ldr: Clean up error codes

The separate enum isn't particularly necessary here, and the values can
just be directly put into the ResultCode instances, given the names are
also self-documenting here.

.travis/clang-format/build.sh

@@ -1,3 +1,3 @@
 #!/bin/bash -ex
 
-docker run -v $(pwd):/yuzu ubuntu:18.04 /bin/bash -ex /yuzu/.travis/clang-format/docker.sh
+docker run --env-file .travis/common/travis-ci.env -v $(pwd):/yuzu -v "$HOME/.ccache":/root/.ccache citraemu/build-environments:linux-clang-format /bin/bash -ex /yuzu/.travis/clang-format/docker.sh

.travis/clang-format/deps.sh

@@ -1,3 +1,3 @@
 #!/bin/sh -ex
 
-docker pull ubuntu:18.04
+docker pull citraemu/build-environments:linux-clang-format

.travis/clang-format/docker.sh

@@ -1,8 +1,5 @@
 #!/bin/bash -ex
 
-apt-get update
-apt-get install -y clang-format-6.0
-
 # Run clang-format
 cd /yuzu
 ./.travis/clang-format/script.sh

.travis/linux-mingw/docker.sh

@@ -57,3 +57,4 @@ done
 
 pip3 install pefile
 python3 .travis/linux-mingw/scan_dll.py package/*.exe "package/"
+python3 .travis/linux-mingw/scan_dll.py package/imageformats/*.dll "package/"

appveyor.yml

@@ -125,17 +125,6 @@ after_build:
           Copy-Item -path "$CMAKE_SOURCE_DIR/license.txt" -destination $RELEASE_DIST
           Copy-Item -path "$CMAKE_SOURCE_DIR/README.md" -destination $RELEASE_DIST
 
-          # copy all the dll dependencies to the release folder
-          . "./.appveyor/UtilityFunctions.ps1"
-          $DLLSearchPath = "C:\msys64\mingw64\bin;$env:PATH"
-          $MingwDLLs = RecursivelyGetDeps $DLLSearchPath "$RELEASE_DIST\yuzu.exe"
-          $MingwDLLs += RecursivelyGetDeps $DLLSearchPath  "$RELEASE_DIST\yuzu_cmd.exe"
-          Write-Host "Detected the following dependencies:"
-          Write-Host $MingwDLLs
-          foreach ($file in $MingwDLLs) {
-            Copy-Item -path "$file" -force -destination "$RELEASE_DIST"
-          }
-
           # copy the qt windows plugin dll to platforms
           Copy-Item -path "C:/msys64/mingw64/share/qt5/plugins/platforms/qwindows.dll" -force -destination "$RELEASE_DIST/platforms"
 
@@ -145,6 +134,18 @@ after_build:
           # copy the qt jpeg imageformat dll to platforms
           Copy-Item -path "C:/msys64/mingw64/share/qt5/plugins/imageformats/qjpeg.dll" -force -destination "$RELEASE_DIST/imageformats"
 
+          # copy all the dll dependencies to the release folder
+          . "./.appveyor/UtilityFunctions.ps1"
+          $DLLSearchPath = "C:\msys64\mingw64\bin;$env:PATH"
+          $MingwDLLs = RecursivelyGetDeps $DLLSearchPath "$RELEASE_DIST\yuzu.exe"
+          $MingwDLLs += RecursivelyGetDeps $DLLSearchPath  "$RELEASE_DIST\yuzu_cmd.exe"
+          $MingwDLLs += RecursivelyGetDeps $DLLSearchPath  "$RELEASE_DIST\imageformats\qjpeg.dll"
+          Write-Host "Detected the following dependencies:"
+          Write-Host $MingwDLLs
+          foreach ($file in $MingwDLLs) {
+            Copy-Item -path "$file" -force -destination "$RELEASE_DIST"
+          }
+
           7z a -tzip $MINGW_BUILD_ZIP $RELEASE_DIST\*
           7z a $MINGW_SEVENZIP $RELEASE_DIST
         }

externals/cmake-modules/GetGitRevisionDescription.cmake.in

@@ -33,6 +33,10 @@ else()
 endif()
 
 if(NOT HEAD_HASH)
-	file(READ "@GIT_DATA@/head-ref" HEAD_HASH LIMIT 1024)
-	string(STRIP "${HEAD_HASH}" HEAD_HASH)
+	if(EXISTS "@GIT_DATA@/head-ref")
+		file(READ "@GIT_DATA@/head-ref" HEAD_HASH LIMIT 1024)
+		string(STRIP "${HEAD_HASH}" HEAD_HASH)
+	else()
+		set(HEAD_HASH "Unknown")
+	endif()
 endif()

src/audio_core/audio_renderer.h

@@ -143,7 +143,7 @@ struct AuxInfo {
     std::array<u8, 24> output_mix_buffers;
     u32_le mix_buffer_count;
     u32_le sample_rate; // Stored in the aux buffer currently
-    u32_le sampe_count;
+    u32_le sample_count;
     u64_le send_buffer_info;
     u64_le send_buffer_base;
 

src/audio_core/stream.cpp

@@ -11,7 +11,6 @@
 #include "audio_core/stream.h"
 #include "common/assert.h"
 #include "common/logging/log.h"
-#include "common/microprofile.h"
 #include "core/core_timing.h"
 #include "core/core_timing_util.h"
 #include "core/settings.h"
@@ -104,10 +103,7 @@ void Stream::PlayNextBuffer() {
     CoreTiming::ScheduleEventThreadsafe(GetBufferReleaseCycles(*active_buffer), release_event, {});
 }
 
-MICROPROFILE_DEFINE(AudioOutput, "Audio", "ReleaseActiveBuffer", MP_RGB(100, 100, 255));
-
 void Stream::ReleaseActiveBuffer() {
-    MICROPROFILE_SCOPE(AudioOutput);
     ASSERT(active_buffer);
     released_buffers.push(std::move(active_buffer));
     release_callback();

src/common/bit_field.h

@@ -117,21 +117,21 @@ private:
     // We don't delete it because we want BitField to be trivially copyable.
     constexpr BitField& operator=(const BitField&) = default;
 
-    // StorageType is T for non-enum types and the underlying type of T if
+    // UnderlyingType is T for non-enum types and the underlying type of T if
     // T is an enumeration. Note that T is wrapped within an enable_if in the
     // former case to workaround compile errors which arise when using
     // std::underlying_type<T>::type directly.
-    using StorageType = typename std::conditional_t<std::is_enum<T>::value, std::underlying_type<T>,
-                                                    std::enable_if<true, T>>::type;
+    using UnderlyingType = typename std::conditional_t<std::is_enum_v<T>, std::underlying_type<T>,
+                                                       std::enable_if<true, T>>::type;
 
-    // Unsigned version of StorageType
-    using StorageTypeU = std::make_unsigned_t<StorageType>;
+    // We store the value as the unsigned type to avoid undefined behaviour on value shifting
+    using StorageType = std::make_unsigned_t<UnderlyingType>;
 
 public:
     /// Constants to allow limited introspection of fields if needed
     static constexpr std::size_t position = Position;
     static constexpr std::size_t bits = Bits;
-    static constexpr StorageType mask = (((StorageTypeU)~0) >> (8 * sizeof(T) - bits)) << position;
+    static constexpr StorageType mask = (((StorageType)~0) >> (8 * sizeof(T) - bits)) << position;
 
     /**
      * Formats a value by masking and shifting it according to the field parameters. A value
@@ -148,11 +148,12 @@ public:
      * union in a constexpr context.
      */
     static constexpr FORCE_INLINE T ExtractValue(const StorageType& storage) {
-        if (std::numeric_limits<T>::is_signed) {
+        if constexpr (std::numeric_limits<UnderlyingType>::is_signed) {
             std::size_t shift = 8 * sizeof(T) - bits;
-            return (T)((storage << (shift - position)) >> shift);
+            return static_cast<T>(static_cast<UnderlyingType>(storage << (shift - position)) >>
+                                  shift);
         } else {
-            return (T)((storage & mask) >> position);
+            return static_cast<T>((storage & mask) >> position);
         }
     }
 

src/common/string_util.cpp

@@ -4,11 +4,10 @@
 
 #include <algorithm>
 #include <cctype>
-#include <cerrno>
 #include <codecvt>
-#include <cstdio>
 #include <cstdlib>
-#include <cstring>
+#include <locale>
+#include <sstream>
 #include "common/common_paths.h"
 #include "common/logging/log.h"
 #include "common/string_util.h"
@@ -33,24 +32,6 @@ std::string ToUpper(std::string str) {
     return str;
 }
 
-// For Debugging. Read out an u8 array.
-std::string ArrayToString(const u8* data, std::size_t size, int line_len, bool spaces) {
-    std::ostringstream oss;
-    oss << std::setfill('0') << std::hex;
-
-    for (int line = 0; size; ++data, --size) {
-        oss << std::setw(2) << (int)*data;
-
-        if (line_len == ++line) {
-            oss << '\n';
-            line = 0;
-        } else if (spaces)
-            oss << ' ';
-    }
-
-    return oss.str();
-}
-
 std::string StringFromBuffer(const std::vector<u8>& data) {
     return std::string(data.begin(), std::find(data.begin(), data.end(), '\0'));
 }
@@ -75,40 +56,6 @@ std::string StripQuotes(const std::string& s) {
         return s;
 }
 
-bool TryParse(const std::string& str, u32* const output) {
-    char* endptr = nullptr;
-
-    // Reset errno to a value other than ERANGE
-    errno = 0;
-
-    unsigned long value = strtoul(str.c_str(), &endptr, 0);
-
-    if (!endptr || *endptr)
-        return false;
-
-    if (errno == ERANGE)
-        return false;
-
-#if ULONG_MAX > UINT_MAX
-    if (value >= 0x100000000ull && value <= 0xFFFFFFFF00000000ull)
-        return false;
-#endif
-
-    *output = static_cast<u32>(value);
-    return true;
-}
-
-bool TryParse(const std::string& str, bool* const output) {
-    if ("1" == str || "true" == ToLower(str))
-        *output = true;
-    else if ("0" == str || "false" == ToLower(str))
-        *output = false;
-    else
-        return false;
-
-    return true;
-}
-
 std::string StringFromBool(bool value) {
     return value ? "True" : "False";
 }

src/common/string_util.h

@@ -5,8 +5,6 @@
 #pragma once
 
 #include <cstddef>
-#include <iomanip>
-#include <sstream>
 #include <string>
 #include <vector>
 #include "common/common_types.h"
@@ -19,44 +17,13 @@ std::string ToLower(std::string str);
 /// Make a string uppercase
 std::string ToUpper(std::string str);
 
-std::string ArrayToString(const u8* data, std::size_t size, int line_len = 20, bool spaces = true);
-
 std::string StringFromBuffer(const std::vector<u8>& data);
 
 std::string StripSpaces(const std::string& s);
 std::string StripQuotes(const std::string& s);
 
-// Thousand separator. Turns 12345678 into 12,345,678
-template <typename I>
-std::string ThousandSeparate(I value, int spaces = 0) {
-    std::ostringstream oss;
-
-// std::locale("") seems to be broken on many platforms
-#if defined _WIN32 || (defined __linux__ && !defined __clang__)
-    oss.imbue(std::locale(""));
-#endif
-    oss << std::setw(spaces) << value;
-
-    return oss.str();
-}
-
 std::string StringFromBool(bool value);
 
-bool TryParse(const std::string& str, bool* output);
-bool TryParse(const std::string& str, u32* output);
-
-template <typename N>
-static bool TryParse(const std::string& str, N* const output) {
-    std::istringstream iss(str);
-
-    N tmp = 0;
-    if (iss >> tmp) {
-        *output = tmp;
-        return true;
-    } else
-        return false;
-}
-
 std::string TabsToSpaces(int tab_size, std::string in);
 
 void SplitString(const std::string& str, char delim, std::vector<std::string>& output);

src/core/file_sys/bis_factory.cpp

@@ -8,8 +8,9 @@
 
 namespace FileSys {
 
-BISFactory::BISFactory(VirtualDir nand_root_, VirtualDir load_root_)
+BISFactory::BISFactory(VirtualDir nand_root_, VirtualDir load_root_, VirtualDir dump_root_)
     : nand_root(std::move(nand_root_)), load_root(std::move(load_root_)),
+      dump_root(std::move(dump_root_)),
       sysnand_cache(std::make_unique<RegisteredCache>(
           GetOrCreateDirectoryRelative(nand_root, "/system/Contents/registered"))),
       usrnand_cache(std::make_unique<RegisteredCache>(
@@ -32,4 +33,10 @@ VirtualDir BISFactory::GetModificationLoadRoot(u64 title_id) const {
     return GetOrCreateDirectoryRelative(load_root, fmt::format("/{:016X}", title_id));
 }
 
+VirtualDir BISFactory::GetModificationDumpRoot(u64 title_id) const {
+    if (title_id == 0)
+        return nullptr;
+    return GetOrCreateDirectoryRelative(dump_root, fmt::format("/{:016X}", title_id));
+}
+
 } // namespace FileSys

src/core/file_sys/bis_factory.h

@@ -17,17 +17,19 @@ class RegisteredCache;
 /// registered caches.
 class BISFactory {
 public:
-    explicit BISFactory(VirtualDir nand_root, VirtualDir load_root);
+    explicit BISFactory(VirtualDir nand_root, VirtualDir load_root, VirtualDir dump_root);
     ~BISFactory();
 
     RegisteredCache* GetSystemNANDContents() const;
     RegisteredCache* GetUserNANDContents() const;
 
     VirtualDir GetModificationLoadRoot(u64 title_id) const;
+    VirtualDir GetModificationDumpRoot(u64 title_id) const;
 
 private:
     VirtualDir nand_root;
     VirtualDir load_root;
+    VirtualDir dump_root;
 
     std::unique_ptr<RegisteredCache> sysnand_cache;
     std::unique_ptr<RegisteredCache> usrnand_cache;

src/core/file_sys/card_image.cpp

@@ -176,7 +176,7 @@ Loader::ResultStatus XCI::AddNCAFromPartition(XCIPartition part) {
     for (const VirtualFile& file : partitions[static_cast<std::size_t>(part)]->GetFiles()) {
         if (file->GetExtension() != "nca")
             continue;
-        auto nca = std::make_shared<NCA>(file);
+        auto nca = std::make_shared<NCA>(file, nullptr, 0, keys);
         // TODO(DarkLordZach): Add proper Rev1+ Support
         if (nca->IsUpdate())
             continue;

src/core/file_sys/card_image.h

@@ -9,6 +9,7 @@
 #include <vector>
 #include "common/common_types.h"
 #include "common/swap.h"
+#include "core/crypto/key_manager.h"
 #include "core/file_sys/vfs.h"
 
 namespace Loader {
@@ -107,5 +108,7 @@ private:
     std::shared_ptr<NSP> secure_partition;
     std::shared_ptr<NCA> program;
     std::vector<std::shared_ptr<NCA>> ncas;
+
+    Core::Crypto::KeyManager keys;
 };
 } // namespace FileSys

src/core/file_sys/content_archive.cpp

@@ -101,8 +101,9 @@ static bool IsValidNCA(const NCAHeader& header) {
     return header.magic == Common::MakeMagic('N', 'C', 'A', '3');
 }
 
-NCA::NCA(VirtualFile file_, VirtualFile bktr_base_romfs_, u64 bktr_base_ivfc_offset)
-    : file(std::move(file_)), bktr_base_romfs(std::move(bktr_base_romfs_)) {
+NCA::NCA(VirtualFile file_, VirtualFile bktr_base_romfs_, u64 bktr_base_ivfc_offset,
+         Core::Crypto::KeyManager keys_)
+    : file(std::move(file_)), bktr_base_romfs(std::move(bktr_base_romfs_)), keys(std::move(keys_)) {
     if (file == nullptr) {
         status = Loader::ResultStatus::ErrorNullFile;
         return;

src/core/file_sys/content_archive.h

@@ -79,7 +79,8 @@ inline bool IsDirectoryExeFS(const std::shared_ptr<VfsDirectory>& pfs) {
 class NCA : public ReadOnlyVfsDirectory {
 public:
     explicit NCA(VirtualFile file, VirtualFile bktr_base_romfs = nullptr,
-                 u64 bktr_base_ivfc_offset = 0);
+                 u64 bktr_base_ivfc_offset = 0,
+                 Core::Crypto::KeyManager keys = Core::Crypto::KeyManager());
     ~NCA() override;
 
     Loader::ResultStatus GetStatus() const;

src/core/file_sys/control_metadata.cpp

@@ -66,4 +66,10 @@ std::string NACP::GetVersionString() const {
     return Common::StringFromFixedZeroTerminatedBuffer(raw->version_string.data(),
                                                        raw->version_string.size());
 }
+
+std::vector<u8> NACP::GetRawBytes() const {
+    std::vector<u8> out(sizeof(RawNACP));
+    std::memcpy(out.data(), raw.get(), sizeof(RawNACP));
+    return out;
+}
 } // namespace FileSys

src/core/file_sys/control_metadata.h

@@ -81,6 +81,7 @@ public:
     u64 GetTitleId() const;
     u64 GetDLCBaseTitleId() const;
     std::string GetVersionString() const;
+    std::vector<u8> GetRawBytes() const;
 
 private:
     std::unique_ptr<RawNACP> raw;

src/core/file_sys/errors.h

@@ -8,25 +8,10 @@
 
 namespace FileSys {
 
-namespace ErrCodes {
-enum {
-    NotFound = 1,
-    TitleNotFound = 1002,
-    SdCardNotFound = 2001,
-    RomFSNotFound = 2520,
-};
-}
-
-constexpr ResultCode ERROR_PATH_NOT_FOUND(ErrorModule::FS, ErrCodes::NotFound);
-
-// TODO(bunnei): Replace these with correct errors for Switch OS
-constexpr ResultCode ERROR_INVALID_PATH(-1);
-constexpr ResultCode ERROR_UNSUPPORTED_OPEN_FLAGS(-1);
-constexpr ResultCode ERROR_INVALID_OPEN_FLAGS(-1);
-constexpr ResultCode ERROR_FILE_NOT_FOUND(-1);
-constexpr ResultCode ERROR_UNEXPECTED_FILE_OR_DIRECTORY(-1);
-constexpr ResultCode ERROR_DIRECTORY_ALREADY_EXISTS(-1);
-constexpr ResultCode ERROR_FILE_ALREADY_EXISTS(-1);
-constexpr ResultCode ERROR_DIRECTORY_NOT_EMPTY(-1);
+constexpr ResultCode ERROR_PATH_NOT_FOUND{ErrorModule::FS, 1};
+constexpr ResultCode ERROR_ENTITY_NOT_FOUND{ErrorModule::FS, 1002};
+constexpr ResultCode ERROR_SD_CARD_NOT_FOUND{ErrorModule::FS, 2001};
+constexpr ResultCode ERROR_INVALID_OFFSET{ErrorModule::FS, 6061};
+constexpr ResultCode ERROR_INVALID_SIZE{ErrorModule::FS, 6062};
 
 } // namespace FileSys

src/core/file_sys/patch_manager.cpp

@@ -19,6 +19,7 @@
 #include "core/file_sys/vfs_vector.h"
 #include "core/hle/service/filesystem/filesystem.h"
 #include "core/loader/loader.h"
+#include "core/settings.h"
 
 namespace FileSys {
 
@@ -119,6 +120,18 @@ std::vector<u8> PatchManager::PatchNSO(const std::vector<u8>& nso) const {
     const auto build_id_raw = Common::HexArrayToString(header.build_id);
     const auto build_id = build_id_raw.substr(0, build_id_raw.find_last_not_of('0') + 1);
 
+    if (Settings::values.dump_nso) {
+        LOG_INFO(Loader, "Dumping NSO for build_id={}, title_id={:016X}", build_id, title_id);
+        const auto dump_dir = Service::FileSystem::GetModificationDumpRoot(title_id);
+        if (dump_dir != nullptr) {
+            const auto nso_dir = GetOrCreateDirectoryRelative(dump_dir, "/nso");
+            const auto file = nso_dir->CreateFile(fmt::format("{}.nso", build_id));
+
+            file->Resize(nso.size());
+            file->WriteBytes(nso);
+        }
+    }
+
     LOG_INFO(Loader, "Patching NSO for build_id={}", build_id);
 
     const auto load_dir = Service::FileSystem::GetModificationLoadRoot(title_id);

src/core/file_sys/registered_cache.cpp

@@ -106,9 +106,12 @@ static ContentRecordType GetCRTypeFromNCAType(NCAContentType type) {
 
 VirtualFile RegisteredCache::OpenFileOrDirectoryConcat(const VirtualDir& dir,
                                                        std::string_view path) const {
-    if (dir->GetFileRelative(path) != nullptr)
-        return dir->GetFileRelative(path);
-    if (dir->GetDirectoryRelative(path) != nullptr) {
+    const auto file = dir->GetFileRelative(path);
+    if (file != nullptr)
+        return file;
+
+    const auto nca_dir = dir->GetDirectoryRelative(path);
+    if (nca_dir != nullptr) {
         const auto nca_dir = dir->GetDirectoryRelative(path);
         VirtualFile file = nullptr;
 
@@ -225,7 +228,7 @@ void RegisteredCache::ProcessFiles(const std::vector<NcaID>& ids) {
 
         if (file == nullptr)
             continue;
-        const auto nca = std::make_shared<NCA>(parser(file, id));
+        const auto nca = std::make_shared<NCA>(parser(file, id), nullptr, 0, keys);
         if (nca->GetStatus() != Loader::ResultStatus::Success ||
             nca->GetType() != NCAContentType::Meta) {
             continue;
@@ -315,7 +318,7 @@ std::unique_ptr<NCA> RegisteredCache::GetEntry(u64 title_id, ContentRecordType t
     const auto raw = GetEntryRaw(title_id, type);
     if (raw == nullptr)
         return nullptr;
-    return std::make_unique<NCA>(raw);
+    return std::make_unique<NCA>(raw, nullptr, 0, keys);
 }
 
 std::unique_ptr<NCA> RegisteredCache::GetEntry(RegisteredCacheEntry entry) const {

src/core/file_sys/registered_cache.h

@@ -12,6 +12,7 @@
 #include <vector>
 #include <boost/container/flat_map.hpp>
 #include "common/common_types.h"
+#include "core/crypto/key_manager.h"
 #include "core/file_sys/vfs.h"
 
 namespace FileSys {
@@ -133,6 +134,8 @@ private:
 
     VirtualDir dir;
     RegisteredCacheParsingFunction parser;
+    Core::Crypto::KeyManager keys;
+
     // maps tid -> NcaID of meta
     boost::container::flat_map<u64, NcaID> meta_id;
     // maps tid -> meta

src/core/file_sys/savedata_factory.cpp

@@ -83,6 +83,24 @@ ResultVal<VirtualDir> SaveDataFactory::Open(SaveDataSpaceId space, SaveDataDescr
     return MakeResult<VirtualDir>(std::move(out));
 }
 
+VirtualDir SaveDataFactory::GetSaveDataSpaceDirectory(SaveDataSpaceId space) const {
+    return dir->GetDirectoryRelative(GetSaveDataSpaceIdPath(space));
+}
+
+std::string SaveDataFactory::GetSaveDataSpaceIdPath(SaveDataSpaceId space) {
+    switch (space) {
+    case SaveDataSpaceId::NandSystem:
+        return "/system/";
+    case SaveDataSpaceId::NandUser:
+        return "/user/";
+    case SaveDataSpaceId::TemporaryStorage:
+        return "/temp/";
+    default:
+        ASSERT_MSG(false, "Unrecognized SaveDataSpaceId: {:02X}", static_cast<u8>(space));
+        return "/unrecognized/"; ///< To prevent corruption when ignoring asserts.
+    }
+}
+
 std::string SaveDataFactory::GetFullPath(SaveDataSpaceId space, SaveDataType type, u64 title_id,
                                          u128 user_id, u64 save_id) {
     // According to switchbrew, if a save is of type SaveData and the title id field is 0, it should
@@ -90,21 +108,7 @@ std::string SaveDataFactory::GetFullPath(SaveDataSpaceId space, SaveDataType typ
     if (type == SaveDataType::SaveData && title_id == 0)
         title_id = Core::CurrentProcess()->GetTitleID();
 
-    std::string out;
-
-    switch (space) {
-    case SaveDataSpaceId::NandSystem:
-        out = "/system/";
-        break;
-    case SaveDataSpaceId::NandUser:
-        out = "/user/";
-        break;
-    case SaveDataSpaceId::TemporaryStorage:
-        out = "/temp/";
-        break;
-    default:
-        ASSERT_MSG(false, "Unrecognized SaveDataSpaceId: {:02X}", static_cast<u8>(space));
-    }
+    std::string out = GetSaveDataSpaceIdPath(space);
 
     switch (type) {
     case SaveDataType::SystemSaveData:

src/core/file_sys/savedata_factory.h

@@ -52,6 +52,9 @@ public:
 
     ResultVal<VirtualDir> Open(SaveDataSpaceId space, SaveDataDescriptor meta);
 
+    VirtualDir GetSaveDataSpaceDirectory(SaveDataSpaceId space) const;
+
+    static std::string GetSaveDataSpaceIdPath(SaveDataSpaceId space);
     static std::string GetFullPath(SaveDataSpaceId space, SaveDataType type, u64 title_id,
                                    u128 user_id, u64 save_id);
 

src/core/file_sys/submission_package.cpp

@@ -252,7 +252,7 @@ void NSP::ReadNCAs(const std::vector<VirtualFile>& files) {
                     continue;
                 }
 
-                auto next_nca = std::make_shared<NCA>(next_file);
+                auto next_nca = std::make_shared<NCA>(next_file, nullptr, 0, keys);
                 if (next_nca->GetType() == NCAContentType::Program)
                     program_status[cnmt.GetTitleID()] = next_nca->GetStatus();
                 if (next_nca->GetStatus() == Loader::ResultStatus::Success ||

src/core/file_sys/submission_package.h

@@ -70,6 +70,8 @@ private:
     std::map<u64, std::map<ContentRecordType, std::shared_ptr<NCA>>> ncas;
     std::vector<VirtualFile> ticket_files;
 
+    Core::Crypto::KeyManager keys;
+
     VirtualFile romfs;
     VirtualDir exefs;
 };

src/core/hle/kernel/errors.h

@@ -8,58 +8,28 @@
 
 namespace Kernel {
 
-namespace ErrCodes {
-enum {
-    // Confirmed Switch OS error codes
-    MaxConnectionsReached = 7,
-    InvalidSize = 101,
-    InvalidAddress = 102,
-    HandleTableFull = 105,
-    InvalidMemoryState = 106,
-    InvalidMemoryPermissions = 108,
-    InvalidMemoryRange = 110,
-    InvalidThreadPriority = 112,
-    InvalidProcessorId = 113,
-    InvalidHandle = 114,
-    InvalidPointer = 115,
-    InvalidCombination = 116,
-    Timeout = 117,
-    SynchronizationCanceled = 118,
-    TooLarge = 119,
-    InvalidEnumValue = 120,
-    NoSuchEntry = 121,
-    AlreadyRegistered = 122,
-    SessionClosed = 123,
-    InvalidState = 125,
-    ResourceLimitExceeded = 132,
-};
-}
+// Confirmed Switch kernel error codes
 
-// WARNING: The kernel is quite inconsistent in it's usage of errors code. Make sure to always
-// double check that the code matches before re-using the constant.
-
-constexpr ResultCode ERR_HANDLE_TABLE_FULL(ErrorModule::Kernel, ErrCodes::HandleTableFull);
-constexpr ResultCode ERR_SESSION_CLOSED_BY_REMOTE(ErrorModule::Kernel, ErrCodes::SessionClosed);
-constexpr ResultCode ERR_PORT_NAME_TOO_LONG(ErrorModule::Kernel, ErrCodes::TooLarge);
-constexpr ResultCode ERR_MAX_CONNECTIONS_REACHED(ErrorModule::Kernel,
-                                                 ErrCodes::MaxConnectionsReached);
-constexpr ResultCode ERR_INVALID_ENUM_VALUE(ErrorModule::Kernel, ErrCodes::InvalidEnumValue);
-constexpr ResultCode ERR_INVALID_COMBINATION_KERNEL(ErrorModule::Kernel,
-                                                    ErrCodes::InvalidCombination);
-constexpr ResultCode ERR_INVALID_ADDRESS(ErrorModule::Kernel, ErrCodes::InvalidAddress);
-constexpr ResultCode ERR_INVALID_ADDRESS_STATE(ErrorModule::Kernel, ErrCodes::InvalidMemoryState);
-constexpr ResultCode ERR_INVALID_MEMORY_PERMISSIONS(ErrorModule::Kernel,
-                                                    ErrCodes::InvalidMemoryPermissions);
-constexpr ResultCode ERR_INVALID_MEMORY_RANGE(ErrorModule::Kernel, ErrCodes::InvalidMemoryRange);
-constexpr ResultCode ERR_INVALID_HANDLE(ErrorModule::Kernel, ErrCodes::InvalidHandle);
-constexpr ResultCode ERR_INVALID_PROCESSOR_ID(ErrorModule::Kernel, ErrCodes::InvalidProcessorId);
-constexpr ResultCode ERR_INVALID_SIZE(ErrorModule::Kernel, ErrCodes::InvalidSize);
-constexpr ResultCode ERR_ALREADY_REGISTERED(ErrorModule::Kernel, ErrCodes::AlreadyRegistered);
-constexpr ResultCode ERR_INVALID_STATE(ErrorModule::Kernel, ErrCodes::InvalidState);
-constexpr ResultCode ERR_INVALID_THREAD_PRIORITY(ErrorModule::Kernel,
-                                                 ErrCodes::InvalidThreadPriority);
-constexpr ResultCode ERR_INVALID_POINTER(ErrorModule::Kernel, ErrCodes::InvalidPointer);
-constexpr ResultCode ERR_NOT_FOUND(ErrorModule::Kernel, ErrCodes::NoSuchEntry);
-constexpr ResultCode RESULT_TIMEOUT(ErrorModule::Kernel, ErrCodes::Timeout);
+constexpr ResultCode ERR_MAX_CONNECTIONS_REACHED{ErrorModule::Kernel, 7};
+constexpr ResultCode ERR_INVALID_SIZE{ErrorModule::Kernel, 101};
+constexpr ResultCode ERR_INVALID_ADDRESS{ErrorModule::Kernel, 102};
+constexpr ResultCode ERR_HANDLE_TABLE_FULL{ErrorModule::Kernel, 105};
+constexpr ResultCode ERR_INVALID_ADDRESS_STATE{ErrorModule::Kernel, 106};
+constexpr ResultCode ERR_INVALID_MEMORY_PERMISSIONS{ErrorModule::Kernel, 108};
+constexpr ResultCode ERR_INVALID_MEMORY_RANGE{ErrorModule::Kernel, 110};
+constexpr ResultCode ERR_INVALID_PROCESSOR_ID{ErrorModule::Kernel, 113};
+constexpr ResultCode ERR_INVALID_THREAD_PRIORITY{ErrorModule::Kernel, 112};
+constexpr ResultCode ERR_INVALID_HANDLE{ErrorModule::Kernel, 114};
+constexpr ResultCode ERR_INVALID_POINTER{ErrorModule::Kernel, 115};
+constexpr ResultCode ERR_INVALID_COMBINATION{ErrorModule::Kernel, 116};
+constexpr ResultCode RESULT_TIMEOUT{ErrorModule::Kernel, 117};
+constexpr ResultCode ERR_SYNCHRONIZATION_CANCELED{ErrorModule::Kernel, 118};
+constexpr ResultCode ERR_OUT_OF_RANGE{ErrorModule::Kernel, 119};
+constexpr ResultCode ERR_INVALID_ENUM_VALUE{ErrorModule::Kernel, 120};
+constexpr ResultCode ERR_NOT_FOUND{ErrorModule::Kernel, 121};
+constexpr ResultCode ERR_ALREADY_REGISTERED{ErrorModule::Kernel, 122};
+constexpr ResultCode ERR_SESSION_CLOSED_BY_REMOTE{ErrorModule::Kernel, 123};
+constexpr ResultCode ERR_INVALID_STATE{ErrorModule::Kernel, 125};
+constexpr ResultCode ERR_RESOURCE_LIMIT_EXCEEDED{ErrorModule::Kernel, 132};
 
 } // namespace Kernel

src/core/hle/kernel/process.cpp

@@ -5,11 +5,9 @@
 #include <algorithm>
 #include <memory>
 #include "common/assert.h"
-#include "common/common_funcs.h"
 #include "common/logging/log.h"
 #include "core/core.h"
 #include "core/file_sys/program_metadata.h"
-#include "core/hle/kernel/errors.h"
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/resource_limit.h"
@@ -17,6 +15,7 @@
 #include "core/hle/kernel/thread.h"
 #include "core/hle/kernel/vm_manager.h"
 #include "core/memory.h"
+#include "core/settings.h"
 
 namespace Kernel {
 
@@ -35,6 +34,11 @@ SharedPtr<Process> Process::Create(KernelCore& kernel, std::string&& name) {
     process->process_id = kernel.CreateNewProcessID();
     process->svc_access_mask.set();
 
+    std::mt19937 rng(Settings::values.rng_seed.value_or(0));
+    std::uniform_int_distribution<u64> distribution;
+    std::generate(process->random_entropy.begin(), process->random_entropy.end(),
+                  [&] { return distribution(rng); });
+
     kernel.AppendNewProcess(process);
     return process;
 }
@@ -241,83 +245,15 @@ void Process::LoadModule(CodeSet module_, VAddr base_addr) {
 }
 
 ResultVal<VAddr> Process::HeapAllocate(VAddr target, u64 size, VMAPermission perms) {
-    if (target < vm_manager.GetHeapRegionBaseAddress() ||
-        target + size > vm_manager.GetHeapRegionEndAddress() || target + size < target) {
-        return ERR_INVALID_ADDRESS;
-    }
-
-    if (heap_memory == nullptr) {
-        // Initialize heap
-        heap_memory = std::make_shared<std::vector<u8>>();
-        heap_start = heap_end = target;
-    } else {
-        vm_manager.UnmapRange(heap_start, heap_end - heap_start);
-    }
-
-    // If necessary, expand backing vector to cover new heap extents.
-    if (target < heap_start) {
-        heap_memory->insert(begin(*heap_memory), heap_start - target, 0);
-        heap_start = target;
-        vm_manager.RefreshMemoryBlockMappings(heap_memory.get());
-    }
-    if (target + size > heap_end) {
-        heap_memory->insert(end(*heap_memory), (target + size) - heap_end, 0);
-        heap_end = target + size;
-        vm_manager.RefreshMemoryBlockMappings(heap_memory.get());
-    }
-    ASSERT(heap_end - heap_start == heap_memory->size());
-
-    CASCADE_RESULT(auto vma, vm_manager.MapMemoryBlock(target, heap_memory, target - heap_start,
-                                                       size, MemoryState::Heap));
-    vm_manager.Reprotect(vma, perms);
-
-    heap_used = size;
-
-    return MakeResult<VAddr>(heap_end - size);
+    return vm_manager.HeapAllocate(target, size, perms);
 }
 
 ResultCode Process::HeapFree(VAddr target, u32 size) {
-    if (target < vm_manager.GetHeapRegionBaseAddress() ||
-        target + size > vm_manager.GetHeapRegionEndAddress() || target + size < target) {
-        return ERR_INVALID_ADDRESS;
-    }
-
-    if (size == 0) {
-        return RESULT_SUCCESS;
-    }
-
-    ResultCode result = vm_manager.UnmapRange(target, size);
-    if (result.IsError())
-        return result;
-
-    heap_used -= size;
-
-    return RESULT_SUCCESS;
+    return vm_manager.HeapFree(target, size);
 }
 
-ResultCode Process::MirrorMemory(VAddr dst_addr, VAddr src_addr, u64 size) {
-    auto vma = vm_manager.FindVMA(src_addr);
-
-    ASSERT_MSG(vma != vm_manager.vma_map.end(), "Invalid memory address");
-    ASSERT_MSG(vma->second.backing_block, "Backing block doesn't exist for address");
-
-    // The returned VMA might be a bigger one encompassing the desired address.
-    auto vma_offset = src_addr - vma->first;
-    ASSERT_MSG(vma_offset + size <= vma->second.size,
-               "Shared memory exceeds bounds of mapped block");
-
-    const std::shared_ptr<std::vector<u8>>& backing_block = vma->second.backing_block;
-    std::size_t backing_block_offset = vma->second.offset + vma_offset;
-
-    CASCADE_RESULT(auto new_vma,
-                   vm_manager.MapMemoryBlock(dst_addr, backing_block, backing_block_offset, size,
-                                             MemoryState::Mapped));
-    // Protect mirror with permissions from old region
-    vm_manager.Reprotect(new_vma, vma->second.permissions);
-    // Remove permissions from old region
-    vm_manager.Reprotect(vma, VMAPermission::None);
-
-    return RESULT_SUCCESS;
+ResultCode Process::MirrorMemory(VAddr dst_addr, VAddr src_addr, u64 size, MemoryState state) {
+    return vm_manager.MirrorMemory(dst_addr, src_addr, size, state);
 }
 
 ResultCode Process::UnmapMemory(VAddr dst_addr, VAddr /*src_addr*/, u64 size) {

src/core/hle/kernel/process.h

@@ -8,6 +8,7 @@
 #include <bitset>
 #include <cstddef>
 #include <memory>
+#include <random>
 #include <string>
 #include <vector>
 #include <boost/container/static_vector.hpp>
@@ -119,6 +120,8 @@ struct CodeSet final {
 
 class Process final : public Object {
 public:
+    static constexpr std::size_t RANDOM_ENTROPY_SIZE = 4;
+
     static SharedPtr<Process> Create(KernelCore& kernel, std::string&& name);
 
     std::string GetTypeName() const override {
@@ -212,6 +215,11 @@ public:
         total_process_running_time_ticks += ticks;
     }
 
+    /// Gets 8 bytes of random data for svcGetInfo RandomEntropy
+    u64 GetRandomEntropy(std::size_t index) const {
+        return random_entropy.at(index);
+    }
+
     /**
      * Loads process-specifics configuration info with metadata provided
      * by an executable.
@@ -251,7 +259,8 @@ public:
     ResultVal<VAddr> HeapAllocate(VAddr target, u64 size, VMAPermission perms);
     ResultCode HeapFree(VAddr target, u32 size);
 
-    ResultCode MirrorMemory(VAddr dst_addr, VAddr src_addr, u64 size);
+    ResultCode MirrorMemory(VAddr dst_addr, VAddr src_addr, u64 size,
+                            MemoryState state = MemoryState::Mapped);
 
     ResultCode UnmapMemory(VAddr dst_addr, VAddr src_addr, u64 size);
 
@@ -292,17 +301,6 @@ private:
     u32 allowed_thread_priority_mask = 0xFFFFFFFF;
     u32 is_virtual_address_memory_enabled = 0;
 
-    // Memory used to back the allocations in the regular heap. A single vector is used to cover
-    // the entire virtual address space extents that bound the allocations, including any holes.
-    // This makes deallocation and reallocation of holes fast and keeps process memory contiguous
-    // in the emulator address space, allowing Memory::GetPointer to be reasonably safe.
-    std::shared_ptr<std::vector<u8>> heap_memory;
-
-    // The left/right bounds of the address space covered by heap_memory.
-    VAddr heap_start = 0;
-    VAddr heap_end = 0;
-    u64 heap_used = 0;
-
     /// The Thread Local Storage area is allocated as processes create threads,
     /// each TLS area is 0x200 bytes, so one page (0x1000) is split up in 8 parts, and each part
     /// holds the TLS for a specific thread. This vector contains which parts are in use for each
@@ -321,6 +319,9 @@ private:
     /// Per-process handle table for storing created object handles in.
     HandleTable handle_table;
 
+    /// Random values for svcGetInfo RandomEntropy
+    std::array<u64, RANDOM_ENTROPY_SIZE> random_entropy;
+
     std::string name;
 };
 

src/core/hle/kernel/svc.cpp

@@ -34,6 +34,7 @@
 #include "core/hle/lock.h"
 #include "core/hle/result.h"
 #include "core/hle/service/service.h"
+#include "core/settings.h"
 
 namespace Kernel {
 namespace {
@@ -122,6 +123,48 @@ static ResultCode SetHeapSize(VAddr* heap_addr, u64 heap_size) {
     return RESULT_SUCCESS;
 }
 
+static ResultCode SetMemoryPermission(VAddr addr, u64 size, u32 prot) {
+    LOG_TRACE(Kernel_SVC, "called, addr=0x{:X}, size=0x{:X}, prot=0x{:X}", addr, size, prot);
+
+    if (!Common::Is4KBAligned(addr)) {
+        return ERR_INVALID_ADDRESS;
+    }
+
+    if (size == 0 || !Common::Is4KBAligned(size)) {
+        return ERR_INVALID_SIZE;
+    }
+
+    if (!IsValidAddressRange(addr, size)) {
+        return ERR_INVALID_ADDRESS_STATE;
+    }
+
+    const auto permission = static_cast<MemoryPermission>(prot);
+    if (permission != MemoryPermission::None && permission != MemoryPermission::Read &&
+        permission != MemoryPermission::ReadWrite) {
+        return ERR_INVALID_MEMORY_PERMISSIONS;
+    }
+
+    auto* const current_process = Core::CurrentProcess();
+    auto& vm_manager = current_process->VMManager();
+
+    if (!IsInsideAddressSpace(vm_manager, addr, size)) {
+        return ERR_INVALID_ADDRESS_STATE;
+    }
+
+    const VMManager::VMAHandle iter = vm_manager.FindVMA(addr);
+    if (iter == vm_manager.vma_map.end()) {
+        return ERR_INVALID_ADDRESS_STATE;
+    }
+
+    LOG_WARNING(Kernel_SVC, "Uniformity check on protected memory is not implemented.");
+    // TODO: Performs a uniformity check to make sure only protected memory is changed (it doesn't
+    // make sense to allow changing permissions on kernel memory itself, etc).
+
+    const auto converted_permissions = SharedMemory::ConvertPermissions(permission);
+
+    return vm_manager.ReprotectRange(addr, size, converted_permissions);
+}
+
 static ResultCode SetMemoryAttribute(VAddr addr, u64 size, u32 state0, u32 state1) {
     LOG_WARNING(Kernel_SVC,
                 "(STUBBED) called, addr=0x{:X}, size=0x{:X}, state0=0x{:X}, state1=0x{:X}", addr,
@@ -171,7 +214,7 @@ static ResultCode ConnectToNamedPort(Handle* out_handle, VAddr port_name_address
     // Read 1 char beyond the max allowed port name to detect names that are too long.
     std::string port_name = Memory::ReadCString(port_name_address, PortNameMaxLength + 1);
     if (port_name.size() > PortNameMaxLength) {
-        return ERR_PORT_NAME_TOO_LONG;
+        return ERR_OUT_OF_RANGE;
     }
 
     LOG_TRACE(Kernel_SVC, "called port_name={}", port_name);
@@ -267,8 +310,9 @@ static ResultCode WaitSynchronization(Handle* index, VAddr handles_address, u64
 
     static constexpr u64 MaxHandles = 0x40;
 
-    if (handle_count > MaxHandles)
-        return ResultCode(ErrorModule::Kernel, ErrCodes::TooLarge);
+    if (handle_count > MaxHandles) {
+        return ERR_OUT_OF_RANGE;
+    }
 
     auto* const thread = GetCurrentThread();
 
@@ -333,8 +377,7 @@ static ResultCode CancelSynchronization(Handle thread_handle) {
     }
 
     ASSERT(thread->GetStatus() == ThreadStatus::WaitSynchAny);
-    thread->SetWaitSynchronizationResult(
-        ResultCode(ErrorModule::Kernel, ErrCodes::SynchronizationCanceled));
+    thread->SetWaitSynchronizationResult(ERR_SYNCHRONIZATION_CANCELED);
     thread->ResumeFromWait();
     return RESULT_SUCCESS;
 }
@@ -395,16 +438,42 @@ struct BreakReason {
 /// Break program execution
 static void Break(u32 reason, u64 info1, u64 info2) {
     BreakReason break_reason{reason};
+    bool has_dumped_buffer{};
 
+    const auto handle_debug_buffer = [&](VAddr addr, u64 sz) {
+        if (sz == 0 || addr == 0 || has_dumped_buffer) {
+            return;
+        }
+
+        // This typically is an error code so we're going to assume this is the case
+        if (sz == sizeof(u32)) {
+            LOG_CRITICAL(Debug_Emulated, "debug_buffer_err_code={:X}", Memory::Read32(addr));
+        } else {
+            // We don't know what's in here so we'll hexdump it
+            std::vector<u8> debug_buffer(sz);
+            Memory::ReadBlock(addr, debug_buffer.data(), sz);
+            std::string hexdump;
+            for (std::size_t i = 0; i < debug_buffer.size(); i++) {
+                hexdump += fmt::format("{:02X} ", debug_buffer[i]);
+                if (i != 0 && i % 16 == 0) {
+                    hexdump += '\n';
+                }
+            }
+            LOG_CRITICAL(Debug_Emulated, "debug_buffer=\n{}", hexdump);
+        }
+        has_dumped_buffer = true;
+    };
     switch (break_reason.break_type) {
     case BreakType::Panic:
         LOG_CRITICAL(Debug_Emulated, "Signalling debugger, PANIC! info1=0x{:016X}, info2=0x{:016X}",
                      info1, info2);
+        handle_debug_buffer(info1, info2);
         break;
     case BreakType::AssertionFailed:
         LOG_CRITICAL(Debug_Emulated,
                      "Signalling debugger, Assertion failed! info1=0x{:016X}, info2=0x{:016X}",
                      info1, info2);
+        handle_debug_buffer(info1, info2);
         break;
     case BreakType::PreNROLoad:
         LOG_WARNING(
@@ -433,6 +502,7 @@ static void Break(u32 reason, u64 info1, u64 info2) {
             Debug_Emulated,
             "Signalling debugger, Unknown break reason {}, info1=0x{:016X}, info2=0x{:016X}",
             static_cast<u32>(break_reason.break_type.Value()), info1, info2);
+        handle_debug_buffer(info1, info2);
         break;
     }
 
@@ -441,6 +511,7 @@ static void Break(u32 reason, u64 info1, u64 info2) {
             Debug_Emulated,
             "Emulated program broke execution! reason=0x{:016X}, info1=0x{:016X}, info2=0x{:016X}",
             reason, info1, info2);
+        handle_debug_buffer(info1, info2);
         ASSERT(false);
 
         Core::CurrentProcess()->PrepareForTermination();
@@ -530,7 +601,16 @@ static ResultCode GetInfo(u64* result, u64 info_id, u64 handle, u64 info_sub_id)
         *result = 0;
         break;
     case GetInfoType::RandomEntropy:
-        *result = 0;
+        if (handle != 0) {
+            return ERR_INVALID_HANDLE;
+        }
+
+        if (info_sub_id >= Process::RANDOM_ENTROPY_SIZE) {
+            return ERR_INVALID_COMBINATION;
+        }
+
+        *result = current_process->GetRandomEntropy(info_sub_id);
+        return RESULT_SUCCESS;
         break;
     case GetInfoType::ASLRRegionBaseAddr:
         *result = vm_manager.GetASLRRegionBaseAddress();
@@ -563,7 +643,7 @@ static ResultCode GetInfo(u64* result, u64 info_id, u64 handle, u64 info_sub_id)
     case GetInfoType::ThreadTickCount: {
         constexpr u64 num_cpus = 4;
         if (info_sub_id != 0xFFFFFFFFFFFFFFFF && info_sub_id >= num_cpus) {
-            return ERR_INVALID_COMBINATION_KERNEL;
+            return ERR_INVALID_COMBINATION;
         }
 
         const auto thread =
@@ -1101,7 +1181,7 @@ static ResultCode CloseHandle(Handle handle) {
 
 /// Reset an event
 static ResultCode ResetSignal(Handle handle) {
-    LOG_WARNING(Kernel_SVC, "(STUBBED) called handle 0x{:08X}", handle);
+    LOG_DEBUG(Kernel_SVC, "called handle 0x{:08X}", handle);
 
     const auto& handle_table = Core::CurrentProcess()->GetHandleTable();
     auto event = handle_table.Get<Event>(handle);
@@ -1156,7 +1236,7 @@ static ResultCode SetThreadCoreMask(Handle thread_handle, u32 core, u64 mask) {
     }
 
     if (mask == 0) {
-        return ResultCode(ErrorModule::Kernel, ErrCodes::InvalidCombination);
+        return ERR_INVALID_COMBINATION;
     }
 
     /// This value is used to only change the affinity mask without changing the current ideal core.
@@ -1165,12 +1245,12 @@ static ResultCode SetThreadCoreMask(Handle thread_handle, u32 core, u64 mask) {
     if (core == OnlyChangeMask) {
         core = thread->GetIdealCore();
     } else if (core >= Core::NUM_CPU_CORES && core != static_cast<u32>(-1)) {
-        return ResultCode(ErrorModule::Kernel, ErrCodes::InvalidProcessorId);
+        return ERR_INVALID_PROCESSOR_ID;
     }
 
     // Error out if the input core isn't enabled in the input mask.
     if (core < Core::NUM_CPU_CORES && (mask & (1ull << core)) == 0) {
-        return ResultCode(ErrorModule::Kernel, ErrCodes::InvalidCombination);
+        return ERR_INVALID_COMBINATION;
     }
 
     thread->ChangeCore(core, mask);
@@ -1259,7 +1339,7 @@ struct FunctionDef {
 static const FunctionDef SVC_Table[] = {
     {0x00, nullptr, "Unknown"},
     {0x01, SvcWrap<SetHeapSize>, "SetHeapSize"},
-    {0x02, nullptr, "SetMemoryPermission"},
+    {0x02, SvcWrap<SetMemoryPermission>, "SetMemoryPermission"},
     {0x03, SvcWrap<SetMemoryAttribute>, "SetMemoryAttribute"},
     {0x04, SvcWrap<MapMemory>, "MapMemory"},
     {0x05, SvcWrap<UnmapMemory>, "UnmapMemory"},

src/core/hle/kernel/svc_wrap.h

@@ -121,6 +121,11 @@ void SvcWrap() {
     FuncReturn(func(Param(0), Param(1), Param(2)).raw);
 }
 
+template <ResultCode func(u64, u64, u32)>
+void SvcWrap() {
+    FuncReturn(func(Param(0), Param(1), static_cast<u32>(Param(2))).raw);
+}
+
 template <ResultCode func(u32, u64, u64, u32)>
 void SvcWrap() {
     FuncReturn(

src/core/hle/kernel/thread.cpp

@@ -142,36 +142,7 @@ void Thread::ResumeFromWait() {
 
     status = ThreadStatus::Ready;
 
-    std::optional<s32> new_processor_id = GetNextProcessorId(affinity_mask);
-    if (!new_processor_id) {
-        new_processor_id = processor_id;
-    }
-    if (ideal_core != -1 &&
-        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);
-
-    if (*new_processor_id != processor_id) {
-        // Remove thread from previous core's scheduler
-        scheduler->RemoveThread(this);
-        next_scheduler->AddThread(this, current_priority);
-    }
-
-    processor_id = *new_processor_id;
-
-    // If the thread was ready, unschedule from the previous core and schedule on the new core
-    scheduler->UnscheduleThread(this, current_priority);
-    next_scheduler->ScheduleThread(this, current_priority);
-
-    // Change thread's scheduler
-    scheduler = next_scheduler;
-
-    Core::System::GetInstance().CpuCore(processor_id).PrepareReschedule();
+    ChangeScheduler();
 }
 
 /**
@@ -364,42 +335,45 @@ void Thread::UpdatePriority() {
 void Thread::ChangeCore(u32 core, u64 mask) {
     ideal_core = core;
     affinity_mask = mask;
+    ChangeScheduler();
+}
 
+void Thread::ChangeScheduler() {
     if (status != ThreadStatus::Ready) {
         return;
     }
 
+    auto& system = Core::System::GetInstance();
     std::optional<s32> new_processor_id{GetNextProcessorId(affinity_mask)};
 
     if (!new_processor_id) {
         new_processor_id = processor_id;
     }
-    if (ideal_core != -1 &&
-        Core::System::GetInstance().Scheduler(ideal_core).GetCurrentThread() == nullptr) {
+    if (ideal_core != -1 && system.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 = system.Scheduler(*new_processor_id);
 
     if (*new_processor_id != processor_id) {
         // Remove thread from previous core's scheduler
         scheduler->RemoveThread(this);
-        next_scheduler->AddThread(this, current_priority);
+        next_scheduler.AddThread(this, current_priority);
     }
 
     processor_id = *new_processor_id;
 
     // If the thread was ready, unschedule from the previous core and schedule on the new core
     scheduler->UnscheduleThread(this, current_priority);
-    next_scheduler->ScheduleThread(this, current_priority);
+    next_scheduler.ScheduleThread(this, current_priority);
 
     // Change thread's scheduler
-    scheduler = next_scheduler;
+    scheduler = &next_scheduler;
 
-    Core::System::GetInstance().CpuCore(processor_id).PrepareReschedule();
+    system.CpuCore(processor_id).PrepareReschedule();
 }
 
 bool Thread::AllWaitObjectsReady() {

src/core/hle/kernel/thread.h

@@ -374,6 +374,8 @@ private:
     explicit Thread(KernelCore& kernel);
     ~Thread() override;
 
+    void ChangeScheduler();
+
     Core::ARM_Interface::ThreadContext context{};
 
     u32 thread_id = 0;

src/core/hle/kernel/vm_manager.cpp

@@ -243,6 +243,85 @@ ResultCode VMManager::ReprotectRange(VAddr target, u64 size, VMAPermission new_p
     return RESULT_SUCCESS;
 }
 
+ResultVal<VAddr> VMManager::HeapAllocate(VAddr target, u64 size, VMAPermission perms) {
+    if (target < GetHeapRegionBaseAddress() || target + size > GetHeapRegionEndAddress() ||
+        target + size < target) {
+        return ERR_INVALID_ADDRESS;
+    }
+
+    if (heap_memory == nullptr) {
+        // Initialize heap
+        heap_memory = std::make_shared<std::vector<u8>>();
+        heap_start = heap_end = target;
+    } else {
+        UnmapRange(heap_start, heap_end - heap_start);
+    }
+
+    // If necessary, expand backing vector to cover new heap extents.
+    if (target < heap_start) {
+        heap_memory->insert(begin(*heap_memory), heap_start - target, 0);
+        heap_start = target;
+        RefreshMemoryBlockMappings(heap_memory.get());
+    }
+    if (target + size > heap_end) {
+        heap_memory->insert(end(*heap_memory), (target + size) - heap_end, 0);
+        heap_end = target + size;
+        RefreshMemoryBlockMappings(heap_memory.get());
+    }
+    ASSERT(heap_end - heap_start == heap_memory->size());
+
+    CASCADE_RESULT(auto vma, MapMemoryBlock(target, heap_memory, target - heap_start, size,
+                                            MemoryState::Heap));
+    Reprotect(vma, perms);
+
+    heap_used = size;
+
+    return MakeResult<VAddr>(heap_end - size);
+}
+
+ResultCode VMManager::HeapFree(VAddr target, u64 size) {
+    if (target < GetHeapRegionBaseAddress() || target + size > GetHeapRegionEndAddress() ||
+        target + size < target) {
+        return ERR_INVALID_ADDRESS;
+    }
+
+    if (size == 0) {
+        return RESULT_SUCCESS;
+    }
+
+    const ResultCode result = UnmapRange(target, size);
+    if (result.IsError()) {
+        return result;
+    }
+
+    heap_used -= size;
+    return RESULT_SUCCESS;
+}
+
+ResultCode VMManager::MirrorMemory(VAddr dst_addr, VAddr src_addr, u64 size, MemoryState state) {
+    const auto vma = FindVMA(src_addr);
+
+    ASSERT_MSG(vma != vma_map.end(), "Invalid memory address");
+    ASSERT_MSG(vma->second.backing_block, "Backing block doesn't exist for address");
+
+    // The returned VMA might be a bigger one encompassing the desired address.
+    const auto vma_offset = src_addr - vma->first;
+    ASSERT_MSG(vma_offset + size <= vma->second.size,
+               "Shared memory exceeds bounds of mapped block");
+
+    const std::shared_ptr<std::vector<u8>>& backing_block = vma->second.backing_block;
+    const std::size_t backing_block_offset = vma->second.offset + vma_offset;
+
+    CASCADE_RESULT(auto new_vma,
+                   MapMemoryBlock(dst_addr, backing_block, backing_block_offset, size, state));
+    // Protect mirror with permissions from old region
+    Reprotect(new_vma, vma->second.permissions);
+    // Remove permissions from old region
+    Reprotect(vma, VMAPermission::None);
+
+    return RESULT_SUCCESS;
+}
+
 void VMManager::RefreshMemoryBlockMappings(const std::vector<u8>* block) {
     // If this ever proves to have a noticeable performance impact, allow users of the function to
     // specify a specific range of addresses to limit the scan to.
@@ -495,8 +574,7 @@ u64 VMManager::GetTotalMemoryUsage() const {
 }
 
 u64 VMManager::GetTotalHeapUsage() const {
-    LOG_WARNING(Kernel, "(STUBBED) called");
-    return 0x0;
+    return heap_used;
 }
 
 VAddr VMManager::GetAddressSpaceBaseAddress() const {

src/core/hle/kernel/vm_manager.h

@@ -186,6 +186,12 @@ public:
     /// Changes the permissions of a range of addresses, splitting VMAs as necessary.
     ResultCode ReprotectRange(VAddr target, u64 size, VMAPermission new_perms);
 
+    ResultVal<VAddr> HeapAllocate(VAddr target, u64 size, VMAPermission perms);
+    ResultCode HeapFree(VAddr target, u64 size);
+
+    ResultCode MirrorMemory(VAddr dst_addr, VAddr src_addr, u64 size,
+                            MemoryState state = MemoryState::Mapped);
+
     /**
      * Scans all VMAs and updates the page table range of any that use the given vector as backing
      * memory. This should be called after any operation that causes reallocation of the vector.
@@ -343,5 +349,15 @@ private:
 
     VAddr tls_io_region_base = 0;
     VAddr tls_io_region_end = 0;
+
+    // Memory used to back the allocations in the regular heap. A single vector is used to cover
+    // the entire virtual address space extents that bound the allocations, including any holes.
+    // This makes deallocation and reallocation of holes fast and keeps process memory contiguous
+    // in the emulator address space, allowing Memory::GetPointer to be reasonably safe.
+    std::shared_ptr<std::vector<u8>> heap_memory;
+    // The left/right bounds of the address space covered by heap_memory.
+    VAddr heap_start = 0;
+    VAddr heap_end = 0;
+    u64 heap_used = 0;
 };
 } // namespace Kernel

src/core/hle/result.h

@@ -19,8 +19,6 @@
 enum class ErrorDescription : u32 {
     Success = 0,
     RemoteProcessDead = 301,
-    InvalidOffset = 6061,
-    InvalidLength = 6062,
 };
 
 /**

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

@@ -242,6 +242,30 @@ void Module::Interface::GetBaasAccountManagerForApplication(Kernel::HLERequestCo
     LOG_DEBUG(Service_ACC, "called");
 }
 
+void Module::Interface::TrySelectUserWithoutInteraction(Kernel::HLERequestContext& ctx) {
+    LOG_DEBUG(Service_ACC, "called");
+    // A u8 is passed into this function which we can safely ignore. It's to determine if we have
+    // access to use the network or not by the looks of it
+    IPC::ResponseBuilder rb{ctx, 6};
+    if (profile_manager->GetUserCount() != 1) {
+        rb.Push(RESULT_SUCCESS);
+        rb.PushRaw<u128>(INVALID_UUID);
+        return;
+    }
+
+    const auto user_list = profile_manager->GetAllUsers();
+    if (std::all_of(user_list.begin(), user_list.end(),
+                    [](const auto& user) { return user.uuid == INVALID_UUID; })) {
+        rb.Push(ResultCode(-1)); // TODO(ogniK): Find the correct error code
+        rb.PushRaw<u128>(INVALID_UUID);
+        return;
+    }
+
+    // Select the first user we have
+    rb.Push(RESULT_SUCCESS);
+    rb.PushRaw<u128>(profile_manager->GetUser(0)->uuid);
+}
+
 Module::Interface::Interface(std::shared_ptr<Module> module,
                              std::shared_ptr<ProfileManager> profile_manager, const char* name)
     : ServiceFramework(name), module(std::move(module)),

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

@@ -27,6 +27,7 @@ public:
         void InitializeApplicationInfo(Kernel::HLERequestContext& ctx);
         void GetBaasAccountManagerForApplication(Kernel::HLERequestContext& ctx);
         void IsUserRegistrationRequestPermitted(Kernel::HLERequestContext& ctx);
+        void TrySelectUserWithoutInteraction(Kernel::HLERequestContext& ctx);
 
     protected:
         std::shared_ptr<Module> module;

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

@@ -17,7 +17,7 @@ ACC_SU::ACC_SU(std::shared_ptr<Module> module, std::shared_ptr<ProfileManager> p
         {5, &ACC_SU::GetProfile, "GetProfile"},
         {6, nullptr, "GetProfileDigest"},
         {50, &ACC_SU::IsUserRegistrationRequestPermitted, "IsUserRegistrationRequestPermitted"},
-        {51, nullptr, "TrySelectUserWithoutInteraction"},
+        {51, &ACC_SU::TrySelectUserWithoutInteraction, "TrySelectUserWithoutInteraction"},
         {60, nullptr, "ListOpenContextStoredUsers"},
         {100, nullptr, "GetUserRegistrationNotifier"},
         {101, nullptr, "GetUserStateChangeNotifier"},

src/core/hle/service/acc/acc_u0.cpp

@@ -17,7 +17,7 @@ ACC_U0::ACC_U0(std::shared_ptr<Module> module, std::shared_ptr<ProfileManager> p
         {5, &ACC_U0::GetProfile, "GetProfile"},
         {6, nullptr, "GetProfileDigest"},
         {50, &ACC_U0::IsUserRegistrationRequestPermitted, "IsUserRegistrationRequestPermitted"},
-        {51, nullptr, "TrySelectUserWithoutInteraction"},
+        {51, &ACC_U0::TrySelectUserWithoutInteraction, "TrySelectUserWithoutInteraction"},
         {60, nullptr, "ListOpenContextStoredUsers"},
         {100, &ACC_U0::InitializeApplicationInfo, "InitializeApplicationInfo"},
         {101, &ACC_U0::GetBaasAccountManagerForApplication, "GetBaasAccountManagerForApplication"},

src/core/hle/service/acc/acc_u1.cpp

@@ -17,7 +17,7 @@ ACC_U1::ACC_U1(std::shared_ptr<Module> module, std::shared_ptr<ProfileManager> p
         {5, &ACC_U1::GetProfile, "GetProfile"},
         {6, nullptr, "GetProfileDigest"},
         {50, &ACC_U1::IsUserRegistrationRequestPermitted, "IsUserRegistrationRequestPermitted"},
-        {51, nullptr, "TrySelectUserWithoutInteraction"},
+        {51, &ACC_U1::TrySelectUserWithoutInteraction, "TrySelectUserWithoutInteraction"},
         {60, nullptr, "ListOpenContextStoredUsers"},
         {100, nullptr, "GetUserRegistrationNotifier"},
         {101, nullptr, "GetUserStateChangeNotifier"},

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

@@ -2,8 +2,11 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <cstring>
 #include <random>
 
+#include <fmt/format.h>
+
 #include "common/file_util.h"
 #include "core/hle/service/acc/profile_manager.h"
 #include "core/settings.h"
@@ -39,6 +42,19 @@ UUID UUID::Generate() {
     return UUID{distribution(gen), distribution(gen)};
 }
 
+std::string UUID::Format() const {
+    return fmt::format("0x{:016X}{:016X}", uuid[1], uuid[0]);
+}
+
+std::string UUID::FormatSwitch() const {
+    std::array<u8, 16> s{};
+    std::memcpy(s.data(), uuid.data(), sizeof(u128));
+    return fmt::format("{:02x}{:02x}{:02x}{:02x}-{:02x}{:02x}-{:02x}{:02x}-{:02x}{:02x}-{:02x}{"
+                       ":02x}{:02x}{:02x}{:02x}{:02x}",
+                       s[0], s[1], s[2], s[3], s[4], s[5], s[6], s[7], s[8], s[9], s[10], s[11],
+                       s[12], s[13], s[14], s[15]);
+}
+
 ProfileManager::ProfileManager() {
     ParseUserSaveFile();
 
@@ -325,11 +341,12 @@ void ProfileManager::ParseUserSaveFile() {
         return;
     }
 
-    for (std::size_t i = 0; i < MAX_USERS; ++i) {
-        const auto& user = data.users[i];
+    for (const auto& user : data.users) {
+        if (user.uuid == UUID(INVALID_UUID)) {
+            continue;
+        }
 
-        if (user.uuid != UUID(INVALID_UUID))
-            AddUser({user.uuid, user.username, user.timestamp, {}, false});
+        AddUser({user.uuid, user.username, user.timestamp, {}, false});
     }
 
     std::stable_partition(profiles.begin(), profiles.end(),

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

@@ -42,18 +42,9 @@ struct UUID {
     void Invalidate() {
         uuid = INVALID_UUID;
     }
-    std::string Format() const {
-        return fmt::format("0x{:016X}{:016X}", uuid[1], uuid[0]);
-    }
 
-    std::string FormatSwitch() const {
-        std::array<u8, 16> s{};
-        std::memcpy(s.data(), uuid.data(), sizeof(u128));
-        return fmt::format("{:02x}{:02x}{:02x}{:02x}-{:02x}{:02x}-{:02x}{:02x}-{:02x}{:02x}-{:02x}{"
-                           ":02x}{:02x}{:02x}{:02x}{:02x}",
-                           s[0], s[1], s[2], s[3], s[4], s[5], s[6], s[7], s[8], s[9], s[10], s[11],
-                           s[12], s[13], s[14], s[15]);
-    }
+    std::string Format() const;
+    std::string FormatSwitch() const;
 };
 static_assert(sizeof(UUID) == 16, "UUID is an invalid size!");
 

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

@@ -203,8 +203,8 @@ ISelfController::ISelfController(std::shared_ptr<NVFlinger::NVFlinger> nvflinger
 ISelfController::~ISelfController() = default;
 
 void ISelfController::SetFocusHandlingMode(Kernel::HLERequestContext& ctx) {
-    // Takes 3 input u8s with each field located immediately after the previous u8, these are
-    // bool flags. No output.
+    // Takes 3 input u8s with each field located immediately after the previous
+    // u8, these are bool flags. No output.
 
     IPC::RequestParser rp{ctx};
 
@@ -258,8 +258,8 @@ void ISelfController::SetOperationModeChangedNotification(Kernel::HLERequestCont
 }
 
 void ISelfController::SetOutOfFocusSuspendingEnabled(Kernel::HLERequestContext& ctx) {
-    // Takes 3 input u8s with each field located immediately after the previous u8, these are
-    // bool flags. No output.
+    // Takes 3 input u8s with each field located immediately after the previous
+    // u8, these are bool flags. No output.
     IPC::RequestParser rp{ctx};
 
     bool enabled = rp.Pop<bool>();
@@ -302,8 +302,8 @@ void ISelfController::SetScreenShotImageOrientation(Kernel::HLERequestContext& c
 }
 
 void ISelfController::CreateManagedDisplayLayer(Kernel::HLERequestContext& ctx) {
-    // TODO(Subv): Find out how AM determines the display to use, for now just create the layer
-    // in the Default display.
+    // TODO(Subv): Find out how AM determines the display to use, for now just
+    // create the layer in the Default display.
     u64 display_id = nvflinger->OpenDisplay("Default");
     u64 layer_id = nvflinger->CreateLayer(display_id);
 
@@ -338,7 +338,54 @@ void ISelfController::GetIdleTimeDetectionExtension(Kernel::HLERequestContext& c
     LOG_WARNING(Service_AM, "(STUBBED) called");
 }
 
-ICommonStateGetter::ICommonStateGetter() : ServiceFramework("ICommonStateGetter") {
+AppletMessageQueue::AppletMessageQueue() {
+    auto& kernel = Core::System::GetInstance().Kernel();
+    on_new_message = Kernel::Event::Create(kernel, Kernel::ResetType::Sticky,
+                                           "AMMessageQueue:OnMessageRecieved");
+    on_operation_mode_changed = Kernel::Event::Create(kernel, Kernel::ResetType::OneShot,
+                                                      "AMMessageQueue:OperationModeChanged");
+}
+
+AppletMessageQueue::~AppletMessageQueue() = default;
+
+const Kernel::SharedPtr<Kernel::Event>& AppletMessageQueue::GetMesssageRecieveEvent() const {
+    return on_new_message;
+}
+
+const Kernel::SharedPtr<Kernel::Event>& AppletMessageQueue::GetOperationModeChangedEvent() const {
+    return on_operation_mode_changed;
+}
+
+void AppletMessageQueue::PushMessage(AppletMessage msg) {
+    messages.push(msg);
+    on_new_message->Signal();
+}
+
+AppletMessageQueue::AppletMessage AppletMessageQueue::PopMessage() {
+    if (messages.empty()) {
+        on_new_message->Clear();
+        return AppletMessage::NoMessage;
+    }
+    auto msg = messages.front();
+    messages.pop();
+    if (messages.empty()) {
+        on_new_message->Clear();
+    }
+    return msg;
+}
+
+std::size_t AppletMessageQueue::GetMessageCount() const {
+    return messages.size();
+}
+
+void AppletMessageQueue::OperationModeChanged() {
+    PushMessage(AppletMessage::OperationModeChanged);
+    PushMessage(AppletMessage::PerformanceModeChanged);
+    on_operation_mode_changed->Signal();
+}
+
+ICommonStateGetter::ICommonStateGetter(std::shared_ptr<AppletMessageQueue> msg_queue)
+    : ServiceFramework("ICommonStateGetter"), msg_queue(std::move(msg_queue)) {
     // clang-format off
     static const FunctionInfo functions[] = {
         {0, &ICommonStateGetter::GetEventHandle, "GetEventHandle"},
@@ -388,21 +435,19 @@ void ICommonStateGetter::GetBootMode(Kernel::HLERequestContext& ctx) {
 }
 
 void ICommonStateGetter::GetEventHandle(Kernel::HLERequestContext& ctx) {
-    event->Signal();
-
     IPC::ResponseBuilder rb{ctx, 2, 1};
     rb.Push(RESULT_SUCCESS);
-    rb.PushCopyObjects(event);
+    rb.PushCopyObjects(msg_queue->GetMesssageRecieveEvent());
 
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    LOG_DEBUG(Service_AM, "called");
 }
 
 void ICommonStateGetter::ReceiveMessage(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 3};
     rb.Push(RESULT_SUCCESS);
-    rb.Push<u32>(15);
+    rb.PushEnum<AppletMessageQueue::AppletMessage>(msg_queue->PopMessage());
 
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    LOG_DEBUG(Service_AM, "called");
 }
 
 void ICommonStateGetter::GetCurrentFocusState(Kernel::HLERequestContext& ctx) {
@@ -414,13 +459,11 @@ void ICommonStateGetter::GetCurrentFocusState(Kernel::HLERequestContext& ctx) {
 }
 
 void ICommonStateGetter::GetDefaultDisplayResolutionChangeEvent(Kernel::HLERequestContext& ctx) {
-    event->Signal();
-
     IPC::ResponseBuilder rb{ctx, 2, 1};
     rb.Push(RESULT_SUCCESS);
-    rb.PushCopyObjects(event);
+    rb.PushCopyObjects(msg_queue->GetOperationModeChangedEvent());
 
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    LOG_DEBUG(Service_AM, "called");
 }
 
 void ICommonStateGetter::GetDefaultDisplayResolution(Kernel::HLERequestContext& ctx) {
@@ -444,7 +487,7 @@ void ICommonStateGetter::GetOperationMode(Kernel::HLERequestContext& ctx) {
     rb.Push(RESULT_SUCCESS);
     rb.Push(static_cast<u8>(use_docked_mode ? OperationMode::Docked : OperationMode::Handheld));
 
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    LOG_DEBUG(Service_AM, "called");
 }
 
 void ICommonStateGetter::GetPerformanceMode(Kernel::HLERequestContext& ctx) {
@@ -454,7 +497,7 @@ void ICommonStateGetter::GetPerformanceMode(Kernel::HLERequestContext& ctx) {
     rb.Push(static_cast<u32>(use_docked_mode ? APM::PerformanceMode::Docked
                                              : APM::PerformanceMode::Handheld));
 
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+    LOG_DEBUG(Service_AM, "called");
 }
 
 class IStorageAccessor final : public ServiceFramework<IStorageAccessor> {
@@ -690,7 +733,7 @@ IApplicationFunctions::IApplicationFunctions() : ServiceFramework("IApplicationF
         {70, nullptr, "RequestToShutdown"},
         {71, nullptr, "RequestToReboot"},
         {80, nullptr, "ExitAndRequestToShowThanksMessage"},
-        {90, nullptr, "EnableApplicationCrashReport"},
+        {90, &IApplicationFunctions::EnableApplicationCrashReport, "EnableApplicationCrashReport"},
         {100, nullptr, "InitializeApplicationCopyrightFrameBuffer"},
         {101, nullptr, "SetApplicationCopyrightImage"},
         {102, nullptr, "SetApplicationCopyrightVisibility"},
@@ -709,6 +752,12 @@ IApplicationFunctions::IApplicationFunctions() : ServiceFramework("IApplicationF
 
 IApplicationFunctions::~IApplicationFunctions() = default;
 
+void IApplicationFunctions::EnableApplicationCrashReport(Kernel::HLERequestContext& ctx) {
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(RESULT_SUCCESS);
+    LOG_WARNING(Service_AM, "(STUBBED) called");
+}
+
 void IApplicationFunctions::BeginBlockingHomeButtonShortAndLongPressed(
     Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2};
@@ -778,7 +827,8 @@ void IApplicationFunctions::EnsureSaveData(Kernel::HLERequestContext& ctx) {
 
 void IApplicationFunctions::SetTerminateResult(Kernel::HLERequestContext& ctx) {
     // Takes an input u32 Result, no output.
-    // For example, in some cases official apps use this with error 0x2A2 then uses svcBreak.
+    // For example, in some cases official apps use this with error 0x2A2 then
+    // uses svcBreak.
 
     IPC::RequestParser rp{ctx};
     u32 result = rp.Pop<u32>();
@@ -840,8 +890,12 @@ void IApplicationFunctions::GetPseudoDeviceId(Kernel::HLERequestContext& ctx) {
 
 void InstallInterfaces(SM::ServiceManager& service_manager,
                        std::shared_ptr<NVFlinger::NVFlinger> nvflinger) {
-    std::make_shared<AppletAE>(nvflinger)->InstallAsService(service_manager);
-    std::make_shared<AppletOE>(nvflinger)->InstallAsService(service_manager);
+    auto message_queue = std::make_shared<AppletMessageQueue>();
+    message_queue->PushMessage(AppletMessageQueue::AppletMessage::FocusStateChanged); // Needed on
+                                                                                      // game boot
+
+    std::make_shared<AppletAE>(nvflinger, message_queue)->InstallAsService(service_manager);
+    std::make_shared<AppletOE>(nvflinger, message_queue)->InstallAsService(service_manager);
     std::make_shared<IdleSys>()->InstallAsService(service_manager);
     std::make_shared<OMM>()->InstallAsService(service_manager);
     std::make_shared<SPSM>()->InstallAsService(service_manager);

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

@@ -5,6 +5,7 @@
 #pragma once
 
 #include <memory>
+#include <queue>
 #include "core/hle/service/service.h"
 
 namespace Kernel {
@@ -39,6 +40,31 @@ enum SystemLanguage {
     TraditionalChinese = 16,
 };
 
+class AppletMessageQueue {
+public:
+    enum class AppletMessage : u32 {
+        NoMessage = 0,
+        FocusStateChanged = 15,
+        OperationModeChanged = 30,
+        PerformanceModeChanged = 31,
+    };
+
+    AppletMessageQueue();
+    ~AppletMessageQueue();
+
+    const Kernel::SharedPtr<Kernel::Event>& GetMesssageRecieveEvent() const;
+    const Kernel::SharedPtr<Kernel::Event>& GetOperationModeChangedEvent() const;
+    void PushMessage(AppletMessage msg);
+    AppletMessage PopMessage();
+    std::size_t GetMessageCount() const;
+    void OperationModeChanged();
+
+private:
+    std::queue<AppletMessage> messages;
+    Kernel::SharedPtr<Kernel::Event> on_new_message;
+    Kernel::SharedPtr<Kernel::Event> on_operation_mode_changed;
+};
+
 class IWindowController final : public ServiceFramework<IWindowController> {
 public:
     IWindowController();
@@ -102,7 +128,7 @@ private:
 
 class ICommonStateGetter final : public ServiceFramework<ICommonStateGetter> {
 public:
-    ICommonStateGetter();
+    explicit ICommonStateGetter(std::shared_ptr<AppletMessageQueue> msg_queue);
     ~ICommonStateGetter() override;
 
 private:
@@ -126,6 +152,7 @@ private:
     void GetDefaultDisplayResolution(Kernel::HLERequestContext& ctx);
 
     Kernel::SharedPtr<Kernel::Event> event;
+    std::shared_ptr<AppletMessageQueue> msg_queue;
 };
 
 class ILibraryAppletCreator final : public ServiceFramework<ILibraryAppletCreator> {
@@ -158,6 +185,7 @@ private:
     void EndBlockingHomeButtonShortAndLongPressed(Kernel::HLERequestContext& ctx);
     void BeginBlockingHomeButton(Kernel::HLERequestContext& ctx);
     void EndBlockingHomeButton(Kernel::HLERequestContext& ctx);
+    void EnableApplicationCrashReport(Kernel::HLERequestContext& ctx);
 };
 
 class IHomeMenuFunctions final : public ServiceFramework<IHomeMenuFunctions> {

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

@@ -12,8 +12,10 @@ namespace Service::AM {
 
 class ILibraryAppletProxy final : public ServiceFramework<ILibraryAppletProxy> {
 public:
-    explicit ILibraryAppletProxy(std::shared_ptr<NVFlinger::NVFlinger> nvflinger)
-        : ServiceFramework("ILibraryAppletProxy"), nvflinger(std::move(nvflinger)) {
+    explicit ILibraryAppletProxy(std::shared_ptr<NVFlinger::NVFlinger> nvflinger,
+                                 std::shared_ptr<AppletMessageQueue> msg_queue)
+        : ServiceFramework("ILibraryAppletProxy"), nvflinger(std::move(nvflinger)),
+          msg_queue(std::move(msg_queue)) {
         static const FunctionInfo functions[] = {
             {0, &ILibraryAppletProxy::GetCommonStateGetter, "GetCommonStateGetter"},
             {1, &ILibraryAppletProxy::GetSelfController, "GetSelfController"},
@@ -32,7 +34,7 @@ private:
     void GetCommonStateGetter(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
-        rb.PushIpcInterface<ICommonStateGetter>();
+        rb.PushIpcInterface<ICommonStateGetter>(msg_queue);
         LOG_DEBUG(Service_AM, "called");
     }
 
@@ -93,12 +95,15 @@ private:
     }
 
     std::shared_ptr<NVFlinger::NVFlinger> nvflinger;
+    std::shared_ptr<AppletMessageQueue> msg_queue;
 };
 
 class ISystemAppletProxy final : public ServiceFramework<ISystemAppletProxy> {
 public:
-    explicit ISystemAppletProxy(std::shared_ptr<NVFlinger::NVFlinger> nvflinger)
-        : ServiceFramework("ISystemAppletProxy"), nvflinger(std::move(nvflinger)) {
+    explicit ISystemAppletProxy(std::shared_ptr<NVFlinger::NVFlinger> nvflinger,
+                                std::shared_ptr<AppletMessageQueue> msg_queue)
+        : ServiceFramework("ISystemAppletProxy"), nvflinger(std::move(nvflinger)),
+          msg_queue(std::move(msg_queue)) {
         static const FunctionInfo functions[] = {
             {0, &ISystemAppletProxy::GetCommonStateGetter, "GetCommonStateGetter"},
             {1, &ISystemAppletProxy::GetSelfController, "GetSelfController"},
@@ -119,7 +124,7 @@ private:
     void GetCommonStateGetter(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
-        rb.PushIpcInterface<ICommonStateGetter>();
+        rb.PushIpcInterface<ICommonStateGetter>(msg_queue);
         LOG_DEBUG(Service_AM, "called");
     }
 
@@ -186,31 +191,34 @@ private:
         LOG_DEBUG(Service_AM, "called");
     }
     std::shared_ptr<NVFlinger::NVFlinger> nvflinger;
+    std::shared_ptr<AppletMessageQueue> msg_queue;
 };
 
 void AppletAE::OpenSystemAppletProxy(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);
-    rb.PushIpcInterface<ISystemAppletProxy>(nvflinger);
+    rb.PushIpcInterface<ISystemAppletProxy>(nvflinger, msg_queue);
     LOG_DEBUG(Service_AM, "called");
 }
 
 void AppletAE::OpenLibraryAppletProxy(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);
-    rb.PushIpcInterface<ILibraryAppletProxy>(nvflinger);
+    rb.PushIpcInterface<ILibraryAppletProxy>(nvflinger, msg_queue);
     LOG_DEBUG(Service_AM, "called");
 }
 
 void AppletAE::OpenLibraryAppletProxyOld(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);
-    rb.PushIpcInterface<ILibraryAppletProxy>(nvflinger);
+    rb.PushIpcInterface<ILibraryAppletProxy>(nvflinger, msg_queue);
     LOG_DEBUG(Service_AM, "called");
 }
 
-AppletAE::AppletAE(std::shared_ptr<NVFlinger::NVFlinger> nvflinger)
-    : ServiceFramework("appletAE"), nvflinger(std::move(nvflinger)) {
+AppletAE::AppletAE(std::shared_ptr<NVFlinger::NVFlinger> nvflinger,
+                   std::shared_ptr<AppletMessageQueue> msg_queue)
+    : ServiceFramework("appletAE"), nvflinger(std::move(nvflinger)),
+      msg_queue(std::move(msg_queue)) {
     // clang-format off
     static const FunctionInfo functions[] = {
         {100, &AppletAE::OpenSystemAppletProxy, "OpenSystemAppletProxy"},
@@ -228,4 +236,8 @@ AppletAE::AppletAE(std::shared_ptr<NVFlinger::NVFlinger> nvflinger)
 
 AppletAE::~AppletAE() = default;
 
+const std::shared_ptr<AppletMessageQueue>& AppletAE::GetMessageQueue() const {
+    return msg_queue;
+}
+
 } // namespace Service::AM

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

@@ -17,15 +17,19 @@ namespace AM {
 
 class AppletAE final : public ServiceFramework<AppletAE> {
 public:
-    explicit AppletAE(std::shared_ptr<NVFlinger::NVFlinger> nvflinger);
+    explicit AppletAE(std::shared_ptr<NVFlinger::NVFlinger> nvflinger,
+                      std::shared_ptr<AppletMessageQueue> msg_queue);
     ~AppletAE() override;
 
+    const std::shared_ptr<AppletMessageQueue>& GetMessageQueue() const;
+
 private:
     void OpenSystemAppletProxy(Kernel::HLERequestContext& ctx);
     void OpenLibraryAppletProxy(Kernel::HLERequestContext& ctx);
     void OpenLibraryAppletProxyOld(Kernel::HLERequestContext& ctx);
 
     std::shared_ptr<NVFlinger::NVFlinger> nvflinger;
+    std::shared_ptr<AppletMessageQueue> msg_queue;
 };
 
 } // namespace AM

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

@@ -12,8 +12,10 @@ namespace Service::AM {
 
 class IApplicationProxy final : public ServiceFramework<IApplicationProxy> {
 public:
-    explicit IApplicationProxy(std::shared_ptr<NVFlinger::NVFlinger> nvflinger)
-        : ServiceFramework("IApplicationProxy"), nvflinger(std::move(nvflinger)) {
+    explicit IApplicationProxy(std::shared_ptr<NVFlinger::NVFlinger> nvflinger,
+                               std::shared_ptr<AppletMessageQueue> msg_queue)
+        : ServiceFramework("IApplicationProxy"), nvflinger(std::move(nvflinger)),
+          msg_queue(std::move(msg_queue)) {
         // clang-format off
         static const FunctionInfo functions[] = {
             {0, &IApplicationProxy::GetCommonStateGetter, "GetCommonStateGetter"},
@@ -70,7 +72,7 @@ private:
     void GetCommonStateGetter(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
-        rb.PushIpcInterface<ICommonStateGetter>();
+        rb.PushIpcInterface<ICommonStateGetter>(msg_queue);
         LOG_DEBUG(Service_AM, "called");
     }
 
@@ -89,17 +91,20 @@ private:
     }
 
     std::shared_ptr<NVFlinger::NVFlinger> nvflinger;
+    std::shared_ptr<AppletMessageQueue> msg_queue;
 };
 
 void AppletOE::OpenApplicationProxy(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);
-    rb.PushIpcInterface<IApplicationProxy>(nvflinger);
+    rb.PushIpcInterface<IApplicationProxy>(nvflinger, msg_queue);
     LOG_DEBUG(Service_AM, "called");
 }
 
-AppletOE::AppletOE(std::shared_ptr<NVFlinger::NVFlinger> nvflinger)
-    : ServiceFramework("appletOE"), nvflinger(std::move(nvflinger)) {
+AppletOE::AppletOE(std::shared_ptr<NVFlinger::NVFlinger> nvflinger,
+                   std::shared_ptr<AppletMessageQueue> msg_queue)
+    : ServiceFramework("appletOE"), nvflinger(std::move(nvflinger)),
+      msg_queue(std::move(msg_queue)) {
     static const FunctionInfo functions[] = {
         {0, &AppletOE::OpenApplicationProxy, "OpenApplicationProxy"},
     };
@@ -108,4 +113,8 @@ AppletOE::AppletOE(std::shared_ptr<NVFlinger::NVFlinger> nvflinger)
 
 AppletOE::~AppletOE() = default;
 
+const std::shared_ptr<AppletMessageQueue>& AppletOE::GetMessageQueue() const {
+    return msg_queue;
+}
+
 } // namespace Service::AM

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

@@ -17,13 +17,17 @@ namespace AM {
 
 class AppletOE final : public ServiceFramework<AppletOE> {
 public:
-    explicit AppletOE(std::shared_ptr<NVFlinger::NVFlinger> nvflinger);
+    explicit AppletOE(std::shared_ptr<NVFlinger::NVFlinger> nvflinger,
+                      std::shared_ptr<AppletMessageQueue> msg_queue);
     ~AppletOE() override;
 
+    const std::shared_ptr<AppletMessageQueue>& GetMessageQueue() const;
+
 private:
     void OpenApplicationProxy(Kernel::HLERequestContext& ctx);
 
     std::shared_ptr<NVFlinger::NVFlinger> nvflinger;
+    std::shared_ptr<AppletMessageQueue> msg_queue;
 };
 
 } // namespace AM

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

@@ -28,13 +28,13 @@ public:
             {1, &IAudioRenderer::GetSampleCount, "GetSampleCount"},
             {2, &IAudioRenderer::GetMixBufferCount, "GetMixBufferCount"},
             {3, &IAudioRenderer::GetState, "GetState"},
-            {4, &IAudioRenderer::RequestUpdate, "RequestUpdate"},
+            {4, &IAudioRenderer::RequestUpdateImpl, "RequestUpdate"},
             {5, &IAudioRenderer::Start, "Start"},
             {6, &IAudioRenderer::Stop, "Stop"},
             {7, &IAudioRenderer::QuerySystemEvent, "QuerySystemEvent"},
-            {8, nullptr, "SetRenderingTimeLimit"},
-            {9, nullptr, "GetRenderingTimeLimit"},
-            {10, nullptr, "RequestUpdateAuto"},
+            {8, &IAudioRenderer::SetRenderingTimeLimit, "SetRenderingTimeLimit"},
+            {9, &IAudioRenderer::GetRenderingTimeLimit, "GetRenderingTimeLimit"},
+            {10, &IAudioRenderer::RequestUpdateImpl, "RequestUpdateAuto"},
             {11, nullptr, "ExecuteAudioRendererRendering"},
         };
         // clang-format on
@@ -79,7 +79,7 @@ private:
         LOG_DEBUG(Service_Audio, "called");
     }
 
-    void RequestUpdate(Kernel::HLERequestContext& ctx) {
+    void RequestUpdateImpl(Kernel::HLERequestContext& ctx) {
         ctx.WriteBuffer(renderer->UpdateAudioRenderer(ctx.ReadBuffer()));
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
@@ -110,8 +110,29 @@ private:
         LOG_WARNING(Service_Audio, "(STUBBED) called");
     }
 
+    void SetRenderingTimeLimit(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        rendering_time_limit_percent = rp.Pop<u32>();
+        ASSERT(rendering_time_limit_percent >= 0 && rendering_time_limit_percent <= 100);
+
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+
+        LOG_DEBUG(Service_Audio, "called. rendering_time_limit_percent={}",
+                  rendering_time_limit_percent);
+    }
+
+    void GetRenderingTimeLimit(Kernel::HLERequestContext& ctx) {
+        LOG_DEBUG(Service_Audio, "called");
+
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(RESULT_SUCCESS);
+        rb.Push(rendering_time_limit_percent);
+    }
+
     Kernel::SharedPtr<Kernel::Event> system_event;
     std::unique_ptr<AudioCore::AudioRenderer> renderer;
+    u32 rendering_time_limit_percent = 100;
 };
 
 class IAudioDevice final : public ServiceFramework<IAudioDevice> {

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

@@ -77,8 +77,8 @@ private:
         IPC::ResponseBuilder rb{ctx, 6};
         rb.Push(RESULT_SUCCESS);
         rb.Push<u32>(consumed);
-        rb.Push<u64>(performance);
         rb.Push<u32>(sample_count);
+        rb.Push<u64>(performance);
         ctx.WriteBuffer(samples.data(), samples.size() * sizeof(s16));
     }
 

src/core/hle/service/btdrv/btdrv.cpp

@@ -2,12 +2,49 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include "common/logging/log.h"
+#include "core/hle/ipc_helpers.h"
+#include "core/hle/kernel/event.h"
+#include "core/hle/kernel/hle_ipc.h"
 #include "core/hle/service/btdrv/btdrv.h"
 #include "core/hle/service/service.h"
 #include "core/hle/service/sm/sm.h"
 
 namespace Service::BtDrv {
 
+class Bt final : public ServiceFramework<Bt> {
+public:
+    explicit Bt() : ServiceFramework{"bt"} {
+        // clang-format off
+        static const FunctionInfo functions[] = {
+            {0, nullptr, "Unknown0"},
+            {1, nullptr, "Unknown1"},
+            {2, nullptr, "Unknown2"},
+            {3, nullptr, "Unknown3"},
+            {4, nullptr, "Unknown4"},
+            {5, nullptr, "Unknown5"},
+            {6, nullptr, "Unknown6"},
+            {7, nullptr, "Unknown7"},
+            {8, nullptr, "Unknown8"},
+            {9, &Bt::RegisterEvent, "RegisterEvent"},
+        };
+        // clang-format on
+        RegisterHandlers(functions);
+    }
+
+private:
+    void RegisterEvent(Kernel::HLERequestContext& ctx) {
+        auto& kernel = Core::System::GetInstance().Kernel();
+        register_event =
+            Kernel::Event::Create(kernel, Kernel::ResetType::OneShot, "BT:RegisterEvent");
+        IPC::ResponseBuilder rb{ctx, 2, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushCopyObjects(register_event);
+        LOG_WARNING(Service_BTM, "(STUBBED) called");
+    }
+    Kernel::SharedPtr<Kernel::Event> register_event;
+};
+
 class BtDrv final : public ServiceFramework<BtDrv> {
 public:
     explicit BtDrv() : ServiceFramework{"btdrv"} {
@@ -67,6 +104,7 @@ public:
 
 void InstallInterfaces(SM::ServiceManager& sm) {
     std::make_shared<BtDrv>()->InstallAsService(sm);
+    std::make_shared<Bt>()->InstallAsService(sm);
 }
 
 } // namespace Service::BtDrv

src/core/hle/service/btm/btm.cpp

@@ -6,13 +6,118 @@
 
 #include "common/logging/log.h"
 #include "core/hle/ipc_helpers.h"
+#include "core/hle/kernel/event.h"
 #include "core/hle/kernel/hle_ipc.h"
 #include "core/hle/service/btm/btm.h"
 #include "core/hle/service/service.h"
-#include "core/hle/service/sm/sm.h"
 
 namespace Service::BTM {
 
+class IBtmUserCore final : public ServiceFramework<IBtmUserCore> {
+public:
+    explicit IBtmUserCore() : ServiceFramework{"IBtmUserCore"} {
+        // clang-format off
+        static const FunctionInfo functions[] = {
+            {0, &IBtmUserCore::GetScanEvent, "GetScanEvent"},
+            {1, nullptr, "Unknown1"},
+            {2, nullptr, "Unknown2"},
+            {3, nullptr, "Unknown3"},
+            {4, nullptr, "Unknown4"},
+            {5, nullptr, "Unknown5"},
+            {6, nullptr, "Unknown6"},
+            {7, nullptr, "Unknown7"},
+            {8, nullptr, "Unknown8"},
+            {9, nullptr, "Unknown9"},
+            {10, nullptr, "Unknown10"},
+            {17, &IBtmUserCore::GetConnectionEvent, "GetConnectionEvent"},
+            {18, nullptr, "Unknown18"},
+            {19, nullptr, "Unknown19"},
+            {20, nullptr, "Unknown20"},
+            {21, nullptr, "Unknown21"},
+            {22, nullptr, "Unknown22"},
+            {23, nullptr, "Unknown23"},
+            {24, nullptr, "Unknown24"},
+            {25, nullptr, "Unknown25"},
+            {26, &IBtmUserCore::GetDiscoveryEvent, "AcquireBleServiceDiscoveryEventImpl"},
+            {27, nullptr, "Unknown27"},
+            {28, nullptr, "Unknown28"},
+            {29, nullptr, "Unknown29"},
+            {30, nullptr, "Unknown30"},
+            {31, nullptr, "Unknown31"},
+            {32, nullptr, "Unknown32"},
+            {33, &IBtmUserCore::GetConfigEvent, "GetConfigEvent"},
+            {34, nullptr, "Unknown34"},
+            {35, nullptr, "Unknown35"},
+            {36, nullptr, "Unknown36"},
+            {37, nullptr, "Unknown37"},
+        };
+        // clang-format on
+        RegisterHandlers(functions);
+    }
+
+private:
+    void GetScanEvent(Kernel::HLERequestContext& ctx) {
+        auto& kernel = Core::System::GetInstance().Kernel();
+        scan_event =
+            Kernel::Event::Create(kernel, Kernel::ResetType::OneShot, "IBtmUserCore:ScanEvent");
+        IPC::ResponseBuilder rb{ctx, 2, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushCopyObjects(scan_event);
+        LOG_WARNING(Service_BTM, "(STUBBED) called");
+    }
+    void GetConnectionEvent(Kernel::HLERequestContext& ctx) {
+        auto& kernel = Core::System::GetInstance().Kernel();
+        connection_event = Kernel::Event::Create(kernel, Kernel::ResetType::OneShot,
+                                                 "IBtmUserCore:ConnectionEvent");
+        IPC::ResponseBuilder rb{ctx, 2, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushCopyObjects(connection_event);
+        LOG_WARNING(Service_BTM, "(STUBBED) called");
+    }
+    void GetDiscoveryEvent(Kernel::HLERequestContext& ctx) {
+        auto& kernel = Core::System::GetInstance().Kernel();
+        service_discovery =
+            Kernel::Event::Create(kernel, Kernel::ResetType::OneShot, "IBtmUserCore:Discovery");
+        IPC::ResponseBuilder rb{ctx, 2, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushCopyObjects(service_discovery);
+        LOG_WARNING(Service_BTM, "(STUBBED) called");
+    }
+    void GetConfigEvent(Kernel::HLERequestContext& ctx) {
+        auto& kernel = Core::System::GetInstance().Kernel();
+        config_event =
+            Kernel::Event::Create(kernel, Kernel::ResetType::OneShot, "IBtmUserCore:ConfigEvent");
+        IPC::ResponseBuilder rb{ctx, 2, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushCopyObjects(config_event);
+        LOG_WARNING(Service_BTM, "(STUBBED) called");
+    }
+    Kernel::SharedPtr<Kernel::Event> scan_event;
+    Kernel::SharedPtr<Kernel::Event> connection_event;
+    Kernel::SharedPtr<Kernel::Event> service_discovery;
+    Kernel::SharedPtr<Kernel::Event> config_event;
+};
+
+class BTM_USR final : public ServiceFramework<BTM_USR> {
+public:
+    explicit BTM_USR() : ServiceFramework{"btm:u"} {
+        // clang-format off
+        static const FunctionInfo functions[] = {
+            {0, &BTM_USR::GetCoreImpl, "GetCoreImpl"},
+        };
+        // clang-format on
+        RegisterHandlers(functions);
+    }
+
+private:
+    void GetCoreImpl(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushIpcInterface<IBtmUserCore>();
+        LOG_DEBUG(Service_BTM, "called");
+    }
+};
+
 class BTM final : public ServiceFramework<BTM> {
 public:
     explicit BTM() : ServiceFramework{"btm"} {
@@ -116,6 +221,7 @@ void InstallInterfaces(SM::ServiceManager& sm) {
     std::make_shared<BTM>()->InstallAsService(sm);
     std::make_shared<BTM_DBG>()->InstallAsService(sm);
     std::make_shared<BTM_SYS>()->InstallAsService(sm);
+    std::make_shared<BTM_USR>()->InstallAsService(sm);
 }
 
 } // namespace Service::BTM

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

@@ -303,25 +303,42 @@ ResultVal<FileSys::VirtualDir> OpenSaveData(FileSys::SaveDataSpaceId space,
               static_cast<u8>(space), save_struct.DebugInfo());
 
     if (save_data_factory == nullptr) {
-        return ResultCode(ErrorModule::FS, FileSys::ErrCodes::TitleNotFound);
+        return FileSys::ERROR_ENTITY_NOT_FOUND;
     }
 
     return save_data_factory->Open(space, save_struct);
 }
 
+ResultVal<FileSys::VirtualDir> OpenSaveDataSpace(FileSys::SaveDataSpaceId space) {
+    LOG_TRACE(Service_FS, "Opening Save Data Space for space_id={:01X}", static_cast<u8>(space));
+
+    if (save_data_factory == nullptr) {
+        return FileSys::ERROR_ENTITY_NOT_FOUND;
+    }
+
+    return MakeResult(save_data_factory->GetSaveDataSpaceDirectory(space));
+}
+
 ResultVal<FileSys::VirtualDir> OpenSDMC() {
     LOG_TRACE(Service_FS, "Opening SDMC");
 
     if (sdmc_factory == nullptr) {
-        return ResultCode(ErrorModule::FS, FileSys::ErrCodes::SdCardNotFound);
+        return FileSys::ERROR_SD_CARD_NOT_FOUND;
     }
 
     return sdmc_factory->Open();
 }
 
-std::unique_ptr<FileSys::RegisteredCacheUnion> GetUnionContents() {
-    return std::make_unique<FileSys::RegisteredCacheUnion>(std::vector<FileSys::RegisteredCache*>{
-        GetSystemNANDContents(), GetUserNANDContents(), GetSDMCContents()});
+std::shared_ptr<FileSys::RegisteredCacheUnion> registered_cache_union;
+
+std::shared_ptr<FileSys::RegisteredCacheUnion> GetUnionContents() {
+    if (registered_cache_union == nullptr) {
+        registered_cache_union =
+            std::make_shared<FileSys::RegisteredCacheUnion>(std::vector<FileSys::RegisteredCache*>{
+                GetSystemNANDContents(), GetUserNANDContents(), GetSDMCContents()});
+    }
+
+    return registered_cache_union;
 }
 
 FileSys::RegisteredCache* GetSystemNANDContents() {
@@ -360,6 +377,15 @@ FileSys::VirtualDir GetModificationLoadRoot(u64 title_id) {
     return bis_factory->GetModificationLoadRoot(title_id);
 }
 
+FileSys::VirtualDir GetModificationDumpRoot(u64 title_id) {
+    LOG_TRACE(Service_FS, "Opening mod dump root for tid={:016X}", title_id);
+
+    if (bis_factory == nullptr)
+        return nullptr;
+
+    return bis_factory->GetModificationDumpRoot(title_id);
+}
+
 void CreateFactories(FileSys::VfsFilesystem& vfs, bool overwrite) {
     if (overwrite) {
         bis_factory = nullptr;
@@ -373,13 +399,21 @@ void CreateFactories(FileSys::VfsFilesystem& vfs, bool overwrite) {
                                           FileSys::Mode::ReadWrite);
     auto load_directory = vfs.OpenDirectory(FileUtil::GetUserPath(FileUtil::UserPath::LoadDir),
                                             FileSys::Mode::ReadWrite);
+    auto dump_directory = vfs.OpenDirectory(FileUtil::GetUserPath(FileUtil::UserPath::DumpDir),
+                                            FileSys::Mode::ReadWrite);
 
-    if (bis_factory == nullptr)
-        bis_factory = std::make_unique<FileSys::BISFactory>(nand_directory, load_directory);
-    if (save_data_factory == nullptr)
+    if (bis_factory == nullptr) {
+        bis_factory =
+            std::make_unique<FileSys::BISFactory>(nand_directory, load_directory, dump_directory);
+    }
+
+    if (save_data_factory == nullptr) {
         save_data_factory = std::make_unique<FileSys::SaveDataFactory>(std::move(nand_directory));
-    if (sdmc_factory == nullptr)
+    }
+
+    if (sdmc_factory == nullptr) {
         sdmc_factory = std::make_unique<FileSys::SDMCFactory>(std::move(sd_directory));
+    }
 }
 
 void InstallInterfaces(SM::ServiceManager& service_manager, FileSys::VfsFilesystem& vfs) {

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

@@ -45,15 +45,17 @@ ResultVal<FileSys::VirtualFile> OpenRomFS(u64 title_id, FileSys::StorageId stora
                                           FileSys::ContentRecordType type);
 ResultVal<FileSys::VirtualDir> OpenSaveData(FileSys::SaveDataSpaceId space,
                                             FileSys::SaveDataDescriptor save_struct);
+ResultVal<FileSys::VirtualDir> OpenSaveDataSpace(FileSys::SaveDataSpaceId space);
 ResultVal<FileSys::VirtualDir> OpenSDMC();
 
-std::unique_ptr<FileSys::RegisteredCacheUnion> GetUnionContents();
+std::shared_ptr<FileSys::RegisteredCacheUnion> GetUnionContents();
 
 FileSys::RegisteredCache* GetSystemNANDContents();
 FileSys::RegisteredCache* GetUserNANDContents();
 FileSys::RegisteredCache* GetSDMCContents();
 
 FileSys::VirtualDir GetModificationLoadRoot(u64 title_id);
+FileSys::VirtualDir GetModificationDumpRoot(u64 title_id);
 
 // Creates the SaveData, SDMC, and BIS Factories. Should be called once and before any function
 // above is called.

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

@@ -11,6 +11,7 @@
 
 #include "common/assert.h"
 #include "common/common_types.h"
+#include "common/hex_util.h"
 #include "common/logging/log.h"
 #include "common/string_util.h"
 #include "core/file_sys/directory.h"
@@ -62,12 +63,12 @@ private:
         // Error checking
         if (length < 0) {
             IPC::ResponseBuilder rb{ctx, 2};
-            rb.Push(ResultCode(ErrorModule::FS, ErrorDescription::InvalidLength));
+            rb.Push(FileSys::ERROR_INVALID_SIZE);
             return;
         }
         if (offset < 0) {
             IPC::ResponseBuilder rb{ctx, 2};
-            rb.Push(ResultCode(ErrorModule::FS, ErrorDescription::InvalidOffset));
+            rb.Push(FileSys::ERROR_INVALID_OFFSET);
             return;
         }
 
@@ -107,12 +108,12 @@ private:
         // Error checking
         if (length < 0) {
             IPC::ResponseBuilder rb{ctx, 2};
-            rb.Push(ResultCode(ErrorModule::FS, ErrorDescription::InvalidLength));
+            rb.Push(FileSys::ERROR_INVALID_SIZE);
             return;
         }
         if (offset < 0) {
             IPC::ResponseBuilder rb{ctx, 2};
-            rb.Push(ResultCode(ErrorModule::FS, ErrorDescription::InvalidOffset));
+            rb.Push(FileSys::ERROR_INVALID_OFFSET);
             return;
         }
 
@@ -138,12 +139,12 @@ private:
         // Error checking
         if (length < 0) {
             IPC::ResponseBuilder rb{ctx, 2};
-            rb.Push(ResultCode(ErrorModule::FS, ErrorDescription::InvalidLength));
+            rb.Push(FileSys::ERROR_INVALID_SIZE);
             return;
         }
         if (offset < 0) {
             IPC::ResponseBuilder rb{ctx, 2};
-            rb.Push(ResultCode(ErrorModule::FS, ErrorDescription::InvalidOffset));
+            rb.Push(FileSys::ERROR_INVALID_OFFSET);
             return;
         }
 
@@ -451,7 +452,147 @@ private:
     VfsDirectoryServiceWrapper backend;
 };
 
+class ISaveDataInfoReader final : public ServiceFramework<ISaveDataInfoReader> {
+public:
+    explicit ISaveDataInfoReader(FileSys::SaveDataSpaceId space)
+        : ServiceFramework("ISaveDataInfoReader") {
+        static const FunctionInfo functions[] = {
+            {0, &ISaveDataInfoReader::ReadSaveDataInfo, "ReadSaveDataInfo"},
+        };
+        RegisterHandlers(functions);
+
+        FindAllSaves(space);
+    }
+
+    void ReadSaveDataInfo(Kernel::HLERequestContext& ctx) {
+        // Calculate how many entries we can fit in the output buffer
+        const u64 count_entries = ctx.GetWriteBufferSize() / sizeof(SaveDataInfo);
+
+        // Cap at total number of entries.
+        const u64 actual_entries = std::min(count_entries, info.size() - next_entry_index);
+
+        // Determine data start and end
+        const auto* begin = reinterpret_cast<u8*>(info.data() + next_entry_index);
+        const auto* end = reinterpret_cast<u8*>(info.data() + next_entry_index + actual_entries);
+        const auto range_size = static_cast<std::size_t>(std::distance(begin, end));
+
+        next_entry_index += actual_entries;
+
+        // Write the data to memory
+        ctx.WriteBuffer(begin, range_size);
+
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(RESULT_SUCCESS);
+        rb.Push<u32>(static_cast<u32>(actual_entries));
+    }
+
+private:
+    static u64 stoull_be(std::string_view str) {
+        if (str.size() != 16)
+            return 0;
+
+        const auto bytes = Common::HexStringToArray<0x8>(str);
+        u64 out{};
+        std::memcpy(&out, bytes.data(), sizeof(u64));
+
+        return Common::swap64(out);
+    }
+
+    void FindAllSaves(FileSys::SaveDataSpaceId space) {
+        const auto save_root = OpenSaveDataSpace(space);
+        ASSERT(save_root.Succeeded());
+
+        for (const auto& type : (*save_root)->GetSubdirectories()) {
+            if (type->GetName() == "save") {
+                for (const auto& save_id : type->GetSubdirectories()) {
+                    for (const auto& user_id : save_id->GetSubdirectories()) {
+                        const auto save_id_numeric = stoull_be(save_id->GetName());
+                        auto user_id_numeric = Common::HexStringToArray<0x10>(user_id->GetName());
+                        std::reverse(user_id_numeric.begin(), user_id_numeric.end());
+
+                        if (save_id_numeric != 0) {
+                            // System Save Data
+                            info.emplace_back(SaveDataInfo{
+                                0,
+                                space,
+                                FileSys::SaveDataType::SystemSaveData,
+                                {},
+                                user_id_numeric,
+                                save_id_numeric,
+                                0,
+                                user_id->GetSize(),
+                                {},
+                            });
+
+                            continue;
+                        }
+
+                        for (const auto& title_id : user_id->GetSubdirectories()) {
+                            const auto device =
+                                std::all_of(user_id_numeric.begin(), user_id_numeric.end(),
+                                            [](u8 val) { return val == 0; });
+                            info.emplace_back(SaveDataInfo{
+                                0,
+                                space,
+                                device ? FileSys::SaveDataType::DeviceSaveData
+                                       : FileSys::SaveDataType::SaveData,
+                                {},
+                                user_id_numeric,
+                                save_id_numeric,
+                                stoull_be(title_id->GetName()),
+                                title_id->GetSize(),
+                                {},
+                            });
+                        }
+                    }
+                }
+            } else if (space == FileSys::SaveDataSpaceId::TemporaryStorage) {
+                // Temporary Storage
+                for (const auto& user_id : type->GetSubdirectories()) {
+                    for (const auto& title_id : user_id->GetSubdirectories()) {
+                        if (!title_id->GetFiles().empty() ||
+                            !title_id->GetSubdirectories().empty()) {
+                            auto user_id_numeric =
+                                Common::HexStringToArray<0x10>(user_id->GetName());
+                            std::reverse(user_id_numeric.begin(), user_id_numeric.end());
+
+                            info.emplace_back(SaveDataInfo{
+                                0,
+                                space,
+                                FileSys::SaveDataType::TemporaryStorage,
+                                {},
+                                user_id_numeric,
+                                stoull_be(type->GetName()),
+                                stoull_be(title_id->GetName()),
+                                title_id->GetSize(),
+                                {},
+                            });
+                        }
+                    }
+                }
+            }
+        }
+    }
+
+    struct SaveDataInfo {
+        u64_le save_id_unknown;
+        FileSys::SaveDataSpaceId space;
+        FileSys::SaveDataType type;
+        INSERT_PADDING_BYTES(0x6);
+        std::array<u8, 0x10> user_id;
+        u64_le save_id;
+        u64_le title_id;
+        u64_le save_image_size;
+        INSERT_PADDING_BYTES(0x28);
+    };
+    static_assert(sizeof(SaveDataInfo) == 0x60, "SaveDataInfo has incorrect size.");
+
+    std::vector<SaveDataInfo> info;
+    u64 next_entry_index = 0;
+};
+
 FSP_SRV::FSP_SRV() : ServiceFramework("fsp-srv") {
+    // clang-format off
     static const FunctionInfo functions[] = {
         {0, nullptr, "MountContent"},
         {1, &FSP_SRV::Initialize, "Initialize"},
@@ -485,7 +626,7 @@ FSP_SRV::FSP_SRV() : ServiceFramework("fsp-srv") {
         {58, nullptr, "ReadSaveDataFileSystemExtraData"},
         {59, nullptr, "WriteSaveDataFileSystemExtraData"},
         {60, nullptr, "OpenSaveDataInfoReader"},
-        {61, nullptr, "OpenSaveDataInfoReaderBySaveDataSpaceId"},
+        {61, &FSP_SRV::OpenSaveDataInfoReaderBySaveDataSpaceId, "OpenSaveDataInfoReaderBySaveDataSpaceId"},
         {62, nullptr, "OpenCacheStorageList"},
         {64, nullptr, "OpenSaveDataInternalStorageFileSystem"},
         {65, nullptr, "UpdateSaveDataMacForDebug"},
@@ -544,6 +685,7 @@ FSP_SRV::FSP_SRV() : ServiceFramework("fsp-srv") {
         {1009, nullptr, "GetAndClearMemoryReportInfo"},
         {1100, nullptr, "OverrideSaveDataTransferTokenSignVerificationKey"},
     };
+    // clang-format on
     RegisterHandlers(functions);
 }
 
@@ -602,7 +744,7 @@ void FSP_SRV::MountSaveData(Kernel::HLERequestContext& ctx) {
 
     if (dir.Failed()) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 0};
-        rb.Push(ResultCode(ErrorModule::FS, FileSys::ErrCodes::TitleNotFound));
+        rb.Push(FileSys::ERROR_ENTITY_NOT_FOUND);
         return;
     }
 
@@ -618,6 +760,15 @@ void FSP_SRV::OpenReadOnlySaveDataFileSystem(Kernel::HLERequestContext& ctx) {
     MountSaveData(ctx);
 }
 
+void FSP_SRV::OpenSaveDataInfoReaderBySaveDataSpaceId(Kernel::HLERequestContext& ctx) {
+    IPC::RequestParser rp{ctx};
+    const auto space = rp.PopRaw<FileSys::SaveDataSpaceId>();
+
+    IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+    rb.Push(RESULT_SUCCESS);
+    rb.PushIpcInterface<ISaveDataInfoReader>(std::make_shared<ISaveDataInfoReader>(space));
+}
+
 void FSP_SRV::GetGlobalAccessLogMode(Kernel::HLERequestContext& ctx) {
     LOG_WARNING(Service_FS, "(STUBBED) called");
 
@@ -685,7 +836,7 @@ void FSP_SRV::OpenRomStorage(Kernel::HLERequestContext& ctx) {
               static_cast<u8>(storage_id), title_id);
 
     IPC::ResponseBuilder rb{ctx, 2};
-    rb.Push(ResultCode(ErrorModule::FS, FileSys::ErrCodes::TitleNotFound));
+    rb.Push(FileSys::ERROR_ENTITY_NOT_FOUND);
 }
 
 } // namespace Service::FileSystem

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

@@ -25,6 +25,7 @@ private:
     void CreateSaveData(Kernel::HLERequestContext& ctx);
     void MountSaveData(Kernel::HLERequestContext& ctx);
     void OpenReadOnlySaveDataFileSystem(Kernel::HLERequestContext& ctx);
+    void OpenSaveDataInfoReaderBySaveDataSpaceId(Kernel::HLERequestContext& ctx);
     void GetGlobalAccessLogMode(Kernel::HLERequestContext& ctx);
     void OpenDataStorageByCurrentProcess(Kernel::HLERequestContext& ctx);
     void OpenDataStorageByDataId(Kernel::HLERequestContext& ctx);

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

@@ -40,6 +40,29 @@ enum class JoystickId : std::size_t {
     Joystick_Right,
 };
 
+static std::size_t NPadIdToIndex(u32 npad_id) {
+    switch (npad_id) {
+    case 0:
+    case 1:
+    case 2:
+    case 3:
+    case 4:
+    case 5:
+    case 6:
+    case 7:
+        return npad_id;
+    case 8:
+    case NPAD_HANDHELD:
+        return 8;
+    case 9:
+    case NPAD_UNKNOWN:
+        return 9;
+    default:
+        UNIMPLEMENTED_MSG("Unknown npad id {}", npad_id);
+        return 0;
+    }
+}
+
 Controller_NPad::Controller_NPad() = default;
 Controller_NPad::~Controller_NPad() = default;
 
@@ -288,10 +311,11 @@ void Controller_NPad::OnUpdate(u8* data, std::size_t data_len) {
         switch (controller_type) {
         case NPadControllerType::Handheld:
             handheld_entry.connection_status.raw = 0;
-            handheld_entry.connection_status.IsConnected.Assign(1);
-            if (!Settings::values.use_docked_mode) {
-                handheld_entry.connection_status.IsWired.Assign(1);
-            }
+            handheld_entry.connection_status.IsWired.Assign(1);
+            handheld_entry.connection_status.IsLeftJoyConnected.Assign(1);
+            handheld_entry.connection_status.IsRightJoyConnected.Assign(1);
+            handheld_entry.connection_status.IsLeftJoyWired.Assign(1);
+            handheld_entry.connection_status.IsRightJoyWired.Assign(1);
             handheld_entry.pad_states.raw = pad_state.raw;
             handheld_entry.l_stick = lstick_entry;
             handheld_entry.r_stick = rstick_entry;
@@ -310,6 +334,7 @@ void Controller_NPad::OnUpdate(u8* data, std::size_t data_len) {
             dual_entry.pad_states.raw = pad_state.raw;
             dual_entry.l_stick = lstick_entry;
             dual_entry.r_stick = rstick_entry;
+            break;
         case NPadControllerType::JoyLeft:
             left_entry.connection_status.raw = 0;
 
@@ -370,16 +395,30 @@ void Controller_NPad::SetSupportedNPadIdTypes(u8* data, std::size_t length) {
     supported_npad_id_types.clear();
     supported_npad_id_types.resize(length / sizeof(u32));
     std::memcpy(supported_npad_id_types.data(), data, length);
+    bool had_controller_update = false;
     for (std::size_t i = 0; i < connected_controllers.size(); i++) {
         auto& controller = connected_controllers[i];
         if (!controller.is_connected) {
             continue;
         }
         if (!IsControllerSupported(PREFERRED_CONTROLLER)) {
-            controller.type = DecideBestController(PREFERRED_CONTROLLER);
-            InitNewlyAddedControler(i);
+            const auto best_type = DecideBestController(PREFERRED_CONTROLLER);
+            const bool is_handheld = (best_type == NPadControllerType::Handheld ||
+                                      PREFERRED_CONTROLLER == NPadControllerType::Handheld);
+            if (is_handheld) {
+                controller.type = NPadControllerType::None;
+                controller.is_connected = false;
+                AddNewController(best_type);
+            } else {
+                controller.type = best_type;
+                InitNewlyAddedControler(i);
+            }
+            had_controller_update = true;
         }
     }
+    if (had_controller_update) {
+        styleset_changed_event->Signal();
+    }
 }
 
 void Controller_NPad::GetSupportedNpadIdTypes(u32* data, std::size_t max_length) {
@@ -392,8 +431,10 @@ std::size_t Controller_NPad::GetSupportedNPadIdTypesSize() const {
 }
 
 void Controller_NPad::SetHoldType(NpadHoldType joy_hold_type) {
+    styleset_changed_event->Signal();
     hold_type = joy_hold_type;
 }
+
 Controller_NPad::NpadHoldType Controller_NPad::GetHoldType() const {
     return hold_type;
 }
@@ -455,15 +496,11 @@ void Controller_NPad::AddNewController(NPadControllerType controller) {
 }
 
 void Controller_NPad::ConnectNPad(u32 npad_id) {
-    if (npad_id >= connected_controllers.size())
-        return;
-    connected_controllers[npad_id].is_connected = true;
+    connected_controllers[NPadIdToIndex(npad_id)].is_connected = true;
 }
 
 void Controller_NPad::DisconnectNPad(u32 npad_id) {
-    if (npad_id >= connected_controllers.size())
-        return;
-    connected_controllers[npad_id].is_connected = false;
+    connected_controllers[NPadIdToIndex(npad_id)].is_connected = false;
 }
 
 Controller_NPad::LedPattern Controller_NPad::GetLedPattern(u32 npad_id) {

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

@@ -286,10 +286,10 @@ public:
             {519, nullptr, "GetPalmaOperationResult"},
             {520, nullptr, "ReadPalmaPlayLog"},
             {521, nullptr, "ResetPalmaPlayLog"},
-            {522, nullptr, "SetIsPalmaAllConnectable"},
+            {522, &Hid::SetIsPalmaAllConnectable, "SetIsPalmaAllConnectable"},
             {523, nullptr, "SetIsPalmaPairedConnectable"},
             {524, nullptr, "PairPalma"},
-            {525, nullptr, "SetPalmaBoostMode"},
+            {525, &Hid::SetPalmaBoostMode, "SetPalmaBoostMode"},
             {1000, nullptr, "SetNpadCommunicationMode"},
             {1001, nullptr, "GetNpadCommunicationMode"},
         };
@@ -596,6 +596,18 @@ private:
         rb.Push(RESULT_SUCCESS);
         LOG_WARNING(Service_HID, "(STUBBED) called");
     }
+
+    void SetIsPalmaAllConnectable(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+        LOG_WARNING(Service_HID, "(STUBBED) called");
+    }
+
+    void SetPalmaBoostMode(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+        LOG_WARNING(Service_HID, "(STUBBED) called");
+    }
 };
 
 class HidDbg final : public ServiceFramework<HidDbg> {

src/core/hle/service/ldr/ldr.cpp

@@ -4,7 +4,10 @@
 
 #include <memory>
 #include <fmt/format.h>
+#include <mbedtls/sha256.h>
 
+#include "common/alignment.h"
+#include "common/hex_util.h"
 #include "core/hle/ipc_helpers.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/service/ldr/ldr.h"
@@ -13,6 +16,21 @@
 
 namespace Service::LDR {
 
+constexpr ResultCode ERROR_INVALID_MEMORY_STATE{ErrorModule::Loader, 51};
+constexpr ResultCode ERROR_INVALID_NRO{ErrorModule::Loader, 52};
+constexpr ResultCode ERROR_INVALID_NRR{ErrorModule::Loader, 53};
+constexpr ResultCode ERROR_MISSING_NRR_HASH{ErrorModule::Loader, 54};
+constexpr ResultCode ERROR_MAXIMUM_NRO{ErrorModule::Loader, 55};
+constexpr ResultCode ERROR_MAXIMUM_NRR{ErrorModule::Loader, 56};
+constexpr ResultCode ERROR_ALREADY_LOADED{ErrorModule::Loader, 57};
+constexpr ResultCode ERROR_INVALID_ALIGNMENT{ErrorModule::Loader, 81};
+constexpr ResultCode ERROR_INVALID_SIZE{ErrorModule::Loader, 82};
+constexpr ResultCode ERROR_INVALID_NRO_ADDRESS{ErrorModule::Loader, 84};
+constexpr ResultCode ERROR_INVALID_NRR_ADDRESS{ErrorModule::Loader, 85};
+constexpr ResultCode ERROR_NOT_INITIALIZED{ErrorModule::Loader, 87};
+
+constexpr u64 MAXIMUM_LOADED_RO = 0x40;
+
 class DebugMonitor final : public ServiceFramework<DebugMonitor> {
 public:
     explicit DebugMonitor() : ServiceFramework{"ldr:dmnt"} {
@@ -64,9 +82,9 @@ public:
         // clang-format off
         static const FunctionInfo functions[] = {
             {0, &RelocatableObject::LoadNro, "LoadNro"},
-            {1, nullptr, "UnloadNro"},
+            {1, &RelocatableObject::UnloadNro, "UnloadNro"},
             {2, &RelocatableObject::LoadNrr, "LoadNrr"},
-            {3, nullptr, "UnloadNrr"},
+            {3, &RelocatableObject::UnloadNrr, "UnloadNrr"},
             {4, &RelocatableObject::Initialize, "Initialize"},
         };
         // clang-format on
@@ -75,9 +93,123 @@ public:
     }
 
     void LoadNrr(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        rp.Skip(2, false);
+        const VAddr nrr_addr{rp.Pop<VAddr>()};
+        const u64 nrr_size{rp.Pop<u64>()};
+
+        if (!initialized) {
+            LOG_ERROR(Service_LDR, "LDR:RO not initialized before use!");
+            IPC::ResponseBuilder rb{ctx, 2};
+            rb.Push(ERROR_NOT_INITIALIZED);
+            return;
+        }
+
+        if (nrr.size() >= MAXIMUM_LOADED_RO) {
+            LOG_ERROR(Service_LDR, "Loading new NRR would exceed the maximum number of loaded NRRs "
+                                   "(0x40)! Failing...");
+            IPC::ResponseBuilder rb{ctx, 2};
+            rb.Push(ERROR_MAXIMUM_NRR);
+            return;
+        }
+
+        // NRR Address does not fall on 0x1000 byte boundary
+        if (!Common::Is4KBAligned(nrr_addr)) {
+            LOG_ERROR(Service_LDR, "NRR Address has invalid alignment (actual {:016X})!", nrr_addr);
+            IPC::ResponseBuilder rb{ctx, 2};
+            rb.Push(ERROR_INVALID_ALIGNMENT);
+            return;
+        }
+
+        // NRR Size is zero or causes overflow
+        if (nrr_addr + nrr_size <= nrr_addr || nrr_size == 0 || !Common::Is4KBAligned(nrr_size)) {
+            LOG_ERROR(Service_LDR, "NRR Size is invalid! (nrr_address={:016X}, nrr_size={:016X})",
+                      nrr_addr, nrr_size);
+            IPC::ResponseBuilder rb{ctx, 2};
+            rb.Push(ERROR_INVALID_SIZE);
+            return;
+        }
+        // Read NRR data from memory
+        std::vector<u8> nrr_data(nrr_size);
+        Memory::ReadBlock(nrr_addr, nrr_data.data(), nrr_size);
+        NRRHeader header;
+        std::memcpy(&header, nrr_data.data(), sizeof(NRRHeader));
+
+        if (header.magic != Common::MakeMagic('N', 'R', 'R', '0')) {
+            LOG_ERROR(Service_LDR, "NRR did not have magic 'NRR0' (actual {:08X})!", header.magic);
+            IPC::ResponseBuilder rb{ctx, 2};
+            rb.Push(ERROR_INVALID_NRR);
+            return;
+        }
+
+        if (header.size != nrr_size) {
+            LOG_ERROR(Service_LDR,
+                      "NRR header reported size did not match LoadNrr parameter size! "
+                      "(header_size={:016X}, loadnrr_size={:016X})",
+                      header.size, nrr_size);
+            IPC::ResponseBuilder rb{ctx, 2};
+            rb.Push(ERROR_INVALID_SIZE);
+            return;
+        }
+
+        if (Core::CurrentProcess()->GetTitleID() != header.title_id) {
+            LOG_ERROR(Service_LDR,
+                      "Attempting to load NRR with title ID other than current process. (actual "
+                      "{:016X})!",
+                      header.title_id);
+            IPC::ResponseBuilder rb{ctx, 2};
+            rb.Push(ERROR_INVALID_NRR);
+            return;
+        }
+
+        std::vector<SHA256Hash> hashes;
+
+        // Copy all hashes in the NRR (specified by hash count/hash offset) into vector.
+        for (std::size_t i = header.hash_offset;
+             i < (header.hash_offset + (header.hash_count * sizeof(SHA256Hash))); i += 8) {
+            SHA256Hash hash;
+            std::memcpy(hash.data(), nrr_data.data() + i, sizeof(SHA256Hash));
+            hashes.emplace_back(hash);
+        }
+
+        nrr.insert_or_assign(nrr_addr, std::move(hashes));
+
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+    }
+
+    void UnloadNrr(Kernel::HLERequestContext& ctx) {
+        if (!initialized) {
+            LOG_ERROR(Service_LDR, "LDR:RO not initialized before use!");
+            IPC::ResponseBuilder rb{ctx, 2};
+            rb.Push(ERROR_NOT_INITIALIZED);
+            return;
+        }
+
+        IPC::RequestParser rp{ctx};
+        rp.Skip(2, false);
+        const auto nrr_addr{rp.Pop<VAddr>()};
+
+        if (!Common::Is4KBAligned(nrr_addr)) {
+            LOG_ERROR(Service_LDR, "NRR Address has invalid alignment (actual {:016X})!", nrr_addr);
+            IPC::ResponseBuilder rb{ctx, 2};
+            rb.Push(ERROR_INVALID_ALIGNMENT);
+            return;
+        }
+
+        const auto iter = nrr.find(nrr_addr);
+        if (iter == nrr.end()) {
+            LOG_ERROR(Service_LDR,
+                      "Attempting to unload NRR which has not been loaded! (addr={:016X})",
+                      nrr_addr);
+            IPC::ResponseBuilder rb{ctx, 2};
+            rb.Push(ERROR_INVALID_NRR_ADDRESS);
+            return;
+        }
+
+        nrr.erase(iter);
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_LDR, "(STUBBED) called");
     }
 
     void LoadNro(Kernel::HLERequestContext& ctx) {
@@ -88,33 +220,253 @@ public:
         const VAddr bss_addr{rp.Pop<VAddr>()};
         const u64 bss_size{rp.Pop<u64>()};
 
+        if (!initialized) {
+            LOG_ERROR(Service_LDR, "LDR:RO not initialized before use!");
+            IPC::ResponseBuilder rb{ctx, 2};
+            rb.Push(ERROR_NOT_INITIALIZED);
+            return;
+        }
+
+        if (nro.size() >= MAXIMUM_LOADED_RO) {
+            LOG_ERROR(Service_LDR, "Loading new NRO would exceed the maximum number of loaded NROs "
+                                   "(0x40)! Failing...");
+            IPC::ResponseBuilder rb{ctx, 2};
+            rb.Push(ERROR_MAXIMUM_NRO);
+            return;
+        }
+
+        // NRO Address does not fall on 0x1000 byte boundary
+        if (!Common::Is4KBAligned(nro_addr)) {
+            LOG_ERROR(Service_LDR, "NRO Address has invalid alignment (actual {:016X})!", nro_addr);
+            IPC::ResponseBuilder rb{ctx, 2};
+            rb.Push(ERROR_INVALID_ALIGNMENT);
+            return;
+        }
+
+        // NRO Size or BSS Size is zero or causes overflow
+        const auto nro_size_valid =
+            nro_size != 0 && nro_addr + nro_size > nro_addr && Common::Is4KBAligned(nro_size);
+        const auto bss_size_valid =
+            nro_size + bss_size >= nro_size && (bss_size == 0 || bss_addr + bss_size > bss_addr);
+
+        if (!nro_size_valid || !bss_size_valid) {
+            LOG_ERROR(Service_LDR,
+                      "NRO Size or BSS Size is invalid! (nro_address={:016X}, nro_size={:016X}, "
+                      "bss_address={:016X}, bss_size={:016X})",
+                      nro_addr, nro_size, bss_addr, bss_size);
+            IPC::ResponseBuilder rb{ctx, 2};
+            rb.Push(ERROR_INVALID_SIZE);
+            return;
+        }
+
         // Read NRO data from memory
         std::vector<u8> nro_data(nro_size);
         Memory::ReadBlock(nro_addr, nro_data.data(), nro_size);
 
-        // Load NRO as new executable module
-        const VAddr addr{*Core::CurrentProcess()->VMManager().FindFreeRegion(nro_size + bss_size)};
-        Loader::AppLoader_NRO::LoadNro(nro_data, fmt::format("nro-{:08x}", addr), addr);
+        SHA256Hash hash{};
+        mbedtls_sha256(nro_data.data(), nro_data.size(), hash.data(), 0);
 
-        // TODO(bunnei): This is an incomplete implementation. It was tested with Super Mario Party.
-        // It is currently missing:
-        // - Signature checks with LoadNRR
-        // - Checking if a module has already been loaded
-        // - Using/validating BSS, etc. params (these are used from NRO header instead)
-        // - Error checking
-        // - ...Probably other things
+        // NRO Hash is already loaded
+        if (std::any_of(nro.begin(), nro.end(), [&hash](const std::pair<VAddr, NROInfo>& info) {
+                return info.second.hash == hash;
+            })) {
+            LOG_ERROR(Service_LDR, "NRO is already loaded!");
+            IPC::ResponseBuilder rb{ctx, 2};
+            rb.Push(ERROR_ALREADY_LOADED);
+            return;
+        }
+
+        // NRO Hash is not in any loaded NRR
+        if (!IsValidNROHash(hash)) {
+            LOG_ERROR(Service_LDR,
+                      "NRO hash is not present in any currently loaded NRRs (hash={})!",
+                      Common::HexArrayToString(hash));
+            IPC::ResponseBuilder rb{ctx, 2};
+            rb.Push(ERROR_MISSING_NRR_HASH);
+            return;
+        }
+
+        NROHeader header;
+        std::memcpy(&header, nro_data.data(), sizeof(NROHeader));
+
+        if (!IsValidNRO(header, nro_size, bss_size)) {
+            LOG_ERROR(Service_LDR, "NRO was invalid!");
+            IPC::ResponseBuilder rb{ctx, 2};
+            rb.Push(ERROR_INVALID_NRO);
+            return;
+        }
+
+        // Load NRO as new executable module
+        auto* process = Core::CurrentProcess();
+        auto& vm_manager = process->VMManager();
+        auto map_address = vm_manager.FindFreeRegion(nro_size + bss_size);
+
+        if (!map_address.Succeeded() ||
+            *map_address + nro_size + bss_size > vm_manager.GetAddressSpaceEndAddress()) {
+
+            LOG_ERROR(Service_LDR,
+                      "General error while allocation memory or no available memory to allocate!");
+            IPC::ResponseBuilder rb{ctx, 2};
+            rb.Push(ERROR_INVALID_MEMORY_STATE);
+            return;
+        }
+
+        ASSERT(process->MirrorMemory(*map_address, nro_addr, nro_size,
+                                     Kernel::MemoryState::ModuleCodeStatic) == RESULT_SUCCESS);
+        ASSERT(process->UnmapMemory(nro_addr, 0, nro_size) == RESULT_SUCCESS);
+
+        if (bss_size > 0) {
+            ASSERT(process->MirrorMemory(*map_address + nro_size, bss_addr, bss_size,
+                                         Kernel::MemoryState::ModuleCodeStatic) == RESULT_SUCCESS);
+            ASSERT(process->UnmapMemory(bss_addr, 0, bss_size) == RESULT_SUCCESS);
+        }
+
+        vm_manager.ReprotectRange(*map_address, header.text_size,
+                                  Kernel::VMAPermission::ReadExecute);
+        vm_manager.ReprotectRange(*map_address + header.ro_offset, header.ro_size,
+                                  Kernel::VMAPermission::Read);
+        vm_manager.ReprotectRange(*map_address + header.rw_offset, header.rw_size,
+                                  Kernel::VMAPermission::ReadWrite);
+
+        Core::System::GetInstance().ArmInterface(0).ClearInstructionCache();
+        Core::System::GetInstance().ArmInterface(1).ClearInstructionCache();
+        Core::System::GetInstance().ArmInterface(2).ClearInstructionCache();
+        Core::System::GetInstance().ArmInterface(3).ClearInstructionCache();
+
+        nro.insert_or_assign(*map_address, NROInfo{hash, nro_size + bss_size});
 
         IPC::ResponseBuilder rb{ctx, 4};
         rb.Push(RESULT_SUCCESS);
-        rb.Push(addr);
-        LOG_WARNING(Service_LDR, "(STUBBED) called");
+        rb.Push(*map_address);
+    }
+
+    void UnloadNro(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        rp.Skip(2, false);
+        const VAddr mapped_addr{rp.PopRaw<VAddr>()};
+        const VAddr heap_addr{rp.PopRaw<VAddr>()};
+
+        if (!initialized) {
+            LOG_ERROR(Service_LDR, "LDR:RO not initialized before use!");
+            IPC::ResponseBuilder rb{ctx, 2};
+            rb.Push(ERROR_NOT_INITIALIZED);
+            return;
+        }
+
+        if (!Common::Is4KBAligned(mapped_addr) || !Common::Is4KBAligned(heap_addr)) {
+            LOG_ERROR(Service_LDR,
+                      "NRO/BSS Address has invalid alignment (actual nro_addr={:016X}, "
+                      "bss_addr={:016X})!",
+                      mapped_addr, heap_addr);
+            IPC::ResponseBuilder rb{ctx, 2};
+            rb.Push(ERROR_INVALID_ALIGNMENT);
+            return;
+        }
+
+        const auto iter = nro.find(mapped_addr);
+        if (iter == nro.end()) {
+            LOG_ERROR(Service_LDR,
+                      "The NRO attempting to unmap was not mapped or has an invalid address "
+                      "(actual {:016X})!",
+                      mapped_addr);
+            IPC::ResponseBuilder rb{ctx, 2};
+            rb.Push(ERROR_INVALID_NRO_ADDRESS);
+            return;
+        }
+
+        auto* process = Core::CurrentProcess();
+        auto& vm_manager = process->VMManager();
+        const auto& nro_size = iter->second.size;
+
+        ASSERT(process->MirrorMemory(heap_addr, mapped_addr, nro_size,
+                                     Kernel::MemoryState::ModuleCodeStatic) == RESULT_SUCCESS);
+        ASSERT(process->UnmapMemory(mapped_addr, 0, nro_size) == RESULT_SUCCESS);
+
+        Core::System::GetInstance().ArmInterface(0).ClearInstructionCache();
+        Core::System::GetInstance().ArmInterface(1).ClearInstructionCache();
+        Core::System::GetInstance().ArmInterface(2).ClearInstructionCache();
+        Core::System::GetInstance().ArmInterface(3).ClearInstructionCache();
+
+        nro.erase(iter);
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
     }
 
     void Initialize(Kernel::HLERequestContext& ctx) {
+        initialized = true;
+
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
         LOG_WARNING(Service_LDR, "(STUBBED) called");
     }
+
+private:
+    using SHA256Hash = std::array<u8, 0x20>;
+
+    struct NROHeader {
+        u32_le entrypoint_insn;
+        u32_le mod_offset;
+        INSERT_PADDING_WORDS(2);
+        u32_le magic;
+        INSERT_PADDING_WORDS(1);
+        u32_le nro_size;
+        INSERT_PADDING_WORDS(1);
+        u32_le text_offset;
+        u32_le text_size;
+        u32_le ro_offset;
+        u32_le ro_size;
+        u32_le rw_offset;
+        u32_le rw_size;
+        u32_le bss_size;
+        INSERT_PADDING_WORDS(1);
+        std::array<u8, 0x20> build_id;
+        INSERT_PADDING_BYTES(0x20);
+    };
+    static_assert(sizeof(NROHeader) == 0x80, "NROHeader has invalid size.");
+
+    struct NRRHeader {
+        u32_le magic;
+        INSERT_PADDING_BYTES(0x1C);
+        u64_le title_id_mask;
+        u64_le title_id_pattern;
+        std::array<u8, 0x100> modulus;
+        std::array<u8, 0x100> signature_1;
+        std::array<u8, 0x100> signature_2;
+        u64_le title_id;
+        u32_le size;
+        INSERT_PADDING_BYTES(4);
+        u32_le hash_offset;
+        u32_le hash_count;
+        INSERT_PADDING_BYTES(8);
+    };
+    static_assert(sizeof(NRRHeader) == 0x350, "NRRHeader has incorrect size.");
+
+    struct NROInfo {
+        SHA256Hash hash;
+        u64 size;
+    };
+
+    bool initialized = false;
+
+    std::map<VAddr, NROInfo> nro;
+    std::map<VAddr, std::vector<SHA256Hash>> nrr;
+
+    bool IsValidNROHash(const SHA256Hash& hash) {
+        return std::any_of(
+            nrr.begin(), nrr.end(), [&hash](const std::pair<VAddr, std::vector<SHA256Hash>>& p) {
+                return std::find(p.second.begin(), p.second.end(), hash) != p.second.end();
+            });
+    }
+
+    static bool IsValidNRO(const NROHeader& header, u64 nro_size, u64 bss_size) {
+        return header.magic == Common::MakeMagic('N', 'R', 'O', '0') &&
+               header.nro_size == nro_size && header.bss_size == bss_size &&
+               header.ro_offset == header.text_offset + header.text_size &&
+               header.rw_offset == header.ro_offset + header.ro_size &&
+               nro_size == header.rw_offset + header.rw_size &&
+               Common::Is4KBAligned(header.text_size) && Common::Is4KBAligned(header.ro_size) &&
+               Common::Is4KBAligned(header.rw_size);
+    }
 };
 
 void InstallInterfaces(SM::ServiceManager& sm) {

src/core/hle/service/nfp/nfp.cpp

@@ -212,7 +212,7 @@ private:
         IPC::ResponseBuilder rb{ctx, 2};
         auto amiibo = nfp_interface.GetAmiiboBuffer();
         TagInfo tag_info{};
-        std::memcpy(tag_info.uuid.data(), amiibo.uuid.data(), sizeof(tag_info.uuid.size()));
+        tag_info.uuid = amiibo.uuid;
         tag_info.uuid_length = static_cast<u8>(tag_info.uuid.size());
 
         tag_info.protocol = 1; // TODO(ogniK): Figure out actual values

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

@@ -2,6 +2,9 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include "common/logging/log.h"
+#include "core/file_sys/control_metadata.h"
+#include "core/file_sys/patch_manager.h"
 #include "core/hle/ipc_helpers.h"
 #include "core/hle/kernel/hle_ipc.h"
 #include "core/hle/service/ns/ns.h"
@@ -118,7 +121,7 @@ public:
             {305, nullptr, "TerminateSystemApplet"},
             {306, nullptr, "LaunchOverlayApplet"},
             {307, nullptr, "TerminateOverlayApplet"},
-            {400, nullptr, "GetApplicationControlData"},
+            {400, &IApplicationManagerInterface::GetApplicationControlData, "GetApplicationControlData"},
             {401, nullptr, "InvalidateAllApplicationControlCache"},
             {402, nullptr, "RequestDownloadApplicationControlData"},
             {403, nullptr, "GetMaxApplicationControlCacheCount"},
@@ -243,6 +246,65 @@ public:
 
         RegisterHandlers(functions);
     }
+
+    void GetApplicationControlData(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        const auto flag = rp.PopRaw<u64>();
+        LOG_DEBUG(Service_NS, "called with flag={:016X}", flag);
+
+        const auto title_id = rp.PopRaw<u64>();
+
+        const auto size = ctx.GetWriteBufferSize();
+
+        const FileSys::PatchManager pm{title_id};
+        const auto control = pm.GetControlMetadata();
+
+        std::vector<u8> out;
+
+        if (control.first != nullptr) {
+            if (size < 0x4000) {
+                LOG_ERROR(Service_NS,
+                          "output buffer is too small! (actual={:016X}, expected_min=0x4000)",
+                          size);
+                IPC::ResponseBuilder rb{ctx, 2};
+                // TODO(DarkLordZach): Find a better error code for this.
+                rb.Push(ResultCode(-1));
+                return;
+            }
+
+            out.resize(0x4000);
+            const auto bytes = control.first->GetRawBytes();
+            std::memcpy(out.data(), bytes.data(), bytes.size());
+        } else {
+            LOG_WARNING(Service_NS, "missing NACP data for title_id={:016X}, defaulting to zeros.",
+                        title_id);
+            out.resize(std::min<u64>(0x4000, size));
+        }
+
+        if (control.second != nullptr) {
+            if (size < 0x4000 + control.second->GetSize()) {
+                LOG_ERROR(Service_NS,
+                          "output buffer is too small! (actual={:016X}, expected_min={:016X})",
+                          size, 0x4000 + control.second->GetSize());
+                IPC::ResponseBuilder rb{ctx, 2};
+                // TODO(DarkLordZach): Find a better error code for this.
+                rb.Push(ResultCode(-1));
+                return;
+            }
+
+            out.resize(0x4000 + control.second->GetSize());
+            control.second->Read(out.data() + 0x4000, control.second->GetSize());
+        } else {
+            LOG_WARNING(Service_NS, "missing icon data for title_id={:016X}, defaulting to zeros.",
+                        title_id);
+        }
+
+        ctx.WriteBuffer(out);
+
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(RESULT_SUCCESS);
+        rb.Push<u32>(static_cast<u32>(out.size()));
+    }
 };
 
 class IApplicationVersionInterface final : public ServiceFramework<IApplicationVersionInterface> {

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

@@ -351,6 +351,14 @@ void PL_U::GetSharedFontInOrderOfPriority(Kernel::HLERequestContext& ctx) {
         font_sizes.push_back(region.size);
     }
 
+    // Resize buffers if game requests smaller size output.
+    font_codes.resize(
+        std::min<std::size_t>(font_codes.size(), ctx.GetWriteBufferSize(0) / sizeof(u32)));
+    font_offsets.resize(
+        std::min<std::size_t>(font_offsets.size(), ctx.GetWriteBufferSize(1) / sizeof(u32)));
+    font_sizes.resize(
+        std::min<std::size_t>(font_sizes.size(), ctx.GetWriteBufferSize(2) / sizeof(u32)));
+
     ctx.WriteBuffer(font_codes, 0);
     ctx.WriteBuffer(font_offsets, 1);
     ctx.WriteBuffer(font_sizes, 2);

src/core/hle/service/service.cpp

@@ -80,8 +80,8 @@ namespace Service {
  * Creates a function string for logging, complete with the name (or header code, depending
  * on what's passed in) the port name, and all the cmd_buff arguments.
  */
-static std::string MakeFunctionString(const char* name, const char* port_name,
-                                      const u32* cmd_buff) {
+[[maybe_unused]] static std::string MakeFunctionString(const char* name, const char* port_name,
+                                                       const u32* cmd_buff) {
     // Number of params == bits 0-5 + bits 6-11
     int num_params = (cmd_buff[0] & 0x3F) + ((cmd_buff[0] >> 6) & 0x3F);
 

src/core/hle/service/spl/module.cpp

@@ -3,18 +3,23 @@
 // Refer to the license.txt file included.
 
 #include <algorithm>
+#include <chrono>
 #include <cstdlib>
+#include <ctime>
+#include <functional>
 #include <vector>
 #include "common/logging/log.h"
 #include "core/hle/ipc_helpers.h"
 #include "core/hle/service/spl/csrng.h"
 #include "core/hle/service/spl/module.h"
 #include "core/hle/service/spl/spl.h"
+#include "core/settings.h"
 
 namespace Service::SPL {
 
 Module::Interface::Interface(std::shared_ptr<Module> module, const char* name)
-    : ServiceFramework(name), module(std::move(module)) {}
+    : ServiceFramework(name), module(std::move(module)),
+      rng(Settings::values.rng_seed.value_or(std::time(nullptr))) {}
 
 Module::Interface::~Interface() = default;
 
@@ -23,8 +28,9 @@ void Module::Interface::GetRandomBytes(Kernel::HLERequestContext& ctx) {
 
     std::size_t size = ctx.GetWriteBufferSize();
 
+    std::uniform_int_distribution<u16> distribution(0, std::numeric_limits<u8>::max());
     std::vector<u8> data(size);
-    std::generate(data.begin(), data.end(), std::rand);
+    std::generate(data.begin(), data.end(), [&] { return static_cast<u8>(distribution(rng)); });
 
     ctx.WriteBuffer(data);
 

src/core/hle/service/spl/module.h

@@ -4,6 +4,7 @@
 
 #pragma once
 
+#include <random>
 #include "core/hle/service/service.h"
 
 namespace Service::SPL {
@@ -19,6 +20,9 @@ public:
 
     protected:
         std::shared_ptr<Module> module;
+
+    private:
+        std::mt19937 rng;
     };
 };
 

src/core/hle/service/time/interface.cpp

@@ -21,9 +21,10 @@ Time::Time(std::shared_ptr<Module> time, const char* name)
         {102, nullptr, "GetStandardUserSystemClockInitialYear"},
         {200, nullptr, "IsStandardNetworkSystemClockAccuracySufficient"},
         {300, nullptr, "CalculateMonotonicSystemClockBaseTimePoint"},
-        {400, nullptr, "GetClockSnapshot"},
+        {400, &Time::GetClockSnapshot, "GetClockSnapshot"},
         {401, nullptr, "GetClockSnapshotFromSystemClockContext"},
-        {500, nullptr, "CalculateStandardUserSystemClockDifferenceByUser"},
+        {500, &Time::CalculateStandardUserSystemClockDifferenceByUser,
+         "CalculateStandardUserSystemClockDifferenceByUser"},
         {501, nullptr, "CalculateSpanBetween"},
     };
     RegisterHandlers(functions);

src/core/hle/service/time/time.cpp

@@ -15,6 +15,44 @@
 
 namespace Service::Time {
 
+static void PosixToCalendar(u64 posix_time, CalendarTime& calendar_time,
+                            CalendarAdditionalInfo& additional_info,
+                            [[maybe_unused]] const TimeZoneRule& /*rule*/) {
+    const std::time_t time(posix_time);
+    const std::tm* tm = std::localtime(&time);
+    if (tm == nullptr) {
+        calendar_time = {};
+        additional_info = {};
+        return;
+    }
+    calendar_time.year = tm->tm_year + 1900;
+    calendar_time.month = tm->tm_mon + 1;
+    calendar_time.day = tm->tm_mday;
+    calendar_time.hour = tm->tm_hour;
+    calendar_time.minute = tm->tm_min;
+    calendar_time.second = tm->tm_sec;
+
+    additional_info.day_of_week = tm->tm_wday;
+    additional_info.day_of_year = tm->tm_yday;
+    std::memcpy(additional_info.name.data(), "UTC", sizeof("UTC"));
+    additional_info.utc_offset = 0;
+}
+
+static u64 CalendarToPosix(const CalendarTime& calendar_time,
+                           [[maybe_unused]] const TimeZoneRule& /*rule*/) {
+    std::tm time{};
+    time.tm_year = calendar_time.year - 1900;
+    time.tm_mon = calendar_time.month - 1;
+    time.tm_mday = calendar_time.day;
+
+    time.tm_hour = calendar_time.hour;
+    time.tm_min = calendar_time.minute;
+    time.tm_sec = calendar_time.second;
+
+    std::time_t epoch_time = std::mktime(&time);
+    return static_cast<u64>(epoch_time);
+}
+
 class ISystemClock final : public ServiceFramework<ISystemClock> {
 public:
     ISystemClock() : ServiceFramework("ISystemClock") {
@@ -80,8 +118,8 @@ public:
             {5, nullptr, "GetTimeZoneRuleVersion"},
             {100, &ITimeZoneService::ToCalendarTime, "ToCalendarTime"},
             {101, &ITimeZoneService::ToCalendarTimeWithMyRule, "ToCalendarTimeWithMyRule"},
-            {201, nullptr, "ToPosixTime"},
-            {202, nullptr, "ToPosixTimeWithMyRule"},
+            {201, &ITimeZoneService::ToPosixTime, "ToPosixTime"},
+            {202, &ITimeZoneService::ToPosixTimeWithMyRule, "ToPosixTimeWithMyRule"},
         };
         RegisterHandlers(functions);
     }
@@ -151,24 +189,29 @@ private:
         rb.PushRaw(additional_info);
     }
 
-    void PosixToCalendar(u64 posix_time, CalendarTime& calendar_time,
-                         CalendarAdditionalInfo& additional_info, const TimeZoneRule& /*rule*/) {
-        std::time_t t(posix_time);
-        std::tm* tm = std::localtime(&t);
-        if (!tm) {
-            return;
-        }
-        calendar_time.year = tm->tm_year + 1900;
-        calendar_time.month = tm->tm_mon + 1;
-        calendar_time.day = tm->tm_mday;
-        calendar_time.hour = tm->tm_hour;
-        calendar_time.minute = tm->tm_min;
-        calendar_time.second = tm->tm_sec;
+    void ToPosixTime(Kernel::HLERequestContext& ctx) {
+        // TODO(ogniK): Figure out how to handle multiple times
+        LOG_WARNING(Service_Time, "(STUBBED) called");
+        IPC::RequestParser rp{ctx};
+        auto calendar_time = rp.PopRaw<CalendarTime>();
+        auto posix_time = CalendarToPosix(calendar_time, {});
 
-        additional_info.day_of_week = tm->tm_wday;
-        additional_info.day_of_year = tm->tm_yday;
-        std::memcpy(additional_info.name.data(), "UTC", sizeof("UTC"));
-        additional_info.utc_offset = 0;
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushRaw<u32>(1); // Amount of times we're returning
+        ctx.WriteBuffer(&posix_time, sizeof(u64));
+    }
+
+    void ToPosixTimeWithMyRule(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_Time, "(STUBBED) called");
+        IPC::RequestParser rp{ctx};
+        auto calendar_time = rp.PopRaw<CalendarTime>();
+        auto posix_time = CalendarToPosix(calendar_time, {});
+
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushRaw<u32>(1); // Amount of times we're returning
+        ctx.WriteBuffer(&posix_time, sizeof(u64));
     }
 };
 
@@ -207,6 +250,70 @@ void Module::Interface::GetStandardLocalSystemClock(Kernel::HLERequestContext& c
     LOG_DEBUG(Service_Time, "called");
 }
 
+void Module::Interface::GetClockSnapshot(Kernel::HLERequestContext& ctx) {
+    LOG_DEBUG(Service_Time, "called");
+
+    IPC::RequestParser rp{ctx};
+    auto unknown_u8 = rp.PopRaw<u8>();
+
+    ClockSnapshot clock_snapshot{};
+
+    const s64 time_since_epoch{std::chrono::duration_cast<std::chrono::seconds>(
+                                   std::chrono::system_clock::now().time_since_epoch())
+                                   .count()};
+    CalendarTime calendar_time{};
+    const std::time_t time(time_since_epoch);
+    const std::tm* tm = std::localtime(&time);
+    if (tm == nullptr) {
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(ResultCode(-1)); // TODO(ogniK): Find appropriate error code
+        return;
+    }
+    SteadyClockTimePoint steady_clock_time_point{CoreTiming::cyclesToMs(CoreTiming::GetTicks()) /
+                                                 1000};
+
+    LocationName location_name{"UTC"};
+    calendar_time.year = tm->tm_year + 1900;
+    calendar_time.month = tm->tm_mon + 1;
+    calendar_time.day = tm->tm_mday;
+    calendar_time.hour = tm->tm_hour;
+    calendar_time.minute = tm->tm_min;
+    calendar_time.second = tm->tm_sec;
+    clock_snapshot.system_posix_time = time_since_epoch;
+    clock_snapshot.network_posix_time = time_since_epoch;
+    clock_snapshot.system_calendar_time = calendar_time;
+    clock_snapshot.network_calendar_time = calendar_time;
+
+    CalendarAdditionalInfo additional_info{};
+    PosixToCalendar(time_since_epoch, calendar_time, additional_info, {});
+
+    clock_snapshot.system_calendar_info = additional_info;
+    clock_snapshot.network_calendar_info = additional_info;
+
+    clock_snapshot.steady_clock_timepoint = steady_clock_time_point;
+    clock_snapshot.location_name = location_name;
+    clock_snapshot.clock_auto_adjustment_enabled = 1;
+    clock_snapshot.ipc_u8 = unknown_u8;
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(RESULT_SUCCESS);
+    ctx.WriteBuffer(&clock_snapshot, sizeof(ClockSnapshot));
+}
+
+void Module::Interface::CalculateStandardUserSystemClockDifferenceByUser(
+    Kernel::HLERequestContext& ctx) {
+    LOG_DEBUG(Service_Time, "called");
+
+    IPC::RequestParser rp{ctx};
+    const auto snapshot_a = rp.PopRaw<ClockSnapshot>();
+    const auto snapshot_b = rp.PopRaw<ClockSnapshot>();
+    const u64 difference =
+        snapshot_b.user_clock_context.offset - snapshot_a.user_clock_context.offset;
+
+    IPC::ResponseBuilder rb{ctx, 4};
+    rb.Push(RESULT_SUCCESS);
+    rb.PushRaw<u64>(difference);
+}
+
 Module::Interface::Interface(std::shared_ptr<Module> time, const char* name)
     : ServiceFramework(name), time(std::move(time)) {}
 

src/core/hle/service/time/time.h

@@ -5,6 +5,7 @@
 #pragma once
 
 #include <array>
+#include "common/common_funcs.h"
 #include "core/hle/service/service.h"
 
 namespace Service::Time {
@@ -53,6 +54,23 @@ struct SystemClockContext {
 static_assert(sizeof(SystemClockContext) == 0x20,
               "SystemClockContext structure has incorrect size");
 
+struct ClockSnapshot {
+    SystemClockContext user_clock_context;
+    SystemClockContext network_clock_context;
+    s64_le system_posix_time;
+    s64_le network_posix_time;
+    CalendarTime system_calendar_time;
+    CalendarTime network_calendar_time;
+    CalendarAdditionalInfo system_calendar_info;
+    CalendarAdditionalInfo network_calendar_info;
+    SteadyClockTimePoint steady_clock_timepoint;
+    LocationName location_name;
+    u8 clock_auto_adjustment_enabled;
+    u8 ipc_u8;
+    INSERT_PADDING_BYTES(2);
+};
+static_assert(sizeof(ClockSnapshot) == 0xd0, "ClockSnapshot is an invalid size");
+
 class Module final {
 public:
     class Interface : public ServiceFramework<Interface> {
@@ -65,6 +83,8 @@ public:
         void GetStandardSteadyClock(Kernel::HLERequestContext& ctx);
         void GetTimeZoneService(Kernel::HLERequestContext& ctx);
         void GetStandardLocalSystemClock(Kernel::HLERequestContext& ctx);
+        void GetClockSnapshot(Kernel::HLERequestContext& ctx);
+        void CalculateStandardUserSystemClockDifferenceByUser(Kernel::HLERequestContext& ctx);
 
     protected:
         std::shared_ptr<Module> time;

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

@@ -237,6 +237,22 @@ private:
     Data data{};
 };
 
+/// Represents a parcel containing one int '0' as its data
+/// Used by DetachBuffer and Disconnect
+class IGBPEmptyResponseParcel : public Parcel {
+protected:
+    void SerializeData() override {
+        Write(data);
+    }
+
+private:
+    struct Data {
+        u32_le unk_0;
+    };
+
+    Data data{};
+};
+
 class IGBPSetPreallocatedBufferRequestParcel : public Parcel {
 public:
     explicit IGBPSetPreallocatedBufferRequestParcel(std::vector<u8> buffer)
@@ -554,6 +570,12 @@ private:
             ctx.WriteBuffer(response.Serialize());
         } else if (transaction == TransactionId::CancelBuffer) {
             LOG_CRITICAL(Service_VI, "(STUBBED) called, transaction=CancelBuffer");
+        } else if (transaction == TransactionId::Disconnect ||
+                   transaction == TransactionId::DetachBuffer) {
+            const auto buffer = ctx.ReadBuffer();
+
+            IGBPEmptyResponseParcel response{};
+            ctx.WriteBuffer(response.Serialize());
         } else {
             ASSERT_MSG(false, "Unimplemented");
         }

src/core/loader/nro.cpp

@@ -12,10 +12,12 @@
 #include "common/swap.h"
 #include "core/core.h"
 #include "core/file_sys/control_metadata.h"
+#include "core/file_sys/romfs_factory.h"
 #include "core/file_sys/vfs_offset.h"
 #include "core/gdbstub/gdbstub.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/vm_manager.h"
+#include "core/hle/service/filesystem/filesystem.h"
 #include "core/loader/nro.h"
 #include "core/loader/nso.h"
 #include "core/memory.h"
@@ -168,17 +170,20 @@ static constexpr u32 PageAlignSize(u32 size) {
                     arg_data.size());
     }
 
-    // Read MOD header
-    ModHeader mod_header{};
     // Default .bss to NRO header bss size if MOD0 section doesn't exist
     u32 bss_size{PageAlignSize(nro_header.bss_size)};
+
+    // Read MOD header
+    ModHeader mod_header{};
     std::memcpy(&mod_header, program_image.data() + nro_header.module_header_offset,
                 sizeof(ModHeader));
+
     const bool has_mod_header{mod_header.magic == Common::MakeMagic('M', 'O', 'D', '0')};
     if (has_mod_header) {
         // Resize program image to include .bss section and page align each section
         bss_size = PageAlignSize(mod_header.bss_end_offset - mod_header.bss_start_offset);
     }
+
     codeset.DataSegment().size += bss_size;
     program_image.resize(static_cast<u32>(program_image.size()) + bss_size);
 
@@ -208,6 +213,9 @@ ResultStatus AppLoader_NRO::Load(Kernel::Process& process) {
         return ResultStatus::ErrorLoadingNRO;
     }
 
+    if (romfs != nullptr)
+        Service::FileSystem::RegisterRomFS(std::make_unique<FileSys::RomFSFactory>(*this));
+
     process.Run(base_address, Kernel::THREADPRIO_DEFAULT, Memory::DEFAULT_STACK_SIZE);
 
     is_loaded = true;

src/core/loader/nso.cpp

@@ -154,7 +154,7 @@ std::optional<VAddr> AppLoader_NSO::LoadModule(const FileSys::VfsFile& file, VAd
     program_image.resize(image_size);
 
     // Apply patches if necessary
-    if (pm && pm->HasNSOPatch(nso_header.build_id)) {
+    if (pm && (pm->HasNSOPatch(nso_header.build_id) || Settings::values.dump_nso)) {
         std::vector<u8> pi_header(program_image.size() + 0x100);
         std::memcpy(pi_header.data(), &nso_header, sizeof(NsoHeader));
         std::memcpy(pi_header.data() + 0x100, program_image.data(), program_image.size());

src/core/settings.h

@@ -6,6 +6,7 @@
 
 #include <array>
 #include <atomic>
+#include <optional>
 #include <string>
 #include "common/common_types.h"
 
@@ -114,8 +115,9 @@ struct Values {
     // System
     bool use_docked_mode;
     bool enable_nfc;
-    int current_user;
-    int language_index;
+    std::optional<u32> rng_seed;
+    s32 current_user;
+    s32 language_index;
 
     // Controls
     std::array<std::string, NativeButton::NumButtons> buttons;
@@ -157,6 +159,7 @@ struct Values {
     bool use_gdbstub;
     u16 gdbstub_port;
     std::string program_args;
+    bool dump_nso;
 
     // WebService
     bool enable_telemetry;

src/video_core/CMakeLists.txt

@@ -21,6 +21,7 @@ add_library(video_core STATIC
     macro_interpreter.h
     memory_manager.cpp
     memory_manager.h
+    rasterizer_cache.cpp
     rasterizer_cache.h
     rasterizer_interface.h
     renderer_base.cpp
@@ -33,6 +34,7 @@ add_library(video_core STATIC
     renderer_opengl/gl_rasterizer.h
     renderer_opengl/gl_rasterizer_cache.cpp
     renderer_opengl/gl_rasterizer_cache.h
+    renderer_opengl/gl_resource_manager.cpp
     renderer_opengl/gl_resource_manager.h
     renderer_opengl/gl_shader_cache.cpp
     renderer_opengl/gl_shader_cache.h

src/video_core/engines/maxwell_3d.cpp

@@ -34,9 +34,41 @@ void Maxwell3D::InitializeRegisterDefaults() {
     // Depth range near/far is not always set, but is expected to be the default 0.0f, 1.0f. This is
     // needed for ARMS.
     for (std::size_t viewport{}; viewport < Regs::NumViewports; ++viewport) {
-        regs.viewport[viewport].depth_range_near = 0.0f;
-        regs.viewport[viewport].depth_range_far = 1.0f;
+        regs.viewports[viewport].depth_range_near = 0.0f;
+        regs.viewports[viewport].depth_range_far = 1.0f;
     }
+    // Doom and Bomberman seems to use the uninitialized registers and just enable blend
+    // so initialize blend registers with sane values
+    regs.blend.equation_rgb = Regs::Blend::Equation::Add;
+    regs.blend.factor_source_rgb = Regs::Blend::Factor::One;
+    regs.blend.factor_dest_rgb = Regs::Blend::Factor::Zero;
+    regs.blend.equation_a = Regs::Blend::Equation::Add;
+    regs.blend.factor_source_a = Regs::Blend::Factor::One;
+    regs.blend.factor_dest_a = Regs::Blend::Factor::Zero;
+    for (std::size_t blend_index = 0; blend_index < Regs::NumRenderTargets; blend_index++) {
+        regs.independent_blend[blend_index].equation_rgb = Regs::Blend::Equation::Add;
+        regs.independent_blend[blend_index].factor_source_rgb = Regs::Blend::Factor::One;
+        regs.independent_blend[blend_index].factor_dest_rgb = Regs::Blend::Factor::Zero;
+        regs.independent_blend[blend_index].equation_a = Regs::Blend::Equation::Add;
+        regs.independent_blend[blend_index].factor_source_a = Regs::Blend::Factor::One;
+        regs.independent_blend[blend_index].factor_dest_a = Regs::Blend::Factor::Zero;
+    }
+    regs.stencil_front_op_fail = Regs::StencilOp::Keep;
+    regs.stencil_front_op_zfail = Regs::StencilOp::Keep;
+    regs.stencil_front_op_zpass = Regs::StencilOp::Keep;
+    regs.stencil_front_func_func = Regs::ComparisonOp::Always;
+    regs.stencil_front_func_mask = 0xFFFFFFFF;
+    regs.stencil_front_mask = 0xFFFFFFFF;
+    regs.stencil_two_side_enable = 1;
+    regs.stencil_back_op_fail = Regs::StencilOp::Keep;
+    regs.stencil_back_op_zfail = Regs::StencilOp::Keep;
+    regs.stencil_back_op_zpass = Regs::StencilOp::Keep;
+    regs.stencil_back_func_func = Regs::ComparisonOp::Always;
+    regs.stencil_back_func_mask = 0xFFFFFFFF;
+    regs.stencil_back_mask = 0xFFFFFFFF;
+    // TODO(Rodrigo): Most games do not set a point size. I think this is a case of a
+    // register carrying a default value. Assume it's OpenGL's default (1).
+    regs.point_size = 1.0f;
 }
 
 void Maxwell3D::CallMacroMethod(u32 method, std::vector<u32> parameters) {
@@ -92,7 +124,13 @@ void Maxwell3D::WriteReg(u32 method, u32 value, u32 remaining_params) {
         debug_context->OnEvent(Tegra::DebugContext::Event::MaxwellCommandLoaded, nullptr);
     }
 
-    regs.reg_array[method] = value;
+    if (regs.reg_array[method] != value) {
+        regs.reg_array[method] = value;
+        if (method >= MAXWELL3D_REG_INDEX(vertex_attrib_format) &&
+            method < MAXWELL3D_REG_INDEX(vertex_attrib_format) + regs.vertex_attrib_format.size()) {
+            dirty_flags.vertex_attrib_format = true;
+        }
+    }
 
     switch (method) {
     case MAXWELL3D_REG_INDEX(macros.data): {

src/video_core/engines/maxwell_3d.h

@@ -345,6 +345,14 @@ public:
             Invert = 6,
             IncrWrap = 7,
             DecrWrap = 8,
+            KeepOGL = 0x1E00,
+            ZeroOGL = 0,
+            ReplaceOGL = 0x1E01,
+            IncrOGL = 0x1E02,
+            DecrOGL = 0x1E03,
+            InvertOGL = 0x150A,
+            IncrWrapOGL = 0x8507,
+            DecrWrapOGL = 0x8508,
         };
 
         enum class MemoryLayout : u32 {
@@ -462,6 +470,77 @@ public:
             }
         };
 
+        struct ColorMask {
+            union {
+                u32 raw;
+                BitField<0, 4, u32> R;
+                BitField<4, 4, u32> G;
+                BitField<8, 4, u32> B;
+                BitField<12, 4, u32> A;
+            };
+        };
+
+        struct ViewportTransform {
+            f32 scale_x;
+            f32 scale_y;
+            f32 scale_z;
+            f32 translate_x;
+            f32 translate_y;
+            f32 translate_z;
+            INSERT_PADDING_WORDS(2);
+
+            MathUtil::Rectangle<s32> GetRect() const {
+                return {
+                    GetX(),               // left
+                    GetY() + GetHeight(), // top
+                    GetX() + GetWidth(),  // right
+                    GetY()                // bottom
+                };
+            };
+
+            s32 GetX() const {
+                return static_cast<s32>(std::max(0.0f, translate_x - std::fabs(scale_x)));
+            }
+
+            s32 GetY() const {
+                return static_cast<s32>(std::max(0.0f, translate_y - std::fabs(scale_y)));
+            }
+
+            s32 GetWidth() const {
+                return static_cast<s32>(translate_x + std::fabs(scale_x)) - GetX();
+            }
+
+            s32 GetHeight() const {
+                return static_cast<s32>(translate_y + std::fabs(scale_y)) - GetY();
+            }
+        };
+
+        struct ScissorTest {
+            u32 enable;
+            union {
+                BitField<0, 16, u32> min_x;
+                BitField<16, 16, u32> max_x;
+            };
+            union {
+                BitField<0, 16, u32> min_y;
+                BitField<16, 16, u32> max_y;
+            };
+            u32 fill;
+        };
+
+        struct ViewPort {
+            union {
+                BitField<0, 16, u32> x;
+                BitField<16, 16, u32> width;
+            };
+            union {
+                BitField<0, 16, u32> y;
+                BitField<16, 16, u32> height;
+            };
+            float depth_range_near;
+            float depth_range_far;
+        };
+
         bool IsShaderConfigEnabled(std::size_t index) const {
             // The VertexB is always enabled.
             if (index == static_cast<std::size_t>(Regs::ShaderProgram::VertexB)) {
@@ -487,55 +566,11 @@ public:
 
                 INSERT_PADDING_WORDS(0x2E);
 
-                RenderTargetConfig rt[NumRenderTargets];
+                std::array<RenderTargetConfig, NumRenderTargets> rt;
 
-                struct {
-                    f32 scale_x;
-                    f32 scale_y;
-                    f32 scale_z;
-                    f32 translate_x;
-                    f32 translate_y;
-                    f32 translate_z;
-                    INSERT_PADDING_WORDS(2);
+                std::array<ViewportTransform, NumViewports> viewport_transform;
 
-                    MathUtil::Rectangle<s32> GetRect() const {
-                        return {
-                            GetX(),               // left
-                            GetY() + GetHeight(), // top
-                            GetX() + GetWidth(),  // right
-                            GetY()                // bottom
-                        };
-                    };
-
-                    s32 GetX() const {
-                        return static_cast<s32>(std::max(0.0f, translate_x - std::fabs(scale_x)));
-                    }
-
-                    s32 GetY() const {
-                        return static_cast<s32>(std::max(0.0f, translate_y - std::fabs(scale_y)));
-                    }
-
-                    s32 GetWidth() const {
-                        return static_cast<s32>(translate_x + std::fabs(scale_x)) - GetX();
-                    }
-
-                    s32 GetHeight() const {
-                        return static_cast<s32>(translate_y + std::fabs(scale_y)) - GetY();
-                    }
-                } viewport_transform[NumViewports];
-
-                struct {
-                    union {
-                        BitField<0, 16, u32> x;
-                        BitField<16, 16, u32> width;
-                    };
-                    union {
-                        BitField<0, 16, u32> y;
-                        BitField<16, 16, u32> height;
-                    };
-                    float depth_range_near;
-                    float depth_range_far;
-                } viewport[NumViewports];
+                std::array<ViewPort, NumViewports> viewports;
 
                 INSERT_PADDING_WORDS(0x1D);
 
@@ -553,25 +588,19 @@ public:
 
                 INSERT_PADDING_WORDS(0x17);
 
-                struct {
-                    u32 enable;
-                    union {
-                        BitField<0, 16, u32> min_x;
-                        BitField<16, 16, u32> max_x;
-                    };
-                    union {
-                        BitField<0, 16, u32> min_y;
-                        BitField<16, 16, u32> max_y;
-                    };
-                } scissor_test;
+                std::array<ScissorTest, NumViewports> scissor_test;
 
-                INSERT_PADDING_WORDS(0x52);
+                INSERT_PADDING_WORDS(0x15);
 
                 s32 stencil_back_func_ref;
                 u32 stencil_back_mask;
                 u32 stencil_back_func_mask;
 
-                INSERT_PADDING_WORDS(0x13);
+                INSERT_PADDING_WORDS(0xC);
+
+                u32 color_mask_common;
+
+                INSERT_PADDING_WORDS(0x6);
 
                 u32 rt_separate_frag_data;
 
@@ -646,8 +675,14 @@ public:
                 ComparisonOp depth_test_func;
                 float alpha_test_ref;
                 ComparisonOp alpha_test_func;
-
-                INSERT_PADDING_WORDS(0x9);
+                u32 draw_tfb_stride;
+                struct {
+                    float r;
+                    float g;
+                    float b;
+                    float a;
+                } blend_color;
+                INSERT_PADDING_WORDS(0x4);
 
                 struct {
                     u32 separate_alpha;
@@ -672,7 +707,9 @@ public:
                 u32 stencil_front_func_mask;
                 u32 stencil_front_mask;
 
-                INSERT_PADDING_WORDS(0x3);
+                INSERT_PADDING_WORDS(0x2);
+
+                u32 frag_color_clamp;
 
                 union {
                     BitField<4, 1, u32> triangle_rast_flip;
@@ -690,7 +727,12 @@ public:
 
                 u32 zeta_enable;
 
-                INSERT_PADDING_WORDS(0x8);
+                union {
+                    BitField<0, 1, u32> alpha_to_coverage;
+                    BitField<4, 1, u32> alpha_to_one;
+                } multisample_control;
+
+                INSERT_PADDING_WORDS(0x7);
 
                 struct {
                     u32 tsc_address_high;
@@ -841,8 +883,9 @@ public:
                     BitField<6, 4, u32> RT;
                     BitField<10, 11, u32> layer;
                 } clear_buffers;
-
-                INSERT_PADDING_WORDS(0x4B);
+                INSERT_PADDING_WORDS(0xB);
+                std::array<ColorMask, NumRenderTargets> color_mask;
+                INSERT_PADDING_WORDS(0x38);
 
                 struct {
                     u32 query_address_high;
@@ -983,6 +1026,12 @@ public:
     State state{};
     MemoryManager& memory_manager;
 
+    struct DirtyFlags {
+        bool vertex_attrib_format = true;
+    };
+
+    DirtyFlags dirty_flags;
+
     /// Reads a register value located at the input method address
     u32 GetRegisterValue(u32 method) const;
 
@@ -1065,8 +1114,8 @@ private:
 ASSERT_REG_POSITION(macros, 0x45);
 ASSERT_REG_POSITION(tfb_enabled, 0x1D1);
 ASSERT_REG_POSITION(rt, 0x200);
-ASSERT_REG_POSITION(viewport_transform[0], 0x280);
-ASSERT_REG_POSITION(viewport, 0x300);
+ASSERT_REG_POSITION(viewport_transform, 0x280);
+ASSERT_REG_POSITION(viewports, 0x300);
 ASSERT_REG_POSITION(vertex_buffer, 0x35D);
 ASSERT_REG_POSITION(clear_color[0], 0x360);
 ASSERT_REG_POSITION(clear_depth, 0x364);
@@ -1075,6 +1124,7 @@ ASSERT_REG_POSITION(scissor_test, 0x380);
 ASSERT_REG_POSITION(stencil_back_func_ref, 0x3D5);
 ASSERT_REG_POSITION(stencil_back_mask, 0x3D6);
 ASSERT_REG_POSITION(stencil_back_func_mask, 0x3D7);
+ASSERT_REG_POSITION(color_mask_common, 0x3E4);
 ASSERT_REG_POSITION(rt_separate_frag_data, 0x3EB);
 ASSERT_REG_POSITION(zeta, 0x3F8);
 ASSERT_REG_POSITION(vertex_attrib_format, 0x458);
@@ -1087,6 +1137,10 @@ ASSERT_REG_POSITION(depth_write_enabled, 0x4BA);
 ASSERT_REG_POSITION(alpha_test_enabled, 0x4BB);
 ASSERT_REG_POSITION(d3d_cull_mode, 0x4C2);
 ASSERT_REG_POSITION(depth_test_func, 0x4C3);
+ASSERT_REG_POSITION(alpha_test_ref, 0x4C4);
+ASSERT_REG_POSITION(alpha_test_func, 0x4C5);
+ASSERT_REG_POSITION(draw_tfb_stride, 0x4C6);
+ASSERT_REG_POSITION(blend_color, 0x4C7);
 ASSERT_REG_POSITION(blend, 0x4CF);
 ASSERT_REG_POSITION(stencil_enable, 0x4E0);
 ASSERT_REG_POSITION(stencil_front_op_fail, 0x4E1);
@@ -1096,10 +1150,12 @@ ASSERT_REG_POSITION(stencil_front_func_func, 0x4E4);
 ASSERT_REG_POSITION(stencil_front_func_ref, 0x4E5);
 ASSERT_REG_POSITION(stencil_front_func_mask, 0x4E6);
 ASSERT_REG_POSITION(stencil_front_mask, 0x4E7);
+ASSERT_REG_POSITION(frag_color_clamp, 0x4EA);
 ASSERT_REG_POSITION(screen_y_control, 0x4EB);
 ASSERT_REG_POSITION(vb_element_base, 0x50D);
 ASSERT_REG_POSITION(point_size, 0x546);
 ASSERT_REG_POSITION(zeta_enable, 0x54E);
+ASSERT_REG_POSITION(multisample_control, 0x54F);
 ASSERT_REG_POSITION(tsc, 0x557);
 ASSERT_REG_POSITION(tic, 0x55D);
 ASSERT_REG_POSITION(stencil_two_side_enable, 0x565);
@@ -1117,6 +1173,7 @@ ASSERT_REG_POSITION(instanced_arrays, 0x620);
 ASSERT_REG_POSITION(cull, 0x646);
 ASSERT_REG_POSITION(logic_op, 0x671);
 ASSERT_REG_POSITION(clear_buffers, 0x674);
+ASSERT_REG_POSITION(color_mask, 0x680);
 ASSERT_REG_POSITION(query, 0x6C0);
 ASSERT_REG_POSITION(vertex_array[0], 0x700);
 ASSERT_REG_POSITION(independent_blend, 0x780);

src/video_core/memory_manager.cpp

@@ -4,18 +4,21 @@
 
 #include "common/alignment.h"
 #include "common/assert.h"
+#include "common/logging/log.h"
 #include "video_core/memory_manager.h"
 
 namespace Tegra {
 
 GPUVAddr MemoryManager::AllocateSpace(u64 size, u64 align) {
-    std::optional<GPUVAddr> gpu_addr = FindFreeBlock(size, align);
-    ASSERT(gpu_addr);
+    const std::optional<GPUVAddr> gpu_addr{FindFreeBlock(0, size, align, PageStatus::Unmapped)};
 
-    for (u64 offset = 0; offset < size; offset += PAGE_SIZE) {
-        VAddr& slot = PageSlot(*gpu_addr + offset);
+    ASSERT_MSG(gpu_addr, "unable to find available GPU memory");
+
+    for (u64 offset{}; offset < size; offset += PAGE_SIZE) {
+        VAddr& slot{PageSlot(*gpu_addr + offset)};
 
         ASSERT(slot == static_cast<u64>(PageStatus::Unmapped));
+
         slot = static_cast<u64>(PageStatus::Allocated);
     }
 
@@ -23,10 +26,11 @@ GPUVAddr MemoryManager::AllocateSpace(u64 size, u64 align) {
 }
 
 GPUVAddr MemoryManager::AllocateSpace(GPUVAddr gpu_addr, u64 size, u64 align) {
-    for (u64 offset = 0; offset < size; offset += PAGE_SIZE) {
-        VAddr& slot = PageSlot(gpu_addr + offset);
+    for (u64 offset{}; offset < size; offset += PAGE_SIZE) {
+        VAddr& slot{PageSlot(gpu_addr + offset)};
 
         ASSERT(slot == static_cast<u64>(PageStatus::Unmapped));
+
         slot = static_cast<u64>(PageStatus::Allocated);
     }
 
@@ -34,17 +38,19 @@ GPUVAddr MemoryManager::AllocateSpace(GPUVAddr gpu_addr, u64 size, u64 align) {
 }
 
 GPUVAddr MemoryManager::MapBufferEx(VAddr cpu_addr, u64 size) {
-    std::optional<GPUVAddr> gpu_addr = FindFreeBlock(size, PAGE_SIZE);
-    ASSERT(gpu_addr);
+    const std::optional<GPUVAddr> gpu_addr{FindFreeBlock(0, size, PAGE_SIZE, PageStatus::Unmapped)};
 
-    for (u64 offset = 0; offset < size; offset += PAGE_SIZE) {
-        VAddr& slot = PageSlot(*gpu_addr + offset);
+    ASSERT_MSG(gpu_addr, "unable to find available GPU memory");
+
+    for (u64 offset{}; offset < size; offset += PAGE_SIZE) {
+        VAddr& slot{PageSlot(*gpu_addr + offset)};
 
         ASSERT(slot == static_cast<u64>(PageStatus::Unmapped));
+
         slot = cpu_addr + offset;
     }
 
-    MappedRegion region{cpu_addr, *gpu_addr, size};
+    const MappedRegion region{cpu_addr, *gpu_addr, size};
     mapped_regions.push_back(region);
 
     return *gpu_addr;
@@ -53,14 +59,31 @@ GPUVAddr MemoryManager::MapBufferEx(VAddr cpu_addr, u64 size) {
 GPUVAddr MemoryManager::MapBufferEx(VAddr cpu_addr, GPUVAddr gpu_addr, u64 size) {
     ASSERT((gpu_addr & PAGE_MASK) == 0);
 
-    for (u64 offset = 0; offset < size; offset += PAGE_SIZE) {
-        VAddr& slot = PageSlot(gpu_addr + offset);
+    if (PageSlot(gpu_addr) != static_cast<u64>(PageStatus::Allocated)) {
+        // Page has been already mapped. In this case, we must find a new area of memory to use that
+        // is different than the specified one. Super Mario Odyssey hits this scenario when changing
+        // areas, but we do not want to overwrite the old pages.
+        // TODO(bunnei): We need to write a hardware test to confirm this behavior.
+
+        LOG_ERROR(HW_GPU, "attempting to map addr 0x{:016X}, which is not available!", gpu_addr);
+
+        const std::optional<GPUVAddr> new_gpu_addr{
+            FindFreeBlock(gpu_addr, size, PAGE_SIZE, PageStatus::Allocated)};
+
+        ASSERT_MSG(new_gpu_addr, "unable to find available GPU memory");
+
+        gpu_addr = *new_gpu_addr;
+    }
+
+    for (u64 offset{}; offset < size; offset += PAGE_SIZE) {
+        VAddr& slot{PageSlot(gpu_addr + offset)};
 
         ASSERT(slot == static_cast<u64>(PageStatus::Allocated));
+
         slot = cpu_addr + offset;
     }
 
-    MappedRegion region{cpu_addr, gpu_addr, size};
+    const MappedRegion region{cpu_addr, gpu_addr, size};
     mapped_regions.push_back(region);
 
     return gpu_addr;
@@ -69,11 +92,12 @@ GPUVAddr MemoryManager::MapBufferEx(VAddr cpu_addr, GPUVAddr gpu_addr, u64 size)
 GPUVAddr MemoryManager::UnmapBuffer(GPUVAddr gpu_addr, u64 size) {
     ASSERT((gpu_addr & PAGE_MASK) == 0);
 
-    for (u64 offset = 0; offset < size; offset += PAGE_SIZE) {
-        VAddr& slot = PageSlot(gpu_addr + offset);
+    for (u64 offset{}; offset < size; offset += PAGE_SIZE) {
+        VAddr& slot{PageSlot(gpu_addr + offset)};
 
         ASSERT(slot != static_cast<u64>(PageStatus::Allocated) &&
                slot != static_cast<u64>(PageStatus::Unmapped));
+
         slot = static_cast<u64>(PageStatus::Unmapped);
     }
 
@@ -97,13 +121,14 @@ GPUVAddr MemoryManager::GetRegionEnd(GPUVAddr region_start) const {
     return {};
 }
 
-std::optional<GPUVAddr> MemoryManager::FindFreeBlock(u64 size, u64 align) {
-    GPUVAddr gpu_addr = 0;
-    u64 free_space = 0;
+std::optional<GPUVAddr> MemoryManager::FindFreeBlock(GPUVAddr region_start, u64 size, u64 align,
+                                                     PageStatus status) {
+    GPUVAddr gpu_addr{region_start};
+    u64 free_space{};
     align = (align + PAGE_MASK) & ~PAGE_MASK;
 
     while (gpu_addr + free_space < MAX_ADDRESS) {
-        if (!IsPageMapped(gpu_addr + free_space)) {
+        if (PageSlot(gpu_addr + free_space) == static_cast<u64>(status)) {
             free_space += PAGE_SIZE;
             if (free_space >= size) {
                 return gpu_addr;
@@ -119,7 +144,7 @@ std::optional<GPUVAddr> MemoryManager::FindFreeBlock(u64 size, u64 align) {
 }
 
 std::optional<VAddr> MemoryManager::GpuToCpuAddress(GPUVAddr gpu_addr) {
-    VAddr base_addr = PageSlot(gpu_addr);
+    const VAddr base_addr{PageSlot(gpu_addr)};
 
     if (base_addr == static_cast<u64>(PageStatus::Allocated) ||
         base_addr == static_cast<u64>(PageStatus::Unmapped)) {
@@ -133,19 +158,15 @@ std::vector<GPUVAddr> MemoryManager::CpuToGpuAddress(VAddr cpu_addr) const {
     std::vector<GPUVAddr> results;
     for (const auto& region : mapped_regions) {
         if (cpu_addr >= region.cpu_addr && cpu_addr < (region.cpu_addr + region.size)) {
-            u64 offset = cpu_addr - region.cpu_addr;
+            const u64 offset{cpu_addr - region.cpu_addr};
             results.push_back(region.gpu_addr + offset);
         }
     }
     return results;
 }
 
-bool MemoryManager::IsPageMapped(GPUVAddr gpu_addr) {
-    return PageSlot(gpu_addr) != static_cast<u64>(PageStatus::Unmapped);
-}
-
 VAddr& MemoryManager::PageSlot(GPUVAddr gpu_addr) {
-    auto& block = page_table[(gpu_addr >> (PAGE_BITS + PAGE_TABLE_BITS)) & PAGE_TABLE_MASK];
+    auto& block{page_table[(gpu_addr >> (PAGE_BITS + PAGE_TABLE_BITS)) & PAGE_TABLE_MASK]};
     if (!block) {
         block = std::make_unique<PageBlock>();
         block->fill(static_cast<VAddr>(PageStatus::Unmapped));

src/video_core/memory_manager.h

@@ -34,15 +34,15 @@ public:
     static constexpr u64 PAGE_MASK = PAGE_SIZE - 1;
 
 private:
-    std::optional<GPUVAddr> FindFreeBlock(u64 size, u64 align = 1);
-    bool IsPageMapped(GPUVAddr gpu_addr);
-    VAddr& PageSlot(GPUVAddr gpu_addr);
-
     enum class PageStatus : u64 {
         Unmapped = 0xFFFFFFFFFFFFFFFFULL,
         Allocated = 0xFFFFFFFFFFFFFFFEULL,
     };
 
+    std::optional<GPUVAddr> FindFreeBlock(GPUVAddr region_start, u64 size, u64 align,
+                                          PageStatus status);
+    VAddr& PageSlot(GPUVAddr gpu_addr);
+
     static constexpr u64 MAX_ADDRESS{0x10000000000ULL};
     static constexpr u64 PAGE_TABLE_BITS{10};
     static constexpr u64 PAGE_TABLE_SIZE{1 << PAGE_TABLE_BITS};

src/video_core/rasterizer_cache.cpp

@@ -0,0 +1,7 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "video_core/rasterizer_cache.h"
+
+RasterizerCacheObject::~RasterizerCacheObject() = default;

src/video_core/rasterizer_cache.h

@@ -5,18 +5,19 @@
 #pragma once
 
 #include <set>
+#include <unordered_map>
 
 #include <boost/icl/interval_map.hpp>
 #include <boost/range/iterator_range_core.hpp>
 
 #include "common/common_types.h"
-#include "core/core.h"
 #include "core/settings.h"
 #include "video_core/rasterizer_interface.h"
-#include "video_core/renderer_base.h"
 
 class RasterizerCacheObject {
 public:
+    virtual ~RasterizerCacheObject();
+
     /// Gets the address of the shader in guest memory, required for cache management
     virtual VAddr GetAddr() const = 0;
 
@@ -64,6 +65,8 @@ class RasterizerCache : NonCopyable {
     friend class RasterizerCacheObject;
 
 public:
+    explicit RasterizerCache(VideoCore::RasterizerInterface& rasterizer) : rasterizer{rasterizer} {}
+
     /// Write any cached resources overlapping the specified region back to memory
     void FlushRegion(Tegra::GPUVAddr addr, size_t size) {
         const auto& objects{GetSortedObjectsFromRegion(addr, size)};
@@ -86,45 +89,39 @@ public:
 
     /// Invalidates everything in the cache
     void InvalidateAll() {
-        while (object_cache.begin() != object_cache.end()) {
-            Unregister(*object_cache.begin()->second.begin());
+        while (interval_cache.begin() != interval_cache.end()) {
+            Unregister(*interval_cache.begin()->second.begin());
         }
     }
 
 protected:
     /// Tries to get an object from the cache with the specified address
     T TryGet(VAddr addr) const {
-        const ObjectInterval interval{addr};
-        for (auto& pair : boost::make_iterator_range(object_cache.equal_range(interval))) {
-            for (auto& cached_object : pair.second) {
-                if (cached_object->GetAddr() == addr) {
-                    return cached_object;
-                }
-            }
-        }
+        const auto iter = map_cache.find(addr);
+        if (iter != map_cache.end())
+            return iter->second;
         return nullptr;
     }
 
     /// Register an object into the cache
     void Register(const T& object) {
         object->SetIsRegistered(true);
-        object_cache.add({GetInterval(object), ObjectSet{object}});
-        auto& rasterizer = Core::System::GetInstance().Renderer().Rasterizer();
+        interval_cache.add({GetInterval(object), ObjectSet{object}});
+        map_cache.insert({object->GetAddr(), object});
         rasterizer.UpdatePagesCachedCount(object->GetAddr(), object->GetSizeInBytes(), 1);
     }
 
     /// Unregisters an object from the cache
     void Unregister(const T& object) {
         object->SetIsRegistered(false);
-        auto& rasterizer = Core::System::GetInstance().Renderer().Rasterizer();
         rasterizer.UpdatePagesCachedCount(object->GetAddr(), object->GetSizeInBytes(), -1);
-
         // Only flush if use_accurate_gpu_emulation is enabled, as it incurs a performance hit
         if (Settings::values.use_accurate_gpu_emulation) {
             FlushObject(object);
         }
 
-        object_cache.subtract({GetInterval(object), ObjectSet{object}});
+        interval_cache.subtract({GetInterval(object), ObjectSet{object}});
+        map_cache.erase(object->GetAddr());
     }
 
     /// Returns a ticks counter used for tracking when cached objects were last modified
@@ -141,7 +138,7 @@ private:
 
         std::vector<T> objects;
         const ObjectInterval interval{addr, addr + size};
-        for (auto& pair : boost::make_iterator_range(object_cache.equal_range(interval))) {
+        for (auto& pair : boost::make_iterator_range(interval_cache.equal_range(interval))) {
             for (auto& cached_object : pair.second) {
                 if (!cached_object) {
                     continue;
@@ -167,14 +164,17 @@ private:
     }
 
     using ObjectSet = std::set<T>;
-    using ObjectCache = boost::icl::interval_map<VAddr, ObjectSet>;
-    using ObjectInterval = typename ObjectCache::interval_type;
+    using ObjectCache = std::unordered_map<VAddr, T>;
+    using IntervalCache = boost::icl::interval_map<VAddr, ObjectSet>;
+    using ObjectInterval = typename IntervalCache::interval_type;
 
     static auto GetInterval(const T& object) {
         return ObjectInterval::right_open(object->GetAddr(),
                                           object->GetAddr() + object->GetSizeInBytes());
     }
 
-    ObjectCache object_cache; ///< Cache of objects
-    u64 modified_ticks{};     ///< Counter of cache state ticks, used for in-order flushing
+    ObjectCache map_cache;
+    IntervalCache interval_cache; ///< Cache of objects
+    u64 modified_ticks{};         ///< Counter of cache state ticks, used for in-order flushing
+    VideoCore::RasterizerInterface& rasterizer;
 };

src/video_core/renderer_base.cpp

@@ -5,7 +5,6 @@
 #include "core/frontend/emu_window.h"
 #include "core/settings.h"
 #include "video_core/renderer_base.h"
-#include "video_core/renderer_opengl/gl_rasterizer.h"
 
 namespace VideoCore {
 

src/video_core/renderer_opengl/gl_buffer_cache.cpp

@@ -9,10 +9,12 @@
 #include "core/core.h"
 #include "core/memory.h"
 #include "video_core/renderer_opengl/gl_buffer_cache.h"
+#include "video_core/renderer_opengl/gl_rasterizer.h"
 
 namespace OpenGL {
 
-OGLBufferCache::OGLBufferCache(std::size_t size) : stream_buffer(GL_ARRAY_BUFFER, size) {}
+OGLBufferCache::OGLBufferCache(RasterizerOpenGL& rasterizer, std::size_t size)
+    : RasterizerCache{rasterizer}, stream_buffer(GL_ARRAY_BUFFER, size) {}
 
 GLintptr OGLBufferCache::UploadMemory(Tegra::GPUVAddr gpu_addr, std::size_t size,
                                       std::size_t alignment, bool cache) {

src/video_core/renderer_opengl/gl_buffer_cache.h

@@ -15,6 +15,8 @@
 
 namespace OpenGL {
 
+class RasterizerOpenGL;
+
 struct CachedBufferEntry final : public RasterizerCacheObject {
     VAddr GetAddr() const override {
         return addr;
@@ -35,7 +37,7 @@ struct CachedBufferEntry final : public RasterizerCacheObject {
 
 class OGLBufferCache final : public RasterizerCache<std::shared_ptr<CachedBufferEntry>> {
 public:
-    explicit OGLBufferCache(std::size_t size);
+    explicit OGLBufferCache(RasterizerOpenGL& rasterizer, std::size_t size);
 
     /// Uploads data from a guest GPU address. Returns host's buffer offset where it's been
     /// allocated.

src/video_core/renderer_opengl/gl_primitive_assembler.cpp

@@ -6,6 +6,7 @@
 #include <array>
 #include "common/assert.h"
 #include "common/common_types.h"
+#include "core/core.h"
 #include "core/memory.h"
 #include "video_core/renderer_opengl/gl_buffer_cache.h"
 #include "video_core/renderer_opengl/gl_primitive_assembler.h"

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -33,7 +33,8 @@ using Maxwell = Tegra::Engines::Maxwell3D::Regs;
 using PixelFormat = VideoCore::Surface::PixelFormat;
 using SurfaceType = VideoCore::Surface::SurfaceType;
 
-MICROPROFILE_DEFINE(OpenGL_VAO, "OpenGL", "Vertex Array Setup", MP_RGB(128, 128, 192));
+MICROPROFILE_DEFINE(OpenGL_VAO, "OpenGL", "Vertex Format Setup", MP_RGB(128, 128, 192));
+MICROPROFILE_DEFINE(OpenGL_VB, "OpenGL", "Vertex Buffer Setup", MP_RGB(128, 128, 192));
 MICROPROFILE_DEFINE(OpenGL_Shader, "OpenGL", "Shader Setup", MP_RGB(128, 128, 192));
 MICROPROFILE_DEFINE(OpenGL_UBO, "OpenGL", "Const Buffer Setup", MP_RGB(128, 128, 192));
 MICROPROFILE_DEFINE(OpenGL_Index, "OpenGL", "Index Buffer Setup", MP_RGB(128, 128, 192));
@@ -79,7 +80,8 @@ struct DrawParameters {
 };
 
 RasterizerOpenGL::RasterizerOpenGL(Core::Frontend::EmuWindow& window, ScreenInfo& info)
-    : emu_window{window}, screen_info{info}, buffer_cache(STREAM_BUFFER_SIZE) {
+    : res_cache{*this}, shader_cache{*this}, emu_window{window}, screen_info{info},
+      buffer_cache(*this, STREAM_BUFFER_SIZE) {
     // Create sampler objects
     for (std::size_t i = 0; i < texture_samplers.size(); ++i) {
         texture_samplers[i].Create();
@@ -105,8 +107,6 @@ RasterizerOpenGL::RasterizerOpenGL(Core::Frontend::EmuWindow& window, ScreenInfo
 
     ASSERT_MSG(has_ARB_separate_shader_objects, "has_ARB_separate_shader_objects is unsupported");
     OpenGLState::ApplyDefaultState();
-    // Clipping plane 0 is always enabled for PICA fixed clip plane z <= 0
-    state.clip_distance[0] = true;
 
     // Create render framebuffer
     framebuffer.Create();
@@ -122,18 +122,23 @@ RasterizerOpenGL::RasterizerOpenGL(Core::Frontend::EmuWindow& window, ScreenInfo
 
 RasterizerOpenGL::~RasterizerOpenGL() {}
 
-void RasterizerOpenGL::SetupVertexArrays() {
-    MICROPROFILE_SCOPE(OpenGL_VAO);
-    const auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
+void RasterizerOpenGL::SetupVertexFormat() {
+    auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
     const auto& regs = gpu.regs;
 
+    if (!gpu.dirty_flags.vertex_attrib_format)
+        return;
+    gpu.dirty_flags.vertex_attrib_format = false;
+
+    MICROPROFILE_SCOPE(OpenGL_VAO);
+
     auto [iter, is_cache_miss] = vertex_array_cache.try_emplace(regs.vertex_attrib_format);
     auto& VAO = iter->second;
 
     if (is_cache_miss) {
         VAO.Create();
         state.draw.vertex_array = VAO.handle;
-        state.Apply();
+        state.ApplyVertexBufferState();
 
         // The index buffer binding is stored within the VAO. Stupid OpenGL, but easy to work
         // around.
@@ -175,8 +180,13 @@ void RasterizerOpenGL::SetupVertexArrays() {
         }
     }
     state.draw.vertex_array = VAO.handle;
-    state.draw.vertex_buffer = buffer_cache.GetHandle();
-    state.Apply();
+    state.ApplyVertexBufferState();
+}
+
+void RasterizerOpenGL::SetupVertexBuffer() {
+    MICROPROFILE_SCOPE(OpenGL_VB);
+    const auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
+    const auto& regs = gpu.regs;
 
     // Upload all guest vertex arrays sequentially to our buffer
     for (u32 index = 0; index < Maxwell::NumVertexArrays; ++index) {
@@ -203,6 +213,9 @@ void RasterizerOpenGL::SetupVertexArrays() {
             glVertexBindingDivisor(index, 0);
         }
     }
+
+    // Implicit set by glBindVertexBuffer. Stupid glstate handling...
+    state.draw.vertex_buffer = buffer_cache.GetHandle();
 }
 
 DrawParameters RasterizerOpenGL::SetupDraw() {
@@ -327,8 +340,6 @@ void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) {
             index++;
         }
     }
-
-    state.Apply();
 }
 
 std::size_t RasterizerOpenGL::CalculateVertexArraysSize() const {
@@ -397,8 +408,8 @@ void RasterizerOpenGL::UpdatePagesCachedCount(VAddr addr, u64 size, int delta) {
         cached_pages.add({pages_interval, delta});
 }
 
-void RasterizerOpenGL::ConfigureFramebuffers(bool using_color_fb, bool using_depth_fb,
-                                             bool preserve_contents,
+void RasterizerOpenGL::ConfigureFramebuffers(OpenGLState& current_state, bool using_color_fb,
+                                             bool using_depth_fb, bool preserve_contents,
                                              std::optional<std::size_t> single_color_target) {
     MICROPROFILE_SCOPE(OpenGL_Framebuffer);
     const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
@@ -414,9 +425,9 @@ void RasterizerOpenGL::ConfigureFramebuffers(bool using_color_fb, bool using_dep
     ASSERT_MSG(regs.rt_separate_frag_data == 0, "Unimplemented");
 
     // Bind the framebuffer surfaces
-    state.draw.draw_framebuffer = framebuffer.handle;
-    state.Apply();
-    state.framebuffer_srgb.enabled = regs.framebuffer_srgb != 0;
+    current_state.draw.draw_framebuffer = framebuffer.handle;
+    current_state.ApplyFramebufferState();
+    current_state.framebuffer_srgb.enabled = regs.framebuffer_srgb != 0;
 
     if (using_color_fb) {
         if (single_color_target) {
@@ -494,10 +505,7 @@ void RasterizerOpenGL::ConfigureFramebuffers(bool using_color_fb, bool using_dep
         glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0,
                                0);
     }
-
-    SyncViewport();
-
-    state.Apply();
+    SyncViewport(current_state);
 }
 
 void RasterizerOpenGL::Clear() {
@@ -510,22 +518,23 @@ void RasterizerOpenGL::Clear() {
     bool use_stencil{};
 
     OpenGLState clear_state;
-    clear_state.draw.draw_framebuffer = framebuffer.handle;
-    clear_state.color_mask.red_enabled = regs.clear_buffers.R ? GL_TRUE : GL_FALSE;
-    clear_state.color_mask.green_enabled = regs.clear_buffers.G ? GL_TRUE : GL_FALSE;
-    clear_state.color_mask.blue_enabled = regs.clear_buffers.B ? GL_TRUE : GL_FALSE;
-    clear_state.color_mask.alpha_enabled = regs.clear_buffers.A ? GL_TRUE : GL_FALSE;
-
     if (regs.clear_buffers.R || regs.clear_buffers.G || regs.clear_buffers.B ||
         regs.clear_buffers.A) {
         use_color = true;
     }
+    if (use_color) {
+        clear_state.color_mask[0].red_enabled = regs.clear_buffers.R ? GL_TRUE : GL_FALSE;
+        clear_state.color_mask[0].green_enabled = regs.clear_buffers.G ? GL_TRUE : GL_FALSE;
+        clear_state.color_mask[0].blue_enabled = regs.clear_buffers.B ? GL_TRUE : GL_FALSE;
+        clear_state.color_mask[0].alpha_enabled = regs.clear_buffers.A ? GL_TRUE : GL_FALSE;
+    }
     if (regs.clear_buffers.Z) {
         ASSERT_MSG(regs.zeta_enable != 0, "Tried to clear Z but buffer is not enabled!");
         use_depth = 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.
+        // ignored when clearing the buffer in the Switch, but OpenGL obeys it so we set it to
+        // true.
         clear_state.depth.test_enabled = true;
         clear_state.depth.test_func = GL_ALWAYS;
     }
@@ -542,11 +551,8 @@ void RasterizerOpenGL::Clear() {
 
     ScopeAcquireGLContext acquire_context{emu_window};
 
-    ConfigureFramebuffers(use_color, use_depth || use_stencil, false,
+    ConfigureFramebuffers(clear_state, use_color, use_depth || use_stencil, false,
                           regs.clear_buffers.RT.Value());
-    // Copy the sRGB setting to the clear state to avoid problem with
-    // specific driver implementations
-    clear_state.framebuffer_srgb.enabled = state.framebuffer_srgb.enabled;
     clear_state.Apply();
 
     if (use_color) {
@@ -572,15 +578,16 @@ void RasterizerOpenGL::DrawArrays() {
 
     ScopeAcquireGLContext acquire_context{emu_window};
 
-    ConfigureFramebuffers();
-
+    ConfigureFramebuffers(state);
+    SyncColorMask();
+    SyncFragmentColorClampState();
+    SyncMultiSampleState();
     SyncDepthTestState();
     SyncStencilTestState();
     SyncBlendState();
     SyncLogicOpState();
     SyncCullMode();
     SyncPrimitiveRestart();
-    SyncDepthRange();
     SyncScissorTest();
     // Alpha Testing is synced on shaders.
     SyncTransformFeedback();
@@ -594,7 +601,7 @@ void RasterizerOpenGL::DrawArrays() {
     const bool is_indexed = accelerate_draw == AccelDraw::Indexed;
 
     state.draw.vertex_buffer = buffer_cache.GetHandle();
-    state.Apply();
+    state.ApplyVertexBufferState();
 
     std::size_t buffer_size = CalculateVertexArraysSize();
 
@@ -621,7 +628,8 @@ void RasterizerOpenGL::DrawArrays() {
 
     buffer_cache.Map(buffer_size);
 
-    SetupVertexArrays();
+    SetupVertexFormat();
+    SetupVertexBuffer();
     DrawParameters params = SetupDraw();
     SetupShaders(params.primitive_mode);
 
@@ -634,7 +642,7 @@ void RasterizerOpenGL::DrawArrays() {
     params.DispatchDraw();
 
     // Disable scissor test
-    state.scissor.enabled = false;
+    state.viewports[0].scissor.enabled = false;
 
     accelerate_draw = AccelDraw::Disabled;
 
@@ -727,7 +735,6 @@ void RasterizerOpenGL::SamplerInfo::Create() {
 
 void RasterizerOpenGL::SamplerInfo::SyncWithConfig(const Tegra::Texture::TSCEntry& config) {
     const GLuint s = sampler.handle;
-
     if (mag_filter != config.mag_filter) {
         mag_filter = config.mag_filter;
         glSamplerParameteri(
@@ -769,15 +776,51 @@ void RasterizerOpenGL::SamplerInfo::SyncWithConfig(const Tegra::Texture::TSCEntr
                             MaxwellToGL::DepthCompareFunc(depth_compare_func));
     }
 
-    if (wrap_u == Tegra::Texture::WrapMode::Border || wrap_v == Tegra::Texture::WrapMode::Border ||
-        wrap_p == Tegra::Texture::WrapMode::Border) {
-        const GLvec4 new_border_color = {{config.border_color_r, config.border_color_g,
-                                          config.border_color_b, config.border_color_a}};
-        if (border_color != new_border_color) {
-            border_color = new_border_color;
-            glSamplerParameterfv(s, GL_TEXTURE_BORDER_COLOR, border_color.data());
+    GLvec4 new_border_color;
+    if (config.srgb_conversion) {
+        new_border_color[0] = config.srgb_border_color_r / 255.0f;
+        new_border_color[1] = config.srgb_border_color_g / 255.0f;
+        new_border_color[2] = config.srgb_border_color_g / 255.0f;
+    } else {
+        new_border_color[0] = config.border_color_r;
+        new_border_color[1] = config.border_color_g;
+        new_border_color[2] = config.border_color_b;
+    }
+    new_border_color[3] = config.border_color_a;
+
+    if (border_color != new_border_color) {
+        border_color = new_border_color;
+        glSamplerParameterfv(s, GL_TEXTURE_BORDER_COLOR, border_color.data());
+    }
+
+    const float anisotropic_max = static_cast<float>(1 << config.max_anisotropy.Value());
+    if (anisotropic_max != max_anisotropic) {
+        max_anisotropic = anisotropic_max;
+        if (GLAD_GL_ARB_texture_filter_anisotropic) {
+            glSamplerParameterf(s, GL_TEXTURE_MAX_ANISOTROPY, max_anisotropic);
+        } else if (GLAD_GL_EXT_texture_filter_anisotropic) {
+            glSamplerParameterf(s, GL_TEXTURE_MAX_ANISOTROPY_EXT, max_anisotropic);
         }
     }
+    const float lod_min = static_cast<float>(config.min_lod_clamp.Value()) / 256.0f;
+    if (lod_min != min_lod) {
+        min_lod = lod_min;
+        glSamplerParameterf(s, GL_TEXTURE_MIN_LOD, min_lod);
+    }
+
+    const float lod_max = static_cast<float>(config.max_lod_clamp.Value()) / 256.0f;
+    if (lod_max != max_lod) {
+        max_lod = lod_max;
+        glSamplerParameterf(s, GL_TEXTURE_MAX_LOD, max_lod);
+    }
+    const u32 bias = config.mip_lod_bias.Value();
+    // Sign extend the 13-bit value.
+    const u32 mask = 1U << (13 - 1);
+    const float bias_lod = static_cast<s32>((bias ^ mask) - mask) / 256.f;
+    if (lod_bias != bias_lod) {
+        lod_bias = bias_lod;
+        glSamplerParameterf(s, GL_TEXTURE_LOD_BIAS, lod_bias);
+    }
 }
 
 u32 RasterizerOpenGL::SetupConstBuffers(Maxwell::ShaderStage stage, Shader& shader,
@@ -897,14 +940,18 @@ u32 RasterizerOpenGL::SetupTextures(Maxwell::ShaderStage stage, Shader& shader,
     return current_unit + static_cast<u32>(entries.size());
 }
 
-void RasterizerOpenGL::SyncViewport() {
+void RasterizerOpenGL::SyncViewport(OpenGLState& current_state) {
     const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
-    const MathUtil::Rectangle<s32> viewport_rect{regs.viewport_transform[0].GetRect()};
-
-    state.viewport.x = viewport_rect.left;
-    state.viewport.y = viewport_rect.bottom;
-    state.viewport.width = static_cast<GLsizei>(viewport_rect.GetWidth());
-    state.viewport.height = static_cast<GLsizei>(viewport_rect.GetHeight());
+    for (std::size_t i = 0; i < Tegra::Engines::Maxwell3D::Regs::NumViewports; i++) {
+        const MathUtil::Rectangle<s32> viewport_rect{regs.viewport_transform[i].GetRect()};
+        auto& viewport = current_state.viewports[i];
+        viewport.x = viewport_rect.left;
+        viewport.y = viewport_rect.bottom;
+        viewport.width = static_cast<GLfloat>(viewport_rect.GetWidth());
+        viewport.height = static_cast<GLfloat>(viewport_rect.GetHeight());
+        viewport.depth_range_far = regs.viewports[i].depth_range_far;
+        viewport.depth_range_near = regs.viewports[i].depth_range_near;
+    }
 }
 
 void RasterizerOpenGL::SyncClipEnabled() {
@@ -946,13 +993,6 @@ void RasterizerOpenGL::SyncPrimitiveRestart() {
     state.primitive_restart.index = regs.primitive_restart.index;
 }
 
-void RasterizerOpenGL::SyncDepthRange() {
-    const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
-
-    state.depth.depth_range_near = regs.viewport->depth_range_near;
-    state.depth.depth_range_far = regs.viewport->depth_range_far;
-}
-
 void RasterizerOpenGL::SyncDepthTestState() {
     const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
 
@@ -973,9 +1013,6 @@ void RasterizerOpenGL::SyncStencilTestState() {
         return;
     }
 
-    // TODO(bunnei): Verify behavior when this is not set
-    ASSERT(regs.stencil_two_side_enable);
-
     state.stencil.front.test_func = MaxwellToGL::ComparisonOp(regs.stencil_front_func_func);
     state.stencil.front.test_ref = regs.stencil_front_func_ref;
     state.stencil.front.test_mask = regs.stencil_front_func_mask;
@@ -983,42 +1020,95 @@ void RasterizerOpenGL::SyncStencilTestState() {
     state.stencil.front.action_depth_fail = MaxwellToGL::StencilOp(regs.stencil_front_op_zfail);
     state.stencil.front.action_depth_pass = MaxwellToGL::StencilOp(regs.stencil_front_op_zpass);
     state.stencil.front.write_mask = regs.stencil_front_mask;
+    if (regs.stencil_two_side_enable) {
+        state.stencil.back.test_func = MaxwellToGL::ComparisonOp(regs.stencil_back_func_func);
+        state.stencil.back.test_ref = regs.stencil_back_func_ref;
+        state.stencil.back.test_mask = regs.stencil_back_func_mask;
+        state.stencil.back.action_stencil_fail = MaxwellToGL::StencilOp(regs.stencil_back_op_fail);
+        state.stencil.back.action_depth_fail = MaxwellToGL::StencilOp(regs.stencil_back_op_zfail);
+        state.stencil.back.action_depth_pass = MaxwellToGL::StencilOp(regs.stencil_back_op_zpass);
+        state.stencil.back.write_mask = regs.stencil_back_mask;
+    } else {
+        state.stencil.back.test_func = GL_ALWAYS;
+        state.stencil.back.test_ref = 0;
+        state.stencil.back.test_mask = 0xFFFFFFFF;
+        state.stencil.back.write_mask = 0xFFFFFFFF;
+        state.stencil.back.action_stencil_fail = GL_KEEP;
+        state.stencil.back.action_depth_fail = GL_KEEP;
+        state.stencil.back.action_depth_pass = GL_KEEP;
+    }
+}
 
-    state.stencil.back.test_func = MaxwellToGL::ComparisonOp(regs.stencil_back_func_func);
-    state.stencil.back.test_ref = regs.stencil_back_func_ref;
-    state.stencil.back.test_mask = regs.stencil_back_func_mask;
-    state.stencil.back.action_stencil_fail = MaxwellToGL::StencilOp(regs.stencil_back_op_fail);
-    state.stencil.back.action_depth_fail = MaxwellToGL::StencilOp(regs.stencil_back_op_zfail);
-    state.stencil.back.action_depth_pass = MaxwellToGL::StencilOp(regs.stencil_back_op_zpass);
-    state.stencil.back.write_mask = regs.stencil_back_mask;
+void RasterizerOpenGL::SyncColorMask() {
+    const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+    const std::size_t count =
+        regs.independent_blend_enable ? Tegra::Engines::Maxwell3D::Regs::NumRenderTargets : 1;
+    for (std::size_t i = 0; i < count; i++) {
+        const auto& source = regs.color_mask[regs.color_mask_common ? 0 : i];
+        auto& dest = state.color_mask[i];
+        dest.red_enabled = (source.R == 0) ? GL_FALSE : GL_TRUE;
+        dest.green_enabled = (source.G == 0) ? GL_FALSE : GL_TRUE;
+        dest.blue_enabled = (source.B == 0) ? GL_FALSE : GL_TRUE;
+        dest.alpha_enabled = (source.A == 0) ? GL_FALSE : GL_TRUE;
+    }
+}
+
+void RasterizerOpenGL::SyncMultiSampleState() {
+    const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+    state.multisample_control.alpha_to_coverage = regs.multisample_control.alpha_to_coverage != 0;
+    state.multisample_control.alpha_to_one = regs.multisample_control.alpha_to_one != 0;
+}
+
+void RasterizerOpenGL::SyncFragmentColorClampState() {
+    const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+    state.fragment_color_clamp.enabled = regs.frag_color_clamp != 0;
 }
 
 void RasterizerOpenGL::SyncBlendState() {
     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;
+    state.blend_color.red = regs.blend_color.r;
+    state.blend_color.green = regs.blend_color.g;
+    state.blend_color.blue = regs.blend_color.b;
+    state.blend_color.alpha = regs.blend_color.a;
 
-    if (!state.blend.enabled)
+    state.independant_blend.enabled = regs.independent_blend_enable;
+    if (!state.independant_blend.enabled) {
+        auto& blend = state.blend[0];
+        const auto& src = regs.blend;
+        blend.enabled = src.enable[0] != 0;
+        if (blend.enabled) {
+            blend.rgb_equation = MaxwellToGL::BlendEquation(src.equation_rgb);
+            blend.src_rgb_func = MaxwellToGL::BlendFunc(src.factor_source_rgb);
+            blend.dst_rgb_func = MaxwellToGL::BlendFunc(src.factor_dest_rgb);
+            blend.a_equation = MaxwellToGL::BlendEquation(src.equation_a);
+            blend.src_a_func = MaxwellToGL::BlendFunc(src.factor_source_a);
+            blend.dst_a_func = MaxwellToGL::BlendFunc(src.factor_dest_a);
+        }
+        for (std::size_t i = 1; i < Tegra::Engines::Maxwell3D::Regs::NumRenderTargets; i++) {
+            state.blend[i].enabled = false;
+        }
         return;
+    }
 
-    ASSERT_MSG(regs.logic_op.enable == 0,
-               "Blending and logic op can't be enabled at the same time.");
-
-    ASSERT_MSG(regs.independent_blend_enable == 1, "Only independent blending is implemented");
-    ASSERT_MSG(!regs.independent_blend[0].separate_alpha, "Unimplemented");
-    state.blend.rgb_equation = MaxwellToGL::BlendEquation(regs.independent_blend[0].equation_rgb);
-    state.blend.src_rgb_func = MaxwellToGL::BlendFunc(regs.independent_blend[0].factor_source_rgb);
-    state.blend.dst_rgb_func = MaxwellToGL::BlendFunc(regs.independent_blend[0].factor_dest_rgb);
-    state.blend.a_equation = MaxwellToGL::BlendEquation(regs.independent_blend[0].equation_a);
-    state.blend.src_a_func = MaxwellToGL::BlendFunc(regs.independent_blend[0].factor_source_a);
-    state.blend.dst_a_func = MaxwellToGL::BlendFunc(regs.independent_blend[0].factor_dest_a);
+    for (std::size_t i = 0; i < Tegra::Engines::Maxwell3D::Regs::NumRenderTargets; i++) {
+        auto& blend = state.blend[i];
+        const auto& src = regs.independent_blend[i];
+        blend.enabled = regs.blend.enable[i] != 0;
+        if (!blend.enabled)
+            continue;
+        blend.rgb_equation = MaxwellToGL::BlendEquation(src.equation_rgb);
+        blend.src_rgb_func = MaxwellToGL::BlendFunc(src.factor_source_rgb);
+        blend.dst_rgb_func = MaxwellToGL::BlendFunc(src.factor_dest_rgb);
+        blend.a_equation = MaxwellToGL::BlendEquation(src.equation_a);
+        blend.src_a_func = MaxwellToGL::BlendFunc(src.factor_source_a);
+        blend.dst_a_func = MaxwellToGL::BlendFunc(src.factor_dest_a);
+    }
 }
 
 void RasterizerOpenGL::SyncLogicOpState() {
     const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
 
-    // TODO(Subv): Support more than just render target 0.
     state.logic_op.enabled = regs.logic_op.enable != 0;
 
     if (!state.logic_op.enabled)
@@ -1032,17 +1122,19 @@ void RasterizerOpenGL::SyncLogicOpState() {
 
 void RasterizerOpenGL::SyncScissorTest() {
     const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
-
-    state.scissor.enabled = (regs.scissor_test.enable != 0);
-    // TODO(Blinkhawk): Figure if the hardware supports scissor testing per viewport and how it's
-    // implemented.
-    if (regs.scissor_test.enable != 0) {
-        const u32 width = regs.scissor_test.max_x - regs.scissor_test.min_x;
-        const u32 height = regs.scissor_test.max_y - regs.scissor_test.min_y;
-        state.scissor.x = regs.scissor_test.min_x;
-        state.scissor.y = regs.scissor_test.min_y;
-        state.scissor.width = width;
-        state.scissor.height = height;
+    for (std::size_t i = 0; i < Tegra::Engines::Maxwell3D::Regs::NumViewports; i++) {
+        const auto& src = regs.scissor_test[i];
+        auto& dst = state.viewports[i].scissor;
+        dst.enabled = (src.enable != 0);
+        if (dst.enabled == 0) {
+            return;
+        }
+        const u32 width = src.max_x - src.min_x;
+        const u32 height = src.max_y - src.min_y;
+        dst.x = src.min_x;
+        dst.y = src.min_y;
+        dst.width = width;
+        dst.height = height;
     }
 }
 
@@ -1057,20 +1149,15 @@ void RasterizerOpenGL::SyncTransformFeedback() {
 
 void RasterizerOpenGL::SyncPointState() {
     const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
-
-    // TODO(Rodrigo): Most games do not set a point size. I think this is a case of a
-    // register carrying a default value. For now, if the point size is zero, assume it's
-    // OpenGL's default (1).
-    state.point.size = regs.point_size == 0 ? 1 : regs.point_size;
+    state.point.size = regs.point_size;
 }
 
 void RasterizerOpenGL::CheckAlphaTests() {
     const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
 
     if (regs.alpha_test_enabled != 0 && regs.rt_control.count > 1) {
-        LOG_CRITICAL(
-            Render_OpenGL,
-            "Alpha Testing is enabled with Multiple Render Targets, this behavior is undefined.");
+        LOG_CRITICAL(Render_OpenGL, "Alpha Testing is enabled with Multiple Render Targets, "
+                                    "this behavior is undefined.");
         UNREACHABLE();
     }
 }

src/video_core/renderer_opengl/gl_rasterizer.h

@@ -88,7 +88,7 @@ private:
         /// SamplerInfo struct.
         void Create();
         /// Syncs the sampler object with the config, updating any necessary state.
-        void SyncWithConfig(const Tegra::Texture::TSCEntry& config);
+        void SyncWithConfig(const Tegra::Texture::TSCEntry& info);
 
     private:
         Tegra::Texture::TextureFilter mag_filter;
@@ -100,6 +100,10 @@ private:
         bool uses_depth_compare;
         Tegra::Texture::DepthCompareFunc depth_compare_func;
         GLvec4 border_color;
+        float min_lod;
+        float max_lod;
+        float lod_bias;
+        float max_anisotropic;
     };
 
     /**
@@ -109,8 +113,8 @@ private:
      * @param preserve_contents If true, tries to preserve data from a previously used framebuffer.
      * @param single_color_target Specifies if a single color buffer target should be used.
      */
-    void ConfigureFramebuffers(bool use_color_fb = true, bool using_depth_fb = true,
-                               bool preserve_contents = true,
+    void ConfigureFramebuffers(OpenGLState& current_state, bool use_color_fb = true,
+                               bool using_depth_fb = true, bool preserve_contents = true,
                                std::optional<std::size_t> single_color_target = {});
 
     /*
@@ -133,8 +137,8 @@ private:
     u32 SetupTextures(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage, Shader& shader,
                       GLenum primitive_mode, u32 current_unit);
 
-    /// Syncs the viewport to match the guest state
-    void SyncViewport();
+    /// Syncs the viewport and depth range to match the guest state
+    void SyncViewport(OpenGLState& current_state);
 
     /// Syncs the clip enabled status to match the guest state
     void SyncClipEnabled();
@@ -148,9 +152,6 @@ private:
     /// Syncs the primitve restart to match the guest state
     void SyncPrimitiveRestart();
 
-    /// Syncs the depth range to match the guest state
-    void SyncDepthRange();
-
     /// Syncs the depth test state to match the guest state
     void SyncDepthTestState();
 
@@ -163,6 +164,12 @@ private:
     /// Syncs the LogicOp state to match the guest state
     void SyncLogicOpState();
 
+    /// Syncs the the color clamp state
+    void SyncFragmentColorClampState();
+
+    /// Syncs the alpha coverage and alpha to one
+    void SyncMultiSampleState();
+
     /// Syncs the scissor test state to match the guest state
     void SyncScissorTest();
 
@@ -172,6 +179,9 @@ private:
     /// Syncs the point state to match the guest state
     void SyncPointState();
 
+    /// Syncs Color Mask
+    void SyncColorMask();
+
     /// Check asserts for alpha testing.
     void CheckAlphaTests();
 
@@ -207,7 +217,8 @@ private:
 
     std::size_t CalculateIndexBufferSize() const;
 
-    void SetupVertexArrays();
+    void SetupVertexFormat();
+    void SetupVertexBuffer();
 
     DrawParameters SetupDraw();
 

src/video_core/renderer_opengl/gl_rasterizer_cache.cpp

@@ -15,7 +15,9 @@
 #include "core/memory.h"
 #include "core/settings.h"
 #include "video_core/engines/maxwell_3d.h"
+#include "video_core/renderer_opengl/gl_rasterizer.h"
 #include "video_core/renderer_opengl/gl_rasterizer_cache.h"
+#include "video_core/renderer_opengl/gl_state.h"
 #include "video_core/renderer_opengl/utils.h"
 #include "video_core/surface.h"
 #include "video_core/textures/astc.h"
@@ -312,6 +314,8 @@ static constexpr std::array<FormatTuple, VideoCore::Surface::MaxPixelFormat> tex
     {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_5X4_SRGB
     {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false},        // ASTC_2D_5X5
     {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_5X5_SRGB
+    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false},        // ASTC_2D_10X8
+    {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_10X8_SRGB
 
     // Depth formats
     {GL_DEPTH_COMPONENT32F, GL_DEPTH_COMPONENT, GL_FLOAT, ComponentType::Float, false}, // Z32F
@@ -377,11 +381,8 @@ void MortonCopy(u32 stride, u32 block_height, u32 height, u32 block_depth, u32 d
     const u32 tile_size_y{GetDefaultBlockHeight(format)};
 
     if (morton_to_gl) {
-        const std::vector<u8> data =
-            Tegra::Texture::UnswizzleTexture(addr, tile_size_x, tile_size_y, bytes_per_pixel,
-                                             stride, height, depth, block_height, block_depth);
-        const std::size_t size_to_copy{std::min(gl_buffer_size, data.size())};
-        memcpy(gl_buffer, data.data(), size_to_copy);
+        Tegra::Texture::UnswizzleTexture(gl_buffer, addr, tile_size_x, tile_size_y, bytes_per_pixel,
+                                         stride, height, depth, block_height, block_depth);
     } else {
         Tegra::Texture::CopySwizzledData((stride + tile_size_x - 1) / tile_size_x,
                                          (height + tile_size_y - 1) / tile_size_y, depth,
@@ -454,6 +455,8 @@ static constexpr GLConversionArray morton_to_gl_fns = {
         MortonCopy<true, PixelFormat::ASTC_2D_5X4_SRGB>,
         MortonCopy<true, PixelFormat::ASTC_2D_5X5>,
         MortonCopy<true, PixelFormat::ASTC_2D_5X5_SRGB>,
+        MortonCopy<true, PixelFormat::ASTC_2D_10X8>,
+        MortonCopy<true, PixelFormat::ASTC_2D_10X8_SRGB>,
         MortonCopy<true, PixelFormat::Z32F>,
         MortonCopy<true, PixelFormat::Z16>,
         MortonCopy<true, PixelFormat::Z24S8>,
@@ -524,6 +527,8 @@ static constexpr GLConversionArray gl_to_morton_fns = {
         nullptr,
         nullptr,
         nullptr,
+        nullptr,
+        nullptr,
         MortonCopy<false, PixelFormat::Z32F>,
         MortonCopy<false, PixelFormat::Z16>,
         MortonCopy<false, PixelFormat::Z24S8>,
@@ -542,8 +547,8 @@ void SwizzleFunc(const GLConversionArray& functions, const SurfaceParams& params
     if (params.is_layered) {
         u64 offset = params.GetMipmapLevelOffset(mip_level);
         u64 offset_gl = 0;
-        u64 layer_size = params.LayerMemorySize();
-        u64 gl_size = params.LayerSizeGL(mip_level);
+        const u64 layer_size = params.LayerMemorySize();
+        const u64 gl_size = params.LayerSizeGL(mip_level);
         for (u32 i = 0; i < params.depth; i++) {
             functions[static_cast<std::size_t>(params.pixel_format)](
                 params.MipWidth(mip_level), params.MipBlockHeight(mip_level),
@@ -553,7 +558,7 @@ void SwizzleFunc(const GLConversionArray& functions, const SurfaceParams& params
             offset_gl += gl_size;
         }
     } else {
-        u64 offset = params.GetMipmapLevelOffset(mip_level);
+        const u64 offset = params.GetMipmapLevelOffset(mip_level);
         functions[static_cast<std::size_t>(params.pixel_format)](
             params.MipWidth(mip_level), params.MipBlockHeight(mip_level),
             params.MipHeight(mip_level), params.MipBlockDepth(mip_level), depth, gl_buffer.data(),
@@ -561,9 +566,11 @@ void SwizzleFunc(const GLConversionArray& functions, const SurfaceParams& params
     }
 }
 
+MICROPROFILE_DEFINE(OpenGL_BlitSurface, "OpenGL", "BlitSurface", MP_RGB(128, 192, 64));
 static bool BlitSurface(const Surface& src_surface, const Surface& dst_surface,
                         GLuint read_fb_handle, GLuint draw_fb_handle, GLenum src_attachment = 0,
                         GLenum dst_attachment = 0, std::size_t cubemap_face = 0) {
+    MICROPROFILE_SCOPE(OpenGL_BlitSurface);
 
     const auto& src_params{src_surface->GetSurfaceParams()};
     const auto& dst_params{dst_surface->GetSurfaceParams()};
@@ -576,7 +583,7 @@ static bool BlitSurface(const Surface& src_surface, const Surface& dst_surface,
     state.draw.draw_framebuffer = draw_fb_handle;
     // Set sRGB enabled if the destination surfaces need it
     state.framebuffer_srgb.enabled = dst_params.srgb_conversion;
-    state.Apply();
+    state.ApplyFramebufferState();
 
     u32 buffers{};
 
@@ -703,18 +710,20 @@ static void FastCopySurface(const Surface& src_surface, const Surface& dst_surfa
                        0, 0, width, height, 1);
 }
 
+MICROPROFILE_DEFINE(OpenGL_CopySurface, "OpenGL", "CopySurface", MP_RGB(128, 192, 64));
 static void CopySurface(const Surface& src_surface, const Surface& dst_surface,
-                        GLuint copy_pbo_handle, GLenum src_attachment = 0,
-                        GLenum dst_attachment = 0, std::size_t cubemap_face = 0) {
+                        const GLuint copy_pbo_handle, const GLenum src_attachment = 0,
+                        const GLenum dst_attachment = 0, const std::size_t cubemap_face = 0) {
+    MICROPROFILE_SCOPE(OpenGL_CopySurface);
     ASSERT_MSG(dst_attachment == 0, "Unimplemented");
 
     const auto& src_params{src_surface->GetSurfaceParams()};
     const auto& dst_params{dst_surface->GetSurfaceParams()};
 
-    auto source_format = GetFormatTuple(src_params.pixel_format, src_params.component_type);
-    auto dest_format = GetFormatTuple(dst_params.pixel_format, dst_params.component_type);
+    const auto source_format = GetFormatTuple(src_params.pixel_format, src_params.component_type);
+    const auto dest_format = GetFormatTuple(dst_params.pixel_format, dst_params.component_type);
 
-    std::size_t buffer_size = std::max(src_params.size_in_bytes, dst_params.size_in_bytes);
+    const std::size_t buffer_size = std::max(src_params.size_in_bytes, dst_params.size_in_bytes);
 
     glBindBuffer(GL_PIXEL_PACK_BUFFER, copy_pbo_handle);
     glBufferData(GL_PIXEL_PACK_BUFFER, buffer_size, nullptr, GL_STREAM_DRAW_ARB);
@@ -738,13 +747,10 @@ static void CopySurface(const Surface& src_surface, const Surface& dst_surface,
             LOG_DEBUG(HW_GPU, "Trying to upload extra texture data from the CPU during "
                               "reinterpretation but the texture is tiled.");
         }
-        std::size_t remaining_size = dst_params.size_in_bytes - src_params.size_in_bytes;
-        std::vector<u8> data(remaining_size);
-        std::memcpy(data.data(), Memory::GetPointer(dst_params.addr + src_params.size_in_bytes),
-                    data.size());
+        const std::size_t remaining_size = dst_params.size_in_bytes - src_params.size_in_bytes;
 
         glBufferSubData(GL_PIXEL_PACK_BUFFER, src_params.size_in_bytes, remaining_size,
-                        data.data());
+                        Memory::GetPointer(dst_params.addr + src_params.size_in_bytes));
     }
 
     glBindBuffer(GL_PIXEL_PACK_BUFFER, 0);
@@ -926,7 +932,9 @@ static void ConvertFormatAsNeeded_LoadGLBuffer(std::vector<u8>& data, PixelForma
     case PixelFormat::ASTC_2D_8X8_SRGB:
     case PixelFormat::ASTC_2D_8X5_SRGB:
     case PixelFormat::ASTC_2D_5X4_SRGB:
-    case PixelFormat::ASTC_2D_5X5_SRGB: {
+    case PixelFormat::ASTC_2D_5X5_SRGB:
+    case PixelFormat::ASTC_2D_10X8:
+    case PixelFormat::ASTC_2D_10X8_SRGB: {
         // Convert ASTC pixel formats to RGBA8, as most desktop GPUs do not support ASTC.
         u32 block_width{};
         u32 block_height{};
@@ -961,7 +969,11 @@ static void ConvertFormatAsNeeded_FlushGLBuffer(std::vector<u8>& data, PixelForm
     case PixelFormat::ASTC_2D_4X4:
     case PixelFormat::ASTC_2D_8X8:
     case PixelFormat::ASTC_2D_4X4_SRGB:
-    case PixelFormat::ASTC_2D_8X8_SRGB: {
+    case PixelFormat::ASTC_2D_8X8_SRGB:
+    case PixelFormat::ASTC_2D_5X5:
+    case PixelFormat::ASTC_2D_5X5_SRGB:
+    case PixelFormat::ASTC_2D_10X8:
+    case PixelFormat::ASTC_2D_10X8_SRGB: {
         LOG_CRITICAL(HW_GPU, "Conversion of format {} after texture flushing is not implemented",
                      static_cast<u32>(pixel_format));
         UNREACHABLE();
@@ -974,7 +986,7 @@ static void ConvertFormatAsNeeded_FlushGLBuffer(std::vector<u8>& data, PixelForm
     }
 }
 
-MICROPROFILE_DEFINE(OpenGL_SurfaceLoad, "OpenGL", "Surface Load", MP_RGB(128, 64, 192));
+MICROPROFILE_DEFINE(OpenGL_SurfaceLoad, "OpenGL", "Surface Load", MP_RGB(128, 192, 64));
 void CachedSurface::LoadGLBuffer() {
     MICROPROFILE_SCOPE(OpenGL_SurfaceLoad);
     gl_buffer.resize(params.max_mip_level);
@@ -1156,7 +1168,7 @@ void CachedSurface::UploadGLMipmapTexture(u32 mip_map, GLuint read_fb_handle,
     glPixelStorei(GL_UNPACK_ROW_LENGTH, 0);
 }
 
-MICROPROFILE_DEFINE(OpenGL_TextureUL, "OpenGL", "Texture Upload", MP_RGB(128, 64, 192));
+MICROPROFILE_DEFINE(OpenGL_TextureUL, "OpenGL", "Texture Upload", MP_RGB(128, 192, 64));
 void CachedSurface::UploadGLTexture(GLuint read_fb_handle, GLuint draw_fb_handle) {
     if (params.type == SurfaceType::Fill)
         return;
@@ -1167,7 +1179,8 @@ void CachedSurface::UploadGLTexture(GLuint read_fb_handle, GLuint draw_fb_handle
         UploadGLMipmapTexture(i, read_fb_handle, draw_fb_handle);
 }
 
-RasterizerCacheOpenGL::RasterizerCacheOpenGL() {
+RasterizerCacheOpenGL::RasterizerCacheOpenGL(RasterizerOpenGL& rasterizer)
+    : RasterizerCache{rasterizer} {
     read_framebuffer.Create();
     draw_framebuffer.Create();
     copy_pbo.Create();
@@ -1329,6 +1342,7 @@ Surface RasterizerCacheOpenGL::RecreateSurface(const Surface& old_surface,
         break;
     case SurfaceTarget::TextureCubemap:
     case SurfaceTarget::Texture3D:
+    case SurfaceTarget::Texture2DArray:
     case SurfaceTarget::TextureCubeArray:
         AccurateCopySurface(old_surface, new_surface);
         break;

src/video_core/renderer_opengl/gl_rasterizer_cache.h

@@ -264,6 +264,8 @@ struct hash<SurfaceReserveKey> {
 
 namespace OpenGL {
 
+class RasterizerOpenGL;
+
 class CachedSurface final : public RasterizerCacheObject {
 public:
     CachedSurface(const SurfaceParams& params);
@@ -311,7 +313,7 @@ private:
 
 class RasterizerCacheOpenGL final : public RasterizerCache<Surface> {
 public:
-    RasterizerCacheOpenGL();
+    explicit RasterizerCacheOpenGL(RasterizerOpenGL& rasterizer);
 
     /// Get a surface based on the texture configuration
     Surface GetTextureSurface(const Tegra::Texture::FullTextureInfo& config,

src/video_core/renderer_opengl/gl_resource_manager.cpp

@@ -0,0 +1,186 @@
+// Copyright 2015 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <utility>
+#include <glad/glad.h>
+#include "common/common_types.h"
+#include "common/microprofile.h"
+#include "video_core/renderer_opengl/gl_resource_manager.h"
+#include "video_core/renderer_opengl/gl_shader_util.h"
+#include "video_core/renderer_opengl/gl_state.h"
+
+MICROPROFILE_DEFINE(OpenGL_ResourceCreation, "OpenGL", "Resource Creation", MP_RGB(128, 128, 192));
+MICROPROFILE_DEFINE(OpenGL_ResourceDeletion, "OpenGL", "Resource Deletion", MP_RGB(128, 128, 192));
+
+namespace OpenGL {
+
+void OGLTexture::Create() {
+    if (handle != 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceCreation);
+    glGenTextures(1, &handle);
+}
+
+void OGLTexture::Release() {
+    if (handle == 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceDeletion);
+    glDeleteTextures(1, &handle);
+    OpenGLState::GetCurState().UnbindTexture(handle).Apply();
+    handle = 0;
+}
+
+void OGLSampler::Create() {
+    if (handle != 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceCreation);
+    glGenSamplers(1, &handle);
+}
+
+void OGLSampler::Release() {
+    if (handle == 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceDeletion);
+    glDeleteSamplers(1, &handle);
+    OpenGLState::GetCurState().ResetSampler(handle).Apply();
+    handle = 0;
+}
+
+void OGLShader::Create(const char* source, GLenum type) {
+    if (handle != 0)
+        return;
+    if (source == nullptr)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceCreation);
+    handle = GLShader::LoadShader(source, type);
+}
+
+void OGLShader::Release() {
+    if (handle == 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceDeletion);
+    glDeleteShader(handle);
+    handle = 0;
+}
+
+void OGLProgram::CreateFromSource(const char* vert_shader, const char* geo_shader,
+                                  const char* frag_shader, bool separable_program) {
+    OGLShader vert, geo, frag;
+    if (vert_shader)
+        vert.Create(vert_shader, GL_VERTEX_SHADER);
+    if (geo_shader)
+        geo.Create(geo_shader, GL_GEOMETRY_SHADER);
+    if (frag_shader)
+        frag.Create(frag_shader, GL_FRAGMENT_SHADER);
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceCreation);
+    Create(separable_program, vert.handle, geo.handle, frag.handle);
+}
+
+void OGLProgram::Release() {
+    if (handle == 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceDeletion);
+    glDeleteProgram(handle);
+    OpenGLState::GetCurState().ResetProgram(handle).Apply();
+    handle = 0;
+}
+
+void OGLPipeline::Create() {
+    if (handle != 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceCreation);
+    glGenProgramPipelines(1, &handle);
+}
+
+void OGLPipeline::Release() {
+    if (handle == 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceDeletion);
+    glDeleteProgramPipelines(1, &handle);
+    OpenGLState::GetCurState().ResetPipeline(handle).Apply();
+    handle = 0;
+}
+
+void OGLBuffer::Create() {
+    if (handle != 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceCreation);
+    glGenBuffers(1, &handle);
+}
+
+void OGLBuffer::Release() {
+    if (handle == 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceDeletion);
+    glDeleteBuffers(1, &handle);
+    OpenGLState::GetCurState().ResetBuffer(handle).Apply();
+    handle = 0;
+}
+
+void OGLSync::Create() {
+    if (handle != 0)
+        return;
+
+    // Don't profile here, this one is expected to happen ingame.
+    handle = glFenceSync(GL_SYNC_GPU_COMMANDS_COMPLETE, 0);
+}
+
+void OGLSync::Release() {
+    if (handle == 0)
+        return;
+
+    // Don't profile here, this one is expected to happen ingame.
+    glDeleteSync(handle);
+    handle = 0;
+}
+
+void OGLVertexArray::Create() {
+    if (handle != 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceCreation);
+    glGenVertexArrays(1, &handle);
+}
+
+void OGLVertexArray::Release() {
+    if (handle == 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceDeletion);
+    glDeleteVertexArrays(1, &handle);
+    OpenGLState::GetCurState().ResetVertexArray(handle).Apply();
+    handle = 0;
+}
+
+void OGLFramebuffer::Create() {
+    if (handle != 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceCreation);
+    glGenFramebuffers(1, &handle);
+}
+
+void OGLFramebuffer::Release() {
+    if (handle == 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceDeletion);
+    glDeleteFramebuffers(1, &handle);
+    OpenGLState::GetCurState().ResetFramebuffer(handle).Apply();
+    handle = 0;
+}
+
+} // namespace OpenGL

src/video_core/renderer_opengl/gl_resource_manager.h

@@ -8,7 +8,6 @@
 #include <glad/glad.h>
 #include "common/common_types.h"
 #include "video_core/renderer_opengl/gl_shader_util.h"
-#include "video_core/renderer_opengl/gl_state.h"
 
 namespace OpenGL {
 
@@ -29,20 +28,10 @@ public:
     }
 
     /// Creates a new internal OpenGL resource and stores the handle
-    void Create() {
-        if (handle != 0)
-            return;
-        glGenTextures(1, &handle);
-    }
+    void Create();
 
     /// Deletes the internal OpenGL resource
-    void Release() {
-        if (handle == 0)
-            return;
-        glDeleteTextures(1, &handle);
-        OpenGLState::GetCurState().UnbindTexture(handle).Apply();
-        handle = 0;
-    }
+    void Release();
 
     GLuint handle = 0;
 };
@@ -64,20 +53,10 @@ public:
     }
 
     /// Creates a new internal OpenGL resource and stores the handle
-    void Create() {
-        if (handle != 0)
-            return;
-        glGenSamplers(1, &handle);
-    }
+    void Create();
 
     /// Deletes the internal OpenGL resource
-    void Release() {
-        if (handle == 0)
-            return;
-        glDeleteSamplers(1, &handle);
-        OpenGLState::GetCurState().ResetSampler(handle).Apply();
-        handle = 0;
-    }
+    void Release();
 
     GLuint handle = 0;
 };
@@ -98,20 +77,9 @@ public:
         return *this;
     }
 
-    void Create(const char* source, GLenum type) {
-        if (handle != 0)
-            return;
-        if (source == nullptr)
-            return;
-        handle = GLShader::LoadShader(source, type);
-    }
+    void Create(const char* source, GLenum type);
 
-    void Release() {
-        if (handle == 0)
-            return;
-        glDeleteShader(handle);
-        handle = 0;
-    }
+    void Release();
 
     GLuint handle = 0;
 };
@@ -141,25 +109,10 @@ public:
 
     /// Creates a new internal OpenGL resource and stores the handle
     void CreateFromSource(const char* vert_shader, const char* geo_shader, const char* frag_shader,
-                          bool separable_program = false) {
-        OGLShader vert, geo, frag;
-        if (vert_shader)
-            vert.Create(vert_shader, GL_VERTEX_SHADER);
-        if (geo_shader)
-            geo.Create(geo_shader, GL_GEOMETRY_SHADER);
-        if (frag_shader)
-            frag.Create(frag_shader, GL_FRAGMENT_SHADER);
-        Create(separable_program, vert.handle, geo.handle, frag.handle);
-    }
+                          bool separable_program = false);
 
     /// Deletes the internal OpenGL resource
-    void Release() {
-        if (handle == 0)
-            return;
-        glDeleteProgram(handle);
-        OpenGLState::GetCurState().ResetProgram(handle).Apply();
-        handle = 0;
-    }
+    void Release();
 
     GLuint handle = 0;
 };
@@ -178,20 +131,10 @@ public:
     }
 
     /// Creates a new internal OpenGL resource and stores the handle
-    void Create() {
-        if (handle != 0)
-            return;
-        glGenProgramPipelines(1, &handle);
-    }
+    void Create();
 
     /// Deletes the internal OpenGL resource
-    void Release() {
-        if (handle == 0)
-            return;
-        glDeleteProgramPipelines(1, &handle);
-        OpenGLState::GetCurState().ResetPipeline(handle).Apply();
-        handle = 0;
-    }
+    void Release();
 
     GLuint handle = 0;
 };
@@ -213,20 +156,10 @@ public:
     }
 
     /// Creates a new internal OpenGL resource and stores the handle
-    void Create() {
-        if (handle != 0)
-            return;
-        glGenBuffers(1, &handle);
-    }
+    void Create();
 
     /// Deletes the internal OpenGL resource
-    void Release() {
-        if (handle == 0)
-            return;
-        glDeleteBuffers(1, &handle);
-        OpenGLState::GetCurState().ResetBuffer(handle).Apply();
-        handle = 0;
-    }
+    void Release();
 
     GLuint handle = 0;
 };
@@ -247,19 +180,10 @@ public:
     }
 
     /// Creates a new internal OpenGL resource and stores the handle
-    void Create() {
-        if (handle != 0)
-            return;
-        handle = glFenceSync(GL_SYNC_GPU_COMMANDS_COMPLETE, 0);
-    }
+    void Create();
 
     /// Deletes the internal OpenGL resource
-    void Release() {
-        if (handle == 0)
-            return;
-        glDeleteSync(handle);
-        handle = 0;
-    }
+    void Release();
 
     GLsync handle = 0;
 };
@@ -281,20 +205,10 @@ public:
     }
 
     /// Creates a new internal OpenGL resource and stores the handle
-    void Create() {
-        if (handle != 0)
-            return;
-        glGenVertexArrays(1, &handle);
-    }
+    void Create();
 
     /// Deletes the internal OpenGL resource
-    void Release() {
-        if (handle == 0)
-            return;
-        glDeleteVertexArrays(1, &handle);
-        OpenGLState::GetCurState().ResetVertexArray(handle).Apply();
-        handle = 0;
-    }
+    void Release();
 
     GLuint handle = 0;
 };
@@ -316,20 +230,10 @@ public:
     }
 
     /// Creates a new internal OpenGL resource and stores the handle
-    void Create() {
-        if (handle != 0)
-            return;
-        glGenFramebuffers(1, &handle);
-    }
+    void Create();
 
     /// Deletes the internal OpenGL resource
-    void Release() {
-        if (handle == 0)
-            return;
-        glDeleteFramebuffers(1, &handle);
-        OpenGLState::GetCurState().ResetFramebuffer(handle).Apply();
-        handle = 0;
-    }
+    void Release();
 
     GLuint handle = 0;
 };

src/video_core/renderer_opengl/gl_shader_cache.cpp

@@ -6,10 +6,10 @@
 #include "core/core.h"
 #include "core/memory.h"
 #include "video_core/engines/maxwell_3d.h"
+#include "video_core/renderer_opengl/gl_rasterizer.h"
 #include "video_core/renderer_opengl/gl_shader_cache.h"
 #include "video_core/renderer_opengl/gl_shader_manager.h"
 #include "video_core/renderer_opengl/utils.h"
-#include "video_core/utils.h"
 
 namespace OpenGL {
 
@@ -121,12 +121,16 @@ GLint CachedShader::GetUniformLocation(const GLShader::SamplerEntry& sampler) {
 }
 
 GLuint CachedShader::LazyGeometryProgram(OGLProgram& target_program,
-                                         const std::string& glsl_topology,
+                                         const std::string& glsl_topology, u32 max_vertices,
                                          const std::string& debug_name) {
     if (target_program.handle != 0) {
         return target_program.handle;
     }
-    const std::string source{geometry_programs.code + "layout (" + glsl_topology + ") in;\n"};
+    std::string source = "#version 430 core\n";
+    source += "layout (" + glsl_topology + ") in;\n";
+    source += "#define MAX_VERTEX_INPUT " + std::to_string(max_vertices) + '\n';
+    source += geometry_programs.code;
+
     OGLShader shader;
     shader.Create(source.c_str(), GL_GEOMETRY_SHADER);
     target_program.Create(true, shader.handle);
@@ -135,6 +139,8 @@ GLuint CachedShader::LazyGeometryProgram(OGLProgram& target_program,
     return target_program.handle;
 };
 
+ShaderCacheOpenGL::ShaderCacheOpenGL(RasterizerOpenGL& rasterizer) : RasterizerCache{rasterizer} {}
+
 Shader ShaderCacheOpenGL::GetStageProgram(Maxwell::ShaderProgram program) {
     const VAddr program_addr{GetShaderAddress(program)};
 

src/video_core/renderer_opengl/gl_shader_cache.h

@@ -16,6 +16,8 @@
 namespace OpenGL {
 
 class CachedShader;
+class RasterizerOpenGL;
+
 using Shader = std::shared_ptr<CachedShader>;
 using Maxwell = Tegra::Engines::Maxwell3D::Regs;
 
@@ -46,22 +48,23 @@ public:
         }
         switch (primitive_mode) {
         case GL_POINTS:
-            return LazyGeometryProgram(geometry_programs.points, "points", "ShaderPoints");
+            return LazyGeometryProgram(geometry_programs.points, "points", 1, "ShaderPoints");
         case GL_LINES:
         case GL_LINE_STRIP:
-            return LazyGeometryProgram(geometry_programs.lines, "lines", "ShaderLines");
+            return LazyGeometryProgram(geometry_programs.lines, "lines", 2, "ShaderLines");
         case GL_LINES_ADJACENCY:
         case GL_LINE_STRIP_ADJACENCY:
-            return LazyGeometryProgram(geometry_programs.lines_adjacency, "lines_adjacency",
+            return LazyGeometryProgram(geometry_programs.lines_adjacency, "lines_adjacency", 4,
                                        "ShaderLinesAdjacency");
         case GL_TRIANGLES:
         case GL_TRIANGLE_STRIP:
         case GL_TRIANGLE_FAN:
-            return LazyGeometryProgram(geometry_programs.triangles, "triangles", "ShaderTriangles");
+            return LazyGeometryProgram(geometry_programs.triangles, "triangles", 3,
+                                       "ShaderTriangles");
         case GL_TRIANGLES_ADJACENCY:
         case GL_TRIANGLE_STRIP_ADJACENCY:
             return LazyGeometryProgram(geometry_programs.triangles_adjacency, "triangles_adjacency",
-                                       "ShaderLines");
+                                       6, "ShaderTrianglesAdjacency");
         default:
             UNREACHABLE_MSG("Unknown primitive mode.");
         }
@@ -76,7 +79,7 @@ public:
 private:
     /// Generates a geometry shader or returns one that already exists.
     GLuint LazyGeometryProgram(OGLProgram& target_program, const std::string& glsl_topology,
-                               const std::string& debug_name);
+                               u32 max_vertices, const std::string& debug_name);
 
     VAddr addr;
     Maxwell::ShaderProgram program_type;
@@ -104,6 +107,8 @@ private:
 
 class ShaderCacheOpenGL final : public RasterizerCache<Shader> {
 public:
+    explicit ShaderCacheOpenGL(RasterizerOpenGL& rasterizer);
+
     /// Gets the current specified shader stage program
     Shader GetStageProgram(Maxwell::ShaderProgram program);
 };

src/video_core/renderer_opengl/gl_shader_decompiler.cpp

@@ -494,10 +494,10 @@ public:
             // instruction for now.
             if (stage == Maxwell3D::Regs::ShaderStage::Geometry) {
                 // TODO(Rodrigo): nouveau sets some attributes after setting emitting a geometry
-                // shader. These instructions use a dirty register as buffer index. To avoid some
-                // drivers from complaining for the out of boundary writes, guard them.
-                const std::string buf_index{"min(" + GetRegisterAsInteger(buf_reg) + ", " +
-                                            std::to_string(MAX_GEOMETRY_BUFFERS - 1) + ')'};
+                // shader. These instructions use a dirty register as buffer index, to avoid some
+                // drivers from complaining about out of boundary writes, guard them.
+                const std::string buf_index{"((" + GetRegisterAsInteger(buf_reg) + ") % " +
+                                            std::to_string(MAX_GEOMETRY_BUFFERS) + ')'};
                 shader.AddLine("amem[" + buf_index + "][" +
                                std::to_string(static_cast<u32>(attribute)) + ']' +
                                GetSwizzle(elem) + " = " + src + ';');
@@ -811,7 +811,11 @@ private:
                                   std::optional<Register> vertex = {}) {
         auto GeometryPass = [&](const std::string& name) {
             if (stage == Maxwell3D::Regs::ShaderStage::Geometry && vertex) {
-                return "gs_" + name + '[' + GetRegisterAsInteger(*vertex, 0, false) + ']';
+                // TODO(Rodrigo): Guard geometry inputs against out of bound reads. Some games set
+                // an 0x80000000 index for those and the shader fails to build. Find out why this
+                // happens and what's its intent.
+                return "gs_" + name + '[' + GetRegisterAsInteger(*vertex, 0, false) +
+                       " % MAX_VERTEX_INPUT]";
             }
             return name;
         };
@@ -2742,12 +2746,12 @@ private:
                 }
                 case 3: {
                     if (is_array) {
-                        UNIMPLEMENTED_MSG("3-coordinate arrays not fully implemented");
-                        const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
-                        const std::string y = regs.GetRegisterAsFloat(instr.gpr20);
-                        coord = "vec2 coords = vec2(" + x + ", " + y + ");";
-                        texture_type = Tegra::Shader::TextureType::Texture2D;
-                        is_array = false;
+                        const std::string index = regs.GetRegisterAsInteger(instr.gpr8);
+                        const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                        const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 2);
+                        const std::string z = regs.GetRegisterAsFloat(instr.gpr20);
+                        coord =
+                            "vec4 coords = vec4(" + x + ", " + y + ", " + z + ", " + index + ");";
                     } else {
                         const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
                         const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
@@ -2777,7 +2781,11 @@ private:
                     break;
                 }
                 case Tegra::Shader::TextureProcessMode::LZ: {
-                    texture = "textureLod(" + sampler + ", coords, 0.0)";
+                    if (depth_compare && is_array) {
+                        texture = "texture(" + sampler + ", coords)";
+                    } else {
+                        texture = "textureLod(" + sampler + ", coords, 0.0)";
+                    }
                     break;
                 }
                 case Tegra::Shader::TextureProcessMode::LL: {

src/video_core/renderer_opengl/gl_shader_gen.cpp

@@ -82,8 +82,8 @@ void main() {
 }
 
 ProgramResult GenerateGeometryShader(const ShaderSetup& setup) {
-    std::string out = "#version 430 core\n";
-    out += "#extension GL_ARB_separate_shader_objects : enable\n\n";
+    // Version is intentionally skipped in shader generation, it's added by the lazy compilation.
+    std::string out = "#extension GL_ARB_separate_shader_objects : enable\n\n";
     out += Decompiler::GetCommonDeclarations();
     out += "bool exec_geometry();\n";
 

src/video_core/renderer_opengl/gl_shader_manager.h

@@ -7,6 +7,7 @@
 #include <glad/glad.h>
 
 #include "video_core/renderer_opengl/gl_resource_manager.h"
+#include "video_core/renderer_opengl/gl_state.h"
 #include "video_core/renderer_opengl/maxwell_to_gl.h"
 
 namespace OpenGL::GLShader {
@@ -66,6 +67,7 @@ public:
         glUseProgramStages(pipeline.handle, GL_FRAGMENT_SHADER_BIT, fs);
         state.draw.shader_program = 0;
         state.draw.program_pipeline = pipeline.handle;
+        state.geometry_shaders.enabled = (gs != 0);
     }
 
 private:

src/video_core/renderer_opengl/gl_state.cpp

@@ -14,7 +14,10 @@ OpenGLState OpenGLState::cur_state;
 bool OpenGLState::s_rgb_used;
 OpenGLState::OpenGLState() {
     // These all match default OpenGL values
+    geometry_shaders.enabled = false;
     framebuffer_srgb.enabled = false;
+    multisample_control.alpha_to_coverage = false;
+    multisample_control.alpha_to_one = false;
     cull.enabled = false;
     cull.mode = GL_BACK;
     cull.front_face = GL_CCW;
@@ -22,17 +25,15 @@ OpenGLState::OpenGLState() {
     depth.test_enabled = false;
     depth.test_func = GL_LESS;
     depth.write_mask = GL_TRUE;
-    depth.depth_range_near = 0.0f;
-    depth.depth_range_far = 1.0f;
 
     primitive_restart.enabled = false;
     primitive_restart.index = 0;
-
-    color_mask.red_enabled = GL_TRUE;
-    color_mask.green_enabled = GL_TRUE;
-    color_mask.blue_enabled = GL_TRUE;
-    color_mask.alpha_enabled = GL_TRUE;
-
+    for (auto& item : color_mask) {
+        item.red_enabled = GL_TRUE;
+        item.green_enabled = GL_TRUE;
+        item.blue_enabled = GL_TRUE;
+        item.alpha_enabled = GL_TRUE;
+    }
     stencil.test_enabled = false;
     auto reset_stencil = [](auto& config) {
         config.test_func = GL_ALWAYS;
@@ -45,19 +46,33 @@ OpenGLState::OpenGLState() {
     };
     reset_stencil(stencil.front);
     reset_stencil(stencil.back);
-
-    blend.enabled = true;
-    blend.rgb_equation = GL_FUNC_ADD;
-    blend.a_equation = GL_FUNC_ADD;
-    blend.src_rgb_func = GL_ONE;
-    blend.dst_rgb_func = GL_ZERO;
-    blend.src_a_func = GL_ONE;
-    blend.dst_a_func = GL_ZERO;
-    blend.color.red = 0.0f;
-    blend.color.green = 0.0f;
-    blend.color.blue = 0.0f;
-    blend.color.alpha = 0.0f;
-
+    for (auto& item : viewports) {
+        item.x = 0;
+        item.y = 0;
+        item.width = 0;
+        item.height = 0;
+        item.depth_range_near = 0.0f;
+        item.depth_range_far = 1.0f;
+        item.scissor.enabled = false;
+        item.scissor.x = 0;
+        item.scissor.y = 0;
+        item.scissor.width = 0;
+        item.scissor.height = 0;
+    }
+    for (auto& item : blend) {
+        item.enabled = true;
+        item.rgb_equation = GL_FUNC_ADD;
+        item.a_equation = GL_FUNC_ADD;
+        item.src_rgb_func = GL_ONE;
+        item.dst_rgb_func = GL_ZERO;
+        item.src_a_func = GL_ONE;
+        item.dst_a_func = GL_ZERO;
+    }
+    independant_blend.enabled = false;
+    blend_color.red = 0.0f;
+    blend_color.green = 0.0f;
+    blend_color.blue = 0.0f;
+    blend_color.alpha = 0.0f;
     logic_op.enabled = false;
     logic_op.operation = GL_COPY;
 
@@ -73,20 +88,10 @@ OpenGLState::OpenGLState() {
     draw.shader_program = 0;
     draw.program_pipeline = 0;
 
-    scissor.enabled = false;
-    scissor.x = 0;
-    scissor.y = 0;
-    scissor.width = 0;
-    scissor.height = 0;
-
-    viewport.x = 0;
-    viewport.y = 0;
-    viewport.width = 0;
-    viewport.height = 0;
-
     clip_distance = {};
 
     point.size = 1;
+    fragment_color_clamp.enabled = false;
 }
 
 void OpenGLState::ApplyDefaultState() {
@@ -134,6 +139,32 @@ void OpenGLState::ApplyCulling() const {
     }
 }
 
+void OpenGLState::ApplyColorMask() const {
+    if (GLAD_GL_ARB_viewport_array && independant_blend.enabled) {
+        for (size_t i = 0; i < Tegra::Engines::Maxwell3D::Regs::NumRenderTargets; i++) {
+            const auto& updated = color_mask[i];
+            const auto& current = cur_state.color_mask[i];
+            if (updated.red_enabled != current.red_enabled ||
+                updated.green_enabled != current.green_enabled ||
+                updated.blue_enabled != current.blue_enabled ||
+                updated.alpha_enabled != current.alpha_enabled) {
+                glColorMaski(static_cast<GLuint>(i), updated.red_enabled, updated.green_enabled,
+                             updated.blue_enabled, updated.alpha_enabled);
+            }
+        }
+    } else {
+        const auto& updated = color_mask[0];
+        const auto& current = cur_state.color_mask[0];
+        if (updated.red_enabled != current.red_enabled ||
+            updated.green_enabled != current.green_enabled ||
+            updated.blue_enabled != current.blue_enabled ||
+            updated.alpha_enabled != current.alpha_enabled) {
+            glColorMask(updated.red_enabled, updated.green_enabled, updated.blue_enabled,
+                        updated.alpha_enabled);
+        }
+    }
+}
+
 void OpenGLState::ApplyDepth() const {
     // Depth test
     const bool depth_test_changed = depth.test_enabled != cur_state.depth.test_enabled;
@@ -152,11 +183,6 @@ void OpenGLState::ApplyDepth() const {
     if (depth.write_mask != cur_state.depth.write_mask) {
         glDepthMask(depth.write_mask);
     }
-    // Depth range
-    if (depth.depth_range_near != cur_state.depth.depth_range_near ||
-        depth.depth_range_far != cur_state.depth.depth_range_far) {
-        glDepthRange(depth.depth_range_near, depth.depth_range_far);
-    }
 }
 
 void OpenGLState::ApplyPrimitiveRestart() const {
@@ -208,60 +234,143 @@ void OpenGLState::ApplyStencilTest() const {
     }
 }
 
-void OpenGLState::ApplyScissorTest() const {
-    const bool scissor_changed = scissor.enabled != cur_state.scissor.enabled;
-    if (scissor_changed) {
-        if (scissor.enabled) {
-            glEnable(GL_SCISSOR_TEST);
-        } else {
-            glDisable(GL_SCISSOR_TEST);
+void OpenGLState::ApplyViewport() const {
+    if (GLAD_GL_ARB_viewport_array && geometry_shaders.enabled) {
+        for (GLuint i = 0; i < static_cast<GLuint>(Tegra::Engines::Maxwell3D::Regs::NumViewports);
+             i++) {
+            const auto& current = cur_state.viewports[i];
+            const auto& updated = viewports[i];
+            if (updated.x != current.x || updated.y != current.y ||
+                updated.width != current.width || updated.height != current.height) {
+                glViewportIndexedf(i, updated.x, updated.y, updated.width, updated.height);
+            }
+            if (updated.depth_range_near != current.depth_range_near ||
+                updated.depth_range_far != current.depth_range_far) {
+                glDepthRangeIndexed(i, updated.depth_range_near, updated.depth_range_far);
+            }
+            const bool scissor_changed = updated.scissor.enabled != current.scissor.enabled;
+            if (scissor_changed) {
+                if (updated.scissor.enabled) {
+                    glEnablei(GL_SCISSOR_TEST, i);
+                } else {
+                    glDisablei(GL_SCISSOR_TEST, i);
+                }
+            }
+            if (updated.scissor.enabled &&
+                (scissor_changed || updated.scissor.x != current.scissor.x ||
+                 updated.scissor.y != current.scissor.y ||
+                 updated.scissor.width != current.scissor.width ||
+                 updated.scissor.height != current.scissor.height)) {
+                glScissorIndexed(i, updated.scissor.x, updated.scissor.y, updated.scissor.width,
+                                 updated.scissor.height);
+            }
+        }
+    } else {
+        const auto& current = cur_state.viewports[0];
+        const auto& updated = viewports[0];
+        if (updated.x != current.x || updated.y != current.y || updated.width != current.width ||
+            updated.height != current.height) {
+            glViewport(static_cast<GLint>(updated.x), static_cast<GLint>(updated.y),
+                       static_cast<GLsizei>(updated.width), static_cast<GLsizei>(updated.height));
+        }
+        if (updated.depth_range_near != current.depth_range_near ||
+            updated.depth_range_far != current.depth_range_far) {
+            glDepthRange(updated.depth_range_near, updated.depth_range_far);
+        }
+        const bool scissor_changed = updated.scissor.enabled != current.scissor.enabled;
+        if (scissor_changed) {
+            if (updated.scissor.enabled) {
+                glEnable(GL_SCISSOR_TEST);
+            } else {
+                glDisable(GL_SCISSOR_TEST);
+            }
+        }
+        if (updated.scissor.enabled && (scissor_changed || updated.scissor.x != current.scissor.x ||
+                                        updated.scissor.y != current.scissor.y ||
+                                        updated.scissor.width != current.scissor.width ||
+                                        updated.scissor.height != current.scissor.height)) {
+            glScissor(updated.scissor.x, updated.scissor.y, updated.scissor.width,
+                      updated.scissor.height);
         }
-    }
-    if (scissor_changed || scissor_changed || scissor.x != cur_state.scissor.x ||
-        scissor.y != cur_state.scissor.y || scissor.width != cur_state.scissor.width ||
-        scissor.height != cur_state.scissor.height) {
-        glScissor(scissor.x, scissor.y, scissor.width, scissor.height);
     }
 }
 
-void OpenGLState::ApplyBlending() const {
-    const bool blend_changed = blend.enabled != cur_state.blend.enabled;
+void OpenGLState::ApplyGlobalBlending() const {
+    const Blend& current = cur_state.blend[0];
+    const Blend& updated = blend[0];
+    const bool blend_changed = updated.enabled != current.enabled;
     if (blend_changed) {
-        if (blend.enabled) {
-            ASSERT(!logic_op.enabled);
+        if (updated.enabled) {
             glEnable(GL_BLEND);
         } else {
             glDisable(GL_BLEND);
         }
     }
-    if (blend.enabled) {
-        if (blend_changed || blend.color.red != cur_state.blend.color.red ||
-            blend.color.green != cur_state.blend.color.green ||
-            blend.color.blue != cur_state.blend.color.blue ||
-            blend.color.alpha != cur_state.blend.color.alpha) {
-            glBlendColor(blend.color.red, blend.color.green, blend.color.blue, blend.color.alpha);
-        }
+    if (!updated.enabled) {
+        return;
+    }
+    if (blend_changed || updated.src_rgb_func != current.src_rgb_func ||
+        updated.dst_rgb_func != current.dst_rgb_func || updated.src_a_func != current.src_a_func ||
+        updated.dst_a_func != current.dst_a_func) {
+        glBlendFuncSeparate(updated.src_rgb_func, updated.dst_rgb_func, updated.src_a_func,
+                            updated.dst_a_func);
+    }
 
-        if (blend_changed || blend.src_rgb_func != cur_state.blend.src_rgb_func ||
-            blend.dst_rgb_func != cur_state.blend.dst_rgb_func ||
-            blend.src_a_func != cur_state.blend.src_a_func ||
-            blend.dst_a_func != cur_state.blend.dst_a_func) {
-            glBlendFuncSeparate(blend.src_rgb_func, blend.dst_rgb_func, blend.src_a_func,
-                                blend.dst_a_func);
-        }
+    if (blend_changed || updated.rgb_equation != current.rgb_equation ||
+        updated.a_equation != current.a_equation) {
+        glBlendEquationSeparate(updated.rgb_equation, updated.a_equation);
+    }
+}
 
-        if (blend_changed || blend.rgb_equation != cur_state.blend.rgb_equation ||
-            blend.a_equation != cur_state.blend.a_equation) {
-            glBlendEquationSeparate(blend.rgb_equation, blend.a_equation);
+void OpenGLState::ApplyTargetBlending(std::size_t target, bool force) const {
+    const Blend& updated = blend[target];
+    const Blend& current = cur_state.blend[target];
+    const bool blend_changed = updated.enabled != current.enabled || force;
+    if (blend_changed) {
+        if (updated.enabled) {
+            glEnablei(GL_BLEND, static_cast<GLuint>(target));
+        } else {
+            glDisablei(GL_BLEND, static_cast<GLuint>(target));
         }
     }
+    if (!updated.enabled) {
+        return;
+    }
+    if (blend_changed || updated.src_rgb_func != current.src_rgb_func ||
+        updated.dst_rgb_func != current.dst_rgb_func || updated.src_a_func != current.src_a_func ||
+        updated.dst_a_func != current.dst_a_func) {
+        glBlendFuncSeparateiARB(static_cast<GLuint>(target), updated.src_rgb_func,
+                                updated.dst_rgb_func, updated.src_a_func, updated.dst_a_func);
+    }
+
+    if (blend_changed || updated.rgb_equation != current.rgb_equation ||
+        updated.a_equation != current.a_equation) {
+        glBlendEquationSeparateiARB(static_cast<GLuint>(target), updated.rgb_equation,
+                                    updated.a_equation);
+    }
+}
+
+void OpenGLState::ApplyBlending() const {
+    if (independant_blend.enabled) {
+        for (size_t i = 0; i < Tegra::Engines::Maxwell3D::Regs::NumRenderTargets; i++) {
+            ApplyTargetBlending(i,
+                                independant_blend.enabled != cur_state.independant_blend.enabled);
+        }
+    } else {
+        ApplyGlobalBlending();
+    }
+    if (blend_color.red != cur_state.blend_color.red ||
+        blend_color.green != cur_state.blend_color.green ||
+        blend_color.blue != cur_state.blend_color.blue ||
+        blend_color.alpha != cur_state.blend_color.alpha) {
+        glBlendColor(blend_color.red, blend_color.green, blend_color.blue, blend_color.alpha);
+    }
 }
 
 void OpenGLState::ApplyLogicOp() const {
     const bool logic_op_changed = logic_op.enabled != cur_state.logic_op.enabled;
     if (logic_op_changed) {
         if (logic_op.enabled) {
-            ASSERT(!blend.enabled);
             glEnable(GL_COLOR_LOGIC_OP);
         } else {
             glDisable(GL_COLOR_LOGIC_OP);
@@ -315,7 +424,7 @@ void OpenGLState::ApplySamplers() const {
     }
 }
 
-void OpenGLState::Apply() const {
+void OpenGLState::ApplyFramebufferState() const {
     // Framebuffer
     if (draw.read_framebuffer != cur_state.draw.read_framebuffer) {
         glBindFramebuffer(GL_READ_FRAMEBUFFER, draw.read_framebuffer);
@@ -323,7 +432,9 @@ void OpenGLState::Apply() const {
     if (draw.draw_framebuffer != cur_state.draw.draw_framebuffer) {
         glBindFramebuffer(GL_DRAW_FRAMEBUFFER, draw.draw_framebuffer);
     }
+}
 
+void OpenGLState::ApplyVertexBufferState() const {
     // Vertex array
     if (draw.vertex_array != cur_state.draw.vertex_array) {
         glBindVertexArray(draw.vertex_array);
@@ -333,7 +444,11 @@ void OpenGLState::Apply() const {
     if (draw.vertex_buffer != cur_state.draw.vertex_buffer) {
         glBindBuffer(GL_ARRAY_BUFFER, draw.vertex_buffer);
     }
+}
 
+void OpenGLState::Apply() const {
+    ApplyFramebufferState();
+    ApplyVertexBufferState();
     // Uniform buffer
     if (draw.uniform_buffer != cur_state.draw.uniform_buffer) {
         glBindBuffer(GL_UNIFORM_BUFFER, draw.uniform_buffer);
@@ -348,12 +463,6 @@ void OpenGLState::Apply() const {
     if (draw.program_pipeline != cur_state.draw.program_pipeline) {
         glBindProgramPipeline(draw.program_pipeline);
     }
-    // Viewport
-    if (viewport.x != cur_state.viewport.x || viewport.y != cur_state.viewport.y ||
-        viewport.width != cur_state.viewport.width ||
-        viewport.height != cur_state.viewport.height) {
-        glViewport(viewport.x, viewport.y, viewport.width, viewport.height);
-    }
     // Clip distance
     for (std::size_t i = 0; i < clip_distance.size(); ++i) {
         if (clip_distance[i] != cur_state.clip_distance[i]) {
@@ -364,19 +473,33 @@ void OpenGLState::Apply() const {
             }
         }
     }
-    // Color mask
-    if (color_mask.red_enabled != cur_state.color_mask.red_enabled ||
-        color_mask.green_enabled != cur_state.color_mask.green_enabled ||
-        color_mask.blue_enabled != cur_state.color_mask.blue_enabled ||
-        color_mask.alpha_enabled != cur_state.color_mask.alpha_enabled) {
-        glColorMask(color_mask.red_enabled, color_mask.green_enabled, color_mask.blue_enabled,
-                    color_mask.alpha_enabled);
-    }
     // Point
     if (point.size != cur_state.point.size) {
         glPointSize(point.size);
     }
-    ApplyScissorTest();
+    if (GLAD_GL_ARB_color_buffer_float) {
+        if (fragment_color_clamp.enabled != cur_state.fragment_color_clamp.enabled) {
+            glClampColor(GL_CLAMP_FRAGMENT_COLOR_ARB,
+                         fragment_color_clamp.enabled ? GL_TRUE : GL_FALSE);
+        }
+    }
+    if (multisample_control.alpha_to_coverage != cur_state.multisample_control.alpha_to_coverage) {
+        if (multisample_control.alpha_to_coverage) {
+            glEnable(GL_SAMPLE_ALPHA_TO_COVERAGE);
+        } else {
+            glDisable(GL_SAMPLE_ALPHA_TO_COVERAGE);
+        }
+    }
+    if (multisample_control.alpha_to_one != cur_state.multisample_control.alpha_to_one) {
+        if (multisample_control.alpha_to_one) {
+            glEnable(GL_SAMPLE_ALPHA_TO_ONE);
+        } else {
+            glDisable(GL_SAMPLE_ALPHA_TO_ONE);
+        }
+    }
+
+    ApplyColorMask();
+    ApplyViewport();
     ApplyStencilTest();
     ApplySRgb();
     ApplyCulling();

src/video_core/renderer_opengl/gl_state.h

@@ -39,6 +39,19 @@ public:
         bool enabled; // GL_FRAMEBUFFER_SRGB
     } framebuffer_srgb;
 
+    struct {
+        bool alpha_to_coverage; // GL_ALPHA_TO_COVERAGE
+        bool alpha_to_one;      // GL_ALPHA_TO_ONE
+    } multisample_control;
+
+    struct {
+        bool enabled; // GL_CLAMP_FRAGMENT_COLOR_ARB
+    } fragment_color_clamp;
+
+    struct {
+        bool enabled; // viewports arrays are only supported when geometry shaders are enabled.
+    } geometry_shaders;
+
     struct {
         bool enabled;      // GL_CULL_FACE
         GLenum mode;       // GL_CULL_FACE_MODE
@@ -46,11 +59,9 @@ public:
     } cull;
 
     struct {
-        bool test_enabled;        // GL_DEPTH_TEST
-        GLenum test_func;         // GL_DEPTH_FUNC
-        GLboolean write_mask;     // GL_DEPTH_WRITEMASK
-        GLfloat depth_range_near; // GL_DEPTH_RANGE
-        GLfloat depth_range_far;  // GL_DEPTH_RANGE
+        bool test_enabled;    // GL_DEPTH_TEST
+        GLenum test_func;     // GL_DEPTH_FUNC
+        GLboolean write_mask; // GL_DEPTH_WRITEMASK
     } depth;
 
     struct {
@@ -58,13 +69,14 @@ public:
         GLuint index;
     } primitive_restart; // GL_PRIMITIVE_RESTART
 
-    struct {
+    struct ColorMask {
         GLboolean red_enabled;
         GLboolean green_enabled;
         GLboolean blue_enabled;
         GLboolean alpha_enabled;
-    } color_mask; // GL_COLOR_WRITEMASK
-
+    };
+    std::array<ColorMask, Tegra::Engines::Maxwell3D::Regs::NumRenderTargets>
+        color_mask; // GL_COLOR_WRITEMASK
     struct {
         bool test_enabled; // GL_STENCIL_TEST
         struct {
@@ -78,7 +90,7 @@ public:
         } front, back;
     } stencil;
 
-    struct {
+    struct Blend {
         bool enabled;        // GL_BLEND
         GLenum rgb_equation; // GL_BLEND_EQUATION_RGB
         GLenum a_equation;   // GL_BLEND_EQUATION_ALPHA
@@ -86,14 +98,19 @@ public:
         GLenum dst_rgb_func; // GL_BLEND_DST_RGB
         GLenum src_a_func;   // GL_BLEND_SRC_ALPHA
         GLenum dst_a_func;   // GL_BLEND_DST_ALPHA
+    };
+    std::array<Blend, Tegra::Engines::Maxwell3D::Regs::NumRenderTargets> blend;
 
-        struct {
-            GLclampf red;
-            GLclampf green;
-            GLclampf blue;
-            GLclampf alpha;
-        } color; // GL_BLEND_COLOR
-    } blend;
+    struct {
+        bool enabled;
+    } independant_blend;
+
+    struct {
+        GLclampf red;
+        GLclampf green;
+        GLclampf blue;
+        GLclampf alpha;
+    } blend_color; // GL_BLEND_COLOR
 
     struct {
         bool enabled; // GL_LOGIC_OP_MODE
@@ -138,20 +155,22 @@ public:
         GLuint program_pipeline; // GL_PROGRAM_PIPELINE_BINDING
     } draw;
 
-    struct {
-        bool enabled; // GL_SCISSOR_TEST
-        GLint x;
-        GLint y;
-        GLsizei width;
-        GLsizei height;
-    } scissor;
-
-    struct {
-        GLint x;
-        GLint y;
-        GLsizei width;
-        GLsizei height;
-    } viewport;
+    struct viewport {
+        GLfloat x;
+        GLfloat y;
+        GLfloat width;
+        GLfloat height;
+        GLfloat depth_range_near; // GL_DEPTH_RANGE
+        GLfloat depth_range_far;  // GL_DEPTH_RANGE
+        struct {
+            bool enabled; // GL_SCISSOR_TEST
+            GLint x;
+            GLint y;
+            GLsizei width;
+            GLsizei height;
+        } scissor;
+    };
+    std::array<viewport, Tegra::Engines::Maxwell3D::Regs::NumViewports> viewports;
 
     struct {
         float size; // GL_POINT_SIZE
@@ -173,6 +192,10 @@ public:
     }
     /// Apply this state as the current OpenGL state
     void Apply() const;
+    /// Apply only the state afecting the framebuffer
+    void ApplyFramebufferState() const;
+    /// Apply only the state afecting the vertex buffer
+    void ApplyVertexBufferState() const;
     /// Set the initial OpenGL state
     static void ApplyDefaultState();
     /// Resets any references to the given resource
@@ -191,10 +214,13 @@ private:
     static bool s_rgb_used;
     void ApplySRgb() const;
     void ApplyCulling() const;
+    void ApplyColorMask() const;
     void ApplyDepth() const;
     void ApplyPrimitiveRestart() const;
     void ApplyStencilTest() const;
-    void ApplyScissorTest() const;
+    void ApplyViewport() const;
+    void ApplyTargetBlending(std::size_t target, bool force) const;
+    void ApplyGlobalBlending() const;
     void ApplyBlending() const;
     void ApplyLogicOp() const;
     void ApplyTextures() const;

src/video_core/renderer_opengl/gl_stream_buffer.cpp

@@ -6,9 +6,13 @@
 #include <vector>
 #include "common/alignment.h"
 #include "common/assert.h"
+#include "common/microprofile.h"
 #include "video_core/renderer_opengl/gl_state.h"
 #include "video_core/renderer_opengl/gl_stream_buffer.h"
 
+MICROPROFILE_DEFINE(OpenGL_StreamBuffer, "OpenGL", "Stream Buffer Orphaning",
+                    MP_RGB(128, 128, 192));
+
 namespace OpenGL {
 
 OGLStreamBuffer::OGLStreamBuffer(GLenum target, GLsizeiptr size, bool prefer_coherent)
@@ -75,6 +79,7 @@ std::tuple<u8*, GLintptr, bool> OGLStreamBuffer::Map(GLsizeiptr size, GLintptr a
     }
 
     if (invalidate || !persistent) {
+        MICROPROFILE_SCOPE(OpenGL_StreamBuffer);
         GLbitfield flags = GL_MAP_WRITE_BIT | (persistent ? GL_MAP_PERSISTENT_BIT : 0) |
                            (coherent ? GL_MAP_COHERENT_BIT : GL_MAP_FLUSH_EXPLICIT_BIT) |
                            (invalidate ? GL_MAP_INVALIDATE_BUFFER_BIT : GL_MAP_UNSYNCHRONIZED_BIT);

src/video_core/renderer_opengl/maxwell_to_gl.h

@@ -159,10 +159,8 @@ inline GLenum TextureFilterMode(Tegra::Texture::TextureFilter filter_mode,
         }
     }
     }
-    LOG_CRITICAL(Render_OpenGL, "Unimplemented texture filter mode={}",
-                 static_cast<u32>(filter_mode));
-    UNREACHABLE();
-    return {};
+    LOG_ERROR(Render_OpenGL, "Unimplemented texture filter mode={}", static_cast<u32>(filter_mode));
+    return GL_LINEAR;
 }
 
 inline GLenum WrapMode(Tegra::Texture::WrapMode wrap_mode) {
@@ -182,10 +180,15 @@ inline GLenum WrapMode(Tegra::Texture::WrapMode wrap_mode) {
         return GL_CLAMP_TO_BORDER;
     case Tegra::Texture::WrapMode::MirrorOnceClampToEdge:
         return GL_MIRROR_CLAMP_TO_EDGE;
+    case Tegra::Texture::WrapMode::MirrorOnceBorder:
+        if (GL_EXT_texture_mirror_clamp) {
+            return GL_MIRROR_CLAMP_TO_BORDER_EXT;
+        } else {
+            return GL_MIRROR_CLAMP_TO_EDGE;
+        }
     }
-    LOG_CRITICAL(Render_OpenGL, "Unimplemented texture wrap mode={}", static_cast<u32>(wrap_mode));
-    UNREACHABLE();
-    return {};
+    LOG_ERROR(Render_OpenGL, "Unimplemented texture wrap mode={}", static_cast<u32>(wrap_mode));
+    return GL_REPEAT;
 }
 
 inline GLenum DepthCompareFunc(Tegra::Texture::DepthCompareFunc func) {
@@ -207,10 +210,9 @@ inline GLenum DepthCompareFunc(Tegra::Texture::DepthCompareFunc func) {
     case Tegra::Texture::DepthCompareFunc::Always:
         return GL_ALWAYS;
     }
-    LOG_CRITICAL(Render_OpenGL, "Unimplemented texture depth compare function ={}",
-                 static_cast<u32>(func));
-    UNREACHABLE();
-    return {};
+    LOG_ERROR(Render_OpenGL, "Unimplemented texture depth compare function ={}",
+              static_cast<u32>(func));
+    return GL_GREATER;
 }
 
 inline GLenum BlendEquation(Maxwell::Blend::Equation equation) {
@@ -226,9 +228,8 @@ inline GLenum BlendEquation(Maxwell::Blend::Equation equation) {
     case Maxwell::Blend::Equation::Max:
         return GL_MAX;
     }
-    LOG_CRITICAL(Render_OpenGL, "Unimplemented blend equation={}", static_cast<u32>(equation));
-    UNREACHABLE();
-    return {};
+    LOG_ERROR(Render_OpenGL, "Unimplemented blend equation={}", static_cast<u32>(equation));
+    return GL_FUNC_ADD;
 }
 
 inline GLenum BlendFunc(Maxwell::Blend::Factor factor) {
@@ -291,9 +292,8 @@ inline GLenum BlendFunc(Maxwell::Blend::Factor factor) {
     case Maxwell::Blend::Factor::OneMinusConstantAlphaGL:
         return GL_ONE_MINUS_CONSTANT_ALPHA;
     }
-    LOG_CRITICAL(Render_OpenGL, "Unimplemented blend factor={}", static_cast<u32>(factor));
-    UNREACHABLE();
-    return {};
+    LOG_ERROR(Render_OpenGL, "Unimplemented blend factor={}", static_cast<u32>(factor));
+    return GL_ZERO;
 }
 
 inline GLenum SwizzleSource(Tegra::Texture::SwizzleSource source) {
@@ -312,9 +312,8 @@ inline GLenum SwizzleSource(Tegra::Texture::SwizzleSource source) {
     case Tegra::Texture::SwizzleSource::OneFloat:
         return GL_ONE;
     }
-    LOG_CRITICAL(Render_OpenGL, "Unimplemented swizzle source={}", static_cast<u32>(source));
-    UNREACHABLE();
-    return {};
+    LOG_ERROR(Render_OpenGL, "Unimplemented swizzle source={}", static_cast<u32>(source));
+    return GL_ZERO;
 }
 
 inline GLenum ComparisonOp(Maxwell::ComparisonOp comparison) {
@@ -344,33 +343,39 @@ inline GLenum ComparisonOp(Maxwell::ComparisonOp comparison) {
     case Maxwell::ComparisonOp::AlwaysOld:
         return GL_ALWAYS;
     }
-    LOG_CRITICAL(Render_OpenGL, "Unimplemented comparison op={}", static_cast<u32>(comparison));
-    UNREACHABLE();
-    return {};
+    LOG_ERROR(Render_OpenGL, "Unimplemented comparison op={}", static_cast<u32>(comparison));
+    return GL_ALWAYS;
 }
 
 inline GLenum StencilOp(Maxwell::StencilOp stencil) {
     switch (stencil) {
     case Maxwell::StencilOp::Keep:
+    case Maxwell::StencilOp::KeepOGL:
         return GL_KEEP;
     case Maxwell::StencilOp::Zero:
+    case Maxwell::StencilOp::ZeroOGL:
         return GL_ZERO;
     case Maxwell::StencilOp::Replace:
+    case Maxwell::StencilOp::ReplaceOGL:
         return GL_REPLACE;
     case Maxwell::StencilOp::Incr:
+    case Maxwell::StencilOp::IncrOGL:
         return GL_INCR;
     case Maxwell::StencilOp::Decr:
+    case Maxwell::StencilOp::DecrOGL:
         return GL_DECR;
     case Maxwell::StencilOp::Invert:
+    case Maxwell::StencilOp::InvertOGL:
         return GL_INVERT;
     case Maxwell::StencilOp::IncrWrap:
+    case Maxwell::StencilOp::IncrWrapOGL:
         return GL_INCR_WRAP;
     case Maxwell::StencilOp::DecrWrap:
+    case Maxwell::StencilOp::DecrWrapOGL:
         return GL_DECR_WRAP;
     }
-    LOG_CRITICAL(Render_OpenGL, "Unimplemented stencil op={}", static_cast<u32>(stencil));
-    UNREACHABLE();
-    return {};
+    LOG_ERROR(Render_OpenGL, "Unimplemented stencil op={}", static_cast<u32>(stencil));
+    return GL_KEEP;
 }
 
 inline GLenum FrontFace(Maxwell::Cull::FrontFace front_face) {
@@ -380,9 +385,8 @@ inline GLenum FrontFace(Maxwell::Cull::FrontFace front_face) {
     case Maxwell::Cull::FrontFace::CounterClockWise:
         return GL_CCW;
     }
-    LOG_CRITICAL(Render_OpenGL, "Unimplemented front face cull={}", static_cast<u32>(front_face));
-    UNREACHABLE();
-    return {};
+    LOG_ERROR(Render_OpenGL, "Unimplemented front face cull={}", static_cast<u32>(front_face));
+    return GL_CCW;
 }
 
 inline GLenum CullFace(Maxwell::Cull::CullFace cull_face) {
@@ -394,9 +398,8 @@ inline GLenum CullFace(Maxwell::Cull::CullFace cull_face) {
     case Maxwell::Cull::CullFace::FrontAndBack:
         return GL_FRONT_AND_BACK;
     }
-    LOG_CRITICAL(Render_OpenGL, "Unimplemented cull face={}", static_cast<u32>(cull_face));
-    UNREACHABLE();
-    return {};
+    LOG_ERROR(Render_OpenGL, "Unimplemented cull face={}", static_cast<u32>(cull_face));
+    return GL_BACK;
 }
 
 inline GLenum LogicOp(Maxwell::LogicOperation operation) {
@@ -434,9 +437,8 @@ inline GLenum LogicOp(Maxwell::LogicOperation operation) {
     case Maxwell::LogicOperation::Set:
         return GL_SET;
     }
-    LOG_CRITICAL(Render_OpenGL, "Unimplemented logic operation={}", static_cast<u32>(operation));
-    UNREACHABLE();
-    return {};
+    LOG_ERROR(Render_OpenGL, "Unimplemented logic operation={}", static_cast<u32>(operation));
+    return GL_COPY;
 }
 
 } // namespace MaxwellToGL