video_core: fixed arithmetic overflow warnings & improved code style

- Fixed all warnings, for renderer_opengl items, which were indicating a
possible incorrect behavior from integral promotion rules and types
larger than those in which arithmetic is typically performed.
- Added const for variables where possible and meaningful.
- Added constexpr where possible.

src/audio_core/audio_renderer.cpp

@@ -17,10 +17,10 @@ AudioRenderer::AudioRenderer(AudioRendererParameter params,
                              Kernel::SharedPtr<Kernel::Event> buffer_event)
     : worker_params{params}, buffer_event{buffer_event}, voices(params.voice_count) {
 
-    audio_core = std::make_unique<AudioCore::AudioOut>();
-    stream = audio_core->OpenStream(STREAM_SAMPLE_RATE, STREAM_NUM_CHANNELS, "AudioRenderer",
-                                    [=]() { buffer_event->Signal(); });
-    audio_core->StartStream(stream);
+    audio_out = std::make_unique<AudioCore::AudioOut>();
+    stream = audio_out->OpenStream(STREAM_SAMPLE_RATE, STREAM_NUM_CHANNELS, "AudioRenderer",
+                                   [=]() { buffer_event->Signal(); });
+    audio_out->StartStream(stream);
 
     QueueMixedBuffer(0);
     QueueMixedBuffer(1);
@@ -236,11 +236,11 @@ void AudioRenderer::QueueMixedBuffer(Buffer::Tag tag) {
             }
         }
     }
-    audio_core->QueueBuffer(stream, tag, std::move(buffer));
+    audio_out->QueueBuffer(stream, tag, std::move(buffer));
 }
 
 void AudioRenderer::ReleaseAndQueueBuffers() {
-    const auto released_buffers{audio_core->GetTagsAndReleaseBuffers(stream, 2)};
+    const auto released_buffers{audio_out->GetTagsAndReleaseBuffers(stream, 2)};
     for (const auto& tag : released_buffers) {
         QueueMixedBuffer(tag);
     }

src/audio_core/audio_renderer.h

@@ -204,7 +204,7 @@ private:
     AudioRendererParameter worker_params;
     Kernel::SharedPtr<Kernel::Event> buffer_event;
     std::vector<VoiceState> voices;
-    std::unique_ptr<AudioCore::AudioOut> audio_core;
+    std::unique_ptr<AudioCore::AudioOut> audio_out;
     AudioCore::StreamPtr stream;
 };
 

src/audio_core/stream.cpp

@@ -10,6 +10,7 @@
 #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"
@@ -94,7 +95,10 @@ 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/CMakeLists.txt

@@ -1,13 +1,16 @@
 # Generate cpp with Git revision from template
-# Also if this is a CI build, add the build name (ie: Nightly, Bleeding Edge) to the scm_rev file as well
+# Also if this is a CI build, add the build name (ie: Nightly, Canary) to the scm_rev file as well
 set(REPO_NAME "")
+set(BUILD_VERSION "0")
 if ($ENV{CI})
   if ($ENV{TRAVIS})
     set(BUILD_REPOSITORY $ENV{TRAVIS_REPO_SLUG})
+    set(BUILD_TAG $ENV{TRAVIS_TAG})
   elseif($ENV{APPVEYOR})
     set(BUILD_REPOSITORY $ENV{APPVEYOR_REPO_NAME})
+    set(BUILD_TAG $ENV{APPVEYOR_REPO_TAG_NAME})
   endif()
-  # regex capture the string nightly or bleeding-edge into CMAKE_MATCH_1
+  # regex capture the string nightly or canary into CMAKE_MATCH_1
   string(REGEX MATCH "yuzu-emu/yuzu-?(.*)" OUTVAR ${BUILD_REPOSITORY})
   if (${CMAKE_MATCH_COUNT} GREATER 0)
     # capitalize the first letter of each word in the repo name.
@@ -16,10 +19,21 @@ if ($ENV{CI})
       string(SUBSTRING ${WORD} 0 1 FIRST_LETTER)
       string(SUBSTRING ${WORD} 1 -1 REMAINDER)
       string(TOUPPER ${FIRST_LETTER} FIRST_LETTER)
-      # this leaves a trailing space on the last word, but we actually want that
-      # because of how it's styled in the title bar.
-      set(REPO_NAME "${REPO_NAME}${FIRST_LETTER}${REMAINDER} ")
+      set(REPO_NAME "${REPO_NAME}${FIRST_LETTER}${REMAINDER}")
     endforeach()
+    if (BUILD_TAG)
+      string(REGEX MATCH "${CMAKE_MATCH_1}-([0-9]+)" OUTVAR ${BUILD_TAG})
+      if (${CMAKE_MATCH_COUNT} GREATER 0)
+        set(BUILD_VERSION ${CMAKE_MATCH_1})
+      endif()
+      if (BUILD_VERSION)
+        # This leaves a trailing space on the last word, but we actually want that
+        # because of how it's styled in the title bar.
+        set(BUILD_FULLNAME "${REPO_NAME} #${BUILD_VERSION} ")
+      else()
+        set(BUILD_FULLNAME "")
+      endif()
+    endif()
   endif()
 endif()
 configure_file("${CMAKE_CURRENT_SOURCE_DIR}/scm_rev.cpp.in" "${CMAKE_CURRENT_SOURCE_DIR}/scm_rev.cpp" @ONLY)

src/common/logging/filter.h

@@ -19,7 +19,7 @@ namespace Log {
 class Filter {
 public:
     /// Initializes the filter with all classes having `default_level` as the minimum level.
-    Filter(Level default_level = Level::Info);
+    explicit Filter(Level default_level = Level::Info);
 
     /// Resets the filter so that all classes have `level` as the minimum displayed level.
     void ResetAll(Level level);
@@ -49,6 +49,6 @@ public:
     bool IsDebug() const;
 
 private:
-    std::array<Level, (size_t)Class::Count> class_levels;
+    std::array<Level, static_cast<size_t>(Class::Count)> class_levels;
 };
 } // namespace Log

src/common/logging/log.h

@@ -12,14 +12,14 @@ namespace Log {
 /// Specifies the severity or level of detail of the log message.
 enum class Level : u8 {
     Trace,    ///< Extremely detailed and repetitive debugging information that is likely to
-              ///  pollute logs.
+              ///< pollute logs.
     Debug,    ///< Less detailed debugging information.
     Info,     ///< Status information from important points during execution.
     Warning,  ///< Minor or potential problems found during execution of a task.
     Error,    ///< Major problems found during execution of a task that prevent it from being
-              ///  completed.
-    Critical, ///< Major problems during execution that threathen the stability of the entire
-              ///  application.
+              ///< completed.
+    Critical, ///< Major problems during execution that threaten the stability of the entire
+              ///< application.
 
     Count ///< Total number of logging levels
 };
@@ -49,7 +49,7 @@ enum class Class : ClassType {
     Kernel,            ///< The HLE implementation of the CTR kernel
     Kernel_SVC,        ///< Kernel system calls
     Service,           ///< HLE implementation of system services. Each major service
-                       ///  should have its own subclass.
+                       ///< should have its own subclass.
     Service_ACC,       ///< The ACC (Accounts) service
     Service_AM,        ///< The AM (Applet manager) service
     Service_AOC,       ///< The AOC (AddOn Content) service

src/common/logging/text_formatter.h

@@ -15,6 +15,6 @@ struct Entry;
 std::string FormatLogMessage(const Entry& entry);
 /// Formats and prints a log entry to stderr.
 void PrintMessage(const Entry& entry);
-/// Prints the same message as `PrintMessage`, but colored acoording to the severity level.
+/// Prints the same message as `PrintMessage`, but colored according to the severity level.
 void PrintColoredMessage(const Entry& entry);
 } // namespace Log

src/common/scm_rev.cpp.in

@@ -9,6 +9,8 @@
 #define GIT_DESC     "@GIT_DESC@"
 #define BUILD_NAME   "@REPO_NAME@"
 #define BUILD_DATE   "@BUILD_DATE@"
+#define BUILD_FULLNAME "@BUILD_FULLNAME@"
+#define BUILD_VERSION "@BUILD_VERSION@"
 
 namespace Common {
 
@@ -17,6 +19,8 @@ const char g_scm_branch[]   = GIT_BRANCH;
 const char g_scm_desc[]     = GIT_DESC;
 const char g_build_name[]   = BUILD_NAME;
 const char g_build_date[]   = BUILD_DATE;
+const char g_build_fullname[] = BUILD_FULLNAME;
+const char g_build_version[]  = BUILD_VERSION;
 
 } // namespace
 

src/common/scm_rev.h

@@ -11,5 +11,7 @@ extern const char g_scm_branch[];
 extern const char g_scm_desc[];
 extern const char g_build_name[];
 extern const char g_build_date[];
+extern const char g_build_fullname[];
+extern const char g_build_version[];
 
 } // namespace Common

src/core/CMakeLists.txt

@@ -35,8 +35,12 @@ add_library(core STATIC
     file_sys/mode.h
     file_sys/nca_metadata.cpp
     file_sys/nca_metadata.h
+    file_sys/nca_patch.cpp
+    file_sys/nca_patch.h
     file_sys/partition_filesystem.cpp
     file_sys/partition_filesystem.h
+    file_sys/patch_manager.cpp
+    file_sys/patch_manager.h
     file_sys/program_metadata.cpp
     file_sys/program_metadata.h
     file_sys/registered_cache.cpp
@@ -49,6 +53,8 @@ add_library(core STATIC
     file_sys/savedata_factory.h
     file_sys/sdmc_factory.cpp
     file_sys/sdmc_factory.h
+    file_sys/submission_package.cpp
+    file_sys/submission_package.h
     file_sys/vfs.cpp
     file_sys/vfs.h
     file_sys/vfs_concat.cpp
@@ -359,6 +365,8 @@ add_library(core STATIC
     loader/nro.h
     loader/nso.cpp
     loader/nso.h
+    loader/nsp.cpp
+    loader/nsp.h
     loader/xci.cpp
     loader/xci.h
     memory.cpp

src/core/arm/dynarmic/arm_dynarmic.cpp

@@ -7,6 +7,7 @@
 #include <dynarmic/A64/a64.h>
 #include <dynarmic/A64/config.h>
 #include "common/logging/log.h"
+#include "common/microprofile.h"
 #include "core/arm/dynarmic/arm_dynarmic.h"
 #include "core/core.h"
 #include "core/core_cpu.h"
@@ -143,7 +144,10 @@ std::unique_ptr<Dynarmic::A64::Jit> ARM_Dynarmic::MakeJit() const {
     return std::make_unique<Dynarmic::A64::Jit>(config);
 }
 
+MICROPROFILE_DEFINE(ARM_Jit_Dynarmic, "ARM JIT", "Dynarmic", MP_RGB(255, 64, 64));
+
 void ARM_Dynarmic::Run() {
+    MICROPROFILE_SCOPE(ARM_Jit_Dynarmic);
     ASSERT(Memory::GetCurrentPageTable() == current_page_table);
 
     jit->Run();

src/core/arm/unicorn/arm_unicorn.cpp

@@ -193,10 +193,10 @@ void ARM_Unicorn::Step() {
     ExecuteInstructions(1);
 }
 
-MICROPROFILE_DEFINE(ARM_Jit, "ARM JIT", "ARM JIT", MP_RGB(255, 64, 64));
+MICROPROFILE_DEFINE(ARM_Jit_Unicorn, "ARM JIT", "Unicorn", MP_RGB(255, 64, 64));
 
 void ARM_Unicorn::ExecuteInstructions(int num_instructions) {
-    MICROPROFILE_SCOPE(ARM_Jit);
+    MICROPROFILE_SCOPE(ARM_Jit_Unicorn);
     CHECKED(uc_emu_start(uc, GetPC(), 1ULL << 63, 0, num_instructions));
     CoreTiming::AddTicks(num_instructions);
     if (GDBStub::IsServerEnabled()) {

src/core/core.cpp

@@ -24,7 +24,6 @@
 #include "core/hle/kernel/scheduler.h"
 #include "core/hle/kernel/thread.h"
 #include "core/hle/service/service.h"
-#include "core/hle/service/sm/controller.h"
 #include "core/hle/service/sm/sm.h"
 #include "core/loader/loader.h"
 #include "core/perf_stats.h"
@@ -137,7 +136,7 @@ struct System::Impl {
         if (virtual_filesystem == nullptr)
             virtual_filesystem = std::make_shared<FileSys::RealVfsFilesystem>();
 
-        current_process = Kernel::Process::Create(kernel, "main");
+        kernel.MakeCurrentProcess(Kernel::Process::Create(kernel, "main"));
 
         cpu_barrier = std::make_shared<CpuBarrier>();
         cpu_exclusive_monitor = Cpu::MakeExclusiveMonitor(cpu_cores.size());
@@ -203,7 +202,7 @@ struct System::Impl {
             return init_result;
         }
 
-        const Loader::ResultStatus load_result{app_loader->Load(current_process)};
+        const Loader::ResultStatus load_result{app_loader->Load(kernel.CurrentProcess())};
         if (load_result != Loader::ResultStatus::Success) {
             LOG_CRITICAL(Core, "Failed to load ROM (Error {})!", static_cast<int>(load_result));
             Shutdown();
@@ -282,7 +281,6 @@ struct System::Impl {
     std::unique_ptr<VideoCore::RendererBase> renderer;
     std::unique_ptr<Tegra::GPU> gpu_core;
     std::shared_ptr<Tegra::DebugContext> debug_context;
-    Kernel::SharedPtr<Kernel::Process> current_process;
     std::shared_ptr<ExclusiveMonitor> cpu_exclusive_monitor;
     std::shared_ptr<CpuBarrier> cpu_barrier;
     std::array<std::shared_ptr<Cpu>, NUM_CPU_CORES> cpu_cores;
@@ -364,7 +362,11 @@ const std::shared_ptr<Kernel::Scheduler>& System::Scheduler(size_t core_index) {
 }
 
 Kernel::SharedPtr<Kernel::Process>& System::CurrentProcess() {
-    return impl->current_process;
+    return impl->kernel.CurrentProcess();
+}
+
+const Kernel::SharedPtr<Kernel::Process>& System::CurrentProcess() const {
+    return impl->kernel.CurrentProcess();
 }
 
 ARM_Interface& System::ArmInterface(size_t core_index) {
@@ -441,8 +443,8 @@ void System::SetGPUDebugContext(std::shared_ptr<Tegra::DebugContext> context) {
     impl->debug_context = std::move(context);
 }
 
-std::shared_ptr<Tegra::DebugContext> System::GetGPUDebugContext() const {
-    return impl->debug_context;
+Tegra::DebugContext* System::GetGPUDebugContext() const {
+    return impl->debug_context.get();
 }
 
 void System::SetFilesystem(std::shared_ptr<FileSys::VfsFilesystem> vfs) {

src/core/core.h

@@ -174,9 +174,12 @@ public:
     /// Gets the scheduler for the CPU core with the specified index
     const std::shared_ptr<Kernel::Scheduler>& Scheduler(size_t core_index);
 
-    /// Gets the current process
+    /// Provides a reference to the current process
     Kernel::SharedPtr<Kernel::Process>& CurrentProcess();
 
+    /// Provides a constant reference to the current process.
+    const Kernel::SharedPtr<Kernel::Process>& CurrentProcess() const;
+
     /// Provides a reference to the kernel instance.
     Kernel::KernelCore& Kernel();
 
@@ -209,7 +212,7 @@ public:
 
     void SetGPUDebugContext(std::shared_ptr<Tegra::DebugContext> context);
 
-    std::shared_ptr<Tegra::DebugContext> GetGPUDebugContext() const;
+    Tegra::DebugContext* GetGPUDebugContext() const;
 
     void SetFilesystem(std::shared_ptr<FileSys::VfsFilesystem> vfs);
 

src/core/crypto/aes_util.cpp

@@ -82,11 +82,25 @@ void AESCipher<Key, KeySize>::Transcode(const u8* src, size_t size, u8* dest, Op
         }
     } else {
         const auto block_size = mbedtls_cipher_get_block_size(context);
+        if (size < block_size) {
+            std::vector<u8> block(block_size);
+            std::memcpy(block.data(), src, size);
+            Transcode(block.data(), block.size(), block.data(), op);
+            std::memcpy(dest, block.data(), size);
+            return;
+        }
 
         for (size_t offset = 0; offset < size; offset += block_size) {
             auto length = std::min<size_t>(block_size, size - offset);
             mbedtls_cipher_update(context, src + offset, length, dest + offset, &written);
             if (written != length) {
+                if (length < block_size) {
+                    std::vector<u8> block(block_size);
+                    std::memcpy(block.data(), src + offset, length);
+                    Transcode(block.data(), block.size(), block.data(), op);
+                    std::memcpy(dest + offset, block.data(), length);
+                    return;
+                }
                 LOG_WARNING(Crypto, "Not all data was decrypted requested={:016X}, actual={:016X}.",
                             length, written);
             }

src/core/crypto/ctr_encryption_layer.cpp

@@ -21,7 +21,7 @@ size_t CTREncryptionLayer::Read(u8* data, size_t length, size_t offset) const {
         UpdateIV(base_offset + offset);
         std::vector<u8> raw = base->ReadBytes(length, offset);
         cipher.Transcode(raw.data(), raw.size(), data, Op::Decrypt);
-        return raw.size();
+        return length;
     }
 
     // offset does not fall on block boundary (0x10)

src/core/crypto/key_manager.cpp

@@ -8,12 +8,15 @@
 #include <locale>
 #include <sstream>
 #include <string_view>
+#include <tuple>
+#include <vector>
 #include "common/common_paths.h"
 #include "common/file_util.h"
 #include "common/hex_util.h"
 #include "common/logging/log.h"
 #include "core/crypto/aes_util.h"
 #include "core/crypto/key_manager.h"
+#include "core/loader/loader.h"
 #include "core/settings.h"
 
 namespace Core::Crypto {
@@ -228,18 +231,28 @@ void KeyManager::WriteKeyToFile(bool title_key, std::string_view keyname,
 }
 
 void KeyManager::SetKey(S128KeyType id, Key128 key, u64 field1, u64 field2) {
-    const auto iter = std::find_if(
+    if (s128_keys.find({id, field1, field2}) != s128_keys.end())
+        return;
+    if (id == S128KeyType::Titlekey) {
+        Key128 rights_id;
+        std::memcpy(rights_id.data(), &field2, sizeof(u64));
+        std::memcpy(rights_id.data() + sizeof(u64), &field1, sizeof(u64));
+        WriteKeyToFile(true, Common::HexArrayToString(rights_id), key);
+    }
+    const auto iter2 = std::find_if(
         s128_file_id.begin(), s128_file_id.end(),
         [&id, &field1, &field2](const std::pair<std::string, KeyIndex<S128KeyType>> elem) {
             return std::tie(elem.second.type, elem.second.field1, elem.second.field2) ==
                    std::tie(id, field1, field2);
         });
-    if (iter != s128_file_id.end())
-        WriteKeyToFile(id == S128KeyType::Titlekey, iter->first, key);
+    if (iter2 != s128_file_id.end())
+        WriteKeyToFile(false, iter2->first, key);
     s128_keys[{id, field1, field2}] = key;
 }
 
 void KeyManager::SetKey(S256KeyType id, Key256 key, u64 field1, u64 field2) {
+    if (s256_keys.find({id, field1, field2}) != s256_keys.end())
+        return;
     const auto iter = std::find_if(
         s256_file_id.begin(), s256_file_id.end(),
         [&id, &field1, &field2](const std::pair<std::string, KeyIndex<S256KeyType>> elem) {

src/core/crypto/key_manager.h

@@ -6,16 +6,19 @@
 
 #include <array>
 #include <string>
-#include <string_view>
-#include <type_traits>
-#include <vector>
 #include <boost/container/flat_map.hpp>
+#include <boost/optional.hpp>
 #include <fmt/format.h>
 #include "common/common_types.h"
-#include "core/loader/loader.h"
+
+namespace Loader {
+enum class ResultStatus : u16;
+}
 
 namespace Core::Crypto {
 
+constexpr u64 TICKET_FILE_TITLEKEY_OFFSET = 0x180;
+
 using Key128 = std::array<u8, 0x10>;
 using Key256 = std::array<u8, 0x20>;
 using SHA256Hash = std::array<u8, 0x20>;

src/core/file_sys/bis_factory.cpp

@@ -3,6 +3,7 @@
 // Refer to the license.txt file included.
 
 #include "core/file_sys/bis_factory.h"
+#include "core/file_sys/registered_cache.h"
 
 namespace FileSys {
 
@@ -13,6 +14,8 @@ BISFactory::BISFactory(VirtualDir nand_root_)
       usrnand_cache(std::make_shared<RegisteredCache>(
           GetOrCreateDirectoryRelative(nand_root, "/user/Contents/registered"))) {}
 
+BISFactory::~BISFactory() = default;
+
 std::shared_ptr<RegisteredCache> BISFactory::GetSystemNANDContents() const {
     return sysnand_cache;
 }

src/core/file_sys/bis_factory.h

@@ -5,17 +5,20 @@
 #pragma once
 
 #include <memory>
-#include "core/loader/loader.h"
-#include "registered_cache.h"
+
+#include "core/file_sys/vfs.h"
 
 namespace FileSys {
 
+class RegisteredCache;
+
 /// File system interface to the Built-In Storage
 /// This is currently missing accessors to BIS partitions, but seemed like a good place for the NAND
 /// registered caches.
 class BISFactory {
 public:
     explicit BISFactory(VirtualDir nand_root);
+    ~BISFactory();
 
     std::shared_ptr<RegisteredCache> GetSystemNANDContents() const;
     std::shared_ptr<RegisteredCache> GetUserNANDContents() const;

src/core/file_sys/card_image.cpp

@@ -9,7 +9,10 @@
 
 #include "common/logging/log.h"
 #include "core/file_sys/card_image.h"
+#include "core/file_sys/content_archive.h"
+#include "core/file_sys/nca_metadata.h"
 #include "core/file_sys/partition_filesystem.h"
+#include "core/file_sys/submission_package.h"
 #include "core/file_sys/vfs_offset.h"
 #include "core/loader/loader.h"
 
@@ -43,15 +46,19 @@ XCI::XCI(VirtualFile file_) : file(std::move(file_)), partitions(0x4) {
             partitions[static_cast<size_t>(partition)] = std::make_shared<PartitionFilesystem>(raw);
     }
 
-    program_nca_status = Loader::ResultStatus::ErrorXCIMissingProgramNCA;
+    secure_partition = std::make_shared<NSP>(
+        main_hfs.GetFile(partition_names[static_cast<size_t>(XCIPartition::Secure)]));
 
-    auto result = AddNCAFromPartition(XCIPartition::Secure);
-    if (result != Loader::ResultStatus::Success) {
-        status = result;
-        return;
-    }
+    const auto secure_ncas = secure_partition->GetNCAsCollapsed();
+    std::copy(secure_ncas.begin(), secure_ncas.end(), std::back_inserter(ncas));
 
-    result = AddNCAFromPartition(XCIPartition::Update);
+    program =
+        secure_partition->GetNCA(secure_partition->GetProgramTitleID(), ContentRecordType::Program);
+    program_nca_status = secure_partition->GetProgramStatus(secure_partition->GetProgramTitleID());
+    if (program_nca_status == Loader::ResultStatus::ErrorNSPMissingProgramNCA)
+        program_nca_status = Loader::ResultStatus::ErrorXCIMissingProgramNCA;
+
+    auto result = AddNCAFromPartition(XCIPartition::Update);
     if (result != Loader::ResultStatus::Success) {
         status = result;
         return;
@@ -74,6 +81,8 @@ XCI::XCI(VirtualFile file_) : file(std::move(file_)), partitions(0x4) {
     status = Loader::ResultStatus::Success;
 }
 
+XCI::~XCI() = default;
+
 Loader::ResultStatus XCI::GetStatus() const {
     return status;
 }
@@ -86,6 +95,10 @@ VirtualDir XCI::GetPartition(XCIPartition partition) const {
     return partitions[static_cast<size_t>(partition)];
 }
 
+std::shared_ptr<NSP> XCI::GetSecurePartitionNSP() const {
+    return secure_partition;
+}
+
 VirtualDir XCI::GetSecurePartition() const {
     return GetPartition(XCIPartition::Secure);
 }
@@ -102,6 +115,20 @@ VirtualDir XCI::GetLogoPartition() const {
     return GetPartition(XCIPartition::Logo);
 }
 
+u64 XCI::GetProgramTitleID() const {
+    return secure_partition->GetProgramTitleID();
+}
+
+std::shared_ptr<NCA> XCI::GetProgramNCA() const {
+    return program;
+}
+
+VirtualFile XCI::GetProgramNCAFile() const {
+    if (GetProgramNCA() == nullptr)
+        return nullptr;
+    return GetProgramNCA()->GetBaseFile();
+}
+
 const std::vector<std::shared_ptr<NCA>>& XCI::GetNCAs() const {
     return ncas;
 }

src/core/file_sys/card_image.h

@@ -5,15 +5,22 @@
 #pragma once
 
 #include <array>
+#include <memory>
 #include <vector>
 #include "common/common_types.h"
 #include "common/swap.h"
-#include "core/file_sys/content_archive.h"
 #include "core/file_sys/vfs.h"
-#include "core/loader/loader.h"
+
+namespace Loader {
+enum class ResultStatus : u16;
+}
 
 namespace FileSys {
 
+class NCA;
+enum class NCAContentType : u8;
+class NSP;
+
 enum class GamecardSize : u8 {
     S_1GB = 0xFA,
     S_2GB = 0xF8,
@@ -57,6 +64,7 @@ enum class XCIPartition : u8 { Update, Normal, Secure, Logo };
 class XCI : public ReadOnlyVfsDirectory {
 public:
     explicit XCI(VirtualFile file);
+    ~XCI() override;
 
     Loader::ResultStatus GetStatus() const;
     Loader::ResultStatus GetProgramNCAStatus() const;
@@ -64,11 +72,16 @@ public:
     u8 GetFormatVersion() const;
 
     VirtualDir GetPartition(XCIPartition partition) const;
+    std::shared_ptr<NSP> GetSecurePartitionNSP() const;
     VirtualDir GetSecurePartition() const;
     VirtualDir GetNormalPartition() const;
     VirtualDir GetUpdatePartition() const;
     VirtualDir GetLogoPartition() const;
 
+    u64 GetProgramTitleID() const;
+
+    std::shared_ptr<NCA> GetProgramNCA() const;
+    VirtualFile GetProgramNCAFile() const;
     const std::vector<std::shared_ptr<NCA>>& GetNCAs() const;
     std::shared_ptr<NCA> GetNCAByType(NCAContentType type) const;
     VirtualFile GetNCAFileByType(NCAContentType type) const;
@@ -94,6 +107,8 @@ private:
     Loader::ResultStatus program_nca_status;
 
     std::vector<VirtualDir> partitions;
+    std::shared_ptr<NSP> secure_partition;
+    std::shared_ptr<NCA> program;
     std::vector<std::shared_ptr<NCA>> ncas;
 };
 } // namespace FileSys

src/core/file_sys/content_archive.cpp

@@ -3,12 +3,17 @@
 // Refer to the license.txt file included.
 
 #include <algorithm>
+#include <cstring>
 #include <utility>
+
 #include <boost/optional.hpp>
+
 #include "common/logging/log.h"
 #include "core/crypto/aes_util.h"
 #include "core/crypto/ctr_encryption_layer.h"
 #include "core/file_sys/content_archive.h"
+#include "core/file_sys/nca_patch.h"
+#include "core/file_sys/partition_filesystem.h"
 #include "core/file_sys/romfs.h"
 #include "core/file_sys/vfs_offset.h"
 #include "core/loader/loader.h"
@@ -64,10 +69,31 @@ struct RomFSSuperblock {
 };
 static_assert(sizeof(RomFSSuperblock) == 0x200, "RomFSSuperblock has incorrect size.");
 
+struct BKTRHeader {
+    u64_le offset;
+    u64_le size;
+    u32_le magic;
+    INSERT_PADDING_BYTES(0x4);
+    u32_le number_entries;
+    INSERT_PADDING_BYTES(0x4);
+};
+static_assert(sizeof(BKTRHeader) == 0x20, "BKTRHeader has incorrect size.");
+
+struct BKTRSuperblock {
+    NCASectionHeaderBlock header_block;
+    IVFCHeader ivfc;
+    INSERT_PADDING_BYTES(0x18);
+    BKTRHeader relocation;
+    BKTRHeader subsection;
+    INSERT_PADDING_BYTES(0xC0);
+};
+static_assert(sizeof(BKTRSuperblock) == 0x200, "BKTRSuperblock has incorrect size.");
+
 union NCASectionHeader {
     NCASectionRaw raw;
     PFS0Superblock pfs0;
     RomFSSuperblock romfs;
+    BKTRSuperblock bktr;
 };
 static_assert(sizeof(NCASectionHeader) == 0x200, "NCASectionHeader has incorrect size.");
 
@@ -100,7 +126,7 @@ boost::optional<Core::Crypto::Key128> NCA::GetKeyAreaKey(NCASectionCryptoType ty
     Core::Crypto::Key128 out;
     if (type == NCASectionCryptoType::XTS)
         std::copy(key_area.begin(), key_area.begin() + 0x10, out.begin());
-    else if (type == NCASectionCryptoType::CTR)
+    else if (type == NCASectionCryptoType::CTR || type == NCASectionCryptoType::BKTR)
         std::copy(key_area.begin() + 0x20, key_area.begin() + 0x30, out.begin());
     else
         LOG_CRITICAL(Crypto, "Called GetKeyAreaKey on invalid NCASectionCryptoType type={:02X}",
@@ -150,6 +176,9 @@ VirtualFile NCA::Decrypt(NCASectionHeader s_header, VirtualFile in, u64 starting
         LOG_DEBUG(Crypto, "called with mode=NONE");
         return in;
     case NCASectionCryptoType::CTR:
+    // During normal BKTR decryption, this entire function is skipped. This is for the metadata,
+    // which uses the same CTR as usual.
+    case NCASectionCryptoType::BKTR:
         LOG_DEBUG(Crypto, "called with mode=CTR, starting_offset={:016X}", starting_offset);
         {
             boost::optional<Core::Crypto::Key128> key = boost::none;
@@ -186,7 +215,9 @@ VirtualFile NCA::Decrypt(NCASectionHeader s_header, VirtualFile in, u64 starting
     }
 }
 
-NCA::NCA(VirtualFile file_) : file(std::move(file_)) {
+NCA::NCA(VirtualFile file_, VirtualFile bktr_base_romfs_, u64 bktr_base_ivfc_offset)
+    : file(std::move(file_)),
+      bktr_base_romfs(bktr_base_romfs_ ? std::move(bktr_base_romfs_) : nullptr) {
     status = Loader::ResultStatus::Success;
 
     if (file == nullptr) {
@@ -261,22 +292,21 @@ NCA::NCA(VirtualFile file_) : file(std::move(file_)) {
     is_update = std::find_if(sections.begin(), sections.end(), [](const NCASectionHeader& header) {
                     return header.raw.header.crypto_type == NCASectionCryptoType::BKTR;
                 }) != sections.end();
+    ivfc_offset = 0;
 
     for (std::ptrdiff_t i = 0; i < number_sections; ++i) {
         auto section = sections[i];
 
         if (section.raw.header.filesystem_type == NCASectionFilesystemType::ROMFS) {
-            const size_t romfs_offset =
-                header.section_tables[i].media_offset * MEDIA_OFFSET_MULTIPLIER +
-                section.romfs.ivfc.levels[IVFC_MAX_LEVEL - 1].offset;
+            const size_t base_offset =
+                header.section_tables[i].media_offset * MEDIA_OFFSET_MULTIPLIER;
+            ivfc_offset = section.romfs.ivfc.levels[IVFC_MAX_LEVEL - 1].offset;
+            const size_t romfs_offset = base_offset + ivfc_offset;
             const size_t romfs_size = section.romfs.ivfc.levels[IVFC_MAX_LEVEL - 1].size;
-            auto dec =
-                Decrypt(section, std::make_shared<OffsetVfsFile>(file, romfs_size, romfs_offset),
-                        romfs_offset);
-            if (dec != nullptr) {
-                files.push_back(std::move(dec));
-                romfs = files.back();
-            } else {
+            auto raw = std::make_shared<OffsetVfsFile>(file, romfs_size, romfs_offset);
+            auto dec = Decrypt(section, raw, romfs_offset);
+
+            if (dec == nullptr) {
                 if (status != Loader::ResultStatus::Success)
                     return;
                 if (has_rights_id)
@@ -285,6 +315,117 @@ NCA::NCA(VirtualFile file_) : file(std::move(file_)) {
                     status = Loader::ResultStatus::ErrorIncorrectKeyAreaKey;
                 return;
             }
+
+            if (section.raw.header.crypto_type == NCASectionCryptoType::BKTR) {
+                if (section.bktr.relocation.magic != Common::MakeMagic('B', 'K', 'T', 'R') ||
+                    section.bktr.subsection.magic != Common::MakeMagic('B', 'K', 'T', 'R')) {
+                    status = Loader::ResultStatus::ErrorBadBKTRHeader;
+                    return;
+                }
+
+                if (section.bktr.relocation.offset + section.bktr.relocation.size !=
+                    section.bktr.subsection.offset) {
+                    status = Loader::ResultStatus::ErrorBKTRSubsectionNotAfterRelocation;
+                    return;
+                }
+
+                const u64 size =
+                    MEDIA_OFFSET_MULTIPLIER * (header.section_tables[i].media_end_offset -
+                                               header.section_tables[i].media_offset);
+                if (section.bktr.subsection.offset + section.bktr.subsection.size != size) {
+                    status = Loader::ResultStatus::ErrorBKTRSubsectionNotAtEnd;
+                    return;
+                }
+
+                const u64 offset = section.romfs.ivfc.levels[IVFC_MAX_LEVEL - 1].offset;
+                RelocationBlock relocation_block{};
+                if (dec->ReadObject(&relocation_block, section.bktr.relocation.offset - offset) !=
+                    sizeof(RelocationBlock)) {
+                    status = Loader::ResultStatus::ErrorBadRelocationBlock;
+                    return;
+                }
+                SubsectionBlock subsection_block{};
+                if (dec->ReadObject(&subsection_block, section.bktr.subsection.offset - offset) !=
+                    sizeof(RelocationBlock)) {
+                    status = Loader::ResultStatus::ErrorBadSubsectionBlock;
+                    return;
+                }
+
+                std::vector<RelocationBucketRaw> relocation_buckets_raw(
+                    (section.bktr.relocation.size - sizeof(RelocationBlock)) /
+                    sizeof(RelocationBucketRaw));
+                if (dec->ReadBytes(relocation_buckets_raw.data(),
+                                   section.bktr.relocation.size - sizeof(RelocationBlock),
+                                   section.bktr.relocation.offset + sizeof(RelocationBlock) -
+                                       offset) !=
+                    section.bktr.relocation.size - sizeof(RelocationBlock)) {
+                    status = Loader::ResultStatus::ErrorBadRelocationBuckets;
+                    return;
+                }
+
+                std::vector<SubsectionBucketRaw> subsection_buckets_raw(
+                    (section.bktr.subsection.size - sizeof(SubsectionBlock)) /
+                    sizeof(SubsectionBucketRaw));
+                if (dec->ReadBytes(subsection_buckets_raw.data(),
+                                   section.bktr.subsection.size - sizeof(SubsectionBlock),
+                                   section.bktr.subsection.offset + sizeof(SubsectionBlock) -
+                                       offset) !=
+                    section.bktr.subsection.size - sizeof(SubsectionBlock)) {
+                    status = Loader::ResultStatus::ErrorBadSubsectionBuckets;
+                    return;
+                }
+
+                std::vector<RelocationBucket> relocation_buckets(relocation_buckets_raw.size());
+                std::transform(relocation_buckets_raw.begin(), relocation_buckets_raw.end(),
+                               relocation_buckets.begin(), &ConvertRelocationBucketRaw);
+                std::vector<SubsectionBucket> subsection_buckets(subsection_buckets_raw.size());
+                std::transform(subsection_buckets_raw.begin(), subsection_buckets_raw.end(),
+                               subsection_buckets.begin(), &ConvertSubsectionBucketRaw);
+
+                u32 ctr_low;
+                std::memcpy(&ctr_low, section.raw.section_ctr.data(), sizeof(ctr_low));
+                subsection_buckets.back().entries.push_back(
+                    {section.bktr.relocation.offset, {0}, ctr_low});
+                subsection_buckets.back().entries.push_back({size, {0}, 0});
+
+                boost::optional<Core::Crypto::Key128> key = boost::none;
+                if (encrypted) {
+                    if (has_rights_id) {
+                        status = Loader::ResultStatus::Success;
+                        key = GetTitlekey();
+                        if (key == boost::none) {
+                            status = Loader::ResultStatus::ErrorMissingTitlekey;
+                            return;
+                        }
+                    } else {
+                        key = GetKeyAreaKey(NCASectionCryptoType::BKTR);
+                        if (key == boost::none) {
+                            status = Loader::ResultStatus::ErrorMissingKeyAreaKey;
+                            return;
+                        }
+                    }
+                }
+
+                if (bktr_base_romfs == nullptr) {
+                    status = Loader::ResultStatus::ErrorMissingBKTRBaseRomFS;
+                    return;
+                }
+
+                auto bktr = std::make_shared<BKTR>(
+                    bktr_base_romfs, std::make_shared<OffsetVfsFile>(file, romfs_size, base_offset),
+                    relocation_block, relocation_buckets, subsection_block, subsection_buckets,
+                    encrypted, encrypted ? key.get() : Core::Crypto::Key128{}, base_offset,
+                    bktr_base_ivfc_offset, section.raw.section_ctr);
+
+                // BKTR applies to entire IVFC, so make an offset version to level 6
+
+                files.push_back(std::make_shared<OffsetVfsFile>(
+                    bktr, romfs_size, section.romfs.ivfc.levels[IVFC_MAX_LEVEL - 1].offset));
+                romfs = files.back();
+            } else {
+                files.push_back(std::move(dec));
+                romfs = files.back();
+            }
         } else if (section.raw.header.filesystem_type == NCASectionFilesystemType::PFS0) {
             u64 offset = (static_cast<u64>(header.section_tables[i].media_offset) *
                           MEDIA_OFFSET_MULTIPLIER) +
@@ -300,6 +441,12 @@ NCA::NCA(VirtualFile file_) : file(std::move(file_)) {
                     dirs.push_back(std::move(npfs));
                     if (IsDirectoryExeFS(dirs.back()))
                         exefs = dirs.back();
+                } else {
+                    if (has_rights_id)
+                        status = Loader::ResultStatus::ErrorIncorrectTitlekeyOrTitlekek;
+                    else
+                        status = Loader::ResultStatus::ErrorIncorrectKeyAreaKey;
+                    return;
                 }
             } else {
                 if (status != Loader::ResultStatus::Success)
@@ -345,11 +492,15 @@ NCAContentType NCA::GetType() const {
 }
 
 u64 NCA::GetTitleId() const {
-    if (status != Loader::ResultStatus::Success)
-        return {};
+    if (is_update || status == Loader::ResultStatus::ErrorMissingBKTRBaseRomFS)
+        return header.title_id | 0x800;
     return header.title_id;
 }
 
+bool NCA::IsUpdate() const {
+    return is_update;
+}
+
 VirtualFile NCA::GetRomFS() const {
     return romfs;
 }
@@ -362,8 +513,8 @@ VirtualFile NCA::GetBaseFile() const {
     return file;
 }
 
-bool NCA::IsUpdate() const {
-    return is_update;
+u64 NCA::GetBaseIVFCOffset() const {
+    return ivfc_offset;
 }
 
 bool NCA::ReplaceFileWithSubdirectory(VirtualFile file, VirtualDir dir) {

src/core/file_sys/content_archive.h

@@ -12,10 +12,12 @@
 #include "common/common_funcs.h"
 #include "common/common_types.h"
 #include "common/swap.h"
-#include "control_metadata.h"
 #include "core/crypto/key_manager.h"
-#include "core/file_sys/partition_filesystem.h"
-#include "core/loader/loader.h"
+#include "core/file_sys/vfs.h"
+
+namespace Loader {
+enum class ResultStatus : u16;
+}
 
 namespace FileSys {
 
@@ -77,7 +79,8 @@ bool IsValidNCA(const NCAHeader& header);
 // After construction, use GetStatus to determine if the file is valid and ready to be used.
 class NCA : public ReadOnlyVfsDirectory {
 public:
-    explicit NCA(VirtualFile file);
+    explicit NCA(VirtualFile file, VirtualFile bktr_base_romfs = nullptr,
+                 u64 bktr_base_ivfc_offset = 0);
     Loader::ResultStatus GetStatus() const;
 
     std::vector<std::shared_ptr<VfsFile>> GetFiles() const override;
@@ -87,13 +90,15 @@ public:
 
     NCAContentType GetType() const;
     u64 GetTitleId() const;
+    bool IsUpdate() const;
 
     VirtualFile GetRomFS() const;
     VirtualDir GetExeFS() const;
 
     VirtualFile GetBaseFile() const;
 
-    bool IsUpdate() const;
+    // Returns the base ivfc offset used in BKTR patching.
+    u64 GetBaseIVFCOffset() const;
 
 protected:
     bool ReplaceFileWithSubdirectory(VirtualFile file, VirtualDir dir) override;
@@ -110,14 +115,16 @@ private:
     VirtualFile romfs = nullptr;
     VirtualDir exefs = nullptr;
     VirtualFile file;
+    VirtualFile bktr_base_romfs;
+    u64 ivfc_offset;
 
     NCAHeader header{};
     bool has_rights_id{};
-    bool is_update{};
 
     Loader::ResultStatus status{};
 
     bool encrypted;
+    bool is_update;
 
     Core::Crypto::KeyManager keys;
 };

src/core/file_sys/control_metadata.cpp

@@ -21,7 +21,17 @@ NACP::NACP(VirtualFile file) : raw(std::make_unique<RawNACP>()) {
 }
 
 const LanguageEntry& NACP::GetLanguageEntry(Language language) const {
-    return raw->language_entries.at(static_cast<u8>(language));
+    if (language != Language::Default) {
+        return raw->language_entries.at(static_cast<u8>(language));
+    } else {
+        for (const auto& language_entry : raw->language_entries) {
+            if (!language_entry.GetApplicationName().empty())
+                return language_entry;
+        }
+
+        // Fallback to English
+        return GetLanguageEntry(Language::AmericanEnglish);
+    }
 }
 
 std::string NACP::GetApplicationName(Language language) const {

src/core/file_sys/control_metadata.h

@@ -8,6 +8,8 @@
 #include <memory>
 #include <string>
 #include "common/common_funcs.h"
+#include "common/common_types.h"
+#include "common/swap.h"
 #include "core/file_sys/vfs.h"
 
 namespace FileSys {
@@ -60,6 +62,8 @@ enum class Language : u8 {
     Korean = 12,
     Taiwanese = 13,
     Chinese = 14,
+
+    Default = 255,
 };
 
 static constexpr std::array<const char*, 15> LANGUAGE_NAMES = {
@@ -74,9 +78,9 @@ static constexpr std::array<const char*, 15> LANGUAGE_NAMES = {
 class NACP {
 public:
     explicit NACP(VirtualFile file);
-    const LanguageEntry& GetLanguageEntry(Language language = Language::AmericanEnglish) const;
-    std::string GetApplicationName(Language language = Language::AmericanEnglish) const;
-    std::string GetDeveloperName(Language language = Language::AmericanEnglish) const;
+    const LanguageEntry& GetLanguageEntry(Language language = Language::Default) const;
+    std::string GetApplicationName(Language language = Language::Default) const;
+    std::string GetDeveloperName(Language language = Language::Default) const;
     u64 GetTitleId() const;
     std::string GetVersionString() const;
 

src/core/file_sys/nca_metadata.cpp

@@ -3,10 +3,9 @@
 // Refer to the license.txt file included.
 
 #include <cstring>
-#include "common/common_funcs.h"
+#include "common/common_types.h"
 #include "common/logging/log.h"
 #include "common/swap.h"
-#include "content_archive.h"
 #include "core/file_sys/nca_metadata.h"
 
 namespace FileSys {

src/core/file_sys/nca_metadata.h

@@ -4,7 +4,6 @@
 
 #pragma once
 
-#include <cstring>
 #include <memory>
 #include <vector>
 #include "common/common_funcs.h"

src/core/file_sys/nca_patch.cpp

@@ -0,0 +1,210 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <algorithm>
+#include <cstddef>
+#include <cstring>
+
+#include "common/assert.h"
+#include "core/crypto/aes_util.h"
+#include "core/file_sys/nca_patch.h"
+
+namespace FileSys {
+
+BKTR::BKTR(VirtualFile base_romfs_, VirtualFile bktr_romfs_, RelocationBlock relocation_,
+           std::vector<RelocationBucket> relocation_buckets_, SubsectionBlock subsection_,
+           std::vector<SubsectionBucket> subsection_buckets_, bool is_encrypted_,
+           Core::Crypto::Key128 key_, u64 base_offset_, u64 ivfc_offset_,
+           std::array<u8, 8> section_ctr_)
+    : relocation(relocation_), relocation_buckets(std::move(relocation_buckets_)),
+      subsection(subsection_), subsection_buckets(std::move(subsection_buckets_)),
+      base_romfs(std::move(base_romfs_)), bktr_romfs(std::move(bktr_romfs_)),
+      encrypted(is_encrypted_), key(key_), base_offset(base_offset_), ivfc_offset(ivfc_offset_),
+      section_ctr(section_ctr_) {
+    for (size_t i = 0; i < relocation.number_buckets - 1; ++i) {
+        relocation_buckets[i].entries.push_back({relocation.base_offsets[i + 1], 0, 0});
+    }
+
+    for (size_t i = 0; i < subsection.number_buckets - 1; ++i) {
+        subsection_buckets[i].entries.push_back({subsection_buckets[i + 1].entries[0].address_patch,
+                                                 {0},
+                                                 subsection_buckets[i + 1].entries[0].ctr});
+    }
+
+    relocation_buckets.back().entries.push_back({relocation.size, 0, 0});
+}
+
+BKTR::~BKTR() = default;
+
+size_t BKTR::Read(u8* data, size_t length, size_t offset) const {
+    // Read out of bounds.
+    if (offset >= relocation.size)
+        return 0;
+    const auto relocation = GetRelocationEntry(offset);
+    const auto section_offset = offset - relocation.address_patch + relocation.address_source;
+    const auto bktr_read = relocation.from_patch;
+
+    const auto next_relocation = GetNextRelocationEntry(offset);
+
+    if (offset + length > next_relocation.address_patch) {
+        const u64 partition = next_relocation.address_patch - offset;
+        return Read(data, partition, offset) +
+               Read(data + partition, length - partition, offset + partition);
+    }
+
+    if (!bktr_read) {
+        ASSERT_MSG(section_offset >= ivfc_offset, "Offset calculation negative.");
+        return base_romfs->Read(data, length, section_offset - ivfc_offset);
+    }
+
+    if (!encrypted) {
+        return bktr_romfs->Read(data, length, section_offset);
+    }
+
+    const auto subsection = GetSubsectionEntry(section_offset);
+    Core::Crypto::AESCipher<Core::Crypto::Key128> cipher(key, Core::Crypto::Mode::CTR);
+
+    // Calculate AES IV
+    std::vector<u8> iv(16);
+    auto subsection_ctr = subsection.ctr;
+    auto offset_iv = section_offset + base_offset;
+    for (size_t i = 0; i < section_ctr.size(); ++i)
+        iv[i] = section_ctr[0x8 - i - 1];
+    offset_iv >>= 4;
+    for (size_t i = 0; i < sizeof(u64); ++i) {
+        iv[0xF - i] = static_cast<u8>(offset_iv & 0xFF);
+        offset_iv >>= 8;
+    }
+    for (size_t i = 0; i < sizeof(u32); ++i) {
+        iv[0x7 - i] = static_cast<u8>(subsection_ctr & 0xFF);
+        subsection_ctr >>= 8;
+    }
+    cipher.SetIV(iv);
+
+    const auto next_subsection = GetNextSubsectionEntry(section_offset);
+
+    if (section_offset + length > next_subsection.address_patch) {
+        const u64 partition = next_subsection.address_patch - section_offset;
+        return Read(data, partition, offset) +
+               Read(data + partition, length - partition, offset + partition);
+    }
+
+    const auto block_offset = section_offset & 0xF;
+    if (block_offset != 0) {
+        auto block = bktr_romfs->ReadBytes(0x10, section_offset & ~0xF);
+        cipher.Transcode(block.data(), block.size(), block.data(), Core::Crypto::Op::Decrypt);
+        if (length + block_offset < 0x10) {
+            std::memcpy(data, block.data() + block_offset, std::min(length, block.size()));
+            return std::min(length, block.size());
+        }
+
+        const auto read = 0x10 - block_offset;
+        std::memcpy(data, block.data() + block_offset, read);
+        return read + Read(data + read, length - read, offset + read);
+    }
+
+    const auto raw_read = bktr_romfs->Read(data, length, section_offset);
+    cipher.Transcode(data, raw_read, data, Core::Crypto::Op::Decrypt);
+    return raw_read;
+}
+
+template <bool Subsection, typename BlockType, typename BucketType>
+std::pair<size_t, size_t> BKTR::SearchBucketEntry(u64 offset, BlockType block,
+                                                  BucketType buckets) const {
+    if constexpr (Subsection) {
+        const auto last_bucket = buckets[block.number_buckets - 1];
+        if (offset >= last_bucket.entries[last_bucket.number_entries].address_patch)
+            return {block.number_buckets - 1, last_bucket.number_entries};
+    } else {
+        ASSERT_MSG(offset <= block.size, "Offset is out of bounds in BKTR relocation block.");
+    }
+
+    size_t bucket_id = std::count_if(block.base_offsets.begin() + 1,
+                                     block.base_offsets.begin() + block.number_buckets,
+                                     [&offset](u64 base_offset) { return base_offset <= offset; });
+
+    const auto bucket = buckets[bucket_id];
+
+    if (bucket.number_entries == 1)
+        return {bucket_id, 0};
+
+    size_t low = 0;
+    size_t mid = 0;
+    size_t high = bucket.number_entries - 1;
+    while (low <= high) {
+        mid = (low + high) / 2;
+        if (bucket.entries[mid].address_patch > offset) {
+            high = mid - 1;
+        } else {
+            if (mid == bucket.number_entries - 1 ||
+                bucket.entries[mid + 1].address_patch > offset) {
+                return {bucket_id, mid};
+            }
+
+            low = mid + 1;
+        }
+    }
+
+    UNREACHABLE_MSG("Offset could not be found in BKTR block.");
+}
+
+RelocationEntry BKTR::GetRelocationEntry(u64 offset) const {
+    const auto res = SearchBucketEntry<false>(offset, relocation, relocation_buckets);
+    return relocation_buckets[res.first].entries[res.second];
+}
+
+RelocationEntry BKTR::GetNextRelocationEntry(u64 offset) const {
+    const auto res = SearchBucketEntry<false>(offset, relocation, relocation_buckets);
+    const auto bucket = relocation_buckets[res.first];
+    if (res.second + 1 < bucket.entries.size())
+        return bucket.entries[res.second + 1];
+    return relocation_buckets[res.first + 1].entries[0];
+}
+
+SubsectionEntry BKTR::GetSubsectionEntry(u64 offset) const {
+    const auto res = SearchBucketEntry<true>(offset, subsection, subsection_buckets);
+    return subsection_buckets[res.first].entries[res.second];
+}
+
+SubsectionEntry BKTR::GetNextSubsectionEntry(u64 offset) const {
+    const auto res = SearchBucketEntry<true>(offset, subsection, subsection_buckets);
+    const auto bucket = subsection_buckets[res.first];
+    if (res.second + 1 < bucket.entries.size())
+        return bucket.entries[res.second + 1];
+    return subsection_buckets[res.first + 1].entries[0];
+}
+
+std::string BKTR::GetName() const {
+    return base_romfs->GetName();
+}
+
+size_t BKTR::GetSize() const {
+    return relocation.size;
+}
+
+bool BKTR::Resize(size_t new_size) {
+    return false;
+}
+
+std::shared_ptr<VfsDirectory> BKTR::GetContainingDirectory() const {
+    return base_romfs->GetContainingDirectory();
+}
+
+bool BKTR::IsWritable() const {
+    return false;
+}
+
+bool BKTR::IsReadable() const {
+    return true;
+}
+
+size_t BKTR::Write(const u8* data, size_t length, size_t offset) {
+    return 0;
+}
+
+bool BKTR::Rename(std::string_view name) {
+    return base_romfs->Rename(name);
+}
+
+} // namespace FileSys

src/core/file_sys/nca_patch.h

@@ -0,0 +1,150 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include <memory>
+#include <vector>
+
+#include "common/common_funcs.h"
+#include "common/common_types.h"
+#include "common/swap.h"
+#include "core/crypto/key_manager.h"
+
+namespace FileSys {
+
+#pragma pack(push, 1)
+struct RelocationEntry {
+    u64_le address_patch;
+    u64_le address_source;
+    u32 from_patch;
+};
+#pragma pack(pop)
+static_assert(sizeof(RelocationEntry) == 0x14, "RelocationEntry has incorrect size.");
+
+struct RelocationBucketRaw {
+    INSERT_PADDING_BYTES(4);
+    u32_le number_entries;
+    u64_le end_offset;
+    std::array<RelocationEntry, 0x332> relocation_entries;
+    INSERT_PADDING_BYTES(8);
+};
+static_assert(sizeof(RelocationBucketRaw) == 0x4000, "RelocationBucketRaw has incorrect size.");
+
+// Vector version of RelocationBucketRaw
+struct RelocationBucket {
+    u32 number_entries;
+    u64 end_offset;
+    std::vector<RelocationEntry> entries;
+};
+
+struct RelocationBlock {
+    INSERT_PADDING_BYTES(4);
+    u32_le number_buckets;
+    u64_le size;
+    std::array<u64, 0x7FE> base_offsets;
+};
+static_assert(sizeof(RelocationBlock) == 0x4000, "RelocationBlock has incorrect size.");
+
+struct SubsectionEntry {
+    u64_le address_patch;
+    INSERT_PADDING_BYTES(0x4);
+    u32_le ctr;
+};
+static_assert(sizeof(SubsectionEntry) == 0x10, "SubsectionEntry has incorrect size.");
+
+struct SubsectionBucketRaw {
+    INSERT_PADDING_BYTES(4);
+    u32_le number_entries;
+    u64_le end_offset;
+    std::array<SubsectionEntry, 0x3FF> subsection_entries;
+};
+static_assert(sizeof(SubsectionBucketRaw) == 0x4000, "SubsectionBucketRaw has incorrect size.");
+
+// Vector version of SubsectionBucketRaw
+struct SubsectionBucket {
+    u32 number_entries;
+    u64 end_offset;
+    std::vector<SubsectionEntry> entries;
+};
+
+struct SubsectionBlock {
+    INSERT_PADDING_BYTES(4);
+    u32_le number_buckets;
+    u64_le size;
+    std::array<u64, 0x7FE> base_offsets;
+};
+static_assert(sizeof(SubsectionBlock) == 0x4000, "SubsectionBlock has incorrect size.");
+
+inline RelocationBucket ConvertRelocationBucketRaw(RelocationBucketRaw raw) {
+    return {raw.number_entries,
+            raw.end_offset,
+            {raw.relocation_entries.begin(), raw.relocation_entries.begin() + raw.number_entries}};
+}
+
+inline SubsectionBucket ConvertSubsectionBucketRaw(SubsectionBucketRaw raw) {
+    return {raw.number_entries,
+            raw.end_offset,
+            {raw.subsection_entries.begin(), raw.subsection_entries.begin() + raw.number_entries}};
+}
+
+class BKTR : public VfsFile {
+public:
+    BKTR(VirtualFile base_romfs, VirtualFile bktr_romfs, RelocationBlock relocation,
+         std::vector<RelocationBucket> relocation_buckets, SubsectionBlock subsection,
+         std::vector<SubsectionBucket> subsection_buckets, bool is_encrypted,
+         Core::Crypto::Key128 key, u64 base_offset, u64 ivfc_offset, std::array<u8, 8> section_ctr);
+    ~BKTR() override;
+
+    size_t Read(u8* data, size_t length, size_t offset) const override;
+
+    std::string GetName() const override;
+
+    size_t GetSize() const override;
+
+    bool Resize(size_t new_size) override;
+
+    std::shared_ptr<VfsDirectory> GetContainingDirectory() const override;
+
+    bool IsWritable() const override;
+
+    bool IsReadable() const override;
+
+    size_t Write(const u8* data, size_t length, size_t offset) override;
+
+    bool Rename(std::string_view name) override;
+
+private:
+    template <bool Subsection, typename BlockType, typename BucketType>
+    std::pair<size_t, size_t> SearchBucketEntry(u64 offset, BlockType block,
+                                                BucketType buckets) const;
+
+    RelocationEntry GetRelocationEntry(u64 offset) const;
+    RelocationEntry GetNextRelocationEntry(u64 offset) const;
+
+    SubsectionEntry GetSubsectionEntry(u64 offset) const;
+    SubsectionEntry GetNextSubsectionEntry(u64 offset) const;
+
+    RelocationBlock relocation;
+    std::vector<RelocationBucket> relocation_buckets;
+    SubsectionBlock subsection;
+    std::vector<SubsectionBucket> subsection_buckets;
+
+    // Should be the raw base romfs, decrypted.
+    VirtualFile base_romfs;
+    // Should be the raw BKTR romfs, (located at media_offset with size media_size).
+    VirtualFile bktr_romfs;
+
+    bool encrypted;
+    Core::Crypto::Key128 key;
+
+    // Base offset into NCA, used for IV calculation.
+    u64 base_offset;
+    // Distance between IVFC start and RomFS start, used for base reads
+    u64 ivfc_offset;
+    std::array<u8, 8> section_ctr;
+};
+
+} // namespace FileSys

src/core/file_sys/patch_manager.cpp

@@ -0,0 +1,157 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <array>
+#include <cstddef>
+
+#include "common/logging/log.h"
+#include "core/file_sys/content_archive.h"
+#include "core/file_sys/control_metadata.h"
+#include "core/file_sys/patch_manager.h"
+#include "core/file_sys/registered_cache.h"
+#include "core/file_sys/romfs.h"
+#include "core/hle/service/filesystem/filesystem.h"
+#include "core/loader/loader.h"
+
+namespace FileSys {
+
+constexpr u64 SINGLE_BYTE_MODULUS = 0x100;
+
+std::string FormatTitleVersion(u32 version, TitleVersionFormat format) {
+    std::array<u8, sizeof(u32)> bytes{};
+    bytes[0] = version % SINGLE_BYTE_MODULUS;
+    for (size_t i = 1; i < bytes.size(); ++i) {
+        version /= SINGLE_BYTE_MODULUS;
+        bytes[i] = version % SINGLE_BYTE_MODULUS;
+    }
+
+    if (format == TitleVersionFormat::FourElements)
+        return fmt::format("v{}.{}.{}.{}", bytes[3], bytes[2], bytes[1], bytes[0]);
+    return fmt::format("v{}.{}.{}", bytes[3], bytes[2], bytes[1]);
+}
+
+constexpr std::array<const char*, 1> PATCH_TYPE_NAMES{
+    "Update",
+};
+
+std::string FormatPatchTypeName(PatchType type) {
+    return PATCH_TYPE_NAMES.at(static_cast<size_t>(type));
+}
+
+PatchManager::PatchManager(u64 title_id) : title_id(title_id) {}
+
+VirtualDir PatchManager::PatchExeFS(VirtualDir exefs) const {
+    LOG_INFO(Loader, "Patching ExeFS for title_id={:016X}", title_id);
+
+    if (exefs == nullptr)
+        return exefs;
+
+    const auto installed = Service::FileSystem::GetUnionContents();
+
+    // Game Updates
+    const auto update_tid = GetUpdateTitleID(title_id);
+    const auto update = installed->GetEntry(update_tid, ContentRecordType::Program);
+    if (update != nullptr) {
+        if (update->GetStatus() == Loader::ResultStatus::ErrorMissingBKTRBaseRomFS &&
+            update->GetExeFS() != nullptr) {
+            LOG_INFO(Loader, "    ExeFS: Update ({}) applied successfully",
+                     FormatTitleVersion(installed->GetEntryVersion(update_tid).get_value_or(0)));
+            exefs = update->GetExeFS();
+        }
+    }
+
+    return exefs;
+}
+
+VirtualFile PatchManager::PatchRomFS(VirtualFile romfs, u64 ivfc_offset,
+                                     ContentRecordType type) const {
+    LOG_INFO(Loader, "Patching RomFS for title_id={:016X}, type={:02X}", title_id,
+             static_cast<u8>(type));
+
+    if (romfs == nullptr)
+        return romfs;
+
+    const auto installed = Service::FileSystem::GetUnionContents();
+
+    // Game Updates
+    const auto update_tid = GetUpdateTitleID(title_id);
+    const auto update = installed->GetEntryRaw(update_tid, type);
+    if (update != nullptr) {
+        const auto new_nca = std::make_shared<NCA>(update, romfs, ivfc_offset);
+        if (new_nca->GetStatus() == Loader::ResultStatus::Success &&
+            new_nca->GetRomFS() != nullptr) {
+            LOG_INFO(Loader, "    RomFS: Update ({}) applied successfully",
+                     FormatTitleVersion(installed->GetEntryVersion(update_tid).get_value_or(0)));
+            romfs = new_nca->GetRomFS();
+        }
+    }
+
+    return romfs;
+}
+
+std::map<PatchType, std::string> PatchManager::GetPatchVersionNames() const {
+    std::map<PatchType, std::string> out;
+    const auto installed = Service::FileSystem::GetUnionContents();
+
+    const auto update_tid = GetUpdateTitleID(title_id);
+    PatchManager update{update_tid};
+    auto [nacp, discard_icon_file] = update.GetControlMetadata();
+
+    if (nacp != nullptr) {
+        out[PatchType::Update] = nacp->GetVersionString();
+    } else {
+        if (installed->HasEntry(update_tid, ContentRecordType::Program)) {
+            const auto meta_ver = installed->GetEntryVersion(update_tid);
+            if (meta_ver == boost::none || meta_ver.get() == 0) {
+                out[PatchType::Update] = "";
+            } else {
+                out[PatchType::Update] =
+                    FormatTitleVersion(meta_ver.get(), TitleVersionFormat::ThreeElements);
+            }
+        }
+    }
+
+    return out;
+}
+
+std::pair<std::shared_ptr<NACP>, VirtualFile> PatchManager::GetControlMetadata() const {
+    const auto& installed{Service::FileSystem::GetUnionContents()};
+
+    const auto base_control_nca = installed->GetEntry(title_id, ContentRecordType::Control);
+    if (base_control_nca == nullptr)
+        return {};
+
+    return ParseControlNCA(base_control_nca);
+}
+
+std::pair<std::shared_ptr<NACP>, VirtualFile> PatchManager::ParseControlNCA(
+    const std::shared_ptr<NCA>& nca) const {
+    const auto base_romfs = nca->GetRomFS();
+    if (base_romfs == nullptr)
+        return {};
+
+    const auto romfs = PatchRomFS(base_romfs, nca->GetBaseIVFCOffset(), ContentRecordType::Control);
+    if (romfs == nullptr)
+        return {};
+
+    const auto extracted = ExtractRomFS(romfs);
+    if (extracted == nullptr)
+        return {};
+
+    auto nacp_file = extracted->GetFile("control.nacp");
+    if (nacp_file == nullptr)
+        nacp_file = extracted->GetFile("Control.nacp");
+
+    const auto nacp = nacp_file == nullptr ? nullptr : std::make_shared<NACP>(nacp_file);
+
+    VirtualFile icon_file;
+    for (const auto& language : FileSys::LANGUAGE_NAMES) {
+        icon_file = extracted->GetFile("icon_" + std::string(language) + ".dat");
+        if (icon_file != nullptr)
+            break;
+    }
+
+    return {nacp, icon_file};
+}
+} // namespace FileSys

src/core/file_sys/patch_manager.h

@@ -0,0 +1,63 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <map>
+#include <memory>
+#include <string>
+#include "common/common_types.h"
+#include "core/file_sys/nca_metadata.h"
+#include "core/file_sys/vfs.h"
+
+namespace FileSys {
+
+class NCA;
+class NACP;
+
+enum class TitleVersionFormat : u8 {
+    ThreeElements, ///< vX.Y.Z
+    FourElements,  ///< vX.Y.Z.W
+};
+
+std::string FormatTitleVersion(u32 version,
+                               TitleVersionFormat format = TitleVersionFormat::ThreeElements);
+
+enum class PatchType {
+    Update,
+};
+
+std::string FormatPatchTypeName(PatchType type);
+
+// A centralized class to manage patches to games.
+class PatchManager {
+public:
+    explicit PatchManager(u64 title_id);
+
+    // Currently tracked ExeFS patches:
+    // - Game Updates
+    VirtualDir PatchExeFS(VirtualDir exefs) const;
+
+    // Currently tracked RomFS patches:
+    // - Game Updates
+    VirtualFile PatchRomFS(VirtualFile base, u64 ivfc_offset,
+                           ContentRecordType type = ContentRecordType::Program) const;
+
+    // Returns a vector of pairs between patch names and patch versions.
+    // i.e. Update v80 will return {Update, 80}
+    std::map<PatchType, std::string> GetPatchVersionNames() const;
+
+    // Given title_id of the program, attempts to get the control data of the update and parse it,
+    // falling back to the base control data.
+    std::pair<std::shared_ptr<NACP>, VirtualFile> GetControlMetadata() const;
+
+    // Version of GetControlMetadata that takes an arbitrary NCA
+    std::pair<std::shared_ptr<NACP>, VirtualFile> ParseControlNCA(
+        const std::shared_ptr<NCA>& nca) const;
+
+private:
+    u64 title_id;
+};
+
+} // namespace FileSys

src/core/file_sys/program_metadata.cpp

@@ -2,7 +2,10 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#include "common/file_util.h"
+#include <cstddef>
+#include <cstring>
+#include <vector>
+
 #include "common/logging/log.h"
 #include "core/file_sys/program_metadata.h"
 #include "core/loader/loader.h"

src/core/file_sys/program_metadata.h

@@ -5,12 +5,10 @@
 #pragma once
 
 #include <array>
-#include <string>
-#include <vector>
 #include "common/bit_field.h"
 #include "common/common_types.h"
 #include "common/swap.h"
-#include "partition_filesystem.h"
+#include "core/file_sys/vfs.h"
 
 namespace Loader {
 enum class ResultStatus : u16;

src/core/file_sys/registered_cache.cpp

@@ -8,10 +8,14 @@
 #include "common/file_util.h"
 #include "common/hex_util.h"
 #include "common/logging/log.h"
+#include "core/crypto/key_manager.h"
 #include "core/file_sys/card_image.h"
+#include "core/file_sys/content_archive.h"
 #include "core/file_sys/nca_metadata.h"
 #include "core/file_sys/registered_cache.h"
+#include "core/file_sys/submission_package.h"
 #include "core/file_sys/vfs_concat.h"
+#include "core/loader/loader.h"
 
 namespace FileSys {
 std::string RegisteredCacheEntry::DebugInfo() const {
@@ -276,6 +280,18 @@ VirtualFile RegisteredCache::GetEntryUnparsed(RegisteredCacheEntry entry) const
     return GetEntryUnparsed(entry.title_id, entry.type);
 }
 
+boost::optional<u32> RegisteredCache::GetEntryVersion(u64 title_id) const {
+    const auto meta_iter = meta.find(title_id);
+    if (meta_iter != meta.end())
+        return meta_iter->second.GetTitleVersion();
+
+    const auto yuzu_meta_iter = yuzu_meta.find(title_id);
+    if (yuzu_meta_iter != yuzu_meta.end())
+        return yuzu_meta_iter->second.GetTitleVersion();
+
+    return boost::none;
+}
+
 VirtualFile RegisteredCache::GetEntryRaw(u64 title_id, ContentRecordType type) const {
     const auto id = GetNcaIDFromMetadata(title_id, type);
     if (id == boost::none)
@@ -355,17 +371,21 @@ std::vector<RegisteredCacheEntry> RegisteredCache::ListEntriesFilter(
     return out;
 }
 
-static std::shared_ptr<NCA> GetNCAFromXCIForID(std::shared_ptr<XCI> xci, const NcaID& id) {
-    const auto filename = fmt::format("{}.nca", Common::HexArrayToString(id, false));
-    const auto iter =
-        std::find_if(xci->GetNCAs().begin(), xci->GetNCAs().end(),
-                     [&filename](std::shared_ptr<NCA> nca) { return nca->GetName() == filename; });
-    return iter == xci->GetNCAs().end() ? nullptr : *iter;
+static std::shared_ptr<NCA> GetNCAFromNSPForID(std::shared_ptr<NSP> nsp, const NcaID& id) {
+    const auto file = nsp->GetFile(fmt::format("{}.nca", Common::HexArrayToString(id, false)));
+    if (file == nullptr)
+        return nullptr;
+    return std::make_shared<NCA>(file);
 }
 
 InstallResult RegisteredCache::InstallEntry(std::shared_ptr<XCI> xci, bool overwrite_if_exists,
                                             const VfsCopyFunction& copy) {
-    const auto& ncas = xci->GetNCAs();
+    return InstallEntry(xci->GetSecurePartitionNSP(), overwrite_if_exists, copy);
+}
+
+InstallResult RegisteredCache::InstallEntry(std::shared_ptr<NSP> nsp, bool overwrite_if_exists,
+                                            const VfsCopyFunction& copy) {
+    const auto& ncas = nsp->GetNCAsCollapsed();
     const auto& meta_iter = std::find_if(ncas.begin(), ncas.end(), [](std::shared_ptr<NCA> nca) {
         return nca->GetType() == NCAContentType::Meta;
     });
@@ -389,7 +409,7 @@ InstallResult RegisteredCache::InstallEntry(std::shared_ptr<XCI> xci, bool overw
     const auto cnmt_file = section0->GetFiles()[0];
     const CNMT cnmt(cnmt_file);
     for (const auto& record : cnmt.GetContentRecords()) {
-        const auto nca = GetNCAFromXCIForID(xci, record.nca_id);
+        const auto nca = GetNCAFromNSPForID(nsp, record.nca_id);
         if (nca == nullptr)
             return InstallResult::ErrorCopyFailed;
         const auto res2 = RawInstallNCA(nca, copy, overwrite_if_exists, record.nca_id);
@@ -490,4 +510,107 @@ bool RegisteredCache::RawInstallYuzuMeta(const CNMT& cnmt) {
                                    kv.second.GetTitleID() == cnmt.GetTitleID();
                         }) != yuzu_meta.end();
 }
+
+RegisteredCacheUnion::RegisteredCacheUnion(std::vector<std::shared_ptr<RegisteredCache>> caches)
+    : caches(std::move(caches)) {}
+
+void RegisteredCacheUnion::Refresh() {
+    for (const auto& c : caches)
+        c->Refresh();
+}
+
+bool RegisteredCacheUnion::HasEntry(u64 title_id, ContentRecordType type) const {
+    return std::any_of(caches.begin(), caches.end(), [title_id, type](const auto& cache) {
+        return cache->HasEntry(title_id, type);
+    });
+}
+
+bool RegisteredCacheUnion::HasEntry(RegisteredCacheEntry entry) const {
+    return HasEntry(entry.title_id, entry.type);
+}
+
+boost::optional<u32> RegisteredCacheUnion::GetEntryVersion(u64 title_id) const {
+    for (const auto& c : caches) {
+        const auto res = c->GetEntryVersion(title_id);
+        if (res != boost::none)
+            return res;
+    }
+
+    return boost::none;
+}
+
+VirtualFile RegisteredCacheUnion::GetEntryUnparsed(u64 title_id, ContentRecordType type) const {
+    for (const auto& c : caches) {
+        const auto res = c->GetEntryUnparsed(title_id, type);
+        if (res != nullptr)
+            return res;
+    }
+
+    return nullptr;
+}
+
+VirtualFile RegisteredCacheUnion::GetEntryUnparsed(RegisteredCacheEntry entry) const {
+    return GetEntryUnparsed(entry.title_id, entry.type);
+}
+
+VirtualFile RegisteredCacheUnion::GetEntryRaw(u64 title_id, ContentRecordType type) const {
+    for (const auto& c : caches) {
+        const auto res = c->GetEntryRaw(title_id, type);
+        if (res != nullptr)
+            return res;
+    }
+
+    return nullptr;
+}
+
+VirtualFile RegisteredCacheUnion::GetEntryRaw(RegisteredCacheEntry entry) const {
+    return GetEntryRaw(entry.title_id, entry.type);
+}
+
+std::shared_ptr<NCA> RegisteredCacheUnion::GetEntry(u64 title_id, ContentRecordType type) const {
+    const auto raw = GetEntryRaw(title_id, type);
+    if (raw == nullptr)
+        return nullptr;
+    return std::make_shared<NCA>(raw);
+}
+
+std::shared_ptr<NCA> RegisteredCacheUnion::GetEntry(RegisteredCacheEntry entry) const {
+    return GetEntry(entry.title_id, entry.type);
+}
+
+std::vector<RegisteredCacheEntry> RegisteredCacheUnion::ListEntries() const {
+    std::vector<RegisteredCacheEntry> out;
+    for (const auto& c : caches) {
+        c->IterateAllMetadata<RegisteredCacheEntry>(
+            out,
+            [](const CNMT& c, const ContentRecord& r) {
+                return RegisteredCacheEntry{c.GetTitleID(), r.type};
+            },
+            [](const CNMT& c, const ContentRecord& r) { return true; });
+    }
+    return out;
+}
+
+std::vector<RegisteredCacheEntry> RegisteredCacheUnion::ListEntriesFilter(
+    boost::optional<TitleType> title_type, boost::optional<ContentRecordType> record_type,
+    boost::optional<u64> title_id) const {
+    std::vector<RegisteredCacheEntry> out;
+    for (const auto& c : caches) {
+        c->IterateAllMetadata<RegisteredCacheEntry>(
+            out,
+            [](const CNMT& c, const ContentRecord& r) {
+                return RegisteredCacheEntry{c.GetTitleID(), r.type};
+            },
+            [&title_type, &record_type, &title_id](const CNMT& c, const ContentRecord& r) {
+                if (title_type != boost::none && title_type.get() != c.GetType())
+                    return false;
+                if (record_type != boost::none && record_type.get() != r.type)
+                    return false;
+                if (title_id != boost::none && title_id.get() != c.GetTitleID())
+                    return false;
+                return true;
+            });
+    }
+    return out;
+}
 } // namespace FileSys

src/core/file_sys/registered_cache.h

@@ -11,15 +11,19 @@
 #include <string>
 #include <vector>
 #include <boost/container/flat_map.hpp>
-#include "common/common_funcs.h"
 #include "common/common_types.h"
-#include "content_archive.h"
-#include "core/file_sys/nca_metadata.h"
 #include "core/file_sys/vfs.h"
 
 namespace FileSys {
-class XCI;
 class CNMT;
+class NCA;
+class NSP;
+class XCI;
+
+enum class ContentRecordType : u8;
+enum class TitleType : u8;
+
+struct ContentRecord;
 
 using NcaID = std::array<u8, 0x10>;
 using RegisteredCacheParsingFunction = std::function<VirtualFile(const VirtualFile&, const NcaID&)>;
@@ -39,6 +43,10 @@ struct RegisteredCacheEntry {
     std::string DebugInfo() const;
 };
 
+constexpr u64 GetUpdateTitleID(u64 base_title_id) {
+    return base_title_id | 0x800;
+}
+
 // boost flat_map requires operator< for O(log(n)) lookups.
 bool operator<(const RegisteredCacheEntry& lhs, const RegisteredCacheEntry& rhs);
 
@@ -56,6 +64,8 @@ bool operator<(const RegisteredCacheEntry& lhs, const RegisteredCacheEntry& rhs)
  * 4GB splitting can be ignored.)
  */
 class RegisteredCache {
+    friend class RegisteredCacheUnion;
+
 public:
     // Parsing function defines the conversion from raw file to NCA. If there are other steps
     // besides creating the NCA from the file (e.g. NAX0 on SD Card), that should go in a custom
@@ -70,6 +80,8 @@ public:
     bool HasEntry(u64 title_id, ContentRecordType type) const;
     bool HasEntry(RegisteredCacheEntry entry) const;
 
+    boost::optional<u32> GetEntryVersion(u64 title_id) const;
+
     VirtualFile GetEntryUnparsed(u64 title_id, ContentRecordType type) const;
     VirtualFile GetEntryUnparsed(RegisteredCacheEntry entry) const;
 
@@ -86,10 +98,12 @@ public:
         boost::optional<ContentRecordType> record_type = boost::none,
         boost::optional<u64> title_id = boost::none) const;
 
-    // Raw copies all the ncas from the xci to the csache. Does some quick checks to make sure there
-    // is a meta NCA and all of them are accessible.
+    // Raw copies all the ncas from the xci/nsp to the csache. Does some quick checks to make sure
+    // there is a meta NCA and all of them are accessible.
     InstallResult InstallEntry(std::shared_ptr<XCI> xci, bool overwrite_if_exists = false,
                                const VfsCopyFunction& copy = &VfsRawCopy);
+    InstallResult InstallEntry(std::shared_ptr<NSP> nsp, bool overwrite_if_exists = false,
+                               const VfsCopyFunction& copy = &VfsRawCopy);
 
     // Due to the fact that we must use Meta-type NCAs to determine the existance of files, this
     // poses quite a challenge. Instead of creating a new meta NCA for this file, yuzu will create a
@@ -125,4 +139,36 @@ private:
     boost::container::flat_map<u64, CNMT> yuzu_meta;
 };
 
+// Combines multiple RegisteredCaches (i.e. SysNAND, UserNAND, SDMC) into one interface.
+class RegisteredCacheUnion {
+public:
+    explicit RegisteredCacheUnion(std::vector<std::shared_ptr<RegisteredCache>> caches);
+
+    void Refresh();
+
+    bool HasEntry(u64 title_id, ContentRecordType type) const;
+    bool HasEntry(RegisteredCacheEntry entry) const;
+
+    boost::optional<u32> GetEntryVersion(u64 title_id) const;
+
+    VirtualFile GetEntryUnparsed(u64 title_id, ContentRecordType type) const;
+    VirtualFile GetEntryUnparsed(RegisteredCacheEntry entry) const;
+
+    VirtualFile GetEntryRaw(u64 title_id, ContentRecordType type) const;
+    VirtualFile GetEntryRaw(RegisteredCacheEntry entry) const;
+
+    std::shared_ptr<NCA> GetEntry(u64 title_id, ContentRecordType type) const;
+    std::shared_ptr<NCA> GetEntry(RegisteredCacheEntry entry) const;
+
+    std::vector<RegisteredCacheEntry> ListEntries() const;
+    // If a parameter is not boost::none, it will be filtered for from all entries.
+    std::vector<RegisteredCacheEntry> ListEntriesFilter(
+        boost::optional<TitleType> title_type = boost::none,
+        boost::optional<ContentRecordType> record_type = boost::none,
+        boost::optional<u64> title_id = boost::none) const;
+
+private:
+    std::vector<std::shared_ptr<RegisteredCache>> caches;
+};
+
 } // namespace FileSys

src/core/file_sys/romfs.h

@@ -6,6 +6,7 @@
 
 #include <array>
 #include "common/common_funcs.h"
+#include "common/common_types.h"
 #include "common/swap.h"
 #include "core/file_sys/vfs.h"
 

src/core/file_sys/romfs_factory.cpp

@@ -2,11 +2,14 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#include <algorithm>
 #include <memory>
+#include "common/assert.h"
 #include "common/common_types.h"
 #include "common/logging/log.h"
+#include "core/core.h"
+#include "core/file_sys/content_archive.h"
 #include "core/file_sys/nca_metadata.h"
+#include "core/file_sys/patch_manager.h"
 #include "core/file_sys/registered_cache.h"
 #include "core/file_sys/romfs_factory.h"
 #include "core/hle/kernel/process.h"
@@ -20,10 +23,17 @@ RomFSFactory::RomFSFactory(Loader::AppLoader& app_loader) {
     if (app_loader.ReadRomFS(file) != Loader::ResultStatus::Success) {
         LOG_ERROR(Service_FS, "Unable to read RomFS!");
     }
+
+    updatable = app_loader.IsRomFSUpdatable();
+    ivfc_offset = app_loader.ReadRomFSIVFCOffset();
 }
 
 ResultVal<VirtualFile> RomFSFactory::OpenCurrentProcess() {
-    return MakeResult<VirtualFile>(file);
+    if (!updatable)
+        return MakeResult<VirtualFile>(file);
+
+    const PatchManager patch_manager(Core::CurrentProcess()->program_id);
+    return MakeResult<VirtualFile>(patch_manager.PatchRomFS(file, ivfc_offset));
 }
 
 ResultVal<VirtualFile> RomFSFactory::Open(u64 title_id, StorageId storage, ContentRecordType type) {

src/core/file_sys/romfs_factory.h

@@ -36,6 +36,8 @@ public:
 
 private:
     VirtualFile file;
+    bool updatable;
+    u64 ivfc_offset;
 };
 
 } // namespace FileSys

src/core/file_sys/savedata_factory.cpp

@@ -3,6 +3,7 @@
 // Refer to the license.txt file included.
 
 #include <memory>
+#include "common/assert.h"
 #include "common/common_types.h"
 #include "common/logging/log.h"
 #include "core/core.h"

src/core/file_sys/savedata_factory.h

@@ -6,6 +6,7 @@
 
 #include <memory>
 #include <string>
+#include "common/common_funcs.h"
 #include "common/common_types.h"
 #include "common/swap.h"
 #include "core/file_sys/vfs.h"

src/core/file_sys/submission_package.cpp

@@ -0,0 +1,245 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <algorithm>
+#include <cstring>
+#include <string_view>
+
+#include <fmt/ostream.h>
+
+#include "common/hex_util.h"
+#include "common/logging/log.h"
+#include "core/crypto/key_manager.h"
+#include "core/file_sys/content_archive.h"
+#include "core/file_sys/nca_metadata.h"
+#include "core/file_sys/partition_filesystem.h"
+#include "core/file_sys/submission_package.h"
+#include "core/loader/loader.h"
+
+namespace FileSys {
+NSP::NSP(VirtualFile file_)
+    : file(std::move(file_)), status{Loader::ResultStatus::Success},
+      pfs(std::make_shared<PartitionFilesystem>(file)) {
+    if (pfs->GetStatus() != Loader::ResultStatus::Success) {
+        status = pfs->GetStatus();
+        return;
+    }
+
+    if (IsDirectoryExeFS(pfs)) {
+        extracted = true;
+        exefs = pfs;
+
+        const auto& files = pfs->GetFiles();
+        const auto romfs_iter =
+            std::find_if(files.begin(), files.end(), [](const FileSys::VirtualFile& file) {
+                return file->GetName().find(".romfs") != std::string::npos;
+            });
+        if (romfs_iter != files.end())
+            romfs = *romfs_iter;
+        return;
+    }
+
+    extracted = false;
+    const auto files = pfs->GetFiles();
+
+    Core::Crypto::KeyManager keys;
+    for (const auto& ticket_file : files) {
+        if (ticket_file->GetExtension() == "tik") {
+            if (ticket_file == nullptr ||
+                ticket_file->GetSize() <
+                    Core::Crypto::TICKET_FILE_TITLEKEY_OFFSET + sizeof(Core::Crypto::Key128)) {
+                continue;
+            }
+
+            Core::Crypto::Key128 key{};
+            ticket_file->Read(key.data(), key.size(), Core::Crypto::TICKET_FILE_TITLEKEY_OFFSET);
+            std::string_view name_only(ticket_file->GetName());
+            name_only.remove_suffix(4);
+            const auto rights_id_raw = Common::HexStringToArray<16>(name_only);
+            u128 rights_id;
+            std::memcpy(rights_id.data(), rights_id_raw.data(), sizeof(u128));
+            keys.SetKey(Core::Crypto::S128KeyType::Titlekey, key, rights_id[1], rights_id[0]);
+        }
+    }
+
+    for (const auto& outer_file : files) {
+        if (outer_file->GetName().substr(outer_file->GetName().size() - 9) == ".cnmt.nca") {
+            const auto nca = std::make_shared<NCA>(outer_file);
+            if (nca->GetStatus() != Loader::ResultStatus::Success) {
+                program_status[nca->GetTitleId()] = nca->GetStatus();
+                continue;
+            }
+
+            const auto section0 = nca->GetSubdirectories()[0];
+
+            for (const auto& inner_file : section0->GetFiles()) {
+                if (inner_file->GetExtension() != "cnmt")
+                    continue;
+
+                const CNMT cnmt(inner_file);
+                auto& ncas_title = ncas[cnmt.GetTitleID()];
+
+                ncas_title[ContentRecordType::Meta] = nca;
+                for (const auto& rec : cnmt.GetContentRecords()) {
+                    const auto id_string = Common::HexArrayToString(rec.nca_id, false);
+                    const auto next_file = pfs->GetFile(fmt::format("{}.nca", id_string));
+                    if (next_file == nullptr) {
+                        LOG_WARNING(Service_FS,
+                                    "NCA with ID {}.nca is listed in content metadata, but cannot "
+                                    "be found in PFS. NSP appears to be corrupted.",
+                                    id_string);
+                        continue;
+                    }
+
+                    auto next_nca = std::make_shared<NCA>(next_file);
+                    if (next_nca->GetType() == NCAContentType::Program)
+                        program_status[cnmt.GetTitleID()] = next_nca->GetStatus();
+                    if (next_nca->GetStatus() == Loader::ResultStatus::Success)
+                        ncas_title[rec.type] = std::move(next_nca);
+                }
+
+                break;
+            }
+        }
+    }
+}
+
+NSP::~NSP() = default;
+
+Loader::ResultStatus NSP::GetStatus() const {
+    return status;
+}
+
+Loader::ResultStatus NSP::GetProgramStatus(u64 title_id) const {
+    const auto iter = program_status.find(title_id);
+    if (iter == program_status.end())
+        return Loader::ResultStatus::ErrorNSPMissingProgramNCA;
+    return iter->second;
+}
+
+u64 NSP::GetFirstTitleID() const {
+    if (program_status.empty())
+        return 0;
+    return program_status.begin()->first;
+}
+
+u64 NSP::GetProgramTitleID() const {
+    const auto out = GetFirstTitleID();
+    if ((out & 0x800) == 0)
+        return out;
+
+    const auto ids = GetTitleIDs();
+    const auto iter =
+        std::find_if(ids.begin(), ids.end(), [](u64 tid) { return (tid & 0x800) == 0; });
+    return iter == ids.end() ? out : *iter;
+}
+
+std::vector<u64> NSP::GetTitleIDs() const {
+    std::vector<u64> out;
+    out.reserve(ncas.size());
+    for (const auto& kv : ncas)
+        out.push_back(kv.first);
+    return out;
+}
+
+bool NSP::IsExtractedType() const {
+    return extracted;
+}
+
+VirtualFile NSP::GetRomFS() const {
+    return romfs;
+}
+
+VirtualDir NSP::GetExeFS() const {
+    return exefs;
+}
+
+std::vector<std::shared_ptr<NCA>> NSP::GetNCAsCollapsed() const {
+    if (extracted)
+        LOG_WARNING(Service_FS, "called on an NSP that is of type extracted.");
+    std::vector<std::shared_ptr<NCA>> out;
+    for (const auto& map : ncas) {
+        for (const auto& inner_map : map.second)
+            out.push_back(inner_map.second);
+    }
+    return out;
+}
+
+std::multimap<u64, std::shared_ptr<NCA>> NSP::GetNCAsByTitleID() const {
+    if (extracted)
+        LOG_WARNING(Service_FS, "called on an NSP that is of type extracted.");
+    std::multimap<u64, std::shared_ptr<NCA>> out;
+    for (const auto& map : ncas) {
+        for (const auto& inner_map : map.second)
+            out.emplace(map.first, inner_map.second);
+    }
+    return out;
+}
+
+std::map<u64, std::map<ContentRecordType, std::shared_ptr<NCA>>> NSP::GetNCAs() const {
+    return ncas;
+}
+
+std::shared_ptr<NCA> NSP::GetNCA(u64 title_id, ContentRecordType type) const {
+    if (extracted)
+        LOG_WARNING(Service_FS, "called on an NSP that is of type extracted.");
+
+    const auto title_id_iter = ncas.find(title_id);
+    if (title_id_iter == ncas.end())
+        return nullptr;
+
+    const auto type_iter = title_id_iter->second.find(type);
+    if (type_iter == title_id_iter->second.end())
+        return nullptr;
+
+    return type_iter->second;
+}
+
+VirtualFile NSP::GetNCAFile(u64 title_id, ContentRecordType type) const {
+    if (extracted)
+        LOG_WARNING(Service_FS, "called on an NSP that is of type extracted.");
+    const auto nca = GetNCA(title_id, type);
+    if (nca != nullptr)
+        return nca->GetBaseFile();
+    return nullptr;
+}
+
+std::vector<Core::Crypto::Key128> NSP::GetTitlekey() const {
+    if (extracted)
+        LOG_WARNING(Service_FS, "called on an NSP that is of type extracted.");
+    std::vector<Core::Crypto::Key128> out;
+    for (const auto& ticket_file : ticket_files) {
+        if (ticket_file == nullptr ||
+            ticket_file->GetSize() <
+                Core::Crypto::TICKET_FILE_TITLEKEY_OFFSET + sizeof(Core::Crypto::Key128)) {
+            continue;
+        }
+
+        out.emplace_back();
+        ticket_file->Read(out.back().data(), out.back().size(),
+                          Core::Crypto::TICKET_FILE_TITLEKEY_OFFSET);
+    }
+    return out;
+}
+
+std::vector<VirtualFile> NSP::GetFiles() const {
+    return pfs->GetFiles();
+}
+
+std::vector<VirtualDir> NSP::GetSubdirectories() const {
+    return pfs->GetSubdirectories();
+}
+
+std::string NSP::GetName() const {
+    return file->GetName();
+}
+
+VirtualDir NSP::GetParentDirectory() const {
+    return file->GetContainingDirectory();
+}
+
+bool NSP::ReplaceFileWithSubdirectory(VirtualFile file, VirtualDir dir) {
+    return false;
+}
+} // namespace FileSys

src/core/file_sys/submission_package.h

@@ -0,0 +1,76 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <map>
+#include <memory>
+#include <vector>
+#include "common/common_types.h"
+#include "core/file_sys/vfs.h"
+
+namespace Loader {
+enum class ResultStatus : u16;
+}
+
+namespace FileSys {
+
+class NCA;
+class PartitionFilesystem;
+
+enum class ContentRecordType : u8;
+
+class NSP : public ReadOnlyVfsDirectory {
+public:
+    explicit NSP(VirtualFile file);
+    ~NSP();
+
+    Loader::ResultStatus GetStatus() const;
+    Loader::ResultStatus GetProgramStatus(u64 title_id) const;
+    // Should only be used when one title id can be assured.
+    u64 GetFirstTitleID() const;
+    u64 GetProgramTitleID() const;
+    std::vector<u64> GetTitleIDs() const;
+
+    bool IsExtractedType() const;
+
+    // Common (Can be safely called on both types)
+    VirtualFile GetRomFS() const;
+    VirtualDir GetExeFS() const;
+
+    // Type 0 Only (Collection of NCAs + Certificate + Ticket + Meta XML)
+    std::vector<std::shared_ptr<NCA>> GetNCAsCollapsed() const;
+    std::multimap<u64, std::shared_ptr<NCA>> GetNCAsByTitleID() const;
+    std::map<u64, std::map<ContentRecordType, std::shared_ptr<NCA>>> GetNCAs() const;
+    std::shared_ptr<NCA> GetNCA(u64 title_id, ContentRecordType type) const;
+    VirtualFile GetNCAFile(u64 title_id, ContentRecordType type) const;
+    std::vector<Core::Crypto::Key128> GetTitlekey() const;
+
+    std::vector<VirtualFile> GetFiles() const override;
+
+    std::vector<VirtualDir> GetSubdirectories() const override;
+
+    std::string GetName() const override;
+
+    VirtualDir GetParentDirectory() const override;
+
+protected:
+    bool ReplaceFileWithSubdirectory(VirtualFile file, VirtualDir dir) override;
+
+private:
+    VirtualFile file;
+
+    bool extracted;
+    Loader::ResultStatus status;
+    std::map<u64, Loader::ResultStatus> program_status;
+
+    std::shared_ptr<PartitionFilesystem> pfs;
+    // Map title id -> {map type -> NCA}
+    std::map<u64, std::map<ContentRecordType, std::shared_ptr<NCA>>> ncas;
+    std::vector<VirtualFile> ticket_files;
+
+    VirtualFile romfs;
+    VirtualDir exefs;
+};
+} // namespace FileSys

src/core/hle/kernel/kernel.cpp

@@ -116,6 +116,7 @@ struct KernelCore::Impl {
         next_thread_id = 1;
 
         process_list.clear();
+        current_process.reset();
 
         handle_table.Clear();
         resource_limits.fill(nullptr);
@@ -206,6 +207,7 @@ struct KernelCore::Impl {
 
     // Lists all processes that exist in the current session.
     std::vector<SharedPtr<Process>> process_list;
+    SharedPtr<Process> current_process;
 
     Kernel::HandleTable handle_table;
     std::array<SharedPtr<ResourceLimit>, 4> resource_limits;
@@ -264,6 +266,18 @@ void KernelCore::AppendNewProcess(SharedPtr<Process> process) {
     impl->process_list.push_back(std::move(process));
 }
 
+void KernelCore::MakeCurrentProcess(SharedPtr<Process> process) {
+    impl->current_process = std::move(process);
+}
+
+SharedPtr<Process>& KernelCore::CurrentProcess() {
+    return impl->current_process;
+}
+
+const SharedPtr<Process>& KernelCore::CurrentProcess() const {
+    return impl->current_process;
+}
+
 void KernelCore::AddNamedPort(std::string name, SharedPtr<ClientPort> port) {
     impl->named_ports.emplace(std::move(name), std::move(port));
 }

src/core/hle/kernel/kernel.h

@@ -65,6 +65,15 @@ public:
     /// Adds the given shared pointer to an internal list of active processes.
     void AppendNewProcess(SharedPtr<Process> process);
 
+    /// Makes the given process the new current process.
+    void MakeCurrentProcess(SharedPtr<Process> process);
+
+    /// Retrieves a reference to the current process.
+    SharedPtr<Process>& CurrentProcess();
+
+    /// Retrieves a const reference to the current process.
+    const SharedPtr<Process>& CurrentProcess() const;
+
     /// Adds a port to the named port table
     void AddNamedPort(std::string name, SharedPtr<ClientPort> port);
 

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

@@ -10,6 +10,7 @@
 #include "core/file_sys/bis_factory.h"
 #include "core/file_sys/errors.h"
 #include "core/file_sys/mode.h"
+#include "core/file_sys/registered_cache.h"
 #include "core/file_sys/romfs_factory.h"
 #include "core/file_sys/savedata_factory.h"
 #include "core/file_sys/sdmc_factory.h"
@@ -307,6 +308,12 @@ ResultVal<FileSys::VirtualDir> OpenSDMC() {
     return sdmc_factory->Open();
 }
 
+std::shared_ptr<FileSys::RegisteredCacheUnion> GetUnionContents() {
+    return std::make_shared<FileSys::RegisteredCacheUnion>(
+        std::vector<std::shared_ptr<FileSys::RegisteredCache>>{
+            GetSystemNANDContents(), GetUserNANDContents(), GetSDMCContents()});
+}
+
 std::shared_ptr<FileSys::RegisteredCache> GetSystemNANDContents() {
     LOG_TRACE(Service_FS, "Opening System NAND Contents");
 

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

@@ -13,6 +13,7 @@
 namespace FileSys {
 class BISFactory;
 class RegisteredCache;
+class RegisteredCacheUnion;
 class RomFSFactory;
 class SaveDataFactory;
 class SDMCFactory;
@@ -45,6 +46,8 @@ ResultVal<FileSys::VirtualDir> OpenSaveData(FileSys::SaveDataSpaceId space,
                                             FileSys::SaveDataDescriptor save_struct);
 ResultVal<FileSys::VirtualDir> OpenSDMC();
 
+std::shared_ptr<FileSys::RegisteredCacheUnion> GetUnionContents();
+
 std::shared_ptr<FileSys::RegisteredCache> GetSystemNANDContents();
 std::shared_ptr<FileSys::RegisteredCache> GetUserNANDContents();
 std::shared_ptr<FileSys::RegisteredCache> GetSDMCContents();

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

@@ -5,7 +5,9 @@
 #include "common/common_paths.h"
 #include "common/file_util.h"
 #include "core/core.h"
-#include "core/file_sys/bis_factory.h"
+#include "core/file_sys/content_archive.h"
+#include "core/file_sys/nca_metadata.h"
+#include "core/file_sys/registered_cache.h"
 #include "core/file_sys/romfs.h"
 #include "core/hle/ipc_helpers.h"
 #include "core/hle/service/filesystem/filesystem.h"

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

@@ -57,4 +57,6 @@ Controller::Controller() : ServiceFramework("IpcController") {
     RegisterHandlers(functions);
 }
 
+Controller::~Controller() = default;
+
 } // namespace Service::SM

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

@@ -11,7 +11,7 @@ namespace Service::SM {
 class Controller final : public ServiceFramework<Controller> {
 public:
     Controller();
-    ~Controller() = default;
+    ~Controller() override;
 
 private:
     void ConvertSessionToDomain(Kernel::HLERequestContext& ctx);

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

@@ -15,6 +15,7 @@
 
 namespace Service::SM {
 
+ServiceManager::ServiceManager() = default;
 ServiceManager::~ServiceManager() = default;
 
 void ServiceManager::InvokeControlRequest(Kernel::HLERequestContext& context) {

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

@@ -46,6 +46,7 @@ class ServiceManager {
 public:
     static void InstallInterfaces(std::shared_ptr<ServiceManager> self);
 
+    ServiceManager();
     ~ServiceManager();
 
     ResultVal<Kernel::SharedPtr<Kernel::ServerPort>> RegisterService(std::string name,

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

@@ -3,6 +3,9 @@
 // Refer to the license.txt file included.
 
 #include "core/hle/ipc_helpers.h"
+#include "core/hle/kernel/hle_ipc.h"
+#include "core/hle/service/service.h"
+#include "core/hle/service/sm/sm.h"
 #include "core/hle/service/ssl/ssl.h"
 
 namespace Service::SSL {
@@ -81,36 +84,43 @@ private:
     }
 };
 
-void SSL::CreateContext(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_SSL, "(STUBBED) called");
+class SSL final : public ServiceFramework<SSL> {
+public:
+    explicit SSL() : ServiceFramework{"ssl"} {
+        // clang-format off
+        static const FunctionInfo functions[] = {
+            {0, &SSL::CreateContext, "CreateContext"},
+            {1, nullptr, "GetContextCount"},
+            {2, nullptr, "GetCertificates"},
+            {3, nullptr, "GetCertificateBufSize"},
+            {4, nullptr, "DebugIoctl"},
+            {5, &SSL::SetInterfaceVersion, "SetInterfaceVersion"},
+            {6, nullptr, "FlushSessionCache"},
+        };
+        // clang-format on
 
-    IPC::ResponseBuilder rb{ctx, 2, 0, 1};
-    rb.Push(RESULT_SUCCESS);
-    rb.PushIpcInterface<ISslContext>();
-}
+        RegisterHandlers(functions);
+    }
 
-SSL::SSL() : ServiceFramework("ssl") {
-    static const FunctionInfo functions[] = {
-        {0, &SSL::CreateContext, "CreateContext"},
-        {1, nullptr, "GetContextCount"},
-        {2, nullptr, "GetCertificates"},
-        {3, nullptr, "GetCertificateBufSize"},
-        {4, nullptr, "DebugIoctl"},
-        {5, &SSL::SetInterfaceVersion, "SetInterfaceVersion"},
-        {6, nullptr, "FlushSessionCache"},
-    };
-    RegisterHandlers(functions);
-}
+private:
+    void CreateContext(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_SSL, "(STUBBED) called");
 
-void SSL::SetInterfaceVersion(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_SSL, "(STUBBED) called");
-    IPC::RequestParser rp{ctx};
-    u32 unk1 = rp.Pop<u32>(); // Probably minor/major?
-    u32 unk2 = rp.Pop<u32>(); // TODO(ogniK): Figure out what this does
+        IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushIpcInterface<ISslContext>();
+    }
 
-    IPC::ResponseBuilder rb{ctx, 2};
-    rb.Push(RESULT_SUCCESS);
-}
+    void SetInterfaceVersion(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_SSL, "(STUBBED) called");
+        IPC::RequestParser rp{ctx};
+        u32 unk1 = rp.Pop<u32>(); // Probably minor/major?
+        u32 unk2 = rp.Pop<u32>(); // TODO(ogniK): Figure out what this does
+
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+    }
+};
 
 void InstallInterfaces(SM::ServiceManager& service_manager) {
     std::make_shared<SSL>()->InstallAsService(service_manager);

src/core/hle/service/ssl/ssl.h

@@ -4,20 +4,12 @@
 
 #pragma once
 
-#include "core/hle/service/service.h"
+namespace Service::SM {
+class ServiceManager;
+}
 
 namespace Service::SSL {
 
-class SSL final : public ServiceFramework<SSL> {
-public:
-    explicit SSL();
-    ~SSL() = default;
-
-private:
-    void CreateContext(Kernel::HLERequestContext& ctx);
-    void SetInterfaceVersion(Kernel::HLERequestContext& ctx);
-};
-
 /// Registers all SSL services with the specified service manager.
 void InstallInterfaces(SM::ServiceManager& service_manager);
 

src/core/loader/deconstructed_rom_directory.cpp

@@ -9,6 +9,7 @@
 #include "core/core.h"
 #include "core/file_sys/content_archive.h"
 #include "core/file_sys/control_metadata.h"
+#include "core/file_sys/patch_manager.h"
 #include "core/file_sys/romfs_factory.h"
 #include "core/gdbstub/gdbstub.h"
 #include "core/hle/kernel/kernel.h"
@@ -21,10 +22,19 @@
 
 namespace Loader {
 
-AppLoader_DeconstructedRomDirectory::AppLoader_DeconstructedRomDirectory(FileSys::VirtualFile file_)
-    : AppLoader(std::move(file_)) {
+AppLoader_DeconstructedRomDirectory::AppLoader_DeconstructedRomDirectory(FileSys::VirtualFile file_,
+                                                                         bool override_update)
+    : AppLoader(std::move(file_)), override_update(override_update) {
     const auto dir = file->GetContainingDirectory();
 
+    // Title ID
+    const auto npdm = dir->GetFile("main.npdm");
+    if (npdm != nullptr) {
+        const auto res = metadata.Load(npdm);
+        if (res == ResultStatus::Success)
+            title_id = metadata.GetTitleID();
+    }
+
     // Icon
     FileSys::VirtualFile icon_file = nullptr;
     for (const auto& language : FileSys::LANGUAGE_NAMES) {
@@ -61,14 +71,14 @@ AppLoader_DeconstructedRomDirectory::AppLoader_DeconstructedRomDirectory(FileSys
 
     if (nacp_file != nullptr) {
         FileSys::NACP nacp(nacp_file);
-        title_id = nacp.GetTitleId();
         name = nacp.GetApplicationName();
     }
 }
 
 AppLoader_DeconstructedRomDirectory::AppLoader_DeconstructedRomDirectory(
-    FileSys::VirtualDir directory)
-    : AppLoader(directory->GetFile("main")), dir(std::move(directory)) {}
+    FileSys::VirtualDir directory, bool override_update)
+    : AppLoader(directory->GetFile("main")), dir(std::move(directory)),
+      override_update(override_update) {}
 
 FileType AppLoader_DeconstructedRomDirectory::IdentifyType(const FileSys::VirtualFile& file) {
     if (FileSys::IsDirectoryExeFS(file->GetContainingDirectory())) {
@@ -90,7 +100,8 @@ ResultStatus AppLoader_DeconstructedRomDirectory::Load(
         dir = file->GetContainingDirectory();
     }
 
-    const FileSys::VirtualFile npdm = dir->GetFile("main.npdm");
+    // Read meta to determine title ID
+    FileSys::VirtualFile npdm = dir->GetFile("main.npdm");
     if (npdm == nullptr)
         return ResultStatus::ErrorMissingNPDM;
 
@@ -98,6 +109,21 @@ ResultStatus AppLoader_DeconstructedRomDirectory::Load(
     if (result != ResultStatus::Success) {
         return result;
     }
+
+    if (override_update) {
+        const FileSys::PatchManager patch_manager(metadata.GetTitleID());
+        dir = patch_manager.PatchExeFS(dir);
+    }
+
+    // Reread in case PatchExeFS affected the main.npdm
+    npdm = dir->GetFile("main.npdm");
+    if (npdm == nullptr)
+        return ResultStatus::ErrorMissingNPDM;
+
+    ResultStatus result2 = metadata.Load(npdm);
+    if (result2 != ResultStatus::Success) {
+        return result2;
+    }
     metadata.Print();
 
     const FileSys::ProgramAddressSpaceType arch_bits{metadata.GetAddressSpaceType()};
@@ -159,8 +185,6 @@ ResultStatus AppLoader_DeconstructedRomDirectory::ReadIcon(std::vector<u8>& buff
 }
 
 ResultStatus AppLoader_DeconstructedRomDirectory::ReadProgramId(u64& out_program_id) {
-    if (name.empty())
-        return ResultStatus::ErrorNoControl;
     out_program_id = title_id;
     return ResultStatus::Success;
 }
@@ -172,4 +196,8 @@ ResultStatus AppLoader_DeconstructedRomDirectory::ReadTitle(std::string& title)
     return ResultStatus::Success;
 }
 
+bool AppLoader_DeconstructedRomDirectory::IsRomFSUpdatable() const {
+    return false;
+}
+
 } // namespace Loader

src/core/loader/deconstructed_rom_directory.h

@@ -20,10 +20,12 @@ namespace Loader {
  */
 class AppLoader_DeconstructedRomDirectory final : public AppLoader {
 public:
-    explicit AppLoader_DeconstructedRomDirectory(FileSys::VirtualFile main_file);
+    explicit AppLoader_DeconstructedRomDirectory(FileSys::VirtualFile main_file,
+                                                 bool override_update = false);
 
     // Overload to accept exefs directory. Must contain 'main' and 'main.npdm'
-    explicit AppLoader_DeconstructedRomDirectory(FileSys::VirtualDir directory);
+    explicit AppLoader_DeconstructedRomDirectory(FileSys::VirtualDir directory,
+                                                 bool override_update = false);
 
     /**
      * Returns the type of the file
@@ -42,6 +44,7 @@ public:
     ResultStatus ReadIcon(std::vector<u8>& buffer) override;
     ResultStatus ReadProgramId(u64& out_program_id) override;
     ResultStatus ReadTitle(std::string& title) override;
+    bool IsRomFSUpdatable() const override;
 
 private:
     FileSys::ProgramMetadata metadata;
@@ -51,6 +54,7 @@ private:
     std::vector<u8> icon_data;
     std::string name;
     u64 title_id{};
+    bool override_update;
 };
 
 } // namespace Loader

src/core/loader/loader.cpp

@@ -15,6 +15,7 @@
 #include "core/loader/nca.h"
 #include "core/loader/nro.h"
 #include "core/loader/nso.h"
+#include "core/loader/nsp.h"
 #include "core/loader/xci.h"
 
 namespace Loader {
@@ -34,6 +35,7 @@ FileType IdentifyFile(FileSys::VirtualFile file) {
     CHECK_TYPE(NCA)
     CHECK_TYPE(XCI)
     CHECK_TYPE(NAX)
+    CHECK_TYPE(NSP)
 
 #undef CHECK_TYPE
 
@@ -59,6 +61,8 @@ FileType GuessFromFilename(const std::string& name) {
         return FileType::NCA;
     if (extension == "xci")
         return FileType::XCI;
+    if (extension == "nsp")
+        return FileType::NSP;
 
     return FileType::Unknown;
 }
@@ -77,6 +81,8 @@ std::string GetFileTypeString(FileType type) {
         return "XCI";
     case FileType::NAX:
         return "NAX";
+    case FileType::NSP:
+        return "NSP";
     case FileType::DeconstructedRomDirectory:
         return "Directory";
     case FileType::Error:
@@ -87,7 +93,7 @@ std::string GetFileTypeString(FileType type) {
     return "unknown";
 }
 
-constexpr std::array<const char*, 49> RESULT_MESSAGES{
+constexpr std::array<const char*, 58> RESULT_MESSAGES{
     "The operation completed successfully.",
     "The loader requested to load is already loaded.",
     "The operation is not implemented.",
@@ -137,6 +143,15 @@ constexpr std::array<const char*, 49> RESULT_MESSAGES{
     "The AES Key Generation Source could not be found.",
     "The SD Save Key Source could not be found.",
     "The SD NCA Key Source could not be found.",
+    "The NSP file is missing a Program-type NCA.",
+    "The BKTR-type NCA has a bad BKTR header.",
+    "The BKTR Subsection entry is not located immediately after the Relocation entry.",
+    "The BKTR Subsection entry is not at the end of the media block.",
+    "The BKTR-type NCA has a bad Relocation block.",
+    "The BKTR-type NCA has a bad Subsection block.",
+    "The BKTR-type NCA has a bad Relocation bucket.",
+    "The BKTR-type NCA has a bad Subsection bucket.",
+    "The BKTR-type NCA is missing the base RomFS.",
 };
 
 std::ostream& operator<<(std::ostream& os, ResultStatus status) {
@@ -182,6 +197,10 @@ static std::unique_ptr<AppLoader> GetFileLoader(FileSys::VirtualFile file, FileT
     case FileType::NAX:
         return std::make_unique<AppLoader_NAX>(std::move(file));
 
+    // NX NSP (Nintendo Submission Package) file format
+    case FileType::NSP:
+        return std::make_unique<AppLoader_NSP>(std::move(file));
+
     // NX deconstructed ROM directory.
     case FileType::DeconstructedRomDirectory:
         return std::make_unique<AppLoader_DeconstructedRomDirectory>(std::move(file));

src/core/loader/loader.h

@@ -29,6 +29,7 @@ enum class FileType {
     NSO,
     NRO,
     NCA,
+    NSP,
     XCI,
     NAX,
     DeconstructedRomDirectory,
@@ -105,6 +106,15 @@ enum class ResultStatus : u16 {
     ErrorMissingAESKeyGenerationSource,
     ErrorMissingSDSaveKeySource,
     ErrorMissingSDNCAKeySource,
+    ErrorNSPMissingProgramNCA,
+    ErrorBadBKTRHeader,
+    ErrorBKTRSubsectionNotAfterRelocation,
+    ErrorBKTRSubsectionNotAtEnd,
+    ErrorBadRelocationBlock,
+    ErrorBadSubsectionBlock,
+    ErrorBadRelocationBuckets,
+    ErrorBadSubsectionBuckets,
+    ErrorMissingBKTRBaseRomFS,
 };
 
 std::ostream& operator<<(std::ostream& os, ResultStatus status);
@@ -195,13 +205,22 @@ public:
     }
 
     /**
-     * Get the update RomFS of the application
-     * Since the RomFS can be huge, we return a file reference instead of copying to a buffer
-     * @param file The file containing the RomFS
-     * @return ResultStatus result of function
+     * Get whether or not updates can be applied to the RomFS.
+     * By default, this is true, however for formats where it cannot be guaranteed that the RomFS is
+     * the base game it should be set to false.
+     * @return bool whether or not updatable.
      */
-    virtual ResultStatus ReadUpdateRomFS(FileSys::VirtualFile& file) {
-        return ResultStatus::ErrorNotImplemented;
+    virtual bool IsRomFSUpdatable() const {
+        return true;
+    }
+
+    /**
+     * Gets the difference between the start of the IVFC header and the start of level 6 (RomFS)
+     * data. Needed for bktr patching.
+     * @return IVFC offset for romfs.
+     */
+    virtual u64 ReadRomFSIVFCOffset() const {
+        return 0;
     }
 
     /**

src/core/loader/nca.cpp

@@ -48,7 +48,7 @@ ResultStatus AppLoader_NCA::Load(Kernel::SharedPtr<Kernel::Process>& process) {
     if (exefs == nullptr)
         return ResultStatus::ErrorNoExeFS;
 
-    directory_loader = std::make_unique<AppLoader_DeconstructedRomDirectory>(exefs);
+    directory_loader = std::make_unique<AppLoader_DeconstructedRomDirectory>(exefs, true);
 
     const auto load_result = directory_loader->Load(process);
     if (load_result != ResultStatus::Success)
@@ -71,6 +71,12 @@ ResultStatus AppLoader_NCA::ReadRomFS(FileSys::VirtualFile& dir) {
     return ResultStatus::Success;
 }
 
+u64 AppLoader_NCA::ReadRomFSIVFCOffset() const {
+    if (nca == nullptr)
+        return 0;
+    return nca->GetBaseIVFCOffset();
+}
+
 ResultStatus AppLoader_NCA::ReadProgramId(u64& out_program_id) {
     if (nca == nullptr || nca->GetStatus() != ResultStatus::Success)
         return ResultStatus::ErrorNotInitialized;

src/core/loader/nca.h

@@ -37,6 +37,7 @@ public:
     ResultStatus Load(Kernel::SharedPtr<Kernel::Process>& process) override;
 
     ResultStatus ReadRomFS(FileSys::VirtualFile& dir) override;
+    u64 ReadRomFSIVFCOffset() const override;
     ResultStatus ReadProgramId(u64& out_program_id) override;
 
 private:

src/core/loader/nro.cpp

@@ -232,4 +232,9 @@ ResultStatus AppLoader_NRO::ReadTitle(std::string& title) {
     title = nacp->GetApplicationName();
     return ResultStatus::Success;
 }
+
+bool AppLoader_NRO::IsRomFSUpdatable() const {
+    return false;
+}
+
 } // namespace Loader

src/core/loader/nro.h

@@ -39,6 +39,7 @@ public:
     ResultStatus ReadProgramId(u64& out_program_id) override;
     ResultStatus ReadRomFS(FileSys::VirtualFile& dir) override;
     ResultStatus ReadTitle(std::string& title) override;
+    bool IsRomFSUpdatable() const override;
 
 private:
     bool LoadNro(FileSys::VirtualFile file, VAddr load_base);

src/core/loader/nsp.cpp

@@ -0,0 +1,125 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <vector>
+
+#include "common/common_types.h"
+#include "core/file_sys/card_image.h"
+#include "core/file_sys/content_archive.h"
+#include "core/file_sys/control_metadata.h"
+#include "core/file_sys/nca_metadata.h"
+#include "core/file_sys/patch_manager.h"
+#include "core/file_sys/registered_cache.h"
+#include "core/file_sys/romfs.h"
+#include "core/file_sys/submission_package.h"
+#include "core/hle/kernel/process.h"
+#include "core/loader/deconstructed_rom_directory.h"
+#include "core/loader/nca.h"
+#include "core/loader/nsp.h"
+
+namespace Loader {
+
+AppLoader_NSP::AppLoader_NSP(FileSys::VirtualFile file)
+    : AppLoader(file), nsp(std::make_unique<FileSys::NSP>(file)),
+      title_id(nsp->GetProgramTitleID()) {
+
+    if (nsp->GetStatus() != ResultStatus::Success)
+        return;
+    if (nsp->IsExtractedType())
+        return;
+
+    const auto control_nca =
+        nsp->GetNCA(nsp->GetProgramTitleID(), FileSys::ContentRecordType::Control);
+    if (control_nca == nullptr || control_nca->GetStatus() != ResultStatus::Success)
+        return;
+
+    std::tie(nacp_file, icon_file) =
+        FileSys::PatchManager(nsp->GetProgramTitleID()).ParseControlNCA(control_nca);
+}
+
+AppLoader_NSP::~AppLoader_NSP() = default;
+
+FileType AppLoader_NSP::IdentifyType(const FileSys::VirtualFile& file) {
+    FileSys::NSP nsp(file);
+
+    if (nsp.GetStatus() == ResultStatus::Success) {
+        // Extracted Type case
+        if (nsp.IsExtractedType() && nsp.GetExeFS() != nullptr &&
+            FileSys::IsDirectoryExeFS(nsp.GetExeFS()) && nsp.GetRomFS() != nullptr) {
+            return FileType::NSP;
+        }
+
+        // Non-Ectracted Type case
+        if (!nsp.IsExtractedType() &&
+            nsp.GetNCA(nsp.GetFirstTitleID(), FileSys::ContentRecordType::Program) != nullptr &&
+            AppLoader_NCA::IdentifyType(nsp.GetNCAFile(
+                nsp.GetFirstTitleID(), FileSys::ContentRecordType::Program)) == FileType::NCA) {
+            return FileType::NSP;
+        }
+    }
+
+    return FileType::Error;
+}
+
+ResultStatus AppLoader_NSP::Load(Kernel::SharedPtr<Kernel::Process>& process) {
+    if (is_loaded) {
+        return ResultStatus::ErrorAlreadyLoaded;
+    }
+
+    if (nsp->IsExtractedType()) {
+        secondary_loader = std::make_unique<AppLoader_DeconstructedRomDirectory>(nsp->GetExeFS());
+    } else {
+        if (title_id == 0)
+            return ResultStatus::ErrorNSPMissingProgramNCA;
+
+        secondary_loader = std::make_unique<AppLoader_NCA>(
+            nsp->GetNCAFile(title_id, FileSys::ContentRecordType::Program));
+
+        if (nsp->GetStatus() != ResultStatus::Success)
+            return nsp->GetStatus();
+
+        if (nsp->GetProgramStatus(title_id) != ResultStatus::Success)
+            return nsp->GetProgramStatus(title_id);
+
+        if (nsp->GetNCA(title_id, FileSys::ContentRecordType::Program) == nullptr) {
+            if (!Core::Crypto::KeyManager::KeyFileExists(false))
+                return ResultStatus::ErrorMissingProductionKeyFile;
+            return ResultStatus::ErrorNSPMissingProgramNCA;
+        }
+    }
+
+    const auto result = secondary_loader->Load(process);
+    if (result != ResultStatus::Success)
+        return result;
+
+    is_loaded = true;
+
+    return ResultStatus::Success;
+}
+
+ResultStatus AppLoader_NSP::ReadRomFS(FileSys::VirtualFile& dir) {
+    return secondary_loader->ReadRomFS(dir);
+}
+
+ResultStatus AppLoader_NSP::ReadProgramId(u64& out_program_id) {
+    if (title_id == 0)
+        return ResultStatus::ErrorNotInitialized;
+    out_program_id = title_id;
+    return ResultStatus::Success;
+}
+
+ResultStatus AppLoader_NSP::ReadIcon(std::vector<u8>& buffer) {
+    if (icon_file == nullptr)
+        return ResultStatus::ErrorNoControl;
+    buffer = icon_file->ReadAllBytes();
+    return ResultStatus::Success;
+}
+
+ResultStatus AppLoader_NSP::ReadTitle(std::string& title) {
+    if (nacp_file == nullptr)
+        return ResultStatus::ErrorNoControl;
+    title = nacp_file->GetApplicationName();
+    return ResultStatus::Success;
+}
+} // namespace Loader

src/core/loader/nsp.h

@@ -0,0 +1,54 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <memory>
+#include "common/common_types.h"
+#include "core/file_sys/vfs.h"
+#include "core/loader/loader.h"
+
+namespace FileSys {
+class NACP;
+class NSP;
+} // namespace FileSys
+
+namespace Loader {
+
+class AppLoader_NCA;
+
+/// Loads an XCI file
+class AppLoader_NSP final : public AppLoader {
+public:
+    explicit AppLoader_NSP(FileSys::VirtualFile file);
+    ~AppLoader_NSP() override;
+
+    /**
+     * Returns the type of the file
+     * @param file std::shared_ptr<VfsFile> open file
+     * @return FileType found, or FileType::Error if this loader doesn't know it
+     */
+    static FileType IdentifyType(const FileSys::VirtualFile& file);
+
+    FileType GetFileType() override {
+        return IdentifyType(file);
+    }
+
+    ResultStatus Load(Kernel::SharedPtr<Kernel::Process>& process) override;
+
+    ResultStatus ReadRomFS(FileSys::VirtualFile& dir) override;
+    ResultStatus ReadProgramId(u64& out_program_id) override;
+    ResultStatus ReadIcon(std::vector<u8>& buffer) override;
+    ResultStatus ReadTitle(std::string& title) override;
+
+private:
+    std::unique_ptr<FileSys::NSP> nsp;
+    std::unique_ptr<AppLoader> secondary_loader;
+
+    FileSys::VirtualFile icon_file;
+    std::shared_ptr<FileSys::NACP> nacp_file;
+    u64 title_id;
+};
+
+} // namespace Loader

src/core/loader/xci.cpp

@@ -8,7 +8,9 @@
 #include "core/file_sys/card_image.h"
 #include "core/file_sys/content_archive.h"
 #include "core/file_sys/control_metadata.h"
+#include "core/file_sys/patch_manager.h"
 #include "core/file_sys/romfs.h"
+#include "core/file_sys/submission_package.h"
 #include "core/hle/kernel/process.h"
 #include "core/loader/nca.h"
 #include "core/loader/xci.h"
@@ -17,25 +19,16 @@ namespace Loader {
 
 AppLoader_XCI::AppLoader_XCI(FileSys::VirtualFile file)
     : AppLoader(file), xci(std::make_unique<FileSys::XCI>(file)),
-      nca_loader(std::make_unique<AppLoader_NCA>(
-          xci->GetNCAFileByType(FileSys::NCAContentType::Program))) {
+      nca_loader(std::make_unique<AppLoader_NCA>(xci->GetProgramNCAFile())) {
     if (xci->GetStatus() != ResultStatus::Success)
         return;
+
     const auto control_nca = xci->GetNCAByType(FileSys::NCAContentType::Control);
     if (control_nca == nullptr || control_nca->GetStatus() != ResultStatus::Success)
         return;
-    const auto romfs = FileSys::ExtractRomFS(control_nca->GetRomFS());
-    if (romfs == nullptr)
-        return;
-    for (const auto& language : FileSys::LANGUAGE_NAMES) {
-        icon_file = romfs->GetFile("icon_" + std::string(language) + ".dat");
-        if (icon_file != nullptr)
-            break;
-    }
-    const auto nacp_raw = romfs->GetFile("control.nacp");
-    if (nacp_raw == nullptr)
-        return;
-    nacp_file = std::make_shared<FileSys::NACP>(nacp_raw);
+
+    std::tie(nacp_file, icon_file) =
+        FileSys::PatchManager(xci->GetProgramTitleID()).ParseControlNCA(control_nca);
 }
 
 AppLoader_XCI::~AppLoader_XCI() = default;
@@ -64,11 +57,11 @@ ResultStatus AppLoader_XCI::Load(Kernel::SharedPtr<Kernel::Process>& process) {
     if (xci->GetProgramNCAStatus() != ResultStatus::Success)
         return xci->GetProgramNCAStatus();
 
-    const auto nca = xci->GetNCAFileByType(FileSys::NCAContentType::Program);
+    const auto nca = xci->GetProgramNCA();
     if (nca == nullptr && !Core::Crypto::KeyManager::KeyFileExists(false))
         return ResultStatus::ErrorMissingProductionKeyFile;
 
-    auto result = nca_loader->Load(process);
+    const auto result = nca_loader->Load(process);
     if (result != ResultStatus::Success)
         return result;
 

src/core/settings.h

@@ -127,6 +127,8 @@ struct Values {
 
     // Data Storage
     bool use_virtual_sd;
+    std::string nand_dir;
+    std::string sdmc_dir;
 
     // Renderer
     float resolution_factor;

src/core/telemetry_session.cpp

@@ -7,6 +7,8 @@
 #include "common/file_util.h"
 
 #include "core/core.h"
+#include "core/file_sys/control_metadata.h"
+#include "core/file_sys/patch_manager.h"
 #include "core/loader/loader.h"
 #include "core/settings.h"
 #include "core/telemetry_session.h"
@@ -88,12 +90,28 @@ TelemetrySession::TelemetrySession() {
                             std::chrono::system_clock::now().time_since_epoch())
                             .count()};
     AddField(Telemetry::FieldType::Session, "Init_Time", init_time);
-    std::string program_name;
-    const Loader::ResultStatus res{System::GetInstance().GetAppLoader().ReadTitle(program_name)};
+
+    u64 program_id{};
+    const Loader::ResultStatus res{System::GetInstance().GetAppLoader().ReadProgramId(program_id)};
     if (res == Loader::ResultStatus::Success) {
-        AddField(Telemetry::FieldType::Session, "ProgramName", program_name);
+        AddField(Telemetry::FieldType::Session, "ProgramId", program_id);
+
+        std::string name;
+        System::GetInstance().GetAppLoader().ReadTitle(name);
+
+        if (name.empty()) {
+            auto [nacp, icon_file] = FileSys::PatchManager(program_id).GetControlMetadata();
+            if (nacp != nullptr)
+                name = nacp->GetApplicationName();
+        }
+
+        if (!name.empty())
+            AddField(Telemetry::FieldType::Session, "ProgramName", name);
     }
 
+    AddField(Telemetry::FieldType::Session, "ProgramFormat",
+             static_cast<u8>(System::GetInstance().GetAppLoader().GetFileType()));
+
     // Log application information
     Telemetry::AppendBuildInfo(field_collection);
 

src/video_core/CMakeLists.txt

@@ -22,6 +22,8 @@ add_library(video_core STATIC
     rasterizer_interface.h
     renderer_base.cpp
     renderer_base.h
+    renderer_opengl/gl_buffer_cache.cpp
+    renderer_opengl/gl_buffer_cache.h
     renderer_opengl/gl_rasterizer.cpp
     renderer_opengl/gl_rasterizer.h
     renderer_opengl/gl_rasterizer_cache.cpp

src/video_core/command_processor.cpp

@@ -34,6 +34,8 @@ void GPU::WriteReg(u32 method, u32 subchannel, u32 value, u32 remaining_params)
               "{:08X} remaining params {}",
               method, subchannel, value, remaining_params);
 
+    ASSERT(subchannel < bound_engines.size());
+
     if (method == static_cast<u32>(BufferMethods::BindObject)) {
         // Bind the current subchannel to the desired engine id.
         LOG_DEBUG(HW_GPU, "Binding subchannel {} to engine {}", subchannel, value);
@@ -47,8 +49,6 @@ void GPU::WriteReg(u32 method, u32 subchannel, u32 value, u32 remaining_params)
         return;
     }
 
-    ASSERT(bound_engines.find(subchannel) != bound_engines.end());
-
     const EngineID engine = bound_engines[subchannel];
 
     switch (engine) {

src/video_core/engines/maxwell_3d.cpp

@@ -293,10 +293,6 @@ Texture::TICEntry Maxwell3D::GetTICEntry(u32 tic_index) const {
                    tic_entry.header_version == Texture::TICHeaderVersion::Pitch,
                "TIC versions other than BlockLinear or Pitch are unimplemented");
 
-    ASSERT_MSG((tic_entry.texture_type == Texture::TextureType::Texture2D) ||
-                   (tic_entry.texture_type == Texture::TextureType::Texture2DNoMipmap),
-               "Texture types other than Texture2D are unimplemented");
-
     auto r_type = tic_entry.r_type.Value();
     auto g_type = tic_entry.g_type.Value();
     auto b_type = tic_entry.b_type.Value();

src/video_core/engines/maxwell_3d.h

@@ -127,6 +127,7 @@ public:
                 BitField<21, 6, Size> size;
                 BitField<27, 3, Type> type;
                 BitField<31, 1, u32> bgra;
+                u32 hex;
             };
 
             u32 ComponentCount() const {
@@ -262,6 +263,10 @@ public:
             bool IsValid() const {
                 return size != Size::Invalid;
             }
+
+            bool operator<(const VertexAttribute& other) const {
+                return hex < other.hex;
+            }
         };
 
         enum class PrimitiveTopology : u32 {
@@ -545,7 +550,7 @@ public:
 
                 INSERT_PADDING_WORDS(0x5B);
 
-                VertexAttribute vertex_attrib_format[NumVertexAttributes];
+                std::array<VertexAttribute, NumVertexAttributes> vertex_attrib_format;
 
                 INSERT_PADDING_WORDS(0xF);
 
@@ -964,7 +969,7 @@ 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(zeta, 0x3F8);
-ASSERT_REG_POSITION(vertex_attrib_format[0], 0x458);
+ASSERT_REG_POSITION(vertex_attrib_format, 0x458);
 ASSERT_REG_POSITION(rt_control, 0x487);
 ASSERT_REG_POSITION(zeta_width, 0x48a);
 ASSERT_REG_POSITION(zeta_height, 0x48b);

src/video_core/engines/shader_bytecode.h

@@ -76,6 +76,7 @@ union Attribute {
         Position = 7,
         Attribute_0 = 8,
         Attribute_31 = 39,
+        PointCoord = 46,
         // This attribute contains a tuple of (~, ~, InstanceId, VertexId) when inside a vertex
         // shader, and a tuple of (TessCoord.x, TessCoord.y, TessCoord.z, ~) when inside a Tess Eval
         // shader.
@@ -246,6 +247,17 @@ enum class TextureType : u64 {
 enum class IpaInterpMode : u64 { Linear = 0, Perspective = 1, Flat = 2, Sc = 3 };
 enum class IpaSampleMode : u64 { Default = 0, Centroid = 1, Offset = 2 };
 
+struct IpaMode {
+    IpaInterpMode interpolation_mode;
+    IpaSampleMode sampling_mode;
+    inline bool operator==(const IpaMode& a) {
+        return (a.interpolation_mode == interpolation_mode) && (a.sampling_mode == sampling_mode);
+    }
+    inline bool operator!=(const IpaMode& a) {
+        return !((*this) == a);
+    }
+};
+
 union Instruction {
     Instruction& operator=(const Instruction& instr) {
         value = instr.value;

src/video_core/gpu.h

@@ -4,8 +4,8 @@
 
 #pragma once
 
+#include <array>
 #include <memory>
-#include <unordered_map>
 #include "common/common_types.h"
 #include "core/hle/service/nvflinger/buffer_queue.h"
 #include "video_core/memory_manager.h"
@@ -136,7 +136,7 @@ private:
     std::unique_ptr<Tegra::MemoryManager> memory_manager;
 
     /// Mapping of command subchannels to their bound engine ids.
-    std::unordered_map<u32, EngineID> bound_engines;
+    std::array<EngineID, 8> bound_engines = {};
 
     /// 3D engine
     std::unique_ptr<Engines::Maxwell3D> maxwell_3d;

src/video_core/renderer_opengl/gl_buffer_cache.cpp

@@ -0,0 +1,92 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <cstring>
+#include <memory>
+
+#include "common/alignment.h"
+#include "core/core.h"
+#include "core/memory.h"
+#include "video_core/renderer_opengl/gl_buffer_cache.h"
+
+namespace OpenGL {
+
+OGLBufferCache::OGLBufferCache(size_t size) : stream_buffer(GL_ARRAY_BUFFER, size) {}
+
+GLintptr OGLBufferCache::UploadMemory(Tegra::GPUVAddr gpu_addr, size_t size, size_t alignment,
+                                      bool cache) {
+    auto& memory_manager = Core::System::GetInstance().GPU().MemoryManager();
+    const boost::optional<VAddr> cpu_addr{memory_manager.GpuToCpuAddress(gpu_addr)};
+
+    // Cache management is a big overhead, so only cache entries with a given size.
+    // TODO: Figure out which size is the best for given games.
+    cache &= size >= 2048;
+
+    if (cache) {
+        auto entry = TryGet(*cpu_addr);
+        if (entry) {
+            if (entry->size >= size && entry->alignment == alignment) {
+                return entry->offset;
+            }
+            Unregister(entry);
+        }
+    }
+
+    AlignBuffer(alignment);
+    GLintptr uploaded_offset = buffer_offset;
+
+    Memory::ReadBlock(*cpu_addr, buffer_ptr, size);
+
+    buffer_ptr += size;
+    buffer_offset += size;
+
+    if (cache) {
+        auto entry = std::make_shared<CachedBufferEntry>();
+        entry->offset = uploaded_offset;
+        entry->size = size;
+        entry->alignment = alignment;
+        entry->addr = *cpu_addr;
+        Register(entry);
+    }
+
+    return uploaded_offset;
+}
+
+GLintptr OGLBufferCache::UploadHostMemory(const void* raw_pointer, size_t size, size_t alignment) {
+    AlignBuffer(alignment);
+    std::memcpy(buffer_ptr, raw_pointer, size);
+    GLintptr uploaded_offset = buffer_offset;
+
+    buffer_ptr += size;
+    buffer_offset += size;
+    return uploaded_offset;
+}
+
+void OGLBufferCache::Map(size_t max_size) {
+    bool invalidate;
+    std::tie(buffer_ptr, buffer_offset_base, invalidate) =
+        stream_buffer.Map(static_cast<GLsizeiptr>(max_size), 4);
+    buffer_offset = buffer_offset_base;
+
+    if (invalidate) {
+        InvalidateAll();
+    }
+}
+void OGLBufferCache::Unmap() {
+    stream_buffer.Unmap(buffer_offset - buffer_offset_base);
+}
+
+GLuint OGLBufferCache::GetHandle() const {
+    return stream_buffer.GetHandle();
+}
+
+void OGLBufferCache::AlignBuffer(size_t alignment) {
+    // Align the offset, not the mapped pointer
+    GLintptr offset_aligned =
+        static_cast<GLintptr>(Common::AlignUp(static_cast<size_t>(buffer_offset), alignment));
+    buffer_ptr += offset_aligned - buffer_offset;
+    buffer_offset = offset_aligned;
+}
+
+} // namespace OpenGL

src/video_core/renderer_opengl/gl_buffer_cache.h

@@ -0,0 +1,57 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <cstddef>
+#include <memory>
+
+#include "common/common_types.h"
+#include "video_core/rasterizer_cache.h"
+#include "video_core/renderer_opengl/gl_resource_manager.h"
+#include "video_core/renderer_opengl/gl_stream_buffer.h"
+
+namespace OpenGL {
+
+struct CachedBufferEntry final {
+    VAddr GetAddr() const {
+        return addr;
+    }
+
+    size_t GetSizeInBytes() const {
+        return size;
+    }
+
+    VAddr addr;
+    size_t size;
+    GLintptr offset;
+    size_t alignment;
+};
+
+class OGLBufferCache final : public RasterizerCache<std::shared_ptr<CachedBufferEntry>> {
+public:
+    explicit OGLBufferCache(size_t size);
+
+    GLintptr UploadMemory(Tegra::GPUVAddr gpu_addr, size_t size, size_t alignment = 4,
+                          bool cache = true);
+
+    GLintptr UploadHostMemory(const void* raw_pointer, size_t size, size_t alignment = 4);
+
+    void Map(size_t max_size);
+    void Unmap();
+
+    GLuint GetHandle() const;
+
+protected:
+    void AlignBuffer(size_t alignment);
+
+private:
+    OGLStreamBuffer stream_buffer;
+
+    u8* buffer_ptr = nullptr;
+    GLintptr buffer_offset = 0;
+    GLintptr buffer_offset_base = 0;
+};
+
+} // namespace OpenGL

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -33,14 +33,17 @@ using PixelFormat = SurfaceParams::PixelFormat;
 using SurfaceType = SurfaceParams::SurfaceType;
 
 MICROPROFILE_DEFINE(OpenGL_VAO, "OpenGL", "Vertex Array Setup", MP_RGB(128, 128, 192));
-MICROPROFILE_DEFINE(OpenGL_VS, "OpenGL", "Vertex Shader Setup", MP_RGB(128, 128, 192));
-MICROPROFILE_DEFINE(OpenGL_FS, "OpenGL", "Fragment Shader 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));
+MICROPROFILE_DEFINE(OpenGL_Texture, "OpenGL", "Texture Setup", MP_RGB(128, 128, 192));
+MICROPROFILE_DEFINE(OpenGL_Framebuffer, "OpenGL", "Framebuffer Setup", MP_RGB(128, 128, 192));
 MICROPROFILE_DEFINE(OpenGL_Drawing, "OpenGL", "Drawing", MP_RGB(128, 128, 192));
-MICROPROFILE_DEFINE(OpenGL_Blits, "OpenGL", "Blits", MP_RGB(100, 100, 255));
+MICROPROFILE_DEFINE(OpenGL_Blits, "OpenGL", "Blits", MP_RGB(128, 128, 192));
 MICROPROFILE_DEFINE(OpenGL_CacheManagement, "OpenGL", "Cache Mgmt", MP_RGB(100, 255, 100));
 
 RasterizerOpenGL::RasterizerOpenGL(Core::Frontend::EmuWindow& window, ScreenInfo& info)
-    : emu_window{window}, screen_info{info}, stream_buffer(GL_ARRAY_BUFFER, STREAM_BUFFER_SIZE) {
+    : emu_window{window}, screen_info{info}, buffer_cache(STREAM_BUFFER_SIZE) {
     // Create sampler objects
     for (size_t i = 0; i < texture_samplers.size(); ++i) {
         texture_samplers[i].Create();
@@ -67,28 +70,13 @@ RasterizerOpenGL::RasterizerOpenGL(Core::Frontend::EmuWindow& window, ScreenInfo
     // Clipping plane 0 is always enabled for PICA fixed clip plane z <= 0
     state.clip_distance[0] = true;
 
-    // Generate VAO and UBO
-    sw_vao.Create();
-    uniform_buffer.Create();
-
-    state.draw.vertex_array = sw_vao.handle;
-    state.draw.uniform_buffer = uniform_buffer.handle;
-    state.Apply();
-
     // Create render framebuffer
     framebuffer.Create();
 
-    hw_vao.Create();
-
-    state.draw.vertex_buffer = stream_buffer.GetHandle();
-
     shader_program_manager = std::make_unique<GLShader::ProgramManager>();
     state.draw.shader_program = 0;
-    state.draw.vertex_array = hw_vao.handle;
     state.Apply();
 
-    glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, stream_buffer.GetHandle());
-
     glEnable(GL_BLEND);
 
     glGetIntegerv(GL_UNIFORM_BUFFER_OFFSET_ALIGNMENT, &uniform_buffer_alignment);
@@ -98,14 +86,60 @@ RasterizerOpenGL::RasterizerOpenGL(Core::Frontend::EmuWindow& window, ScreenInfo
 
 RasterizerOpenGL::~RasterizerOpenGL() {}
 
-std::pair<u8*, GLintptr> RasterizerOpenGL::SetupVertexArrays(u8* array_ptr,
-                                                             GLintptr buffer_offset) {
+void RasterizerOpenGL::SetupVertexArrays() {
     MICROPROFILE_SCOPE(OpenGL_VAO);
     const auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
     const auto& regs = gpu.regs;
 
-    state.draw.vertex_array = hw_vao.handle;
-    state.draw.vertex_buffer = stream_buffer.GetHandle();
+    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();
+
+        // The index buffer binding is stored within the VAO. Stupid OpenGL, but easy to work
+        // around.
+        glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, buffer_cache.GetHandle());
+
+        // Use the vertex array as-is, assumes that the data is formatted correctly for OpenGL.
+        // Enables the first 16 vertex attributes always, as we don't know which ones are actually
+        // used until shader time. Note, Tegra technically supports 32, but we're capping this to 16
+        // for now to avoid OpenGL errors.
+        // TODO(Subv): Analyze the shader to identify which attributes are actually used and don't
+        // assume every shader uses them all.
+        for (unsigned index = 0; index < 16; ++index) {
+            const auto& attrib = regs.vertex_attrib_format[index];
+
+            // Ignore invalid attributes.
+            if (!attrib.IsValid())
+                continue;
+
+            const auto& buffer = regs.vertex_array[attrib.buffer];
+            LOG_TRACE(HW_GPU,
+                      "vertex attrib {}, count={}, size={}, type={}, offset={}, normalize={}",
+                      index, attrib.ComponentCount(), attrib.SizeString(), attrib.TypeString(),
+                      attrib.offset.Value(), attrib.IsNormalized());
+
+            ASSERT(buffer.IsEnabled());
+
+            glEnableVertexAttribArray(index);
+            if (attrib.type == Tegra::Engines::Maxwell3D::Regs::VertexAttribute::Type::SignedInt ||
+                attrib.type ==
+                    Tegra::Engines::Maxwell3D::Regs::VertexAttribute::Type::UnsignedInt) {
+                glVertexAttribIFormat(index, attrib.ComponentCount(),
+                                      MaxwellToGL::VertexType(attrib), attrib.offset);
+            } else {
+                glVertexAttribFormat(index, attrib.ComponentCount(),
+                                     MaxwellToGL::VertexType(attrib),
+                                     attrib.IsNormalized() ? GL_TRUE : GL_FALSE, attrib.offset);
+            }
+            glVertexAttribBinding(index, attrib.buffer);
+        }
+    }
+    state.draw.vertex_array = VAO.handle;
+    state.draw.vertex_buffer = buffer_cache.GetHandle();
     state.Apply();
 
     // Upload all guest vertex arrays sequentially to our buffer
@@ -117,69 +151,27 @@ std::pair<u8*, GLintptr> RasterizerOpenGL::SetupVertexArrays(u8* array_ptr,
         Tegra::GPUVAddr start = vertex_array.StartAddress();
         const Tegra::GPUVAddr end = regs.vertex_array_limit[index].LimitAddress();
 
-        if (regs.instanced_arrays.IsInstancingEnabled(index) && vertex_array.divisor != 0) {
-            start += vertex_array.stride * (gpu.state.current_instance / vertex_array.divisor);
-        }
-
         ASSERT(end > start);
-        u64 size = end - start + 1;
-
-        GLintptr vertex_buffer_offset;
-        std::tie(array_ptr, buffer_offset, vertex_buffer_offset) =
-            UploadMemory(array_ptr, buffer_offset, start, size);
+        const u64 size = end - start + 1;
+        const GLintptr vertex_buffer_offset = buffer_cache.UploadMemory(start, size);
 
         // Bind the vertex array to the buffer at the current offset.
-        glBindVertexBuffer(index, stream_buffer.GetHandle(), vertex_buffer_offset,
+        glBindVertexBuffer(index, buffer_cache.GetHandle(), vertex_buffer_offset,
                            vertex_array.stride);
 
         if (regs.instanced_arrays.IsInstancingEnabled(index) && vertex_array.divisor != 0) {
-            // Tell OpenGL that this is an instanced vertex buffer to prevent accessing different
-            // indexes on each vertex. We do the instance indexing manually by incrementing the
-            // start address of the vertex buffer.
-            glVertexBindingDivisor(index, 1);
+            // Enable vertex buffer instancing with the specified divisor.
+            glVertexBindingDivisor(index, vertex_array.divisor);
         } else {
             // Disable the vertex buffer instancing.
             glVertexBindingDivisor(index, 0);
         }
     }
-
-    // Use the vertex array as-is, assumes that the data is formatted correctly for OpenGL.
-    // Enables the first 16 vertex attributes always, as we don't know which ones are actually used
-    // until shader time. Note, Tegra technically supports 32, but we're capping this to 16 for now
-    // to avoid OpenGL errors.
-    // TODO(Subv): Analyze the shader to identify which attributes are actually used and don't
-    // assume every shader uses them all.
-    for (unsigned index = 0; index < 16; ++index) {
-        auto& attrib = regs.vertex_attrib_format[index];
-
-        // Ignore invalid attributes.
-        if (!attrib.IsValid())
-            continue;
-
-        auto& buffer = regs.vertex_array[attrib.buffer];
-        LOG_TRACE(HW_GPU, "vertex attrib {}, count={}, size={}, type={}, offset={}, normalize={}",
-                  index, attrib.ComponentCount(), attrib.SizeString(), attrib.TypeString(),
-                  attrib.offset.Value(), attrib.IsNormalized());
-
-        ASSERT(buffer.IsEnabled());
-
-        glEnableVertexAttribArray(index);
-        if (attrib.type == Tegra::Engines::Maxwell3D::Regs::VertexAttribute::Type::SignedInt ||
-            attrib.type == Tegra::Engines::Maxwell3D::Regs::VertexAttribute::Type::UnsignedInt) {
-            glVertexAttribIFormat(index, attrib.ComponentCount(), MaxwellToGL::VertexType(attrib),
-                                  attrib.offset);
-        } else {
-            glVertexAttribFormat(index, attrib.ComponentCount(), MaxwellToGL::VertexType(attrib),
-                                 attrib.IsNormalized() ? GL_TRUE : GL_FALSE, attrib.offset);
-        }
-        glVertexAttribBinding(index, attrib.buffer);
-    }
-
-    return {array_ptr, buffer_offset};
 }
 
-std::pair<u8*, GLintptr> RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset) {
-    auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
+void RasterizerOpenGL::SetupShaders() {
+    MICROPROFILE_SCOPE(OpenGL_Shader);
+    const auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
 
     // Next available bindpoints to use when uploading the const buffers and textures to the GLSL
     // shaders. The constbuffer bindpoint starts after the shader stage configuration bind points.
@@ -187,7 +179,7 @@ std::pair<u8*, GLintptr> RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr
     u32 current_texture_bindpoint = 0;
 
     for (size_t index = 0; index < Maxwell::MaxShaderProgram; ++index) {
-        auto& shader_config = gpu.regs.shader_config[index];
+        const auto& shader_config = gpu.regs.shader_config[index];
         const Maxwell::ShaderProgram program{static_cast<Maxwell::ShaderProgram>(index)};
 
         // Skip stages that are not enabled
@@ -195,21 +187,15 @@ std::pair<u8*, GLintptr> RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr
             continue;
         }
 
-        std::tie(buffer_ptr, buffer_offset) =
-            AlignBuffer(buffer_ptr, buffer_offset, static_cast<size_t>(uniform_buffer_alignment));
-
         const size_t stage{index == 0 ? 0 : index - 1}; // Stage indices are 0 - 5
 
         GLShader::MaxwellUniformData ubo{};
         ubo.SetFromRegs(gpu.state.shader_stages[stage]);
-        std::memcpy(buffer_ptr, &ubo, sizeof(ubo));
+        const GLintptr offset = buffer_cache.UploadHostMemory(
+            &ubo, sizeof(ubo), static_cast<size_t>(uniform_buffer_alignment));
 
         // Bind the buffer
-        glBindBufferRange(GL_UNIFORM_BUFFER, stage, stream_buffer.GetHandle(), buffer_offset,
-                          sizeof(ubo));
-
-        buffer_ptr += sizeof(ubo);
-        buffer_offset += sizeof(ubo);
+        glBindBufferRange(GL_UNIFORM_BUFFER, stage, buffer_cache.GetHandle(), offset, sizeof(ubo));
 
         Shader shader{shader_cache.GetStageProgram(program)};
 
@@ -230,9 +216,8 @@ std::pair<u8*, GLintptr> RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr
         }
 
         // Configure the const buffers for this shader stage.
-        std::tie(buffer_ptr, buffer_offset, current_constbuffer_bindpoint) =
-            SetupConstBuffers(buffer_ptr, buffer_offset, static_cast<Maxwell::ShaderStage>(stage),
-                              shader, current_constbuffer_bindpoint);
+        current_constbuffer_bindpoint = SetupConstBuffers(static_cast<Maxwell::ShaderStage>(stage),
+                                                          shader, current_constbuffer_bindpoint);
 
         // Configure the textures for this shader stage.
         current_texture_bindpoint = SetupTextures(static_cast<Maxwell::ShaderStage>(stage), shader,
@@ -245,9 +230,9 @@ std::pair<u8*, GLintptr> RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr
         }
     }
 
-    shader_program_manager->UseTrivialGeometryShader();
+    state.Apply();
 
-    return {buffer_ptr, buffer_offset};
+    shader_program_manager->UseTrivialGeometryShader();
 }
 
 size_t RasterizerOpenGL::CalculateVertexArraysSize() const {
@@ -312,6 +297,7 @@ void RasterizerOpenGL::UpdatePagesCachedCount(VAddr addr, u64 size, int delta) {
 std::pair<Surface, Surface> RasterizerOpenGL::ConfigureFramebuffers(bool using_color_fb,
                                                                     bool using_depth_fb,
                                                                     bool preserve_contents) {
+    MICROPROFILE_SCOPE(OpenGL_Framebuffer);
     const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
 
     if (regs.rt[0].format == Tegra::RenderTargetFormat::NONE) {
@@ -434,41 +420,17 @@ void RasterizerOpenGL::Clear() {
     glClear(clear_mask);
 }
 
-std::pair<u8*, GLintptr> RasterizerOpenGL::AlignBuffer(u8* buffer_ptr, GLintptr buffer_offset,
-                                                       size_t alignment) {
-    // Align the offset, not the mapped pointer
-    GLintptr offset_aligned =
-        static_cast<GLintptr>(Common::AlignUp(static_cast<size_t>(buffer_offset), alignment));
-    return {buffer_ptr + (offset_aligned - buffer_offset), offset_aligned};
-}
-
-std::tuple<u8*, GLintptr, GLintptr> RasterizerOpenGL::UploadMemory(u8* buffer_ptr,
-                                                                   GLintptr buffer_offset,
-                                                                   Tegra::GPUVAddr gpu_addr,
-                                                                   size_t size, size_t alignment) {
-    std::tie(buffer_ptr, buffer_offset) = AlignBuffer(buffer_ptr, buffer_offset, alignment);
-    GLintptr uploaded_offset = buffer_offset;
-
-    auto& memory_manager = Core::System::GetInstance().GPU().MemoryManager();
-    const boost::optional<VAddr> cpu_addr{memory_manager.GpuToCpuAddress(gpu_addr)};
-    Memory::ReadBlock(*cpu_addr, buffer_ptr, size);
-
-    buffer_ptr += size;
-    buffer_offset += size;
-
-    return {buffer_ptr, buffer_offset, uploaded_offset};
-}
-
 void RasterizerOpenGL::DrawArrays() {
     if (accelerate_draw == AccelDraw::Disabled)
         return;
 
     MICROPROFILE_SCOPE(OpenGL_Drawing);
-    const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+    const auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
+    const auto& regs = gpu.regs;
 
     ScopeAcquireGLContext acquire_context{emu_window};
 
-    auto [dirty_color_surface, dirty_depth_surface] =
+    const auto [dirty_color_surface, dirty_depth_surface] =
         ConfigureFramebuffers(true, regs.zeta.Address() != 0 && regs.zeta_enable != 0, true);
 
     SyncDepthTestState();
@@ -482,9 +444,10 @@ void RasterizerOpenGL::DrawArrays() {
 
     // Draw the vertex batch
     const bool is_indexed = accelerate_draw == AccelDraw::Indexed;
-    const u64 index_buffer_size{regs.index_array.count * regs.index_array.FormatSizeInBytes()};
+    const u64 index_buffer_size{static_cast<u64>(regs.index_array.count) *
+                                static_cast<u64>(regs.index_array.FormatSizeInBytes())};
 
-    state.draw.vertex_buffer = stream_buffer.GetHandle();
+    state.draw.vertex_buffer = buffer_cache.GetHandle();
     state.Apply();
 
     size_t buffer_size = CalculateVertexArraysSize();
@@ -501,24 +464,21 @@ void RasterizerOpenGL::DrawArrays() {
     // Add space for at least 18 constant buffers
     buffer_size += Maxwell::MaxConstBuffers * (MaxConstbufferSize + uniform_buffer_alignment);
 
-    u8* buffer_ptr;
-    GLintptr buffer_offset;
-    std::tie(buffer_ptr, buffer_offset, std::ignore) =
-        stream_buffer.Map(static_cast<GLsizeiptr>(buffer_size), 4);
-    u8* buffer_ptr_base = buffer_ptr;
+    buffer_cache.Map(buffer_size);
 
-    std::tie(buffer_ptr, buffer_offset) = SetupVertexArrays(buffer_ptr, buffer_offset);
+    SetupVertexArrays();
 
     // If indexed mode, copy the index buffer
     GLintptr index_buffer_offset = 0;
     if (is_indexed) {
-        std::tie(buffer_ptr, buffer_offset, index_buffer_offset) = UploadMemory(
-            buffer_ptr, buffer_offset, regs.index_array.StartAddress(), index_buffer_size);
+        MICROPROFILE_SCOPE(OpenGL_Index);
+        index_buffer_offset =
+            buffer_cache.UploadMemory(regs.index_array.StartAddress(), index_buffer_size);
     }
 
-    std::tie(buffer_ptr, buffer_offset) = SetupShaders(buffer_ptr, buffer_offset);
+    SetupShaders();
 
-    stream_buffer.Unmap(buffer_ptr - buffer_ptr_base);
+    buffer_cache.Unmap();
 
     shader_program_manager->ApplyTo(state);
     state.Apply();
@@ -528,13 +488,29 @@ void RasterizerOpenGL::DrawArrays() {
         const GLint base_vertex{static_cast<GLint>(regs.vb_element_base)};
 
         // Adjust the index buffer offset so it points to the first desired index.
-        index_buffer_offset += regs.index_array.first * regs.index_array.FormatSizeInBytes();
+        index_buffer_offset += static_cast<GLintptr>(regs.index_array.first) *
+                               static_cast<GLintptr>(regs.index_array.FormatSizeInBytes());
 
-        glDrawElementsBaseVertex(primitive_mode, regs.index_array.count,
-                                 MaxwellToGL::IndexFormat(regs.index_array.format),
-                                 reinterpret_cast<const void*>(index_buffer_offset), base_vertex);
+        if (gpu.state.current_instance > 0) {
+            glDrawElementsInstancedBaseVertexBaseInstance(
+                primitive_mode, regs.index_array.count,
+                MaxwellToGL::IndexFormat(regs.index_array.format),
+                reinterpret_cast<const void*>(index_buffer_offset), 1, base_vertex,
+                gpu.state.current_instance);
+        } else {
+            glDrawElementsBaseVertex(primitive_mode, regs.index_array.count,
+                                     MaxwellToGL::IndexFormat(regs.index_array.format),
+                                     reinterpret_cast<const void*>(index_buffer_offset),
+                                     base_vertex);
+        }
     } else {
-        glDrawArrays(primitive_mode, regs.vertex_buffer.first, regs.vertex_buffer.count);
+        if (gpu.state.current_instance > 0) {
+            glDrawArraysInstancedBaseInstance(primitive_mode, regs.vertex_buffer.first,
+                                              regs.vertex_buffer.count, 1,
+                                              gpu.state.current_instance);
+        } else {
+            glDrawArrays(primitive_mode, regs.vertex_buffer.first, regs.vertex_buffer.count);
+        }
     }
 
     // Disable scissor test
@@ -551,22 +527,18 @@ void RasterizerOpenGL::DrawArrays() {
 
 void RasterizerOpenGL::NotifyMaxwellRegisterChanged(u32 method) {}
 
-void RasterizerOpenGL::FlushAll() {
-    MICROPROFILE_SCOPE(OpenGL_CacheManagement);
-}
+void RasterizerOpenGL::FlushAll() {}
 
-void RasterizerOpenGL::FlushRegion(VAddr addr, u64 size) {
-    MICROPROFILE_SCOPE(OpenGL_CacheManagement);
-}
+void RasterizerOpenGL::FlushRegion(VAddr addr, u64 size) {}
 
 void RasterizerOpenGL::InvalidateRegion(VAddr addr, u64 size) {
     MICROPROFILE_SCOPE(OpenGL_CacheManagement);
     res_cache.InvalidateRegion(addr, size);
     shader_cache.InvalidateRegion(addr, size);
+    buffer_cache.InvalidateRegion(addr, size);
 }
 
 void RasterizerOpenGL::FlushAndInvalidateRegion(VAddr addr, u64 size) {
-    MICROPROFILE_SCOPE(OpenGL_CacheManagement);
     InvalidateRegion(addr, size);
 }
 
@@ -614,7 +586,7 @@ bool RasterizerOpenGL::AccelerateDisplay(const Tegra::FramebufferConfig& config,
 void RasterizerOpenGL::SamplerInfo::Create() {
     sampler.Create();
     mag_filter = min_filter = Tegra::Texture::TextureFilter::Linear;
-    wrap_u = wrap_v = Tegra::Texture::WrapMode::Wrap;
+    wrap_u = wrap_v = wrap_p = Tegra::Texture::WrapMode::Wrap;
 
     // default is GL_LINEAR_MIPMAP_LINEAR
     glSamplerParameteri(sampler.handle, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
@@ -622,7 +594,7 @@ void RasterizerOpenGL::SamplerInfo::Create() {
 }
 
 void RasterizerOpenGL::SamplerInfo::SyncWithConfig(const Tegra::Texture::TSCEntry& config) {
-    GLuint s = sampler.handle;
+    const GLuint s = sampler.handle;
 
     if (mag_filter != config.mag_filter) {
         mag_filter = config.mag_filter;
@@ -641,8 +613,13 @@ void RasterizerOpenGL::SamplerInfo::SyncWithConfig(const Tegra::Texture::TSCEntr
         wrap_v = config.wrap_v;
         glSamplerParameteri(s, GL_TEXTURE_WRAP_T, MaxwellToGL::WrapMode(wrap_v));
     }
+    if (wrap_p != config.wrap_p) {
+        wrap_p = config.wrap_p;
+        glSamplerParameteri(s, GL_TEXTURE_WRAP_R, MaxwellToGL::WrapMode(wrap_p));
+    }
 
-    if (wrap_u == Tegra::Texture::WrapMode::Border || wrap_v == Tegra::Texture::WrapMode::Border) {
+    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) {
@@ -652,11 +629,9 @@ void RasterizerOpenGL::SamplerInfo::SyncWithConfig(const Tegra::Texture::TSCEntr
     }
 }
 
-std::tuple<u8*, GLintptr, u32> RasterizerOpenGL::SetupConstBuffers(u8* buffer_ptr,
-                                                                   GLintptr buffer_offset,
-                                                                   Maxwell::ShaderStage stage,
-                                                                   Shader& shader,
-                                                                   u32 current_bindpoint) {
+u32 RasterizerOpenGL::SetupConstBuffers(Maxwell::ShaderStage stage, Shader& shader,
+                                        u32 current_bindpoint) {
+    MICROPROFILE_SCOPE(OpenGL_UBO);
     const auto& gpu = Core::System::GetInstance().GPU();
     const auto& maxwell3d = gpu.Maxwell3D();
     const auto& shader_stage = maxwell3d.state.shader_stages[static_cast<size_t>(stage)];
@@ -692,26 +667,23 @@ std::tuple<u8*, GLintptr, u32> RasterizerOpenGL::SetupConstBuffers(u8* buffer_pt
         size = Common::AlignUp(size, sizeof(GLvec4));
         ASSERT_MSG(size <= MaxConstbufferSize, "Constbuffer too big");
 
-        GLintptr const_buffer_offset;
-        std::tie(buffer_ptr, buffer_offset, const_buffer_offset) =
-            UploadMemory(buffer_ptr, buffer_offset, buffer.address, size,
-                         static_cast<size_t>(uniform_buffer_alignment));
+        GLintptr const_buffer_offset = buffer_cache.UploadMemory(
+            buffer.address, size, static_cast<size_t>(uniform_buffer_alignment));
 
         glBindBufferRange(GL_UNIFORM_BUFFER, current_bindpoint + bindpoint,
-                          stream_buffer.GetHandle(), const_buffer_offset, size);
+                          buffer_cache.GetHandle(), const_buffer_offset, size);
 
         // Now configure the bindpoint of the buffer inside the shader
         glUniformBlockBinding(shader->GetProgramHandle(),
-                              shader->GetProgramResourceIndex(used_buffer.GetName()),
+                              shader->GetProgramResourceIndex(used_buffer),
                               current_bindpoint + bindpoint);
     }
 
-    state.Apply();
-
-    return {buffer_ptr, buffer_offset, current_bindpoint + static_cast<u32>(entries.size())};
+    return current_bindpoint + static_cast<u32>(entries.size());
 }
 
 u32 RasterizerOpenGL::SetupTextures(Maxwell::ShaderStage stage, Shader& shader, u32 current_unit) {
+    MICROPROFILE_SCOPE(OpenGL_Texture);
     const auto& gpu = Core::System::GetInstance().GPU();
     const auto& maxwell3d = gpu.Maxwell3D();
     const auto& entries = shader->GetShaderEntries().texture_samplers;
@@ -721,24 +693,25 @@ u32 RasterizerOpenGL::SetupTextures(Maxwell::ShaderStage stage, Shader& shader,
 
     for (u32 bindpoint = 0; bindpoint < entries.size(); ++bindpoint) {
         const auto& entry = entries[bindpoint];
-        u32 current_bindpoint = current_unit + bindpoint;
+        const u32 current_bindpoint = current_unit + bindpoint;
 
         // Bind the uniform to the sampler.
 
-        glProgramUniform1i(shader->GetProgramHandle(), shader->GetUniformLocation(entry.GetName()),
+        glProgramUniform1i(shader->GetProgramHandle(), shader->GetUniformLocation(entry),
                            current_bindpoint);
 
         const auto texture = maxwell3d.GetStageTexture(entry.GetStage(), entry.GetOffset());
 
         if (!texture.enabled) {
-            state.texture_units[current_bindpoint].texture_2d = 0;
+            state.texture_units[current_bindpoint].texture = 0;
             continue;
         }
 
         texture_samplers[current_bindpoint].SyncWithConfig(texture.tsc);
         Surface surface = res_cache.GetTextureSurface(texture);
         if (surface != nullptr) {
-            state.texture_units[current_bindpoint].texture_2d = surface->Texture().handle;
+            state.texture_units[current_bindpoint].texture = surface->Texture().handle;
+            state.texture_units[current_bindpoint].target = surface->Target();
             state.texture_units[current_bindpoint].swizzle.r =
                 MaxwellToGL::SwizzleSource(texture.tic.x_source);
             state.texture_units[current_bindpoint].swizzle.g =
@@ -749,12 +722,10 @@ u32 RasterizerOpenGL::SetupTextures(Maxwell::ShaderStage stage, Shader& shader,
                 MaxwellToGL::SwizzleSource(texture.tic.w_source);
         } else {
             // Can occur when texture addr is null or its memory is unmapped/invalid
-            state.texture_units[current_bindpoint].texture_2d = 0;
+            state.texture_units[current_bindpoint].texture = 0;
         }
     }
 
-    state.Apply();
-
     return current_unit + static_cast<u32>(entries.size());
 }
 

src/video_core/renderer_opengl/gl_rasterizer.h

@@ -6,6 +6,7 @@
 
 #include <array>
 #include <cstddef>
+#include <map>
 #include <memory>
 #include <tuple>
 #include <utility>
@@ -18,7 +19,9 @@
 #include "common/common_types.h"
 #include "video_core/engines/maxwell_3d.h"
 #include "video_core/memory_manager.h"
+#include "video_core/rasterizer_cache.h"
 #include "video_core/rasterizer_interface.h"
+#include "video_core/renderer_opengl/gl_buffer_cache.h"
 #include "video_core/renderer_opengl/gl_rasterizer_cache.h"
 #include "video_core/renderer_opengl/gl_resource_manager.h"
 #include "video_core/renderer_opengl/gl_shader_cache.h"
@@ -90,6 +93,7 @@ private:
         Tegra::Texture::TextureFilter min_filter;
         Tegra::Texture::WrapMode wrap_u;
         Tegra::Texture::WrapMode wrap_v;
+        Tegra::Texture::WrapMode wrap_p;
         GLvec4 border_color;
     };
 
@@ -109,9 +113,8 @@ private:
      * @param current_bindpoint The offset at which to start counting new buffer bindpoints.
      * @returns The next available bindpoint for use in the next shader stage.
      */
-    std::tuple<u8*, GLintptr, u32> SetupConstBuffers(
-        u8* buffer_ptr, GLintptr buffer_offset, Tegra::Engines::Maxwell3D::Regs::ShaderStage stage,
-        Shader& shader, u32 current_bindpoint);
+    u32 SetupConstBuffers(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage, Shader& shader,
+                          u32 current_bindpoint);
 
     /*
      * Configures the current textures to use for the draw command.
@@ -167,28 +170,23 @@ private:
     ScreenInfo& screen_info;
 
     std::unique_ptr<GLShader::ProgramManager> shader_program_manager;
-    OGLVertexArray sw_vao;
-    OGLVertexArray hw_vao;
+    std::map<std::array<Tegra::Engines::Maxwell3D::Regs::VertexAttribute,
+                        Tegra::Engines::Maxwell3D::Regs::NumVertexAttributes>,
+             OGLVertexArray>
+        vertex_array_cache;
 
     std::array<SamplerInfo, GLShader::NumTextureSamplers> texture_samplers;
 
     static constexpr size_t STREAM_BUFFER_SIZE = 128 * 1024 * 1024;
-    OGLStreamBuffer stream_buffer;
-    OGLBuffer uniform_buffer;
+    OGLBufferCache buffer_cache;
     OGLFramebuffer framebuffer;
     GLint uniform_buffer_alignment;
 
     size_t CalculateVertexArraysSize() const;
 
-    std::pair<u8*, GLintptr> SetupVertexArrays(u8* array_ptr, GLintptr buffer_offset);
+    void SetupVertexArrays();
 
-    std::pair<u8*, GLintptr> SetupShaders(u8* buffer_ptr, GLintptr buffer_offset);
-
-    std::pair<u8*, GLintptr> AlignBuffer(u8* buffer_ptr, GLintptr buffer_offset, size_t alignment);
-
-    std::tuple<u8*, GLintptr, GLintptr> UploadMemory(u8* buffer_ptr, GLintptr buffer_offset,
-                                                     Tegra::GPUVAddr gpu_addr, size_t size,
-                                                     size_t alignment = 4);
+    void SetupShaders();
 
     enum class AccelDraw { Disabled, Arrays, Indexed };
     AccelDraw accelerate_draw = AccelDraw::Disabled;

src/video_core/renderer_opengl/gl_rasterizer_cache.cpp

@@ -7,6 +7,7 @@
 
 #include "common/alignment.h"
 #include "common/assert.h"
+#include "common/logging/log.h"
 #include "common/microprofile.h"
 #include "common/scope_exit.h"
 #include "core/core.h"
@@ -51,10 +52,12 @@ static VAddr TryGetCpuAddr(Tegra::GPUVAddr gpu_addr) {
     params.type = GetFormatType(params.pixel_format);
     params.width = Common::AlignUp(config.tic.Width(), GetCompressionFactor(params.pixel_format));
     params.height = Common::AlignUp(config.tic.Height(), GetCompressionFactor(params.pixel_format));
+    params.depth = config.tic.Depth();
     params.unaligned_height = config.tic.Height();
     params.size_in_bytes = params.SizeInBytes();
-    params.cache_width = Common::AlignUp(params.width, 16);
-    params.cache_height = Common::AlignUp(params.height, 16);
+    params.cache_width = Common::AlignUp(params.width, 8);
+    params.cache_height = Common::AlignUp(params.height, 8);
+    params.target = SurfaceTargetFromTextureType(config.tic.texture_type);
     return params;
 }
 
@@ -69,10 +72,12 @@ static VAddr TryGetCpuAddr(Tegra::GPUVAddr gpu_addr) {
     params.type = GetFormatType(params.pixel_format);
     params.width = config.width;
     params.height = config.height;
+    params.depth = 1;
     params.unaligned_height = config.height;
     params.size_in_bytes = params.SizeInBytes();
-    params.cache_width = Common::AlignUp(params.width, 16);
-    params.cache_height = Common::AlignUp(params.height, 16);
+    params.cache_width = Common::AlignUp(params.width, 8);
+    params.cache_height = Common::AlignUp(params.height, 8);
+    params.target = SurfaceTarget::Texture2D;
     return params;
 }
 
@@ -86,13 +91,14 @@ static VAddr TryGetCpuAddr(Tegra::GPUVAddr gpu_addr) {
     params.pixel_format = PixelFormatFromDepthFormat(format);
     params.component_type = ComponentTypeFromDepthFormat(format);
     params.type = GetFormatType(params.pixel_format);
-    params.size_in_bytes = params.SizeInBytes();
     params.width = zeta_width;
     params.height = zeta_height;
+    params.depth = 1;
     params.unaligned_height = zeta_height;
     params.size_in_bytes = params.SizeInBytes();
-    params.cache_width = Common::AlignUp(params.width, 16);
-    params.cache_height = Common::AlignUp(params.height, 16);
+    params.cache_width = Common::AlignUp(params.width, 8);
+    params.cache_height = Common::AlignUp(params.height, 8);
+    params.target = SurfaceTarget::Texture2D;
     return params;
 }
 
@@ -166,6 +172,26 @@ static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_form
      ComponentType::Float, false}, // Z32FS8
 }};
 
+static GLenum SurfaceTargetToGL(SurfaceParams::SurfaceTarget target) {
+    switch (target) {
+    case SurfaceParams::SurfaceTarget::Texture1D:
+        return GL_TEXTURE_1D;
+    case SurfaceParams::SurfaceTarget::Texture2D:
+        return GL_TEXTURE_2D;
+    case SurfaceParams::SurfaceTarget::Texture3D:
+        return GL_TEXTURE_3D;
+    case SurfaceParams::SurfaceTarget::Texture1DArray:
+        return GL_TEXTURE_1D_ARRAY;
+    case SurfaceParams::SurfaceTarget::Texture2DArray:
+        return GL_TEXTURE_2D_ARRAY;
+    case SurfaceParams::SurfaceTarget::TextureCubemap:
+        return GL_TEXTURE_CUBE_MAP;
+    }
+    LOG_CRITICAL(Render_OpenGL, "Unimplemented texture target={}", static_cast<u32>(target));
+    UNREACHABLE();
+    return {};
+}
+
 static const FormatTuple& GetFormatTuple(PixelFormat pixel_format, ComponentType component_type) {
     ASSERT(static_cast<size_t>(pixel_format) < tex_format_tuples.size());
     auto& format = tex_format_tuples[static_cast<unsigned int>(pixel_format)];
@@ -220,7 +246,8 @@ static bool IsFormatBCn(PixelFormat format) {
 }
 
 template <bool morton_to_gl, PixelFormat format>
-void MortonCopy(u32 stride, u32 block_height, u32 height, std::vector<u8>& gl_buffer, VAddr addr) {
+void MortonCopy(u32 stride, u32 block_height, u32 height, u8* gl_buffer, size_t gl_buffer_size,
+                VAddr addr) {
     constexpr u32 bytes_per_pixel = SurfaceParams::GetFormatBpp(format) / CHAR_BIT;
     constexpr u32 gl_bytes_per_pixel = CachedSurface::GetGLBytesPerPixel(format);
 
@@ -230,18 +257,18 @@ void MortonCopy(u32 stride, u32 block_height, u32 height, std::vector<u8>& gl_bu
         const u32 tile_size{IsFormatBCn(format) ? 4U : 1U};
         const std::vector<u8> data = Tegra::Texture::UnswizzleTexture(
             addr, tile_size, bytes_per_pixel, stride, height, block_height);
-        const size_t size_to_copy{std::min(gl_buffer.size(), data.size())};
-        gl_buffer.assign(data.begin(), data.begin() + size_to_copy);
+        const size_t size_to_copy{std::min(gl_buffer_size, data.size())};
+        memcpy(gl_buffer, data.data(), size_to_copy);
     } else {
         // TODO(bunnei): Assumes the default rendering GOB size of 16 (128 lines). We should
         // check the configuration for this and perform more generic un/swizzle
         LOG_WARNING(Render_OpenGL, "need to use correct swizzle/GOB parameters!");
         VideoCore::MortonCopyPixels128(stride, height, bytes_per_pixel, gl_bytes_per_pixel,
-                                       Memory::GetPointer(addr), gl_buffer.data(), morton_to_gl);
+                                       Memory::GetPointer(addr), gl_buffer, morton_to_gl);
     }
 }
 
-static constexpr std::array<void (*)(u32, u32, u32, std::vector<u8>&, VAddr),
+static constexpr std::array<void (*)(u32, u32, u32, u8*, size_t, VAddr),
                             SurfaceParams::MaxPixelFormat>
     morton_to_gl_fns = {
         // clang-format off
@@ -298,7 +325,7 @@ static constexpr std::array<void (*)(u32, u32, u32, std::vector<u8>&, VAddr),
         // clang-format on
 };
 
-static constexpr std::array<void (*)(u32, u32, u32, std::vector<u8>&, VAddr),
+static constexpr std::array<void (*)(u32, u32, u32, u8*, size_t, VAddr),
                             SurfaceParams::MaxPixelFormat>
     gl_to_morton_fns = {
         // clang-format off
@@ -357,33 +384,6 @@ static constexpr std::array<void (*)(u32, u32, u32, std::vector<u8>&, VAddr),
         // clang-format on
 };
 
-// Allocate an uninitialized texture of appropriate size and format for the surface
-static void AllocateSurfaceTexture(GLuint texture, const FormatTuple& format_tuple, u32 width,
-                                   u32 height) {
-    OpenGLState cur_state = OpenGLState::GetCurState();
-
-    // Keep track of previous texture bindings
-    GLuint old_tex = cur_state.texture_units[0].texture_2d;
-    cur_state.texture_units[0].texture_2d = texture;
-    cur_state.Apply();
-    glActiveTexture(GL_TEXTURE0);
-
-    if (!format_tuple.compressed) {
-        // Only pre-create the texture for non-compressed textures.
-        glTexImage2D(GL_TEXTURE_2D, 0, format_tuple.internal_format, width, height, 0,
-                     format_tuple.format, format_tuple.type, nullptr);
-    }
-
-    glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_LEVEL, 0);
-    glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
-    glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
-    glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
-
-    // Restore previous texture bindings
-    cur_state.texture_units[0].texture_2d = old_tex;
-    cur_state.Apply();
-}
-
 static bool BlitTextures(GLuint src_tex, const MathUtil::Rectangle<u32>& src_rect, GLuint dst_tex,
                          const MathUtil::Rectangle<u32>& dst_rect, SurfaceType type,
                          GLuint read_fb_handle, GLuint draw_fb_handle) {
@@ -438,12 +438,56 @@ static bool BlitTextures(GLuint src_tex, const MathUtil::Rectangle<u32>& src_rec
     return true;
 }
 
-CachedSurface::CachedSurface(const SurfaceParams& params) : params(params) {
+CachedSurface::CachedSurface(const SurfaceParams& params)
+    : params(params), gl_target(SurfaceTargetToGL(params.target)) {
     texture.Create();
     const auto& rect{params.GetRect()};
-    AllocateSurfaceTexture(texture.handle,
-                           GetFormatTuple(params.pixel_format, params.component_type),
-                           rect.GetWidth(), rect.GetHeight());
+
+    // Keep track of previous texture bindings
+    OpenGLState cur_state = OpenGLState::GetCurState();
+    const auto& old_tex = cur_state.texture_units[0];
+    SCOPE_EXIT({
+        cur_state.texture_units[0] = old_tex;
+        cur_state.Apply();
+    });
+
+    cur_state.texture_units[0].texture = texture.handle;
+    cur_state.texture_units[0].target = SurfaceTargetToGL(params.target);
+    cur_state.Apply();
+    glActiveTexture(GL_TEXTURE0);
+
+    const auto& format_tuple = GetFormatTuple(params.pixel_format, params.component_type);
+    if (!format_tuple.compressed) {
+        // Only pre-create the texture for non-compressed textures.
+        switch (params.target) {
+        case SurfaceParams::SurfaceTarget::Texture1D:
+            glTexImage1D(SurfaceTargetToGL(params.target), 0, format_tuple.internal_format,
+                         rect.GetWidth(), 0, format_tuple.format, format_tuple.type, nullptr);
+            break;
+        case SurfaceParams::SurfaceTarget::Texture2D:
+            glTexImage2D(SurfaceTargetToGL(params.target), 0, format_tuple.internal_format,
+                         rect.GetWidth(), rect.GetHeight(), 0, format_tuple.format,
+                         format_tuple.type, nullptr);
+            break;
+        case SurfaceParams::SurfaceTarget::Texture3D:
+        case SurfaceParams::SurfaceTarget::Texture2DArray:
+            glTexImage3D(SurfaceTargetToGL(params.target), 0, format_tuple.internal_format,
+                         rect.GetWidth(), rect.GetHeight(), params.depth, 0, format_tuple.format,
+                         format_tuple.type, nullptr);
+            break;
+        default:
+            LOG_CRITICAL(Render_OpenGL, "Unimplemented surface target={}",
+                         static_cast<u32>(params.target));
+            UNREACHABLE();
+            glTexImage2D(GL_TEXTURE_2D, 0, format_tuple.internal_format, rect.GetWidth(),
+                         rect.GetHeight(), 0, format_tuple.format, format_tuple.type, nullptr);
+        }
+    }
+
+    glTexParameteri(SurfaceTargetToGL(params.target), GL_TEXTURE_MAX_LEVEL, 0);
+    glTexParameteri(SurfaceTargetToGL(params.target), GL_TEXTURE_MIN_FILTER, GL_LINEAR);
+    glTexParameteri(SurfaceTargetToGL(params.target), GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
+    glTexParameteri(SurfaceTargetToGL(params.target), GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
 }
 
 static void ConvertS8Z24ToZ24S8(std::vector<u8>& data, u32 width, u32 height) {
@@ -461,7 +505,7 @@ static void ConvertS8Z24ToZ24S8(std::vector<u8>& data, u32 width, u32 height) {
 
     S8Z24 input_pixel{};
     Z24S8 output_pixel{};
-    const auto bpp{CachedSurface::GetGLBytesPerPixel(PixelFormat::S8Z24)};
+    constexpr auto bpp{CachedSurface::GetGLBytesPerPixel(PixelFormat::S8Z24)};
     for (size_t y = 0; y < height; ++y) {
         for (size_t x = 0; x < width; ++x) {
             const size_t offset{bpp * (y * width + x)};
@@ -474,7 +518,7 @@ static void ConvertS8Z24ToZ24S8(std::vector<u8>& data, u32 width, u32 height) {
 }
 
 static void ConvertG8R8ToR8G8(std::vector<u8>& data, u32 width, u32 height) {
-    const auto bpp{CachedSurface::GetGLBytesPerPixel(PixelFormat::G8R8U)};
+    constexpr auto bpp{CachedSurface::GetGLBytesPerPixel(PixelFormat::G8R8U)};
     for (size_t y = 0; y < height; ++y) {
         for (size_t x = 0; x < width; ++x) {
             const size_t offset{bpp * (y * width + x)};
@@ -514,23 +558,6 @@ static void ConvertFormatAsNeeded_LoadGLBuffer(std::vector<u8>& data, PixelForma
     }
 }
 
-/**
- * Helper function to perform software conversion (as needed) when flushing a buffer to Switch
- * memory. This is for Maxwell pixel formats that cannot be represented as-is in OpenGL or with
- * typical desktop GPUs.
- */
-static void ConvertFormatAsNeeded_FlushGLBuffer(std::vector<u8>& /*data*/, PixelFormat pixel_format,
-                                                u32 /*width*/, u32 /*height*/) {
-    switch (pixel_format) {
-    case PixelFormat::ASTC_2D_4X4:
-    case PixelFormat::S8Z24:
-        LOG_CRITICAL(Render_OpenGL, "Unimplemented pixel_format={}",
-                     static_cast<u32>(pixel_format));
-        UNREACHABLE();
-        break;
-    }
-}
-
 MICROPROFILE_DEFINE(OpenGL_SurfaceLoad, "OpenGL", "Surface Load", MP_RGB(128, 64, 192));
 void CachedSurface::LoadGLBuffer() {
     ASSERT(params.type != SurfaceType::Fill);
@@ -545,13 +572,25 @@ void CachedSurface::LoadGLBuffer() {
     MICROPROFILE_SCOPE(OpenGL_SurfaceLoad);
 
     if (params.is_tiled) {
-        gl_buffer.resize(copy_size);
+        // TODO(bunnei): This only unswizzles and copies a 2D texture - we do not yet know how to do
+        // this for 3D textures, etc.
+        switch (params.target) {
+        case SurfaceParams::SurfaceTarget::Texture2D:
+            // Pass impl. to the fallback code below
+            break;
+        default:
+            LOG_CRITICAL(HW_GPU, "Unimplemented tiled load for target={}",
+                         static_cast<u32>(params.target));
+            UNREACHABLE();
+        }
 
+        gl_buffer.resize(static_cast<size_t>(params.depth) * copy_size);
         morton_to_gl_fns[static_cast<size_t>(params.pixel_format)](
-            params.width, params.block_height, params.height, gl_buffer, params.addr);
+            params.width, params.block_height, params.height, gl_buffer.data(), copy_size,
+            params.addr);
     } else {
-        const u8* const texture_src_data_end = texture_src_data + copy_size;
-
+        const u8* const texture_src_data_end{texture_src_data +
+                                             (static_cast<size_t>(params.depth) * copy_size)};
         gl_buffer.assign(texture_src_data, texture_src_data_end);
     }
 
@@ -560,23 +599,7 @@ void CachedSurface::LoadGLBuffer() {
 
 MICROPROFILE_DEFINE(OpenGL_SurfaceFlush, "OpenGL", "Surface Flush", MP_RGB(128, 192, 64));
 void CachedSurface::FlushGLBuffer() {
-    u8* const dst_buffer = Memory::GetPointer(params.addr);
-
-    ASSERT(dst_buffer);
-    ASSERT(gl_buffer.size() ==
-           params.width * params.height * GetGLBytesPerPixel(params.pixel_format));
-
-    MICROPROFILE_SCOPE(OpenGL_SurfaceFlush);
-
-    ConvertFormatAsNeeded_FlushGLBuffer(gl_buffer, params.pixel_format, params.width,
-                                        params.height);
-
-    if (!params.is_tiled) {
-        std::memcpy(dst_buffer, gl_buffer.data(), params.size_in_bytes);
-    } else {
-        gl_to_morton_fns[static_cast<size_t>(params.pixel_format)](
-            params.width, params.block_height, params.height, gl_buffer, params.addr);
-    }
+    ASSERT_MSG(false, "Unimplemented");
 }
 
 MICROPROFILE_DEFINE(OpenGL_TextureUL, "OpenGL", "Texture Upload", MP_RGB(128, 64, 192));
@@ -586,22 +609,29 @@ void CachedSurface::UploadGLTexture(GLuint read_fb_handle, GLuint draw_fb_handle
 
     MICROPROFILE_SCOPE(OpenGL_TextureUL);
 
-    ASSERT(gl_buffer.size() ==
-           params.width * params.height * GetGLBytesPerPixel(params.pixel_format));
+    ASSERT(gl_buffer.size() == static_cast<size_t>(params.width) * params.height *
+                                   GetGLBytesPerPixel(params.pixel_format) * params.depth);
 
     const auto& rect{params.GetRect()};
 
     // Load data from memory to the surface
-    GLint x0 = static_cast<GLint>(rect.left);
-    GLint y0 = static_cast<GLint>(rect.bottom);
-    size_t buffer_offset = (y0 * params.width + x0) * GetGLBytesPerPixel(params.pixel_format);
+    const GLint x0 = static_cast<GLint>(rect.left);
+    const GLint y0 = static_cast<GLint>(rect.bottom);
+    const size_t buffer_offset =
+        static_cast<size_t>(static_cast<size_t>(y0) * params.width + static_cast<size_t>(x0)) *
+        GetGLBytesPerPixel(params.pixel_format);
 
     const FormatTuple& tuple = GetFormatTuple(params.pixel_format, params.component_type);
-    GLuint target_tex = texture.handle;
+    const GLuint target_tex = texture.handle;
     OpenGLState cur_state = OpenGLState::GetCurState();
 
-    GLuint old_tex = cur_state.texture_units[0].texture_2d;
-    cur_state.texture_units[0].texture_2d = target_tex;
+    const auto& old_tex = cur_state.texture_units[0];
+    SCOPE_EXIT({
+        cur_state.texture_units[0] = old_tex;
+        cur_state.Apply();
+    });
+    cur_state.texture_units[0].texture = target_tex;
+    cur_state.texture_units[0].target = SurfaceTargetToGL(params.target);
     cur_state.Apply();
 
     // Ensure no bad interactions with GL_UNPACK_ALIGNMENT
@@ -610,74 +640,68 @@ void CachedSurface::UploadGLTexture(GLuint read_fb_handle, GLuint draw_fb_handle
 
     glActiveTexture(GL_TEXTURE0);
     if (tuple.compressed) {
-        glCompressedTexImage2D(
-            GL_TEXTURE_2D, 0, tuple.internal_format, static_cast<GLsizei>(params.width),
-            static_cast<GLsizei>(params.height), 0, static_cast<GLsizei>(params.size_in_bytes),
-            &gl_buffer[buffer_offset]);
+        switch (params.target) {
+        case SurfaceParams::SurfaceTarget::Texture2D:
+            glCompressedTexImage2D(
+                SurfaceTargetToGL(params.target), 0, tuple.internal_format,
+                static_cast<GLsizei>(params.width), static_cast<GLsizei>(params.height), 0,
+                static_cast<GLsizei>(params.size_in_bytes), &gl_buffer[buffer_offset]);
+            break;
+        case SurfaceParams::SurfaceTarget::Texture3D:
+        case SurfaceParams::SurfaceTarget::Texture2DArray:
+            glCompressedTexImage3D(
+                SurfaceTargetToGL(params.target), 0, tuple.internal_format,
+                static_cast<GLsizei>(params.width), static_cast<GLsizei>(params.height),
+                static_cast<GLsizei>(params.depth), 0, static_cast<GLsizei>(params.size_in_bytes),
+                &gl_buffer[buffer_offset]);
+            break;
+        default:
+            LOG_CRITICAL(Render_OpenGL, "Unimplemented surface target={}",
+                         static_cast<u32>(params.target));
+            UNREACHABLE();
+            glCompressedTexImage2D(
+                GL_TEXTURE_2D, 0, tuple.internal_format, static_cast<GLsizei>(params.width),
+                static_cast<GLsizei>(params.height), 0, static_cast<GLsizei>(params.size_in_bytes),
+                &gl_buffer[buffer_offset]);
+        }
     } else {
-        glTexSubImage2D(GL_TEXTURE_2D, 0, x0, y0, static_cast<GLsizei>(rect.GetWidth()),
-                        static_cast<GLsizei>(rect.GetHeight()), tuple.format, tuple.type,
-                        &gl_buffer[buffer_offset]);
+
+        switch (params.target) {
+        case SurfaceParams::SurfaceTarget::Texture1D:
+            glTexSubImage1D(SurfaceTargetToGL(params.target), 0, x0,
+                            static_cast<GLsizei>(rect.GetWidth()), tuple.format, tuple.type,
+                            &gl_buffer[buffer_offset]);
+            break;
+        case SurfaceParams::SurfaceTarget::Texture2D:
+            glTexSubImage2D(SurfaceTargetToGL(params.target), 0, x0, y0,
+                            static_cast<GLsizei>(rect.GetWidth()),
+                            static_cast<GLsizei>(rect.GetHeight()), tuple.format, tuple.type,
+                            &gl_buffer[buffer_offset]);
+            break;
+        case SurfaceParams::SurfaceTarget::Texture3D:
+        case SurfaceParams::SurfaceTarget::Texture2DArray:
+            glTexSubImage3D(SurfaceTargetToGL(params.target), 0, x0, y0, 0,
+                            static_cast<GLsizei>(rect.GetWidth()),
+                            static_cast<GLsizei>(rect.GetHeight()), params.depth, tuple.format,
+                            tuple.type, &gl_buffer[buffer_offset]);
+            break;
+        default:
+            LOG_CRITICAL(Render_OpenGL, "Unimplemented surface target={}",
+                         static_cast<u32>(params.target));
+            UNREACHABLE();
+            glTexSubImage2D(GL_TEXTURE_2D, 0, x0, y0, static_cast<GLsizei>(rect.GetWidth()),
+                            static_cast<GLsizei>(rect.GetHeight()), tuple.format, tuple.type,
+                            &gl_buffer[buffer_offset]);
+        }
     }
 
     glPixelStorei(GL_UNPACK_ROW_LENGTH, 0);
-
-    cur_state.texture_units[0].texture_2d = old_tex;
-    cur_state.Apply();
-}
-
-MICROPROFILE_DEFINE(OpenGL_TextureDL, "OpenGL", "Texture Download", MP_RGB(128, 192, 64));
-void CachedSurface::DownloadGLTexture(GLuint read_fb_handle, GLuint draw_fb_handle) {
-    if (params.type == SurfaceType::Fill)
-        return;
-
-    MICROPROFILE_SCOPE(OpenGL_TextureDL);
-
-    gl_buffer.resize(params.width * params.height * GetGLBytesPerPixel(params.pixel_format));
-
-    OpenGLState state = OpenGLState::GetCurState();
-    OpenGLState prev_state = state;
-    SCOPE_EXIT({ prev_state.Apply(); });
-
-    const FormatTuple& tuple = GetFormatTuple(params.pixel_format, params.component_type);
-
-    // Ensure no bad interactions with GL_PACK_ALIGNMENT
-    ASSERT(params.width * GetGLBytesPerPixel(params.pixel_format) % 4 == 0);
-    glPixelStorei(GL_PACK_ROW_LENGTH, static_cast<GLint>(params.width));
-
-    const auto& rect{params.GetRect()};
-    size_t buffer_offset =
-        (rect.bottom * params.width + rect.left) * GetGLBytesPerPixel(params.pixel_format);
-
-    state.UnbindTexture(texture.handle);
-    state.draw.read_framebuffer = read_fb_handle;
-    state.Apply();
-
-    if (params.type == SurfaceType::ColorTexture) {
-        glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D,
-                               texture.handle, 0);
-        glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0,
-                               0);
-    } else if (params.type == SurfaceType::Depth) {
-        glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, 0, 0);
-        glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_DEPTH_ATTACHMENT, GL_TEXTURE_2D,
-                               texture.handle, 0);
-        glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0, 0);
-    } else {
-        glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, 0, 0);
-        glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D,
-                               texture.handle, 0);
-    }
-    glReadPixels(static_cast<GLint>(rect.left), static_cast<GLint>(rect.bottom),
-                 static_cast<GLsizei>(rect.GetWidth()), static_cast<GLsizei>(rect.GetHeight()),
-                 tuple.format, tuple.type, &gl_buffer[buffer_offset]);
-
-    glPixelStorei(GL_PACK_ROW_LENGTH, 0);
 }
 
 RasterizerCacheOpenGL::RasterizerCacheOpenGL() {
     read_framebuffer.Create();
     draw_framebuffer.Create();
+    copy_pbo.Create();
 }
 
 Surface RasterizerCacheOpenGL::GetTextureSurface(const Tegra::Texture::FullTextureInfo& config) {
@@ -690,7 +714,7 @@ SurfaceSurfaceRect_Tuple RasterizerCacheOpenGL::GetFramebufferSurfaces(bool usin
     const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
 
     // TODO(bunnei): This is hard corded to use just the first render buffer
-    LOG_WARNING(Render_OpenGL, "hard-coded for render target 0!");
+    LOG_TRACE(Render_OpenGL, "hard-coded for render target 0!");
 
     // get color and depth surfaces
     SurfaceParams color_params{};
@@ -748,7 +772,6 @@ void RasterizerCacheOpenGL::LoadSurface(const Surface& surface) {
 }
 
 void RasterizerCacheOpenGL::FlushSurface(const Surface& surface) {
-    surface->DownloadGLTexture(read_framebuffer.handle, draw_framebuffer.handle);
     surface->FlushGLBuffer();
 }
 
@@ -809,8 +832,8 @@ Surface RasterizerCacheOpenGL::RecreateSurface(const Surface& surface,
     // If format is unchanged, we can do a faster blit without reinterpreting pixel data
     if (params.pixel_format == new_params.pixel_format) {
         BlitTextures(surface->Texture().handle, params.GetRect(), new_surface->Texture().handle,
-                     new_surface->GetSurfaceParams().GetRect(), params.type,
-                     read_framebuffer.handle, draw_framebuffer.handle);
+                     params.GetRect(), params.type, read_framebuffer.handle,
+                     draw_framebuffer.handle);
         return new_surface;
     }
 
@@ -821,12 +844,7 @@ Surface RasterizerCacheOpenGL::RecreateSurface(const Surface& surface,
 
         size_t buffer_size = std::max(params.SizeInBytes(), new_params.SizeInBytes());
 
-        // Use a Pixel Buffer Object to download the previous texture and then upload it to the new
-        // one using the new format.
-        OGLBuffer pbo;
-        pbo.Create();
-
-        glBindBuffer(GL_PIXEL_PACK_BUFFER, pbo.handle);
+        glBindBuffer(GL_PIXEL_PACK_BUFFER, copy_pbo.handle);
         glBufferData(GL_PIXEL_PACK_BUFFER, buffer_size, nullptr, GL_STREAM_DRAW_ARB);
         if (source_format.compressed) {
             glGetCompressedTextureImage(surface->Texture().handle, 0,
@@ -845,8 +863,8 @@ Surface RasterizerCacheOpenGL::RecreateSurface(const Surface& surface,
                 // of the data in this case. Games like Super Mario Odyssey seem to hit this case
                 // when drawing, it re-uses the memory of a previous texture as a bigger framebuffer
                 // but it doesn't clear it beforehand, the texture is already full of zeros.
-                LOG_CRITICAL(HW_GPU, "Trying to upload extra texture data from the CPU during "
-                                     "reinterpretation but the texture is tiled.");
+                LOG_DEBUG(HW_GPU, "Trying to upload extra texture data from the CPU during "
+                                  "reinterpretation but the texture is tiled.");
             }
             size_t remaining_size = new_params.SizeInBytes() - params.SizeInBytes();
             std::vector<u8> data(remaining_size);
@@ -859,21 +877,38 @@ Surface RasterizerCacheOpenGL::RecreateSurface(const Surface& surface,
 
         const auto& dest_rect{new_params.GetRect()};
 
-        glBindBuffer(GL_PIXEL_UNPACK_BUFFER, pbo.handle);
+        glBindBuffer(GL_PIXEL_UNPACK_BUFFER, copy_pbo.handle);
         if (dest_format.compressed) {
-            glCompressedTexSubImage2D(
-                GL_TEXTURE_2D, 0, 0, 0, static_cast<GLsizei>(dest_rect.GetWidth()),
-                static_cast<GLsizei>(dest_rect.GetHeight()), dest_format.format,
-                static_cast<GLsizei>(new_params.SizeInBytes()), nullptr);
+            LOG_CRITICAL(HW_GPU, "Compressed copy is unimplemented!");
+            UNREACHABLE();
         } else {
-            glTextureSubImage2D(new_surface->Texture().handle, 0, 0, 0,
-                                static_cast<GLsizei>(dest_rect.GetWidth()),
-                                static_cast<GLsizei>(dest_rect.GetHeight()), dest_format.format,
-                                dest_format.type, nullptr);
+            switch (new_params.target) {
+            case SurfaceParams::SurfaceTarget::Texture1D:
+                glTextureSubImage1D(new_surface->Texture().handle, 0, 0,
+                                    static_cast<GLsizei>(dest_rect.GetWidth()), dest_format.format,
+                                    dest_format.type, nullptr);
+                break;
+            case SurfaceParams::SurfaceTarget::Texture2D:
+                glTextureSubImage2D(new_surface->Texture().handle, 0, 0, 0,
+                                    static_cast<GLsizei>(dest_rect.GetWidth()),
+                                    static_cast<GLsizei>(dest_rect.GetHeight()), dest_format.format,
+                                    dest_format.type, nullptr);
+                break;
+            case SurfaceParams::SurfaceTarget::Texture3D:
+            case SurfaceParams::SurfaceTarget::Texture2DArray:
+                glTextureSubImage3D(new_surface->Texture().handle, 0, 0, 0, 0,
+                                    static_cast<GLsizei>(dest_rect.GetWidth()),
+                                    static_cast<GLsizei>(dest_rect.GetHeight()),
+                                    static_cast<GLsizei>(new_params.depth), dest_format.format,
+                                    dest_format.type, nullptr);
+                break;
+            default:
+                LOG_CRITICAL(Render_OpenGL, "Unimplemented surface target={}",
+                             static_cast<u32>(params.target));
+                UNREACHABLE();
+            }
         }
         glBindBuffer(GL_PIXEL_UNPACK_BUFFER, 0);
-
-        pbo.Release();
     }
 
     return new_surface;

src/video_core/renderer_opengl/gl_rasterizer_cache.h

@@ -109,6 +109,33 @@ struct SurfaceParams {
         Invalid = 4,
     };
 
+    enum class SurfaceTarget {
+        Texture1D,
+        Texture2D,
+        Texture3D,
+        Texture1DArray,
+        Texture2DArray,
+        TextureCubemap,
+    };
+
+    static SurfaceTarget SurfaceTargetFromTextureType(Tegra::Texture::TextureType texture_type) {
+        switch (texture_type) {
+        case Tegra::Texture::TextureType::Texture1D:
+            return SurfaceTarget::Texture1D;
+        case Tegra::Texture::TextureType::Texture2D:
+        case Tegra::Texture::TextureType::Texture2DNoMipmap:
+            return SurfaceTarget::Texture2D;
+        case Tegra::Texture::TextureType::Texture1DArray:
+            return SurfaceTarget::Texture1DArray;
+        case Tegra::Texture::TextureType::Texture2DArray:
+            return SurfaceTarget::Texture2DArray;
+        default:
+            LOG_CRITICAL(HW_GPU, "Unimplemented texture_type={}", static_cast<u32>(texture_type));
+            UNREACHABLE();
+            return SurfaceTarget::Texture2D;
+        }
+    }
+
     /**
      * Gets the compression factor for the specified PixelFormat. This applies to just the
      * "compressed width" and "compressed height", not the overall compression factor of a
@@ -635,7 +662,7 @@ struct SurfaceParams {
         ASSERT(width % compression_factor == 0);
         ASSERT(height % compression_factor == 0);
         return (width / compression_factor) * (height / compression_factor) *
-               GetFormatBpp(pixel_format) / CHAR_BIT;
+               GetFormatBpp(pixel_format) * depth / CHAR_BIT;
     }
 
     /// Creates SurfaceParams from a texture configuration
@@ -664,8 +691,10 @@ struct SurfaceParams {
     SurfaceType type;
     u32 width;
     u32 height;
+    u32 depth;
     u32 unaligned_height;
     size_t size_in_bytes;
+    SurfaceTarget target;
 
     // Parameters used for caching only
     u32 cache_width;
@@ -709,6 +738,10 @@ public:
         return texture;
     }
 
+    GLenum Target() const {
+        return gl_target;
+    }
+
     static constexpr unsigned int GetGLBytesPerPixel(SurfaceParams::PixelFormat format) {
         if (format == SurfaceParams::PixelFormat::Invalid)
             return 0;
@@ -724,14 +757,14 @@ public:
     void LoadGLBuffer();
     void FlushGLBuffer();
 
-    // Upload/Download data in gl_buffer in/to this surface's texture
+    // Upload data in gl_buffer to this surface's texture
     void UploadGLTexture(GLuint read_fb_handle, GLuint draw_fb_handle);
-    void DownloadGLTexture(GLuint read_fb_handle, GLuint draw_fb_handle);
 
 private:
     OGLTexture texture;
     std::vector<u8> gl_buffer;
     SurfaceParams params;
+    GLenum gl_target;
 };
 
 class RasterizerCacheOpenGL final : public RasterizerCache<Surface> {
@@ -774,6 +807,10 @@ private:
 
     OGLFramebuffer read_framebuffer;
     OGLFramebuffer draw_framebuffer;
+
+    /// Use a Pixel Buffer Object to download the previous texture and then upload it to the new one
+    /// using the new format.
+    OGLBuffer copy_pbo;
 };
 
 } // namespace OpenGL

src/video_core/renderer_opengl/gl_shader_cache.cpp

@@ -13,8 +13,8 @@ namespace OpenGL {
 
 /// Gets the address for the specified shader stage program
 static VAddr GetShaderAddress(Maxwell::ShaderProgram program) {
-    auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
-    auto& shader_config = gpu.regs.shader_config[static_cast<size_t>(program)];
+    const auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
+    const auto& shader_config = gpu.regs.shader_config[static_cast<size_t>(program)];
     return *gpu.memory_manager.GpuToCpuAddress(gpu.regs.code_address.CodeAddress() +
                                                shader_config.offset);
 }
@@ -85,23 +85,23 @@ CachedShader::CachedShader(VAddr addr, Maxwell::ShaderProgram program_type)
     SetShaderUniformBlockBindings(program.handle);
 }
 
-GLuint CachedShader::GetProgramResourceIndex(const std::string& name) {
-    auto search{resource_cache.find(name)};
+GLuint CachedShader::GetProgramResourceIndex(const GLShader::ConstBufferEntry& buffer) {
+    const auto search{resource_cache.find(buffer.GetHash())};
     if (search == resource_cache.end()) {
         const GLuint index{
-            glGetProgramResourceIndex(program.handle, GL_UNIFORM_BLOCK, name.c_str())};
-        resource_cache[name] = index;
+            glGetProgramResourceIndex(program.handle, GL_UNIFORM_BLOCK, buffer.GetName().c_str())};
+        resource_cache[buffer.GetHash()] = index;
         return index;
     }
 
     return search->second;
 }
 
-GLint CachedShader::GetUniformLocation(const std::string& name) {
-    auto search{uniform_cache.find(name)};
+GLint CachedShader::GetUniformLocation(const GLShader::SamplerEntry& sampler) {
+    const auto search{uniform_cache.find(sampler.GetHash())};
     if (search == uniform_cache.end()) {
-        const GLint index{glGetUniformLocation(program.handle, name.c_str())};
-        uniform_cache[name] = index;
+        const GLint index{glGetUniformLocation(program.handle, sampler.GetName().c_str())};
+        uniform_cache[sampler.GetHash()] = index;
         return index;
     }
 

src/video_core/renderer_opengl/gl_shader_cache.h

@@ -4,8 +4,8 @@
 
 #pragma once
 
+#include <map>
 #include <memory>
-#include <unordered_map>
 
 #include "common/common_types.h"
 #include "video_core/rasterizer_cache.h"
@@ -43,10 +43,10 @@ public:
     }
 
     /// Gets the GL program resource location for the specified resource, caching as needed
-    GLuint GetProgramResourceIndex(const std::string& name);
+    GLuint GetProgramResourceIndex(const GLShader::ConstBufferEntry& buffer);
 
     /// Gets the GL uniform location for the specified resource, caching as needed
-    GLint GetUniformLocation(const std::string& name);
+    GLint GetUniformLocation(const GLShader::SamplerEntry& sampler);
 
 private:
     VAddr addr;
@@ -55,8 +55,8 @@ private:
     GLShader::ShaderEntries entries;
     OGLProgram program;
 
-    std::unordered_map<std::string, GLuint> resource_cache;
-    std::unordered_map<std::string, GLint> uniform_cache;
+    std::map<u32, GLuint> resource_cache;
+    std::map<u32, GLint> uniform_cache;
 };
 
 class ShaderCacheOpenGL final : public RasterizerCache<Shader> {

src/video_core/renderer_opengl/gl_shader_decompiler.cpp

@@ -113,7 +113,7 @@ private:
 
     /// Scans a range of code for labels and determines the exit method.
     ExitMethod Scan(u32 begin, u32 end, std::set<u32>& labels) {
-        auto [iter, inserted] =
+        const auto [iter, inserted] =
             exit_method_map.emplace(std::make_pair(begin, end), ExitMethod::Undetermined);
         ExitMethod& exit_method = iter->second;
         if (!inserted)
@@ -131,22 +131,22 @@ private:
                     if (instr.pred.pred_index == static_cast<u64>(Pred::UnusedIndex)) {
                         return exit_method = ExitMethod::AlwaysEnd;
                     } else {
-                        ExitMethod not_met = Scan(offset + 1, end, labels);
+                        const ExitMethod not_met = Scan(offset + 1, end, labels);
                         return exit_method = ParallelExit(ExitMethod::AlwaysEnd, not_met);
                     }
                 }
                 case OpCode::Id::BRA: {
-                    u32 target = offset + instr.bra.GetBranchTarget();
+                    const u32 target = offset + instr.bra.GetBranchTarget();
                     labels.insert(target);
-                    ExitMethod no_jmp = Scan(offset + 1, end, labels);
-                    ExitMethod jmp = Scan(target, end, labels);
+                    const ExitMethod no_jmp = Scan(offset + 1, end, labels);
+                    const ExitMethod jmp = Scan(target, end, labels);
                     return exit_method = ParallelExit(no_jmp, jmp);
                 }
                 case OpCode::Id::SSY: {
                     // The SSY instruction uses a similar encoding as the BRA instruction.
                     ASSERT_MSG(instr.bra.constant_buffer == 0,
                                "Constant buffer SSY is not supported");
-                    u32 target = offset + instr.bra.GetBranchTarget();
+                    const u32 target = offset + instr.bra.GetBranchTarget();
                     labels.insert(target);
                     // Continue scanning for an exit method.
                     break;
@@ -247,6 +247,7 @@ public:
                         const Maxwell3D::Regs::ShaderStage& stage, const std::string& suffix)
         : shader{shader}, declarations{declarations}, stage{stage}, suffix{suffix} {
         BuildRegisterList();
+        BuildInputList();
     }
 
     /**
@@ -343,9 +344,10 @@ public:
      * @param elem The element to use for the operation.
      * @param attribute The input attribute to use as the source value.
      */
-    void SetRegisterToInputAttibute(const Register& reg, u64 elem, Attribute::Index attribute) {
-        std::string dest = GetRegisterAsFloat(reg);
-        std::string src = GetInputAttribute(attribute) + GetSwizzle(elem);
+    void SetRegisterToInputAttibute(const Register& reg, u64 elem, Attribute::Index attribute,
+                                    const Tegra::Shader::IpaMode& input_mode) {
+        const std::string dest = GetRegisterAsFloat(reg);
+        const std::string src = GetInputAttribute(attribute, input_mode) + GetSwizzle(elem);
         shader.AddLine(dest + " = " + src + ';');
     }
 
@@ -357,8 +359,8 @@ public:
      * @param reg The register to use as the source value.
      */
     void SetOutputAttributeToRegister(Attribute::Index attribute, u64 elem, const Register& reg) {
-        std::string dest = GetOutputAttribute(attribute);
-        std::string src = GetRegisterAsFloat(reg);
+        const std::string dest = GetOutputAttribute(attribute);
+        const std::string src = GetRegisterAsFloat(reg);
 
         if (!dest.empty()) {
             // Can happen with unknown/unimplemented output attributes, in which case we ignore the
@@ -391,9 +393,9 @@ public:
                                    GLSLRegister::Type type) {
         declr_const_buffers[cbuf_index].MarkAsUsedIndirect(cbuf_index, stage);
 
-        std::string final_offset = fmt::format("({} + {})", index_str, offset / 4);
-        std::string value = 'c' + std::to_string(cbuf_index) + '[' + final_offset + " / 4][" +
-                            final_offset + " % 4]";
+        const std::string final_offset = fmt::format("({} + {})", index_str, offset / 4);
+        const std::string value = 'c' + std::to_string(cbuf_index) + '[' + final_offset + " / 4][" +
+                                  final_offset + " % 4]";
 
         if (type == GLSLRegister::Type::Float) {
             return value;
@@ -412,12 +414,13 @@ public:
         }
         declarations.AddNewLine();
 
-        for (const auto& index : declr_input_attribute) {
+        for (const auto element : declr_input_attribute) {
             // TODO(bunnei): Use proper number of elements for these
-            declarations.AddLine("layout(location = " +
-                                 std::to_string(static_cast<u32>(index) -
-                                                static_cast<u32>(Attribute::Index::Attribute_0)) +
-                                 ") in vec4 " + GetInputAttribute(index) + ';');
+            u32 idx =
+                static_cast<u32>(element.first) - static_cast<u32>(Attribute::Index::Attribute_0);
+            declarations.AddLine("layout(location = " + std::to_string(idx) + ")" +
+                                 GetInputFlags(element.first) + "in vec4 " +
+                                 GetInputAttribute(element.first, element.second) + ';');
         }
         declarations.AddNewLine();
 
@@ -440,13 +443,12 @@ public:
         }
         declarations.AddNewLine();
 
-        // Append the sampler2D array for the used textures.
-        size_t num_samplers = GetSamplers().size();
-        if (num_samplers > 0) {
-            declarations.AddLine("uniform sampler2D " + SamplerEntry::GetArrayName(stage) + '[' +
-                                 std::to_string(num_samplers) + "];");
-            declarations.AddNewLine();
+        const auto& samplers = GetSamplers();
+        for (const auto& sampler : samplers) {
+            declarations.AddLine("uniform " + sampler.GetTypeString() + ' ' + sampler.GetName() +
+                                 ';');
         }
+        declarations.AddNewLine();
     }
 
     /// Returns a list of constant buffer declarations
@@ -458,27 +460,29 @@ public:
     }
 
     /// Returns a list of samplers used in the shader
-    std::vector<SamplerEntry> GetSamplers() const {
+    const std::vector<SamplerEntry>& GetSamplers() const {
         return used_samplers;
     }
 
     /// Returns the GLSL sampler used for the input shader sampler, and creates a new one if
     /// necessary.
-    std::string AccessSampler(const Sampler& sampler) {
-        size_t offset = static_cast<size_t>(sampler.index.Value());
+    std::string AccessSampler(const Sampler& sampler, Tegra::Shader::TextureType type,
+                              bool is_array) {
+        const size_t offset = static_cast<size_t>(sampler.index.Value());
 
         // If this sampler has already been used, return the existing mapping.
-        auto itr =
+        const auto itr =
             std::find_if(used_samplers.begin(), used_samplers.end(),
                          [&](const SamplerEntry& entry) { return entry.GetOffset() == offset; });
 
         if (itr != used_samplers.end()) {
+            ASSERT(itr->GetType() == type && itr->IsArray() == is_array);
             return itr->GetName();
         }
 
         // Otherwise create a new mapping for this sampler
-        size_t next_index = used_samplers.size();
-        SamplerEntry entry{stage, offset, next_index};
+        const size_t next_index = used_samplers.size();
+        const SamplerEntry entry{stage, offset, next_index, type, is_array};
         used_samplers.emplace_back(entry);
         return entry.GetName();
     }
@@ -532,11 +536,24 @@ private:
         }
     }
 
+    void BuildInputList() {
+        const u32 size = static_cast<u32>(Attribute::Index::Attribute_31) -
+                         static_cast<u32>(Attribute::Index::Attribute_0) + 1;
+        declr_input_attribute.reserve(size);
+    }
+
     /// Generates code representing an input attribute register.
-    std::string GetInputAttribute(Attribute::Index attribute) {
+    std::string GetInputAttribute(Attribute::Index attribute,
+                                  const Tegra::Shader::IpaMode& input_mode) {
         switch (attribute) {
         case Attribute::Index::Position:
-            return "position";
+            if (stage != Maxwell3D::Regs::ShaderStage::Fragment) {
+                return "position";
+            } else {
+                return "vec4(gl_FragCoord.x, gl_FragCoord.y, gl_FragCoord.z, 1.0)";
+            }
+        case Attribute::Index::PointCoord:
+            return "vec4(gl_PointCoord.x, gl_PointCoord.y, 0, 0)";
         case Attribute::Index::TessCoordInstanceIDVertexID:
             // TODO(Subv): Find out what the values are for the first two elements when inside a
             // vertex shader, and what's the value of the fourth element when inside a Tess Eval
@@ -552,7 +569,14 @@ private:
                             static_cast<u32>(Attribute::Index::Attribute_0)};
             if (attribute >= Attribute::Index::Attribute_0 &&
                 attribute <= Attribute::Index::Attribute_31) {
-                declr_input_attribute.insert(attribute);
+                if (declr_input_attribute.count(attribute) == 0) {
+                    declr_input_attribute[attribute] = input_mode;
+                } else {
+                    if (declr_input_attribute[attribute] != input_mode) {
+                        LOG_CRITICAL(HW_GPU, "Same Input multiple input modes");
+                        UNREACHABLE();
+                    }
+                }
                 return "input_attribute_" + std::to_string(index);
             }
 
@@ -563,6 +587,49 @@ private:
         return "vec4(0, 0, 0, 0)";
     }
 
+    std::string GetInputFlags(const Attribute::Index attribute) {
+        const Tegra::Shader::IpaSampleMode sample_mode =
+            declr_input_attribute[attribute].sampling_mode;
+        const Tegra::Shader::IpaInterpMode interp_mode =
+            declr_input_attribute[attribute].interpolation_mode;
+        std::string out;
+        switch (interp_mode) {
+        case Tegra::Shader::IpaInterpMode::Flat: {
+            out += "flat ";
+            break;
+        }
+        case Tegra::Shader::IpaInterpMode::Linear: {
+            out += "noperspective ";
+            break;
+        }
+        case Tegra::Shader::IpaInterpMode::Perspective: {
+            // Default, Smooth
+            break;
+        }
+        default: {
+            LOG_CRITICAL(HW_GPU, "Unhandled Ipa InterpMode: {}", static_cast<u32>(interp_mode));
+            UNREACHABLE();
+        }
+        }
+        switch (sample_mode) {
+        case Tegra::Shader::IpaSampleMode::Centroid: {
+            // Note not implemented, it can be implemented with the "centroid " keyword in glsl;
+            LOG_CRITICAL(HW_GPU, "Ipa Sampler Mode: centroid, not implemented");
+            UNREACHABLE();
+            break;
+        }
+        case Tegra::Shader::IpaSampleMode::Default: {
+            // Default, n/a
+            break;
+        }
+        default: {
+            LOG_CRITICAL(HW_GPU, "Unhandled Ipa SampleMode: {}", static_cast<u32>(sample_mode));
+            UNREACHABLE();
+        }
+        }
+        return out;
+    }
+
     /// Generates code representing an output attribute register.
     std::string GetOutputAttribute(Attribute::Index attribute) {
         switch (attribute) {
@@ -593,7 +660,7 @@ private:
     ShaderWriter& shader;
     ShaderWriter& declarations;
     std::vector<GLSLRegister> regs;
-    std::set<Attribute::Index> declr_input_attribute;
+    std::unordered_map<Attribute::Index, Tegra::Shader::IpaMode> declr_input_attribute;
     std::set<Attribute::Index> declr_output_attribute;
     std::array<ConstBufferEntry, Maxwell3D::Regs::MaxConstBuffers> declr_const_buffers;
     std::vector<SamplerEntry> used_samplers;
@@ -632,7 +699,7 @@ private:
         };
 
         bool IsColorComponentOutputEnabled(u32 render_target, u32 component) const {
-            u32 bit = render_target * 4 + component;
+            const u32 bit = render_target * 4 + component;
             return enabled_color_outputs & (1 << bit);
         }
     };
@@ -640,7 +707,7 @@ private:
 
     /// Gets the Subroutine object corresponding to the specified address.
     const Subroutine& GetSubroutine(u32 begin, u32 end) const {
-        auto iter = subroutines.find(Subroutine{begin, end, suffix});
+        const auto iter = subroutines.find(Subroutine{begin, end, suffix});
         ASSERT(iter != subroutines.end());
         return *iter;
     }
@@ -656,8 +723,8 @@ private:
     }
 
     /// Generates code representing a texture sampler.
-    std::string GetSampler(const Sampler& sampler) {
-        return regs.AccessSampler(sampler);
+    std::string GetSampler(const Sampler& sampler, Tegra::Shader::TextureType type, bool is_array) {
+        return regs.AccessSampler(sampler, type, is_array);
     }
 
     /**
@@ -685,7 +752,7 @@ private:
         // Can't assign to the constant predicate.
         ASSERT(pred != static_cast<u64>(Pred::UnusedIndex));
 
-        std::string variable = 'p' + std::to_string(pred) + '_' + suffix;
+        const std::string variable = 'p' + std::to_string(pred) + '_' + suffix;
         shader.AddLine(variable + " = " + value + ';');
         declr_predicates.insert(std::move(variable));
     }
@@ -966,7 +1033,11 @@ private:
         if (header.writes_depth) {
             // The depth output is always 2 registers after the last color output, and current_reg
             // already contains one past the last color register.
-            shader.AddLine("gl_FragDepth = " + regs.GetRegisterAsFloat(current_reg + 1) + ';');
+
+            shader.AddLine(
+                "gl_FragDepth = " +
+                regs.GetRegisterAsFloat(static_cast<Tegra::Shader::Register>(current_reg) + 1) +
+                ';');
         }
     }
 
@@ -1368,7 +1439,7 @@ private:
                 if (instr.alu_integer.negate_b)
                     op_b = "-(" + op_b + ')';
 
-                std::string shift = std::to_string(instr.alu_integer.shift_amount.Value());
+                const std::string shift = std::to_string(instr.alu_integer.shift_amount.Value());
 
                 regs.SetRegisterToInteger(instr.gpr0, true, 0,
                                           "((" + op_a + " << " + shift + ") + " + op_b + ')', 1, 1);
@@ -1386,7 +1457,7 @@ private:
             case OpCode::Id::SEL_C:
             case OpCode::Id::SEL_R:
             case OpCode::Id::SEL_IMM: {
-                std::string condition =
+                const std::string condition =
                     GetPredicateCondition(instr.sel.pred, instr.sel.neg_pred != 0);
                 regs.SetRegisterToInteger(instr.gpr0, true, 0,
                                           '(' + condition + ") ? " + op_a + " : " + op_b, 1, 1);
@@ -1408,8 +1479,9 @@ private:
             case OpCode::Id::LOP3_C:
             case OpCode::Id::LOP3_R:
             case OpCode::Id::LOP3_IMM: {
-                std::string op_c = regs.GetRegisterAsInteger(instr.gpr39);
+                const std::string op_c = regs.GetRegisterAsInteger(instr.gpr39);
                 std::string lut;
+
                 if (opcode->GetId() == OpCode::Id::LOP3_R) {
                     lut = '(' + std::to_string(instr.alu.lop3.GetImmLut28()) + ')';
                 } else {
@@ -1424,9 +1496,9 @@ private:
             case OpCode::Id::IMNMX_IMM: {
                 ASSERT_MSG(instr.imnmx.exchange == Tegra::Shader::IMinMaxExchange::None,
                            "Unimplemented");
-                std::string condition =
+                const std::string condition =
                     GetPredicateCondition(instr.imnmx.pred, instr.imnmx.negate_pred != 0);
-                std::string parameters = op_a + ',' + op_b;
+                const std::string parameters = op_a + ',' + op_b;
                 regs.SetRegisterToInteger(instr.gpr0, instr.imnmx.is_signed, 0,
                                           '(' + condition + ") ? min(" + parameters + ") : max(" +
                                               parameters + ')',
@@ -1443,7 +1515,7 @@ private:
             break;
         }
         case OpCode::Type::Ffma: {
-            std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
+            const std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
             std::string op_b = instr.ffma.negate_b ? "-" : "";
             std::string op_c = instr.ffma.negate_c ? "-" : "";
 
@@ -1634,8 +1706,12 @@ private:
             switch (opcode->GetId()) {
             case OpCode::Id::LD_A: {
                 ASSERT_MSG(instr.attribute.fmt20.size == 0, "untested");
+                // Note: Shouldn't this be interp mode flat? As in no interpolation made.
+
+                Tegra::Shader::IpaMode input_mode{Tegra::Shader::IpaInterpMode::Perspective,
+                                                  Tegra::Shader::IpaSampleMode::Default};
                 regs.SetRegisterToInputAttibute(instr.gpr0, instr.attribute.fmt20.element,
-                                                instr.attribute.fmt20.index);
+                                                instr.attribute.fmt20.index, input_mode);
                 break;
             }
             case OpCode::Id::LD_C: {
@@ -1649,7 +1725,7 @@ private:
                 shader.AddLine("uint index = (" + regs.GetRegisterAsInteger(instr.gpr8, 0, false) +
                                " / 4) & (MAX_CONSTBUFFER_ELEMENTS - 1);");
 
-                std::string op_a =
+                const std::string op_a =
                     regs.GetUniformIndirect(instr.cbuf36.index, instr.cbuf36.offset + 0, "index",
                                             GLSLRegister::Type::Float);
 
@@ -1659,7 +1735,7 @@ private:
                     break;
 
                 case Tegra::Shader::UniformType::Double: {
-                    std::string op_b =
+                    const std::string op_b =
                         regs.GetUniformIndirect(instr.cbuf36.index, instr.cbuf36.offset + 4,
                                                 "index", GLSLRegister::Type::Float);
                     regs.SetRegisterToFloat(instr.gpr0, 0, op_a, 1, 1);
@@ -1683,10 +1759,35 @@ private:
                 break;
             }
             case OpCode::Id::TEX: {
-                const std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
-                const std::string op_b = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
-                const std::string sampler = GetSampler(instr.sampler);
-                const std::string coord = "vec2 coords = vec2(" + op_a + ", " + op_b + ");";
+                ASSERT_MSG(instr.tex.array == 0, "TEX arrays unimplemented");
+                Tegra::Shader::TextureType texture_type{instr.tex.texture_type};
+                std::string coord;
+
+                switch (texture_type) {
+                case Tegra::Shader::TextureType::Texture1D: {
+                    const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
+                    coord = "float coords = " + x + ';';
+                    break;
+                }
+                case Tegra::Shader::TextureType::Texture2D: {
+                    const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
+                    const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                    coord = "vec2 coords = vec2(" + x + ", " + y + ");";
+                    break;
+                }
+                default:
+                    LOG_CRITICAL(HW_GPU, "Unhandled texture type {}",
+                                 static_cast<u32>(texture_type));
+                    UNREACHABLE();
+
+                    // Fallback to interpreting as a 2D texture for now
+                    const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
+                    const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                    coord = "vec2 coords = vec2(" + x + ", " + y + ");";
+                    texture_type = Tegra::Shader::TextureType::Texture2D;
+                }
+
+                const std::string sampler = GetSampler(instr.sampler, texture_type, false);
                 // Add an extra scope and declare the texture coords inside to prevent
                 // overwriting them in case they are used as outputs of the texs instruction.
                 shader.AddLine("{");
@@ -1708,20 +1809,65 @@ private:
                 break;
             }
             case OpCode::Id::TEXS: {
-                const std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
-                const std::string op_b = regs.GetRegisterAsFloat(instr.gpr20);
-                const std::string sampler = GetSampler(instr.sampler);
-                const std::string coord = "vec2 coords = vec2(" + op_a + ", " + op_b + ");";
+                std::string coord;
+                Tegra::Shader::TextureType texture_type{instr.texs.GetTextureType()};
+                bool is_array{instr.texs.IsArrayTexture()};
 
+                switch (texture_type) {
+                case Tegra::Shader::TextureType::Texture2D: {
+                    if (is_array) {
+                        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.gpr20);
+                        coord = "vec3 coords = vec3(" + x + ", " + y + ", " + index + ");";
+                    } else {
+                        const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
+                        const std::string y = regs.GetRegisterAsFloat(instr.gpr20);
+                        coord = "vec2 coords = vec2(" + x + ", " + y + ");";
+                    }
+                    break;
+                }
+                default:
+                    LOG_CRITICAL(HW_GPU, "Unhandled texture type {}",
+                                 static_cast<u32>(texture_type));
+                    UNREACHABLE();
+
+                    // Fallback to interpreting as a 2D texture for now
+                    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 sampler = GetSampler(instr.sampler, texture_type, is_array);
                 const std::string texture = "texture(" + sampler + ", coords)";
                 WriteTexsInstruction(instr, coord, texture);
                 break;
             }
             case OpCode::Id::TLDS: {
-                const std::string op_a = regs.GetRegisterAsInteger(instr.gpr8);
-                const std::string op_b = regs.GetRegisterAsInteger(instr.gpr20);
-                const std::string sampler = GetSampler(instr.sampler);
-                const std::string coord = "ivec2 coords = ivec2(" + op_a + ", " + op_b + ");";
+                ASSERT(instr.tlds.GetTextureType() == Tegra::Shader::TextureType::Texture2D);
+                ASSERT(instr.tlds.IsArrayTexture() == false);
+                std::string coord;
+
+                switch (instr.tlds.GetTextureType()) {
+                case Tegra::Shader::TextureType::Texture2D: {
+                    if (instr.tlds.IsArrayTexture()) {
+                        LOG_CRITICAL(HW_GPU, "Unhandled 2d array texture");
+                        UNREACHABLE();
+                    } else {
+                        const std::string x = regs.GetRegisterAsInteger(instr.gpr8);
+                        const std::string y = regs.GetRegisterAsInteger(instr.gpr20);
+                        coord = "ivec2 coords = ivec2(" + x + ", " + y + ");";
+                    }
+                    break;
+                }
+                default:
+                    LOG_CRITICAL(HW_GPU, "Unhandled texture type {}",
+                                 static_cast<u32>(instr.tlds.GetTextureType()));
+                    UNREACHABLE();
+                }
+                const std::string sampler = GetSampler(instr.sampler, instr.tlds.GetTextureType(),
+                                                       instr.tlds.IsArrayTexture());
                 const std::string texture = "texelFetch(" + sampler + ", coords, 0)";
                 WriteTexsInstruction(instr, coord, texture);
                 break;
@@ -1729,12 +1875,12 @@ private:
             case OpCode::Id::TLD4: {
                 ASSERT(instr.tld4.texture_type == Tegra::Shader::TextureType::Texture2D);
                 ASSERT(instr.tld4.array == 0);
-                std::string coord{};
+                std::string coord;
 
                 switch (instr.tld4.texture_type) {
                 case Tegra::Shader::TextureType::Texture2D: {
-                    std::string x = regs.GetRegisterAsFloat(instr.gpr8);
-                    std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                    const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
+                    const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
                     coord = "vec2 coords = vec2(" + x + ", " + y + ");";
                     break;
                 }
@@ -1744,7 +1890,8 @@ private:
                     UNREACHABLE();
                 }
 
-                const std::string sampler = GetSampler(instr.sampler);
+                const std::string sampler =
+                    GetSampler(instr.sampler, instr.tld4.texture_type, false);
                 // Add an extra scope and declare the texture coords inside to prevent
                 // overwriting them in case they are used as outputs of the texs instruction.
                 shader.AddLine("{");
@@ -1770,7 +1917,8 @@ private:
                 const std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
                 const std::string op_b = regs.GetRegisterAsFloat(instr.gpr20);
                 // TODO(Subv): Figure out how the sampler type is encoded in the TLD4S instruction.
-                const std::string sampler = GetSampler(instr.sampler);
+                const std::string sampler =
+                    GetSampler(instr.sampler, Tegra::Shader::TextureType::Texture2D, false);
                 const std::string coord = "vec2 coords = vec2(" + op_a + ", " + op_b + ");";
                 const std::string texture = "textureGather(" + sampler + ", coords, " +
                                             std::to_string(instr.tld4s.component) + ')';
@@ -1816,12 +1964,12 @@ private:
             // We can't use the constant predicate as destination.
             ASSERT(instr.fsetp.pred3 != static_cast<u64>(Pred::UnusedIndex));
 
-            std::string second_pred =
+            const std::string second_pred =
                 GetPredicateCondition(instr.fsetp.pred39, instr.fsetp.neg_pred != 0);
 
-            std::string combiner = GetPredicateCombiner(instr.fsetp.op);
+            const std::string combiner = GetPredicateCombiner(instr.fsetp.op);
 
-            std::string predicate = GetPredicateComparison(instr.fsetp.cond, op_a, op_b);
+            const std::string predicate = GetPredicateComparison(instr.fsetp.cond, op_a, op_b);
             // Set the primary predicate to the result of Predicate OP SecondPredicate
             SetPredicate(instr.fsetp.pred3,
                          '(' + predicate + ") " + combiner + " (" + second_pred + ')');
@@ -1835,7 +1983,8 @@ private:
             break;
         }
         case OpCode::Type::IntegerSetPredicate: {
-            std::string op_a = regs.GetRegisterAsInteger(instr.gpr8, 0, instr.isetp.is_signed);
+            const std::string op_a =
+                regs.GetRegisterAsInteger(instr.gpr8, 0, instr.isetp.is_signed);
             std::string op_b;
 
             if (instr.is_b_imm) {
@@ -1852,12 +2001,12 @@ private:
             // We can't use the constant predicate as destination.
             ASSERT(instr.isetp.pred3 != static_cast<u64>(Pred::UnusedIndex));
 
-            std::string second_pred =
+            const std::string second_pred =
                 GetPredicateCondition(instr.isetp.pred39, instr.isetp.neg_pred != 0);
 
-            std::string combiner = GetPredicateCombiner(instr.isetp.op);
+            const std::string combiner = GetPredicateCombiner(instr.isetp.op);
 
-            std::string predicate = GetPredicateComparison(instr.isetp.cond, op_a, op_b);
+            const std::string predicate = GetPredicateComparison(instr.isetp.cond, op_a, op_b);
             // Set the primary predicate to the result of Predicate OP SecondPredicate
             SetPredicate(instr.isetp.pred3,
                          '(' + predicate + ") " + combiner + " (" + second_pred + ')');
@@ -1871,20 +2020,20 @@ private:
             break;
         }
         case OpCode::Type::PredicateSetPredicate: {
-            std::string op_a =
+            const std::string op_a =
                 GetPredicateCondition(instr.psetp.pred12, instr.psetp.neg_pred12 != 0);
-            std::string op_b =
+            const std::string op_b =
                 GetPredicateCondition(instr.psetp.pred29, instr.psetp.neg_pred29 != 0);
 
             // We can't use the constant predicate as destination.
             ASSERT(instr.psetp.pred3 != static_cast<u64>(Pred::UnusedIndex));
 
-            std::string second_pred =
+            const std::string second_pred =
                 GetPredicateCondition(instr.psetp.pred39, instr.psetp.neg_pred39 != 0);
 
-            std::string combiner = GetPredicateCombiner(instr.psetp.op);
+            const std::string combiner = GetPredicateCombiner(instr.psetp.op);
 
-            std::string predicate =
+            const std::string predicate =
                 '(' + op_a + ") " + GetPredicateCombiner(instr.psetp.cond) + " (" + op_b + ')';
 
             // Set the primary predicate to the result of Predicate OP SecondPredicate
@@ -1910,7 +2059,7 @@ private:
             std::string op_b = instr.fset.neg_b ? "-" : "";
 
             if (instr.is_b_imm) {
-                std::string imm = GetImmediate19(instr);
+                const std::string imm = GetImmediate19(instr);
                 if (instr.fset.neg_imm)
                     op_b += "(-" + imm + ')';
                 else
@@ -1930,13 +2079,14 @@ private:
 
             // The fset instruction sets a register to 1.0 or -1 (depending on the bf bit) if the
             // condition is true, and to 0 otherwise.
-            std::string second_pred =
+            const std::string second_pred =
                 GetPredicateCondition(instr.fset.pred39, instr.fset.neg_pred != 0);
 
-            std::string combiner = GetPredicateCombiner(instr.fset.op);
+            const std::string combiner = GetPredicateCombiner(instr.fset.op);
 
-            std::string predicate = "((" + GetPredicateComparison(instr.fset.cond, op_a, op_b) +
-                                    ") " + combiner + " (" + second_pred + "))";
+            const std::string predicate = "((" +
+                                          GetPredicateComparison(instr.fset.cond, op_a, op_b) +
+                                          ") " + combiner + " (" + second_pred + "))";
 
             if (instr.fset.bf) {
                 regs.SetRegisterToFloat(instr.gpr0, 0, predicate + " ? 1.0 : 0.0", 1, 1);
@@ -1947,7 +2097,7 @@ private:
             break;
         }
         case OpCode::Type::IntegerSet: {
-            std::string op_a = regs.GetRegisterAsInteger(instr.gpr8, 0, instr.iset.is_signed);
+            const std::string op_a = regs.GetRegisterAsInteger(instr.gpr8, 0, instr.iset.is_signed);
 
             std::string op_b;
 
@@ -1964,13 +2114,14 @@ private:
 
             // The iset instruction sets a register to 1.0 or -1 (depending on the bf bit) if the
             // condition is true, and to 0 otherwise.
-            std::string second_pred =
+            const std::string second_pred =
                 GetPredicateCondition(instr.iset.pred39, instr.iset.neg_pred != 0);
 
-            std::string combiner = GetPredicateCombiner(instr.iset.op);
+            const std::string combiner = GetPredicateCombiner(instr.iset.op);
 
-            std::string predicate = "((" + GetPredicateComparison(instr.iset.cond, op_a, op_b) +
-                                    ") " + combiner + " (" + second_pred + "))";
+            const std::string predicate = "((" +
+                                          GetPredicateComparison(instr.iset.cond, op_a, op_b) +
+                                          ") " + combiner + " (" + second_pred + "))";
 
             if (instr.iset.bf) {
                 regs.SetRegisterToFloat(instr.gpr0, 0, predicate + " ? 1.0 : 0.0", 1, 1);
@@ -2120,49 +2271,22 @@ private:
             case OpCode::Id::BRA: {
                 ASSERT_MSG(instr.bra.constant_buffer == 0,
                            "BRA with constant buffers are not implemented");
-                u32 target = offset + instr.bra.GetBranchTarget();
+                const u32 target = offset + instr.bra.GetBranchTarget();
                 shader.AddLine("{ jmp_to = " + std::to_string(target) + "u; break; }");
                 break;
             }
             case OpCode::Id::IPA: {
                 const auto& attribute = instr.attribute.fmt28;
                 const auto& reg = instr.gpr0;
-                ASSERT_MSG(instr.ipa.sample_mode == Tegra::Shader::IpaSampleMode::Default,
-                           "Unhandled IPA sample mode: {}",
-                           static_cast<u32>(instr.ipa.sample_mode.Value()));
-                ASSERT_MSG(instr.ipa.saturate == 0, "IPA saturate not implemented");
-                switch (instr.ipa.interp_mode) {
-                case Tegra::Shader::IpaInterpMode::Linear:
-                    if (stage == Maxwell3D::Regs::ShaderStage::Fragment &&
-                        attribute.index == Attribute::Index::Position) {
-                        switch (attribute.element) {
-                        case 0:
-                            shader.AddLine(regs.GetRegisterAsFloat(reg) + " = gl_FragCoord.x;");
-                            break;
-                        case 1:
-                            shader.AddLine(regs.GetRegisterAsFloat(reg) + " = gl_FragCoord.y;");
-                            break;
-                        case 2:
-                            shader.AddLine(regs.GetRegisterAsFloat(reg) + " = gl_FragCoord.z;");
-                            break;
-                        case 3:
-                            shader.AddLine(regs.GetRegisterAsFloat(reg) + " = 1.0;");
-                            break;
-                        }
-                    } else {
-                        regs.SetRegisterToInputAttibute(reg, attribute.element, attribute.index);
-                    }
-                    break;
-                case Tegra::Shader::IpaInterpMode::Perspective:
-                    regs.SetRegisterToInputAttibute(reg, attribute.element, attribute.index);
-                    break;
-                default:
-                    LOG_CRITICAL(HW_GPU, "Unhandled IPA mode: {}",
-                                 static_cast<u32>(instr.ipa.interp_mode.Value()));
-                    UNREACHABLE();
-                    regs.SetRegisterToInputAttibute(reg, attribute.element, attribute.index);
-                }
 
+                Tegra::Shader::IpaMode input_mode{instr.ipa.interp_mode.Value(),
+                                                  instr.ipa.sample_mode.Value()};
+                regs.SetRegisterToInputAttibute(reg, attribute.element, attribute.index,
+                                                input_mode);
+
+                if (instr.ipa.saturate) {
+                    regs.SetRegisterToFloat(reg, 0, regs.GetRegisterAsFloat(reg), 1, 1, true);
+                }
                 break;
             }
             case OpCode::Id::SSY: {
@@ -2171,7 +2295,7 @@ private:
                 // has a similar structure to the BRA opcode.
                 ASSERT_MSG(instr.bra.constant_buffer == 0, "Constant buffer SSY is not supported");
 
-                u32 target = offset + instr.bra.GetBranchTarget();
+                const u32 target = offset + instr.bra.GetBranchTarget();
                 EmitPushToSSYStack(target);
                 break;
             }
@@ -2265,10 +2389,10 @@ private:
                     shader.AddLine("case " + std::to_string(label) + "u: {");
                     ++shader.scope;
 
-                    auto next_it = labels.lower_bound(label + 1);
-                    u32 next_label = next_it == labels.end() ? subroutine.end : *next_it;
+                    const auto next_it = labels.lower_bound(label + 1);
+                    const u32 next_label = next_it == labels.end() ? subroutine.end : *next_it;
 
-                    u32 compile_end = CompileRange(label, next_label);
+                    const u32 compile_end = CompileRange(label, next_label);
                     if (compile_end > next_label && compile_end != PROGRAM_END) {
                         // This happens only when there is a label inside a IF/LOOP block
                         shader.AddLine(" jmp_to = " + std::to_string(compile_end) + "u; break; }");
@@ -2331,7 +2455,8 @@ boost::optional<ProgramResult> DecompileProgram(const ProgramCode& program_code,
                                                 Maxwell3D::Regs::ShaderStage stage,
                                                 const std::string& suffix) {
     try {
-        auto subroutines = ControlFlowAnalyzer(program_code, main_offset, suffix).GetSubroutines();
+        const auto subroutines =
+            ControlFlowAnalyzer(program_code, main_offset, suffix).GetSubroutines();
         GLSLGenerator generator(subroutines, program_code, main_offset, stage, suffix);
         return ProgramResult{generator.GetShaderCode(), generator.GetEntries()};
     } catch (const DecompileFail& exception) {

src/video_core/renderer_opengl/gl_shader_gen.cpp

@@ -42,6 +42,7 @@ layout (std140) uniform vs_config {
 };
 
 void main() {
+    position = vec4(0.0, 0.0, 0.0, 0.0);
     exec_vertex();
 )";
 

src/video_core/renderer_opengl/gl_shader_gen.h

@@ -9,6 +9,7 @@
 #include <vector>
 
 #include "common/common_types.h"
+#include "video_core/engines/shader_bytecode.h"
 
 namespace OpenGL::GLShader {
 
@@ -53,6 +54,10 @@ public:
         return BufferBaseNames[static_cast<size_t>(stage)] + std::to_string(index);
     }
 
+    u32 GetHash() const {
+        return (static_cast<u32>(stage) << 16) | index;
+    }
+
 private:
     static constexpr std::array<const char*, Maxwell::MaxShaderStage> BufferBaseNames = {
         "buffer_vs_c", "buffer_tessc_c", "buffer_tesse_c", "buffer_gs_c", "buffer_fs_c",
@@ -69,8 +74,9 @@ class SamplerEntry {
     using Maxwell = Tegra::Engines::Maxwell3D::Regs;
 
 public:
-    SamplerEntry(Maxwell::ShaderStage stage, size_t offset, size_t index)
-        : offset(offset), stage(stage), sampler_index(index) {}
+    SamplerEntry(Maxwell::ShaderStage stage, size_t offset, size_t index,
+                 Tegra::Shader::TextureType type, bool is_array)
+        : offset(offset), stage(stage), sampler_index(index), type(type), is_array(is_array) {}
 
     size_t GetOffset() const {
         return offset;
@@ -85,8 +91,45 @@ public:
     }
 
     std::string GetName() const {
-        return std::string(TextureSamplerNames[static_cast<size_t>(stage)]) + '[' +
-               std::to_string(sampler_index) + ']';
+        return std::string(TextureSamplerNames[static_cast<size_t>(stage)]) + '_' +
+               std::to_string(sampler_index);
+    }
+
+    std::string GetTypeString() const {
+        using Tegra::Shader::TextureType;
+        std::string glsl_type;
+
+        switch (type) {
+        case TextureType::Texture1D:
+            glsl_type = "sampler1D";
+            break;
+        case TextureType::Texture2D:
+            glsl_type = "sampler2D";
+            break;
+        case TextureType::Texture3D:
+            glsl_type = "sampler3D";
+            break;
+        case TextureType::TextureCube:
+            glsl_type = "samplerCube";
+            break;
+        default:
+            UNIMPLEMENTED();
+        }
+        if (is_array)
+            glsl_type += "Array";
+        return glsl_type;
+    }
+
+    Tegra::Shader::TextureType GetType() const {
+        return type;
+    }
+
+    bool IsArray() const {
+        return is_array;
+    }
+
+    u32 GetHash() const {
+        return (static_cast<u32>(stage) << 16) | static_cast<u32>(sampler_index);
     }
 
     static std::string GetArrayName(Maxwell::ShaderStage stage) {
@@ -97,11 +140,14 @@ private:
     static constexpr std::array<const char*, Maxwell::MaxShaderStage> TextureSamplerNames = {
         "tex_vs", "tex_tessc", "tex_tesse", "tex_gs", "tex_fs",
     };
+
     /// Offset in TSC memory from which to read the sampler object, as specified by the sampling
     /// instruction.
     size_t offset;
-    Maxwell::ShaderStage stage; ///< Shader stage where this sampler was used.
-    size_t sampler_index;       ///< Value used to index into the generated GLSL sampler array.
+    Maxwell::ShaderStage stage;      ///< Shader stage where this sampler was used.
+    size_t sampler_index;            ///< Value used to index into the generated GLSL sampler array.
+    Tegra::Shader::TextureType type; ///< The type used to sample this texture (Texture2D, etc)
+    bool is_array; ///< Whether the texture is being sampled as an array texture or not.
 };
 
 struct ShaderEntries {

src/video_core/renderer_opengl/gl_shader_util.cpp

@@ -25,7 +25,7 @@ GLuint LoadShader(const char* source, GLenum type) {
     default:
         UNREACHABLE();
     }
-    GLuint shader_id = glCreateShader(type);
+    const GLuint shader_id = glCreateShader(type);
     glShaderSource(shader_id, 1, &source, nullptr);
     LOG_DEBUG(Render_OpenGL, "Compiling {} shader...", debug_type);
     glCompileShader(shader_id);

src/video_core/renderer_opengl/gl_state.cpp

@@ -200,9 +200,9 @@ void OpenGLState::Apply() const {
         const auto& texture_unit = texture_units[i];
         const auto& cur_state_texture_unit = cur_state.texture_units[i];
 
-        if (texture_unit.texture_2d != cur_state_texture_unit.texture_2d) {
+        if (texture_unit.texture != cur_state_texture_unit.texture) {
             glActiveTexture(TextureUnits::MaxwellTexture(static_cast<int>(i)).Enum());
-            glBindTexture(GL_TEXTURE_2D, texture_unit.texture_2d);
+            glBindTexture(texture_unit.target, texture_unit.texture);
         }
         if (texture_unit.sampler != cur_state_texture_unit.sampler) {
             glBindSampler(static_cast<GLuint>(i), texture_unit.sampler);
@@ -214,7 +214,7 @@ void OpenGLState::Apply() const {
             texture_unit.swizzle.a != cur_state_texture_unit.swizzle.a) {
             std::array<GLint, 4> mask = {texture_unit.swizzle.r, texture_unit.swizzle.g,
                                          texture_unit.swizzle.b, texture_unit.swizzle.a};
-            glTexParameteriv(GL_TEXTURE_2D, GL_TEXTURE_SWIZZLE_RGBA, mask.data());
+            glTexParameteriv(texture_unit.target, GL_TEXTURE_SWIZZLE_RGBA, mask.data());
         }
     }
 
@@ -287,7 +287,7 @@ void OpenGLState::Apply() const {
 
 OpenGLState& OpenGLState::UnbindTexture(GLuint handle) {
     for (auto& unit : texture_units) {
-        if (unit.texture_2d == handle) {
+        if (unit.texture == handle) {
             unit.Unbind();
         }
     }

src/video_core/renderer_opengl/gl_state.h

@@ -94,8 +94,9 @@ public:
 
     // 3 texture units - one for each that is used in PICA fragment shader emulation
     struct TextureUnit {
-        GLuint texture_2d; // GL_TEXTURE_BINDING_2D
-        GLuint sampler;    // GL_SAMPLER_BINDING
+        GLuint texture; // GL_TEXTURE_BINDING_2D
+        GLuint sampler; // GL_SAMPLER_BINDING
+        GLenum target;
         struct {
             GLint r; // GL_TEXTURE_SWIZZLE_R
             GLint g; // GL_TEXTURE_SWIZZLE_G
@@ -104,7 +105,7 @@ public:
         } swizzle;
 
         void Unbind() {
-            texture_2d = 0;
+            texture = 0;
             swizzle.r = GL_RED;
             swizzle.g = GL_GREEN;
             swizzle.b = GL_BLUE;
@@ -114,6 +115,7 @@ public:
         void Reset() {
             Unbind();
             sampler = 0;
+            target = GL_TEXTURE_2D;
         }
     };
     std::array<TextureUnit, 32> texture_units;

src/video_core/renderer_opengl/gl_stream_buffer.cpp

@@ -29,7 +29,7 @@ OGLStreamBuffer::OGLStreamBuffer(GLenum target, GLsizeiptr size, bool prefer_coh
     if (GLAD_GL_ARB_buffer_storage) {
         persistent = true;
         coherent = prefer_coherent;
-        GLbitfield flags =
+        const GLbitfield flags =
             GL_MAP_WRITE_BIT | GL_MAP_PERSISTENT_BIT | (coherent ? GL_MAP_COHERENT_BIT : 0);
         glBufferStorage(gl_target, allocate_size, nullptr, flags);
         mapped_ptr = static_cast<u8*>(glMapBufferRange(

src/video_core/renderer_opengl/renderer_opengl.cpp

@@ -177,7 +177,7 @@ void RendererOpenGL::LoadFBToScreenInfo(const Tegra::FramebufferConfig& framebuf
                                        Memory::GetPointer(framebuffer_addr),
                                        gl_framebuffer_data.data(), true);
 
-        state.texture_units[0].texture_2d = screen_info.texture.resource.handle;
+        state.texture_units[0].texture = screen_info.texture.resource.handle;
         state.Apply();
 
         glActiveTexture(GL_TEXTURE0);
@@ -194,7 +194,7 @@ void RendererOpenGL::LoadFBToScreenInfo(const Tegra::FramebufferConfig& framebuf
 
         glPixelStorei(GL_UNPACK_ROW_LENGTH, 0);
 
-        state.texture_units[0].texture_2d = 0;
+        state.texture_units[0].texture = 0;
         state.Apply();
     }
 }
@@ -205,7 +205,7 @@ void RendererOpenGL::LoadFBToScreenInfo(const Tegra::FramebufferConfig& framebuf
  */
 void RendererOpenGL::LoadColorToActiveGLTexture(u8 color_r, u8 color_g, u8 color_b, u8 color_a,
                                                 const TextureInfo& texture) {
-    state.texture_units[0].texture_2d = texture.resource.handle;
+    state.texture_units[0].texture = texture.resource.handle;
     state.Apply();
 
     glActiveTexture(GL_TEXTURE0);
@@ -214,7 +214,7 @@ void RendererOpenGL::LoadColorToActiveGLTexture(u8 color_r, u8 color_g, u8 color
     // Update existing texture
     glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, 1, 1, 0, GL_RGBA, GL_UNSIGNED_BYTE, framebuffer_data);
 
-    state.texture_units[0].texture_2d = 0;
+    state.texture_units[0].texture = 0;
     state.Apply();
 }
 
@@ -260,7 +260,7 @@ void RendererOpenGL::InitOpenGLObjects() {
     // Allocation of storage is deferred until the first frame, when we
     // know the framebuffer size.
 
-    state.texture_units[0].texture_2d = screen_info.texture.resource.handle;
+    state.texture_units[0].texture = screen_info.texture.resource.handle;
     state.Apply();
 
     glActiveTexture(GL_TEXTURE0);
@@ -272,7 +272,7 @@ void RendererOpenGL::InitOpenGLObjects() {
 
     screen_info.display_texture = screen_info.texture.resource.handle;
 
-    state.texture_units[0].texture_2d = 0;
+    state.texture_units[0].texture = 0;
     state.Apply();
 
     // Clear screen to black
@@ -305,14 +305,14 @@ void RendererOpenGL::ConfigureFramebufferTexture(TextureInfo& texture,
         UNREACHABLE();
     }
 
-    state.texture_units[0].texture_2d = texture.resource.handle;
+    state.texture_units[0].texture = texture.resource.handle;
     state.Apply();
 
     glActiveTexture(GL_TEXTURE0);
     glTexImage2D(GL_TEXTURE_2D, 0, internal_format, texture.width, texture.height, 0,
                  texture.gl_format, texture.gl_type, nullptr);
 
-    state.texture_units[0].texture_2d = 0;
+    state.texture_units[0].texture = 0;
     state.Apply();
 }
 
@@ -354,14 +354,14 @@ void RendererOpenGL::DrawScreenTriangles(const ScreenInfo& screen_info, float x,
         ScreenRectVertex(x + w, y + h, texcoords.bottom * scale_u, right * scale_v),
     }};
 
-    state.texture_units[0].texture_2d = screen_info.display_texture;
+    state.texture_units[0].texture = screen_info.display_texture;
     state.texture_units[0].swizzle = {GL_RED, GL_GREEN, GL_BLUE, GL_ALPHA};
     state.Apply();
 
     glBufferSubData(GL_ARRAY_BUFFER, 0, sizeof(vertices), vertices.data());
     glDrawArrays(GL_TRIANGLE_STRIP, 0, 4);
 
-    state.texture_units[0].texture_2d = 0;
+    state.texture_units[0].texture = 0;
     state.Apply();
 }
 

src/video_core/textures/texture.h

@@ -170,8 +170,12 @@ struct TICEntry {
         BitField<0, 16, u32> width_minus_1;
         BitField<23, 4, TextureType> texture_type;
     };
-    u16 height_minus_1;
-    INSERT_PADDING_BYTES(10);
+    union {
+        BitField<0, 16, u32> height_minus_1;
+        BitField<16, 15, u32> depth_minus_1;
+    };
+
+    INSERT_PADDING_BYTES(8);
 
     GPUVAddr Address() const {
         return static_cast<GPUVAddr>((static_cast<GPUVAddr>(address_high) << 32) | address_low);
@@ -192,6 +196,10 @@ struct TICEntry {
         return height_minus_1 + 1;
     }
 
+    u32 Depth() const {
+        return depth_minus_1 + 1;
+    }
+
     u32 BlockHeight() const {
         ASSERT(header_version == TICHeaderVersion::BlockLinear ||
                header_version == TICHeaderVersion::BlockLinearColorKey);

src/yuzu/CMakeLists.txt

@@ -43,6 +43,8 @@ add_executable(yuzu
     game_list.cpp
     game_list.h
     game_list_p.h
+    game_list_worker.cpp
+    game_list_worker.h
     hotkeys.cpp
     hotkeys.h
     main.cpp

src/yuzu/about_dialog.cpp

@@ -11,7 +11,7 @@ AboutDialog::AboutDialog(QWidget* parent) : QDialog(parent), ui(new Ui::AboutDia
     ui->setupUi(this);
     ui->labelLogo->setPixmap(QIcon::fromTheme("yuzu").pixmap(200));
     ui->labelBuildInfo->setText(
-        ui->labelBuildInfo->text().arg(Common::g_build_name, Common::g_scm_branch,
+        ui->labelBuildInfo->text().arg(Common::g_build_fullname, Common::g_scm_branch,
                                        Common::g_scm_desc, QString(Common::g_build_date).left(10)));
 }
 

src/yuzu/bootmanager.cpp

@@ -256,6 +256,7 @@ void GRenderWindow::InitRenderTarget() {
     QGLFormat fmt;
     fmt.setVersion(3, 3);
     fmt.setProfile(QGLFormat::CoreProfile);
+    fmt.setSwapInterval(false);
 
     // Requests a forward-compatible context, which is required to get a 3.2+ context on OS X
     fmt.setOption(QGL::NoDeprecatedFunctions);

src/yuzu/configuration/config.cpp

@@ -102,6 +102,20 @@ void Config::ReadValues() {
 
     qt_config->beginGroup("Data Storage");
     Settings::values.use_virtual_sd = qt_config->value("use_virtual_sd", true).toBool();
+    FileUtil::GetUserPath(
+        FileUtil::UserPath::NANDDir,
+        qt_config
+            ->value("nand_directory",
+                    QString::fromStdString(FileUtil::GetUserPath(FileUtil::UserPath::NANDDir)))
+            .toString()
+            .toStdString());
+    FileUtil::GetUserPath(
+        FileUtil::UserPath::SDMCDir,
+        qt_config
+            ->value("sdmc_directory",
+                    QString::fromStdString(FileUtil::GetUserPath(FileUtil::UserPath::SDMCDir)))
+            .toString()
+            .toStdString());
     qt_config->endGroup();
 
     qt_config->beginGroup("System");
@@ -222,6 +236,10 @@ void Config::SaveValues() {
 
     qt_config->beginGroup("Data Storage");
     qt_config->setValue("use_virtual_sd", Settings::values.use_virtual_sd);
+    qt_config->setValue("nand_directory",
+                        QString::fromStdString(FileUtil::GetUserPath(FileUtil::UserPath::NANDDir)));
+    qt_config->setValue("sdmc_directory",
+                        QString::fromStdString(FileUtil::GetUserPath(FileUtil::UserPath::SDMCDir)));
     qt_config->endGroup();
 
     qt_config->beginGroup("System");

src/yuzu/game_list.cpp

@@ -13,21 +13,15 @@
 #include <QKeyEvent>
 #include <QMenu>
 #include <QThreadPool>
-#include <boost/container/flat_map.hpp>
 #include <fmt/format.h>
 #include "common/common_paths.h"
 #include "common/common_types.h"
 #include "common/file_util.h"
 #include "common/logging/log.h"
-#include "core/file_sys/content_archive.h"
-#include "core/file_sys/control_metadata.h"
-#include "core/file_sys/registered_cache.h"
-#include "core/file_sys/romfs.h"
-#include "core/file_sys/vfs_real.h"
-#include "core/hle/service/filesystem/filesystem.h"
-#include "core/loader/loader.h"
+#include "core/file_sys/patch_manager.h"
 #include "yuzu/game_list.h"
 #include "yuzu/game_list_p.h"
+#include "yuzu/game_list_worker.h"
 #include "yuzu/main.h"
 #include "yuzu/ui_settings.h"
 
@@ -232,6 +226,7 @@ GameList::GameList(FileSys::VirtualFilesystem vfs, GMainWindow* parent)
     item_model->insertColumns(0, COLUMN_COUNT);
     item_model->setHeaderData(COLUMN_NAME, Qt::Horizontal, "Name");
     item_model->setHeaderData(COLUMN_COMPATIBILITY, Qt::Horizontal, "Compatibility");
+    item_model->setHeaderData(COLUMN_ADD_ONS, Qt::Horizontal, "Add-ons");
     item_model->setHeaderData(COLUMN_FILE_TYPE, Qt::Horizontal, "File type");
     item_model->setHeaderData(COLUMN_SIZE, Qt::Horizontal, "Size");
 
@@ -432,27 +427,7 @@ void GameList::LoadInterfaceLayout() {
     item_model->sort(header->sortIndicatorSection(), header->sortIndicatorOrder());
 }
 
-const QStringList GameList::supported_file_extensions = {"nso", "nro", "nca", "xci"};
-
-static bool HasSupportedFileExtension(const std::string& file_name) {
-    const QFileInfo file = QFileInfo(QString::fromStdString(file_name));
-    return GameList::supported_file_extensions.contains(file.suffix(), Qt::CaseInsensitive);
-}
-
-static bool IsExtractedNCAMain(const std::string& file_name) {
-    return QFileInfo(QString::fromStdString(file_name)).fileName() == "main";
-}
-
-static QString FormatGameName(const std::string& physical_name) {
-    const QString physical_name_as_qstring = QString::fromStdString(physical_name);
-    const QFileInfo file_info(physical_name_as_qstring);
-
-    if (IsExtractedNCAMain(physical_name)) {
-        return file_info.dir().path();
-    }
-
-    return physical_name_as_qstring;
-}
+const QStringList GameList::supported_file_extensions = {"nso", "nro", "nca", "xci", "nsp"};
 
 void GameList::RefreshGameDirectory() {
     if (!UISettings::values.gamedir.isEmpty() && current_worker != nullptr) {
@@ -461,175 +436,3 @@ void GameList::RefreshGameDirectory() {
         PopulateAsync(UISettings::values.gamedir, UISettings::values.gamedir_deepscan);
     }
 }
-
-static void GetMetadataFromControlNCA(const std::shared_ptr<FileSys::NCA>& nca,
-                                      std::vector<u8>& icon, std::string& name) {
-    const auto control_dir = FileSys::ExtractRomFS(nca->GetRomFS());
-    if (control_dir == nullptr)
-        return;
-
-    const auto nacp_file = control_dir->GetFile("control.nacp");
-    if (nacp_file == nullptr)
-        return;
-    FileSys::NACP nacp(nacp_file);
-    name = nacp.GetApplicationName();
-
-    FileSys::VirtualFile icon_file = nullptr;
-    for (const auto& language : FileSys::LANGUAGE_NAMES) {
-        icon_file = control_dir->GetFile("icon_" + std::string(language) + ".dat");
-        if (icon_file != nullptr) {
-            icon = icon_file->ReadAllBytes();
-            break;
-        }
-    }
-}
-
-GameListWorker::GameListWorker(
-    FileSys::VirtualFilesystem vfs, QString dir_path, bool deep_scan,
-    const std::unordered_map<std::string, std::pair<QString, QString>>& compatibility_list)
-    : vfs(std::move(vfs)), dir_path(std::move(dir_path)), deep_scan(deep_scan),
-      compatibility_list(compatibility_list) {}
-
-GameListWorker::~GameListWorker() = default;
-
-void GameListWorker::AddInstalledTitlesToGameList(std::shared_ptr<FileSys::RegisteredCache> cache) {
-    const auto installed_games = cache->ListEntriesFilter(FileSys::TitleType::Application,
-                                                          FileSys::ContentRecordType::Program);
-
-    for (const auto& game : installed_games) {
-        const auto& file = cache->GetEntryUnparsed(game);
-        std::unique_ptr<Loader::AppLoader> loader = Loader::GetLoader(file);
-        if (!loader)
-            continue;
-
-        std::vector<u8> icon;
-        std::string name;
-        u64 program_id = 0;
-        loader->ReadProgramId(program_id);
-
-        const auto& control = cache->GetEntry(game.title_id, FileSys::ContentRecordType::Control);
-        if (control != nullptr)
-            GetMetadataFromControlNCA(control, icon, name);
-        emit EntryReady({
-            new GameListItemPath(
-                FormatGameName(file->GetFullPath()), icon, QString::fromStdString(name),
-                QString::fromStdString(Loader::GetFileTypeString(loader->GetFileType())),
-                program_id),
-            new GameListItem(
-                QString::fromStdString(Loader::GetFileTypeString(loader->GetFileType()))),
-            new GameListItemSize(file->GetSize()),
-        });
-    }
-
-    const auto control_data = cache->ListEntriesFilter(FileSys::TitleType::Application,
-                                                       FileSys::ContentRecordType::Control);
-
-    for (const auto& entry : control_data) {
-        const auto nca = cache->GetEntry(entry);
-        if (nca != nullptr)
-            nca_control_map.insert_or_assign(entry.title_id, nca);
-    }
-}
-
-void GameListWorker::FillControlMap(const std::string& dir_path) {
-    const auto nca_control_callback = [this](u64* num_entries_out, const std::string& directory,
-                                             const std::string& virtual_name) -> bool {
-        std::string physical_name = directory + DIR_SEP + virtual_name;
-
-        if (stop_processing)
-            return false; // Breaks the callback loop.
-
-        bool is_dir = FileUtil::IsDirectory(physical_name);
-        QFileInfo file_info(physical_name.c_str());
-        if (!is_dir && file_info.suffix().toStdString() == "nca") {
-            auto nca =
-                std::make_shared<FileSys::NCA>(vfs->OpenFile(physical_name, FileSys::Mode::Read));
-            if (nca->GetType() == FileSys::NCAContentType::Control)
-                nca_control_map.insert_or_assign(nca->GetTitleId(), nca);
-        }
-        return true;
-    };
-
-    FileUtil::ForeachDirectoryEntry(nullptr, dir_path, nca_control_callback);
-}
-
-void GameListWorker::AddFstEntriesToGameList(const std::string& dir_path, unsigned int recursion) {
-    const auto callback = [this, recursion](u64* num_entries_out, const std::string& directory,
-                                            const std::string& virtual_name) -> bool {
-        std::string physical_name = directory + DIR_SEP + virtual_name;
-
-        if (stop_processing)
-            return false; // Breaks the callback loop.
-
-        bool is_dir = FileUtil::IsDirectory(physical_name);
-        if (!is_dir &&
-            (HasSupportedFileExtension(physical_name) || IsExtractedNCAMain(physical_name))) {
-            std::unique_ptr<Loader::AppLoader> loader =
-                Loader::GetLoader(vfs->OpenFile(physical_name, FileSys::Mode::Read));
-            if (!loader || ((loader->GetFileType() == Loader::FileType::Unknown ||
-                             loader->GetFileType() == Loader::FileType::Error) &&
-                            !UISettings::values.show_unknown))
-                return true;
-
-            std::vector<u8> icon;
-            const auto res1 = loader->ReadIcon(icon);
-
-            u64 program_id = 0;
-            const auto res2 = loader->ReadProgramId(program_id);
-
-            std::string name = " ";
-            const auto res3 = loader->ReadTitle(name);
-
-            if (res1 != Loader::ResultStatus::Success && res3 != Loader::ResultStatus::Success &&
-                res2 == Loader::ResultStatus::Success) {
-                // Use from metadata pool.
-                if (nca_control_map.find(program_id) != nca_control_map.end()) {
-                    const auto nca = nca_control_map[program_id];
-                    GetMetadataFromControlNCA(nca, icon, name);
-                }
-            }
-
-            auto it = FindMatchingCompatibilityEntry(compatibility_list, program_id);
-
-            // The game list uses this as compatibility number for untested games
-            QString compatibility("99");
-            if (it != compatibility_list.end())
-                compatibility = it->second.first;
-
-            emit EntryReady({
-                new GameListItemPath(
-                    FormatGameName(physical_name), icon, QString::fromStdString(name),
-                    QString::fromStdString(Loader::GetFileTypeString(loader->GetFileType())),
-                    program_id),
-                new GameListItemCompat(compatibility),
-                new GameListItem(
-                    QString::fromStdString(Loader::GetFileTypeString(loader->GetFileType()))),
-                new GameListItemSize(FileUtil::GetSize(physical_name)),
-            });
-        } else if (is_dir && recursion > 0) {
-            watch_list.append(QString::fromStdString(physical_name));
-            AddFstEntriesToGameList(physical_name, recursion - 1);
-        }
-
-        return true;
-    };
-
-    FileUtil::ForeachDirectoryEntry(nullptr, dir_path, callback);
-}
-
-void GameListWorker::run() {
-    stop_processing = false;
-    watch_list.append(dir_path);
-    FillControlMap(dir_path.toStdString());
-    AddInstalledTitlesToGameList(Service::FileSystem::GetUserNANDContents());
-    AddInstalledTitlesToGameList(Service::FileSystem::GetSystemNANDContents());
-    AddInstalledTitlesToGameList(Service::FileSystem::GetSDMCContents());
-    AddFstEntriesToGameList(dir_path.toStdString(), deep_scan ? 256 : 0);
-    nca_control_map.clear();
-    emit Finished(watch_list);
-}
-
-void GameListWorker::Cancel() {
-    this->disconnect();
-    stop_processing = true;
-}

src/yuzu/game_list.h

@@ -38,6 +38,7 @@ public:
     enum {
         COLUMN_NAME,
         COLUMN_COMPATIBILITY,
+        COLUMN_ADD_ONS,
         COLUMN_FILE_TYPE,
         COLUMN_SIZE,
         COLUMN_COUNT, // Number of columns

src/yuzu/game_list_p.h

@@ -6,9 +6,7 @@
 
 #include <algorithm>
 #include <array>
-#include <atomic>
 #include <map>
-#include <memory>
 #include <string>
 #include <unordered_map>
 #include <utility>
@@ -16,7 +14,6 @@
 #include <QCoreApplication>
 #include <QImage>
 #include <QObject>
-#include <QRunnable>
 #include <QStandardItem>
 #include <QString>
 
@@ -26,12 +23,6 @@
 #include "yuzu/ui_settings.h"
 #include "yuzu/util/util.h"
 
-namespace FileSys {
-class NCA;
-class RegisteredCache;
-class VfsFilesystem;
-} // namespace FileSys
-
 /**
  * Gets the default icon (for games without valid SMDH)
  * @param large If true, returns large icon (48x48), otherwise returns small icon (24x24)
@@ -43,17 +34,6 @@ static QPixmap GetDefaultIcon(u32 size) {
     return icon;
 }
 
-static auto FindMatchingCompatibilityEntry(
-    const std::unordered_map<std::string, std::pair<QString, QString>>& compatibility_list,
-    u64 program_id) {
-    return std::find_if(
-        compatibility_list.begin(), compatibility_list.end(),
-        [program_id](const std::pair<std::string, std::pair<QString, QString>>& element) {
-            std::string pid = fmt::format("{:016X}", program_id);
-            return element.first == pid;
-        });
-}
-
 class GameListItem : public QStandardItem {
 
 public:
@@ -197,49 +177,13 @@ public:
     }
 };
 
-/**
- * Asynchronous worker object for populating the game list.
- * Communicates with other threads through Qt's signal/slot system.
- */
-class GameListWorker : public QObject, public QRunnable {
-    Q_OBJECT
-
-public:
-    GameListWorker(
-        std::shared_ptr<FileSys::VfsFilesystem> vfs, QString dir_path, bool deep_scan,
-        const std::unordered_map<std::string, std::pair<QString, QString>>& compatibility_list);
-    ~GameListWorker() override;
-
-public slots:
-    /// Starts the processing of directory tree information.
-    void run() override;
-    /// Tells the worker that it should no longer continue processing. Thread-safe.
-    void Cancel();
-
-signals:
-    /**
-     * The `EntryReady` signal is emitted once an entry has been prepared and is ready
-     * to be added to the game list.
-     * @param entry_items a list with `QStandardItem`s that make up the columns of the new entry.
-     */
-    void EntryReady(QList<QStandardItem*> entry_items);
-
-    /**
-     * After the worker has traversed the game directory looking for entries, this signal is emmited
-     * with a list of folders that should be watched for changes as well.
-     */
-    void Finished(QStringList watch_list);
-
-private:
-    std::shared_ptr<FileSys::VfsFilesystem> vfs;
-    std::map<u64, std::shared_ptr<FileSys::NCA>> nca_control_map;
-    QStringList watch_list;
-    QString dir_path;
-    bool deep_scan;
-    const std::unordered_map<std::string, std::pair<QString, QString>>& compatibility_list;
-    std::atomic_bool stop_processing;
-
-    void AddInstalledTitlesToGameList(std::shared_ptr<FileSys::RegisteredCache> cache);
-    void FillControlMap(const std::string& dir_path);
-    void AddFstEntriesToGameList(const std::string& dir_path, unsigned int recursion = 0);
-};
+inline auto FindMatchingCompatibilityEntry(
+    const std::unordered_map<std::string, std::pair<QString, QString>>& compatibility_list,
+    u64 program_id) {
+    return std::find_if(
+        compatibility_list.begin(), compatibility_list.end(),
+        [program_id](const std::pair<std::string, std::pair<QString, QString>>& element) {
+            std::string pid = fmt::format("{:016X}", program_id);
+            return element.first == pid;
+        });
+}

src/yuzu/game_list_worker.cpp

@@ -0,0 +1,239 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <memory>
+#include <string>
+#include <utility>
+#include <vector>
+
+#include <QDir>
+#include <QFileInfo>
+
+#include "common/common_paths.h"
+#include "common/file_util.h"
+#include "core/file_sys/content_archive.h"
+#include "core/file_sys/control_metadata.h"
+#include "core/file_sys/mode.h"
+#include "core/file_sys/nca_metadata.h"
+#include "core/file_sys/patch_manager.h"
+#include "core/file_sys/registered_cache.h"
+#include "core/hle/service/filesystem/filesystem.h"
+#include "core/loader/loader.h"
+#include "yuzu/game_list.h"
+#include "yuzu/game_list_p.h"
+#include "yuzu/game_list_worker.h"
+#include "yuzu/ui_settings.h"
+
+namespace {
+void GetMetadataFromControlNCA(const FileSys::PatchManager& patch_manager,
+                               const std::shared_ptr<FileSys::NCA>& nca, std::vector<u8>& icon,
+                               std::string& name) {
+    auto [nacp, icon_file] = patch_manager.ParseControlNCA(nca);
+    if (icon_file != nullptr)
+        icon = icon_file->ReadAllBytes();
+    if (nacp != nullptr)
+        name = nacp->GetApplicationName();
+}
+
+bool HasSupportedFileExtension(const std::string& file_name) {
+    const QFileInfo file = QFileInfo(QString::fromStdString(file_name));
+    return GameList::supported_file_extensions.contains(file.suffix(), Qt::CaseInsensitive);
+}
+
+bool IsExtractedNCAMain(const std::string& file_name) {
+    return QFileInfo(QString::fromStdString(file_name)).fileName() == "main";
+}
+
+QString FormatGameName(const std::string& physical_name) {
+    const QString physical_name_as_qstring = QString::fromStdString(physical_name);
+    const QFileInfo file_info(physical_name_as_qstring);
+
+    if (IsExtractedNCAMain(physical_name)) {
+        return file_info.dir().path();
+    }
+
+    return physical_name_as_qstring;
+}
+
+QString FormatPatchNameVersions(const FileSys::PatchManager& patch_manager, bool updatable = true) {
+    QString out;
+    for (const auto& kv : patch_manager.GetPatchVersionNames()) {
+        if (!updatable && kv.first == FileSys::PatchType::Update)
+            continue;
+
+        if (kv.second.empty()) {
+            out.append(fmt::format("{}\n", FileSys::FormatPatchTypeName(kv.first)).c_str());
+        } else {
+            out.append(fmt::format("{} ({})\n", FileSys::FormatPatchTypeName(kv.first), kv.second)
+                           .c_str());
+        }
+    }
+
+    out.chop(1);
+    return out;
+}
+} // Anonymous namespace
+
+GameListWorker::GameListWorker(
+    FileSys::VirtualFilesystem vfs, QString dir_path, bool deep_scan,
+    const std::unordered_map<std::string, std::pair<QString, QString>>& compatibility_list)
+    : vfs(std::move(vfs)), dir_path(std::move(dir_path)), deep_scan(deep_scan),
+      compatibility_list(compatibility_list) {}
+
+GameListWorker::~GameListWorker() = default;
+
+void GameListWorker::AddInstalledTitlesToGameList() {
+    const auto cache = Service::FileSystem::GetUnionContents();
+    const auto installed_games = cache->ListEntriesFilter(FileSys::TitleType::Application,
+                                                          FileSys::ContentRecordType::Program);
+
+    for (const auto& game : installed_games) {
+        const auto& file = cache->GetEntryUnparsed(game);
+        std::unique_ptr<Loader::AppLoader> loader = Loader::GetLoader(file);
+        if (!loader)
+            continue;
+
+        std::vector<u8> icon;
+        std::string name;
+        u64 program_id = 0;
+        loader->ReadProgramId(program_id);
+
+        const FileSys::PatchManager patch{program_id};
+        const auto& control = cache->GetEntry(game.title_id, FileSys::ContentRecordType::Control);
+        if (control != nullptr)
+            GetMetadataFromControlNCA(patch, control, icon, name);
+
+        auto it = FindMatchingCompatibilityEntry(compatibility_list, program_id);
+
+        // The game list uses this as compatibility number for untested games
+        QString compatibility("99");
+        if (it != compatibility_list.end())
+            compatibility = it->second.first;
+
+        emit EntryReady({
+            new GameListItemPath(
+                FormatGameName(file->GetFullPath()), icon, QString::fromStdString(name),
+                QString::fromStdString(Loader::GetFileTypeString(loader->GetFileType())),
+                program_id),
+            new GameListItemCompat(compatibility),
+            new GameListItem(FormatPatchNameVersions(patch)),
+            new GameListItem(
+                QString::fromStdString(Loader::GetFileTypeString(loader->GetFileType()))),
+            new GameListItemSize(file->GetSize()),
+        });
+    }
+
+    const auto control_data = cache->ListEntriesFilter(FileSys::TitleType::Application,
+                                                       FileSys::ContentRecordType::Control);
+
+    for (const auto& entry : control_data) {
+        const auto nca = cache->GetEntry(entry);
+        if (nca != nullptr)
+            nca_control_map.insert_or_assign(entry.title_id, nca);
+    }
+}
+
+void GameListWorker::FillControlMap(const std::string& dir_path) {
+    const auto nca_control_callback = [this](u64* num_entries_out, const std::string& directory,
+                                             const std::string& virtual_name) -> bool {
+        std::string physical_name = directory + DIR_SEP + virtual_name;
+
+        if (stop_processing)
+            return false; // Breaks the callback loop.
+
+        bool is_dir = FileUtil::IsDirectory(physical_name);
+        QFileInfo file_info(physical_name.c_str());
+        if (!is_dir && file_info.suffix().toStdString() == "nca") {
+            auto nca =
+                std::make_shared<FileSys::NCA>(vfs->OpenFile(physical_name, FileSys::Mode::Read));
+            if (nca->GetType() == FileSys::NCAContentType::Control)
+                nca_control_map.insert_or_assign(nca->GetTitleId(), nca);
+        }
+        return true;
+    };
+
+    FileUtil::ForeachDirectoryEntry(nullptr, dir_path, nca_control_callback);
+}
+
+void GameListWorker::AddFstEntriesToGameList(const std::string& dir_path, unsigned int recursion) {
+    const auto callback = [this, recursion](u64* num_entries_out, const std::string& directory,
+                                            const std::string& virtual_name) -> bool {
+        std::string physical_name = directory + DIR_SEP + virtual_name;
+
+        if (stop_processing)
+            return false; // Breaks the callback loop.
+
+        bool is_dir = FileUtil::IsDirectory(physical_name);
+        if (!is_dir &&
+            (HasSupportedFileExtension(physical_name) || IsExtractedNCAMain(physical_name))) {
+            std::unique_ptr<Loader::AppLoader> loader =
+                Loader::GetLoader(vfs->OpenFile(physical_name, FileSys::Mode::Read));
+            if (!loader || ((loader->GetFileType() == Loader::FileType::Unknown ||
+                             loader->GetFileType() == Loader::FileType::Error) &&
+                            !UISettings::values.show_unknown))
+                return true;
+
+            std::vector<u8> icon;
+            const auto res1 = loader->ReadIcon(icon);
+
+            u64 program_id = 0;
+            const auto res2 = loader->ReadProgramId(program_id);
+
+            std::string name = " ";
+            const auto res3 = loader->ReadTitle(name);
+
+            const FileSys::PatchManager patch{program_id};
+
+            if (res1 != Loader::ResultStatus::Success && res3 != Loader::ResultStatus::Success &&
+                res2 == Loader::ResultStatus::Success) {
+                // Use from metadata pool.
+                if (nca_control_map.find(program_id) != nca_control_map.end()) {
+                    const auto nca = nca_control_map[program_id];
+                    GetMetadataFromControlNCA(patch, nca, icon, name);
+                }
+            }
+
+            auto it = FindMatchingCompatibilityEntry(compatibility_list, program_id);
+
+            // The game list uses this as compatibility number for untested games
+            QString compatibility("99");
+            if (it != compatibility_list.end())
+                compatibility = it->second.first;
+
+            emit EntryReady({
+                new GameListItemPath(
+                    FormatGameName(physical_name), icon, QString::fromStdString(name),
+                    QString::fromStdString(Loader::GetFileTypeString(loader->GetFileType())),
+                    program_id),
+                new GameListItemCompat(compatibility),
+                new GameListItem(FormatPatchNameVersions(patch, loader->IsRomFSUpdatable())),
+                new GameListItem(
+                    QString::fromStdString(Loader::GetFileTypeString(loader->GetFileType()))),
+                new GameListItemSize(FileUtil::GetSize(physical_name)),
+            });
+        } else if (is_dir && recursion > 0) {
+            watch_list.append(QString::fromStdString(physical_name));
+            AddFstEntriesToGameList(physical_name, recursion - 1);
+        }
+
+        return true;
+    };
+
+    FileUtil::ForeachDirectoryEntry(nullptr, dir_path, callback);
+}
+
+void GameListWorker::run() {
+    stop_processing = false;
+    watch_list.append(dir_path);
+    FillControlMap(dir_path.toStdString());
+    AddInstalledTitlesToGameList();
+    AddFstEntriesToGameList(dir_path.toStdString(), deep_scan ? 256 : 0);
+    nca_control_map.clear();
+    emit Finished(watch_list);
+}
+
+void GameListWorker::Cancel() {
+    this->disconnect();
+    stop_processing = true;
+}