ${DEST_DIR} ${ARGN} /NJH /NJS /NDL /NFL /NC /NS /NP || cmd /c "exit /b 0"
     )
-endfunction()
+endfunction()

src/common/assert.h

@@ -52,5 +52,5 @@ __declspec(noinline, noreturn)
 #define DEBUG_ASSERT_MSG(_a_, _desc_, ...)
 #endif
 
-#define UNIMPLEMENTED() DEBUG_ASSERT_MSG(false, "Unimplemented code!")
+#define UNIMPLEMENTED() LOG_CRITICAL(Debug, "Unimplemented code!")
 #define UNIMPLEMENTED_MSG(...) ASSERT_MSG(false, __VA_ARGS__)

src/common/common_funcs.h

@@ -4,7 +4,7 @@
 
 #pragma once
 
-#if !defined(ARCHITECTURE_x86_64) && !defined(_M_ARM)
+#if !defined(ARCHITECTURE_x86_64) && !defined(ARCHITECTURE_ARM)
 #include <cstdlib> // for exit
 #endif
 #include "common/common_types.h"
@@ -30,7 +30,7 @@
 
 #ifdef ARCHITECTURE_x86_64
 #define Crash() __asm__ __volatile__("int $3")
-#elif defined(_M_ARM)
+#elif defined(ARCHITECTURE_ARM)
 #define Crash() __asm__ __volatile__("trap")
 #else
 #define Crash() exit(1)

src/common/file_util.cpp

@@ -826,6 +826,20 @@ std::string GetParentPath(const std::string& path) {
     return out.erase(name_index);
 }
 
+std::string GetPathWithoutTop(std::string path) {
+    if (path.empty())
+        return "";
+    while (path[0] == '\\' || path[0] == '/') {
+        path = path.substr(1);
+        if (path.empty())
+            return "";
+    }
+    const auto name_bck_index = path.find_first_of('\\');
+    const auto name_fwd_index = path.find_first_of('/');
+    return path.substr(std::min<size_t>(name_bck_index, name_fwd_index) + 1);
+    return path.substr(std::min<size_t>(name_bck_index, name_fwd_index) + 1);
+}
+
 std::string GetFilename(std::string path) {
     std::replace(path.begin(), path.end(), '\\', '/');
     auto name_index = path.find_last_of('/');

src/common/file_util.h

@@ -154,9 +154,12 @@ void SplitFilename83(const std::string& filename, std::array<char, 9>& short_nam
 // i.e. "C:\Users\Yuzu\Documents\save.bin" becomes {"C:", "Users", "Yuzu", "Documents", "save.bin" }
 std::vector<std::string> SplitPathComponents(const std::string& filename);
 
-// Gets all of the text prior to the last '/' or '\' in the path.
+// Gets all of the text up to the last '/' or '\' in the path.
 std::string GetParentPath(const std::string& path);
 
+// Gets all of the text after the first '/' or '\' in the path.
+std::string GetPathWithoutTop(std::string path);
+
 // Gets the filename of the path
 std::string GetFilename(std::string path);
 

src/common/logging/backend.cpp

@@ -5,6 +5,7 @@
 #include <algorithm>
 #include <array>
 #include <chrono>
+#include <climits>
 #include <condition_variable>
 #include <memory>
 #include <thread>
@@ -83,8 +84,10 @@ private:
                 }
             };
             while (true) {
-                std::unique_lock<std::mutex> lock(message_mutex);
-                message_cv.wait(lock, [&] { return !running || message_queue.Pop(entry); });
+                {
+                    std::unique_lock<std::mutex> lock(message_mutex);
+                    message_cv.wait(lock, [&] { return !running || message_queue.Pop(entry); });
+                }
                 if (!running) {
                     break;
                 }
@@ -92,7 +95,7 @@ private:
             }
             // Drain the logging queue. Only writes out up to MAX_LOGS_TO_WRITE to prevent a case
             // where a system is repeatedly spamming logs even on close.
-            constexpr int MAX_LOGS_TO_WRITE = 100;
+            const int MAX_LOGS_TO_WRITE = filter.IsDebug() ? INT_MAX : 100;
             int logs_written = 0;
             while (logs_written++ < MAX_LOGS_TO_WRITE && message_queue.Pop(entry)) {
                 write_logs(entry);
@@ -282,4 +285,4 @@ void FmtLogMessageImpl(Class log_class, Level log_level, const char* filename,
 
     Impl::Instance().PushEntry(std::move(entry));
 }
-} // namespace Log
+} // namespace Log

src/common/logging/filter.cpp

@@ -94,4 +94,11 @@ bool Filter::ParseFilterRule(const std::string::const_iterator begin,
 bool Filter::CheckMessage(Class log_class, Level level) const {
     return static_cast<u8>(level) >= static_cast<u8>(class_levels[static_cast<size_t>(log_class)]);
 }
+
+bool Filter::IsDebug() const {
+    return std::any_of(class_levels.begin(), class_levels.end(), [](const Level& l) {
+        return static_cast<u8>(l) <= static_cast<u8>(Level::Debug);
+    });
+}
+
 } // namespace Log

src/common/logging/filter.h

@@ -47,6 +47,9 @@ public:
     /// Matches class/level combination against the filter, returning true if it passed.
     bool CheckMessage(Class log_class, Level level) const;
 
+    /// Returns true if any logging classes are set to debug
+    bool IsDebug() const;
+
 private:
     std::array<Level, (size_t)Class::Count> class_levels;
 };

src/common/logging/log.h

@@ -103,7 +103,7 @@ template <typename... Args>
 void FmtLogMessage(Class log_class, Level log_level, const char* filename, unsigned int line_num,
                    const char* function, const char* format, const Args&... args) {
     FmtLogMessageImpl(log_class, log_level, filename, line_num, function, format,
-                      fmt::make_args(args...));
+                      fmt::make_format_args(args...));
 }
 
 } // namespace Log

src/common/memory_util.cpp

@@ -16,7 +16,7 @@
 #include <sys/mman.h>
 #endif
 
-#if !defined(_WIN32) && defined(ARCHITECTURE_X64) && !defined(MAP_32BIT)
+#if !defined(_WIN32) && defined(ARCHITECTURE_x86_64) && !defined(MAP_32BIT)
 #include <unistd.h>
 #define PAGE_MASK (getpagesize() - 1)
 #define round_page(x) ((((unsigned long)(x)) + PAGE_MASK) & ~(PAGE_MASK))
@@ -30,7 +30,7 @@ void* AllocateExecutableMemory(size_t size, bool low) {
     void* ptr = VirtualAlloc(nullptr, size, MEM_COMMIT, PAGE_EXECUTE_READWRITE);
 #else
     static char* map_hint = nullptr;
-#if defined(ARCHITECTURE_X64) && !defined(MAP_32BIT)
+#if defined(ARCHITECTURE_x86_64) && !defined(MAP_32BIT)
     // This OS has no flag to enforce allocation below the 4 GB boundary,
     // but if we hint that we want a low address it is very likely we will
     // get one.
@@ -42,7 +42,7 @@ void* AllocateExecutableMemory(size_t size, bool low) {
 #endif
     void* ptr = mmap(map_hint, size, PROT_READ | PROT_WRITE | PROT_EXEC,
                      MAP_ANON | MAP_PRIVATE
-#if defined(ARCHITECTURE_X64) && defined(MAP_32BIT)
+#if defined(ARCHITECTURE_x86_64) && defined(MAP_32BIT)
                          | (low ? MAP_32BIT : 0)
 #endif
                          ,
@@ -57,7 +57,7 @@ void* AllocateExecutableMemory(size_t size, bool low) {
 #endif
         LOG_ERROR(Common_Memory, "Failed to allocate executable memory");
     }
-#if !defined(_WIN32) && defined(ARCHITECTURE_X64) && !defined(MAP_32BIT)
+#if !defined(_WIN32) && defined(ARCHITECTURE_x86_64) && !defined(MAP_32BIT)
     else {
         if (low) {
             map_hint += size;

src/common/string_util.cpp

@@ -134,7 +134,7 @@ bool SplitPath(const std::string& full_path, std::string* _pPath, std::string* _
     size_t dir_end = full_path.find_last_of("/"
 // windows needs the : included for something like just "C:" to be considered a directory
 #ifdef _WIN32
-                                            ":"
+                                            "\\:"
 #endif
     );
     if (std::string::npos == dir_end)

src/common/swap.h

@@ -69,7 +69,7 @@ inline u32 swap32(u32 _data) {
 inline u64 swap64(u64 _data) {
     return _byteswap_uint64(_data);
 }
-#elif _M_ARM
+#elif ARCHITECTURE_ARM
 inline u16 swap16(u16 _data) {
     u32 data = _data;
     __asm__("rev16 %0, %1\n" : "=l"(data) : "l"(data));

src/common/telemetry.h

@@ -52,27 +52,14 @@ public:
 template <typename T>
 class Field : public FieldInterface {
 public:
-    Field(FieldType type, std::string name, const T& value)
-        : name(std::move(name)), type(type), value(value) {}
-
-    Field(FieldType type, std::string name, T&& value)
+    Field(FieldType type, std::string name, T value)
         : name(std::move(name)), type(type), value(std::move(value)) {}
 
-    Field(const Field& other) : Field(other.type, other.name, other.value) {}
+    Field(const Field&) = default;
+    Field& operator=(const Field&) = default;
 
-    Field& operator=(const Field& other) {
-        type = other.type;
-        name = other.name;
-        value = other.value;
-        return *this;
-    }
-
-    Field& operator=(Field&& other) {
-        type = other.type;
-        name = std::move(other.name);
-        value = std::move(other.value);
-        return *this;
-    }
+    Field(Field&&) = default;
+    Field& operator=(Field&& other) = default;
 
     void Accept(VisitorInterface& visitor) const override;
 
@@ -94,11 +81,11 @@ public:
         return value;
     }
 
-    inline bool operator==(const Field<T>& other) {
+    bool operator==(const Field& other) const {
         return (type == other.type) && (name == other.name) && (value == other.value);
     }
 
-    inline bool operator!=(const Field<T>& other) {
+    bool operator!=(const Field& other) const {
         return !(*this == other);
     }
 

src/core/CMakeLists.txt

@@ -12,15 +12,17 @@ add_library(core STATIC
     file_sys/content_archive.h
     file_sys/directory.h
     file_sys/errors.h
-    file_sys/filesystem.cpp
-    file_sys/filesystem.h
+    file_sys/mode.h
     file_sys/partition_filesystem.cpp
     file_sys/partition_filesystem.h
-    file_sys/path_parser.cpp
-    file_sys/path_parser.h
     file_sys/program_metadata.cpp
     file_sys/program_metadata.h
-    file_sys/storage.h
+    file_sys/romfs_factory.cpp
+    file_sys/romfs_factory.h
+    file_sys/savedata_factory.cpp
+    file_sys/savedata_factory.h
+    file_sys/sdmc_factory.cpp
+    file_sys/sdmc_factory.h
     file_sys/vfs.cpp
     file_sys/vfs.h
     file_sys/vfs_offset.cpp

src/core/arm/arm_interface.h

@@ -116,6 +116,8 @@ public:
      */
     virtual void LoadContext(const ThreadContext& ctx) = 0;
 
+    virtual void ClearExclusiveState() = 0;
+
     /// Prepare core for thread reschedule (if needed to correctly handle state)
     virtual void PrepareReschedule() = 0;
 };

src/core/arm/dynarmic/arm_dynarmic.cpp

@@ -226,6 +226,10 @@ void ARM_Dynarmic::ClearInstructionCache() {
     jit->ClearCache();
 }
 
+void ARM_Dynarmic::ClearExclusiveState() {
+    jit->ClearExclusiveState();
+}
+
 void ARM_Dynarmic::PageTableChanged() {
     jit = MakeJit(cb);
     current_page_table = Memory::GetCurrentPageTable();

src/core/arm/dynarmic/arm_dynarmic.h

@@ -39,6 +39,7 @@ public:
     void LoadContext(const ThreadContext& ctx) override;
 
     void PrepareReschedule() override;
+    void ClearExclusiveState() override;
 
     void ClearInstructionCache() override;
     void PageTableChanged() override;

src/core/arm/unicorn/arm_unicorn.cpp

@@ -193,11 +193,11 @@ void ARM_Unicorn::ExecuteInstructions(int num_instructions) {
         }
         Kernel::Thread* thread = Kernel::GetCurrentThread();
         SaveContext(thread->context);
-        if (last_bkpt_hit) {
+        if (last_bkpt_hit || (num_instructions == 1)) {
             last_bkpt_hit = false;
             GDBStub::Break();
+            GDBStub::SendTrap(thread, 5);
         }
-        GDBStub::SendTrap(thread, 5);
     }
 }
 
@@ -263,6 +263,8 @@ void ARM_Unicorn::PrepareReschedule() {
     CHECKED(uc_emu_stop(uc));
 }
 
+void ARM_Unicorn::ClearExclusiveState() {}
+
 void ARM_Unicorn::ClearInstructionCache() {}
 
 void ARM_Unicorn::RecordBreak(GDBStub::BreakpointAddress bkpt) {

src/core/arm/unicorn/arm_unicorn.h

@@ -31,6 +31,7 @@ public:
     void SaveContext(ThreadContext& ctx) override;
     void LoadContext(const ThreadContext& ctx) override;
     void PrepareReschedule() override;
+    void ClearExclusiveState() override;
     void ExecuteInstructions(int num_instructions);
     void Run() override;
     void Step() override;

src/core/core.cpp

@@ -26,11 +26,13 @@ namespace Core {
 
 /*static*/ System System::s_instance;
 
+System::System() = default;
+
 System::~System() = default;
 
 /// Runs a CPU core while the system is powered on
 static void RunCpuCore(std::shared_ptr<Cpu> cpu_state) {
-    while (Core::System().GetInstance().IsPoweredOn()) {
+    while (Core::System::GetInstance().IsPoweredOn()) {
         cpu_state->RunLoop(true);
     }
 }

src/core/core.h

@@ -168,6 +168,8 @@ public:
     }
 
 private:
+    System();
+
     /// Returns the currently running CPU core
     Cpu& CurrentCpuCore();
 

src/core/file_sys/content_archive.cpp

@@ -17,7 +17,10 @@ constexpr u64 SECTION_HEADER_OFFSET = 0x400;
 
 constexpr u32 IVFC_MAX_LEVEL = 6;
 
-enum class NCASectionFilesystemType : u8 { PFS0 = 0x2, ROMFS = 0x3 };
+enum class NCASectionFilesystemType : u8 {
+    PFS0 = 0x2,
+    ROMFS = 0x3,
+};
 
 struct NCASectionHeaderBlock {
     INSERT_PADDING_BYTES(3);

src/core/file_sys/content_archive.h

@@ -11,7 +11,13 @@
 
 namespace FileSys {
 
-enum class NCAContentType : u8 { Program = 0, Meta = 1, Control = 2, Manual = 3, Data = 4 };
+enum class NCAContentType : u8 {
+    Program = 0,
+    Meta = 1,
+    Control = 2,
+    Manual = 3,
+    Data = 4,
+};
 
 struct NCASectionTableEntry {
     u32_le media_offset;
@@ -42,12 +48,12 @@ struct NCAHeader {
 };
 static_assert(sizeof(NCAHeader) == 0x400, "NCAHeader has incorrect size.");
 
-static bool IsDirectoryExeFS(std::shared_ptr<FileSys::VfsDirectory> pfs) {
+inline bool IsDirectoryExeFS(std::shared_ptr<FileSys::VfsDirectory> pfs) {
     // According to switchbrew, an exefs must only contain these two files:
     return pfs->GetFile("main") != nullptr && pfs->GetFile("main.npdm") != nullptr;
 }
 
-static bool IsValidNCA(const NCAHeader& header) {
+inline bool IsValidNCA(const NCAHeader& header) {
     return header.magic == Common::MakeMagic('N', 'C', 'A', '2') ||
            header.magic == Common::MakeMagic('N', 'C', 'A', '3');
 }

src/core/file_sys/directory.h

@@ -8,13 +8,17 @@
 #include <cstddef>
 #include "common/common_funcs.h"
 #include "common/common_types.h"
-#include "core/file_sys/filesystem.h"
 
 ////////////////////////////////////////////////////////////////////////////////////////////////////
 // FileSys namespace
 
 namespace FileSys {
 
+enum EntryType : u8 {
+    Directory = 0,
+    File = 1,
+};
+
 // Structure of a directory entry, from
 // http://switchbrew.org/index.php?title=Filesystem_services#DirectoryEntry
 const size_t FILENAME_LENGTH = 0x300;

src/core/file_sys/errors.h

@@ -11,6 +11,9 @@ namespace FileSys {
 namespace ErrCodes {
 enum {
     NotFound = 1,
+    TitleNotFound = 1002,
+    SdCardNotFound = 2001,
+    RomFSNotFound = 2520,
 };
 }
 

src/core/file_sys/filesystem.cpp

@@ -1,122 +0,0 @@
-// Copyright 2018 yuzu emulator team
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include <cstddef>
-#include <iomanip>
-#include <sstream>
-#include "common/logging/log.h"
-#include "common/string_util.h"
-#include "core/file_sys/filesystem.h"
-#include "core/memory.h"
-
-namespace FileSys {
-
-Path::Path(LowPathType type, u32 size, u32 pointer) : type(type) {
-    switch (type) {
-    case Binary: {
-        binary.resize(size);
-        Memory::ReadBlock(pointer, binary.data(), binary.size());
-        break;
-    }
-
-    case Char: {
-        string.resize(size - 1); // Data is always null-terminated.
-        Memory::ReadBlock(pointer, &string[0], string.size());
-        break;
-    }
-
-    case Wchar: {
-        u16str.resize(size / 2 - 1); // Data is always null-terminated.
-        Memory::ReadBlock(pointer, &u16str[0], u16str.size() * sizeof(char16_t));
-        break;
-    }
-
-    default:
-        break;
-    }
-}
-
-std::string Path::DebugStr() const {
-    switch (GetType()) {
-    case Invalid:
-    default:
-        return "[Invalid]";
-    case Empty:
-        return "[Empty]";
-    case Binary: {
-        std::stringstream res;
-        res << "[Binary: ";
-        for (unsigned byte : binary)
-            res << std::hex << std::setw(2) << std::setfill('0') << byte;
-        res << ']';
-        return res.str();
-    }
-    case Char:
-        return "[Char: " + AsString() + ']';
-    case Wchar:
-        return "[Wchar: " + AsString() + ']';
-    }
-}
-
-std::string Path::AsString() const {
-    switch (GetType()) {
-    case Char:
-        return string;
-    case Wchar:
-        return Common::UTF16ToUTF8(u16str);
-    case Empty:
-        return {};
-    case Invalid:
-    case Binary:
-    default:
-        // TODO(yuriks): Add assert
-        LOG_ERROR(Service_FS, "LowPathType cannot be converted to string!");
-        return {};
-    }
-}
-
-std::u16string Path::AsU16Str() const {
-    switch (GetType()) {
-    case Char:
-        return Common::UTF8ToUTF16(string);
-    case Wchar:
-        return u16str;
-    case Empty:
-        return {};
-    case Invalid:
-    case Binary:
-        // TODO(yuriks): Add assert
-        LOG_ERROR(Service_FS, "LowPathType cannot be converted to u16string!");
-        return {};
-    }
-
-    UNREACHABLE();
-}
-
-std::vector<u8> Path::AsBinary() const {
-    switch (GetType()) {
-    case Binary:
-        return binary;
-    case Char:
-        return std::vector<u8>(string.begin(), string.end());
-    case Wchar: {
-        // use two u8 for each character of u16str
-        std::vector<u8> to_return(u16str.size() * 2);
-        for (size_t i = 0; i < u16str.size(); ++i) {
-            u16 tmp_char = u16str.at(i);
-            to_return[i * 2] = (tmp_char & 0xFF00) >> 8;
-            to_return[i * 2 + 1] = (tmp_char & 0x00FF);
-        }
-        return to_return;
-    }
-    case Empty:
-        return {};
-    case Invalid:
-    default:
-        // TODO(yuriks): Add assert
-        LOG_ERROR(Service_FS, "LowPathType cannot be converted to binary!");
-        return {};
-    }
-}
-} // namespace FileSys

src/core/file_sys/filesystem.h

@@ -1,69 +0,0 @@
-// Copyright 2018 yuzu emulator team
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <memory>
-#include <string>
-#include <utility>
-#include <vector>
-#include "common/bit_field.h"
-#include "common/common_types.h"
-#include "common/swap.h"
-#include "core/hle/result.h"
-
-namespace FileSys {
-
-class StorageBackend;
-class DirectoryBackend;
-
-// Path string type
-enum LowPathType : u32 {
-    Invalid = 0,
-    Empty = 1,
-    Binary = 2,
-    Char = 3,
-    Wchar = 4,
-};
-
-enum EntryType : u8 {
-    Directory = 0,
-    File = 1,
-};
-
-enum class Mode : u32 {
-    Read = 1,
-    Write = 2,
-    Append = 4,
-};
-
-class Path {
-public:
-    Path() : type(Invalid) {}
-    Path(const char* path) : type(Char), string(path) {}
-    Path(std::vector<u8> binary_data) : type(Binary), binary(std::move(binary_data)) {}
-    Path(LowPathType type, u32 size, u32 pointer);
-
-    LowPathType GetType() const {
-        return type;
-    }
-
-    /**
-     * Gets the string representation of the path for debugging
-     * @return String representation of the path for debugging
-     */
-    std::string DebugStr() const;
-
-    std::string AsString() const;
-    std::u16string AsU16Str() const;
-    std::vector<u8> AsBinary() const;
-
-private:
-    LowPathType type;
-    std::vector<u8> binary;
-    std::string string;
-    std::u16string u16str;
-};
-
-} // namespace FileSys

src/core/file_sys/mode.h

@@ -0,0 +1,17 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "common/common_types.h"
+
+namespace FileSys {
+
+enum class Mode : u32 {
+    Read = 1,
+    Write = 2,
+    Append = 4,
+};
+
+} // namespace FileSys

src/core/file_sys/path_parser.cpp

@@ -1,98 +0,0 @@
-// Copyright 2016 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include <algorithm>
-#include <set>
-#include "common/file_util.h"
-#include "common/string_util.h"
-#include "core/file_sys/path_parser.h"
-
-namespace FileSys {
-
-PathParser::PathParser(const Path& path) {
-    if (path.GetType() != LowPathType::Char && path.GetType() != LowPathType::Wchar) {
-        is_valid = false;
-        return;
-    }
-
-    auto path_string = path.AsString();
-    if (path_string.size() == 0 || path_string[0] != '/') {
-        is_valid = false;
-        return;
-    }
-
-    // Filter out invalid characters for the host system.
-    // Although some of these characters are valid on 3DS, they are unlikely to be used by games.
-    if (std::find_if(path_string.begin(), path_string.end(), [](char c) {
-            static const std::set<char> invalid_chars{'<', '>', '\\', '|', ':', '\"', '*', '?'};
-            return invalid_chars.find(c) != invalid_chars.end();
-        }) != path_string.end()) {
-        is_valid = false;
-        return;
-    }
-
-    Common::SplitString(path_string, '/', path_sequence);
-
-    auto begin = path_sequence.begin();
-    auto end = path_sequence.end();
-    end = std::remove_if(begin, end, [](std::string& str) { return str == "" || str == "."; });
-    path_sequence = std::vector<std::string>(begin, end);
-
-    // checks if the path is out of bounds.
-    int level = 0;
-    for (auto& node : path_sequence) {
-        if (node == "..") {
-            --level;
-            if (level < 0) {
-                is_valid = false;
-                return;
-            }
-        } else {
-            ++level;
-        }
-    }
-
-    is_valid = true;
-    is_root = level == 0;
-}
-
-PathParser::HostStatus PathParser::GetHostStatus(const std::string& mount_point) const {
-    auto path = mount_point;
-    if (!FileUtil::IsDirectory(path))
-        return InvalidMountPoint;
-    if (path_sequence.empty()) {
-        return DirectoryFound;
-    }
-
-    for (auto iter = path_sequence.begin(); iter != path_sequence.end() - 1; iter++) {
-        if (path.back() != '/')
-            path += '/';
-        path += *iter;
-
-        if (!FileUtil::Exists(path))
-            return PathNotFound;
-        if (FileUtil::IsDirectory(path))
-            continue;
-        return FileInPath;
-    }
-
-    path += "/" + path_sequence.back();
-    if (!FileUtil::Exists(path))
-        return NotFound;
-    if (FileUtil::IsDirectory(path))
-        return DirectoryFound;
-    return FileFound;
-}
-
-std::string PathParser::BuildHostPath(const std::string& mount_point) const {
-    std::string path = mount_point;
-    for (auto& node : path_sequence) {
-        if (path.back() != '/')
-            path += '/';
-        path += node;
-    }
-    return path;
-}
-
-} // namespace FileSys

src/core/file_sys/path_parser.h

@@ -1,61 +0,0 @@
-// Copyright 2016 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <string>
-#include <vector>
-#include "core/file_sys/filesystem.h"
-
-namespace FileSys {
-
-/**
- * A helper class parsing and verifying a string-type Path.
- * Every archives with a sub file system should use this class to parse the path argument and check
- * the status of the file / directory in question on the host file system.
- */
-class PathParser {
-public:
-    explicit PathParser(const Path& path);
-
-    /**
-     * Checks if the Path is valid.
-     * This function should be called once a PathParser is constructed.
-     * A Path is valid if:
-     *  - it is a string path (with type LowPathType::Char or LowPathType::Wchar),
-     *  - it starts with "/" (this seems a hard requirement in real 3DS),
-     *  - it doesn't contain invalid characters, and
-     *  - it doesn't go out of the root directory using "..".
-     */
-    bool IsValid() const {
-        return is_valid;
-    }
-
-    /// Checks if the Path represents the root directory.
-    bool IsRootDirectory() const {
-        return is_root;
-    }
-
-    enum HostStatus {
-        InvalidMountPoint,
-        PathNotFound,   // "/a/b/c" when "a" doesn't exist
-        FileInPath,     // "/a/b/c" when "a" is a file
-        FileFound,      // "/a/b/c" when "c" is a file
-        DirectoryFound, // "/a/b/c" when "c" is a directory
-        NotFound        // "/a/b/c" when "a/b/" exists but "c" doesn't exist
-    };
-
-    /// Checks the status of the specified file / directory by the Path on the host file system.
-    HostStatus GetHostStatus(const std::string& mount_point) const;
-
-    /// Builds a full path on the host file system.
-    std::string BuildHostPath(const std::string& mount_point) const;
-
-private:
-    std::vector<std::string> path_sequence;
-    bool is_valid{};
-    bool is_root{};
-};
-
-} // namespace FileSys

src/core/file_sys/romfs_factory.cpp

@@ -0,0 +1,25 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <algorithm>
+#include <memory>
+#include "common/common_types.h"
+#include "common/logging/log.h"
+#include "core/file_sys/romfs_factory.h"
+
+namespace FileSys {
+
+RomFSFactory::RomFSFactory(Loader::AppLoader& app_loader) {
+    // Load the RomFS from the app
+    if (Loader::ResultStatus::Success != app_loader.ReadRomFS(file)) {
+        LOG_ERROR(Service_FS, "Unable to read RomFS!");
+    }
+}
+
+ResultVal<VirtualFile> RomFSFactory::Open(u64 title_id) {
+    // TODO(DarkLordZach): Use title id.
+    return MakeResult<VirtualFile>(file);
+}
+
+} // namespace FileSys

src/core/file_sys/romfs_factory.h

@@ -0,0 +1,25 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <memory>
+#include "common/common_types.h"
+#include "core/hle/result.h"
+#include "core/loader/loader.h"
+
+namespace FileSys {
+
+/// File system interface to the RomFS archive
+class RomFSFactory {
+public:
+    explicit RomFSFactory(Loader::AppLoader& app_loader);
+
+    ResultVal<VirtualFile> Open(u64 title_id);
+
+private:
+    VirtualFile file;
+};
+
+} // namespace FileSys

src/core/file_sys/savedata_factory.cpp

@@ -0,0 +1,106 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <memory>
+#include "common/common_types.h"
+#include "common/logging/log.h"
+#include "core/core.h"
+#include "core/file_sys/savedata_factory.h"
+#include "core/hle/kernel/process.h"
+
+namespace FileSys {
+
+std::string SaveDataDescriptor::DebugInfo() {
+    return fmt::format("[type={:02X}, title_id={:016X}, user_id={:016X}{:016X}, save_id={:016X}]",
+                       static_cast<u8>(type), title_id, user_id[1], user_id[0], save_id);
+}
+
+SaveDataFactory::SaveDataFactory(VirtualDir save_directory) : dir(std::move(save_directory)) {}
+
+ResultVal<VirtualDir> SaveDataFactory::Open(SaveDataSpaceId space, SaveDataDescriptor meta) {
+    if (meta.type == SaveDataType::SystemSaveData || meta.type == SaveDataType::SaveData) {
+        if (meta.zero_1 != 0) {
+            LOG_WARNING(Service_FS,
+                        "Possibly incorrect SaveDataDescriptor, type is "
+                        "SystemSaveData||SaveData but offset 0x28 is non-zero ({:016X}).",
+                        meta.zero_1);
+        }
+        if (meta.zero_2 != 0) {
+            LOG_WARNING(Service_FS,
+                        "Possibly incorrect SaveDataDescriptor, type is "
+                        "SystemSaveData||SaveData but offset 0x30 is non-zero ({:016X}).",
+                        meta.zero_2);
+        }
+        if (meta.zero_3 != 0) {
+            LOG_WARNING(Service_FS,
+                        "Possibly incorrect SaveDataDescriptor, type is "
+                        "SystemSaveData||SaveData but offset 0x38 is non-zero ({:016X}).",
+                        meta.zero_3);
+        }
+    }
+
+    if (meta.type == SaveDataType::SystemSaveData && meta.title_id != 0) {
+        LOG_WARNING(Service_FS,
+                    "Possibly incorrect SaveDataDescriptor, type is SystemSaveData but title_id is "
+                    "non-zero ({:016X}).",
+                    meta.title_id);
+    }
+
+    std::string save_directory =
+        GetFullPath(space, meta.type, meta.title_id, meta.user_id, meta.save_id);
+
+    // TODO(DarkLordZach): Try to not create when opening, there are dedicated create save methods.
+    // But, user_ids don't match so this works for now.
+
+    auto out = dir->GetDirectoryRelative(save_directory);
+
+    if (out == nullptr) {
+        // TODO(bunnei): This is a work-around to always create a save data directory if it does not
+        // already exist. This is a hack, as we do not understand yet how this works on hardware.
+        // Without a save data directory, many games will assert on boot. This should not have any
+        // bad side-effects.
+        out = dir->CreateDirectoryRelative(save_directory);
+    }
+
+    // Return an error if the save data doesn't actually exist.
+    if (out == nullptr) {
+        // TODO(Subv): Find out correct error code.
+        return ResultCode(-1);
+    }
+
+    return MakeResult<VirtualDir>(std::move(out));
+}
+
+std::string SaveDataFactory::GetFullPath(SaveDataSpaceId space, SaveDataType type, u64 title_id,
+                                         u128 user_id, u64 save_id) const {
+    // According to switchbrew, if a save is of type SaveData and the title id field is 0, it should
+    // be interpreted as the title id of the current process.
+    if (type == SaveDataType::SaveData && title_id == 0)
+        title_id = Core::CurrentProcess()->program_id;
+
+    std::string out;
+
+    switch (space) {
+    case SaveDataSpaceId::NandSystem:
+        out = "/system/save/";
+        break;
+    case SaveDataSpaceId::NandUser:
+        out = "/user/save/";
+        break;
+    default:
+        ASSERT_MSG(false, "Unrecognized SaveDataSpaceId: {:02X}", static_cast<u8>(space));
+    }
+
+    switch (type) {
+    case SaveDataType::SystemSaveData:
+        return fmt::format("{}{:016X}/{:016X}{:016X}", out, save_id, user_id[1], user_id[0]);
+    case SaveDataType::SaveData:
+        return fmt::format("{}{:016X}/{:016X}{:016X}/{:016X}", out, 0, user_id[1], user_id[0],
+                           title_id);
+    default:
+        ASSERT_MSG(false, "Unrecognized SaveDataType: {:02X}", static_cast<u8>(type));
+    }
+}
+
+} // namespace FileSys

src/core/file_sys/savedata_factory.h

@@ -0,0 +1,58 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <memory>
+#include <string>
+#include "common/common_types.h"
+#include "core/hle/result.h"
+
+namespace FileSys {
+
+enum class SaveDataSpaceId : u8 {
+    NandSystem = 0,
+    NandUser = 1,
+    SdCard = 2,
+    TemporaryStorage = 3,
+};
+
+enum class SaveDataType : u8 {
+    SystemSaveData = 0,
+    SaveData = 1,
+    BcatDeliveryCacheStorage = 2,
+    DeviceSaveData = 3,
+    TemporaryStorage = 4,
+    CacheStorage = 5,
+};
+
+struct SaveDataDescriptor {
+    u64_le title_id;
+    u128 user_id;
+    u64_le save_id;
+    SaveDataType type;
+    INSERT_PADDING_BYTES(7);
+    u64_le zero_1;
+    u64_le zero_2;
+    u64_le zero_3;
+
+    std::string DebugInfo();
+};
+static_assert(sizeof(SaveDataDescriptor) == 0x40, "SaveDataDescriptor has incorrect size.");
+
+/// File system interface to the SaveData archive
+class SaveDataFactory {
+public:
+    explicit SaveDataFactory(VirtualDir dir);
+
+    ResultVal<VirtualDir> Open(SaveDataSpaceId space, SaveDataDescriptor meta);
+
+private:
+    VirtualDir dir;
+
+    std::string GetFullPath(SaveDataSpaceId space, SaveDataType type, u64 title_id, u128 user_id,
+                            u64 save_id) const;
+};
+
+} // namespace FileSys

src/core/file_sys/sdmc_factory.cpp

@@ -0,0 +1,17 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <memory>
+#include "core/core.h"
+#include "core/file_sys/sdmc_factory.h"
+
+namespace FileSys {
+
+SDMCFactory::SDMCFactory(VirtualDir dir) : dir(std::move(dir)) {}
+
+ResultVal<VirtualDir> SDMCFactory::Open() {
+    return MakeResult<VirtualDir>(dir);
+}
+
+} // namespace FileSys

src/core/file_sys/sdmc_factory.h

@@ -0,0 +1,22 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "core/hle/result.h"
+
+namespace FileSys {
+
+/// File system interface to the SDCard archive
+class SDMCFactory {
+public:
+    explicit SDMCFactory(VirtualDir dir);
+
+    ResultVal<VirtualDir> Open();
+
+private:
+    VirtualDir dir;
+};
+
+} // namespace FileSys

src/core/file_sys/storage.h

@@ -1,63 +0,0 @@
-// Copyright 2018 yuzu emulator team
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <cstddef>
-#include "common/common_types.h"
-#include "core/hle/result.h"
-
-namespace FileSys {
-
-class StorageBackend : NonCopyable {
-public:
-    StorageBackend() {}
-    virtual ~StorageBackend() {}
-
-    /**
-     * Read data from the file
-     * @param offset Offset in bytes to start reading data from
-     * @param length Length in bytes of data to read from file
-     * @param buffer Buffer to read data into
-     * @return Number of bytes read, or error code
-     */
-    virtual ResultVal<size_t> Read(u64 offset, size_t length, u8* buffer) const = 0;
-
-    /**
-     * Write data to the file
-     * @param offset Offset in bytes to start writing data to
-     * @param length Length in bytes of data to write to file
-     * @param flush The flush parameters (0 == do not flush)
-     * @param buffer Buffer to read data from
-     * @return Number of bytes written, or error code
-     */
-    virtual ResultVal<size_t> Write(u64 offset, size_t length, bool flush,
-                                    const u8* buffer) const = 0;
-
-    /**
-     * Flushes the file
-     */
-    virtual void Flush() const = 0;
-
-    /**
-     * Set the size of the file in bytes
-     * @param size New size of the file
-     * @return true if successful
-     */
-    virtual bool SetSize(u64 size) const = 0;
-
-    /**
-     * Get the size of the file in bytes
-     * @return Size of the file in bytes
-     */
-    virtual u64 GetSize() const = 0;
-
-    /**
-     * Close the file
-     * @return true if the file closed correctly
-     */
-    virtual bool Close() const = 0;
-};
-
-} // namespace FileSys

src/core/file_sys/vfs.cpp

@@ -5,6 +5,7 @@
 #include <algorithm>
 #include <numeric>
 #include "common/file_util.h"
+#include "common/logging/backend.h"
 #include "core/file_sys/vfs.h"
 
 namespace FileSys {
@@ -54,10 +55,10 @@ std::shared_ptr<VfsFile> VfsDirectory::GetFileRelative(const std::string& path)
     if (vec.size() == 1)
         return GetFile(vec[0]);
     auto dir = GetSubdirectory(vec[0]);
-    for (size_t i = 1; i < vec.size() - 1; ++i) {
+    for (size_t component = 1; component < vec.size() - 1; ++component) {
         if (dir == nullptr)
             return nullptr;
-        dir = dir->GetSubdirectory(vec[i]);
+        dir = dir->GetSubdirectory(vec[component]);
     }
     if (dir == nullptr)
         return nullptr;
@@ -76,11 +77,14 @@ std::shared_ptr<VfsDirectory> VfsDirectory::GetDirectoryRelative(const std::stri
     vec.erase(std::remove_if(vec.begin(), vec.end(), [](const auto& str) { return str.empty(); }),
               vec.end());
     if (vec.empty())
-        // return std::shared_ptr<VfsDirectory>(this);
+        // TODO(DarkLordZach): Return this directory if path is '/' or similar. Can't currently
+        // because of const-ness
         return nullptr;
     auto dir = GetSubdirectory(vec[0]);
-    for (size_t i = 1; i < vec.size(); ++i) {
-        dir = dir->GetSubdirectory(vec[i]);
+    for (size_t component = 1; component < vec.size(); ++component) {
+        if (dir == nullptr)
+            return nullptr;
+        dir = dir->GetSubdirectory(vec[component]);
     }
     return dir;
 }
@@ -124,6 +128,53 @@ size_t VfsDirectory::GetSize() const {
     return file_total + subdir_total;
 }
 
+std::shared_ptr<VfsFile> VfsDirectory::CreateFileRelative(const std::string& path) {
+    auto vec = FileUtil::SplitPathComponents(path);
+    vec.erase(std::remove_if(vec.begin(), vec.end(), [](const auto& str) { return str.empty(); }),
+              vec.end());
+    if (vec.empty())
+        return nullptr;
+    if (vec.size() == 1)
+        return CreateFile(vec[0]);
+    auto dir = GetSubdirectory(vec[0]);
+    if (dir == nullptr) {
+        dir = CreateSubdirectory(vec[0]);
+        if (dir == nullptr)
+            return nullptr;
+    }
+
+    return dir->CreateFileRelative(FileUtil::GetPathWithoutTop(path));
+}
+
+std::shared_ptr<VfsFile> VfsDirectory::CreateFileAbsolute(const std::string& path) {
+    if (IsRoot())
+        return CreateFileRelative(path);
+    return GetParentDirectory()->CreateFileAbsolute(path);
+}
+
+std::shared_ptr<VfsDirectory> VfsDirectory::CreateDirectoryRelative(const std::string& path) {
+    auto vec = FileUtil::SplitPathComponents(path);
+    vec.erase(std::remove_if(vec.begin(), vec.end(), [](const auto& str) { return str.empty(); }),
+              vec.end());
+    if (vec.empty())
+        return nullptr;
+    if (vec.size() == 1)
+        return CreateSubdirectory(vec[0]);
+    auto dir = GetSubdirectory(vec[0]);
+    if (dir == nullptr) {
+        dir = CreateSubdirectory(vec[0]);
+        if (dir == nullptr)
+            return nullptr;
+    }
+    return dir->CreateDirectoryRelative(FileUtil::GetPathWithoutTop(path));
+}
+
+std::shared_ptr<VfsDirectory> VfsDirectory::CreateDirectoryAbsolute(const std::string& path) {
+    if (IsRoot())
+        return CreateDirectoryRelative(path);
+    return GetParentDirectory()->CreateDirectoryAbsolute(path);
+}
+
 bool VfsDirectory::DeleteSubdirectoryRecursive(const std::string& name) {
     auto dir = GetSubdirectory(name);
     if (dir == nullptr)

src/core/file_sys/vfs.h

@@ -173,6 +173,23 @@ struct VfsDirectory : NonCopyable {
     // operation failed.
     virtual std::shared_ptr<VfsFile> CreateFile(const std::string& name) = 0;
 
+    // Creates a new file at the path relative to this directory. Also creates directories if
+    // they do not exist and is supported by this implementation. Returns nullptr on any failure.
+    virtual std::shared_ptr<VfsFile> CreateFileRelative(const std::string& path);
+
+    // Creates a new file at the path relative to root of this directory. Also creates directories
+    // if they do not exist and is supported by this implementation. Returns nullptr on any failure.
+    virtual std::shared_ptr<VfsFile> CreateFileAbsolute(const std::string& path);
+
+    // Creates a new directory at the path relative to this directory. Also creates directories if
+    // they do not exist and is supported by this implementation. Returns nullptr on any failure.
+    virtual std::shared_ptr<VfsDirectory> CreateDirectoryRelative(const std::string& path);
+
+    // Creates a new directory at the path relative to root of this directory. Also creates
+    // directories if they do not exist and is supported by this implementation. Returns nullptr on
+    // any failure.
+    virtual std::shared_ptr<VfsDirectory> CreateDirectoryAbsolute(const std::string& path);
+
     // Deletes the subdirectory with name and returns true on success.
     virtual bool DeleteSubdirectory(const std::string& name) = 0;
     // Deletes all subdirectories and files of subdirectory with name recirsively and then deletes

src/core/file_sys/vfs_real.cpp

@@ -76,6 +76,10 @@ bool RealVfsFile::Rename(const std::string& name) {
     return out;
 }
 
+bool RealVfsFile::Close() {
+    return backing.Close();
+}
+
 RealVfsDirectory::RealVfsDirectory(const std::string& path_, Mode perms_)
     : path(FileUtil::RemoveTrailingSlash(path_)), parent_path(FileUtil::GetParentPath(path)),
       path_components(FileUtil::SplitPathComponents(path)),
@@ -84,20 +88,19 @@ RealVfsDirectory::RealVfsDirectory(const std::string& path_, Mode perms_)
     if (!FileUtil::Exists(path) && (perms == Mode::Write || perms == Mode::Append))
         FileUtil::CreateDir(path);
     unsigned size;
-    if (perms != Mode::Append) {
-        FileUtil::ForeachDirectoryEntry(
-            &size, path,
-            [this](unsigned* entries_out, const std::string& directory,
-                   const std::string& filename) {
-                std::string full_path = directory + DIR_SEP + filename;
-                if (FileUtil::IsDirectory(full_path))
-                    subdirectories.emplace_back(
-                        std::make_shared<RealVfsDirectory>(full_path, perms));
-                else
-                    files.emplace_back(std::make_shared<RealVfsFile>(full_path, perms));
-                return true;
-            });
-    }
+    if (perms == Mode::Append)
+        return;
+
+    FileUtil::ForeachDirectoryEntry(
+        &size, path,
+        [this](unsigned* entries_out, const std::string& directory, const std::string& filename) {
+            std::string full_path = directory + DIR_SEP + filename;
+            if (FileUtil::IsDirectory(full_path))
+                subdirectories.emplace_back(std::make_shared<RealVfsDirectory>(full_path, perms));
+            else
+                files.emplace_back(std::make_shared<RealVfsFile>(full_path, perms));
+            return true;
+        });
 }
 
 std::vector<std::shared_ptr<VfsFile>> RealVfsDirectory::GetFiles() const {
@@ -146,6 +149,12 @@ bool RealVfsDirectory::DeleteSubdirectory(const std::string& name) {
 }
 
 bool RealVfsDirectory::DeleteFile(const std::string& name) {
+    auto file = GetFile(name);
+    if (file == nullptr)
+        return false;
+    files.erase(std::find(files.begin(), files.end(), file));
+    auto real_file = std::static_pointer_cast<RealVfsFile>(file);
+    real_file->Close();
     return FileUtil::Delete(path + DIR_SEP + name);
 }
 

src/core/file_sys/vfs_real.h

@@ -5,13 +5,15 @@
 #pragma once
 
 #include "common/file_util.h"
-#include "core/file_sys/filesystem.h"
+#include "core/file_sys/mode.h"
 #include "core/file_sys/vfs.h"
 
 namespace FileSys {
 
 // An implmentation of VfsFile that represents a file on the user's computer.
 struct RealVfsFile : public VfsFile {
+    friend struct RealVfsDirectory;
+
     RealVfsFile(const std::string& name, Mode perms = Mode::Read);
 
     std::string GetName() const override;
@@ -25,6 +27,8 @@ struct RealVfsFile : public VfsFile {
     bool Rename(const std::string& name) override;
 
 private:
+    bool Close();
+
     FileUtil::IOFile backing;
     std::string path;
     std::string parent_path;

src/core/gdbstub/gdbstub.cpp

@@ -61,10 +61,16 @@ const u32 SIGTERM = 15;
 const u32 MSG_WAITALL = 8;
 #endif
 
-const u32 X30_REGISTER = 30;
+const u32 LR_REGISTER = 30;
 const u32 SP_REGISTER = 31;
 const u32 PC_REGISTER = 32;
 const u32 CPSR_REGISTER = 33;
+const u32 UC_ARM64_REG_Q0 = 34;
+const u32 FPSCR_REGISTER = 66;
+
+// TODO/WiP - Used while working on support for FPU
+const u32 TODO_DUMMY_REG_997 = 997;
+const u32 TODO_DUMMY_REG_998 = 998;
 
 // For sample XML files see the GDB source /gdb/features
 // GDB also wants the l character at the start
@@ -130,6 +136,8 @@ static const char* target_xml =
     </flags>
     <reg name="cpsr" bitsize="32" type="cpsr_flags"/>
   </feature>
+  <feature name="org.gnu.gdb.aarch64.fpu">
+  </feature>
 </target>
 )";
 
@@ -144,6 +152,7 @@ static u32 latest_signal = 0;
 static bool memory_break = false;
 
 static Kernel::Thread* current_thread = nullptr;
+static u32 current_core = 0;
 
 // Binding to a port within the reserved ports range (0-1023) requires root permissions,
 // so default to a port outside of that range.
@@ -171,13 +180,34 @@ static std::map<u64, Breakpoint> breakpoints_execute;
 static std::map<u64, Breakpoint> breakpoints_read;
 static std::map<u64, Breakpoint> breakpoints_write;
 
+struct Module {
+    std::string name;
+    PAddr beg;
+    PAddr end;
+};
+
+static std::vector<Module> modules;
+
+void RegisterModule(std::string name, PAddr beg, PAddr end, bool add_elf_ext) {
+    Module module;
+    if (add_elf_ext) {
+        Common::SplitPath(name, nullptr, &module.name, nullptr);
+        module.name += ".elf";
+    } else {
+        module.name = std::move(name);
+    }
+    module.beg = beg;
+    module.end = end;
+    modules.push_back(std::move(module));
+}
+
 static Kernel::Thread* FindThreadById(int id) {
-    for (int core = 0; core < Core::NUM_CPU_CORES; core++) {
-        auto threads = Core::System::GetInstance().Scheduler(core)->GetThreadList();
-        for (auto thread : threads) {
+    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() == id) {
-                current_thread = thread.get();
-                return current_thread;
+                current_core = core;
+                return thread.get();
             }
         }
     }
@@ -197,6 +227,8 @@ static u64 RegRead(int id, Kernel::Thread* thread = nullptr) {
         return thread->context.pc;
     } else if (id == CPSR_REGISTER) {
         return thread->context.cpsr;
+    } else if (id > CPSR_REGISTER && id < FPSCR_REGISTER) {
+        return thread->context.fpu_registers[id - UC_ARM64_REG_Q0][0];
     } else {
         return 0;
     }
@@ -215,6 +247,8 @@ static void RegWrite(int id, u64 val, Kernel::Thread* thread = nullptr) {
         thread->context.pc = val;
     } else if (id == CPSR_REGISTER) {
         thread->context.cpsr = val;
+    } else if (id > CPSR_REGISTER && id < FPSCR_REGISTER) {
+        thread->context.fpu_registers[id - (CPSR_REGISTER + 1)][0] = val;
     }
 }
 
@@ -534,7 +568,11 @@ static void HandleQuery() {
         SendReply("T0");
     } else if (strncmp(query, "Supported", strlen("Supported")) == 0) {
         // PacketSize needs to be large enough for target xml
-        SendReply("PacketSize=2000;qXfer:features:read+");
+        std::string buffer = "PacketSize=2000;qXfer:features:read+;qXfer:threads:read+";
+        if (!modules.empty()) {
+            buffer += ";qXfer:libraries:read+";
+        }
+        SendReply(buffer.c_str());
     } else if (strncmp(query, "Xfer:features:read:target.xml:",
                        strlen("Xfer:features:read:target.xml:")) == 0) {
         SendReply(target_xml);
@@ -543,9 +581,9 @@ static void HandleQuery() {
         SendReply(buffer.c_str());
     } else if (strncmp(query, "fThreadInfo", strlen("fThreadInfo")) == 0) {
         std::string val = "m";
-        for (int core = 0; core < Core::NUM_CPU_CORES; core++) {
-            auto threads = Core::System::GetInstance().Scheduler(core)->GetThreadList();
-            for (auto thread : threads) {
+        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 += ",";
             }
@@ -554,6 +592,31 @@ static void HandleQuery() {
         SendReply(val.c_str());
     } else if (strncmp(query, "sThreadInfo", strlen("sThreadInfo")) == 0) {
         SendReply("l");
+    } else if (strncmp(query, "Xfer:threads:read", strlen("Xfer:threads:read")) == 0) {
+        std::string buffer;
+        buffer += "l<?xml version=\"1.0\"?>";
+        buffer += "<threads>";
+        for (u32 core = 0; core < Core::NUM_CPU_CORES; core++) {
+            const auto& threads = Core::System::GetInstance().Scheduler(core)->GetThreadList();
+            for (const auto& thread : threads) {
+                buffer +=
+                    fmt::format(R"*(<thread id="{:x}" core="{:d}" name="Thread {:x}"></thread>)*",
+                                thread->GetThreadId(), core, thread->GetThreadId());
+            }
+        }
+        buffer += "</threads>";
+        SendReply(buffer.c_str());
+    } else if (strncmp(query, "Xfer:libraries:read", strlen("Xfer:libraries:read")) == 0) {
+        std::string buffer;
+        buffer += "l<?xml version=\"1.0\"?>";
+        buffer += "<library-list>";
+        for (const auto& module : modules) {
+            buffer +=
+                fmt::format(R"*("<library name = "{}"><segment address = "0x{:x}"/></library>)*",
+                            module.name, module.beg);
+        }
+        buffer += "</library-list>";
+        SendReply(buffer.c_str());
     } else {
         SendReply("");
     }
@@ -561,33 +624,27 @@ static void HandleQuery() {
 
 /// Handle set thread command from gdb client.
 static void HandleSetThread() {
-    if (memcmp(command_buffer, "Hc", 2) == 0 || memcmp(command_buffer, "Hg", 2) == 0) {
-        int thread_id = -1;
-        if (command_buffer[2] != '-') {
-            thread_id = static_cast<int>(HexToInt(
-                command_buffer + 2,
-                command_length - 2 /*strlen(reinterpret_cast<char*>(command_buffer) + 2)*/));
-        }
-        if (thread_id >= 1) {
-            current_thread = FindThreadById(thread_id);
-        }
-        if (!current_thread) {
-            thread_id = 1;
-            current_thread = FindThreadById(thread_id);
-        }
-        if (current_thread) {
-            SendReply("OK");
-            return;
-        }
+    int thread_id = -1;
+    if (command_buffer[2] != '-') {
+        thread_id = static_cast<int>(HexToInt(command_buffer + 2, command_length - 2));
+    }
+    if (thread_id >= 1) {
+        current_thread = FindThreadById(thread_id);
+    }
+    if (!current_thread) {
+        thread_id = 1;
+        current_thread = FindThreadById(thread_id);
+    }
+    if (current_thread) {
+        SendReply("OK");
+        return;
     }
     SendReply("E01");
 }
 
 /// Handle thread alive command from gdb client.
 static void HandleThreadAlive() {
-    int thread_id = static_cast<int>(
-        HexToInt(command_buffer + 1,
-                 command_length - 1 /*strlen(reinterpret_cast<char*>(command_buffer) + 1)*/));
+    int thread_id = static_cast<int>(HexToInt(command_buffer + 1, command_length - 1));
     if (thread_id == 0) {
         thread_id = 1;
     }
@@ -610,16 +667,23 @@ static void SendSignal(Kernel::Thread* thread, u32 signal, bool full = true) {
 
     latest_signal = signal;
 
-    std::string buffer;
-    if (full) {
-        buffer = fmt::format("T{:02x}{:02x}:{:016x};{:02x}:{:016x};", latest_signal, PC_REGISTER,
-                             Common::swap64(RegRead(PC_REGISTER, thread)), SP_REGISTER,
-                             Common::swap64(RegRead(SP_REGISTER, thread)));
-    } else {
-        buffer = fmt::format("T{:02x};", latest_signal);
+    if (!thread) {
+        full = false;
     }
 
-    buffer += fmt::format("thread:{:x};", thread->GetThreadId());
+    std::string buffer;
+    if (full) {
+        buffer = fmt::format("T{:02x}{:02x}:{:016x};{:02x}:{:016x};{:02x}:{:016x}", latest_signal,
+                             PC_REGISTER, Common::swap64(RegRead(PC_REGISTER, thread)), SP_REGISTER,
+                             Common::swap64(RegRead(SP_REGISTER, thread)), LR_REGISTER,
+                             Common::swap64(RegRead(LR_REGISTER, thread)));
+    } else {
+        buffer = fmt::format("T{:02x}", latest_signal);
+    }
+
+    if (thread) {
+        buffer += fmt::format(";thread:{:x};", thread->GetThreadId());
+    }
 
     SendReply(buffer.c_str());
 }
@@ -711,8 +775,12 @@ static void ReadRegister() {
         LongToGdbHex(reply, RegRead(id, current_thread));
     } else if (id == CPSR_REGISTER) {
         IntToGdbHex(reply, (u32)RegRead(id, current_thread));
+    } else if (id >= UC_ARM64_REG_Q0 && id < FPSCR_REGISTER) {
+        LongToGdbHex(reply, RegRead(id, current_thread));
+    } else if (id == FPSCR_REGISTER) {
+        LongToGdbHex(reply, RegRead(TODO_DUMMY_REG_998, current_thread));
     } else {
-        return SendReply("E01");
+        LongToGdbHex(reply, RegRead(TODO_DUMMY_REG_997, current_thread));
     }
 
     SendReply(reinterpret_cast<char*>(reply));
@@ -729,7 +797,7 @@ static void ReadRegisters() {
         LongToGdbHex(bufptr + reg * 16, RegRead(reg, current_thread));
     }
 
-    bufptr += (32 * 16);
+    bufptr += 32 * 16;
 
     LongToGdbHex(bufptr, RegRead(PC_REGISTER, current_thread));
 
@@ -739,6 +807,16 @@ static void ReadRegisters() {
 
     bufptr += 8;
 
+    for (int reg = UC_ARM64_REG_Q0; reg <= UC_ARM64_REG_Q0 + 31; reg++) {
+        LongToGdbHex(bufptr + reg * 16, RegRead(reg, current_thread));
+    }
+
+    bufptr += 32 * 32;
+
+    LongToGdbHex(bufptr, RegRead(TODO_DUMMY_REG_998, current_thread));
+
+    bufptr += 8;
+
     SendReply(reinterpret_cast<char*>(buffer));
 }
 
@@ -759,10 +837,17 @@ static void WriteRegister() {
         RegWrite(id, GdbHexToLong(buffer_ptr), current_thread);
     } else if (id == CPSR_REGISTER) {
         RegWrite(id, GdbHexToInt(buffer_ptr), current_thread);
+    } else if (id >= UC_ARM64_REG_Q0 && id < FPSCR_REGISTER) {
+        RegWrite(id, GdbHexToLong(buffer_ptr), current_thread);
+    } else if (id == FPSCR_REGISTER) {
+        RegWrite(TODO_DUMMY_REG_998, GdbHexToLong(buffer_ptr), current_thread);
     } else {
-        return SendReply("E01");
+        RegWrite(TODO_DUMMY_REG_997, GdbHexToLong(buffer_ptr), current_thread);
     }
 
+    // Update Unicorn context skipping scheduler, no running threads at this point
+    Core::System::GetInstance().ArmInterface(current_core).LoadContext(current_thread->context);
+
     SendReply("OK");
 }
 
@@ -773,18 +858,25 @@ static void WriteRegisters() {
     if (command_buffer[0] != 'G')
         return SendReply("E01");
 
-    for (int i = 0, reg = 0; reg <= CPSR_REGISTER; i++, reg++) {
+    for (int i = 0, reg = 0; reg <= FPSCR_REGISTER; i++, reg++) {
         if (reg <= SP_REGISTER) {
             RegWrite(reg, GdbHexToLong(buffer_ptr + i * 16), current_thread);
         } else if (reg == PC_REGISTER) {
             RegWrite(PC_REGISTER, GdbHexToLong(buffer_ptr + i * 16), current_thread);
         } else if (reg == CPSR_REGISTER) {
             RegWrite(CPSR_REGISTER, GdbHexToInt(buffer_ptr + i * 16), current_thread);
+        } else if (reg >= UC_ARM64_REG_Q0 && reg < FPSCR_REGISTER) {
+            RegWrite(reg, GdbHexToLong(buffer_ptr + i * 16), current_thread);
+        } else if (reg == FPSCR_REGISTER) {
+            RegWrite(TODO_DUMMY_REG_998, GdbHexToLong(buffer_ptr + i * 16), current_thread);
         } else {
             UNIMPLEMENTED();
         }
     }
 
+    // Update Unicorn context skipping scheduler, no running threads at this point
+    Core::System::GetInstance().ArmInterface(current_core).LoadContext(current_thread->context);
+
     SendReply("OK");
 }
 
@@ -806,6 +898,10 @@ static void ReadMemory() {
         SendReply("E01");
     }
 
+    if (addr < Memory::PROCESS_IMAGE_VADDR || addr >= Memory::MAP_REGION_VADDR_END) {
+        return SendReply("E00");
+    }
+
     if (!Memory::IsValidVirtualAddress(addr)) {
         return SendReply("E00");
     }
@@ -840,16 +936,18 @@ static void WriteMemory() {
 }
 
 void Break(bool is_memory_break) {
-    if (!halt_loop) {
-        halt_loop = true;
-        send_trap = true;
-    }
+    send_trap = true;
 
     memory_break = is_memory_break;
 }
 
 /// Tell the CPU that it should perform a single step.
 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);
+    }
     step_loop = true;
     halt_loop = true;
     send_trap = true;
@@ -1090,6 +1188,8 @@ static void Init(u16 port) {
     breakpoints_read.clear();
     breakpoints_write.clear();
 
+    modules.clear();
+
     // Start gdb server
     LOG_INFO(Debug_GDBStub, "Starting GDB server on port {}...", port);
 
@@ -1192,8 +1292,12 @@ void SetCpuStepFlag(bool is_step) {
 
 void SendTrap(Kernel::Thread* thread, int trap) {
     if (send_trap) {
+        if (!halt_loop || current_thread == thread) {
+            current_thread = thread;
+            SendSignal(thread, trap);
+        }
+        halt_loop = true;
         send_trap = false;
-        SendSignal(thread, trap);
     }
 }
 }; // namespace GDBStub

src/core/gdbstub/gdbstub.h

@@ -6,6 +6,7 @@
 
 #pragma once
 
+#include <string>
 #include "common/common_types.h"
 #include "core/hle/kernel/thread.h"
 
@@ -51,6 +52,9 @@ bool IsServerEnabled();
 /// Returns true if there is an active socket connection.
 bool IsConnected();
 
+/// Register module.
+void RegisterModule(std::string name, PAddr beg, PAddr end, bool add_elf_ext = true);
+
 /**
  * Signal to the gdbstub server that it should halt CPU execution.
  *
@@ -80,10 +84,10 @@ BreakpointAddress GetNextBreakpointFromAddress(PAddr addr, GDBStub::BreakpointTy
  */
 bool CheckBreakpoint(PAddr addr, GDBStub::BreakpointType type);
 
-// If set to true, the CPU will halt at the beginning of the next CPU loop.
+/// If set to true, the CPU will halt at the beginning of the next CPU loop.
 bool GetCpuHaltFlag();
 
-// If set to true and the CPU is halted, the CPU will step one instruction.
+/// If set to true and the CPU is halted, the CPU will step one instruction.
 bool GetCpuStepFlag();
 
 /**

src/core/hle/kernel/address_arbiter.cpp

@@ -65,7 +65,7 @@ static void WakeThreads(std::vector<SharedPtr<Thread>>& waiting_threads, s32 num
 
     // Signal the waiting threads.
     for (size_t i = 0; i < last; i++) {
-        ASSERT(waiting_threads[i]->status = THREADSTATUS_WAIT_ARB);
+        ASSERT(waiting_threads[i]->status == THREADSTATUS_WAIT_ARB);
         waiting_threads[i]->SetWaitSynchronizationResult(RESULT_SUCCESS);
         waiting_threads[i]->arb_wait_address = 0;
         waiting_threads[i]->ResumeFromWait();
@@ -115,7 +115,7 @@ ResultCode ModifyByWaitingCountAndSignalToAddressIfEqual(VAddr address, s32 valu
     s32 updated_value;
     if (waiting_threads.size() == 0) {
         updated_value = value - 1;
-    } else if (num_to_wake <= 0 || waiting_threads.size() <= num_to_wake) {
+    } else if (num_to_wake <= 0 || waiting_threads.size() <= static_cast<u32>(num_to_wake)) {
         updated_value = value + 1;
     } else {
         updated_value = value;
@@ -140,7 +140,9 @@ ResultCode WaitForAddressIfLessThan(VAddr address, s32 value, s64 timeout, bool
 
     s32 cur_value = static_cast<s32>(Memory::Read32(address));
     if (cur_value < value) {
-        Memory::Write32(address, static_cast<u32>(cur_value - 1));
+        if (should_decrement) {
+            Memory::Write32(address, static_cast<u32>(cur_value - 1));
+        }
     } else {
         return ERR_INVALID_STATE;
     }

src/core/hle/kernel/hle_ipc.cpp

@@ -2,6 +2,8 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <utility>
+
 #include <boost/range/algorithm_ext/erase.hpp>
 #include "common/assert.h"
 #include "common/common_funcs.h"
@@ -19,17 +21,18 @@ namespace Kernel {
 
 void SessionRequestHandler::ClientConnected(SharedPtr<ServerSession> server_session) {
     server_session->SetHleHandler(shared_from_this());
-    connected_sessions.push_back(server_session);
+    connected_sessions.push_back(std::move(server_session));
 }
 
-void SessionRequestHandler::ClientDisconnected(SharedPtr<ServerSession> server_session) {
+void SessionRequestHandler::ClientDisconnected(const SharedPtr<ServerSession>& server_session) {
     server_session->SetHleHandler(nullptr);
     boost::range::remove_erase(connected_sessions, server_session);
 }
 
 SharedPtr<Event> HLERequestContext::SleepClientThread(SharedPtr<Thread> thread,
                                                       const std::string& reason, u64 timeout,
-                                                      WakeupCallback&& callback) {
+                                                      WakeupCallback&& callback,
+                                                      Kernel::SharedPtr<Kernel::Event> event) {
 
     // Put the client thread to sleep until the wait event is signaled or the timeout expires.
     thread->wakeup_callback =
@@ -41,7 +44,12 @@ SharedPtr<Event> HLERequestContext::SleepClientThread(SharedPtr<Thread> thread,
         return true;
     };
 
-    auto event = Kernel::Event::Create(Kernel::ResetType::OneShot, "HLE Pause Event: " + reason);
+    if (!event) {
+        // Create event if not provided
+        event = Kernel::Event::Create(Kernel::ResetType::OneShot, "HLE Pause Event: " + reason);
+    }
+
+    event->Clear();
     thread->status = THREADSTATUS_WAIT_HLE_EVENT;
     thread->wait_objects = {event};
     event->AddWaitingThread(thread);
@@ -214,8 +222,8 @@ ResultCode HLERequestContext::WriteToOutgoingCommandBuffer(Thread& thread) {
             (sizeof(IPC::CommandHeader) + sizeof(IPC::HandleDescriptorHeader)) / sizeof(u32);
         ASSERT_MSG(!handle_descriptor_header->send_current_pid, "Sending PID is not implemented");
 
-        ASSERT_MSG(copy_objects.size() == handle_descriptor_header->num_handles_to_copy);
-        ASSERT_MSG(move_objects.size() == handle_descriptor_header->num_handles_to_move);
+        ASSERT(copy_objects.size() == handle_descriptor_header->num_handles_to_copy);
+        ASSERT(move_objects.size() == handle_descriptor_header->num_handles_to_move);
 
         // We don't make a distinction between copy and move handles when translating since HLE
         // services don't deal with handles directly. However, the guest applications might check

src/core/hle/kernel/hle_ipc.h

@@ -59,12 +59,12 @@ public:
      * associated ServerSession.
      * @param server_session ServerSession associated with the connection.
      */
-    void ClientDisconnected(SharedPtr<ServerSession> server_session);
+    void ClientDisconnected(const SharedPtr<ServerSession>& server_session);
 
 protected:
     /// List of sessions that are connected to this handler.
     /// A ServerSession whose server endpoint is an HLE implementation is kept alive by this list
-    // for the duration of the connection.
+    /// for the duration of the connection.
     std::vector<SharedPtr<ServerSession>> connected_sessions;
 };
 
@@ -118,10 +118,12 @@ public:
      * @param callback Callback to be invoked when the thread is resumed. This callback must write
      * the entire command response once again, regardless of the state of it before this function
      * was called.
+     * @param event Event to use to wake up the thread. If unspecified, an event will be created.
      * @returns Event that when signaled will resume the thread and call the callback function.
      */
     SharedPtr<Event> SleepClientThread(SharedPtr<Thread> thread, const std::string& reason,
-                                       u64 timeout, WakeupCallback&& callback);
+                                       u64 timeout, WakeupCallback&& callback,
+                                       Kernel::SharedPtr<Kernel::Event> event = nullptr);
 
     void ParseCommandBuffer(u32_le* src_cmdbuf, bool incoming);
 

src/core/hle/kernel/object_address_table.cpp

@@ -2,6 +2,8 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <utility>
+
 #include "common/assert.h"
 #include "core/hle/kernel/object_address_table.h"
 
@@ -11,7 +13,7 @@ ObjectAddressTable g_object_address_table;
 
 void ObjectAddressTable::Insert(VAddr addr, SharedPtr<Object> obj) {
     ASSERT_MSG(objects.find(addr) == objects.end(), "Object already exists with addr=0x{:X}", addr);
-    objects[addr] = obj;
+    objects[addr] = std::move(obj);
 }
 
 void ObjectAddressTable::Close(VAddr addr) {

src/core/hle/kernel/scheduler.cpp

@@ -2,6 +2,8 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <utility>
+
 #include "core/core.h"
 #include "core/core_timing.h"
 #include "core/hle/kernel/process.h"
@@ -85,6 +87,7 @@ void Scheduler::SwitchContext(Thread* new_thread) {
 
         cpu_core->LoadContext(new_thread->context);
         cpu_core->SetTlsAddress(new_thread->GetTLSAddress());
+        cpu_core->ClearExclusiveState();
     } else {
         current_thread = nullptr;
         // Note: We do not reset the current process and current page table when idling because
@@ -112,7 +115,7 @@ void Scheduler::Reschedule() {
 void Scheduler::AddThread(SharedPtr<Thread> thread, u32 priority) {
     std::lock_guard<std::mutex> lock(scheduler_mutex);
 
-    thread_list.push_back(thread);
+    thread_list.push_back(std::move(thread));
     ready_queue.prepare(priority);
 }
 

src/core/hle/kernel/server_session.cpp

@@ -3,6 +3,7 @@
 // Refer to the license.txt file included.
 
 #include <tuple>
+#include <utility>
 
 #include "core/core.h"
 #include "core/hle/ipc_helpers.h"
@@ -158,7 +159,7 @@ ServerSession::SessionPair ServerSession::CreateSessionPair(const std::string& n
     std::shared_ptr<Session> parent(new Session);
     parent->client = client_session.get();
     parent->server = server_session.get();
-    parent->port = port;
+    parent->port = std::move(port);
 
     client_session->parent = parent;
     server_session->parent = parent;

src/core/hle/kernel/shared_memory.cpp

@@ -2,7 +2,7 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#include <cstring>
+#include <utility>
 #include "common/logging/log.h"
 #include "core/core.h"
 #include "core/hle/kernel/errors.h"
@@ -21,7 +21,7 @@ SharedPtr<SharedMemory> SharedMemory::Create(SharedPtr<Process> owner_process, u
                                              MemoryRegion region, std::string name) {
     SharedPtr<SharedMemory> shared_memory(new SharedMemory);
 
-    shared_memory->owner_process = owner_process;
+    shared_memory->owner_process = std::move(owner_process);
     shared_memory->name = std::move(name);
     shared_memory->size = size;
     shared_memory->permissions = permissions;
@@ -87,7 +87,7 @@ SharedPtr<SharedMemory> SharedMemory::CreateForApplet(std::shared_ptr<std::vecto
     shared_memory->size = size;
     shared_memory->permissions = permissions;
     shared_memory->other_permissions = other_permissions;
-    shared_memory->backing_block = heap_block;
+    shared_memory->backing_block = std::move(heap_block);
     shared_memory->backing_block_offset = offset;
     shared_memory->base_address = Memory::HEAP_VADDR + offset;
 

src/core/hle/kernel/thread.cpp

@@ -165,7 +165,7 @@ void Thread::CancelWakeupTimer() {
 static boost::optional<s32> GetNextProcessorId(u64 mask) {
     for (s32 index = 0; index < Core::NUM_CPU_CORES; ++index) {
         if (mask & (1ULL << index)) {
-            if (!Core::System().GetInstance().Scheduler(index)->GetCurrentThread()) {
+            if (!Core::System::GetInstance().Scheduler(index)->GetCurrentThread()) {
                 // Core is enabled and not running any threads, use this one
                 return index;
             }
@@ -215,14 +215,14 @@ void Thread::ResumeFromWait() {
         new_processor_id = processor_id;
     }
     if (ideal_core != -1 &&
-        Core::System().GetInstance().Scheduler(ideal_core)->GetCurrentThread() == nullptr) {
+        Core::System::GetInstance().Scheduler(ideal_core)->GetCurrentThread() == nullptr) {
         new_processor_id = ideal_core;
     }
 
     ASSERT(*new_processor_id < 4);
 
     // Add thread to new core's scheduler
-    auto& next_scheduler = Core::System().GetInstance().Scheduler(*new_processor_id);
+    auto& next_scheduler = Core::System::GetInstance().Scheduler(*new_processor_id);
 
     if (*new_processor_id != processor_id) {
         // Remove thread from previous core's scheduler
@@ -325,7 +325,7 @@ ResultVal<SharedPtr<Thread>> Thread::Create(std::string name, VAddr entry_point,
     thread->name = std::move(name);
     thread->callback_handle = wakeup_callback_handle_table.Create(thread).Unwrap();
     thread->owner_process = owner_process;
-    thread->scheduler = Core::System().GetInstance().Scheduler(processor_id);
+    thread->scheduler = Core::System::GetInstance().Scheduler(processor_id);
     thread->scheduler->AddThread(thread, priority);
 
     // Find the next available TLS index, and mark it as used
@@ -400,7 +400,7 @@ SharedPtr<Thread> SetupMainThread(VAddr entry_point, u32 priority,
 
     // Initialize new "main" thread
     auto thread_res = Thread::Create("main", entry_point, priority, 0, THREADPROCESSORID_0,
-                                     Memory::STACK_AREA_VADDR_END, owner_process);
+                                     Memory::STACK_AREA_VADDR_END, std::move(owner_process));
 
     SharedPtr<Thread> thread = std::move(thread_res).Unwrap();
 
@@ -481,14 +481,14 @@ void Thread::ChangeCore(u32 core, u64 mask) {
         new_processor_id = processor_id;
     }
     if (ideal_core != -1 &&
-        Core::System().GetInstance().Scheduler(ideal_core)->GetCurrentThread() == nullptr) {
+        Core::System::GetInstance().Scheduler(ideal_core)->GetCurrentThread() == nullptr) {
         new_processor_id = ideal_core;
     }
 
     ASSERT(*new_processor_id < 4);
 
     // Add thread to new core's scheduler
-    auto& next_scheduler = Core::System().GetInstance().Scheduler(*new_processor_id);
+    auto& next_scheduler = Core::System::GetInstance().Scheduler(*new_processor_id);
 
     if (*new_processor_id != processor_id) {
         // Remove thread from previous core's scheduler

src/core/hle/kernel/vm_manager.cpp

@@ -3,6 +3,7 @@
 // Refer to the license.txt file included.
 
 #include <iterator>
+#include <utility>
 #include "common/assert.h"
 #include "common/logging/log.h"
 #include "core/arm/arm_interface.h"
@@ -16,30 +17,20 @@
 namespace Kernel {
 
 static const char* GetMemoryStateName(MemoryState state) {
-    static const char* names[] = {
-        "Unmapped",
-        "Io",
-        "Normal",
-        "CodeStatic",
-        "CodeMutable",
-        "Heap",
-        "Shared",
-        "Unknown1"
-        "ModuleCodeStatic",
-        "ModuleCodeMutable",
-        "IpcBuffer0",
-        "Mapped",
-        "ThreadLocal",
-        "TransferMemoryIsolated",
-        "TransferMemory",
-        "ProcessMemory",
-        "Unknown2"
-        "IpcBuffer1",
-        "IpcBuffer3",
-        "KernelStack",
+    static constexpr const char* names[] = {
+        "Unmapped",         "Io",
+        "Normal",           "CodeStatic",
+        "CodeMutable",      "Heap",
+        "Shared",           "Unknown1",
+        "ModuleCodeStatic", "ModuleCodeMutable",
+        "IpcBuffer0",       "Mapped",
+        "ThreadLocal",      "TransferMemoryIsolated",
+        "TransferMemory",   "ProcessMemory",
+        "Unknown2",         "IpcBuffer1",
+        "IpcBuffer3",       "KernelStack",
     };
 
-    return names[(int)state];
+    return names[static_cast<int>(state)];
 }
 
 bool VirtualMemoryArea::CanBeMergedWith(const VirtualMemoryArea& next) const {
@@ -117,7 +108,7 @@ ResultVal<VMManager::VMAHandle> VMManager::MapMemoryBlock(VAddr target,
     final_vma.type = VMAType::AllocatedMemoryBlock;
     final_vma.permissions = VMAPermission::ReadWrite;
     final_vma.meminfo_state = state;
-    final_vma.backing_block = block;
+    final_vma.backing_block = std::move(block);
     final_vma.offset = offset;
     UpdatePageTableForVMA(final_vma);
 
@@ -160,7 +151,7 @@ ResultVal<VMManager::VMAHandle> VMManager::MapMMIO(VAddr target, PAddr paddr, u6
     final_vma.permissions = VMAPermission::ReadWrite;
     final_vma.meminfo_state = state;
     final_vma.paddr = paddr;
-    final_vma.mmio_handler = mmio_handler;
+    final_vma.mmio_handler = std::move(mmio_handler);
     UpdatePageTableForVMA(final_vma);
 
     return MakeResult<VMAHandle>(MergeAdjacent(vma_handle));

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

@@ -4,9 +4,10 @@
 
 #include <cinttypes>
 #include <stack>
-#include "core/file_sys/filesystem.h"
+#include "core/core.h"
 #include "core/hle/ipc_helpers.h"
 #include "core/hle/kernel/event.h"
+#include "core/hle/kernel/process.h"
 #include "core/hle/service/am/am.h"
 #include "core/hle/service/am/applet_ae.h"
 #include "core/hle/service/am/applet_oe.h"
@@ -614,25 +615,14 @@ void IApplicationFunctions::CreateApplicationAndRequestToStartForQuest(
 
 void IApplicationFunctions::EnsureSaveData(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
-    u128 uid = rp.PopRaw<u128>();
+    u128 uid = rp.PopRaw<u128>(); // What does this do?
 
     LOG_WARNING(Service, "(STUBBED) called uid = {:016X}{:016X}", uid[1], uid[0]);
 
     IPC::ResponseBuilder rb{ctx, 4};
-
-    FileSys::Path unused;
-    auto savedata = FileSystem::OpenFileSystem(FileSystem::Type::SaveData);
-    if (savedata.Failed()) {
-        // Create the save data and return an error indicating that the operation was performed.
-        FileSystem::FormatFileSystem(FileSystem::Type::SaveData);
-        // TODO(Subv): Find out the correct error code for this.
-        rb.Push(ResultCode(ErrorModule::FS, 40));
-    } else {
-        rb.Push(RESULT_SUCCESS);
-    }
-
+    rb.Push(RESULT_SUCCESS);
     rb.Push<u64>(0);
-}
+} // namespace Service::AM
 
 void IApplicationFunctions::SetTerminateResult(Kernel::HLERequestContext& ctx) {
     // Takes an input u32 Result, no output.

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

@@ -27,12 +27,12 @@ public:
             {0, &IAudioOut::GetAudioOutState, "GetAudioOutState"},
             {1, &IAudioOut::StartAudioOut, "StartAudioOut"},
             {2, &IAudioOut::StopAudioOut, "StopAudioOut"},
-            {3, &IAudioOut::AppendAudioOutBuffer, "AppendAudioOutBuffer"},
+            {3, &IAudioOut::AppendAudioOutBufferImpl, "AppendAudioOutBuffer"},
             {4, &IAudioOut::RegisterBufferEvent, "RegisterBufferEvent"},
-            {5, &IAudioOut::GetReleasedAudioOutBuffer, "GetReleasedAudioOutBuffer"},
+            {5, &IAudioOut::GetReleasedAudioOutBufferImpl, "GetReleasedAudioOutBuffer"},
             {6, nullptr, "ContainsAudioOutBuffer"},
-            {7, nullptr, "AppendAudioOutBufferAuto"},
-            {8, nullptr, "GetReleasedAudioOutBufferAuto"},
+            {7, &IAudioOut::AppendAudioOutBufferImpl, "AppendAudioOutBufferAuto"},
+            {8, &IAudioOut::GetReleasedAudioOutBufferImpl, "GetReleasedAudioOutBufferAuto"},
             {9, nullptr, "GetAudioOutBufferCount"},
             {10, nullptr, "GetAudioOutPlayedSampleCount"},
             {11, nullptr, "FlushAudioOutBuffers"},
@@ -96,7 +96,7 @@ private:
         rb.PushCopyObjects(buffer_event);
     }
 
-    void AppendAudioOutBuffer(Kernel::HLERequestContext& ctx) {
+    void AppendAudioOutBufferImpl(Kernel::HLERequestContext& ctx) {
         LOG_WARNING(Service_Audio, "(STUBBED) called");
         IPC::RequestParser rp{ctx};
 
@@ -107,7 +107,7 @@ private:
         rb.Push(RESULT_SUCCESS);
     }
 
-    void GetReleasedAudioOutBuffer(Kernel::HLERequestContext& ctx) {
+    void GetReleasedAudioOutBufferImpl(Kernel::HLERequestContext& ctx) {
         LOG_WARNING(Service_Audio, "(STUBBED) called");
 
         // TODO(st4rk): This is how libtransistor currently implements the
@@ -163,7 +163,7 @@ private:
     AudioState audio_out_state;
 };
 
-void AudOutU::ListAudioOuts(Kernel::HLERequestContext& ctx) {
+void AudOutU::ListAudioOutsImpl(Kernel::HLERequestContext& ctx) {
     LOG_WARNING(Service_Audio, "(STUBBED) called");
     IPC::RequestParser rp{ctx};
 
@@ -179,7 +179,7 @@ void AudOutU::ListAudioOuts(Kernel::HLERequestContext& ctx) {
     rb.Push<u32>(1);
 }
 
-void AudOutU::OpenAudioOut(Kernel::HLERequestContext& ctx) {
+void AudOutU::OpenAudioOutImpl(Kernel::HLERequestContext& ctx) {
     LOG_WARNING(Service_Audio, "(STUBBED) called");
 
     if (!audio_out_interface) {
@@ -196,10 +196,10 @@ void AudOutU::OpenAudioOut(Kernel::HLERequestContext& ctx) {
 }
 
 AudOutU::AudOutU() : ServiceFramework("audout:u") {
-    static const FunctionInfo functions[] = {{0, &AudOutU::ListAudioOuts, "ListAudioOuts"},
-                                             {1, &AudOutU::OpenAudioOut, "OpenAudioOut"},
-                                             {2, nullptr, "ListAudioOutsAuto"},
-                                             {3, nullptr, "OpenAudioOutAuto"}};
+    static const FunctionInfo functions[] = {{0, &AudOutU::ListAudioOutsImpl, "ListAudioOuts"},
+                                             {1, &AudOutU::OpenAudioOutImpl, "OpenAudioOut"},
+                                             {2, &AudOutU::ListAudioOutsImpl, "ListAudioOutsAuto"},
+                                             {3, &AudOutU::OpenAudioOutImpl, "OpenAudioOutAuto"}};
     RegisterHandlers(functions);
 }
 

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

@@ -22,8 +22,8 @@ public:
 private:
     std::shared_ptr<IAudioOut> audio_out_interface;
 
-    void ListAudioOuts(Kernel::HLERequestContext& ctx);
-    void OpenAudioOut(Kernel::HLERequestContext& ctx);
+    void ListAudioOutsImpl(Kernel::HLERequestContext& ctx);
+    void OpenAudioOutImpl(Kernel::HLERequestContext& ctx);
 
     enum class PcmFormat : u32 {
         Invalid = 0,

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

@@ -47,7 +47,7 @@ public:
 
         // Start the audio event
         CoreTiming::ScheduleEvent(audio_ticks, audio_event);
-        voice_status_list.reserve(worker_params.voice_count);
+        voice_status_list.resize(worker_params.voice_count);
     }
     ~IAudioRenderer() {
         CoreTiming::UnscheduleEvent(audio_event, 0);
@@ -87,8 +87,6 @@ private:
                 memory_pool[i].state = MemoryPoolStates::Attached;
             else if (mem_pool_info[i].pool_state == MemoryPoolStates::RequestDetach)
                 memory_pool[i].state = MemoryPoolStates::Detached;
-            else
-                memory_pool[i].state = mem_pool_info[i].pool_state;
         }
         std::memcpy(output.data() + sizeof(UpdateDataHeader), memory_pool.data(),
                     response_data.memory_pools_size);
@@ -183,7 +181,9 @@ private:
             behavior_size = 0xb0;
             memory_pools_size = (config.effect_count + (config.voice_count * 4)) * 0x10;
             voices_size = config.voice_count * 0x10;
+            voice_resource_size = 0x0;
             effects_size = config.effect_count * 0x10;
+            mixes_size = 0x0;
             sinks_size = config.sink_count * 0x20;
             performance_manager_size = 0x10;
             total_size = sizeof(UpdateDataHeader) + behavior_size + memory_pools_size +

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

@@ -2,17 +2,17 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#include "boost/container/flat_map.hpp"
+#pragma optimize("", off)
+
 #include "common/assert.h"
-#include "common/common_paths.h"
 #include "common/file_util.h"
 #include "core/core.h"
 #include "core/file_sys/errors.h"
-#include "core/file_sys/filesystem.h"
+#include "core/file_sys/savedata_factory.h"
+#include "core/file_sys/sdmc_factory.h"
 #include "core/file_sys/vfs.h"
 #include "core/file_sys/vfs_offset.h"
 #include "core/file_sys/vfs_real.h"
-#include "core/hle/kernel/process.h"
 #include "core/hle/service/filesystem/filesystem.h"
 #include "core/hle/service/filesystem/fsp_srv.h"
 
@@ -41,17 +41,29 @@ std::string VfsDirectoryServiceWrapper::GetName() const {
 ResultCode VfsDirectoryServiceWrapper::CreateFile(const std::string& path, u64 size) const {
     auto dir = GetDirectoryRelativeWrapped(backing, FileUtil::GetParentPath(path));
     auto file = dir->CreateFile(FileUtil::GetFilename(path));
-    if (file == nullptr)
+    if (file == nullptr) {
+        // TODO(DarkLordZach): Find a better error code for this
         return ResultCode(-1);
-    if (!file->Resize(size))
+    }
+    if (!file->Resize(size)) {
+        // TODO(DarkLordZach): Find a better error code for this
         return ResultCode(-1);
+    }
     return RESULT_SUCCESS;
 }
 
 ResultCode VfsDirectoryServiceWrapper::DeleteFile(const std::string& path) const {
     auto dir = GetDirectoryRelativeWrapped(backing, FileUtil::GetParentPath(path));
-    if (!backing->DeleteFile(FileUtil::GetFilename(path)))
+    if (path == "/" || path == "\\") {
+        // TODO(DarkLordZach): Why do games call this and what should it do? Works as is but...
+        return RESULT_SUCCESS;
+    }
+    if (dir->GetFile(FileUtil::GetFilename(path)) == nullptr)
+        return FileSys::ERROR_PATH_NOT_FOUND;
+    if (!backing->DeleteFile(FileUtil::GetFilename(path))) {
+        // TODO(DarkLordZach): Find a better error code for this
         return ResultCode(-1);
+    }
     return RESULT_SUCCESS;
 }
 
@@ -60,22 +72,28 @@ ResultCode VfsDirectoryServiceWrapper::CreateDirectory(const std::string& path)
     if (dir == nullptr && FileUtil::GetFilename(FileUtil::GetParentPath(path)).empty())
         dir = backing;
     auto new_dir = dir->CreateSubdirectory(FileUtil::GetFilename(path));
-    if (new_dir == nullptr)
+    if (new_dir == nullptr) {
+        // TODO(DarkLordZach): Find a better error code for this
         return ResultCode(-1);
+    }
     return RESULT_SUCCESS;
 }
 
 ResultCode VfsDirectoryServiceWrapper::DeleteDirectory(const std::string& path) const {
     auto dir = GetDirectoryRelativeWrapped(backing, FileUtil::GetParentPath(path));
-    if (!dir->DeleteSubdirectory(FileUtil::GetFilename(path)))
+    if (!dir->DeleteSubdirectory(FileUtil::GetFilename(path))) {
+        // TODO(DarkLordZach): Find a better error code for this
         return ResultCode(-1);
+    }
     return RESULT_SUCCESS;
 }
 
 ResultCode VfsDirectoryServiceWrapper::DeleteDirectoryRecursively(const std::string& path) const {
     auto dir = GetDirectoryRelativeWrapped(backing, FileUtil::GetParentPath(path));
-    if (!dir->DeleteSubdirectoryRecursive(FileUtil::GetFilename(path)))
+    if (!dir->DeleteSubdirectoryRecursive(FileUtil::GetFilename(path))) {
+        // TODO(DarkLordZach): Find a better error code for this
         return ResultCode(-1);
+    }
     return RESULT_SUCCESS;
 }
 
@@ -86,8 +104,10 @@ ResultCode VfsDirectoryServiceWrapper::RenameFile(const std::string& src_path,
         // Use more-optimized vfs implementation rename.
         if (src == nullptr)
             return FileSys::ERROR_PATH_NOT_FOUND;
-        if (!src->Rename(FileUtil::GetFilename(dest_path)))
+        if (!src->Rename(FileUtil::GetFilename(dest_path))) {
+            // TODO(DarkLordZach): Find a better error code for this
             return ResultCode(-1);
+        }
         return RESULT_SUCCESS;
     }
 
@@ -102,8 +122,10 @@ ResultCode VfsDirectoryServiceWrapper::RenameFile(const std::string& src_path,
     ASSERT_MSG(dest->WriteBytes(src->ReadAllBytes()) == src->GetSize(),
                "Could not write all of the bytes but everything else has succeded.");
 
-    if (!src->GetContainingDirectory()->DeleteFile(FileUtil::GetFilename(src_path)))
+    if (!src->GetContainingDirectory()->DeleteFile(FileUtil::GetFilename(src_path))) {
+        // TODO(DarkLordZach): Find a better error code for this
         return ResultCode(-1);
+    }
 
     return RESULT_SUCCESS;
 }
@@ -115,8 +137,10 @@ ResultCode VfsDirectoryServiceWrapper::RenameDirectory(const std::string& src_pa
         // Use more-optimized vfs implementation rename.
         if (src == nullptr)
             return FileSys::ERROR_PATH_NOT_FOUND;
-        if (!src->Rename(FileUtil::GetFilename(dest_path)))
+        if (!src->Rename(FileUtil::GetFilename(dest_path))) {
+            // TODO(DarkLordZach): Find a better error code for this
             return ResultCode(-1);
+        }
         return RESULT_SUCCESS;
     }
 
@@ -126,16 +150,20 @@ ResultCode VfsDirectoryServiceWrapper::RenameDirectory(const std::string& src_pa
                "don't match -- UNIMPLEMENTED",
                src_path, dest_path);
 
+    // TODO(DarkLordZach): Find a better error code for this
     return ResultCode(-1);
 }
 
 ResultVal<FileSys::VirtualFile> VfsDirectoryServiceWrapper::OpenFile(const std::string& path,
                                                                      FileSys::Mode mode) const {
-    auto file = backing->GetFileRelative(path);
+    auto npath = path;
+    while (npath.size() > 0 && (npath[0] == '/' || npath[0] == '\\'))
+        npath = npath.substr(1);
+    auto file = backing->GetFileRelative(npath);
     if (file == nullptr)
         return FileSys::ERROR_PATH_NOT_FOUND;
 
-    if ((static_cast<u32>(mode) & static_cast<u32>(FileSys::Mode::Append)) != 0) {
+    if (mode == FileSys::Mode::Append) {
         return MakeResult<FileSys::VirtualFile>(
             std::make_shared<FileSys::OffsetVfsFile>(file, 0, file->GetSize()));
     }
@@ -145,8 +173,10 @@ ResultVal<FileSys::VirtualFile> VfsDirectoryServiceWrapper::OpenFile(const std::
 
 ResultVal<FileSys::VirtualDir> VfsDirectoryServiceWrapper::OpenDirectory(const std::string& path) {
     auto dir = GetDirectoryRelativeWrapped(backing, path);
-    if (dir == nullptr)
+    if (dir == nullptr) {
+        // TODO(DarkLordZach): Find a better error code for this
         return ResultCode(-1);
+    }
     return MakeResult(dir);
 }
 
@@ -161,103 +191,92 @@ ResultVal<FileSys::EntryType> VfsDirectoryServiceWrapper::GetEntryType(
     const std::string& path) const {
     auto dir = GetDirectoryRelativeWrapped(backing, FileUtil::GetParentPath(path));
     if (dir == nullptr)
-        return ResultCode(-1);
+        return FileSys::ERROR_PATH_NOT_FOUND;
     auto filename = FileUtil::GetFilename(path);
     if (dir->GetFile(filename) != nullptr)
         return MakeResult(FileSys::EntryType::File);
     if (dir->GetSubdirectory(filename) != nullptr)
         return MakeResult(FileSys::EntryType::Directory);
-    return ResultCode(-1);
+    return FileSys::ERROR_PATH_NOT_FOUND;
 }
 
-// A deferred filesystem for nand save data.
-// This must be deferred because the directory is dependent on title id, which is not set at
-// registration time.
-struct SaveDataDeferredFilesystem : DeferredFilesystem {
-protected:
-    FileSys::VirtualDir CreateFilesystem() override {
-        u64 title_id = Core::CurrentProcess()->program_id;
-        // TODO(DarkLordZach): Users
-        u32 user_id = 0;
-        std::string nand_directory = fmt::format(
-            "{}save/{:016X}/{:08X}/", FileUtil::GetUserPath(D_NAND_IDX), title_id, user_id);
-
-        auto savedata =
-            std::make_shared<FileSys::RealVfsDirectory>(nand_directory, FileSys::Mode::Write);
-        return savedata;
-    }
-};
-
 /**
  * Map of registered file systems, identified by type. Once an file system is registered here, it
  * is never removed until UnregisterFileSystems is called.
  */
-static boost::container::flat_map<Type, std::unique_ptr<DeferredFilesystem>> filesystem_map;
-static FileSys::VirtualFile filesystem_romfs = nullptr;
+static std::unique_ptr<FileSys::RomFSFactory> romfs_factory;
+static std::unique_ptr<FileSys::SaveDataFactory> save_data_factory;
+static std::unique_ptr<FileSys::SDMCFactory> sdmc_factory;
 
-ResultCode RegisterFileSystem(std::unique_ptr<DeferredFilesystem>&& factory, Type type) {
-    auto result = filesystem_map.emplace(type, std::move(factory));
-
-    bool inserted = result.second;
-    ASSERT_MSG(inserted, "Tried to register more than one system with same id code");
-
-    auto& filesystem = result.first->second;
-    LOG_DEBUG(Service_FS, "Registered file system with id code 0x{:08X}", static_cast<u32>(type));
+ResultCode RegisterRomFS(std::unique_ptr<FileSys::RomFSFactory>&& factory) {
+    ASSERT_MSG(romfs_factory == nullptr, "Tried to register a second RomFS");
+    romfs_factory = std::move(factory);
+    LOG_DEBUG(Service_FS, "Registered RomFS");
     return RESULT_SUCCESS;
 }
 
-ResultCode RegisterRomFS(FileSys::VirtualFile filesystem) {
-    ASSERT_MSG(filesystem_romfs == nullptr,
-               "Tried to register more than one system with same id code");
-
-    filesystem_romfs = filesystem;
-    LOG_DEBUG(Service_FS, "Registered file system {} with id code 0x{:08X}", filesystem->GetName(),
-              static_cast<u32>(Type::RomFS));
+ResultCode RegisterSaveData(std::unique_ptr<FileSys::SaveDataFactory>&& factory) {
+    ASSERT_MSG(romfs_factory == nullptr, "Tried to register a second save data");
+    save_data_factory = std::move(factory);
+    LOG_DEBUG(Service_FS, "Registered save data");
     return RESULT_SUCCESS;
 }
 
-ResultVal<FileSys::VirtualDir> OpenFileSystem(Type type) {
-    LOG_TRACE(Service_FS, "Opening FileSystem with type={}", static_cast<u32>(type));
+ResultCode RegisterSDMC(std::unique_ptr<FileSys::SDMCFactory>&& factory) {
+    ASSERT_MSG(sdmc_factory == nullptr, "Tried to register a second SDMC");
+    sdmc_factory = std::move(factory);
+    LOG_DEBUG(Service_FS, "Registered SDMC");
+    return RESULT_SUCCESS;
+}
 
-    auto itr = filesystem_map.find(type);
-    if (itr == filesystem_map.end()) {
+ResultVal<FileSys::VirtualFile> OpenRomFS(u64 title_id) {
+    LOG_TRACE(Service_FS, "Opening RomFS for title_id={:016X}", title_id);
+
+    if (romfs_factory == nullptr) {
         // TODO(bunnei): Find a better error code for this
         return ResultCode(-1);
     }
 
-    return MakeResult(itr->second->Get());
+    return romfs_factory->Open(title_id);
 }
 
-ResultVal<FileSys::VirtualFile> OpenRomFS() {
-    if (filesystem_romfs == nullptr)
-        return ResultCode(-1);
-    return MakeResult(filesystem_romfs);
-}
+ResultVal<FileSys::VirtualDir> OpenSaveData(FileSys::SaveDataSpaceId space,
+                                            FileSys::SaveDataDescriptor save_struct) {
+    LOG_TRACE(Service_FS, "Opening Save Data for space_id={:01X}, save_struct={}",
+              static_cast<u8>(space), save_struct.DebugInfo());
 
-ResultCode FormatFileSystem(Type type) {
-    LOG_TRACE(Service_FS, "Formatting FileSystem with type={}", static_cast<u32>(type));
-
-    auto itr = filesystem_map.find(type);
-    if (itr == filesystem_map.end()) {
-        // TODO(bunnei): Find a better error code for this
-        return ResultCode(-1);
+    if (save_data_factory == nullptr) {
+        return ResultCode(ErrorModule::FS, FileSys::ErrCodes::TitleNotFound);
     }
 
-    return itr->second->Get()->GetParentDirectory()->DeleteSubdirectory(
-               itr->second->Get()->GetName())
-               ? RESULT_SUCCESS
-               : ResultCode(-1);
+    return save_data_factory->Open(space, save_struct);
+}
+
+ResultVal<FileSys::VirtualDir> OpenSDMC() {
+    LOG_TRACE(Service_FS, "Opening SDMC");
+
+    if (sdmc_factory == nullptr) {
+        return ResultCode(ErrorModule::FS, FileSys::ErrCodes::SdCardNotFound);
+    }
+
+    return sdmc_factory->Open();
 }
 
 void RegisterFileSystems() {
-    filesystem_map.clear();
-    filesystem_romfs = nullptr;
+    romfs_factory = nullptr;
+    save_data_factory = nullptr;
+    sdmc_factory = nullptr;
 
-    std::string sd_directory = FileUtil::GetUserPath(D_SDMC_IDX);
-    auto sdcard = std::make_shared<FileSys::RealVfsDirectory>(sd_directory, FileSys::Mode::Write);
-    RegisterFileSystem(std::make_unique<DeferredFilesystem>(sdcard), Type::SDMC);
+    auto nand_directory = std::make_shared<FileSys::RealVfsDirectory>(
+        FileUtil::GetUserPath(D_NAND_IDX), FileSys::Mode::Write);
+    auto sd_directory = std::make_shared<FileSys::RealVfsDirectory>(
+        FileUtil::GetUserPath(D_SDMC_IDX), FileSys::Mode::Write);
 
-    RegisterFileSystem(std::make_unique<SaveDataDeferredFilesystem>(), Type::SaveData);
+    auto savedata = std::make_unique<FileSys::SaveDataFactory>(std::move(nand_directory));
+    save_data_factory = std::move(savedata);
+
+    auto sdcard = std::make_unique<FileSys::SDMCFactory>(std::move(sd_directory));
+    sdmc_factory = std::move(sdcard);
 }
 
 void InstallInterfaces(SM::ServiceManager& service_manager) {

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

@@ -6,16 +6,13 @@
 
 #include <memory>
 #include "common/common_types.h"
-#include "core/file_sys/filesystem.h"
-#include "core/file_sys/vfs.h"
+#include "core/file_sys/directory.h"
+#include "core/file_sys/mode.h"
+#include "core/file_sys/romfs_factory.h"
+#include "core/file_sys/savedata_factory.h"
+#include "core/file_sys/sdmc_factory.h"
 #include "core/hle/result.h"
 
-namespace FileSys {
-class FileSystemBackend;
-class FileSystemFactory;
-class Path;
-} // namespace FileSys
-
 namespace Service {
 
 namespace SM {
@@ -24,12 +21,22 @@ class ServiceManager;
 
 namespace FileSystem {
 
-/// Supported FileSystem types
-enum class Type {
-    RomFS = 1,
-    SaveData = 2,
-    SDMC = 3,
-};
+ResultCode RegisterRomFS(std::unique_ptr<FileSys::RomFSFactory>&& factory);
+ResultCode RegisterSaveData(std::unique_ptr<FileSys::SaveDataFactory>&& factory);
+ResultCode RegisterSDMC(std::unique_ptr<FileSys::SDMCFactory>&& factory);
+
+// TODO(DarkLordZach): BIS Filesystem
+// ResultCode RegisterBIS(std::unique_ptr<FileSys::BISFactory>&& factory);
+ResultVal<FileSys::VirtualFile> OpenRomFS(u64 title_id);
+ResultVal<FileSys::VirtualDir> OpenSaveData(FileSys::SaveDataSpaceId space,
+                                            FileSys::SaveDataDescriptor save_struct);
+ResultVal<FileSys::VirtualDir> OpenSDMC();
+
+// TODO(DarkLordZach): BIS Filesystem
+// ResultVal<std::unique_ptr<FileSys::FileSystemBackend>> OpenBIS();
+
+/// Registers all Filesystem services with the specified service manager.
+void InstallInterfaces(SM::ServiceManager& service_manager);
 
 // A class that wraps a VfsDirectory with methods that return ResultVal and ResultCode instead of
 // pointers and booleans. This makes using a VfsDirectory with switch services much easier and
@@ -126,61 +133,5 @@ private:
     FileSys::VirtualDir backing;
 };
 
-// A class that deferres the creation of a filesystem until a later time.
-// This is useful if construction depends on a variable not known when the filesystem is registered.
-// Construct this with a filesystem (VirtualDir) to avoid the deferrence feature or override the
-//     CreateFilesystem method which will be called on first use.
-struct DeferredFilesystem {
-    DeferredFilesystem() = default;
-
-    explicit DeferredFilesystem(FileSys::VirtualDir vfs_directory) : fs(std::move(vfs_directory)) {}
-
-    FileSys::VirtualDir Get() {
-        if (fs == nullptr)
-            fs = CreateFilesystem();
-
-        return fs;
-    }
-
-    virtual ~DeferredFilesystem() = default;
-
-protected:
-    virtual FileSys::VirtualDir CreateFilesystem() {
-        return fs;
-    }
-
-private:
-    FileSys::VirtualDir fs;
-};
-
-/**
- * Registers a FileSystem, instances of which can later be opened using its IdCode.
- * @param factory FileSystem backend interface to use
- * @param type Type used to access this type of FileSystem
- */
-ResultCode RegisterFileSystem(std::unique_ptr<DeferredFilesystem>&& fs, Type type);
-
-ResultCode RegisterRomFS(FileSys::VirtualFile fs);
-
-/**
- * Opens a file system
- * @param type Type of the file system to open
- * @param path Path to the file system, used with Binary paths
- * @return FileSys::FileSystemBackend interface to the file system
- */
-ResultVal<FileSys::VirtualDir> OpenFileSystem(Type type);
-
-ResultVal<FileSys::VirtualFile> OpenRomFS();
-
-/**
- * Formats a file system
- * @param type Type of the file system to format
- * @return ResultCode of the operation
- */
-ResultCode FormatFileSystem(Type type);
-
-/// Registers all Filesystem services with the specified service manager.
-void InstallInterfaces(SM::ServiceManager& service_manager);
-
 } // namespace FileSystem
 } // namespace Service

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

@@ -7,16 +7,23 @@
 #include "common/string_util.h"
 #include "core/core.h"
 #include "core/file_sys/directory.h"
-#include "core/file_sys/filesystem.h"
-#include "core/file_sys/storage.h"
+#include "core/file_sys/errors.h"
 #include "core/hle/ipc_helpers.h"
-#include "core/hle/kernel/client_port.h"
-#include "core/hle/kernel/client_session.h"
+#include "core/hle/kernel/process.h"
 #include "core/hle/service/filesystem/filesystem.h"
 #include "core/hle/service/filesystem/fsp_srv.h"
 
 namespace Service::FileSystem {
 
+enum class StorageId : u8 {
+    None = 0,
+    Host = 1,
+    GameCard = 2,
+    NandSystem = 3,
+    NandUser = 4,
+    SdCard = 5,
+};
+
 class IStorage final : public ServiceFramework<IStorage> {
 public:
     IStorage(FileSys::VirtualFile backend_)
@@ -52,13 +59,6 @@ private:
 
         // Read the data from the Storage backend
         std::vector<u8> output = backend->ReadBytes(length, offset);
-        auto res = MakeResult<size_t>(output.size());
-        if (res.Failed()) {
-            IPC::ResponseBuilder rb{ctx, 2};
-            rb.Push(res.Code());
-            return;
-        }
-
         // Write the data to memory
         ctx.WriteBuffer(output);
 
@@ -104,19 +104,13 @@ private:
 
         // Read the data from the Storage backend
         std::vector<u8> output = backend->ReadBytes(length, offset);
-        auto res = MakeResult<size_t>(output.size());
-        if (res.Failed()) {
-            IPC::ResponseBuilder rb{ctx, 2};
-            rb.Push(res.Code());
-            return;
-        }
 
         // Write the data to memory
         ctx.WriteBuffer(output);
 
         IPC::ResponseBuilder rb{ctx, 4};
         rb.Push(RESULT_SUCCESS);
-        rb.Push(static_cast<u64>(*res));
+        rb.Push(static_cast<u64>(output.size()));
     }
 
     void Write(Kernel::HLERequestContext& ctx) {
@@ -140,14 +134,20 @@ private:
         }
 
         std::vector<u8> data = ctx.ReadBuffer();
-        data.resize(length);
+        std::vector<u8> actual_data(length);
+
+        ASSERT_MSG(
+            data.size() <= length,
+            "Attempting to write more data than requested (requested={:016X}, actual={:016X}).",
+            length, data.size());
+
+        std::copy(data.begin(), data.end(), actual_data.begin());
         // Write the data to the Storage backend
-        auto res = MakeResult<size_t>(backend->WriteBytes(data, offset));
-        if (res.Failed()) {
-            IPC::ResponseBuilder rb{ctx, 2};
-            rb.Push(res.Code());
-            return;
-        }
+        auto written = backend->WriteBytes(data, offset);
+
+        ASSERT_MSG(written == length,
+                   "Could not write all bytes to file (requested={:016X}, actual={:016X}).", length,
+                   written);
 
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
@@ -205,6 +205,7 @@ public:
         };
         RegisterHandlers(functions);
 
+        // TODO(DarkLordZach): Verify that this is the correct behavior.
         // Build entry index now to save time later.
         BuildEntryIndex(entries, backend->GetFiles(), FileSys::File);
         BuildEntryIndex(entries, backend->GetSubdirectories(), FileSys::Directory);
@@ -512,16 +513,6 @@ FSP_SRV::FSP_SRV() : ServiceFramework("fsp-srv") {
     RegisterHandlers(functions);
 }
 
-void FSP_SRV::TryLoadRomFS() {
-    if (romfs) {
-        return;
-    }
-    auto res = OpenRomFS();
-    if (res.Succeeded()) {
-        romfs = std::move(res.Unwrap());
-    }
-}
-
 void FSP_SRV::Initialize(Kernel::HLERequestContext& ctx) {
     LOG_WARNING(Service_FS, "(STUBBED) called");
 
@@ -532,7 +523,7 @@ void FSP_SRV::Initialize(Kernel::HLERequestContext& ctx) {
 void FSP_SRV::MountSdCard(Kernel::HLERequestContext& ctx) {
     LOG_DEBUG(Service_FS, "called");
 
-    IFileSystem filesystem(OpenFileSystem(Type::SDMC).Unwrap());
+    IFileSystem filesystem(OpenSDMC().Unwrap());
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);
@@ -542,20 +533,34 @@ void FSP_SRV::MountSdCard(Kernel::HLERequestContext& ctx) {
 void FSP_SRV::CreateSaveData(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
 
-    auto save_struct = rp.PopRaw<std::array<u8, 0x40>>();
+    auto save_struct = rp.PopRaw<FileSys::SaveDataDescriptor>();
     auto save_create_struct = rp.PopRaw<std::array<u8, 0x40>>();
     u128 uid = rp.PopRaw<u128>();
 
-    LOG_WARNING(Service_FS, "(STUBBED) called uid = {:016X}{:016X}", uid[1], uid[0]);
+    LOG_WARNING(Service_FS, "(STUBBED) called save_struct = {}, uid = {:016X}{:016X}",
+                save_struct.DebugInfo(), uid[1], uid[0]);
 
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);
 }
 
 void FSP_SRV::MountSaveData(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_FS, "(STUBBED) called");
+    IPC::RequestParser rp{ctx};
 
-    IFileSystem filesystem(OpenFileSystem(Type::SaveData).Unwrap());
+    auto space_id = rp.PopRaw<FileSys::SaveDataSpaceId>();
+    auto unk = rp.Pop<u32>();
+    LOG_INFO(Service_FS, "called with unknown={:08X}", unk);
+    auto save_struct = rp.PopRaw<FileSys::SaveDataDescriptor>();
+
+    auto dir = OpenSaveData(space_id, save_struct);
+
+    if (dir.Failed()) {
+        IPC::ResponseBuilder rb{ctx, 2, 0, 0};
+        rb.Push(ResultCode(ErrorModule::FS, FileSys::ErrCodes::TitleNotFound));
+        return;
+    }
+
+    IFileSystem filesystem(std::move(dir.Unwrap()));
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);
@@ -573,8 +578,8 @@ void FSP_SRV::GetGlobalAccessLogMode(Kernel::HLERequestContext& ctx) {
 void FSP_SRV::OpenDataStorageByCurrentProcess(Kernel::HLERequestContext& ctx) {
     LOG_DEBUG(Service_FS, "called");
 
-    TryLoadRomFS();
-    if (!romfs) {
+    auto romfs = OpenRomFS(Core::System::GetInstance().CurrentProcess()->program_id);
+    if (romfs.Failed()) {
         // TODO (bunnei): Find the right error code to use here
         LOG_CRITICAL(Service_FS, "no file system interface available!");
         IPC::ResponseBuilder rb{ctx, 2};
@@ -582,7 +587,7 @@ void FSP_SRV::OpenDataStorageByCurrentProcess(Kernel::HLERequestContext& ctx) {
         return;
     }
 
-    IStorage storage(romfs);
+    IStorage storage(std::move(romfs.Unwrap()));
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);
@@ -590,8 +595,16 @@ void FSP_SRV::OpenDataStorageByCurrentProcess(Kernel::HLERequestContext& ctx) {
 }
 
 void FSP_SRV::OpenRomStorage(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_FS, "(STUBBED) called, using OpenDataStorageByCurrentProcess");
-    OpenDataStorageByCurrentProcess(ctx);
+    IPC::RequestParser rp{ctx};
+
+    auto storage_id = rp.PopRaw<StorageId>();
+    auto title_id = rp.PopRaw<u64>();
+
+    LOG_DEBUG(Service_FS, "called with storage_id={:02X}, title_id={:016X}",
+              static_cast<u8>(storage_id), title_id);
+
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(ResultCode(ErrorModule::FS, FileSys::ErrCodes::TitleNotFound));
 }
 
 } // namespace Service::FileSystem

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

@@ -19,8 +19,6 @@ public:
     ~FSP_SRV() = default;
 
 private:
-    void TryLoadRomFS();
-
     void Initialize(Kernel::HLERequestContext& ctx);
     void MountSdCard(Kernel::HLERequestContext& ctx);
     void CreateSaveData(Kernel::HLERequestContext& ctx);

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

@@ -5,6 +5,7 @@
 #include <atomic>
 #include "common/logging/log.h"
 #include "core/core_timing.h"
+#include "core/frontend/emu_window.h"
 #include "core/frontend/input.h"
 #include "core/hle/ipc_helpers.h"
 #include "core/hle/kernel/client_port.h"
@@ -18,9 +19,9 @@ namespace Service::HID {
 
 // Updating period for each HID device.
 // TODO(shinyquagsire23): These need better values.
-constexpr u64 pad_update_ticks = CoreTiming::BASE_CLOCK_RATE / 10000;
-constexpr u64 accelerometer_update_ticks = CoreTiming::BASE_CLOCK_RATE / 10000;
-constexpr u64 gyroscope_update_ticks = CoreTiming::BASE_CLOCK_RATE / 10000;
+constexpr u64 pad_update_ticks = CoreTiming::BASE_CLOCK_RATE / 100;
+constexpr u64 accelerometer_update_ticks = CoreTiming::BASE_CLOCK_RATE / 100;
+constexpr u64 gyroscope_update_ticks = CoreTiming::BASE_CLOCK_RATE / 100;
 
 class IAppletResource final : public ServiceFramework<IAppletResource> {
 public:
@@ -63,7 +64,8 @@ private:
         std::transform(Settings::values.analogs.begin() + Settings::NativeAnalog::STICK_HID_BEGIN,
                        Settings::values.analogs.begin() + Settings::NativeAnalog::STICK_HID_END,
                        sticks.begin(), Input::CreateDevice<Input::AnalogDevice>);
-        // TODO(shinyquagsire23): gyro, touch, mouse, keyboard
+        touch_device = Input::CreateDevice<Input::TouchDevice>(Settings::values.touch_device);
+        // TODO(shinyquagsire23): gyro, mouse, keyboard
     }
 
     void UpdatePadCallback(u64 userdata, int cycles_late) {
@@ -75,7 +77,7 @@ private:
 
         // Set up controllers as neon red+blue Joy-Con attached to console
         ControllerHeader& controller_header = mem.controllers[Controller_Handheld].header;
-        controller_header.type = ControllerType_Handheld | ControllerType_JoyconPair;
+        controller_header.type = ControllerType_Handheld;
         controller_header.single_colors_descriptor = ColorDesc_ColorsNonexistent;
         controller_header.right_color_body = JOYCON_BODY_NEON_RED;
         controller_header.right_color_buttons = JOYCON_BUTTONS_NEON_RED;
@@ -84,23 +86,21 @@ private:
 
         for (size_t controller = 0; controller < mem.controllers.size(); controller++) {
             for (int index = 0; index < HID_NUM_LAYOUTS; index++) {
-                // TODO(DarkLordZach): Is this layout/controller config actually invalid?
-                if (controller == Controller_Handheld && index == Layout_Single)
-                    continue;
-
                 ControllerLayout& layout = mem.controllers[controller].layouts[index];
                 layout.header.num_entries = HID_NUM_ENTRIES;
                 layout.header.max_entry_index = HID_NUM_ENTRIES - 1;
 
                 // HID shared memory stores the state of the past 17 samples in a circlular buffer,
                 // each with a timestamp in number of samples since boot.
+                const ControllerInputEntry& last_entry = layout.entries[layout.header.latest_entry];
+
                 layout.header.timestamp_ticks = CoreTiming::GetTicks();
                 layout.header.latest_entry = (layout.header.latest_entry + 1) % HID_NUM_ENTRIES;
 
                 ControllerInputEntry& entry = layout.entries[layout.header.latest_entry];
-                entry.timestamp++;
+                entry.timestamp = last_entry.timestamp + 1;
                 // TODO(shinyquagsire23): Is this always identical to timestamp?
-                entry.timestamp_2++;
+                entry.timestamp_2 = entry.timestamp;
 
                 // TODO(shinyquagsire23): More than just handheld input
                 if (controller != Controller_Handheld)
@@ -153,8 +153,6 @@ private:
             }
         }
 
-        // TODO(bunnei): Properly implement the touch screen, the below will just write empty data
-
         TouchScreen& touchscreen = mem.touchscreen;
         const u64 last_entry = touchscreen.header.latest_entry;
         const u64 curr_entry = (last_entry + 1) % touchscreen.entries.size();
@@ -166,7 +164,26 @@ private:
         touchscreen.header.max_entry_index = touchscreen.entries.size();
         touchscreen.header.timestamp = timestamp;
         touchscreen.entries[curr_entry].header.timestamp = sample_counter;
-        touchscreen.entries[curr_entry].header.num_touches = 0;
+
+        TouchScreenEntryTouch touch_entry{};
+        auto [x, y, pressed] = touch_device->GetStatus();
+        touch_entry.timestamp = timestamp;
+        touch_entry.x = static_cast<u16>(x * Layout::ScreenUndocked::Width);
+        touch_entry.y = static_cast<u16>(y * Layout::ScreenUndocked::Height);
+        touch_entry.touch_index = 0;
+
+        // TODO(DarkLordZach): Maybe try to derive these from EmuWindow?
+        touch_entry.diameter_x = 15;
+        touch_entry.diameter_y = 15;
+        touch_entry.angle = 0;
+
+        // TODO(DarkLordZach): Implement multi-touch support
+        if (pressed) {
+            touchscreen.entries[curr_entry].header.num_touches = 1;
+            touchscreen.entries[curr_entry].touches[0] = touch_entry;
+        } else {
+            touchscreen.entries[curr_entry].header.num_touches = 0;
+        }
 
         // TODO(shinyquagsire23): Properly implement mouse
         Mouse& mouse = mem.mouse;
@@ -252,6 +269,7 @@ private:
     std::array<std::unique_ptr<Input::ButtonDevice>, Settings::NativeButton::NUM_BUTTONS_HID>
         buttons;
     std::array<std::unique_ptr<Input::AnalogDevice>, Settings::NativeAnalog::NUM_STICKS_HID> sticks;
+    std::unique_ptr<Input::TouchDevice> touch_device;
 };
 
 class IActiveVibrationDeviceList final : public ServiceFramework<IActiveVibrationDeviceList> {

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

@@ -148,6 +148,24 @@ private:
 
         LOG_DEBUG(Service_NIFM, "called");
     }
+    void IsWirelessCommunicationEnabled(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_NIFM, "(STUBBED) called");
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(RESULT_SUCCESS);
+        rb.Push<u8>(0);
+    }
+    void IsEthernetCommunicationEnabled(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_NIFM, "(STUBBED) called");
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(RESULT_SUCCESS);
+        rb.Push<u8>(0);
+    }
+    void IsAnyInternetRequestAccepted(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_NIFM, "(STUBBED) called");
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(RESULT_SUCCESS);
+        rb.Push<u8>(0);
+    }
 };
 
 IGeneralService::IGeneralService() : ServiceFramework("IGeneralService") {
@@ -167,11 +185,11 @@ IGeneralService::IGeneralService() : ServiceFramework("IGeneralService") {
         {14, &IGeneralService::CreateTemporaryNetworkProfile, "CreateTemporaryNetworkProfile"},
         {15, nullptr, "GetCurrentIpConfigInfo"},
         {16, nullptr, "SetWirelessCommunicationEnabled"},
-        {17, nullptr, "IsWirelessCommunicationEnabled"},
+        {17, &IGeneralService::IsWirelessCommunicationEnabled, "IsWirelessCommunicationEnabled"},
         {18, nullptr, "GetInternetConnectionStatus"},
         {19, nullptr, "SetEthernetCommunicationEnabled"},
-        {20, nullptr, "IsEthernetCommunicationEnabled"},
-        {21, nullptr, "IsAnyInternetRequestAccepted"},
+        {20, &IGeneralService::IsEthernetCommunicationEnabled, "IsEthernetCommunicationEnabled"},
+        {21, &IGeneralService::IsAnyInternetRequestAccepted, "IsAnyInternetRequestAccepted"},
         {22, nullptr, "IsAnyForegroundRequestAccepted"},
         {23, nullptr, "PutToSleep"},
         {24, nullptr, "WakeUp"},

src/core/hle/service/nvdrv/devices/nvdisp_disp0.cpp

@@ -18,7 +18,8 @@ u32 nvdisp_disp0::ioctl(Ioctl command, const std::vector<u8>& input, std::vector
 }
 
 void nvdisp_disp0::flip(u32 buffer_handle, u32 offset, u32 format, u32 width, u32 height,
-                        u32 stride, NVFlinger::BufferQueue::BufferTransformFlags transform) {
+                        u32 stride, NVFlinger::BufferQueue::BufferTransformFlags transform,
+                        const MathUtil::Rectangle<int>& crop_rect) {
     VAddr addr = nvmap_dev->GetObjectAddress(buffer_handle);
     LOG_WARNING(Service,
                 "Drawing from address {:X} offset {:08X} Width {} Height {} Stride {} Format {}",
@@ -26,7 +27,8 @@ void nvdisp_disp0::flip(u32 buffer_handle, u32 offset, u32 format, u32 width, u3
 
     using PixelFormat = Tegra::FramebufferConfig::PixelFormat;
     const Tegra::FramebufferConfig framebuffer{
-        addr, offset, width, height, stride, static_cast<PixelFormat>(format), transform};
+        addr,      offset,   width, height, stride, static_cast<PixelFormat>(format),
+        transform, crop_rect};
 
     Core::System::GetInstance().perf_stats.EndGameFrame();
 

src/core/hle/service/nvdrv/devices/nvdisp_disp0.h

@@ -7,6 +7,7 @@
 #include <memory>
 #include <vector>
 #include "common/common_types.h"
+#include "common/math_util.h"
 #include "core/hle/service/nvdrv/devices/nvdevice.h"
 #include "core/hle/service/nvflinger/buffer_queue.h"
 
@@ -23,7 +24,8 @@ public:
 
     /// Performs a screen flip, drawing the buffer pointed to by the handle.
     void flip(u32 buffer_handle, u32 offset, u32 format, u32 width, u32 height, u32 stride,
-              NVFlinger::BufferQueue::BufferTransformFlags transform);
+              NVFlinger::BufferQueue::BufferTransformFlags transform,
+              const MathUtil::Rectangle<int>& crop_rect);
 
 private:
     std::shared_ptr<nvmap> nvmap_dev;

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

@@ -29,24 +29,9 @@ u32 nvhost_ctrl::ioctl(Ioctl command, const std::vector<u8>& input, std::vector<
 u32 nvhost_ctrl::NvOsGetConfigU32(const std::vector<u8>& input, std::vector<u8>& output) {
     IocGetConfigParams params{};
     std::memcpy(&params, input.data(), sizeof(params));
-    LOG_DEBUG(Service_NVDRV, "called, setting={}!{}", params.domain_str.data(),
+    LOG_TRACE(Service_NVDRV, "called, setting={}!{}", params.domain_str.data(),
               params.param_str.data());
-
-    if (!strcmp(params.domain_str.data(), "nv")) {
-        if (!strcmp(params.param_str.data(), "NV_MEMORY_PROFILER")) {
-            params.config_str[0] = '0';
-        } else if (!strcmp(params.param_str.data(), "NVN_THROUGH_OPENGL")) {
-            params.config_str[0] = '0';
-        } else if (!strcmp(params.param_str.data(), "NVRM_GPU_PREVENT_USE")) {
-            params.config_str[0] = '0';
-        } else {
-            params.config_str[0] = '\0';
-        }
-    } else {
-        UNIMPLEMENTED(); // unknown domain? Only nv has been seen so far on hardware
-    }
-    std::memcpy(output.data(), &params, sizeof(params));
-    return 0;
+    return 0x30006; // Returns error on production mode
 }
 
 u32 nvhost_ctrl::IocCtrlEventWait(const std::vector<u8>& input, std::vector<u8>& output,

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

@@ -13,8 +13,8 @@ namespace Service {
 namespace NVFlinger {
 
 BufferQueue::BufferQueue(u32 id, u64 layer_id) : id(id), layer_id(layer_id) {
-    native_handle = Kernel::Event::Create(Kernel::ResetType::OneShot, "BufferQueue NativeHandle");
-    native_handle->Signal();
+    buffer_wait_event =
+        Kernel::Event::Create(Kernel::ResetType::Sticky, "BufferQueue NativeHandle");
 }
 
 void BufferQueue::SetPreallocatedBuffer(u32 slot, IGBPBuffer& igbp_buffer) {
@@ -26,10 +26,7 @@ void BufferQueue::SetPreallocatedBuffer(u32 slot, IGBPBuffer& igbp_buffer) {
     LOG_WARNING(Service, "Adding graphics buffer {}", slot);
 
     queue.emplace_back(buffer);
-
-    if (buffer_wait_event) {
-        buffer_wait_event->Signal();
-    }
+    buffer_wait_event->Signal();
 }
 
 boost::optional<u32> BufferQueue::DequeueBuffer(u32 width, u32 height) {
@@ -48,8 +45,6 @@ boost::optional<u32> BufferQueue::DequeueBuffer(u32 width, u32 height) {
         return boost::none;
     }
 
-    buffer_wait_event = nullptr;
-
     itr->status = Buffer::Status::Dequeued;
     return itr->slot;
 }
@@ -62,13 +57,15 @@ const IGBPBuffer& BufferQueue::RequestBuffer(u32 slot) const {
     return itr->igbp_buffer;
 }
 
-void BufferQueue::QueueBuffer(u32 slot, BufferTransformFlags transform) {
+void BufferQueue::QueueBuffer(u32 slot, BufferTransformFlags transform,
+                              const MathUtil::Rectangle<int>& crop_rect) {
     auto itr = std::find_if(queue.begin(), queue.end(),
                             [&](const Buffer& buffer) { return buffer.slot == slot; });
     ASSERT(itr != queue.end());
     ASSERT(itr->status == Buffer::Status::Dequeued);
     itr->status = Buffer::Status::Queued;
     itr->transform = transform;
+    itr->crop_rect = crop_rect;
 }
 
 boost::optional<const BufferQueue::Buffer&> BufferQueue::AcquireBuffer() {
@@ -88,9 +85,7 @@ void BufferQueue::ReleaseBuffer(u32 slot) {
     ASSERT(itr->status == Buffer::Status::Acquired);
     itr->status = Buffer::Status::Free;
 
-    if (buffer_wait_event) {
-        buffer_wait_event->Signal();
-    }
+    buffer_wait_event->Signal();
 }
 
 u32 BufferQueue::Query(QueryType type) {
@@ -106,10 +101,5 @@ u32 BufferQueue::Query(QueryType type) {
     return 0;
 }
 
-void BufferQueue::SetBufferWaitEvent(Kernel::SharedPtr<Kernel::Event>&& wait_event) {
-    ASSERT_MSG(!buffer_wait_event, "buffer_wait_event only supports a single waiting thread!");
-    buffer_wait_event = std::move(wait_event);
-}
-
 } // namespace NVFlinger
 } // namespace Service

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

@@ -6,6 +6,7 @@
 
 #include <vector>
 #include <boost/optional.hpp>
+#include "common/math_util.h"
 #include "common/swap.h"
 #include "core/hle/kernel/event.h"
 
@@ -68,23 +69,24 @@ public:
         Status status = Status::Free;
         IGBPBuffer igbp_buffer;
         BufferTransformFlags transform;
+        MathUtil::Rectangle<int> crop_rect;
     };
 
     void SetPreallocatedBuffer(u32 slot, IGBPBuffer& buffer);
     boost::optional<u32> DequeueBuffer(u32 width, u32 height);
     const IGBPBuffer& RequestBuffer(u32 slot) const;
-    void QueueBuffer(u32 slot, BufferTransformFlags transform);
+    void QueueBuffer(u32 slot, BufferTransformFlags transform,
+                     const MathUtil::Rectangle<int>& crop_rect);
     boost::optional<const Buffer&> AcquireBuffer();
     void ReleaseBuffer(u32 slot);
     u32 Query(QueryType type);
-    void SetBufferWaitEvent(Kernel::SharedPtr<Kernel::Event>&& wait_event);
 
     u32 GetId() const {
         return id;
     }
 
-    Kernel::SharedPtr<Kernel::Event> GetNativeHandle() const {
-        return native_handle;
+    Kernel::SharedPtr<Kernel::Event> GetBufferWaitEvent() const {
+        return buffer_wait_event;
     }
 
 private:
@@ -92,9 +94,6 @@ private:
     u64 layer_id;
 
     std::vector<Buffer> queue;
-    Kernel::SharedPtr<Kernel::Event> native_handle;
-
-    /// Used to signal waiting thread when no buffers are available
     Kernel::SharedPtr<Kernel::Event> buffer_wait_event;
 };
 

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

@@ -149,12 +149,10 @@ void NVFlinger::Compose() {
         ASSERT(nvdisp);
 
         nvdisp->flip(igbp_buffer.gpu_buffer_id, igbp_buffer.offset, igbp_buffer.format,
-                     igbp_buffer.width, igbp_buffer.height, igbp_buffer.stride, buffer->transform);
+                     igbp_buffer.width, igbp_buffer.height, igbp_buffer.stride, buffer->transform,
+                     buffer->crop_rect);
 
         buffer_queue->ReleaseBuffer(buffer->slot);
-
-        // TODO(Subv): Figure out when we should actually signal this event.
-        buffer_queue->GetNativeHandle()->Signal();
     }
 }
 

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

@@ -2,6 +2,8 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#pragma once
+
 #include <memory>
 #include <string>
 #include "core/hle/kernel/event.h"

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

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

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

@@ -7,6 +7,7 @@
 #include <memory>
 #include <boost/optional.hpp>
 #include "common/alignment.h"
+#include "common/math_util.h"
 #include "common/scope_exit.h"
 #include "core/core_timing.h"
 #include "core/hle/ipc_helpers.h"
@@ -27,8 +28,8 @@ struct DisplayInfo {
     char display_name[0x40]{"Default"};
     u64 unknown_1{1};
     u64 unknown_2{1};
-    u64 width{1920};
-    u64 height{1080};
+    u64 width{1280};
+    u64 height{720};
 };
 static_assert(sizeof(DisplayInfo) == 0x60, "DisplayInfo has wrong size");
 
@@ -327,8 +328,8 @@ public:
 
 protected:
     void SerializeData() override {
-        // TODO(Subv): Figure out what this value means, writing non-zero here will make libnx try
-        // to read an IGBPBuffer object from the parcel.
+        // TODO(Subv): Figure out what this value means, writing non-zero here will make libnx
+        // try to read an IGBPBuffer object from the parcel.
         Write<u32_le>(1);
         WriteObject(buffer);
         Write<u32_le>(0);
@@ -360,8 +361,8 @@ public:
         INSERT_PADDING_WORDS(3);
         u32_le timestamp;
         s32_le is_auto_timestamp;
-        s32_le crop_left;
         s32_le crop_top;
+        s32_le crop_left;
         s32_le crop_right;
         s32_le crop_bottom;
         s32_le scaling_mode;
@@ -370,6 +371,10 @@ public:
         INSERT_PADDING_WORDS(2);
         u32_le fence_is_valid;
         std::array<Fence, 2> fences;
+
+        MathUtil::Rectangle<int> GetCropRect() const {
+            return {crop_left, crop_top, crop_right, crop_bottom};
+        }
     };
     static_assert(sizeof(Data) == 80, "ParcelData has wrong size");
 
@@ -495,7 +500,7 @@ private:
                 ctx.WriteBuffer(response.Serialize());
             } else {
                 // Wait the current thread until a buffer becomes available
-                auto wait_event = ctx.SleepClientThread(
+                ctx.SleepClientThread(
                     Kernel::GetCurrentThread(), "IHOSBinderDriver::DequeueBuffer", -1,
                     [=](Kernel::SharedPtr<Kernel::Thread> thread, Kernel::HLERequestContext& ctx,
                         ThreadWakeupReason reason) {
@@ -506,8 +511,8 @@ private:
                         ctx.WriteBuffer(response.Serialize());
                         IPC::ResponseBuilder rb{ctx, 2};
                         rb.Push(RESULT_SUCCESS);
-                    });
-                buffer_queue->SetBufferWaitEvent(std::move(wait_event));
+                    },
+                    buffer_queue->GetBufferWaitEvent());
             }
         } else if (transaction == TransactionId::RequestBuffer) {
             IGBPRequestBufferRequestParcel request{ctx.ReadBuffer()};
@@ -519,7 +524,8 @@ private:
         } else if (transaction == TransactionId::QueueBuffer) {
             IGBPQueueBufferRequestParcel request{ctx.ReadBuffer()};
 
-            buffer_queue->QueueBuffer(request.data.slot, request.data.transform);
+            buffer_queue->QueueBuffer(request.data.slot, request.data.transform,
+                                      request.data.GetCropRect());
 
             IGBPQueueBufferResponseParcel response{1280, 720};
             ctx.WriteBuffer(response.Serialize());
@@ -532,7 +538,7 @@ private:
             IGBPQueryResponseParcel response{value};
             ctx.WriteBuffer(response.Serialize());
         } else if (transaction == TransactionId::CancelBuffer) {
-            LOG_WARNING(Service_VI, "(STUBBED) called, transaction=CancelBuffer");
+            LOG_CRITICAL(Service_VI, "(STUBBED) called, transaction=CancelBuffer");
         } else {
             ASSERT_MSG(false, "Unimplemented");
         }
@@ -565,7 +571,7 @@ private:
         LOG_WARNING(Service_VI, "(STUBBED) called id={}, unknown={:08X}", id, unknown);
         IPC::ResponseBuilder rb{ctx, 2, 1};
         rb.Push(RESULT_SUCCESS);
-        rb.PushCopyObjects(buffer_queue->GetNativeHandle());
+        rb.PushCopyObjects(buffer_queue->GetBufferWaitEvent());
     }
 
     std::shared_ptr<NVFlinger::NVFlinger> nv_flinger;

src/core/loader/deconstructed_rom_directory.cpp

@@ -8,6 +8,7 @@
 #include "common/logging/log.h"
 #include "common/string_util.h"
 #include "core/file_sys/content_archive.h"
+#include "core/gdbstub/gdbstub.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/resource_limit.h"
 #include "core/hle/service/filesystem/filesystem.h"
@@ -88,6 +89,8 @@ ResultStatus AppLoader_DeconstructedRomDirectory::Load(
             next_load_addr = AppLoader_NSO::LoadModule(module_file, load_addr);
         if (next_load_addr) {
             LOG_DEBUG(Loader, "loaded module {} @ 0x{:X}", module, load_addr);
+            // Register module with GDBStub
+            GDBStub::RegisterModule(module, load_addr, next_load_addr - 1, false);
         } else {
             next_load_addr = load_addr;
         }
@@ -108,14 +111,21 @@ ResultStatus AppLoader_DeconstructedRomDirectory::Load(
             return file->GetName().find(".romfs") != std::string::npos;
         });
 
-    // TODO(DarkLordZach): Identify RomFS if its a subdirectory.
-    const auto romfs = (romfs_iter == files.end()) ? nullptr : *romfs_iter;
-
-    if (romfs != nullptr)
-        Service::FileSystem::RegisterRomFS(romfs);
+    // Register the RomFS if a ".romfs" file was found
+    if (romfs_iter != files.end() && *romfs_iter != nullptr) {
+        romfs = *romfs_iter;
+        Service::FileSystem::RegisterRomFS(std::make_unique<FileSys::RomFSFactory>(*this));
+    }
 
     is_loaded = true;
     return ResultStatus::Success;
 }
 
+ResultStatus AppLoader_DeconstructedRomDirectory::ReadRomFS(FileSys::VirtualFile& dir) {
+    if (romfs == nullptr)
+        return ResultStatus::ErrorNotUsed;
+    dir = romfs;
+    return ResultStatus::Success;
+}
+
 } // namespace Loader

src/core/loader/deconstructed_rom_directory.h

@@ -35,8 +35,11 @@ public:
 
     ResultStatus Load(Kernel::SharedPtr<Kernel::Process>& process) override;
 
+    ResultStatus ReadRomFS(FileSys::VirtualFile& dir) override;
+
 private:
     FileSys::ProgramMetadata metadata;
+    FileSys::VirtualFile romfs;
 };
 
 } // namespace Loader

src/core/loader/nca.cpp

@@ -6,9 +6,12 @@
 
 #include "common/file_util.h"
 #include "common/logging/log.h"
+#include "common/string_util.h"
+#include "common/swap.h"
 #include "core/core.h"
 #include "core/file_sys/content_archive.h"
 #include "core/file_sys/program_metadata.h"
+#include "core/gdbstub/gdbstub.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/resource_limit.h"
 #include "core/hle/service/filesystem/filesystem.h"
@@ -70,6 +73,8 @@ ResultStatus AppLoader_NCA::Load(Kernel::SharedPtr<Kernel::Process>& process) {
         next_load_addr = AppLoader_NSO::LoadModule(exefs->GetFile(module), load_addr);
         if (next_load_addr) {
             LOG_DEBUG(Loader, "loaded module {} @ 0x{:X}", module, load_addr);
+            // Register module with GDBStub
+            GDBStub::RegisterModule(module, load_addr, next_load_addr - 1, false);
         } else {
             next_load_addr = load_addr;
         }
@@ -83,12 +88,18 @@ ResultStatus AppLoader_NCA::Load(Kernel::SharedPtr<Kernel::Process>& process) {
     process->Run(Memory::PROCESS_IMAGE_VADDR, metadata.GetMainThreadPriority(),
                  metadata.GetMainThreadStackSize());
 
+    if (nca->GetRomFS() != nullptr && nca->GetRomFS()->GetSize() > 0)
+        Service::FileSystem::RegisterRomFS(std::make_unique<FileSys::RomFSFactory>(*this));
+
     is_loaded = true;
 
-    const auto romfs = nca->GetRomFS();
-    if (romfs != nullptr)
-        Service::FileSystem::RegisterRomFS(romfs);
+    return ResultStatus::Success;
+}
 
+ResultStatus AppLoader_NCA::ReadRomFS(FileSys::VirtualFile& dir) {
+    if (nca == nullptr || nca->GetRomFS() == nullptr || nca->GetRomFS()->GetSize() == 0)
+        return ResultStatus::ErrorNotUsed;
+    dir = nca->GetRomFS();
     return ResultStatus::Success;
 }
 

src/core/loader/nca.h

@@ -31,6 +31,8 @@ public:
 
     ResultStatus Load(Kernel::SharedPtr<Kernel::Process>& process) override;
 
+    ResultStatus ReadRomFS(FileSys::VirtualFile& dir) override;
+
     ~AppLoader_NCA();
 
 private:

src/core/loader/nro.cpp

@@ -9,6 +9,7 @@
 #include "common/logging/log.h"
 #include "common/swap.h"
 #include "core/core.h"
+#include "core/gdbstub/gdbstub.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/resource_limit.h"
 #include "core/loader/nro.h"
@@ -106,6 +107,9 @@ bool AppLoader_NRO::LoadNro(FileSys::VirtualFile file, VAddr load_base) {
     codeset->memory = std::make_shared<std::vector<u8>>(std::move(program_image));
     Core::CurrentProcess()->LoadModule(codeset, load_base);
 
+    // Register module with GDBStub
+    GDBStub::RegisterModule(codeset->name, load_base, load_base);
+
     return true;
 }
 

src/core/loader/nso.cpp

@@ -10,6 +10,7 @@
 #include "common/logging/log.h"
 #include "common/swap.h"
 #include "core/core.h"
+#include "core/gdbstub/gdbstub.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/resource_limit.h"
 #include "core/loader/nso.h"
@@ -146,6 +147,9 @@ VAddr AppLoader_NSO::LoadModule(FileSys::VirtualFile file, VAddr load_base) {
     codeset->memory = std::make_shared<std::vector<u8>>(std::move(program_image));
     Core::CurrentProcess()->LoadModule(codeset, load_base);
 
+    // Register module with GDBStub
+    GDBStub::RegisterModule(codeset->name, load_base, load_base);
+
     return load_base + image_size;
 }
 

src/core/memory.cpp

@@ -5,6 +5,8 @@
 #include <algorithm>
 #include <array>
 #include <cstring>
+#include <utility>
+
 #include <boost/optional.hpp>
 #include "common/assert.h"
 #include "common/common_types.h"
@@ -74,7 +76,7 @@ void MapIoRegion(PageTable& page_table, VAddr base, u64 size, MemoryHookPointer
     MapPages(page_table, base / PAGE_SIZE, size / PAGE_SIZE, nullptr, PageType::Special);
 
     auto interval = boost::icl::discrete_interval<VAddr>::closed(base, base + size - 1);
-    SpecialRegion region{SpecialRegion::Type::IODevice, mmio_handler};
+    SpecialRegion region{SpecialRegion::Type::IODevice, std::move(mmio_handler)};
     page_table.special_regions.add(std::make_pair(interval, std::set<SpecialRegion>{region}));
 }
 
@@ -89,13 +91,13 @@ void UnmapRegion(PageTable& page_table, VAddr base, u64 size) {
 
 void AddDebugHook(PageTable& page_table, VAddr base, u64 size, MemoryHookPointer hook) {
     auto interval = boost::icl::discrete_interval<VAddr>::closed(base, base + size - 1);
-    SpecialRegion region{SpecialRegion::Type::DebugHook, hook};
+    SpecialRegion region{SpecialRegion::Type::DebugHook, std::move(hook)};
     page_table.special_regions.add(std::make_pair(interval, std::set<SpecialRegion>{region}));
 }
 
 void RemoveDebugHook(PageTable& page_table, VAddr base, u64 size, MemoryHookPointer hook) {
     auto interval = boost::icl::discrete_interval<VAddr>::closed(base, base + size - 1);
-    SpecialRegion region{SpecialRegion::Type::DebugHook, hook};
+    SpecialRegion region{SpecialRegion::Type::DebugHook, std::move(hook)};
     page_table.special_regions.subtract(std::make_pair(interval, std::set<SpecialRegion>{region}));
 }
 
@@ -115,11 +117,6 @@ static std::set<MemoryHookPointer> GetSpecialHandlers(const PageTable& page_tabl
     return result;
 }
 
-static std::set<MemoryHookPointer> GetSpecialHandlers(VAddr vaddr, u64 size) {
-    const PageTable& page_table = Core::CurrentProcess()->vm_manager.page_table;
-    return GetSpecialHandlers(page_table, vaddr, size);
-}
-
 /**
  * Gets a pointer to the exact memory at the virtual address (i.e. not page aligned)
  * using a VMA from the current process
@@ -587,8 +584,6 @@ void ZeroBlock(const Kernel::Process& process, const VAddr dest_addr, const size
     size_t page_index = dest_addr >> PAGE_BITS;
     size_t page_offset = dest_addr & PAGE_MASK;
 
-    static const std::array<u8, PAGE_SIZE> zeros = {};
-
     while (remaining_size > 0) {
         const size_t copy_amount =
             std::min(static_cast<size_t>(PAGE_SIZE) - page_offset, remaining_size);

src/tests/CMakeLists.txt

@@ -3,7 +3,6 @@ add_executable(tests
     core/arm/arm_test_common.cpp
     core/arm/arm_test_common.h
     core/core_timing.cpp
-    core/file_sys/path_parser.cpp
     core/memory/memory.cpp
     glad.cpp
     tests.cpp

src/tests/core/file_sys/path_parser.cpp

@@ -1,38 +0,0 @@
-// Copyright 2016 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include <catch.hpp>
-#include "common/file_util.h"
-#include "core/file_sys/path_parser.h"
-
-namespace FileSys {
-
-TEST_CASE("PathParser", "[core][file_sys]") {
-    REQUIRE(!PathParser(Path(std::vector<u8>{})).IsValid());
-    REQUIRE(!PathParser(Path("a")).IsValid());
-    REQUIRE(!PathParser(Path("/|")).IsValid());
-    REQUIRE(PathParser(Path("/a")).IsValid());
-    REQUIRE(!PathParser(Path("/a/b/../../c/../../d")).IsValid());
-    REQUIRE(PathParser(Path("/a/b/../c/../../d")).IsValid());
-    REQUIRE(PathParser(Path("/")).IsRootDirectory());
-    REQUIRE(!PathParser(Path("/a")).IsRootDirectory());
-    REQUIRE(PathParser(Path("/a/..")).IsRootDirectory());
-}
-
-TEST_CASE("PathParser - Host file system", "[core][file_sys]") {
-    std::string test_dir = "./test";
-    FileUtil::CreateDir(test_dir);
-    FileUtil::CreateDir(test_dir + "/z");
-    FileUtil::CreateEmptyFile(test_dir + "/a");
-
-    REQUIRE(PathParser(Path("/a")).GetHostStatus(test_dir) == PathParser::FileFound);
-    REQUIRE(PathParser(Path("/b")).GetHostStatus(test_dir) == PathParser::NotFound);
-    REQUIRE(PathParser(Path("/z")).GetHostStatus(test_dir) == PathParser::DirectoryFound);
-    REQUIRE(PathParser(Path("/a/c")).GetHostStatus(test_dir) == PathParser::FileInPath);
-    REQUIRE(PathParser(Path("/b/c")).GetHostStatus(test_dir) == PathParser::PathNotFound);
-
-    FileUtil::DeleteDirRecursively(test_dir);
-}
-
-} // namespace FileSys

src/video_core/engines/maxwell_3d.cpp

@@ -398,27 +398,6 @@ u32 Maxwell3D::GetRegisterValue(u32 method) const {
     return regs.reg_array[method];
 }
 
-bool Maxwell3D::IsShaderStageEnabled(Regs::ShaderStage stage) const {
-    // The Vertex stage is always enabled.
-    if (stage == Regs::ShaderStage::Vertex)
-        return true;
-
-    switch (stage) {
-    case Regs::ShaderStage::TesselationControl:
-        return regs.shader_config[static_cast<size_t>(Regs::ShaderProgram::TesselationControl)]
-                   .enable != 0;
-    case Regs::ShaderStage::TesselationEval:
-        return regs.shader_config[static_cast<size_t>(Regs::ShaderProgram::TesselationEval)]
-                   .enable != 0;
-    case Regs::ShaderStage::Geometry:
-        return regs.shader_config[static_cast<size_t>(Regs::ShaderProgram::Geometry)].enable != 0;
-    case Regs::ShaderStage::Fragment:
-        return regs.shader_config[static_cast<size_t>(Regs::ShaderProgram::Fragment)].enable != 0;
-    }
-
-    UNREACHABLE();
-}
-
 void Maxwell3D::ProcessClearBuffers() {
     ASSERT(regs.clear_buffers.R == regs.clear_buffers.G &&
            regs.clear_buffers.R == regs.clear_buffers.B &&

src/video_core/engines/maxwell_3d.h

@@ -379,6 +379,14 @@ public:
             }
         };
 
+        bool IsShaderConfigEnabled(size_t index) const {
+            // The VertexB is always enabled.
+            if (index == static_cast<size_t>(Regs::ShaderProgram::VertexB)) {
+                return true;
+            }
+            return shader_config[index].enable != 0;
+        }
+
         union {
             struct {
                 INSERT_PADDING_WORDS(0x45);
@@ -512,7 +520,18 @@ public:
                     u32 enable[NumRenderTargets];
                 } blend;
 
-                INSERT_PADDING_WORDS(0xB);
+                struct {
+                    u32 enable;
+                    u32 front_op_fail;
+                    u32 front_op_zfail;
+                    u32 front_op_zpass;
+                    u32 front_func_func;
+                    u32 front_func_ref;
+                    u32 front_func_mask;
+                    u32 front_mask;
+                } stencil;
+
+                INSERT_PADDING_WORDS(0x3);
 
                 union {
                     BitField<4, 1, u32> triangle_rast_flip;
@@ -548,7 +567,17 @@ public:
                     }
                 } tic;
 
-                INSERT_PADDING_WORDS(0x21);
+                INSERT_PADDING_WORDS(0x5);
+
+                struct {
+                    u32 enable;
+                    u32 back_op_fail;
+                    u32 back_op_zfail;
+                    u32 back_op_zpass;
+                    u32 back_func_func;
+                } stencil_two_side;
+
+                INSERT_PADDING_WORDS(0x17);
 
                 union {
                     BitField<2, 1, u32> coord_origin;
@@ -780,9 +809,6 @@ public:
     /// Returns the texture information for a specific texture in a specific shader stage.
     Texture::FullTextureInfo GetStageTexture(Regs::ShaderStage stage, size_t offset) const;
 
-    /// Returns whether the specified shader stage is enabled or not.
-    bool IsShaderStageEnabled(Regs::ShaderStage stage) const;
-
 private:
     std::unordered_map<u32, std::vector<u32>> uploaded_macros;
 
@@ -846,10 +872,12 @@ ASSERT_REG_POSITION(depth_write_enabled, 0x4BA);
 ASSERT_REG_POSITION(d3d_cull_mode, 0x4C2);
 ASSERT_REG_POSITION(depth_test_func, 0x4C3);
 ASSERT_REG_POSITION(blend, 0x4CF);
+ASSERT_REG_POSITION(stencil, 0x4E0);
 ASSERT_REG_POSITION(screen_y_control, 0x4EB);
 ASSERT_REG_POSITION(vb_element_base, 0x50D);
 ASSERT_REG_POSITION(tsc, 0x557);
 ASSERT_REG_POSITION(tic, 0x55D);
+ASSERT_REG_POSITION(stencil_two_side, 0x565);
 ASSERT_REG_POSITION(point_coord_replace, 0x581);
 ASSERT_REG_POSITION(code_address, 0x582);
 ASSERT_REG_POSITION(draw, 0x585);

src/video_core/engines/shader_bytecode.h

@@ -142,6 +142,7 @@ enum class PredCondition : u64 {
     GreaterThan = 4,
     NotEqual = 5,
     GreaterEqual = 6,
+    LessThanWithNan = 9,
     NotEqualWithNan = 13,
     // TODO(Subv): Other condition types
 };
@@ -194,6 +195,18 @@ enum class UniformType : u64 {
     Double = 5,
 };
 
+enum class IMinMaxExchange : u64 {
+    None = 0,
+    XLo = 1,
+    XMed = 2,
+    XHi = 3,
+};
+
+enum class FlowCondition : u64 {
+    Always = 0xF,
+    Fcsm_Tr = 0x1C, // TODO(bunnei): What is this used for?
+};
+
 union Instruction {
     Instruction& operator=(const Instruction& instr) {
         value = instr.value;
@@ -278,6 +291,13 @@ union Instruction {
         BitField<49, 1, u64> negate_a;
     } alu_integer;
 
+    union {
+        BitField<39, 3, u64> pred;
+        BitField<42, 1, u64> negate_pred;
+        BitField<43, 2, IMinMaxExchange> exchange;
+        BitField<48, 1, u64> is_signed;
+    } imnmx;
+
     union {
         BitField<54, 1, u64> saturate;
         BitField<56, 1, u64> negate_a;
@@ -301,6 +321,10 @@ union Instruction {
         }
     } bfe;
 
+    union {
+        BitField<0, 5, FlowCondition> cond;
+    } flow;
+
     union {
         BitField<48, 1, u64> negate_b;
         BitField<49, 1, u64> negate_c;
@@ -673,7 +697,7 @@ private:
             INST("1101101---------", Id::TLDS, Type::Memory, "TLDS"),
             INST("111000110000----", Id::EXIT, Type::Trivial, "EXIT"),
             INST("11100000--------", Id::IPA, Type::Trivial, "IPA"),
-            INST("001100101-------", Id::FFMA_IMM, Type::Ffma, "FFMA_IMM"),
+            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"),
             INST("010110011-------", Id::FFMA_RR, Type::Ffma, "FFMA_RR"),
@@ -709,9 +733,9 @@ private:
             INST("0100110001100---", Id::FMNMX_C, Type::Arithmetic, "FMNMX_C"),
             INST("0101110001100---", Id::FMNMX_R, Type::Arithmetic, "FMNMX_R"),
             INST("0011100-01100---", Id::FMNMX_IMM, Type::Arithmetic, "FMNMX_IMM"),
-            INST("0100110000100---", Id::IMNMX_C, Type::Arithmetic, "FMNMX_IMM"),
-            INST("0101110000100---", Id::IMNMX_R, Type::Arithmetic, "FMNMX_IMM"),
-            INST("0011100-00100---", Id::IMNMX_IMM, Type::Arithmetic, "FMNMX_IMM"),
+            INST("0100110000100---", Id::IMNMX_C, Type::ArithmeticInteger, "IMNMX_C"),
+            INST("0101110000100---", Id::IMNMX_R, Type::ArithmeticInteger, "IMNMX_R"),
+            INST("0011100-00100---", Id::IMNMX_IMM, Type::ArithmeticInteger, "IMNMX_IMM"),
             INST("0100110000000---", Id::BFE_C, Type::Bfe, "BFE_C"),
             INST("0101110000000---", Id::BFE_R, Type::Bfe, "BFE_R"),
             INST("0011100-00000---", Id::BFE_IMM, Type::Bfe, "BFE_IMM"),

src/video_core/gpu.h

@@ -67,6 +67,7 @@ struct FramebufferConfig {
 
     using TransformFlags = Service::NVFlinger::BufferQueue::BufferTransformFlags;
     TransformFlags transform_flags;
+    MathUtil::Rectangle<int> crop_rect;
 };
 
 namespace Engines {

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -15,6 +15,7 @@
 #include "common/microprofile.h"
 #include "common/scope_exit.h"
 #include "core/core.h"
+#include "core/frontend/emu_window.h"
 #include "core/hle/kernel/process.h"
 #include "core/settings.h"
 #include "video_core/engines/maxwell_3d.h"
@@ -22,6 +23,7 @@
 #include "video_core/renderer_opengl/gl_shader_gen.h"
 #include "video_core/renderer_opengl/maxwell_to_gl.h"
 #include "video_core/renderer_opengl/renderer_opengl.h"
+#include "video_core/video_core.h"
 
 using Maxwell = Tegra::Engines::Maxwell3D::Regs;
 using PixelFormat = SurfaceParams::PixelFormat;
@@ -126,8 +128,8 @@ RasterizerOpenGL::~RasterizerOpenGL() {
 std::pair<u8*, GLintptr> RasterizerOpenGL::SetupVertexArrays(u8* array_ptr,
                                                              GLintptr buffer_offset) {
     MICROPROFILE_SCOPE(OpenGL_VAO);
-    const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
-    const auto& memory_manager = Core::System().GetInstance().GPU().memory_manager;
+    const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+    const auto& memory_manager = Core::System::GetInstance().GPU().memory_manager;
 
     state.draw.vertex_array = hw_vao.handle;
     state.draw.vertex_buffer = stream_buffer->GetHandle();
@@ -181,6 +183,19 @@ std::pair<u8*, GLintptr> RasterizerOpenGL::SetupVertexArrays(u8* array_ptr,
     return {array_ptr, buffer_offset};
 }
 
+static GLShader::ProgramCode GetShaderProgramCode(Maxwell::ShaderProgram program) {
+    auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
+
+    // Fetch program code from memory
+    GLShader::ProgramCode program_code;
+    auto& shader_config = gpu.regs.shader_config[static_cast<size_t>(program)];
+    const u64 gpu_address{gpu.regs.code_address.CodeAddress() + shader_config.offset};
+    const boost::optional<VAddr> cpu_address{gpu.memory_manager.GpuToCpuAddress(gpu_address)};
+    Memory::ReadBlock(*cpu_address, program_code.data(), program_code.size() * sizeof(u64));
+
+    return program_code;
+}
+
 void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset) {
     // Helper function for uploading uniform data
     const auto copy_buffer = [&](GLuint handle, GLintptr offset, GLsizeiptr size) {
@@ -192,27 +207,24 @@ void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset) {
         }
     };
 
-    auto& gpu = Core::System().GetInstance().GPU().Maxwell3D();
-    ASSERT_MSG(!gpu.regs.shader_config[0].enable, "VertexA is unsupported!");
+    auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
 
     // Next available bindpoints to use when uploading the const buffers and textures to the GLSL
     // shaders. The constbuffer bindpoint starts after the shader stage configuration bind points.
     u32 current_constbuffer_bindpoint = uniform_buffers.size();
     u32 current_texture_bindpoint = 0;
 
-    for (unsigned index = 1; index < Maxwell::MaxShaderProgram; ++index) {
+    for (size_t index = 0; index < Maxwell::MaxShaderProgram; ++index) {
         auto& shader_config = gpu.regs.shader_config[index];
         const Maxwell::ShaderProgram program{static_cast<Maxwell::ShaderProgram>(index)};
 
-        const auto& stage = index - 1; // Stage indices are 0 - 5
-
-        const bool is_enabled = gpu.IsShaderStageEnabled(static_cast<Maxwell::ShaderStage>(stage));
-
         // Skip stages that are not enabled
-        if (!is_enabled) {
+        if (!gpu.regs.IsShaderConfigEnabled(index)) {
             continue;
         }
 
+        const size_t stage{index == 0 ? 0 : index - 1}; // Stage indices are 0 - 5
+
         GLShader::MaxwellUniformData ubo{};
         ubo.SetFromRegs(gpu.state.shader_stages[stage]);
         std::memcpy(buffer_ptr, &ubo, sizeof(ubo));
@@ -228,16 +240,21 @@ void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset) {
         buffer_ptr += sizeof(GLShader::MaxwellUniformData);
         buffer_offset += sizeof(GLShader::MaxwellUniformData);
 
-        // Fetch program code from memory
-        GLShader::ProgramCode program_code;
-        const u64 gpu_address{gpu.regs.code_address.CodeAddress() + shader_config.offset};
-        const boost::optional<VAddr> cpu_address{gpu.memory_manager.GpuToCpuAddress(gpu_address)};
-        Memory::ReadBlock(*cpu_address, program_code.data(), program_code.size() * sizeof(u64));
-        GLShader::ShaderSetup setup{std::move(program_code)};
-
+        GLShader::ShaderSetup setup{GetShaderProgramCode(program)};
         GLShader::ShaderEntries shader_resources;
 
         switch (program) {
+        case Maxwell::ShaderProgram::VertexA: {
+            // VertexB is always enabled, so when VertexA is enabled, we have two vertex shaders.
+            // Conventional HW does not support this, so we combine VertexA and VertexB into one
+            // stage here.
+            setup.SetProgramB(GetShaderProgramCode(Maxwell::ShaderProgram::VertexB));
+            GLShader::MaxwellVSConfig vs_config{setup};
+            shader_resources =
+                shader_program_manager->UseProgrammableVertexShader(vs_config, setup);
+            break;
+        }
+
         case Maxwell::ShaderProgram::VertexB: {
             GLShader::MaxwellVSConfig vs_config{setup};
             shader_resources =
@@ -268,13 +285,19 @@ void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset) {
         current_texture_bindpoint =
             SetupTextures(static_cast<Maxwell::ShaderStage>(stage), gl_stage_program,
                           current_texture_bindpoint, shader_resources.texture_samplers);
+
+        // When VertexA is enabled, we have dual vertex shaders
+        if (program == Maxwell::ShaderProgram::VertexA) {
+            // VertexB was combined with VertexA, so we skip the VertexB iteration
+            index++;
+        }
     }
 
     shader_program_manager->UseTrivialGeometryShader();
 }
 
 size_t RasterizerOpenGL::CalculateVertexArraysSize() const {
-    const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
+    const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
 
     size_t size = 0;
     for (u32 index = 0; index < Maxwell::NumVertexArrays; ++index) {
@@ -299,10 +322,7 @@ bool RasterizerOpenGL::AccelerateDrawBatch(bool is_indexed) {
 
 std::pair<Surface, Surface> RasterizerOpenGL::ConfigureFramebuffers(bool using_color_fb,
                                                                     bool using_depth_fb) {
-    const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
-
-    // Sync the depth test state before configuring the framebuffer surfaces.
-    SyncDepthTestState();
+    const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
 
     // TODO(bunnei): Implement this
     const bool has_stencil = false;
@@ -354,7 +374,7 @@ std::pair<Surface, Surface> RasterizerOpenGL::ConfigureFramebuffers(bool using_c
 }
 
 void RasterizerOpenGL::Clear() {
-    const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
+    const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
 
     bool use_color_fb = false;
     bool use_depth_fb = false;
@@ -368,11 +388,20 @@ void RasterizerOpenGL::Clear() {
     if (regs.clear_buffers.Z) {
         clear_mask |= GL_DEPTH_BUFFER_BIT;
         use_depth_fb = true;
+
+        // Always enable the depth write when clearing the depth buffer. The depth write mask is
+        // ignored when clearing the buffer in the Switch, but OpenGL obeys it so we set it to true.
+        state.depth.test_enabled = true;
+        state.depth.write_mask = GL_TRUE;
+        state.depth.test_func = GL_ALWAYS;
+        state.Apply();
     }
 
     if (clear_mask == 0)
         return;
 
+    ScopeAcquireGLContext acquire_context;
+
     auto [dirty_color_surface, dirty_depth_surface] =
         ConfigureFramebuffers(use_color_fb, use_depth_fb);
 
@@ -397,11 +426,14 @@ void RasterizerOpenGL::DrawArrays() {
         return;
 
     MICROPROFILE_SCOPE(OpenGL_Drawing);
-    const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
+    const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+
+    ScopeAcquireGLContext acquire_context;
 
     auto [dirty_color_surface, dirty_depth_surface] =
         ConfigureFramebuffers(true, regs.zeta.Address() != 0);
 
+    SyncDepthTestState();
     SyncBlendState();
     SyncCullMode();
 
@@ -441,7 +473,7 @@ void RasterizerOpenGL::DrawArrays() {
     // If indexed mode, copy the index buffer
     GLintptr index_buffer_offset = 0;
     if (is_indexed) {
-        const auto& memory_manager = Core::System().GetInstance().GPU().memory_manager;
+        const auto& memory_manager = Core::System::GetInstance().GPU().memory_manager;
         const boost::optional<VAddr> index_data_addr{
             memory_manager->GpuToCpuAddress(regs.index_array.StartAddress())};
         Memory::ReadBlock(*index_data_addr, offseted_buffer, index_buffer_size);
@@ -605,9 +637,6 @@ u32 RasterizerOpenGL::SetupConstBuffers(Maxwell::ShaderStage stage, GLuint progr
     auto& gpu = Core::System::GetInstance().GPU();
     auto& maxwell3d = gpu.Get3DEngine();
 
-    ASSERT_MSG(maxwell3d.IsShaderStageEnabled(stage),
-               "Attempted to upload constbuffer of disabled shader stage");
-
     // Reset all buffer draw state for this stage.
     for (auto& buffer : state.draw.const_buffers[static_cast<size_t>(stage)]) {
         buffer.bindpoint = 0;
@@ -674,9 +703,6 @@ u32 RasterizerOpenGL::SetupTextures(Maxwell::ShaderStage stage, GLuint program,
     auto& gpu = Core::System::GetInstance().GPU();
     auto& maxwell3d = gpu.Get3DEngine();
 
-    ASSERT_MSG(maxwell3d.IsShaderStageEnabled(stage),
-               "Attempted to upload textures of disabled shader stage");
-
     ASSERT_MSG(current_unit + entries.size() <= std::size(state.texture_units),
                "Exceeded the number of active textures.");
 
@@ -749,7 +775,7 @@ void RasterizerOpenGL::BindFramebufferSurfaces(const Surface& color_surface,
 }
 
 void RasterizerOpenGL::SyncViewport(const MathUtil::Rectangle<u32>& surfaces_rect) {
-    const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
+    const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
     const MathUtil::Rectangle<s32> viewport_rect{regs.viewport_transform[0].GetRect()};
 
     state.viewport.x = static_cast<GLint>(surfaces_rect.left) + viewport_rect.left;
@@ -767,7 +793,7 @@ void RasterizerOpenGL::SyncClipCoef() {
 }
 
 void RasterizerOpenGL::SyncCullMode() {
-    const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
+    const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
 
     state.cull.enabled = regs.cull.enabled != 0;
 
@@ -775,10 +801,13 @@ void RasterizerOpenGL::SyncCullMode() {
         state.cull.front_face = MaxwellToGL::FrontFace(regs.cull.front_face);
         state.cull.mode = MaxwellToGL::CullFace(regs.cull.cull_face);
 
+        const bool flip_triangles{regs.screen_y_control.triangle_rast_flip == 0 ||
+                                  regs.viewport_transform[0].scale_y < 0.0f};
+
         // If the GPU is configured to flip the rasterized triangles, then we need to flip the
         // notion of front and back. Note: We flip the triangles when the value of the register is 0
         // because OpenGL already does it for us.
-        if (regs.screen_y_control.triangle_rast_flip == 0) {
+        if (flip_triangles) {
             if (state.cull.front_face == GL_CCW)
                 state.cull.front_face = GL_CW;
             else if (state.cull.front_face == GL_CW)
@@ -796,7 +825,7 @@ void RasterizerOpenGL::SyncDepthOffset() {
 }
 
 void RasterizerOpenGL::SyncDepthTestState() {
-    const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
+    const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
 
     state.depth.test_enabled = regs.depth_test_enable != 0;
     state.depth.write_mask = regs.depth_write_enabled ? GL_TRUE : GL_FALSE;
@@ -808,7 +837,7 @@ void RasterizerOpenGL::SyncDepthTestState() {
 }
 
 void RasterizerOpenGL::SyncBlendState() {
-    const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
+    const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
 
     // TODO(Subv): Support more than just render target 0.
     state.blend.enabled = regs.blend.enable[0] != 0;

src/video_core/renderer_opengl/gl_rasterizer_cache.cpp

@@ -102,7 +102,10 @@ static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_form
     {GL_COMPRESSED_RGBA_S3TC_DXT5_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
      true},                                                                                 // DXT45
     {GL_COMPRESSED_RED_RGTC1, GL_RED, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm, true}, // DXN1
+    {GL_COMPRESSED_RGBA_BPTC_UNORM_ARB, GL_RGB, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
+     true},                                                             // BC7U
     {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_4X4
+    {GL_RG8, GL_RG, GL_UNSIGNED_BYTE, ComponentType::UNorm, false},     // G8R8
 
     // DepthStencil formats
     {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8, ComponentType::UNorm,
@@ -110,6 +113,8 @@ static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_form
     {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8, ComponentType::UNorm,
      false},                                                                            // S8Z24
     {GL_DEPTH_COMPONENT32F, GL_DEPTH_COMPONENT, GL_FLOAT, ComponentType::Float, false}, // Z32F
+    {GL_DEPTH_COMPONENT16, GL_DEPTH_COMPONENT, GL_UNSIGNED_SHORT, ComponentType::UNorm,
+     false}, // Z16
 }};
 
 static const FormatTuple& GetFormatTuple(PixelFormat pixel_format, ComponentType component_type) {
@@ -191,8 +196,10 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr),
         MortonCopy<true, PixelFormat::R11FG11FB10F>, MortonCopy<true, PixelFormat::RGBA32UI>,
         MortonCopy<true, PixelFormat::DXT1>,         MortonCopy<true, PixelFormat::DXT23>,
         MortonCopy<true, PixelFormat::DXT45>,        MortonCopy<true, PixelFormat::DXN1>,
-        MortonCopy<true, PixelFormat::ASTC_2D_4X4>,  MortonCopy<true, PixelFormat::Z24S8>,
+        MortonCopy<true, PixelFormat::BC7U>,         MortonCopy<true, PixelFormat::ASTC_2D_4X4>,
+        MortonCopy<true, PixelFormat::G8R8>,         MortonCopy<true, PixelFormat::Z24S8>,
         MortonCopy<true, PixelFormat::S8Z24>,        MortonCopy<true, PixelFormat::Z32F>,
+        MortonCopy<true, PixelFormat::Z16>,
 };
 
 static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr),
@@ -206,15 +213,18 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr),
         MortonCopy<false, PixelFormat::RGBA16F>,
         MortonCopy<false, PixelFormat::R11FG11FB10F>,
         MortonCopy<false, PixelFormat::RGBA32UI>,
-        // TODO(Subv): Swizzling the DXT1/DXT23/DXT45/DXN1 formats is not yet supported
+        // TODO(Subv): Swizzling the DXT1/DXT23/DXT45/DXN1/BC7U formats is not yet supported
         nullptr,
         nullptr,
         nullptr,
         nullptr,
-        MortonCopy<false, PixelFormat::ABGR8>,
+        nullptr,
+        nullptr,
+        MortonCopy<false, PixelFormat::G8R8>,
         MortonCopy<false, PixelFormat::Z24S8>,
         MortonCopy<false, PixelFormat::S8Z24>,
         MortonCopy<false, PixelFormat::Z32F>,
+        MortonCopy<false, PixelFormat::Z16>,
 };
 
 // Allocate an uninitialized texture of appropriate size and format for the surface
@@ -267,9 +277,10 @@ static void ConvertS8Z24ToZ24S8(std::vector<u8>& data, u32 width, u32 height) {
 
     S8Z24 input_pixel{};
     Z24S8 output_pixel{};
+    const auto bpp{CachedSurface::GetGLBytesPerPixel(PixelFormat::S8Z24)};
     for (size_t y = 0; y < height; ++y) {
         for (size_t x = 0; x < width; ++x) {
-            const size_t offset{y * width + x};
+            const size_t offset{bpp * (y * width + x)};
             std::memcpy(&input_pixel, &data[offset], sizeof(S8Z24));
             output_pixel.s8.Assign(input_pixel.s8);
             output_pixel.z24.Assign(input_pixel.z24);
@@ -277,6 +288,19 @@ static void ConvertS8Z24ToZ24S8(std::vector<u8>& data, u32 width, u32 height) {
         }
     }
 }
+
+static void ConvertG8R8ToR8G8(std::vector<u8>& data, u32 width, u32 height) {
+    const auto bpp{CachedSurface::GetGLBytesPerPixel(PixelFormat::G8R8)};
+    for (size_t y = 0; y < height; ++y) {
+        for (size_t x = 0; x < width; ++x) {
+            const size_t offset{bpp * (y * width + x)};
+            const u8 temp{data[offset]};
+            data[offset] = data[offset + 1];
+            data[offset + 1] = temp;
+        }
+    }
+}
+
 /**
  * Helper function to perform software conversion (as needed) when loading a buffer from Switch
  * memory. This is for Maxwell pixel formats that cannot be represented as-is in OpenGL or with
@@ -297,6 +321,11 @@ static void ConvertFormatAsNeeded_LoadGLBuffer(std::vector<u8>& data, PixelForma
         // Convert the S8Z24 depth format to Z24S8, as OpenGL does not support S8Z24.
         ConvertS8Z24ToZ24S8(data, width, height);
         break;
+
+    case PixelFormat::G8R8:
+        // Convert the G8R8 color format to R8G8, as OpenGL does not support G8R8.
+        ConvertG8R8ToR8G8(data, width, height);
+        break;
     }
 }
 
@@ -476,7 +505,7 @@ Surface RasterizerCacheOpenGL::GetTextureSurface(const Tegra::Texture::FullTextu
 
 SurfaceSurfaceRect_Tuple RasterizerCacheOpenGL::GetFramebufferSurfaces(
     bool using_color_fb, bool using_depth_fb, const MathUtil::Rectangle<s32>& viewport) {
-    const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
+    const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
 
     // TODO(bunnei): This is hard corded to use just the first render buffer
     LOG_WARNING(Render_OpenGL, "hard-coded for render target 0!");

src/video_core/renderer_opengl/gl_rasterizer_cache.h

@@ -35,14 +35,17 @@ struct SurfaceParams {
         DXT23 = 9,
         DXT45 = 10,
         DXN1 = 11, // This is also known as BC4
-        ASTC_2D_4X4 = 12,
+        BC7U = 12,
+        ASTC_2D_4X4 = 13,
+        G8R8 = 14,
 
         MaxColorFormat,
 
         // DepthStencil formats
-        Z24S8 = 13,
-        S8Z24 = 14,
-        Z32F = 15,
+        Z24S8 = 15,
+        S8Z24 = 16,
+        Z32F = 17,
+        Z16 = 18,
 
         MaxDepthStencilFormat,
 
@@ -92,10 +95,13 @@ struct SurfaceParams {
             4, // DXT23
             4, // DXT45
             4, // DXN1
+            4, // BC7U
             4, // ASTC_2D_4X4
+            1, // G8R8
             1, // Z24S8
             1, // S8Z24
             1, // Z32F
+            1, // Z16
         }};
 
         ASSERT(static_cast<size_t>(format) < compression_factor_table.size());
@@ -119,10 +125,13 @@ struct SurfaceParams {
             128, // DXT23
             128, // DXT45
             64,  // DXN1
+            128, // BC7U
             32,  // ASTC_2D_4X4
+            16,  // G8R8
             32,  // Z24S8
             32,  // S8Z24
             32,  // Z32F
+            16,  // Z16
         }};
 
         ASSERT(static_cast<size_t>(format) < bpp_table.size());
@@ -140,6 +149,8 @@ struct SurfaceParams {
             return PixelFormat::Z24S8;
         case Tegra::DepthFormat::Z32_FLOAT:
             return PixelFormat::Z32F;
+        case Tegra::DepthFormat::Z16_UNORM:
+            return PixelFormat::Z16;
         default:
             LOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
             UNREACHABLE();
@@ -178,6 +189,8 @@ struct SurfaceParams {
             return PixelFormat::A1B5G5R5;
         case Tegra::Texture::TextureFormat::R8:
             return PixelFormat::R8;
+        case Tegra::Texture::TextureFormat::G8R8:
+            return PixelFormat::G8R8;
         case Tegra::Texture::TextureFormat::R16_G16_B16_A16:
             return PixelFormat::RGBA16F;
         case Tegra::Texture::TextureFormat::BF10GF11RF11:
@@ -192,6 +205,8 @@ struct SurfaceParams {
             return PixelFormat::DXT45;
         case Tegra::Texture::TextureFormat::DXN1:
             return PixelFormat::DXN1;
+        case Tegra::Texture::TextureFormat::BC7U:
+            return PixelFormat::BC7U;
         case Tegra::Texture::TextureFormat::ASTC_2D_4X4:
             return PixelFormat::ASTC_2D_4X4;
         default:
@@ -213,6 +228,8 @@ struct SurfaceParams {
             return Tegra::Texture::TextureFormat::A1B5G5R5;
         case PixelFormat::R8:
             return Tegra::Texture::TextureFormat::R8;
+        case PixelFormat::G8R8:
+            return Tegra::Texture::TextureFormat::G8R8;
         case PixelFormat::RGBA16F:
             return Tegra::Texture::TextureFormat::R16_G16_B16_A16;
         case PixelFormat::R11FG11FB10F:
@@ -227,6 +244,8 @@ struct SurfaceParams {
             return Tegra::Texture::TextureFormat::DXT45;
         case PixelFormat::DXN1:
             return Tegra::Texture::TextureFormat::DXN1;
+        case PixelFormat::BC7U:
+            return Tegra::Texture::TextureFormat::BC7U;
         case PixelFormat::ASTC_2D_4X4:
             return Tegra::Texture::TextureFormat::ASTC_2D_4X4;
         default:
@@ -242,6 +261,8 @@ struct SurfaceParams {
             return Tegra::DepthFormat::Z24_S8_UNORM;
         case PixelFormat::Z32F:
             return Tegra::DepthFormat::Z32_FLOAT;
+        case PixelFormat::Z16:
+            return Tegra::DepthFormat::Z16_UNORM;
         default:
             UNREACHABLE();
         }
@@ -288,6 +309,7 @@ struct SurfaceParams {
 
     static ComponentType ComponentTypeFromDepthFormat(Tegra::DepthFormat format) {
         switch (format) {
+        case Tegra::DepthFormat::Z16_UNORM:
         case Tegra::DepthFormat::S8_Z24_UNORM:
         case Tegra::DepthFormat::Z24_S8_UNORM:
             return ComponentType::UNorm;

src/video_core/renderer_opengl/gl_shader_decompiler.cpp

@@ -42,13 +42,14 @@ enum class ExitMethod {
 struct Subroutine {
     /// Generates a name suitable for GLSL source code.
     std::string GetName() const {
-        return "sub_" + std::to_string(begin) + '_' + std::to_string(end);
+        return "sub_" + std::to_string(begin) + '_' + std::to_string(end) + '_' + suffix;
     }
 
-    u32 begin;              ///< Entry point of the subroutine.
-    u32 end;                ///< Return point of the subroutine.
-    ExitMethod exit_method; ///< Exit method of the subroutine.
-    std::set<u32> labels;   ///< Addresses refereced by JMP instructions.
+    u32 begin;                 ///< Entry point of the subroutine.
+    u32 end;                   ///< Return point of the subroutine.
+    const std::string& suffix; ///< Suffix of the shader, used to make a unique subroutine name
+    ExitMethod exit_method;    ///< Exit method of the subroutine.
+    std::set<u32> labels;      ///< Addresses refereced by JMP instructions.
 
     bool operator<(const Subroutine& rhs) const {
         return std::tie(begin, end) < std::tie(rhs.begin, rhs.end);
@@ -58,11 +59,11 @@ struct Subroutine {
 /// Analyzes shader code and produces a set of subroutines.
 class ControlFlowAnalyzer {
 public:
-    ControlFlowAnalyzer(const ProgramCode& program_code, u32 main_offset)
+    ControlFlowAnalyzer(const ProgramCode& program_code, u32 main_offset, const std::string& suffix)
         : program_code(program_code) {
 
         // Recursively finds all subroutines.
-        const Subroutine& program_main = AddSubroutine(main_offset, PROGRAM_END);
+        const Subroutine& program_main = AddSubroutine(main_offset, PROGRAM_END, suffix);
         if (program_main.exit_method != ExitMethod::AlwaysEnd)
             throw DecompileFail("Program does not always end");
     }
@@ -77,12 +78,12 @@ private:
     std::map<std::pair<u32, u32>, ExitMethod> exit_method_map;
 
     /// Adds and analyzes a new subroutine if it is not added yet.
-    const Subroutine& AddSubroutine(u32 begin, u32 end) {
-        auto iter = subroutines.find(Subroutine{begin, end});
+    const Subroutine& AddSubroutine(u32 begin, u32 end, const std::string& suffix) {
+        auto iter = subroutines.find(Subroutine{begin, end, suffix});
         if (iter != subroutines.end())
             return *iter;
 
-        Subroutine subroutine{begin, end};
+        Subroutine subroutine{begin, end, suffix};
         subroutine.exit_method = Scan(begin, end, subroutine.labels);
         if (subroutine.exit_method == ExitMethod::Undetermined)
             throw DecompileFail("Recursive function detected");
@@ -191,7 +192,8 @@ public:
         UnsignedInteger,
     };
 
-    GLSLRegister(size_t index, ShaderWriter& shader) : index{index}, shader{shader} {}
+    GLSLRegister(size_t index, ShaderWriter& shader, const std::string& suffix)
+        : index{index}, shader{shader}, suffix{suffix} {}
 
     /// Gets the GLSL type string for a register
     static std::string GetTypeString(Type type) {
@@ -216,7 +218,7 @@ public:
     /// Returns a GLSL string representing the current state of the register
     const std::string GetActiveString() {
         declr_type.insert(active_type);
-        return GetPrefixString(active_type) + std::to_string(index);
+        return GetPrefixString(active_type) + std::to_string(index) + '_' + suffix;
     }
 
     /// Returns true if the active type is a float
@@ -251,6 +253,7 @@ private:
     ShaderWriter& shader;
     Type active_type{Type::Float};
     std::set<Type> declr_type;
+    const std::string& suffix;
 };
 
 /**
@@ -262,8 +265,8 @@ private:
 class GLSLRegisterManager {
 public:
     GLSLRegisterManager(ShaderWriter& shader, ShaderWriter& declarations,
-                        const Maxwell3D::Regs::ShaderStage& stage)
-        : shader{shader}, declarations{declarations}, stage{stage} {
+                        const Maxwell3D::Regs::ShaderStage& stage, const std::string& suffix)
+        : shader{shader}, declarations{declarations}, stage{stage}, suffix{suffix} {
         BuildRegisterList();
     }
 
@@ -430,12 +433,12 @@ public:
     }
 
     /// Add declarations for registers
-    void GenerateDeclarations() {
+    void GenerateDeclarations(const std::string& suffix) {
         for (const auto& reg : regs) {
             for (const auto& type : reg.DeclaredTypes()) {
                 declarations.AddLine(GLSLRegister::GetTypeString(type) + ' ' +
-                                     GLSLRegister::GetPrefixString(type) +
-                                     std::to_string(reg.GetIndex()) + " = 0;");
+                                     reg.GetPrefixString(type) + std::to_string(reg.GetIndex()) +
+                                     '_' + suffix + " = 0;");
             }
         }
         declarations.AddNewLine();
@@ -558,7 +561,7 @@ private:
     /// Build the GLSL register list.
     void BuildRegisterList() {
         for (size_t index = 0; index < Register::NumRegisters; ++index) {
-            regs.emplace_back(index, shader);
+            regs.emplace_back(index, shader, suffix);
         }
     }
 
@@ -620,16 +623,17 @@ private:
     std::array<ConstBufferEntry, Maxwell3D::Regs::MaxConstBuffers> declr_const_buffers;
     std::vector<SamplerEntry> used_samplers;
     const Maxwell3D::Regs::ShaderStage& stage;
+    const std::string& suffix;
 };
 
 class GLSLGenerator {
 public:
     GLSLGenerator(const std::set<Subroutine>& subroutines, const ProgramCode& program_code,
-                  u32 main_offset, Maxwell3D::Regs::ShaderStage stage)
+                  u32 main_offset, Maxwell3D::Regs::ShaderStage stage, const std::string& suffix)
         : subroutines(subroutines), program_code(program_code), main_offset(main_offset),
-          stage(stage) {
+          stage(stage), suffix(suffix) {
 
-        Generate();
+        Generate(suffix);
     }
 
     std::string GetShaderCode() {
@@ -644,7 +648,7 @@ public:
 private:
     /// Gets the Subroutine object corresponding to the specified address.
     const Subroutine& GetSubroutine(u32 begin, u32 end) const {
-        auto iter = subroutines.find(Subroutine{begin, end});
+        auto iter = subroutines.find(Subroutine{begin, end, suffix});
         ASSERT(iter != subroutines.end());
         return *iter;
     }
@@ -689,7 +693,7 @@ private:
         // Can't assign to the constant predicate.
         ASSERT(pred != static_cast<u64>(Pred::UnusedIndex));
 
-        std::string variable = 'p' + std::to_string(pred);
+        std::string variable = 'p' + std::to_string(pred) + '_' + suffix;
         shader.AddLine(variable + " = " + value + ';');
         declr_predicates.insert(std::move(variable));
     }
@@ -707,7 +711,7 @@ private:
         if (index == static_cast<u64>(Pred::UnusedIndex))
             variable = "true";
         else
-            variable = 'p' + std::to_string(index);
+            variable = 'p' + std::to_string(index) + '_' + suffix;
 
         if (negate) {
             return "!(" + variable + ')';
@@ -728,10 +732,10 @@ private:
                                        const std::string& op_a, const std::string& op_b) const {
         using Tegra::Shader::PredCondition;
         static const std::unordered_map<PredCondition, const char*> PredicateComparisonStrings = {
-            {PredCondition::LessThan, "<"},         {PredCondition::Equal, "=="},
-            {PredCondition::LessEqual, "<="},       {PredCondition::GreaterThan, ">"},
-            {PredCondition::NotEqual, "!="},        {PredCondition::GreaterEqual, ">="},
-            {PredCondition::NotEqualWithNan, "!="},
+            {PredCondition::LessThan, "<"},        {PredCondition::Equal, "=="},
+            {PredCondition::LessEqual, "<="},      {PredCondition::GreaterThan, ">"},
+            {PredCondition::NotEqual, "!="},       {PredCondition::GreaterEqual, ">="},
+            {PredCondition::LessThanWithNan, "<"}, {PredCondition::NotEqualWithNan, "!="},
         };
 
         const auto& comparison{PredicateComparisonStrings.find(condition)};
@@ -739,7 +743,8 @@ private:
                    "Unknown predicate comparison operation");
 
         std::string predicate{'(' + op_a + ") " + comparison->second + " (" + op_b + ')'};
-        if (condition == PredCondition::NotEqualWithNan) {
+        if (condition == PredCondition::LessThanWithNan ||
+            condition == PredCondition::NotEqualWithNan) {
             predicate += " || isnan(" + op_a + ") || isnan(" + op_b + ')';
         }
 
@@ -1150,6 +1155,20 @@ private:
                 WriteLogicOperation(instr.gpr0, instr.alu.lop.operation, op_a, op_b);
                 break;
             }
+            case OpCode::Id::IMNMX_C:
+            case OpCode::Id::IMNMX_R:
+            case OpCode::Id::IMNMX_IMM: {
+                ASSERT_MSG(instr.imnmx.exchange == Tegra::Shader::IMinMaxExchange::None,
+                           "Unimplemented");
+                std::string condition =
+                    GetPredicateCondition(instr.imnmx.pred, instr.imnmx.negate_pred != 0);
+                std::string parameters = op_a + ',' + op_b;
+                regs.SetRegisterToInteger(instr.gpr0, instr.imnmx.is_signed, 0,
+                                          '(' + condition + ") ? min(" + parameters + ") : max(" +
+                                              parameters + ')',
+                                          1, 1);
+                break;
+            }
             default: {
                 LOG_CRITICAL(HW_GPU, "Unhandled ArithmeticInteger instruction: {}",
                              opcode->GetName());
@@ -1625,16 +1644,32 @@ private:
                     shader.AddLine("color.a = " + regs.GetRegisterAsFloat(3) + ';');
                 }
 
-                shader.AddLine("return true;");
-                if (instr.pred.pred_index == static_cast<u64>(Pred::UnusedIndex)) {
-                    // If this is an unconditional exit then just end processing here, otherwise
-                    // we have to account for the possibility of the condition not being met, so
-                    // continue processing the next instruction.
-                    offset = PROGRAM_END - 1;
+                switch (instr.flow.cond) {
+                case Tegra::Shader::FlowCondition::Always:
+                    shader.AddLine("return true;");
+                    if (instr.pred.pred_index == static_cast<u64>(Pred::UnusedIndex)) {
+                        // If this is an unconditional exit then just end processing here,
+                        // otherwise we have to account for the possibility of the condition
+                        // not being met, so continue processing the next instruction.
+                        offset = PROGRAM_END - 1;
+                    }
+                    break;
+
+                case Tegra::Shader::FlowCondition::Fcsm_Tr:
+                    // TODO(bunnei): What is this used for? If we assume this conditon is not
+                    // satisifed, dual vertex shaders in Farming Simulator make more sense
+                    LOG_CRITICAL(HW_GPU, "Skipping unknown FlowCondition::Fcsm_Tr");
+                    break;
+
+                default:
+                    LOG_CRITICAL(HW_GPU, "Unhandled flow condition: {}",
+                                 static_cast<u32>(instr.flow.cond.Value()));
+                    UNREACHABLE();
                 }
                 break;
             }
             case OpCode::Id::KIL: {
+                ASSERT(instr.flow.cond == Tegra::Shader::FlowCondition::Always);
                 shader.AddLine("discard;");
                 break;
             }
@@ -1655,8 +1690,9 @@ private:
                 // can ignore this when generating GLSL code.
                 break;
             }
-            case OpCode::Id::DEPBAR:
-            case OpCode::Id::SYNC: {
+            case OpCode::Id::SYNC:
+                ASSERT(instr.flow.cond == Tegra::Shader::FlowCondition::Always);
+            case OpCode::Id::DEPBAR: {
                 // TODO(Subv): Find out if we actually have to care about these instructions or if
                 // the GLSL compiler takes care of that for us.
                 LOG_WARNING(HW_GPU, "DEPBAR/SYNC instruction is stubbed");
@@ -1696,7 +1732,7 @@ private:
         return program_counter;
     }
 
-    void Generate() {
+    void Generate(const std::string& suffix) {
         // Add declarations for all subroutines
         for (const auto& subroutine : subroutines) {
             shader.AddLine("bool " + subroutine.GetName() + "();");
@@ -1704,7 +1740,7 @@ private:
         shader.AddNewLine();
 
         // Add the main entry point
-        shader.AddLine("bool exec_shader() {");
+        shader.AddLine("bool exec_" + suffix + "() {");
         ++shader.scope;
         CallSubroutine(GetSubroutine(main_offset, PROGRAM_END));
         --shader.scope;
@@ -1767,7 +1803,7 @@ private:
 
     /// Add declarations for registers
     void GenerateDeclarations() {
-        regs.GenerateDeclarations();
+        regs.GenerateDeclarations(suffix);
 
         for (const auto& pred : declr_predicates) {
             declarations.AddLine("bool " + pred + " = false;");
@@ -1780,27 +1816,30 @@ private:
     const ProgramCode& program_code;
     const u32 main_offset;
     Maxwell3D::Regs::ShaderStage stage;
+    const std::string& suffix;
 
     ShaderWriter shader;
     ShaderWriter declarations;
-    GLSLRegisterManager regs{shader, declarations, stage};
+    GLSLRegisterManager regs{shader, declarations, stage, suffix};
 
     // Declarations
     std::set<std::string> declr_predicates;
 }; // namespace Decompiler
 
 std::string GetCommonDeclarations() {
-    std::string declarations = "bool exec_shader();\n";
+    std::string declarations;
     declarations += "#define MAX_CONSTBUFFER_ELEMENTS " +
                     std::to_string(RasterizerOpenGL::MaxConstbufferSize / (sizeof(GLvec4)));
+    declarations += '\n';
     return declarations;
 }
 
 boost::optional<ProgramResult> DecompileProgram(const ProgramCode& program_code, u32 main_offset,
-                                                Maxwell3D::Regs::ShaderStage stage) {
+                                                Maxwell3D::Regs::ShaderStage stage,
+                                                const std::string& suffix) {
     try {
-        auto subroutines = ControlFlowAnalyzer(program_code, main_offset).GetSubroutines();
-        GLSLGenerator generator(subroutines, program_code, main_offset, stage);
+        auto subroutines = ControlFlowAnalyzer(program_code, main_offset, suffix).GetSubroutines();
+        GLSLGenerator generator(subroutines, program_code, main_offset, stage, suffix);
         return ProgramResult{generator.GetShaderCode(), generator.GetEntries()};
     } catch (const DecompileFail& exception) {
         LOG_ERROR(HW_GPU, "Shader decompilation failed: {}", exception.what());

src/video_core/renderer_opengl/gl_shader_decompiler.h

@@ -20,7 +20,8 @@ using Tegra::Engines::Maxwell3D;
 std::string GetCommonDeclarations();
 
 boost::optional<ProgramResult> DecompileProgram(const ProgramCode& program_code, u32 main_offset,
-                                                Maxwell3D::Regs::ShaderStage stage);
+                                                Maxwell3D::Regs::ShaderStage stage,
+                                                const std::string& suffix);
 
 } // namespace Decompiler
 } // namespace GLShader

src/video_core/renderer_opengl/gl_shader_gen.cpp

@@ -17,10 +17,17 @@ ProgramResult GenerateVertexShader(const ShaderSetup& setup, const MaxwellVSConf
     std::string out = "#version 430 core\n";
     out += "#extension GL_ARB_separate_shader_objects : enable\n\n";
     out += Decompiler::GetCommonDeclarations();
+    out += "bool exec_vertex();\n";
+
+    if (setup.IsDualProgram()) {
+        out += "bool exec_vertex_b();\n";
+    }
+
+    ProgramResult program =
+        Decompiler::DecompileProgram(setup.program.code, PROGRAM_OFFSET,
+                                     Maxwell3D::Regs::ShaderStage::Vertex, "vertex")
+            .get_value_or({});
 
-    ProgramResult program = Decompiler::DecompileProgram(setup.program_code, PROGRAM_OFFSET,
-                                                         Maxwell3D::Regs::ShaderStage::Vertex)
-                                .get_value_or({});
     out += R"(
 
 out gl_PerVertex {
@@ -34,7 +41,14 @@ layout (std140) uniform vs_config {
 };
 
 void main() {
-    exec_shader();
+    exec_vertex();
+)";
+
+    if (setup.IsDualProgram()) {
+        out += "    exec_vertex_b();";
+    }
+
+    out += R"(
 
     // Viewport can be flipped, which is unsupported by glViewport
     position.xy *= viewport_flip.xy;
@@ -44,8 +58,19 @@ void main() {
     // For now, this is here to bring order in lieu of proper emulation
     position.w = 1.0;
 }
+
 )";
+
     out += program.first;
+
+    if (setup.IsDualProgram()) {
+        ProgramResult program_b =
+            Decompiler::DecompileProgram(setup.program.code_b, PROGRAM_OFFSET,
+                                         Maxwell3D::Regs::ShaderStage::Vertex, "vertex_b")
+                .get_value_or({});
+        out += program_b.first;
+    }
+
     return {out, program.second};
 }
 
@@ -53,12 +78,13 @@ ProgramResult GenerateFragmentShader(const ShaderSetup& setup, const MaxwellFSCo
     std::string out = "#version 430 core\n";
     out += "#extension GL_ARB_separate_shader_objects : enable\n\n";
     out += Decompiler::GetCommonDeclarations();
+    out += "bool exec_fragment();\n";
 
-    ProgramResult program = Decompiler::DecompileProgram(setup.program_code, PROGRAM_OFFSET,
-                                                         Maxwell3D::Regs::ShaderStage::Fragment)
-                                .get_value_or({});
+    ProgramResult program =
+        Decompiler::DecompileProgram(setup.program.code, PROGRAM_OFFSET,
+                                     Maxwell3D::Regs::ShaderStage::Fragment, "fragment")
+            .get_value_or({});
     out += R"(
-
 in vec4 position;
 out vec4 color;
 
@@ -67,7 +93,7 @@ layout (std140) uniform fs_config {
 };
 
 void main() {
-    exec_shader();
+    exec_fragment();
 }
 
 )";

src/video_core/renderer_opengl/gl_shader_gen.h

@@ -115,21 +115,48 @@ struct ShaderEntries {
 using ProgramResult = std::pair<std::string, ShaderEntries>;
 
 struct ShaderSetup {
-    ShaderSetup(ProgramCode&& program_code) : program_code(std::move(program_code)) {}
+    ShaderSetup(const ProgramCode& program_code) {
+        program.code = program_code;
+    }
+
+    struct {
+        ProgramCode code;
+        ProgramCode code_b; // Used for dual vertex shaders
+    } program;
 
-    ProgramCode program_code;
     bool program_code_hash_dirty = true;
 
     u64 GetProgramCodeHash() {
         if (program_code_hash_dirty) {
-            program_code_hash = Common::ComputeHash64(&program_code, sizeof(program_code));
+            program_code_hash = GetNewHash();
             program_code_hash_dirty = false;
         }
         return program_code_hash;
     }
 
+    /// Used in scenarios where we have a dual vertex shaders
+    void SetProgramB(const ProgramCode& program_b) {
+        program.code_b = program_b;
+        has_program_b = true;
+    }
+
+    bool IsDualProgram() const {
+        return has_program_b;
+    }
+
 private:
+    u64 GetNewHash() const {
+        if (has_program_b) {
+            // Compute hash over dual shader programs
+            return Common::ComputeHash64(&program, sizeof(program));
+        } else {
+            // Compute hash over a single shader program
+            return Common::ComputeHash64(&program.code, program.code.size());
+        }
+    }
+
     u64 program_code_hash{};
+    bool has_program_b{};
 };
 
 struct MaxwellShaderConfigCommon {

src/video_core/renderer_opengl/gl_shader_manager.cpp

@@ -35,7 +35,7 @@ void SetShaderUniformBlockBindings(GLuint shader) {
 } // namespace Impl
 
 void MaxwellUniformData::SetFromRegs(const Maxwell3D::State::ShaderStageInfo& shader_stage) {
-    const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
+    const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
 
     // TODO(bunnei): Support more than one viewport
     viewport_flip[0] = regs.viewport_transform[0].scale_x < 0.0 ? -1.0f : 1.0f;

src/video_core/renderer_opengl/renderer_opengl.cpp

@@ -92,11 +92,24 @@ static std::array<GLfloat, 3 * 2> MakeOrthographicMatrix(const float width, cons
     return matrix;
 }
 
+ScopeAcquireGLContext::ScopeAcquireGLContext() {
+    if (Settings::values.use_multi_core) {
+        VideoCore::g_emu_window->MakeCurrent();
+    }
+}
+ScopeAcquireGLContext::~ScopeAcquireGLContext() {
+    if (Settings::values.use_multi_core) {
+        VideoCore::g_emu_window->DoneCurrent();
+    }
+}
+
 RendererOpenGL::RendererOpenGL() = default;
 RendererOpenGL::~RendererOpenGL() = default;
 
 /// Swap buffers (render frame)
 void RendererOpenGL::SwapBuffers(boost::optional<const Tegra::FramebufferConfig&> framebuffer) {
+    ScopeAcquireGLContext acquire_context;
+
     Core::System::GetInstance().perf_stats.EndSystemFrame();
 
     // Maintain the rasterizer's state as a priority
@@ -141,6 +154,7 @@ void RendererOpenGL::LoadFBToScreenInfo(const Tegra::FramebufferConfig& framebuf
 
     // Framebuffer orientation handling
     framebuffer_transform_flags = framebuffer.transform_flags;
+    framebuffer_crop_rect = framebuffer.crop_rect;
 
     // Ensure no bad interactions with GL_UNPACK_ALIGNMENT, which by default
     // only allows rows to have a memory alignement of 4.
@@ -307,11 +321,24 @@ void RendererOpenGL::DrawScreenTriangles(const ScreenInfo& screen_info, float x,
         }
     }
 
+    ASSERT_MSG(framebuffer_crop_rect.top == 0, "Unimplemented");
+    ASSERT_MSG(framebuffer_crop_rect.left == 0, "Unimplemented");
+
+    // Scale the output by the crop width/height. This is commonly used with 1280x720 rendering
+    // (e.g. handheld mode) on a 1920x1080 framebuffer.
+    f32 scale_u = 1.f, scale_v = 1.f;
+    if (framebuffer_crop_rect.GetWidth() > 0) {
+        scale_u = static_cast<f32>(framebuffer_crop_rect.GetWidth()) / screen_info.texture.width;
+    }
+    if (framebuffer_crop_rect.GetHeight() > 0) {
+        scale_v = static_cast<f32>(framebuffer_crop_rect.GetHeight()) / screen_info.texture.height;
+    }
+
     std::array<ScreenRectVertex, 4> vertices = {{
-        ScreenRectVertex(x, y, texcoords.top, left),
-        ScreenRectVertex(x + w, y, texcoords.bottom, left),
-        ScreenRectVertex(x, y + h, texcoords.top, right),
-        ScreenRectVertex(x + w, y + h, texcoords.bottom, right),
+        ScreenRectVertex(x, y, texcoords.top * scale_u, left * scale_v),
+        ScreenRectVertex(x + w, y, texcoords.bottom * scale_u, left * scale_v),
+        ScreenRectVertex(x, y + h, texcoords.top * scale_u, right * scale_v),
+        ScreenRectVertex(x + w, y + h, texcoords.bottom * scale_u, right * scale_v),
     }};
 
     state.texture_units[0].texture_2d = screen_info.display_texture;
@@ -418,7 +445,7 @@ static void APIENTRY DebugHandler(GLenum source, GLenum type, GLuint id, GLenum
 
 /// Initialize the renderer
 bool RendererOpenGL::Init() {
-    render_window->MakeCurrent();
+    ScopeAcquireGLContext acquire_context;
 
     if (GLAD_GL_KHR_debug) {
         glEnable(GL_DEBUG_OUTPUT);

src/video_core/renderer_opengl/renderer_opengl.h

@@ -31,6 +31,13 @@ struct ScreenInfo {
     TextureInfo texture;
 };
 
+/// Helper class to acquire/release OpenGL context within a given scope
+class ScopeAcquireGLContext : NonCopyable {
+public:
+    ScopeAcquireGLContext();
+    ~ScopeAcquireGLContext();
+};
+
 class RendererOpenGL : public RendererBase {
 public:
     RendererOpenGL();
@@ -90,4 +97,5 @@ private:
 
     /// Used for transforming the framebuffer orientation
     Tegra::FramebufferConfig::TransformFlags framebuffer_transform_flags;
+    MathUtil::Rectangle<int> framebuffer_crop_rect;
 };

src/video_core/textures/astc.cpp

@@ -25,16 +25,15 @@
 
 class BitStream {
 public:
-    BitStream(unsigned char* ptr, int nBits = 0, int start_offset = 0)
-        : m_BitsWritten(0), m_BitsRead(0), m_NumBits(nBits), m_CurByte(ptr),
-          m_NextBit(start_offset % 8), done(false) {}
+    explicit BitStream(unsigned char* ptr, int nBits = 0, int start_offset = 0)
+        : m_NumBits(nBits), m_CurByte(ptr), m_NextBit(start_offset % 8) {}
+
+    ~BitStream() = default;
 
     int GetBitsWritten() const {
         return m_BitsWritten;
     }
 
-    ~BitStream() {}
-
     void WriteBitsR(unsigned int val, unsigned int nBits) {
         for (unsigned int i = 0; i < nBits; i++) {
             WriteBit((val >> (nBits - i - 1)) & 1);
@@ -95,33 +94,28 @@ private:
         done = done || ++m_BitsWritten >= m_NumBits;
     }
 
-    int m_BitsWritten;
+    int m_BitsWritten = 0;
     const int m_NumBits;
     unsigned char* m_CurByte;
-    int m_NextBit;
-    int m_BitsRead;
+    int m_NextBit = 0;
+    int m_BitsRead = 0;
 
-    bool done;
+    bool done = false;
 };
 
 template <typename IntType>
 class Bits {
-private:
-    const IntType& m_Bits;
-
-    // Don't copy
-    Bits() {}
-    Bits(const Bits&) {}
-    Bits& operator=(const Bits&) {}
-
 public:
-    explicit Bits(IntType& v) : m_Bits(v) {}
+    explicit Bits(const IntType& v) : m_Bits(v) {}
 
-    uint8_t operator[](uint32_t bitPos) {
+    Bits(const Bits&) = delete;
+    Bits& operator=(const Bits&) = delete;
+
+    uint8_t operator[](uint32_t bitPos) const {
         return static_cast<uint8_t>((m_Bits >> bitPos) & 1);
     }
 
-    IntType operator()(uint32_t start, uint32_t end) {
+    IntType operator()(uint32_t start, uint32_t end) const {
         if (start == end) {
             return (*this)[start];
         } else if (start > end) {
@@ -133,6 +127,9 @@ public:
         uint64_t mask = (1 << (end - start + 1)) - 1;
         return (m_Bits >> start) & mask;
     }
+
+private:
+    const IntType& m_Bits;
 };
 
 enum EIntegerEncoding { eIntegerEncoding_JustBits, eIntegerEncoding_Quint, eIntegerEncoding_Trit };
@@ -186,12 +183,12 @@ public:
         m_QuintValue = val;
     }
 
-    bool MatchesEncoding(const IntegerEncodedValue& other) {
+    bool MatchesEncoding(const IntegerEncodedValue& other) const {
         return m_Encoding == other.m_Encoding && m_NumBits == other.m_NumBits;
     }
 
     // Returns the number of bits required to encode nVals values.
-    uint32_t GetBitLength(uint32_t nVals) {
+    uint32_t GetBitLength(uint32_t nVals) const {
         uint32_t totalBits = m_NumBits * nVals;
         if (m_Encoding == eIntegerEncoding_Trit) {
             totalBits += (nVals * 8 + 4) / 5;
@@ -382,19 +379,15 @@ private:
 namespace ASTCC {
 
 struct TexelWeightParams {
-    uint32_t m_Width;
-    uint32_t m_Height;
-    bool m_bDualPlane;
-    uint32_t m_MaxWeight;
-    bool m_bError;
-    bool m_bVoidExtentLDR;
-    bool m_bVoidExtentHDR;
+    uint32_t m_Width = 0;
+    uint32_t m_Height = 0;
+    bool m_bDualPlane = false;
+    uint32_t m_MaxWeight = 0;
+    bool m_bError = false;
+    bool m_bVoidExtentLDR = false;
+    bool m_bVoidExtentHDR = false;
 
-    TexelWeightParams() {
-        memset(this, 0, sizeof(*this));
-    }
-
-    uint32_t GetPackedBitSize() {
+    uint32_t GetPackedBitSize() const {
         // How many indices do we have?
         uint32_t nIdxs = m_Height * m_Width;
         if (m_bDualPlane) {
@@ -413,7 +406,7 @@ struct TexelWeightParams {
     }
 };
 
-TexelWeightParams DecodeBlockInfo(BitStream& strm) {
+static TexelWeightParams DecodeBlockInfo(BitStream& strm) {
     TexelWeightParams params;
 
     // Read the entire block mode all at once
@@ -612,8 +605,8 @@ TexelWeightParams DecodeBlockInfo(BitStream& strm) {
     return params;
 }
 
-void FillVoidExtentLDR(BitStream& strm, uint32_t* const outBuf, uint32_t blockWidth,
-                       uint32_t blockHeight) {
+static void FillVoidExtentLDR(BitStream& strm, uint32_t* const outBuf, uint32_t blockWidth,
+                              uint32_t blockHeight) {
     // Don't actually care about the void extent, just read the bits...
     for (int i = 0; i < 4; ++i) {
         strm.ReadBits(13);
@@ -628,23 +621,25 @@ void FillVoidExtentLDR(BitStream& strm, uint32_t* const outBuf, uint32_t blockWi
     uint32_t rgba = (r >> 8) | (g & 0xFF00) | (static_cast<uint32_t>(b) & 0xFF00) << 8 |
                     (static_cast<uint32_t>(a) & 0xFF00) << 16;
 
-    for (uint32_t j = 0; j < blockHeight; j++)
+    for (uint32_t j = 0; j < blockHeight; j++) {
         for (uint32_t i = 0; i < blockWidth; i++) {
             outBuf[j * blockWidth + i] = rgba;
         }
+    }
 }
 
-void FillError(uint32_t* outBuf, uint32_t blockWidth, uint32_t blockHeight) {
-    for (uint32_t j = 0; j < blockHeight; j++)
+static void FillError(uint32_t* outBuf, uint32_t blockWidth, uint32_t blockHeight) {
+    for (uint32_t j = 0; j < blockHeight; j++) {
         for (uint32_t i = 0; i < blockWidth; i++) {
             outBuf[j * blockWidth + i] = 0xFFFF00FF;
         }
+    }
 }
 
 // Replicates low numBits such that [(toBit - 1):(toBit - 1 - fromBit)]
 // is the same as [(numBits - 1):0] and repeats all the way down.
 template <typename IntType>
-IntType Replicate(const IntType& val, uint32_t numBits, uint32_t toBit) {
+static IntType Replicate(const IntType& val, uint32_t numBits, uint32_t toBit) {
     if (numBits == 0)
         return 0;
     if (toBit == 0)
@@ -668,27 +663,15 @@ IntType Replicate(const IntType& val, uint32_t numBits, uint32_t toBit) {
 
 class Pixel {
 protected:
-    typedef int16_t ChannelType;
-    uint8_t m_BitDepth[4];
-    int16_t color[4];
+    using ChannelType = int16_t;
+    uint8_t m_BitDepth[4] = {8, 8, 8, 8};
+    int16_t color[4] = {};
 
 public:
-    Pixel() {
-        for (int i = 0; i < 4; i++) {
-            m_BitDepth[i] = 8;
-            color[i] = 0;
-        }
-    }
-
-    Pixel(ChannelType a, ChannelType r, ChannelType g, ChannelType b, unsigned bitDepth = 8) {
-        for (int i = 0; i < 4; i++)
-            m_BitDepth[i] = bitDepth;
-
-        color[0] = a;
-        color[1] = r;
-        color[2] = g;
-        color[3] = b;
-    }
+    Pixel() = default;
+    Pixel(ChannelType a, ChannelType r, ChannelType g, ChannelType b, unsigned bitDepth = 8)
+        : m_BitDepth{uint8_t(bitDepth), uint8_t(bitDepth), uint8_t(bitDepth), uint8_t(bitDepth)},
+          color{a, r, g, b} {}
 
     // Changes the depth of each pixel. This scales the values to
     // the appropriate bit depth by either truncating the least
@@ -807,8 +790,8 @@ public:
     }
 };
 
-void DecodeColorValues(uint32_t* out, uint8_t* data, uint32_t* modes, const uint32_t nPartitions,
-                       const uint32_t nBitsForColorData) {
+static void DecodeColorValues(uint32_t* out, uint8_t* data, const uint32_t* modes,
+                              const uint32_t nPartitions, const uint32_t nBitsForColorData) {
     // First figure out how many color values we have
     uint32_t nValues = 0;
     for (uint32_t i = 0; i < nPartitions; i++) {
@@ -844,8 +827,7 @@ void DecodeColorValues(uint32_t* out, uint8_t* data, uint32_t* modes, const uint
     // Once we have the decoded values, we need to dequantize them to the 0-255 range
     // This procedure is outlined in ASTC spec C.2.13
     uint32_t outIdx = 0;
-    std::vector<IntegerEncodedValue>::const_iterator itr;
-    for (itr = decodedColorValues.begin(); itr != decodedColorValues.end(); itr++) {
+    for (auto itr = decodedColorValues.begin(); itr != decodedColorValues.end(); ++itr) {
         // Have we already decoded all that we need?
         if (outIdx >= nValues) {
             break;
@@ -978,7 +960,7 @@ void DecodeColorValues(uint32_t* out, uint8_t* data, uint32_t* modes, const uint
     }
 }
 
-uint32_t UnquantizeTexelWeight(const IntegerEncodedValue& val) {
+static uint32_t UnquantizeTexelWeight(const IntegerEncodedValue& val) {
     uint32_t bitval = val.GetBitValue();
     uint32_t bitlen = val.BaseBitLength();
 
@@ -1067,17 +1049,18 @@ uint32_t UnquantizeTexelWeight(const IntegerEncodedValue& val) {
     return result;
 }
 
-void UnquantizeTexelWeights(uint32_t out[2][144], std::vector<IntegerEncodedValue>& weights,
-                            const TexelWeightParams& params, const uint32_t blockWidth,
-                            const uint32_t blockHeight) {
+static void UnquantizeTexelWeights(uint32_t out[2][144],
+                                   const std::vector<IntegerEncodedValue>& weights,
+                                   const TexelWeightParams& params, const uint32_t blockWidth,
+                                   const uint32_t blockHeight) {
     uint32_t weightIdx = 0;
     uint32_t unquantized[2][144];
-    std::vector<IntegerEncodedValue>::const_iterator itr;
-    for (itr = weights.begin(); itr != weights.end(); itr++) {
+
+    for (auto itr = weights.begin(); itr != weights.end(); ++itr) {
         unquantized[0][weightIdx] = UnquantizeTexelWeight(*itr);
 
         if (params.m_bDualPlane) {
-            itr++;
+            ++itr;
             unquantized[1][weightIdx] = UnquantizeTexelWeight(*itr);
             if (itr == weights.end()) {
                 break;
@@ -1261,8 +1244,8 @@ static inline uint32_t Select2DPartition(int32_t seed, int32_t x, int32_t y, int
 }
 
 // Section C.2.14
-void ComputeEndpoints(Pixel& ep1, Pixel& ep2, const uint32_t*& colorValues,
-                      uint32_t colorEndpointMode) {
+static void ComputeEndpoints(Pixel& ep1, Pixel& ep2, const uint32_t*& colorValues,
+                             uint32_t colorEndpointMode) {
 #define READ_UINT_VALUES(N)                                                                        \
     uint32_t v[N];                                                                                 \
     for (uint32_t i = 0; i < N; i++) {                                                             \
@@ -1382,8 +1365,8 @@ void ComputeEndpoints(Pixel& ep1, Pixel& ep2, const uint32_t*& colorValues,
 #undef READ_INT_VALUES
 }
 
-void DecompressBlock(uint8_t inBuf[16], const uint32_t blockWidth, const uint32_t blockHeight,
-                     uint32_t* outBuf) {
+static void DecompressBlock(uint8_t inBuf[16], const uint32_t blockWidth,
+                            const uint32_t blockHeight, uint32_t* outBuf) {
     BitStream strm(inBuf);
     TexelWeightParams weightParams = DecodeBlockInfo(strm);
 
@@ -1617,8 +1600,7 @@ namespace Tegra::Texture::ASTC {
 std::vector<uint8_t> Decompress(std::vector<uint8_t>& data, uint32_t width, uint32_t height,
                                 uint32_t block_width, uint32_t block_height) {
     uint32_t blockIdx = 0;
-    std::vector<uint8_t> outData;
-    outData.resize(height * width * 4);
+    std::vector<uint8_t> outData(height * width * 4);
     for (uint32_t j = 0; j < height; j += block_height) {
         for (uint32_t i = 0; i < width; i += block_width) {
 

src/video_core/textures/decoders.cpp

@@ -2,6 +2,7 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <cmath>
 #include <cstring>
 #include "common/assert.h"
 #include "core/memory.h"
@@ -17,7 +18,9 @@ namespace Texture {
  * Taken from the Tegra X1 TRM.
  */
 static u32 GetSwizzleOffset(u32 x, u32 y, u32 image_width, u32 bytes_per_pixel, u32 block_height) {
-    u32 image_width_in_gobs = image_width * bytes_per_pixel / 64;
+    // Round up to the next gob
+    const u32 image_width_in_gobs{(image_width * bytes_per_pixel + 63) / 64};
+
     u32 GOB_address = 0 + (y / (8 * block_height)) * 512 * block_height * image_width_in_gobs +
                       (x * bytes_per_pixel / 64) * 512 * block_height +
                       (y % (8 * block_height) / 8) * 512;
@@ -52,6 +55,7 @@ u32 BytesPerPixel(TextureFormat format) {
         return 8;
     case TextureFormat::DXT23:
     case TextureFormat::DXT45:
+    case TextureFormat::BC7U:
         // In this case a 'pixel' actually refers to a 4x4 tile.
         return 16;
     case TextureFormat::ASTC_2D_4X4:
@@ -61,6 +65,7 @@ u32 BytesPerPixel(TextureFormat format) {
         return 4;
     case TextureFormat::A1B5G5R5:
     case TextureFormat::B5G6R5:
+    case TextureFormat::G8R8:
         return 2;
     case TextureFormat::R8:
         return 1;
@@ -76,6 +81,8 @@ u32 BytesPerPixel(TextureFormat format) {
 
 static u32 DepthBytesPerPixel(DepthFormat format) {
     switch (format) {
+    case DepthFormat::Z16_UNORM:
+        return 2;
     case DepthFormat::S8_Z24_UNORM:
     case DepthFormat::Z24_S8_UNORM:
     case DepthFormat::Z32_FLOAT:
@@ -98,6 +105,7 @@ std::vector<u8> UnswizzleTexture(VAddr address, TextureFormat format, u32 width,
     case TextureFormat::DXT23:
     case TextureFormat::DXT45:
     case TextureFormat::DXN1:
+    case TextureFormat::BC7U:
         // In the DXT and DXN formats, each 4x4 tile is swizzled instead of just individual pixel
         // values.
         CopySwizzledData(width / 4, height / 4, bytes_per_pixel, bytes_per_pixel, data,
@@ -108,6 +116,7 @@ std::vector<u8> UnswizzleTexture(VAddr address, TextureFormat format, u32 width,
     case TextureFormat::A1B5G5R5:
     case TextureFormat::B5G6R5:
     case TextureFormat::R8:
+    case TextureFormat::G8R8:
     case TextureFormat::R16_G16_B16_A16:
     case TextureFormat::R32_G32_B32_A32:
     case TextureFormat::BF10GF11RF11:
@@ -131,6 +140,7 @@ std::vector<u8> UnswizzleDepthTexture(VAddr address, DepthFormat format, u32 wid
     std::vector<u8> unswizzled_data(width * height * bytes_per_pixel);
 
     switch (format) {
+    case DepthFormat::Z16_UNORM:
     case DepthFormat::S8_Z24_UNORM:
     case DepthFormat::Z24_S8_UNORM:
     case DepthFormat::Z32_FLOAT:
@@ -155,12 +165,14 @@ std::vector<u8> DecodeTexture(const std::vector<u8>& texture_data, TextureFormat
     case TextureFormat::DXT23:
     case TextureFormat::DXT45:
     case TextureFormat::DXN1:
+    case TextureFormat::BC7U:
     case TextureFormat::ASTC_2D_4X4:
     case TextureFormat::A8R8G8B8:
     case TextureFormat::A2B10G10R10:
     case TextureFormat::A1B5G5R5:
     case TextureFormat::B5G6R5:
     case TextureFormat::R8:
+    case TextureFormat::G8R8:
     case TextureFormat::BF10GF11RF11:
     case TextureFormat::R32_G32_B32_A32:
         // TODO(Subv): For the time being just forward the same data without any decoding.