fix stupid

k_server_session: translate special header for non-HLE requests

.ci/scripts/windows/docker.sh

@@ -56,7 +56,6 @@ for i in package/*.exe; do
   x86_64-w64-mingw32-strip "${i}"
 done
 
-pip3 install pefile
 python3 .ci/scripts/windows/scan_dll.py package/*.exe package/imageformats/*.dll "package/"
 
 # copy FFmpeg libraries

.ci/scripts/windows/install-vulkan-sdk.ps1

@@ -0,0 +1,33 @@
+# SPDX-FileCopyrightText: 2023 yuzu Emulator Project
+# SPDX-License-Identifier: GPL-3.0-or-later
+
+$ErrorActionPreference = "Stop"
+
+$VulkanSDKVer = "1.3.250.1"
+$ExeFile = "VulkanSDK-$VulkanSDKVer-Installer.exe"
+$Uri = "https://sdk.lunarg.com/sdk/download/$VulkanSDKVer/windows/$ExeFile"
+$Destination = "./$ExeFile"
+
+echo "Downloading Vulkan SDK $VulkanSDKVer from $Uri"
+$WebClient = New-Object System.Net.WebClient
+$WebClient.DownloadFile($Uri, $Destination)
+echo "Finished downloading $ExeFile"
+
+$VULKAN_SDK = "C:/VulkanSDK/$VulkanSDKVer"
+$Arguments = "--root `"$VULKAN_SDK`" --accept-licenses --default-answer --confirm-command install"
+
+echo "Installing Vulkan SDK $VulkanSDKVer"
+$InstallProcess = Start-Process -FilePath $Destination -NoNewWindow -PassThru -Wait -ArgumentList $Arguments
+$ExitCode = $InstallProcess.ExitCode
+
+if ($ExitCode -ne 0) {
+    echo "Error installing Vulkan SDK $VulkanSDKVer (Error: $ExitCode)"
+    Exit $ExitCode
+}
+
+echo "Finished installing Vulkan SDK $VulkanSDKVer"
+
+if ("$env:GITHUB_ACTIONS" -eq "true") {
+    echo "VULKAN_SDK=$VULKAN_SDK" | Out-File -FilePath $env:GITHUB_ENV -Encoding utf8 -Append
+    echo "$VULKAN_SDK/Bin" | Out-File -FilePath $env:GITHUB_PATH -Encoding utf8 -Append
+}

.ci/templates/build-msvc.yml

@@ -7,9 +7,12 @@ parameters:
   version: ''
 
 steps:
-- script: choco install vulkan-sdk
-  displayName: 'Install vulkan-sdk'
-- script: refreshenv && mkdir build && cd build && cmake -E env CXXFLAGS="/Gw /GA /Gr /Ob2" cmake -G "Visual Studio 17 2022" -A x64 -DCMAKE_POLICY_DEFAULT_CMP0069=NEW -DYUZU_ENABLE_LTO=ON -DYUZU_USE_BUNDLED_QT=1 -DYUZU_USE_BUNDLED_SDL2=1 -DYUZU_USE_QT_WEB_ENGINE=ON -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON -DYUZU_ENABLE_COMPATIBILITY_REPORTING=${COMPAT} -DYUZU_TESTS=OFF -DUSE_DISCORD_PRESENCE=ON -DENABLE_QT_TRANSLATION=ON -DDISPLAY_VERSION=${{ parameters['version'] }} -DCMAKE_BUILD_TYPE=Release -DYUZU_CRASH_DUMPS=ON .. && cd ..
+- task: Powershell@2
+  displayName: 'Install Vulkan SDK'
+  inputs:
+    targetType: 'filePath'
+    filePath: './.ci/scripts/windows/install-vulkan-sdk.ps1'
+- script: refreshenv && glslangValidator --version && mkdir build && cd build && cmake -E env CXXFLAGS="/Gw /GA /Gr /Ob2" cmake -G "Visual Studio 17 2022" -A x64 -DCMAKE_POLICY_DEFAULT_CMP0069=NEW -DYUZU_ENABLE_LTO=ON -DYUZU_USE_BUNDLED_QT=1 -DYUZU_USE_BUNDLED_SDL2=1 -DYUZU_USE_QT_WEB_ENGINE=ON -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON -DYUZU_ENABLE_COMPATIBILITY_REPORTING=${COMPAT} -DYUZU_TESTS=OFF -DUSE_DISCORD_PRESENCE=ON -DENABLE_QT_TRANSLATION=ON -DDISPLAY_VERSION=${{ parameters['version'] }} -DCMAKE_BUILD_TYPE=Release -DYUZU_CRASH_DUMPS=ON .. && cd ..
   displayName: 'Configure CMake'
 - task: MSBuild@1
   displayName: 'Build'

.github/workflows/android-build.yml

@@ -10,6 +10,7 @@ on:
 jobs:
   android:
     runs-on: ubuntu-latest
+    if: ${{ github.repository == 'yuzu-emu/yuzu-android' }}
     steps:
       - uses: actions/checkout@v3
         with:

.github/workflows/verify.yml

@@ -73,6 +73,10 @@ jobs:
     needs: format
     runs-on: windows-2022
     steps:
+      - uses: actions/checkout@v3
+        with:
+          submodules: recursive
+          fetch-depth: 0
       - name: Set up cache
         uses: actions/cache@v3
         with:
@@ -81,22 +85,22 @@ jobs:
           restore-keys: |
             ${{ runner.os }}-msvc-
       - name: Install dependencies
-        # due to how chocolatey works, only cmd.exe is supported here
-        shell: cmd
+        shell: pwsh
         run: |
-          choco install vulkan-sdk wget
-          call refreshenv
-          wget https://github.com/mbitsnbites/buildcache/releases/download/v0.27.6/buildcache-windows.zip
-          7z x buildcache-windows.zip
-          copy buildcache\bin\buildcache.exe C:\ProgramData\chocolatey\bin
-          rmdir buildcache
-          echo %PATH% >> %GITHUB_PATH%
+          $ErrorActionPreference = "Stop"
+          $BuildCacheVer = "v0.28.4"
+          $File = "buildcache-windows.zip"
+          $Uri = "https://github.com/mbitsnbites/buildcache/releases/download/$BuildCacheVer/$File"
+          $WebClient = New-Object System.Net.WebClient
+          $WebClient.DownloadFile($Uri, $File)
+          7z x $File
+          $CurrentDir = Convert-Path .
+          echo "$CurrentDir/buildcache/bin" | Out-File -FilePath $env:GITHUB_PATH -Encoding utf8 -Append
+      - name: Install Vulkan SDK
+        shell: pwsh
+        run: .\.ci\scripts\windows\install-vulkan-sdk.ps1
       - name: Set up MSVC
         uses: ilammy/msvc-dev-cmd@v1
-      - uses: actions/checkout@v3
-        with:
-          submodules: recursive
-          fetch-depth: 0
       - name: Configure
         env:
           CC: cl.exe

CMakeLists.txt

@@ -285,7 +285,7 @@ find_package(ZLIB 1.2 REQUIRED)
 find_package(zstd 1.5 REQUIRED)
 
 if (NOT YUZU_USE_EXTERNAL_VULKAN_HEADERS)
-    find_package(Vulkan 1.3.246 REQUIRED)
+    find_package(Vulkan 1.3.256 REQUIRED)
 endif()
 
 if (ENABLE_LIBUSB)
@@ -489,7 +489,7 @@ if (ENABLE_SDL2)
     if (YUZU_USE_BUNDLED_SDL2)
         # Detect toolchain and platform
         if ((MSVC_VERSION GREATER_EQUAL 1920 AND MSVC_VERSION LESS 1940) AND ARCHITECTURE_x86_64)
-            set(SDL2_VER "SDL2-2.28.0")
+            set(SDL2_VER "SDL2-2.28.1")
         else()
             message(FATAL_ERROR "No bundled SDL2 binaries for your toolchain. Disable YUZU_USE_BUNDLED_SDL2 and provide your own.")
         endif()

src/android/app/src/main/AndroidManifest.xml

@@ -22,7 +22,7 @@ SPDX-License-Identifier: GPL-3.0-or-later
         android:label="@string/app_name_suffixed"
         android:icon="@drawable/ic_launcher"
         android:allowBackup="true"
-        android:hasFragileUserData="true"
+        android:hasFragileUserData="false"
         android:supportsRtl="true"
         android:isGame="true"
         android:localeConfig="@xml/locales_config"

src/android/app/src/main/java/org/yuzu/yuzu_emu/adapters/HomeSettingAdapter.kt

@@ -12,6 +12,7 @@ import androidx.core.content.res.ResourcesCompat
 import androidx.recyclerview.widget.RecyclerView
 import org.yuzu.yuzu_emu.R
 import org.yuzu.yuzu_emu.databinding.CardHomeOptionBinding
+import org.yuzu.yuzu_emu.fragments.MessageDialogFragment
 import org.yuzu.yuzu_emu.model.HomeSetting
 
 class HomeSettingAdapter(private val activity: AppCompatActivity, var options: List<HomeSetting>) :
@@ -34,7 +35,14 @@ class HomeSettingAdapter(private val activity: AppCompatActivity, var options: L
 
     override fun onClick(view: View) {
         val holder = view.tag as HomeOptionViewHolder
-        holder.option.onClick.invoke()
+        if (holder.option.isEnabled.invoke()) {
+            holder.option.onClick.invoke()
+        } else {
+            MessageDialogFragment.newInstance(
+                holder.option.disabledTitleId,
+                holder.option.disabledMessageId
+            ).show(activity.supportFragmentManager, MessageDialogFragment.TAG)
+        }
     }
 
     inner class HomeOptionViewHolder(val binding: CardHomeOptionBinding) :
@@ -65,6 +73,12 @@ class HomeSettingAdapter(private val activity: AppCompatActivity, var options: L
                             R.drawable.premium_background
                         )
             }
+
+            if (!option.isEnabled.invoke()) {
+                binding.optionTitle.alpha = 0.5f
+                binding.optionDescription.alpha = 0.5f
+                binding.optionIcon.alpha = 0.5f
+            }
         }
     }
 }

src/android/app/src/main/java/org/yuzu/yuzu_emu/fragments/HomeSettingsFragment.kt

@@ -73,102 +73,113 @@ class HomeSettingsFragment : Fragment() {
                 HomeSetting(
                     R.string.advanced_settings,
                     R.string.settings_description,
-                    R.drawable.ic_settings
-                ) { SettingsActivity.launch(requireContext(), SettingsFile.FILE_NAME_CONFIG, "") }
+                    R.drawable.ic_settings,
+                    { SettingsActivity.launch(requireContext(), SettingsFile.FILE_NAME_CONFIG, "") }
+                )
             )
             add(
                 HomeSetting(
                     R.string.open_user_folder,
                     R.string.open_user_folder_description,
-                    R.drawable.ic_folder_open
-                ) { openFileManager() }
+                    R.drawable.ic_folder_open,
+                    { openFileManager() }
+                )
             )
             add(
                 HomeSetting(
                     R.string.preferences_theme,
                     R.string.theme_and_color_description,
-                    R.drawable.ic_palette
-                ) { SettingsActivity.launch(requireContext(), Settings.SECTION_THEME, "") }
-            )
-
-            if (GpuDriverHelper.supportsCustomDriverLoading()) {
-                add(
-                    HomeSetting(
-                        R.string.install_gpu_driver,
-                        R.string.install_gpu_driver_description,
-                        R.drawable.ic_exit
-                    ) { driverInstaller() }
+                    R.drawable.ic_palette,
+                    { SettingsActivity.launch(requireContext(), Settings.SECTION_THEME, "") }
                 )
-            }
-
+            )
+            add(
+                HomeSetting(
+                    R.string.install_gpu_driver,
+                    R.string.install_gpu_driver_description,
+                    R.drawable.ic_exit,
+                    { driverInstaller() },
+                    { GpuDriverHelper.supportsCustomDriverLoading() },
+                    R.string.custom_driver_not_supported,
+                    R.string.custom_driver_not_supported_description
+                )
+            )
             add(
                 HomeSetting(
                     R.string.install_amiibo_keys,
                     R.string.install_amiibo_keys_description,
-                    R.drawable.ic_nfc
-                ) { mainActivity.getAmiiboKey.launch(arrayOf("*/*")) }
+                    R.drawable.ic_nfc,
+                    { mainActivity.getAmiiboKey.launch(arrayOf("*/*")) }
+                )
             )
             add(
                 HomeSetting(
                     R.string.install_game_content,
                     R.string.install_game_content_description,
-                    R.drawable.ic_system_update_alt
-                ) { mainActivity.installGameUpdate.launch(arrayOf("*/*")) }
+                    R.drawable.ic_system_update_alt,
+                    { mainActivity.installGameUpdate.launch(arrayOf("*/*")) }
+                )
             )
             add(
                 HomeSetting(
                     R.string.select_games_folder,
                     R.string.select_games_folder_description,
-                    R.drawable.ic_add
-                ) {
-                    mainActivity.getGamesDirectory.launch(
-                        Intent(Intent.ACTION_OPEN_DOCUMENT_TREE).data
-                    )
-                }
+                    R.drawable.ic_add,
+                    {
+                        mainActivity.getGamesDirectory.launch(
+                            Intent(Intent.ACTION_OPEN_DOCUMENT_TREE).data
+                        )
+                    }
+                )
             )
             add(
                 HomeSetting(
                     R.string.manage_save_data,
                     R.string.import_export_saves_description,
-                    R.drawable.ic_save
-                ) {
-                    ImportExportSavesFragment().show(
-                        parentFragmentManager,
-                        ImportExportSavesFragment.TAG
-                    )
-                }
+                    R.drawable.ic_save,
+                    {
+                        ImportExportSavesFragment().show(
+                            parentFragmentManager,
+                            ImportExportSavesFragment.TAG
+                        )
+                    }
+                )
             )
             add(
                 HomeSetting(
                     R.string.install_prod_keys,
                     R.string.install_prod_keys_description,
-                    R.drawable.ic_unlock
-                ) { mainActivity.getProdKey.launch(arrayOf("*/*")) }
+                    R.drawable.ic_unlock,
+                    { mainActivity.getProdKey.launch(arrayOf("*/*")) }
+                )
             )
             add(
                 HomeSetting(
                     R.string.install_firmware,
                     R.string.install_firmware_description,
-                    R.drawable.ic_firmware
-                ) { mainActivity.getFirmware.launch(arrayOf("application/zip")) }
+                    R.drawable.ic_firmware,
+                    { mainActivity.getFirmware.launch(arrayOf("application/zip")) }
+                )
             )
             add(
                 HomeSetting(
                     R.string.share_log,
                     R.string.share_log_description,
-                    R.drawable.ic_log
-                ) { shareLog() }
+                    R.drawable.ic_log,
+                    { shareLog() }
+                )
             )
             add(
                 HomeSetting(
                     R.string.about,
                     R.string.about_description,
-                    R.drawable.ic_info_outline
-                ) {
-                    exitTransition = MaterialSharedAxis(MaterialSharedAxis.X, true)
-                    parentFragmentManager.primaryNavigationFragment?.findNavController()
-                        ?.navigate(R.id.action_homeSettingsFragment_to_aboutFragment)
-                }
+                    R.drawable.ic_info_outline,
+                    {
+                        exitTransition = MaterialSharedAxis(MaterialSharedAxis.X, true)
+                        parentFragmentManager.primaryNavigationFragment?.findNavController()
+                            ?.navigate(R.id.action_homeSettingsFragment_to_aboutFragment)
+                    }
+                )
             )
         }
 
@@ -178,12 +189,13 @@ class HomeSettingsFragment : Fragment() {
                 HomeSetting(
                     R.string.get_early_access,
                     R.string.get_early_access_description,
-                    R.drawable.ic_diamond
-                ) {
-                    exitTransition = MaterialSharedAxis(MaterialSharedAxis.X, true)
-                    parentFragmentManager.primaryNavigationFragment?.findNavController()
-                        ?.navigate(R.id.action_homeSettingsFragment_to_earlyAccessFragment)
-                }
+                    R.drawable.ic_diamond,
+                    {
+                        exitTransition = MaterialSharedAxis(MaterialSharedAxis.X, true)
+                        parentFragmentManager.primaryNavigationFragment?.findNavController()
+                            ?.navigate(R.id.action_homeSettingsFragment_to_earlyAccessFragment)
+                    }
+                )
             )
         }
 

src/android/app/src/main/java/org/yuzu/yuzu_emu/model/HomeSetting.kt

@@ -7,5 +7,8 @@ data class HomeSetting(
     val titleId: Int,
     val descriptionId: Int,
     val iconId: Int,
-    val onClick: () -> Unit
+    val onClick: () -> Unit,
+    val isEnabled: () -> Boolean = { true },
+    val disabledTitleId: Int = 0,
+    val disabledMessageId: Int = 0
 )

src/android/app/src/main/res/values/strings.xml

@@ -113,6 +113,8 @@
     <string name="install_game_content_success_install">%1$d installed successfully</string>
     <string name="install_game_content_success_overwrite">%1$d overwritten successfully</string>
     <string name="install_game_content_help_link">https://yuzu-emu.org/help/quickstart/#dumping-installed-updates</string>
+    <string name="custom_driver_not_supported">Custom drivers not supported</string>
+    <string name="custom_driver_not_supported_description">Custom driver loading isn\'t currently supported for this device.\nCheck this option again in the future to see if support was added!</string>
 
     <!-- About screen strings -->
     <string name="gaia_is_not_real">Gaia isn\'t real</string>

src/audio_core/device/device_session.cpp

@@ -92,9 +92,9 @@ void DeviceSession::AppendBuffers(std::span<const AudioBuffer> buffers) {
         if (type == Sink::StreamType::In) {
             stream->AppendBuffer(new_buffer, tmp_samples);
         } else {
-            system.ApplicationMemory().ReadBlockUnsafe(buffer.samples, tmp_samples.data(),
-                                                       buffer.size);
-            stream->AppendBuffer(new_buffer, tmp_samples);
+            Core::Memory::CpuGuestMemory<s16, Core::Memory::GuestMemoryFlags::UnsafeRead> samples(
+                system.ApplicationMemory(), buffer.samples, buffer.size / sizeof(s16));
+            stream->AppendBuffer(new_buffer, samples);
         }
     }
 }

src/audio_core/renderer/command/data_source/decode.cpp

@@ -28,7 +28,6 @@ constexpr std::array<u8, 3> PitchBySrcQuality = {4, 8, 4};
 template <typename T>
 static u32 DecodePcm(Core::Memory::Memory& memory, std::span<s16> out_buffer,
                      const DecodeArg& req) {
-    std::array<T, TempBufferSize> tmp_samples{};
     constexpr s32 min{std::numeric_limits<s16>::min()};
     constexpr s32 max{std::numeric_limits<s16>::max()};
 
@@ -49,19 +48,18 @@ static u32 DecodePcm(Core::Memory::Memory& memory, std::span<s16> out_buffer,
         const VAddr source{req.buffer +
                            (((req.start_offset + req.offset) * channel_count) * sizeof(T))};
         const u64 size{channel_count * samples_to_decode};
-        const u64 size_bytes{size * sizeof(T)};
-
-        memory.ReadBlockUnsafe(source, tmp_samples.data(), size_bytes);
 
+        Core::Memory::CpuGuestMemory<T, Core::Memory::GuestMemoryFlags::UnsafeRead> samples(
+            memory, source, size);
         if constexpr (std::is_floating_point_v<T>) {
             for (u32 i = 0; i < samples_to_decode; i++) {
-                auto sample{static_cast<s32>(tmp_samples[i * channel_count + req.target_channel] *
+                auto sample{static_cast<s32>(samples[i * channel_count + req.target_channel] *
                                              std::numeric_limits<s16>::max())};
                 out_buffer[i] = static_cast<s16>(std::clamp(sample, min, max));
             }
         } else {
             for (u32 i = 0; i < samples_to_decode; i++) {
-                out_buffer[i] = tmp_samples[i * channel_count + req.target_channel];
+                out_buffer[i] = samples[i * channel_count + req.target_channel];
             }
         }
     } break;
@@ -74,16 +72,17 @@ static u32 DecodePcm(Core::Memory::Memory& memory, std::span<s16> out_buffer,
         }
 
         const VAddr source{req.buffer + ((req.start_offset + req.offset) * sizeof(T))};
-        memory.ReadBlockUnsafe(source, tmp_samples.data(), samples_to_decode * sizeof(T));
+        Core::Memory::CpuGuestMemory<T, Core::Memory::GuestMemoryFlags::UnsafeRead> samples(
+            memory, source, samples_to_decode);
 
         if constexpr (std::is_floating_point_v<T>) {
             for (u32 i = 0; i < samples_to_decode; i++) {
-                auto sample{static_cast<s32>(tmp_samples[i * channel_count + req.target_channel] *
+                auto sample{static_cast<s32>(samples[i * channel_count + req.target_channel] *
                                              std::numeric_limits<s16>::max())};
                 out_buffer[i] = static_cast<s16>(std::clamp(sample, min, max));
             }
         } else {
-            std::memcpy(out_buffer.data(), tmp_samples.data(), samples_to_decode * sizeof(s16));
+            std::memcpy(out_buffer.data(), samples.data(), samples_to_decode * sizeof(s16));
         }
         break;
     }
@@ -101,7 +100,6 @@ static u32 DecodePcm(Core::Memory::Memory& memory, std::span<s16> out_buffer,
  */
 static u32 DecodeAdpcm(Core::Memory::Memory& memory, std::span<s16> out_buffer,
                        const DecodeArg& req) {
-    std::array<u8, TempBufferSize> wavebuffer{};
     constexpr u32 SamplesPerFrame{14};
     constexpr u32 NibblesPerFrame{16};
 
@@ -139,7 +137,8 @@ static u32 DecodeAdpcm(Core::Memory::Memory& memory, std::span<s16> out_buffer,
     }
 
     const auto size{std::max((samples_to_process / 8U) * SamplesPerFrame, 8U)};
-    memory.ReadBlockUnsafe(req.buffer + position_in_frame / 2, wavebuffer.data(), size);
+    Core::Memory::CpuGuestMemory<u8, Core::Memory::GuestMemoryFlags::UnsafeRead> wavebuffer(
+        memory, req.buffer + position_in_frame / 2, size);
 
     auto context{req.adpcm_context};
     auto header{context->header};

src/audio_core/renderer/command/effect/aux_.cpp

@@ -21,23 +21,13 @@ static void ResetAuxBufferDsp(Core::Memory::Memory& memory, const CpuAddr aux_in
     }
 
     AuxInfo::AuxInfoDsp info{};
-    auto info_ptr{&info};
-    bool host_safe{(aux_info & Core::Memory::YUZU_PAGEMASK) <=
-                   (Core::Memory::YUZU_PAGESIZE - sizeof(AuxInfo::AuxInfoDsp))};
+    memory.ReadBlockUnsafe(aux_info, &info, sizeof(AuxInfo::AuxInfoDsp));
 
-    if (host_safe) [[likely]] {
-        info_ptr = memory.GetPointer<AuxInfo::AuxInfoDsp>(aux_info);
-    } else {
-        memory.ReadBlockUnsafe(aux_info, info_ptr, sizeof(AuxInfo::AuxInfoDsp));
-    }
+    info.read_offset = 0;
+    info.write_offset = 0;
+    info.total_sample_count = 0;
 
-    info_ptr->read_offset = 0;
-    info_ptr->write_offset = 0;
-    info_ptr->total_sample_count = 0;
-
-    if (!host_safe) [[unlikely]] {
-        memory.WriteBlockUnsafe(aux_info, info_ptr, sizeof(AuxInfo::AuxInfoDsp));
-    }
+    memory.WriteBlockUnsafe(aux_info, &info, sizeof(AuxInfo::AuxInfoDsp));
 }
 
 /**
@@ -86,17 +76,9 @@ static u32 WriteAuxBufferDsp(Core::Memory::Memory& memory, CpuAddr send_info_,
     }
 
     AuxInfo::AuxInfoDsp send_info{};
-    auto send_ptr = &send_info;
-    bool host_safe = (send_info_ & Core::Memory::YUZU_PAGEMASK) <=
-                     (Core::Memory::YUZU_PAGESIZE - sizeof(AuxInfo::AuxInfoDsp));
+    memory.ReadBlockUnsafe(send_info_, &send_info, sizeof(AuxInfo::AuxInfoDsp));
 
-    if (host_safe) [[likely]] {
-        send_ptr = memory.GetPointer<AuxInfo::AuxInfoDsp>(send_info_);
-    } else {
-        memory.ReadBlockUnsafe(send_info_, send_ptr, sizeof(AuxInfo::AuxInfoDsp));
-    }
-
-    u32 target_write_offset{send_ptr->write_offset + write_offset};
+    u32 target_write_offset{send_info.write_offset + write_offset};
     if (target_write_offset > count_max) {
         return 0;
     }
@@ -105,15 +87,9 @@ static u32 WriteAuxBufferDsp(Core::Memory::Memory& memory, CpuAddr send_info_,
     u32 read_pos{0};
     while (write_count > 0) {
         u32 to_write{std::min(count_max - target_write_offset, write_count)};
-        const auto write_addr = send_buffer + target_write_offset * sizeof(s32);
-        bool write_safe{(write_addr & Core::Memory::YUZU_PAGEMASK) <=
-                        (Core::Memory::YUZU_PAGESIZE - (write_addr + to_write * sizeof(s32)))};
-        if (write_safe) [[likely]] {
-            auto ptr = memory.GetPointer(write_addr);
-            std::memcpy(ptr, &input[read_pos], to_write * sizeof(s32));
-        } else {
-            memory.WriteBlockUnsafe(send_buffer + target_write_offset * sizeof(s32),
-                                    &input[read_pos], to_write * sizeof(s32));
+        if (to_write > 0) {
+            const auto write_addr = send_buffer + target_write_offset * sizeof(s32);
+            memory.WriteBlockUnsafe(write_addr, &input[read_pos], to_write * sizeof(s32));
         }
         target_write_offset = (target_write_offset + to_write) % count_max;
         write_count -= to_write;
@@ -121,13 +97,10 @@ static u32 WriteAuxBufferDsp(Core::Memory::Memory& memory, CpuAddr send_info_,
     }
 
     if (update_count) {
-        send_ptr->write_offset = (send_ptr->write_offset + update_count) % count_max;
-    }
-
-    if (!host_safe) [[unlikely]] {
-        memory.WriteBlockUnsafe(send_info_, send_ptr, sizeof(AuxInfo::AuxInfoDsp));
+        send_info.write_offset = (send_info.write_offset + update_count) % count_max;
     }
 
+    memory.WriteBlockUnsafe(send_info_, &send_info, sizeof(AuxInfo::AuxInfoDsp));
     return write_count_;
 }
 
@@ -174,17 +147,9 @@ static u32 ReadAuxBufferDsp(Core::Memory::Memory& memory, CpuAddr return_info_,
     }
 
     AuxInfo::AuxInfoDsp return_info{};
-    auto return_ptr = &return_info;
-    bool host_safe = (return_info_ & Core::Memory::YUZU_PAGEMASK) <=
-                     (Core::Memory::YUZU_PAGESIZE - sizeof(AuxInfo::AuxInfoDsp));
+    memory.ReadBlockUnsafe(return_info_, &return_info, sizeof(AuxInfo::AuxInfoDsp));
 
-    if (host_safe) [[likely]] {
-        return_ptr = memory.GetPointer<AuxInfo::AuxInfoDsp>(return_info_);
-    } else {
-        memory.ReadBlockUnsafe(return_info_, return_ptr, sizeof(AuxInfo::AuxInfoDsp));
-    }
-
-    u32 target_read_offset{return_ptr->read_offset + read_offset};
+    u32 target_read_offset{return_info.read_offset + read_offset};
     if (target_read_offset > count_max) {
         return 0;
     }
@@ -193,15 +158,9 @@ static u32 ReadAuxBufferDsp(Core::Memory::Memory& memory, CpuAddr return_info_,
     u32 write_pos{0};
     while (read_count > 0) {
         u32 to_read{std::min(count_max - target_read_offset, read_count)};
-        const auto read_addr = return_buffer + target_read_offset * sizeof(s32);
-        bool read_safe{(read_addr & Core::Memory::YUZU_PAGEMASK) <=
-                       (Core::Memory::YUZU_PAGESIZE - (read_addr + to_read * sizeof(s32)))};
-        if (read_safe) [[likely]] {
-            auto ptr = memory.GetPointer(read_addr);
-            std::memcpy(&output[write_pos], ptr, to_read * sizeof(s32));
-        } else {
-            memory.ReadBlockUnsafe(return_buffer + target_read_offset * sizeof(s32),
-                                   &output[write_pos], to_read * sizeof(s32));
+        if (to_read > 0) {
+            const auto read_addr = return_buffer + target_read_offset * sizeof(s32);
+            memory.ReadBlockUnsafe(read_addr, &output[write_pos], to_read * sizeof(s32));
         }
         target_read_offset = (target_read_offset + to_read) % count_max;
         read_count -= to_read;
@@ -209,13 +168,10 @@ static u32 ReadAuxBufferDsp(Core::Memory::Memory& memory, CpuAddr return_info_,
     }
 
     if (update_count) {
-        return_ptr->read_offset = (return_ptr->read_offset + update_count) % count_max;
-    }
-
-    if (!host_safe) [[unlikely]] {
-        memory.WriteBlockUnsafe(return_info_, return_ptr, sizeof(AuxInfo::AuxInfoDsp));
+        return_info.read_offset = (return_info.read_offset + update_count) % count_max;
     }
 
+    memory.WriteBlockUnsafe(return_info_, &return_info, sizeof(AuxInfo::AuxInfoDsp));
     return read_count_;
 }
 

src/common/page_table.cpp

@@ -66,6 +66,7 @@ void PageTable::Resize(std::size_t address_space_width_in_bits, std::size_t page
                                              << (address_space_width_in_bits - page_size_in_bits)};
     pointers.resize(num_page_table_entries);
     backing_addr.resize(num_page_table_entries);
+    blocks.resize(num_page_table_entries);
     current_address_space_width_in_bits = address_space_width_in_bits;
     page_size = 1ULL << page_size_in_bits;
 }

src/common/page_table.h

@@ -122,6 +122,7 @@ struct PageTable {
      * corresponding attribute element is of type `Memory`.
      */
     VirtualBuffer<PageInfo> pointers;
+    VirtualBuffer<u64> blocks;
 
     VirtualBuffer<u64> backing_addr;
 

src/common/scratch_buffer.h

@@ -40,8 +40,21 @@ public:
     ~ScratchBuffer() = default;
     ScratchBuffer(const ScratchBuffer&) = delete;
     ScratchBuffer& operator=(const ScratchBuffer&) = delete;
-    ScratchBuffer(ScratchBuffer&&) = default;
-    ScratchBuffer& operator=(ScratchBuffer&&) = default;
+
+    ScratchBuffer(ScratchBuffer&& other) noexcept {
+        swap(other);
+        other.last_requested_size = 0;
+        other.buffer_capacity = 0;
+        other.buffer.reset();
+    }
+
+    ScratchBuffer& operator=(ScratchBuffer&& other) noexcept {
+        swap(other);
+        other.last_requested_size = 0;
+        other.buffer_capacity = 0;
+        other.buffer.reset();
+        return *this;
+    }
 
     /// This will only grow the buffer's capacity if size is greater than the current capacity.
     /// The previously held data will remain intact.

src/common/settings.cpp

@@ -27,8 +27,8 @@ std::string GetTimeZoneString() {
     std::string location_name;
     if (time_zone_index == 0) { // Auto
 #if __cpp_lib_chrono >= 201907L
-        const struct std::chrono::tzdb& time_zone_data = std::chrono::get_tzdb();
         try {
+            const struct std::chrono::tzdb& time_zone_data = std::chrono::get_tzdb();
             const std::chrono::time_zone* current_zone = time_zone_data.current_zone();
             std::string_view current_zone_name = current_zone->name();
             location_name = current_zone_name;

src/core/CMakeLists.txt

@@ -285,6 +285,7 @@ add_library(core STATIC
     hle/kernel/kernel.cpp
     hle/kernel/kernel.h
     hle/kernel/memory_types.h
+    hle/kernel/message_buffer.h
     hle/kernel/physical_core.cpp
     hle/kernel/physical_core.h
     hle/kernel/physical_memory.h

src/core/arm/arm_interface.cpp

@@ -185,7 +185,7 @@ void ARM_Interface::Run() {
         // Notify the debugger and go to sleep if a breakpoint was hit,
         // or if the thread is unable to continue for any reason.
         if (True(hr & HaltReason::InstructionBreakpoint) || True(hr & HaltReason::PrefetchAbort)) {
-            if (!True(hr & HaltReason::InstructionBreakpoint)) {
+            if (!True(hr & HaltReason::PrefetchAbort)) {
                 RewindBreakpointInstruction();
             }
             if (system.DebuggerEnabled()) {

src/core/core.cpp

@@ -543,6 +543,7 @@ struct System::Impl {
 
     ExecuteProgramCallback execute_program_callback;
     ExitCallback exit_callback;
+    GameLogCallback game_log_callback;
 
     std::array<u64, Core::Hardware::NUM_CPU_CORES> dynarmic_ticks{};
     std::array<MicroProfileToken, Core::Hardware::NUM_CPU_CORES> microprofile_cpu{};
@@ -1011,6 +1012,18 @@ void System::RegisterExitCallback(ExitCallback&& callback) {
     impl->exit_callback = std::move(callback);
 }
 
+void System::RegisterGameLogCallback(GameLogCallback&& callback) {
+    impl->game_log_callback = std::move(callback);
+}
+
+void System::AppendGameMessage(std::string& text) {
+    if (impl->game_log_callback) {
+        impl->game_log_callback(text);
+    } else {
+        LOG_CRITICAL(Core, "game_log_callback must be initialized by the frontend");
+    }
+}
+
 void System::Exit() {
     if (impl->exit_callback) {
         impl->exit_callback();

src/core/core.h

@@ -462,6 +462,18 @@ public:
      */
     void RegisterExitCallback(ExitCallback&& callback);
 
+    /// Type used for the frontend to designate a callback for System to display log output.
+    using GameLogCallback = std::function<void(std::string)>;
+
+    /**
+     * Registers a callback from the frontend for System to display log output.
+     * @param callback Callback from the frontend to display log output.
+     */
+    void RegisterGameLogCallback(GameLogCallback&& callback);
+
+    /// Inserts a new entry on the game log ouput window
+    void AppendGameMessage(std::string& text);
+
     /// Instructs the frontend to exit the application.
     void Exit();
 

src/core/core_timing.cpp

@@ -70,7 +70,7 @@ void CoreTiming::Initialize(std::function<void()>&& on_thread_init_) {
         -> std::optional<std::chrono::nanoseconds> { return std::nullopt; };
     ev_lost = CreateEvent("_lost_event", empty_timed_callback);
     if (is_multicore) {
-        timer_thread = std::make_unique<std::thread>(ThreadEntry, std::ref(*this));
+        timer_thread = std::make_unique<std::jthread>(ThreadEntry, std::ref(*this));
     }
 }
 
@@ -255,7 +255,6 @@ void CoreTiming::ThreadLoop() {
 #ifdef _WIN32
                     while (!paused && !event.IsSet() && wait_time > 0) {
                         wait_time = *next_time - GetGlobalTimeNs().count();
-
                         if (wait_time >= timer_resolution_ns) {
                             Common::Windows::SleepForOneTick();
                         } else {

src/core/core_timing.h

@@ -163,7 +163,7 @@ private:
     Common::Event pause_event{};
     std::mutex basic_lock;
     std::mutex advance_lock;
-    std::unique_ptr<std::thread> timer_thread;
+    std::unique_ptr<std::jthread> timer_thread;
     std::atomic<bool> paused{};
     std::atomic<bool> paused_set{};
     std::atomic<bool> wait_set{};

src/core/file_sys/content_archive.cpp

@@ -57,11 +57,34 @@ struct NCASectionHeaderBlock {
 };
 static_assert(sizeof(NCASectionHeaderBlock) == 0x8, "NCASectionHeaderBlock has incorrect size.");
 
+struct NCABucketInfo {
+    u64 table_offset;
+    u64 table_size;
+    std::array<u8, 0x10> table_header;
+};
+static_assert(sizeof(NCABucketInfo) == 0x20, "NCABucketInfo has incorrect size.");
+
+struct NCASparseInfo {
+    NCABucketInfo bucket;
+    u64 physical_offset;
+    u16 generation;
+    INSERT_PADDING_BYTES_NOINIT(0x6);
+};
+static_assert(sizeof(NCASparseInfo) == 0x30, "NCASparseInfo has incorrect size.");
+
+struct NCACompressionInfo {
+    NCABucketInfo bucket;
+    INSERT_PADDING_BYTES_NOINIT(0x8);
+};
+static_assert(sizeof(NCACompressionInfo) == 0x28, "NCACompressionInfo has incorrect size.");
+
 struct NCASectionRaw {
     NCASectionHeaderBlock header;
     std::array<u8, 0x138> block_data;
     std::array<u8, 0x8> section_ctr;
-    INSERT_PADDING_BYTES_NOINIT(0xB8);
+    NCASparseInfo sparse_info;
+    NCACompressionInfo compression_info;
+    INSERT_PADDING_BYTES_NOINIT(0x60);
 };
 static_assert(sizeof(NCASectionRaw) == 0x200, "NCASectionRaw has incorrect size.");
 
@@ -225,6 +248,20 @@ bool NCA::ReadSections(const std::vector<NCASectionHeader>& sections, u64 bktr_b
     for (std::size_t i = 0; i < sections.size(); ++i) {
         const auto& section = sections[i];
 
+        if (section.raw.sparse_info.bucket.table_offset != 0 &&
+            section.raw.sparse_info.bucket.table_size != 0) {
+            LOG_ERROR(Loader, "Sparse NCAs are not supported.");
+            status = Loader::ResultStatus::ErrorSparseNCA;
+            return false;
+        }
+
+        if (section.raw.compression_info.bucket.table_offset != 0 &&
+            section.raw.compression_info.bucket.table_size != 0) {
+            LOG_ERROR(Loader, "Compressed NCAs are not supported.");
+            status = Loader::ResultStatus::ErrorCompressedNCA;
+            return false;
+        }
+
         if (section.raw.header.filesystem_type == NCASectionFilesystemType::ROMFS) {
             if (!ReadRomFSSection(section, header.section_tables[i], bktr_base_ivfc_offset)) {
                 return false;

src/core/file_sys/vfs_real.cpp

@@ -283,7 +283,8 @@ std::size_t RealVfsFile::GetSize() const {
     if (size) {
         return *size;
     }
-    return FS::GetSize(path);
+    auto lk = base.RefreshReference(path, perms, *reference);
+    return reference->file ? reference->file->GetSize() : 0;
 }
 
 bool RealVfsFile::Resize(std::size_t new_size) {

src/core/hle/kernel/k_server_session.cpp

@@ -20,12 +20,132 @@
 #include "core/hle/kernel/k_thread.h"
 #include "core/hle/kernel/k_thread_queue.h"
 #include "core/hle/kernel/kernel.h"
+#include "core/hle/kernel/message_buffer.h"
 #include "core/hle/service/hle_ipc.h"
 #include "core/hle/service/ipc_helpers.h"
 #include "core/memory.h"
 
 namespace Kernel {
 
+namespace {
+
+template <bool MoveHandleAllowed>
+Result ProcessMessageSpecialData(KProcess& dst_process, KProcess& src_process, KThread& src_thread,
+                                 MessageBuffer& dst_msg, const MessageBuffer& src_msg,
+                                 MessageBuffer::SpecialHeader& src_special_header) {
+    // Copy the special header to the destination.
+    s32 offset = dst_msg.Set(src_special_header);
+
+    // Copy the process ID.
+    if (src_special_header.GetHasProcessId()) {
+        offset = dst_msg.SetProcessId(offset, src_process.GetProcessId());
+    }
+
+    // Prepare to process handles.
+    auto& dst_handle_table = dst_process.GetHandleTable();
+    auto& src_handle_table = src_process.GetHandleTable();
+    Result result = ResultSuccess;
+
+    // Process copy handles.
+    for (auto i = 0; i < src_special_header.GetCopyHandleCount(); ++i) {
+        // Get the handles.
+        const Handle src_handle = src_msg.GetHandle(offset);
+        Handle dst_handle = Svc::InvalidHandle;
+
+        // If we're in a success state, try to move the handle to the new table.
+        if (R_SUCCEEDED(result) && src_handle != Svc::InvalidHandle) {
+            KScopedAutoObject obj =
+                src_handle_table.GetObjectForIpc(src_handle, std::addressof(src_thread));
+            if (obj.IsNotNull()) {
+                Result add_result =
+                    dst_handle_table.Add(std::addressof(dst_handle), obj.GetPointerUnsafe());
+                if (R_FAILED(add_result)) {
+                    result = add_result;
+                    dst_handle = Svc::InvalidHandle;
+                }
+            } else {
+                result = ResultInvalidHandle;
+            }
+        }
+
+        // Set the handle.
+        offset = dst_msg.SetHandle(offset, dst_handle);
+    }
+
+    // Process move handles.
+    if constexpr (MoveHandleAllowed) {
+        for (auto i = 0; i < src_special_header.GetMoveHandleCount(); ++i) {
+            // Get the handles.
+            const Handle src_handle = src_msg.GetHandle(offset);
+            Handle dst_handle = Svc::InvalidHandle;
+
+            // Whether or not we've succeeded, we need to remove the handles from the source table.
+            if (src_handle != Svc::InvalidHandle) {
+                if (R_SUCCEEDED(result)) {
+                    KScopedAutoObject obj =
+                        src_handle_table.GetObjectForIpcWithoutPseudoHandle(src_handle);
+                    if (obj.IsNotNull()) {
+                        Result add_result = dst_handle_table.Add(std::addressof(dst_handle),
+                                                                 obj.GetPointerUnsafe());
+
+                        src_handle_table.Remove(src_handle);
+
+                        if (R_FAILED(add_result)) {
+                            result = add_result;
+                            dst_handle = Svc::InvalidHandle;
+                        }
+                    } else {
+                        result = ResultInvalidHandle;
+                    }
+                } else {
+                    src_handle_table.Remove(src_handle);
+                }
+            }
+
+            // Set the handle.
+            offset = dst_msg.SetHandle(offset, dst_handle);
+        }
+    }
+
+    R_RETURN(result);
+}
+
+void CleanupSpecialData(KProcess& dst_process, u32* dst_msg_ptr, size_t dst_buffer_size) {
+    // Parse the message.
+    const MessageBuffer dst_msg(dst_msg_ptr, dst_buffer_size);
+    const MessageBuffer::MessageHeader dst_header(dst_msg);
+    const MessageBuffer::SpecialHeader dst_special_header(dst_msg, dst_header);
+
+    // Check that the size is big enough.
+    if (MessageBuffer::GetMessageBufferSize(dst_header, dst_special_header) > dst_buffer_size) {
+        return;
+    }
+
+    // Set the special header.
+    int offset = dst_msg.Set(dst_special_header);
+
+    // Clear the process id, if needed.
+    if (dst_special_header.GetHasProcessId()) {
+        offset = dst_msg.SetProcessId(offset, 0);
+    }
+
+    // Clear handles, as relevant.
+    auto& dst_handle_table = dst_process.GetHandleTable();
+    for (auto i = 0;
+         i < (dst_special_header.GetCopyHandleCount() + dst_special_header.GetMoveHandleCount());
+         ++i) {
+        const Handle handle = dst_msg.GetHandle(offset);
+
+        if (handle != Svc::InvalidHandle) {
+            dst_handle_table.Remove(handle);
+        }
+
+        offset = dst_msg.SetHandle(offset, Svc::InvalidHandle);
+    }
+}
+
+} // namespace
+
 using ThreadQueueImplForKServerSessionRequest = KThreadQueue;
 
 KServerSession::KServerSession(KernelCore& kernel)
@@ -223,12 +343,27 @@ Result KServerSession::SendReply(bool is_hle) {
             // the reply has already been written in this case.
         } else {
             Core::Memory::Memory& memory{client_thread->GetOwnerProcess()->GetMemory()};
-            KThread* server_thread{GetCurrentThreadPointer(m_kernel)};
+            KThread* server_thread = GetCurrentThreadPointer(m_kernel);
+            KProcess& src_process = *client_thread->GetOwnerProcess();
+            KProcess& dst_process = *server_thread->GetOwnerProcess();
             UNIMPLEMENTED_IF(server_thread->GetOwnerProcess() != client_thread->GetOwnerProcess());
 
-            auto* src_msg_buffer = memory.GetPointer(server_thread->GetTlsAddress());
-            auto* dst_msg_buffer = memory.GetPointer(client_message);
+            auto* src_msg_buffer = memory.GetPointer<u32>(server_thread->GetTlsAddress());
+            auto* dst_msg_buffer = memory.GetPointer<u32>(client_message);
             std::memcpy(dst_msg_buffer, src_msg_buffer, client_buffer_size);
+
+            // Translate special header ad-hoc.
+            MessageBuffer src_msg(src_msg_buffer, client_buffer_size);
+            MessageBuffer::MessageHeader src_header(src_msg);
+            MessageBuffer::SpecialHeader src_special_header(src_msg, src_header);
+            if (src_header.GetHasSpecialHeader()) {
+                MessageBuffer dst_msg(dst_msg_buffer, client_buffer_size);
+                result = ProcessMessageSpecialData<true>(dst_process, src_process, *server_thread,
+                                                         dst_msg, src_msg, src_special_header);
+                if (R_FAILED(result)) {
+                    CleanupSpecialData(dst_process, dst_msg_buffer, client_buffer_size);
+                }
+            }
         }
     } else {
         result = ResultSessionClosed;
@@ -330,12 +465,28 @@ Result KServerSession::ReceiveRequest(std::shared_ptr<Service::HLERequestContext
             ->PopulateFromIncomingCommandBuffer(client_thread->GetOwnerProcess()->GetHandleTable(),
                                                 cmd_buf);
     } else {
-        KThread* server_thread{GetCurrentThreadPointer(m_kernel)};
-        UNIMPLEMENTED_IF(server_thread->GetOwnerProcess() != client_thread->GetOwnerProcess());
+        KThread* server_thread = GetCurrentThreadPointer(m_kernel);
+        KProcess& src_process = *client_thread->GetOwnerProcess();
+        KProcess& dst_process = *server_thread->GetOwnerProcess();
+        UNIMPLEMENTED_IF(client_thread->GetOwnerProcess() != server_thread->GetOwnerProcess());
 
-        auto* src_msg_buffer = memory.GetPointer(client_message);
-        auto* dst_msg_buffer = memory.GetPointer(server_thread->GetTlsAddress());
+        auto* src_msg_buffer = memory.GetPointer<u32>(client_message);
+        auto* dst_msg_buffer = memory.GetPointer<u32>(server_thread->GetTlsAddress());
         std::memcpy(dst_msg_buffer, src_msg_buffer, client_buffer_size);
+
+        // Translate special header ad-hoc.
+        // TODO: fix this mess
+        MessageBuffer src_msg(src_msg_buffer, client_buffer_size);
+        MessageBuffer::MessageHeader src_header(src_msg);
+        MessageBuffer::SpecialHeader src_special_header(src_msg, src_header);
+        if (src_header.GetHasSpecialHeader()) {
+            MessageBuffer dst_msg(dst_msg_buffer, client_buffer_size);
+            Result res = ProcessMessageSpecialData<false>(dst_process, src_process, *client_thread,
+                                                          dst_msg, src_msg, src_special_header);
+            if (R_FAILED(res)) {
+                CleanupSpecialData(dst_process, dst_msg_buffer, client_buffer_size);
+            }
+        }
     }
 
     // We succeeded.

src/core/hle/kernel/message_buffer.h

@@ -0,0 +1,612 @@
+// SPDX-FileCopyrightText: 2023 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include "common/alignment.h"
+#include "common/bit_field.h"
+#include "core/hle/kernel/k_thread.h"
+
+namespace Kernel {
+
+constexpr inline size_t MessageBufferSize = 0x100;
+
+class MessageBuffer {
+public:
+    class MessageHeader {
+    private:
+        static constexpr inline u64 NullTag = 0;
+
+    public:
+        enum class ReceiveListCountType : u32 {
+            None = 0,
+            ToMessageBuffer = 1,
+            ToSingleBuffer = 2,
+
+            CountOffset = 2,
+            CountMax = 13,
+        };
+
+    private:
+        union {
+            std::array<u32, 2> raw;
+
+            struct {
+                // Define fields for the first header word.
+                union {
+                    BitField<0, 16, u16> tag;
+                    BitField<16, 4, u32> pointer_count;
+                    BitField<20, 4, u32> send_count;
+                    BitField<24, 4, u32> receive_count;
+                    BitField<28, 4, u32> exchange_count;
+                };
+
+                // Define fields for the second header word.
+                union {
+                    BitField<0, 10, u32> raw_count;
+                    BitField<10, 4, ReceiveListCountType> receive_list_count;
+                    BitField<14, 6, u32> reserved0;
+                    BitField<20, 11, u32> receive_list_offset;
+                    BitField<31, 1, u32> has_special_header;
+                };
+            };
+        } m_header;
+
+    public:
+        constexpr MessageHeader() : m_header{} {}
+
+        constexpr MessageHeader(u16 tag, bool special, s32 ptr, s32 send, s32 recv, s32 exch,
+                                s32 raw, ReceiveListCountType recv_list)
+            : m_header{} {
+            m_header.raw[0] = 0;
+            m_header.raw[1] = 0;
+
+            m_header.tag.Assign(tag);
+            m_header.pointer_count.Assign(ptr);
+            m_header.send_count.Assign(send);
+            m_header.receive_count.Assign(recv);
+            m_header.exchange_count.Assign(exch);
+
+            m_header.raw_count.Assign(raw);
+            m_header.receive_list_count.Assign(recv_list);
+            m_header.has_special_header.Assign(special);
+        }
+
+        explicit MessageHeader(const MessageBuffer& buf) : m_header{} {
+            buf.Get(0, m_header.raw.data(), 2);
+        }
+
+        explicit MessageHeader(const u32* msg) : m_header{{msg[0], msg[1]}} {}
+
+        constexpr u16 GetTag() const {
+            return m_header.tag;
+        }
+
+        constexpr s32 GetPointerCount() const {
+            return m_header.pointer_count;
+        }
+
+        constexpr s32 GetSendCount() const {
+            return m_header.send_count;
+        }
+
+        constexpr s32 GetReceiveCount() const {
+            return m_header.receive_count;
+        }
+
+        constexpr s32 GetExchangeCount() const {
+            return m_header.exchange_count;
+        }
+
+        constexpr s32 GetMapAliasCount() const {
+            return this->GetSendCount() + this->GetReceiveCount() + this->GetExchangeCount();
+        }
+
+        constexpr s32 GetRawCount() const {
+            return m_header.raw_count;
+        }
+
+        constexpr ReceiveListCountType GetReceiveListCount() const {
+            return m_header.receive_list_count;
+        }
+
+        constexpr s32 GetReceiveListOffset() const {
+            return m_header.receive_list_offset;
+        }
+
+        constexpr bool GetHasSpecialHeader() const {
+            return m_header.has_special_header.Value() != 0;
+        }
+
+        constexpr void SetReceiveListCount(ReceiveListCountType recv_list) {
+            m_header.receive_list_count.Assign(recv_list);
+        }
+
+        constexpr const u32* GetData() const {
+            return m_header.raw.data();
+        }
+
+        static constexpr size_t GetDataSize() {
+            return sizeof(m_header);
+        }
+    };
+
+    class SpecialHeader {
+    private:
+        union {
+            std::array<u32, 1> raw;
+
+            // Define fields for the header word.
+            BitField<0, 1, u32> has_process_id;
+            BitField<1, 4, u32> copy_handle_count;
+            BitField<5, 4, u32> move_handle_count;
+        } m_header;
+        bool m_has_header;
+
+    public:
+        constexpr explicit SpecialHeader(bool pid, s32 copy, s32 move)
+            : m_header{}, m_has_header(true) {
+            m_header.has_process_id.Assign(pid);
+            m_header.copy_handle_count.Assign(copy);
+            m_header.move_handle_count.Assign(move);
+        }
+
+        constexpr explicit SpecialHeader(bool pid, s32 copy, s32 move, bool _has_header)
+            : m_header{}, m_has_header(_has_header) {
+            m_header.has_process_id.Assign(pid);
+            m_header.copy_handle_count.Assign(copy);
+            m_header.move_handle_count.Assign(move);
+        }
+
+        explicit SpecialHeader(const MessageBuffer& buf, const MessageHeader& hdr)
+            : m_header{}, m_has_header(hdr.GetHasSpecialHeader()) {
+            if (m_has_header) {
+                buf.Get(static_cast<s32>(MessageHeader::GetDataSize() / sizeof(u32)),
+                        m_header.raw.data(), sizeof(m_header) / sizeof(u32));
+            }
+        }
+
+        constexpr bool GetHasProcessId() const {
+            return m_header.has_process_id.Value() != 0;
+        }
+
+        constexpr s32 GetCopyHandleCount() const {
+            return m_header.copy_handle_count;
+        }
+
+        constexpr s32 GetMoveHandleCount() const {
+            return m_header.move_handle_count;
+        }
+
+        constexpr const u32* GetHeader() const {
+            return m_header.raw.data();
+        }
+
+        constexpr size_t GetHeaderSize() const {
+            if (m_has_header) {
+                return sizeof(m_header);
+            } else {
+                return 0;
+            }
+        }
+
+        constexpr size_t GetDataSize() const {
+            if (m_has_header) {
+                return (this->GetHasProcessId() ? sizeof(u64) : 0) +
+                       (this->GetCopyHandleCount() * sizeof(Handle)) +
+                       (this->GetMoveHandleCount() * sizeof(Handle));
+            } else {
+                return 0;
+            }
+        }
+    };
+
+    class MapAliasDescriptor {
+    public:
+        enum class Attribute : u32 {
+            Ipc = 0,
+            NonSecureIpc = 1,
+            NonDeviceIpc = 3,
+        };
+
+    private:
+        static constexpr u32 SizeLowCount = 32;
+        static constexpr u32 SizeHighCount = 4;
+        static constexpr u32 AddressLowCount = 32;
+        static constexpr u32 AddressMidCount = 4;
+
+        constexpr u32 GetAddressMid(u64 address) {
+            return static_cast<u32>(address >> AddressLowCount) & ((1U << AddressMidCount) - 1);
+        }
+
+        constexpr u32 GetAddressHigh(u64 address) {
+            return static_cast<u32>(address >> (AddressLowCount + AddressMidCount));
+        }
+
+    private:
+        union {
+            std::array<u32, 3> raw;
+
+            struct {
+                // Define fields for the first two words.
+                u32 size_low;
+                u32 address_low;
+
+                // Define fields for the packed descriptor word.
+                union {
+                    BitField<0, 2, Attribute> attributes;
+                    BitField<2, 3, u32> address_high;
+                    BitField<5, 19, u32> reserved;
+                    BitField<24, 4, u32> size_high;
+                    BitField<28, 4, u32> address_mid;
+                };
+            };
+        } m_data;
+
+    public:
+        constexpr MapAliasDescriptor() : m_data{} {}
+
+        MapAliasDescriptor(const void* buffer, size_t _size, Attribute attr = Attribute::Ipc)
+            : m_data{} {
+            const u64 address = reinterpret_cast<u64>(buffer);
+            const u64 size = static_cast<u64>(_size);
+            m_data.size_low = static_cast<u32>(size);
+            m_data.address_low = static_cast<u32>(address);
+            m_data.attributes.Assign(attr);
+            m_data.address_mid.Assign(GetAddressMid(address));
+            m_data.size_high.Assign(static_cast<u32>(size >> SizeLowCount));
+            m_data.address_high.Assign(GetAddressHigh(address));
+        }
+
+        MapAliasDescriptor(const MessageBuffer& buf, s32 index) : m_data{} {
+            buf.Get(index, m_data.raw.data(), 3);
+        }
+
+        constexpr uintptr_t GetAddress() const {
+            return (static_cast<u64>((m_data.address_high << AddressMidCount) | m_data.address_mid)
+                    << AddressLowCount) |
+                   m_data.address_low;
+        }
+
+        constexpr uintptr_t GetSize() const {
+            return (static_cast<u64>(m_data.size_high) << SizeLowCount) | m_data.size_low;
+        }
+
+        constexpr Attribute GetAttribute() const {
+            return m_data.attributes;
+        }
+
+        constexpr const u32* GetData() const {
+            return m_data.raw.data();
+        }
+
+        static constexpr size_t GetDataSize() {
+            return sizeof(m_data);
+        }
+    };
+
+    class PointerDescriptor {
+    private:
+        static constexpr u32 AddressLowCount = 32;
+        static constexpr u32 AddressMidCount = 4;
+
+        constexpr u32 GetAddressMid(u64 address) {
+            return static_cast<u32>(address >> AddressLowCount) & ((1u << AddressMidCount) - 1);
+        }
+
+        constexpr u32 GetAddressHigh(u64 address) {
+            return static_cast<u32>(address >> (AddressLowCount + AddressMidCount));
+        }
+
+    private:
+        union {
+            std::array<u32, 2> raw;
+
+            struct {
+                // Define fields for the packed descriptor word.
+                union {
+                    BitField<0, 4, u32> index;
+                    BitField<4, 2, u32> reserved0;
+                    BitField<6, 3, u32> address_high;
+                    BitField<9, 3, u32> reserved1;
+                    BitField<12, 4, u32> address_mid;
+                    BitField<16, 16, u32> size;
+                };
+
+                // Define fields for the second word.
+                u32 address_low;
+            };
+        } m_data;
+
+    public:
+        constexpr PointerDescriptor() : m_data{} {}
+
+        PointerDescriptor(const void* buffer, size_t size, s32 index) : m_data{} {
+            const u64 address = reinterpret_cast<u64>(buffer);
+
+            m_data.index.Assign(index);
+            m_data.address_high.Assign(GetAddressHigh(address));
+            m_data.address_mid.Assign(GetAddressMid(address));
+            m_data.size.Assign(static_cast<u32>(size));
+
+            m_data.address_low = static_cast<u32>(address);
+        }
+
+        PointerDescriptor(const MessageBuffer& buf, s32 index) : m_data{} {
+            buf.Get(index, m_data.raw.data(), 2);
+        }
+
+        constexpr s32 GetIndex() const {
+            return m_data.index;
+        }
+
+        constexpr uintptr_t GetAddress() const {
+            return (static_cast<u64>((m_data.address_high << AddressMidCount) | m_data.address_mid)
+                    << AddressLowCount) |
+                   m_data.address_low;
+        }
+
+        constexpr size_t GetSize() const {
+            return m_data.size;
+        }
+
+        constexpr const u32* GetData() const {
+            return m_data.raw.data();
+        }
+
+        static constexpr size_t GetDataSize() {
+            return sizeof(m_data);
+        }
+    };
+
+    class ReceiveListEntry {
+    private:
+        static constexpr u32 AddressLowCount = 32;
+
+        constexpr u32 GetAddressHigh(u64 address) {
+            return static_cast<u32>(address >> (AddressLowCount));
+        }
+
+    private:
+        union {
+            std::array<u32, 2> raw;
+
+            struct {
+                // Define fields for the first word.
+                u32 address_low;
+
+                // Define fields for the packed descriptor word.
+                union {
+                    BitField<0, 7, u32> address_high;
+                    BitField<7, 9, u32> reserved;
+                    BitField<16, 16, u32> size;
+                };
+            };
+        } m_data;
+
+    public:
+        constexpr ReceiveListEntry() : m_data{} {}
+
+        ReceiveListEntry(const void* buffer, size_t size) : m_data{} {
+            const u64 address = reinterpret_cast<u64>(buffer);
+
+            m_data.address_low = static_cast<u32>(address);
+
+            m_data.address_high.Assign(GetAddressHigh(address));
+            m_data.size.Assign(static_cast<u32>(size));
+        }
+
+        ReceiveListEntry(u32 a, u32 b) : m_data{{a, b}} {}
+
+        constexpr uintptr_t GetAddress() const {
+            return (static_cast<u64>(m_data.address_high) << AddressLowCount) | m_data.address_low;
+        }
+
+        constexpr size_t GetSize() const {
+            return m_data.size;
+        }
+
+        constexpr const u32* GetData() const {
+            return m_data.raw.data();
+        }
+
+        static constexpr size_t GetDataSize() {
+            return sizeof(m_data);
+        }
+    };
+
+private:
+    u32* m_buffer;
+    size_t m_size;
+
+public:
+    constexpr MessageBuffer(u32* b, size_t sz) : m_buffer(b), m_size(sz) {}
+    constexpr explicit MessageBuffer(u32* b) : m_buffer(b), m_size(MessageBufferSize) {}
+
+    constexpr void* GetBufferForDebug() const {
+        return m_buffer;
+    }
+
+    constexpr size_t GetBufferSize() const {
+        return m_size;
+    }
+
+    void Get(s32 index, u32* dst, size_t count) const {
+        // Ensure that this doesn't get re-ordered.
+        std::atomic_thread_fence(std::memory_order_seq_cst);
+
+        // Get the words.
+        static_assert(sizeof(*dst) == sizeof(*m_buffer));
+
+        memcpy(dst, m_buffer + index, count * sizeof(*dst));
+    }
+
+    s32 Set(s32 index, u32* src, size_t count) const {
+        // Ensure that this doesn't get re-ordered.
+        std::atomic_thread_fence(std::memory_order_seq_cst);
+
+        // Set the words.
+        memcpy(m_buffer + index, src, count * sizeof(*src));
+
+        // Ensure that this doesn't get re-ordered.
+        std::atomic_thread_fence(std::memory_order_seq_cst);
+
+        return static_cast<s32>(index + count);
+    }
+
+    template <typename T>
+    const T& GetRaw(s32 index) const {
+        return *reinterpret_cast<const T*>(m_buffer + index);
+    }
+
+    template <typename T>
+    s32 SetRaw(s32 index, const T& val) const {
+        *reinterpret_cast<const T*>(m_buffer + index) = val;
+        return index + (Common::AlignUp(sizeof(val), sizeof(*m_buffer)) / sizeof(*m_buffer));
+    }
+
+    void GetRawArray(s32 index, void* dst, size_t len) const {
+        memcpy(dst, m_buffer + index, len);
+    }
+
+    void SetRawArray(s32 index, const void* src, size_t len) const {
+        memcpy(m_buffer + index, src, len);
+    }
+
+    void SetNull() const {
+        this->Set(MessageHeader());
+    }
+
+    s32 Set(const MessageHeader& hdr) const {
+        memcpy(m_buffer, hdr.GetData(), hdr.GetDataSize());
+        return static_cast<s32>(hdr.GetDataSize() / sizeof(*m_buffer));
+    }
+
+    s32 Set(const SpecialHeader& spc) const {
+        const s32 index = static_cast<s32>(MessageHeader::GetDataSize() / sizeof(*m_buffer));
+        memcpy(m_buffer + index, spc.GetHeader(), spc.GetHeaderSize());
+        return static_cast<s32>(index + (spc.GetHeaderSize() / sizeof(*m_buffer)));
+    }
+
+    s32 SetHandle(s32 index, const Handle& hnd) const {
+        memcpy(m_buffer + index, std::addressof(hnd), sizeof(hnd));
+        return static_cast<s32>(index + (sizeof(hnd) / sizeof(*m_buffer)));
+    }
+
+    s32 SetProcessId(s32 index, const u64 pid) const {
+        memcpy(m_buffer + index, std::addressof(pid), sizeof(pid));
+        return static_cast<s32>(index + (sizeof(pid) / sizeof(*m_buffer)));
+    }
+
+    s32 Set(s32 index, const MapAliasDescriptor& desc) const {
+        memcpy(m_buffer + index, desc.GetData(), desc.GetDataSize());
+        return static_cast<s32>(index + (desc.GetDataSize() / sizeof(*m_buffer)));
+    }
+
+    s32 Set(s32 index, const PointerDescriptor& desc) const {
+        memcpy(m_buffer + index, desc.GetData(), desc.GetDataSize());
+        return static_cast<s32>(index + (desc.GetDataSize() / sizeof(*m_buffer)));
+    }
+
+    s32 Set(s32 index, const ReceiveListEntry& desc) const {
+        memcpy(m_buffer + index, desc.GetData(), desc.GetDataSize());
+        return static_cast<s32>(index + (desc.GetDataSize() / sizeof(*m_buffer)));
+    }
+
+    s32 Set(s32 index, const u32 val) const {
+        memcpy(m_buffer + index, std::addressof(val), sizeof(val));
+        return static_cast<s32>(index + (sizeof(val) / sizeof(*m_buffer)));
+    }
+
+    Result GetAsyncResult() const {
+        MessageHeader hdr(m_buffer);
+        MessageHeader null{};
+        if (memcmp(hdr.GetData(), null.GetData(), MessageHeader::GetDataSize()) != 0) [[unlikely]] {
+            R_SUCCEED();
+        }
+        return Result(m_buffer[MessageHeader::GetDataSize() / sizeof(*m_buffer)]);
+    }
+
+    void SetAsyncResult(Result res) const {
+        const s32 index = this->Set(MessageHeader());
+        const auto value = res.raw;
+        memcpy(m_buffer + index, std::addressof(value), sizeof(value));
+    }
+
+    u32 Get32(s32 index) const {
+        return m_buffer[index];
+    }
+
+    u64 Get64(s32 index) const {
+        u64 value;
+        memcpy(std::addressof(value), m_buffer + index, sizeof(value));
+        return value;
+    }
+
+    u64 GetProcessId(s32 index) const {
+        return this->Get64(index);
+    }
+
+    Handle GetHandle(s32 index) const {
+        static_assert(sizeof(Handle) == sizeof(*m_buffer));
+        return Handle(m_buffer[index]);
+    }
+
+    static constexpr s32 GetSpecialDataIndex(const MessageHeader& hdr, const SpecialHeader& spc) {
+        return static_cast<s32>((MessageHeader::GetDataSize() / sizeof(u32)) +
+                                (spc.GetHeaderSize() / sizeof(u32)));
+    }
+
+    static constexpr s32 GetPointerDescriptorIndex(const MessageHeader& hdr,
+                                                   const SpecialHeader& spc) {
+        return static_cast<s32>(GetSpecialDataIndex(hdr, spc) + (spc.GetDataSize() / sizeof(u32)));
+    }
+
+    static constexpr s32 GetMapAliasDescriptorIndex(const MessageHeader& hdr,
+                                                    const SpecialHeader& spc) {
+        return GetPointerDescriptorIndex(hdr, spc) +
+               static_cast<s32>(hdr.GetPointerCount() * PointerDescriptor::GetDataSize() /
+                                sizeof(u32));
+    }
+
+    static constexpr s32 GetRawDataIndex(const MessageHeader& hdr, const SpecialHeader& spc) {
+        return GetMapAliasDescriptorIndex(hdr, spc) +
+               static_cast<s32>(hdr.GetMapAliasCount() * MapAliasDescriptor::GetDataSize() /
+                                sizeof(u32));
+    }
+
+    static constexpr s32 GetReceiveListIndex(const MessageHeader& hdr, const SpecialHeader& spc) {
+        if (const s32 recv_list_index = hdr.GetReceiveListOffset()) {
+            return recv_list_index;
+        } else {
+            return GetRawDataIndex(hdr, spc) + hdr.GetRawCount();
+        }
+    }
+
+    static constexpr size_t GetMessageBufferSize(const MessageHeader& hdr,
+                                                 const SpecialHeader& spc) {
+        // Get the size of the plain message.
+        size_t msg_size = GetReceiveListIndex(hdr, spc) * sizeof(u32);
+
+        // Add the size of the receive list.
+        const auto count = hdr.GetReceiveListCount();
+        switch (count) {
+        case MessageHeader::ReceiveListCountType::None:
+            break;
+        case MessageHeader::ReceiveListCountType::ToMessageBuffer:
+            break;
+        case MessageHeader::ReceiveListCountType::ToSingleBuffer:
+            msg_size += ReceiveListEntry::GetDataSize();
+            break;
+        default:
+            msg_size += (static_cast<s32>(count) -
+                         static_cast<s32>(MessageHeader::ReceiveListCountType::CountOffset)) *
+                        ReceiveListEntry::GetDataSize();
+            break;
+        }
+
+        return msg_size;
+    }
+};
+
+} // namespace Kernel

src/core/hle/service/hle_ipc.cpp

@@ -329,8 +329,22 @@ std::vector<u8> HLERequestContext::ReadBufferCopy(std::size_t buffer_index) cons
 }
 
 std::span<const u8> HLERequestContext::ReadBuffer(std::size_t buffer_index) const {
-    static thread_local std::array<Common::ScratchBuffer<u8>, 2> read_buffer_a;
-    static thread_local std::array<Common::ScratchBuffer<u8>, 2> read_buffer_x;
+    static thread_local std::array read_buffer_a{
+        Core::Memory::CpuGuestMemory<u8, Core::Memory::GuestMemoryFlags::SafeRead>(memory, 0, 0),
+        Core::Memory::CpuGuestMemory<u8, Core::Memory::GuestMemoryFlags::SafeRead>(memory, 0, 0),
+    };
+    static thread_local std::array read_buffer_data_a{
+        Common::ScratchBuffer<u8>(),
+        Common::ScratchBuffer<u8>(),
+    };
+    static thread_local std::array read_buffer_x{
+        Core::Memory::CpuGuestMemory<u8, Core::Memory::GuestMemoryFlags::SafeRead>(memory, 0, 0),
+        Core::Memory::CpuGuestMemory<u8, Core::Memory::GuestMemoryFlags::SafeRead>(memory, 0, 0),
+    };
+    static thread_local std::array read_buffer_data_x{
+        Common::ScratchBuffer<u8>(),
+        Common::ScratchBuffer<u8>(),
+    };
 
     const bool is_buffer_a{BufferDescriptorA().size() > buffer_index &&
                            BufferDescriptorA()[buffer_index].Size()};
@@ -339,19 +353,17 @@ std::span<const u8> HLERequestContext::ReadBuffer(std::size_t buffer_index) cons
             BufferDescriptorA().size() > buffer_index, { return {}; },
             "BufferDescriptorA invalid buffer_index {}", buffer_index);
         auto& read_buffer = read_buffer_a[buffer_index];
-        read_buffer.resize_destructive(BufferDescriptorA()[buffer_index].Size());
-        memory.ReadBlock(BufferDescriptorA()[buffer_index].Address(), read_buffer.data(),
-                         read_buffer.size());
-        return read_buffer;
+        return read_buffer.Read(BufferDescriptorA()[buffer_index].Address(),
+                                BufferDescriptorA()[buffer_index].Size(),
+                                &read_buffer_data_a[buffer_index]);
     } else {
         ASSERT_OR_EXECUTE_MSG(
             BufferDescriptorX().size() > buffer_index, { return {}; },
             "BufferDescriptorX invalid buffer_index {}", buffer_index);
         auto& read_buffer = read_buffer_x[buffer_index];
-        read_buffer.resize_destructive(BufferDescriptorX()[buffer_index].Size());
-        memory.ReadBlock(BufferDescriptorX()[buffer_index].Address(), read_buffer.data(),
-                         read_buffer.size());
-        return read_buffer;
+        return read_buffer.Read(BufferDescriptorX()[buffer_index].Address(),
+                                BufferDescriptorX()[buffer_index].Size(),
+                                &read_buffer_data_x[buffer_index]);
     }
 }
 

src/core/hle/service/lm/lm.cpp

@@ -276,6 +276,7 @@ private:
 
         if (text_log) {
             output_log += fmt::format("Log Text: {}\n", *text_log);
+            system.AppendGameMessage(*text_log);
         }
         LOG_DEBUG(Service_LM, "LogManager {} ({}):\n{}", NameOf(entry.severity),
                   DestinationToString(destination), output_log);

src/core/loader/loader.h

@@ -79,6 +79,8 @@ enum class ResultStatus : u16 {
     ErrorBadPFSHeader,
     ErrorIncorrectPFSFileSize,
     ErrorBadNCAHeader,
+    ErrorCompressedNCA,
+    ErrorSparseNCA,
     ErrorMissingProductionKeyFile,
     ErrorMissingHeaderKey,
     ErrorIncorrectHeaderKey,

src/core/memory.cpp

@@ -266,6 +266,22 @@ struct Memory::Impl {
         ReadBlockImpl<true>(*system.ApplicationProcess(), src_addr, dest_buffer, size);
     }
 
+    const u8* GetSpan(const VAddr src_addr, const std::size_t size) const {
+        if (current_page_table->blocks[src_addr >> YUZU_PAGEBITS] ==
+            current_page_table->blocks[(src_addr + size) >> YUZU_PAGEBITS]) {
+            return GetPointerSilent(src_addr);
+        }
+        return nullptr;
+    }
+
+    u8* GetSpan(const VAddr src_addr, const std::size_t size) {
+        if (current_page_table->blocks[src_addr >> YUZU_PAGEBITS] ==
+            current_page_table->blocks[(src_addr + size) >> YUZU_PAGEBITS]) {
+            return GetPointerSilent(src_addr);
+        }
+        return nullptr;
+    }
+
     template <bool UNSAFE>
     void WriteBlockImpl(const Kernel::KProcess& process, const Common::ProcessAddress dest_addr,
                         const void* src_buffer, const std::size_t size) {
@@ -559,7 +575,7 @@ struct Memory::Impl {
             }
         }
 
-        const Common::ProcessAddress end = base + size;
+        const auto end = base + size;
         ASSERT_MSG(end <= page_table.pointers.size(), "out of range mapping at {:016X}",
                    base + page_table.pointers.size());
 
@@ -570,14 +586,18 @@ struct Memory::Impl {
             while (base != end) {
                 page_table.pointers[base].Store(nullptr, type);
                 page_table.backing_addr[base] = 0;
-
+                page_table.blocks[base] = 0;
                 base += 1;
             }
         } else {
+            auto orig_base = base;
             while (base != end) {
-                page_table.pointers[base].Store(
-                    system.DeviceMemory().GetPointer<u8>(target) - (base << YUZU_PAGEBITS), type);
-                page_table.backing_addr[base] = GetInteger(target) - (base << YUZU_PAGEBITS);
+                auto host_ptr =
+                    system.DeviceMemory().GetPointer<u8>(target) - (base << YUZU_PAGEBITS);
+                auto backing = GetInteger(target) - (base << YUZU_PAGEBITS);
+                page_table.pointers[base].Store(host_ptr, type);
+                page_table.backing_addr[base] = backing;
+                page_table.blocks[base] = orig_base << YUZU_PAGEBITS;
 
                 ASSERT_MSG(page_table.pointers[base].Pointer(),
                            "memory mapping base yield a nullptr within the table");
@@ -747,6 +767,14 @@ struct Memory::Impl {
         VAddr last_address;
     };
 
+    void InvalidateRegion(Common::ProcessAddress dest_addr, size_t size) {
+        system.GPU().InvalidateRegion(GetInteger(dest_addr), size);
+    }
+
+    void FlushRegion(Common::ProcessAddress dest_addr, size_t size) {
+        system.GPU().FlushRegion(GetInteger(dest_addr), size);
+    }
+
     Core::System& system;
     Common::PageTable* current_page_table = nullptr;
     std::array<VideoCore::RasterizerDownloadArea, Core::Hardware::NUM_CPU_CORES>
@@ -881,6 +909,14 @@ void Memory::ReadBlockUnsafe(const Common::ProcessAddress src_addr, void* dest_b
     impl->ReadBlockUnsafe(src_addr, dest_buffer, size);
 }
 
+const u8* Memory::GetSpan(const VAddr src_addr, const std::size_t size) const {
+    return impl->GetSpan(src_addr, size);
+}
+
+u8* Memory::GetSpan(const VAddr src_addr, const std::size_t size) {
+    return impl->GetSpan(src_addr, size);
+}
+
 void Memory::WriteBlock(const Common::ProcessAddress dest_addr, const void* src_buffer,
                         const std::size_t size) {
     impl->WriteBlock(dest_addr, src_buffer, size);
@@ -924,4 +960,12 @@ void Memory::MarkRegionDebug(Common::ProcessAddress vaddr, u64 size, bool debug)
     impl->MarkRegionDebug(GetInteger(vaddr), size, debug);
 }
 
+void Memory::InvalidateRegion(Common::ProcessAddress dest_addr, size_t size) {
+    impl->InvalidateRegion(dest_addr, size);
+}
+
+void Memory::FlushRegion(Common::ProcessAddress dest_addr, size_t size) {
+    impl->FlushRegion(dest_addr, size);
+}
+
 } // namespace Core::Memory

src/core/memory.h

@@ -5,8 +5,12 @@
 
 #include <cstddef>
 #include <memory>
+#include <optional>
 #include <span>
 #include <string>
+#include <vector>
+
+#include "common/scratch_buffer.h"
 #include "common/typed_address.h"
 #include "core/hle/result.h"
 
@@ -24,6 +28,10 @@ class PhysicalMemory;
 class KProcess;
 } // namespace Kernel
 
+namespace Tegra {
+class MemoryManager;
+}
+
 namespace Core::Memory {
 
 /**
@@ -343,6 +351,9 @@ public:
      */
     void ReadBlockUnsafe(Common::ProcessAddress src_addr, void* dest_buffer, std::size_t size);
 
+    const u8* GetSpan(const VAddr src_addr, const std::size_t size) const;
+    u8* GetSpan(const VAddr src_addr, const std::size_t size);
+
     /**
      * Writes a range of bytes into the current process' address space at the specified
      * virtual address.
@@ -461,6 +472,8 @@ public:
     void MarkRegionDebug(Common::ProcessAddress vaddr, u64 size, bool debug);
 
     void SetGPUDirtyManagers(std::span<Core::GPUDirtyMemoryManager> managers);
+    void InvalidateRegion(Common::ProcessAddress dest_addr, size_t size);
+    void FlushRegion(Common::ProcessAddress dest_addr, size_t size);
 
 private:
     Core::System& system;
@@ -469,4 +482,203 @@ private:
     std::unique_ptr<Impl> impl;
 };
 
+enum GuestMemoryFlags : u32 {
+    Read = 1 << 0,
+    Write = 1 << 1,
+    Safe = 1 << 2,
+    Cached = 1 << 3,
+
+    SafeRead = Read | Safe,
+    SafeWrite = Write | Safe,
+    SafeReadWrite = SafeRead | SafeWrite,
+    SafeReadCachedWrite = SafeReadWrite | Cached,
+
+    UnsafeRead = Read,
+    UnsafeWrite = Write,
+    UnsafeReadWrite = UnsafeRead | UnsafeWrite,
+    UnsafeReadCachedWrite = UnsafeReadWrite | Cached,
+};
+
+namespace {
+template <typename M, typename T, GuestMemoryFlags FLAGS>
+class GuestMemory {
+    using iterator = T*;
+    using const_iterator = const T*;
+    using value_type = T;
+    using element_type = T;
+    using iterator_category = std::contiguous_iterator_tag;
+
+public:
+    GuestMemory() = delete;
+    explicit GuestMemory(M& memory_, u64 addr_, std::size_t size_,
+                         Common::ScratchBuffer<T>* backup = nullptr)
+        : memory{memory_}, addr{addr_}, size{size_} {
+        static_assert(FLAGS & GuestMemoryFlags::Read || FLAGS & GuestMemoryFlags::Write);
+        if constexpr (FLAGS & GuestMemoryFlags::Read) {
+            Read(addr, size, backup);
+        }
+    }
+
+    ~GuestMemory() = default;
+
+    T* data() noexcept {
+        return data_span.data();
+    }
+
+    const T* data() const noexcept {
+        return data_span.data();
+    }
+
+    [[nodiscard]] T* begin() noexcept {
+        return data();
+    }
+
+    [[nodiscard]] const T* begin() const noexcept {
+        return data();
+    }
+
+    [[nodiscard]] T* end() noexcept {
+        return data() + size;
+    }
+
+    [[nodiscard]] const T* end() const noexcept {
+        return data() + size;
+    }
+
+    T& operator[](size_t index) noexcept {
+        return data_span[index];
+    }
+
+    const T& operator[](size_t index) const noexcept {
+        return data_span[index];
+    }
+
+    void SetAddressAndSize(u64 addr_, std::size_t size_) noexcept {
+        addr = addr_;
+        size = size_;
+        addr_changed = true;
+    }
+
+    std::span<T> Read(u64 addr_, std::size_t size_,
+                      Common::ScratchBuffer<T>* backup = nullptr) noexcept {
+        addr = addr_;
+        size = size_;
+        if (size == 0) {
+            is_data_copy = true;
+            return {};
+        }
+
+        if (TrySetSpan()) {
+            if constexpr (FLAGS & GuestMemoryFlags::Safe) {
+                memory.FlushRegion(addr, size * sizeof(T));
+            }
+        } else {
+            if (backup) {
+                backup->resize_destructive(size);
+                data_span = *backup;
+            } else {
+                data_copy.resize(size);
+                data_span = std::span(data_copy);
+            }
+            is_data_copy = true;
+            span_valid = true;
+            if constexpr (FLAGS & GuestMemoryFlags::Safe) {
+                memory.ReadBlock(addr, data_span.data(), size * sizeof(T));
+            } else {
+                memory.ReadBlockUnsafe(addr, data_span.data(), size * sizeof(T));
+            }
+        }
+        return data_span;
+    }
+
+    void Write(std::span<T> write_data) noexcept {
+        if constexpr (FLAGS & GuestMemoryFlags::Cached) {
+            memory.WriteBlockCached(addr, write_data.data(), size * sizeof(T));
+        } else if constexpr (FLAGS & GuestMemoryFlags::Safe) {
+            memory.WriteBlock(addr, write_data.data(), size * sizeof(T));
+        } else {
+            memory.WriteBlockUnsafe(addr, write_data.data(), size * sizeof(T));
+        }
+    }
+
+    bool TrySetSpan() noexcept {
+        if (u8* ptr = memory.GetSpan(addr, size * sizeof(T)); ptr) {
+            data_span = {reinterpret_cast<T*>(ptr), size};
+            span_valid = true;
+            return true;
+        }
+        return false;
+    }
+
+protected:
+    bool IsDataCopy() const noexcept {
+        return is_data_copy;
+    }
+
+    bool AddressChanged() const noexcept {
+        return addr_changed;
+    }
+
+    M& memory;
+    u64 addr;
+    size_t size;
+    std::span<T> data_span{};
+    std::vector<T> data_copy;
+    bool span_valid{false};
+    bool is_data_copy{false};
+    bool addr_changed{false};
+};
+
+template <typename M, typename T, GuestMemoryFlags FLAGS>
+class GuestMemoryScoped : public GuestMemory<M, T, FLAGS> {
+public:
+    GuestMemoryScoped() = delete;
+    explicit GuestMemoryScoped(M& memory_, u64 addr_, std::size_t size_,
+                               Common::ScratchBuffer<T>* backup = nullptr)
+        : GuestMemory<M, T, FLAGS>(memory_, addr_, size_, backup) {
+        if constexpr (!(FLAGS & GuestMemoryFlags::Read)) {
+            if (!this->TrySetSpan()) {
+                if (backup) {
+                    this->data_span = *backup;
+                    this->span_valid = true;
+                    this->is_data_copy = true;
+                }
+            }
+        }
+    }
+
+    ~GuestMemoryScoped() {
+        if constexpr (FLAGS & GuestMemoryFlags::Write) {
+            if (this->size == 0) [[unlikely]] {
+                return;
+            }
+
+            if (this->AddressChanged() || this->IsDataCopy()) {
+                ASSERT(this->span_valid);
+                if constexpr (FLAGS & GuestMemoryFlags::Cached) {
+                    this->memory.WriteBlockCached(this->addr, this->data_span.data(),
+                                                  this->size * sizeof(T));
+                } else if constexpr (FLAGS & GuestMemoryFlags::Safe) {
+                    this->memory.WriteBlock(this->addr, this->data_span.data(),
+                                            this->size * sizeof(T));
+                } else {
+                    this->memory.WriteBlockUnsafe(this->addr, this->data_span.data(),
+                                                  this->size * sizeof(T));
+                }
+            } else if constexpr (FLAGS & GuestMemoryFlags::Safe) {
+                this->memory.InvalidateRegion(this->addr, this->size * sizeof(T));
+            }
+        }
+    }
+};
+} // namespace
+
+template <typename T, GuestMemoryFlags FLAGS>
+using CpuGuestMemory = GuestMemory<Memory, T, FLAGS>;
+template <typename T, GuestMemoryFlags FLAGS>
+using CpuGuestMemoryScoped = GuestMemoryScoped<Memory, T, FLAGS>;
+template <typename T, GuestMemoryFlags FLAGS>
+using GpuGuestMemory = GuestMemory<Tegra::MemoryManager, T, FLAGS>;
+template <typename T, GuestMemoryFlags FLAGS>
+using GpuGuestMemoryScoped = GuestMemoryScoped<Tegra::MemoryManager, T, FLAGS>;
 } // namespace Core::Memory

src/input_common/drivers/sdl_driver.cpp

@@ -523,6 +523,8 @@ SDLDriver::SDLDriver(std::string input_engine_) : InputEngine(std::move(input_en
     }
 
     SDL_SetHint(SDL_HINT_JOYSTICK_HIDAPI_SWITCH_PLAYER_LED, "1");
+    // Share the same button mapping with non-Nintendo controllers
+    SDL_SetHint(SDL_HINT_GAMECONTROLLER_USE_BUTTON_LABELS, "0");
 
     // Disable hidapi driver for xbox. Already default on Windows, this causes conflict with native
     // driver on Linux.
@@ -800,16 +802,9 @@ ButtonMapping SDLDriver::GetButtonMappingForDevice(const Common::ParamPackage& p
 
     // This list is missing ZL/ZR since those are not considered buttons in SDL GameController.
     // We will add those afterwards
-    // This list also excludes Screenshot since there's not really a mapping for that
     ButtonBindings switch_to_sdl_button;
 
-    if (SDL_GameControllerGetType(controller) == SDL_CONTROLLER_TYPE_NINTENDO_SWITCH_PRO ||
-        SDL_GameControllerGetType(controller) == SDL_CONTROLLER_TYPE_NINTENDO_SWITCH_JOYCON_LEFT ||
-        SDL_GameControllerGetType(controller) == SDL_CONTROLLER_TYPE_NINTENDO_SWITCH_JOYCON_RIGHT) {
-        switch_to_sdl_button = GetNintendoButtonBinding(joystick);
-    } else {
-        switch_to_sdl_button = GetDefaultButtonBinding();
-    }
+    switch_to_sdl_button = GetDefaultButtonBinding(joystick);
 
     // Add the missing bindings for ZL/ZR
     static constexpr ZButtonBindings switch_to_sdl_axis{{
@@ -830,32 +825,9 @@ ButtonMapping SDLDriver::GetButtonMappingForDevice(const Common::ParamPackage& p
     return GetSingleControllerMapping(joystick, switch_to_sdl_button, switch_to_sdl_axis);
 }
 
-ButtonBindings SDLDriver::GetDefaultButtonBinding() const {
-    return {
-        std::pair{Settings::NativeButton::A, SDL_CONTROLLER_BUTTON_B},
-        {Settings::NativeButton::B, SDL_CONTROLLER_BUTTON_A},
-        {Settings::NativeButton::X, SDL_CONTROLLER_BUTTON_Y},
-        {Settings::NativeButton::Y, SDL_CONTROLLER_BUTTON_X},
-        {Settings::NativeButton::LStick, SDL_CONTROLLER_BUTTON_LEFTSTICK},
-        {Settings::NativeButton::RStick, SDL_CONTROLLER_BUTTON_RIGHTSTICK},
-        {Settings::NativeButton::L, SDL_CONTROLLER_BUTTON_LEFTSHOULDER},
-        {Settings::NativeButton::R, SDL_CONTROLLER_BUTTON_RIGHTSHOULDER},
-        {Settings::NativeButton::Plus, SDL_CONTROLLER_BUTTON_START},
-        {Settings::NativeButton::Minus, SDL_CONTROLLER_BUTTON_BACK},
-        {Settings::NativeButton::DLeft, SDL_CONTROLLER_BUTTON_DPAD_LEFT},
-        {Settings::NativeButton::DUp, SDL_CONTROLLER_BUTTON_DPAD_UP},
-        {Settings::NativeButton::DRight, SDL_CONTROLLER_BUTTON_DPAD_RIGHT},
-        {Settings::NativeButton::DDown, SDL_CONTROLLER_BUTTON_DPAD_DOWN},
-        {Settings::NativeButton::SL, SDL_CONTROLLER_BUTTON_LEFTSHOULDER},
-        {Settings::NativeButton::SR, SDL_CONTROLLER_BUTTON_RIGHTSHOULDER},
-        {Settings::NativeButton::Home, SDL_CONTROLLER_BUTTON_GUIDE},
-        {Settings::NativeButton::Screenshot, SDL_CONTROLLER_BUTTON_MISC1},
-    };
-}
-
-ButtonBindings SDLDriver::GetNintendoButtonBinding(
+ButtonBindings SDLDriver::GetDefaultButtonBinding(
     const std::shared_ptr<SDLJoystick>& joystick) const {
-    // Default SL/SR mapping for pro controllers
+    // Default SL/SR mapping for other controllers
     auto sl_button = SDL_CONTROLLER_BUTTON_LEFTSHOULDER;
     auto sr_button = SDL_CONTROLLER_BUTTON_RIGHTSHOULDER;
 
@@ -869,10 +841,10 @@ ButtonBindings SDLDriver::GetNintendoButtonBinding(
     }
 
     return {
-        std::pair{Settings::NativeButton::A, SDL_CONTROLLER_BUTTON_A},
-        {Settings::NativeButton::B, SDL_CONTROLLER_BUTTON_B},
-        {Settings::NativeButton::X, SDL_CONTROLLER_BUTTON_X},
-        {Settings::NativeButton::Y, SDL_CONTROLLER_BUTTON_Y},
+        std::pair{Settings::NativeButton::A, SDL_CONTROLLER_BUTTON_B},
+        {Settings::NativeButton::B, SDL_CONTROLLER_BUTTON_A},
+        {Settings::NativeButton::X, SDL_CONTROLLER_BUTTON_Y},
+        {Settings::NativeButton::Y, SDL_CONTROLLER_BUTTON_X},
         {Settings::NativeButton::LStick, SDL_CONTROLLER_BUTTON_LEFTSTICK},
         {Settings::NativeButton::RStick, SDL_CONTROLLER_BUTTON_RIGHTSTICK},
         {Settings::NativeButton::L, SDL_CONTROLLER_BUTTON_LEFTSHOULDER},

src/input_common/drivers/sdl_driver.h

@@ -100,11 +100,8 @@ private:
                                                     int axis_y, float offset_x,
                                                     float offset_y) const;
 
-    /// Returns the default button bindings list for generic controllers
-    ButtonBindings GetDefaultButtonBinding() const;
-
-    /// Returns the default button bindings list for nintendo controllers
-    ButtonBindings GetNintendoButtonBinding(const std::shared_ptr<SDLJoystick>& joystick) const;
+    /// Returns the default button bindings list
+    ButtonBindings GetDefaultButtonBinding(const std::shared_ptr<SDLJoystick>& joystick) const;
 
     /// Returns the button mappings from a single controller
     ButtonMapping GetSingleControllerMapping(const std::shared_ptr<SDLJoystick>& joystick,

src/input_common/helpers/joycon_driver.cpp

@@ -172,34 +172,88 @@ void JoyconDriver::InputThread(std::stop_token stop_token) {
 }
 
 void JoyconDriver::OnNewData(std::span<u8> buffer) {
-    const u8 report_mode_byte = buffer[0];
     const auto report_mode = static_cast<ReportMode>(buffer[0]);
 
-    if (report_mode_byte > 0x33 || ((1L << report_mode_byte) & 0xf000a00000000U) == 0) {
-        // ParceUSBMode();
-    }
-
-    u8 motion_samples = 0;
-    if (motion_enabled) {
-        // Packages can be a little bit inconsistent. Average the delta time to provide a smoother
-        // motion experience
+    // Packages can be a little bit inconsistent. Average the delta time to provide a smoother
+    // motion experience
+    switch (report_mode) {
+    case ReportMode::STANDARD_FULL_60HZ:
+    case ReportMode::NFC_IR_MODE_60HZ:
+    case ReportMode::SIMPLE_HID_MODE: {
         const auto now = std::chrono::steady_clock::now();
         const auto new_delta_time = static_cast<u64>(
             std::chrono::duration_cast<std::chrono::microseconds>(now - last_update).count());
         delta_time = ((delta_time * 8) + (new_delta_time * 2)) / 10;
         last_update = now;
-        motion_samples = 3;
+        joycon_poller->UpdateColor(color);
+        break;
+    }
+    default:
+        break;
     }
 
-    if (irs_enabled) {
-        motion_samples = 2;
+    const MotionStatus motion_status{
+        .is_enabled = motion_enabled,
+        .delta_time = delta_time,
+        .gyro_sensitivity = gyro_sensitivity,
+        .accelerometer_sensitivity = accelerometer_sensitivity,
+    };
+
+    // TODO: Remove this when calibration is properly loaded and not calculated
+    if (ring_connected && report_mode == ReportMode::STANDARD_FULL_60HZ) {
+        InputReportActive data{};
+        memcpy(&data, buffer.data(), sizeof(InputReportActive));
+        calibration_protocol->GetRingCalibration(ring_calibration, data.ring_input);
     }
 
-    const bool is_not_reply_or_state =
-        report_mode != ReportMode::SUBCMD_REPLY && report_mode != ReportMode::MCU_UPDATE_STATE;
+    const RingStatus ring_status{
+        .is_enabled = ring_connected,
+        .default_value = ring_calibration.default_value,
+        .max_value = ring_calibration.max_value,
+        .min_value = ring_calibration.min_value,
+    };
 
-    joycon_poller->ReadActiveMode(buffer, motion);
+    if (irs_protocol->IsEnabled()) {
+        irs_protocol->RequestImage(buffer);
+        joycon_poller->UpdateCamera(irs_protocol->GetImage(), irs_protocol->GetIrsFormat());
+    }
 
+    if (nfc_protocol->IsPolling()) {
+        if (amiibo_detected) {
+            if (!nfc_protocol->HasAmiibo()) {
+                joycon_poller->UpdateAmiibo({});
+                amiibo_detected = false;
+                return;
+            }
+        }
+
+        if (!amiibo_detected) {
+            Joycon::TagInfo tag_info;
+            const auto result = nfc_protocol->GetTagInfo(tag_info);
+            if (result == Common::Input::DriverResult::Success) {
+                joycon_poller->UpdateAmiibo(tag_info);
+                amiibo_detected = true;
+            }
+        }
+    }
+
+    switch (report_mode) {
+    case ReportMode::STANDARD_FULL_60HZ:
+        joycon_poller->ReadActiveMode(buffer, motion_status, ring_status);
+        break;
+    case ReportMode::NFC_IR_MODE_60HZ:
+        joycon_poller->ReadNfcIRMode(buffer, motion_status);
+        break;
+    case ReportMode::SIMPLE_HID_MODE:
+        joycon_poller->ReadPassiveMode(buffer);
+        break;
+    case ReportMode::SUBCMD_REPLY:
+        LOG_DEBUG(Input, "Unhandled command reply");
+        break;
+    default:
+        LOG_ERROR(Input, "Report mode not Implemented {}", report_mode);
+        break;
+    }
 }
 
 Common::Input::DriverResult JoyconDriver::SetPollingMode() {

src/input_common/helpers/joycon_protocol/poller.cpp

@@ -16,13 +16,6 @@ void JoyconPoller::SetCallbacks(const JoyconCallbacks& callbacks_) {
     callbacks = std::move(callbacks_);
 }
 
-
-void JoyconPoller::ReadButtonAndMotionInput(std::span<u8> buffer, const MotionStatus& motion_status,
-    std::size_t motion_samples, bool is_not_reply_or_state) {
-
-}
-
-
 void JoyconPoller::ReadActiveMode(std::span<u8> buffer, const MotionStatus& motion_status,
                                   const RingStatus& ring_status) {
     InputReportActive data{};

src/input_common/helpers/joycon_protocol/poller.h

@@ -24,10 +24,6 @@ public:
 
     void SetCallbacks(const JoyconCallbacks& callbacks_);
 
-    /// Handles data from active packages
-    void ReadButtonAndMotionInput(std::span<u8> buffer, const MotionStatus& motion_status,
-                                  std::size_t motion_samples, bool is_not_reply_or_state);
-
     /// Handles data from passive packages
     void ReadPassiveMode(std::span<u8> buffer);
 

src/video_core/buffer_cache/buffer_cache.h

@@ -234,9 +234,10 @@ bool BufferCache<P>::DMACopy(GPUVAddr src_address, GPUVAddr dest_address, u64 am
     if (has_new_downloads) {
         memory_tracker.MarkRegionAsGpuModified(*cpu_dest_address, amount);
     }
-    tmp_buffer.resize_destructive(amount);
-    cpu_memory.ReadBlockUnsafe(*cpu_src_address, tmp_buffer.data(), amount);
-    cpu_memory.WriteBlockUnsafe(*cpu_dest_address, tmp_buffer.data(), amount);
+
+    Core::Memory::CpuGuestMemoryScoped<u8, Core::Memory::GuestMemoryFlags::UnsafeReadWrite> tmp(
+        cpu_memory, *cpu_src_address, amount, &tmp_buffer);
+    tmp.SetAddressAndSize(*cpu_dest_address, amount);
     return true;
 }
 

src/video_core/dma_pusher.cpp

@@ -5,6 +5,7 @@
 #include "common/microprofile.h"
 #include "common/settings.h"
 #include "core/core.h"
+#include "core/memory.h"
 #include "video_core/dma_pusher.h"
 #include "video_core/engines/maxwell_3d.h"
 #include "video_core/gpu.h"
@@ -12,6 +13,8 @@
 
 namespace Tegra {
 
+constexpr u32 MacroRegistersStart = 0xE00;
+
 DmaPusher::DmaPusher(Core::System& system_, GPU& gpu_, MemoryManager& memory_manager_,
                      Control::ChannelState& channel_state_)
     : gpu{gpu_}, system{system_}, memory_manager{memory_manager_}, puller{gpu_, memory_manager_,
@@ -74,25 +77,16 @@ bool DmaPusher::Step() {
         }
 
         // Push buffer non-empty, read a word
-        command_headers.resize_destructive(command_list_header.size);
-        constexpr u32 MacroRegistersStart = 0xE00;
-        if (dma_state.method < MacroRegistersStart) {
-            if (Settings::IsGPULevelHigh()) {
-                memory_manager.ReadBlock(dma_state.dma_get, command_headers.data(),
-                                         command_list_header.size * sizeof(u32));
-            } else {
-                memory_manager.ReadBlockUnsafe(dma_state.dma_get, command_headers.data(),
-                                               command_list_header.size * sizeof(u32));
-            }
-        } else {
-            const size_t copy_size = command_list_header.size * sizeof(u32);
+        if (dma_state.method >= MacroRegistersStart) {
             if (subchannels[dma_state.subchannel]) {
-                subchannels[dma_state.subchannel]->current_dirty =
-                    memory_manager.IsMemoryDirty(dma_state.dma_get, copy_size);
+                subchannels[dma_state.subchannel]->current_dirty = memory_manager.IsMemoryDirty(
+                    dma_state.dma_get, command_list_header.size * sizeof(u32));
             }
-            memory_manager.ReadBlockUnsafe(dma_state.dma_get, command_headers.data(), copy_size);
         }
-        ProcessCommands(command_headers);
+        Core::Memory::GpuGuestMemory<Tegra::CommandHeader,
+                                     Core::Memory::GuestMemoryFlags::UnsafeRead>
+            headers(memory_manager, dma_state.dma_get, command_list_header.size, &command_headers);
+        ProcessCommands(headers);
     }
 
     return true;

src/video_core/engines/engine_upload.cpp

@@ -5,6 +5,7 @@
 
 #include "common/algorithm.h"
 #include "common/assert.h"
+#include "core/memory.h"
 #include "video_core/engines/engine_upload.h"
 #include "video_core/memory_manager.h"
 #include "video_core/rasterizer_interface.h"
@@ -46,15 +47,11 @@ void State::ProcessData(const u32* data, size_t num_data) {
 void State::ProcessData(std::span<const u8> read_buffer) {
     const GPUVAddr address{regs.dest.Address()};
     if (is_linear) {
-        if (regs.line_count == 1) {
-            rasterizer->AccelerateInlineToMemory(address, copy_size, read_buffer);
-        } else {
-            for (size_t line = 0; line < regs.line_count; ++line) {
-                const GPUVAddr dest_line = address + line * regs.dest.pitch;
-                std::span<const u8> buffer(read_buffer.data() + line * regs.line_length_in,
-                                           regs.line_length_in);
-                rasterizer->AccelerateInlineToMemory(dest_line, regs.line_length_in, buffer);
-            }
+        for (size_t line = 0; line < regs.line_count; ++line) {
+            const GPUVAddr dest_line = address + line * regs.dest.pitch;
+            std::span<const u8> buffer(read_buffer.data() + line * regs.line_length_in,
+                                       regs.line_length_in);
+            rasterizer->AccelerateInlineToMemory(dest_line, regs.line_length_in, buffer);
         }
     } else {
         u32 width = regs.dest.width;
@@ -70,13 +67,14 @@ void State::ProcessData(std::span<const u8> read_buffer) {
         const std::size_t dst_size = Tegra::Texture::CalculateSize(
             true, bytes_per_pixel, width, regs.dest.height, regs.dest.depth,
             regs.dest.BlockHeight(), regs.dest.BlockDepth());
-        tmp_buffer.resize_destructive(dst_size);
-        memory_manager.ReadBlock(address, tmp_buffer.data(), dst_size);
-        Tegra::Texture::SwizzleSubrect(tmp_buffer, read_buffer, bytes_per_pixel, width,
-                                       regs.dest.height, regs.dest.depth, x_offset, regs.dest.y,
-                                       x_elements, regs.line_count, regs.dest.BlockHeight(),
+
+        Core::Memory::GpuGuestMemoryScoped<u8, Core::Memory::GuestMemoryFlags::SafeReadCachedWrite>
+            tmp(memory_manager, address, dst_size, &tmp_buffer);
+
+        Tegra::Texture::SwizzleSubrect(tmp, read_buffer, bytes_per_pixel, width, regs.dest.height,
+                                       regs.dest.depth, x_offset, regs.dest.y, x_elements,
+                                       regs.line_count, regs.dest.BlockHeight(),
                                        regs.dest.BlockDepth(), regs.line_length_in);
-        memory_manager.WriteBlockCached(address, tmp_buffer.data(), dst_size);
     }
 }
 

src/video_core/engines/kepler_compute.cpp

@@ -84,7 +84,6 @@ Texture::TICEntry KeplerCompute::GetTICEntry(u32 tic_index) const {
 
     Texture::TICEntry tic_entry;
     memory_manager.ReadBlockUnsafe(tic_address_gpu, &tic_entry, sizeof(Texture::TICEntry));
-
     return tic_entry;
 }
 

src/video_core/engines/maxwell_3d.cpp

@@ -9,6 +9,7 @@
 #include "common/settings.h"
 #include "core/core.h"
 #include "core/core_timing.h"
+#include "core/memory.h"
 #include "video_core/dirty_flags.h"
 #include "video_core/engines/draw_manager.h"
 #include "video_core/engines/maxwell_3d.h"
@@ -679,17 +680,14 @@ void Maxwell3D::ProcessCBData(u32 value) {
 Texture::TICEntry Maxwell3D::GetTICEntry(u32 tic_index) const {
     const GPUVAddr tic_address_gpu{regs.tex_header.Address() +
                                    tic_index * sizeof(Texture::TICEntry)};
-
     Texture::TICEntry tic_entry;
     memory_manager.ReadBlockUnsafe(tic_address_gpu, &tic_entry, sizeof(Texture::TICEntry));
-
     return tic_entry;
 }
 
 Texture::TSCEntry Maxwell3D::GetTSCEntry(u32 tsc_index) const {
     const GPUVAddr tsc_address_gpu{regs.tex_sampler.Address() +
                                    tsc_index * sizeof(Texture::TSCEntry)};
-
     Texture::TSCEntry tsc_entry;
     memory_manager.ReadBlockUnsafe(tsc_address_gpu, &tsc_entry, sizeof(Texture::TSCEntry));
     return tsc_entry;

src/video_core/engines/maxwell_dma.cpp

@@ -7,6 +7,7 @@
 #include "common/microprofile.h"
 #include "common/settings.h"
 #include "core/core.h"
+#include "core/memory.h"
 #include "video_core/engines/maxwell_3d.h"
 #include "video_core/engines/maxwell_dma.h"
 #include "video_core/memory_manager.h"
@@ -130,11 +131,12 @@ void MaxwellDMA::Launch() {
                 UNIMPLEMENTED_IF(regs.offset_out % 16 != 0);
                 read_buffer.resize_destructive(16);
                 for (u32 offset = 0; offset < regs.line_length_in; offset += 16) {
-                    memory_manager.ReadBlock(
-                        convert_linear_2_blocklinear_addr(regs.offset_in + offset),
-                        read_buffer.data(), read_buffer.size());
-                    memory_manager.WriteBlockCached(regs.offset_out + offset, read_buffer.data(),
-                                                    read_buffer.size());
+                    Core::Memory::GpuGuestMemoryScoped<
+                        u8, Core::Memory::GuestMemoryFlags::SafeReadCachedWrite>
+                        tmp_write_buffer(memory_manager,
+                                         convert_linear_2_blocklinear_addr(regs.offset_in + offset),
+                                         16, &read_buffer);
+                    tmp_write_buffer.SetAddressAndSize(regs.offset_out + offset, 16);
                 }
             } else if (is_src_pitch && !is_dst_pitch) {
                 UNIMPLEMENTED_IF(regs.line_length_in % 16 != 0);
@@ -142,20 +144,19 @@ void MaxwellDMA::Launch() {
                 UNIMPLEMENTED_IF(regs.offset_out % 16 != 0);
                 read_buffer.resize_destructive(16);
                 for (u32 offset = 0; offset < regs.line_length_in; offset += 16) {
-                    memory_manager.ReadBlock(regs.offset_in + offset, read_buffer.data(),
-                                             read_buffer.size());
-                    memory_manager.WriteBlockCached(
-                        convert_linear_2_blocklinear_addr(regs.offset_out + offset),
-                        read_buffer.data(), read_buffer.size());
+                    Core::Memory::GpuGuestMemoryScoped<
+                        u8, Core::Memory::GuestMemoryFlags::SafeReadCachedWrite>
+                        tmp_write_buffer(memory_manager, regs.offset_in + offset, 16, &read_buffer);
+                    tmp_write_buffer.SetAddressAndSize(
+                        convert_linear_2_blocklinear_addr(regs.offset_out + offset), 16);
                 }
             } else {
                 if (!accelerate.BufferCopy(regs.offset_in, regs.offset_out, regs.line_length_in)) {
-                    read_buffer.resize_destructive(regs.line_length_in);
-                    memory_manager.ReadBlock(regs.offset_in, read_buffer.data(),
-                                             regs.line_length_in,
-                                             VideoCommon::CacheType::NoBufferCache);
-                    memory_manager.WriteBlockCached(regs.offset_out, read_buffer.data(),
-                                                    regs.line_length_in);
+                    Core::Memory::GpuGuestMemoryScoped<
+                        u8, Core::Memory::GuestMemoryFlags::SafeReadCachedWrite>
+                        tmp_write_buffer(memory_manager, regs.offset_in, regs.line_length_in,
+                                         &read_buffer);
+                    tmp_write_buffer.SetAddressAndSize(regs.offset_out, regs.line_length_in);
                 }
             }
         }
@@ -222,17 +223,15 @@ void MaxwellDMA::CopyBlockLinearToPitch() {
         CalculateSize(true, bytes_per_pixel, width, height, depth, block_height, block_depth);
 
     const size_t dst_size = dst_operand.pitch * regs.line_count;
-    read_buffer.resize_destructive(src_size);
-    write_buffer.resize_destructive(dst_size);
 
-    memory_manager.ReadBlock(src_operand.address, read_buffer.data(), src_size);
-    memory_manager.ReadBlock(dst_operand.address, write_buffer.data(), dst_size);
+    Core::Memory::GpuGuestMemory<u8, Core::Memory::GuestMemoryFlags::SafeRead> tmp_read_buffer(
+        memory_manager, src_operand.address, src_size, &read_buffer);
+    Core::Memory::GpuGuestMemoryScoped<u8, Core::Memory::GuestMemoryFlags::SafeReadCachedWrite>
+        tmp_write_buffer(memory_manager, dst_operand.address, dst_size, &write_buffer);
 
-    UnswizzleSubrect(write_buffer, read_buffer, bytes_per_pixel, width, height, depth, x_offset,
-                     src_params.origin.y, x_elements, regs.line_count, block_height, block_depth,
-                     dst_operand.pitch);
-
-    memory_manager.WriteBlockCached(regs.offset_out, write_buffer.data(), dst_size);
+    UnswizzleSubrect(tmp_write_buffer, tmp_read_buffer, bytes_per_pixel, width, height, depth,
+                     x_offset, src_params.origin.y, x_elements, regs.line_count, block_height,
+                     block_depth, dst_operand.pitch);
 }
 
 void MaxwellDMA::CopyPitchToBlockLinear() {
@@ -287,18 +286,17 @@ void MaxwellDMA::CopyPitchToBlockLinear() {
         CalculateSize(true, bytes_per_pixel, width, height, depth, block_height, block_depth);
     const size_t src_size = static_cast<size_t>(regs.pitch_in) * regs.line_count;
 
-    read_buffer.resize_destructive(src_size);
-    write_buffer.resize_destructive(dst_size);
+    GPUVAddr src_addr = regs.offset_in;
+    GPUVAddr dst_addr = regs.offset_out;
+    Core::Memory::GpuGuestMemory<u8, Core::Memory::GuestMemoryFlags::SafeRead> tmp_read_buffer(
+        memory_manager, src_addr, src_size, &read_buffer);
+    Core::Memory::GpuGuestMemoryScoped<u8, Core::Memory::GuestMemoryFlags::SafeReadCachedWrite>
+        tmp_write_buffer(memory_manager, dst_addr, dst_size, &write_buffer);
 
-    memory_manager.ReadBlock(regs.offset_in, read_buffer.data(), src_size);
-    memory_manager.ReadBlockUnsafe(regs.offset_out, write_buffer.data(), dst_size);
-
-    // If the input is linear and the output is tiled, swizzle the input and copy it over.
-    SwizzleSubrect(write_buffer, read_buffer, bytes_per_pixel, width, height, depth, x_offset,
-                   dst_params.origin.y, x_elements, regs.line_count, block_height, block_depth,
-                   regs.pitch_in);
-
-    memory_manager.WriteBlockCached(regs.offset_out, write_buffer.data(), dst_size);
+    //  If the input is linear and the output is tiled, swizzle the input and copy it over.
+    SwizzleSubrect(tmp_write_buffer, tmp_read_buffer, bytes_per_pixel, width, height, depth,
+                   x_offset, dst_params.origin.y, x_elements, regs.line_count, block_height,
+                   block_depth, regs.pitch_in);
 }
 
 void MaxwellDMA::CopyBlockLinearToBlockLinear() {
@@ -342,23 +340,20 @@ void MaxwellDMA::CopyBlockLinearToBlockLinear() {
     const u32 pitch = x_elements * bytes_per_pixel;
     const size_t mid_buffer_size = pitch * regs.line_count;
 
-    read_buffer.resize_destructive(src_size);
-    write_buffer.resize_destructive(dst_size);
-
     intermediate_buffer.resize_destructive(mid_buffer_size);
 
-    memory_manager.ReadBlock(regs.offset_in, read_buffer.data(), src_size);
-    memory_manager.ReadBlock(regs.offset_out, write_buffer.data(), dst_size);
+    Core::Memory::GpuGuestMemory<u8, Core::Memory::GuestMemoryFlags::SafeRead> tmp_read_buffer(
+        memory_manager, regs.offset_in, src_size, &read_buffer);
+    Core::Memory::GpuGuestMemoryScoped<u8, Core::Memory::GuestMemoryFlags::SafeReadCachedWrite>
+        tmp_write_buffer(memory_manager, regs.offset_out, dst_size, &write_buffer);
 
-    UnswizzleSubrect(intermediate_buffer, read_buffer, bytes_per_pixel, src_width, src.height,
+    UnswizzleSubrect(intermediate_buffer, tmp_read_buffer, bytes_per_pixel, src_width, src.height,
                      src.depth, src_x_offset, src.origin.y, x_elements, regs.line_count,
                      src.block_size.height, src.block_size.depth, pitch);
 
-    SwizzleSubrect(write_buffer, intermediate_buffer, bytes_per_pixel, dst_width, dst.height,
+    SwizzleSubrect(tmp_write_buffer, intermediate_buffer, bytes_per_pixel, dst_width, dst.height,
                    dst.depth, dst_x_offset, dst.origin.y, x_elements, regs.line_count,
                    dst.block_size.height, dst.block_size.depth, pitch);
-
-    memory_manager.WriteBlockCached(regs.offset_out, write_buffer.data(), dst_size);
 }
 
 void MaxwellDMA::ReleaseSemaphore() {

src/video_core/engines/sw_blitter/blitter.cpp

@@ -159,11 +159,11 @@ bool SoftwareBlitEngine::Blit(Fermi2D::Surface& src, Fermi2D::Surface& dst,
     const auto src_bytes_per_pixel = BytesPerBlock(PixelFormatFromRenderTargetFormat(src.format));
     const auto dst_bytes_per_pixel = BytesPerBlock(PixelFormatFromRenderTargetFormat(dst.format));
     const size_t src_size = get_surface_size(src, src_bytes_per_pixel);
-    impl->tmp_buffer.resize_destructive(src_size);
-    memory_manager.ReadBlock(src.Address(), impl->tmp_buffer.data(), src_size);
+
+    Core::Memory::GpuGuestMemory<u8, Core::Memory::GuestMemoryFlags::SafeRead> tmp_buffer(
+        memory_manager, src.Address(), src_size, &impl->tmp_buffer);
 
     const size_t src_copy_size = src_extent_x * src_extent_y * src_bytes_per_pixel;
-
     const size_t dst_copy_size = dst_extent_x * dst_extent_y * dst_bytes_per_pixel;
 
     impl->src_buffer.resize_destructive(src_copy_size);
@@ -200,12 +200,11 @@ bool SoftwareBlitEngine::Blit(Fermi2D::Surface& src, Fermi2D::Surface& dst,
 
     impl->dst_buffer.resize_destructive(dst_copy_size);
     if (src.linear == Fermi2D::MemoryLayout::BlockLinear) {
-        UnswizzleSubrect(impl->src_buffer, impl->tmp_buffer, src_bytes_per_pixel, src.width,
-                         src.height, src.depth, config.src_x0, config.src_y0, src_extent_x,
-                         src_extent_y, src.block_height, src.block_depth,
-                         src_extent_x * src_bytes_per_pixel);
+        UnswizzleSubrect(impl->src_buffer, tmp_buffer, src_bytes_per_pixel, src.width, src.height,
+                         src.depth, config.src_x0, config.src_y0, src_extent_x, src_extent_y,
+                         src.block_height, src.block_depth, src_extent_x * src_bytes_per_pixel);
     } else {
-        process_pitch_linear(false, impl->tmp_buffer, impl->src_buffer, src_extent_x, src_extent_y,
+        process_pitch_linear(false, tmp_buffer, impl->src_buffer, src_extent_x, src_extent_y,
                              src.pitch, config.src_x0, config.src_y0, src_bytes_per_pixel);
     }
 
@@ -221,20 +220,18 @@ bool SoftwareBlitEngine::Blit(Fermi2D::Surface& src, Fermi2D::Surface& dst,
     }
 
     const size_t dst_size = get_surface_size(dst, dst_bytes_per_pixel);
-    impl->tmp_buffer.resize_destructive(dst_size);
-    memory_manager.ReadBlock(dst.Address(), impl->tmp_buffer.data(), dst_size);
+    Core::Memory::GpuGuestMemoryScoped<u8, Core::Memory::GuestMemoryFlags::SafeReadWrite>
+        tmp_buffer2(memory_manager, dst.Address(), dst_size, &impl->tmp_buffer);
 
     if (dst.linear == Fermi2D::MemoryLayout::BlockLinear) {
-        SwizzleSubrect(impl->tmp_buffer, impl->dst_buffer, dst_bytes_per_pixel, dst.width,
-                       dst.height, dst.depth, config.dst_x0, config.dst_y0, dst_extent_x,
-                       dst_extent_y, dst.block_height, dst.block_depth,
-                       dst_extent_x * dst_bytes_per_pixel);
+        SwizzleSubrect(tmp_buffer2, impl->dst_buffer, dst_bytes_per_pixel, dst.width, dst.height,
+                       dst.depth, config.dst_x0, config.dst_y0, dst_extent_x, dst_extent_y,
+                       dst.block_height, dst.block_depth, dst_extent_x * dst_bytes_per_pixel);
     } else {
-        process_pitch_linear(true, impl->dst_buffer, impl->tmp_buffer, dst_extent_x, dst_extent_y,
+        process_pitch_linear(true, impl->dst_buffer, tmp_buffer2, dst_extent_x, dst_extent_y,
                              dst.pitch, config.dst_x0, config.dst_y0,
                              static_cast<size_t>(dst_bytes_per_pixel));
     }
-    memory_manager.WriteBlock(dst.Address(), impl->tmp_buffer.data(), dst_size);
     return true;
 }
 

src/video_core/memory_manager.cpp

@@ -10,13 +10,13 @@
 #include "core/device_memory.h"
 #include "core/hle/kernel/k_page_table.h"
 #include "core/hle/kernel/k_process.h"
-#include "core/memory.h"
 #include "video_core/invalidation_accumulator.h"
 #include "video_core/memory_manager.h"
 #include "video_core/rasterizer_interface.h"
 #include "video_core/renderer_base.h"
 
 namespace Tegra {
+using Core::Memory::GuestMemoryFlags;
 
 std::atomic<size_t> MemoryManager::unique_identifier_generator{};
 
@@ -587,13 +587,10 @@ void MemoryManager::InvalidateRegion(GPUVAddr gpu_addr, size_t size,
 
 void MemoryManager::CopyBlock(GPUVAddr gpu_dest_addr, GPUVAddr gpu_src_addr, std::size_t size,
                               VideoCommon::CacheType which) {
-    tmp_buffer.resize_destructive(size);
-    ReadBlock(gpu_src_addr, tmp_buffer.data(), size, which);
-
-    // The output block must be flushed in case it has data modified from the GPU.
-    // Fixes NPC geometry in Zombie Panic in Wonderland DX
+    Core::Memory::GpuGuestMemoryScoped<u8, GuestMemoryFlags::SafeReadWrite> data(
+        *this, gpu_src_addr, size);
+    data.SetAddressAndSize(gpu_dest_addr, size);
     FlushRegion(gpu_dest_addr, size, which);
-    WriteBlock(gpu_dest_addr, tmp_buffer.data(), size, which);
 }
 
 bool MemoryManager::IsGranularRange(GPUVAddr gpu_addr, std::size_t size) const {
@@ -758,4 +755,23 @@ void MemoryManager::FlushCaching() {
     accumulator->Clear();
 }
 
+const u8* MemoryManager::GetSpan(const GPUVAddr src_addr, const std::size_t size) const {
+    auto cpu_addr = GpuToCpuAddress(src_addr);
+    if (cpu_addr) {
+        return memory.GetSpan(*cpu_addr, size);
+    }
+    return nullptr;
+}
+
+u8* MemoryManager::GetSpan(const GPUVAddr src_addr, const std::size_t size) {
+    if (!IsContinuousRange(src_addr, size)) {
+        return nullptr;
+    }
+    auto cpu_addr = GpuToCpuAddress(src_addr);
+    if (cpu_addr) {
+        return memory.GetSpan(*cpu_addr, size);
+    }
+    return nullptr;
+}
+
 } // namespace Tegra

src/video_core/memory_manager.h

@@ -15,6 +15,7 @@
 #include "common/range_map.h"
 #include "common/scratch_buffer.h"
 #include "common/virtual_buffer.h"
+#include "core/memory.h"
 #include "video_core/cache_types.h"
 #include "video_core/pte_kind.h"
 
@@ -62,6 +63,20 @@ public:
     [[nodiscard]] u8* GetPointer(GPUVAddr addr);
     [[nodiscard]] const u8* GetPointer(GPUVAddr addr) const;
 
+    template <typename T>
+    [[nodiscard]] T* GetPointer(GPUVAddr addr) {
+        const auto address{GpuToCpuAddress(addr)};
+        if (!address) {
+            return {};
+        }
+        return memory.GetPointer(*address);
+    }
+
+    template <typename T>
+    [[nodiscard]] const T* GetPointer(GPUVAddr addr) const {
+        return GetPointer<T*>(addr);
+    }
+
     /**
      * ReadBlock and WriteBlock are full read and write operations over virtual
      * GPU Memory. It's important to use these when GPU memory may not be continuous
@@ -139,6 +154,9 @@ public:
 
     void FlushCaching();
 
+    const u8* GetSpan(const GPUVAddr src_addr, const std::size_t size) const;
+    u8* GetSpan(const GPUVAddr src_addr, const std::size_t size);
+
 private:
     template <bool is_big_pages, typename FuncMapped, typename FuncReserved, typename FuncUnmapped>
     inline void MemoryOperation(GPUVAddr gpu_src_addr, std::size_t size, FuncMapped&& func_mapped,

src/video_core/texture_cache/texture_cache.h

@@ -8,6 +8,7 @@
 
 #include "common/alignment.h"
 #include "common/settings.h"
+#include "core/memory.h"
 #include "video_core/control/channel_state.h"
 #include "video_core/dirty_flags.h"
 #include "video_core/engines/kepler_compute.h"
@@ -1026,19 +1027,19 @@ void TextureCache<P>::UploadImageContents(Image& image, StagingBuffer& staging)
         runtime.AccelerateImageUpload(image, staging, uploads);
         return;
     }
-    const size_t guest_size_bytes = image.guest_size_bytes;
-    swizzle_data_buffer.resize_destructive(guest_size_bytes);
-    gpu_memory->ReadBlockUnsafe(gpu_addr, swizzle_data_buffer.data(), guest_size_bytes);
+
+    Core::Memory::GpuGuestMemory<u8, Core::Memory::GuestMemoryFlags::UnsafeRead> swizzle_data(
+        *gpu_memory, gpu_addr, image.guest_size_bytes, &swizzle_data_buffer);
 
     if (True(image.flags & ImageFlagBits::Converted)) {
         unswizzle_data_buffer.resize_destructive(image.unswizzled_size_bytes);
-        auto copies = UnswizzleImage(*gpu_memory, gpu_addr, image.info, swizzle_data_buffer,
-                                     unswizzle_data_buffer);
+        auto copies =
+            UnswizzleImage(*gpu_memory, gpu_addr, image.info, swizzle_data, unswizzle_data_buffer);
         ConvertImage(unswizzle_data_buffer, image.info, mapped_span, copies);
         image.UploadMemory(staging, copies);
     } else {
         const auto copies =
-            UnswizzleImage(*gpu_memory, gpu_addr, image.info, swizzle_data_buffer, mapped_span);
+            UnswizzleImage(*gpu_memory, gpu_addr, image.info, swizzle_data, mapped_span);
         image.UploadMemory(staging, copies);
     }
 }
@@ -1231,11 +1232,12 @@ void TextureCache<P>::QueueAsyncDecode(Image& image, ImageId image_id) {
     decode->image_id = image_id;
     async_decodes.push_back(std::move(decode));
 
-    Common::ScratchBuffer<u8> local_unswizzle_data_buffer(image.unswizzled_size_bytes);
-    const size_t guest_size_bytes = image.guest_size_bytes;
-    swizzle_data_buffer.resize_destructive(guest_size_bytes);
-    gpu_memory->ReadBlockUnsafe(image.gpu_addr, swizzle_data_buffer.data(), guest_size_bytes);
-    auto copies = UnswizzleImage(*gpu_memory, image.gpu_addr, image.info, swizzle_data_buffer,
+    static Common::ScratchBuffer<u8> local_unswizzle_data_buffer;
+    local_unswizzle_data_buffer.resize_destructive(image.unswizzled_size_bytes);
+    Core::Memory::GpuGuestMemory<u8, Core::Memory::GuestMemoryFlags::UnsafeRead> swizzle_data(
+        *gpu_memory, image.gpu_addr, image.guest_size_bytes, &swizzle_data_buffer);
+
+    auto copies = UnswizzleImage(*gpu_memory, image.gpu_addr, image.info, swizzle_data,
                                  local_unswizzle_data_buffer);
     const size_t out_size = MapSizeBytes(image);
 

src/video_core/texture_cache/util.cpp

@@ -20,6 +20,7 @@
 #include "common/div_ceil.h"
 #include "common/scratch_buffer.h"
 #include "common/settings.h"
+#include "core/memory.h"
 #include "video_core/compatible_formats.h"
 #include "video_core/engines/maxwell_3d.h"
 #include "video_core/memory_manager.h"
@@ -544,17 +545,15 @@ void SwizzleBlockLinearImage(Tegra::MemoryManager& gpu_memory, GPUVAddr gpu_addr
                        tile_size.height, info.tile_width_spacing);
     const size_t subresource_size = sizes[level];
 
-    tmp_buffer.resize_destructive(subresource_size);
-    const std::span<u8> dst(tmp_buffer);
-
     for (s32 layer = 0; layer < info.resources.layers; ++layer) {
         const std::span<const u8> src = input.subspan(host_offset);
-        gpu_memory.ReadBlockUnsafe(gpu_addr + guest_offset, dst.data(), dst.size_bytes());
+        {
+            Core::Memory::GpuGuestMemoryScoped<u8, Core::Memory::GuestMemoryFlags::UnsafeReadWrite>
+                dst(gpu_memory, gpu_addr + guest_offset, subresource_size, &tmp_buffer);
 
-        SwizzleTexture(dst, src, bytes_per_block, num_tiles.width, num_tiles.height,
-                       num_tiles.depth, block.height, block.depth);
-
-        gpu_memory.WriteBlockUnsafe(gpu_addr + guest_offset, dst.data(), dst.size_bytes());
+            SwizzleTexture(dst, src, bytes_per_block, num_tiles.width, num_tiles.height,
+                           num_tiles.depth, block.height, block.depth);
+        }
 
         host_offset += host_bytes_per_layer;
         guest_offset += layer_stride;
@@ -837,6 +836,7 @@ boost::container::small_vector<BufferImageCopy, 16> UnswizzleImage(Tegra::Memory
     const Extent3D size = info.size;
 
     if (info.type == ImageType::Linear) {
+        ASSERT(output.size_bytes() >= guest_size_bytes);
         gpu_memory.ReadBlockUnsafe(gpu_addr, output.data(), guest_size_bytes);
 
         ASSERT((info.pitch >> bpp_log2) << bpp_log2 == info.pitch);
@@ -904,16 +904,6 @@ boost::container::small_vector<BufferImageCopy, 16> UnswizzleImage(Tegra::Memory
     return copies;
 }
 
-BufferCopy UploadBufferCopy(Tegra::MemoryManager& gpu_memory, GPUVAddr gpu_addr,
-                            const ImageBase& image, std::span<u8> output) {
-    gpu_memory.ReadBlockUnsafe(gpu_addr, output.data(), image.guest_size_bytes);
-    return BufferCopy{
-        .src_offset = 0,
-        .dst_offset = 0,
-        .size = image.guest_size_bytes,
-    };
-}
-
 void ConvertImage(std::span<const u8> input, const ImageInfo& info, std::span<u8> output,
                   std::span<BufferImageCopy> copies) {
     u32 output_offset = 0;

src/video_core/texture_cache/util.h

@@ -66,9 +66,6 @@ struct OverlapResult {
     Tegra::MemoryManager& gpu_memory, GPUVAddr gpu_addr, const ImageInfo& info,
     std::span<const u8> input, std::span<u8> output);
 
-[[nodiscard]] BufferCopy UploadBufferCopy(Tegra::MemoryManager& gpu_memory, GPUVAddr gpu_addr,
-                                          const ImageBase& image, std::span<u8> output);
-
 void ConvertImage(std::span<const u8> input, const ImageInfo& info, std::span<u8> output,
                   std::span<BufferImageCopy> copies);
 

src/yuzu/debugger/console.cpp

@@ -7,6 +7,11 @@
 #include <wincon.h>
 #endif
 
+#include <QAction>
+#include <QLayout>
+#include <QString>
+#include <QScrollBar>
+
 #include "common/logging/backend.h"
 #include "yuzu/debugger/console.h"
 #include "yuzu/uisettings.h"
@@ -46,4 +51,66 @@ void ToggleConsole() {
     SetColorConsoleBackendEnabled(UISettings::values.show_console.GetValue());
 #endif
 }
+
+
+GameLogOuput::GameLogOuput(QWidget* parent)
+    : QWidget(parent, Qt::Dialog) {
+    setObjectName(QStringLiteral("Game Log Output"));
+    setWindowTitle(tr("Game Log Output"));
+    resize(500, 350);
+    setMinimumSize(500, 350);
+    // Remove the "?" button from the titlebar and enable the maximize button
+    setWindowFlags((windowFlags() & ~Qt::WindowContextHelpButtonHint) |
+                   Qt::WindowMaximizeButtonHint);
+
+    log_output = new QPlainTextEdit(this);
+    log_output->setReadOnly(true);
+    QLayout* layout = new QVBoxLayout(this);
+    layout->setContentsMargins(0, 0, 0, 0);
+    layout->addWidget(log_output);
+    setLayout(layout);
+}
+
+void GameLogOuput::AppendMessage(const QString& text) {
+    log_queue.Push(text);
+    // Scrolls to the bottom
+    //log_output->verticalScrollBar()->setValue(
+    //    log_output->verticalScrollBar()->maximum());
+}
+
+
+void GameLogOuput::paintEvent(QPaintEvent* event) {
+    QWidget::paintEvent(event);
+    while (!log_queue.Empty()) {
+        QString text{};
+        log_queue.Pop(text);
+        log_output->appendPlainText(text);
+    }
+}
+
+QAction* GameLogOuput::toggleViewAction() {
+    if (toggle_view_action == nullptr) {
+        toggle_view_action = new QAction(tr("&Game Log Ouput"), this);
+        toggle_view_action->setCheckable(true);
+        toggle_view_action->setChecked(isVisible());
+        connect(toggle_view_action, &QAction::toggled, this, &GameLogOuput::setVisible);
+    }
+
+    return toggle_view_action;
+}
+
+void GameLogOuput::showEvent(QShowEvent* ev) {
+    if (toggle_view_action) {
+        toggle_view_action->setChecked(isVisible());
+    }
+    QWidget::showEvent(ev);
+}
+
+void GameLogOuput::hideEvent(QHideEvent* ev) {
+    if (toggle_view_action) {
+        toggle_view_action->setChecked(isVisible());
+    }
+    QWidget::hideEvent(ev);
+}
+
 } // namespace Debugger

src/yuzu/debugger/console.h

@@ -3,6 +3,16 @@
 
 #pragma once
 
+#include <mutex>
+#include <QWidget>
+#include <QPlainTextEdit>
+
+#include "common/threadsafe_queue.h"
+
+namespace Core::HID {
+class HIDCore;
+} // namespace Core::HID
+
 namespace Debugger {
 
 /**
@@ -10,4 +20,27 @@ namespace Debugger {
  * get a real qt logging window which would work for all platforms.
  */
 void ToggleConsole();
+
+class GameLogOuput : public QWidget {
+    Q_OBJECT
+
+public:
+    explicit GameLogOuput(QWidget* parent = nullptr);
+
+    void AppendMessage(const QString& text);
+
+    /// Returns a QAction that can be used to toggle visibility of this dialog.
+    QAction* toggleViewAction();
+
+protected:
+    void showEvent(QShowEvent* ev) override;
+    void hideEvent(QHideEvent* ev) override;
+    void paintEvent(QPaintEvent* event) override;
+
+private:
+    QAction* toggle_view_action = nullptr;
+    QPlainTextEdit* log_output = nullptr;
+    /// Queue of vibration request to controllers
+    Common::SPSCQueue<QString> log_queue;
+};
 } // namespace Debugger

src/yuzu/main.cpp

@@ -178,6 +178,8 @@ constexpr int default_mouse_hide_timeout = 2500;
 constexpr int default_mouse_center_timeout = 10;
 constexpr int default_input_update_timeout = 1;
 
+constexpr size_t CopyBufferSize = 1_MiB;
+
 /**
  * "Callouts" are one-time instructional messages shown to the user. In the config settings, there
  * is a bitfield "callout_flags" options, used to track if a message has already been shown to the
@@ -1246,6 +1248,10 @@ void GMainWindow::InitializeDebugWidgets() {
     controller_dialog->hide();
     debug_menu->addAction(controller_dialog->toggleViewAction());
 
+    game_log_ouput_widget = new Debugger::GameLogOuput(this);
+    game_log_ouput_widget->hide();
+    debug_menu->addAction(game_log_ouput_widget->toggleViewAction());
+
     connect(this, &GMainWindow::EmulationStarting, waitTreeWidget,
             &WaitTreeWidget::OnEmulationStarting);
     connect(this, &GMainWindow::EmulationStopping, waitTreeWidget,
@@ -1874,6 +1880,11 @@ void GMainWindow::BootGame(const QString& filename, u64 program_id, std::size_t
         render_window->Exit();
     });
 
+    // Register an Exit callback such that Core can exit the currently running application.
+    system->RegisterGameLogCallback([this](std::string text) {
+        game_log_ouput_widget->AppendMessage(QString::fromStdString(text));
+    });
+
     connect(render_window, &GRenderWindow::Closed, this, &GMainWindow::OnStopGame);
     connect(render_window, &GRenderWindow::MouseActivity, this, &GMainWindow::OnMouseActivity);
     // BlockingQueuedConnection is important here, it makes sure we've finished refreshing our views
@@ -2929,10 +2940,10 @@ void GMainWindow::OnMenuInstallToNAND() {
 
     int remaining = filenames.size();
 
-    // This would only overflow above 2^43 bytes (8.796 TB)
+    // This would only overflow above 2^51 bytes (2.252 PB)
     int total_size = 0;
     for (const QString& file : files) {
-        total_size += static_cast<int>(QFile(file).size() / 0x1000);
+        total_size += static_cast<int>(QFile(file).size() / CopyBufferSize);
     }
     if (total_size < 0) {
         LOG_CRITICAL(Frontend, "Attempting to install too many files, aborting.");
@@ -3032,7 +3043,7 @@ InstallResult GMainWindow::InstallNSPXCI(const QString& filename) {
             return false;
         }
 
-        std::vector<u8> buffer(1_MiB);
+        std::vector<u8> buffer(CopyBufferSize);
 
         for (std::size_t i = 0; i < src->GetSize(); i += buffer.size()) {
             if (install_progress->wasCanceled()) {
@@ -3088,7 +3099,7 @@ InstallResult GMainWindow::InstallNCA(const QString& filename) {
             return false;
         }
 
-        std::array<u8, 0x1000> buffer{};
+        std::vector<u8> buffer(CopyBufferSize);
 
         for (std::size_t i = 0; i < src->GetSize(); i += buffer.size()) {
             if (install_progress->wasCanceled()) {

src/yuzu/main.h

@@ -72,6 +72,10 @@ struct KeyboardInitializeParameters;
 struct ProfileSelectParameters;
 } // namespace Core::Frontend
 
+namespace Debugger {
+class GameLogOuput;
+}
+
 namespace DiscordRPC {
 class DiscordInterface;
 }
@@ -468,6 +472,7 @@ private:
     MicroProfileDialog* microProfileDialog;
     WaitTreeWidget* waitTreeWidget;
     ControllerDialog* controller_dialog;
+    Debugger::GameLogOuput* game_log_ouput_widget;
 
     QAction* actions_recent_files[max_recent_files_item];