[opus] proper limit and error of opus decoder instances

Reduces the page entries from 32 bytes to 8 and rewrites `VirtualBuffer` to be more efficient in memory usage and specifically for large zero regions.

The page table will now only reserve 1GiB instead of 4GiB and of this memory it should only use at most ~8MiB.

This PR has the side effect of using Eden on Windows on low memory systems much more plausible since it would previously require ~10GiB of committable memory at front (despite using ~5-6 at most, inadvertently stalling other processes) where as now it should only require around the amount that it'll actually use.

Co-authored-by: Lizzie <lizzie@eden-emu.dev>
Reviewed-on: https://git.eden-emu.dev/eden-emu/eden/pulls/4219
Reviewed-by: Maufeat <sahyno1996@gmail.com>
Reviewed-by: lizzie <lizzie@eden-emu.dev>
Reviewed-by: CamilleLaVey <camillelavey99@gmail.com>

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

@@ -3,7 +3,7 @@
 
     <!-- General application strings -->
     <string name="app_name" translatable="false">Eden</string>
-    <string name="app_disclaimer">This software will run games for the Nintendo Switch game console. No game titles or keys are included.&lt;br /&gt;&lt;br /&gt;Before you begin, please locate your <![CDATA[<b> prod.keys </b>]]> file on your device storage.&lt;br /&gt;&lt;br /&gt;<![CDATA[<a href=\"https://yuzu-mirror.github.io/help/quickstart\">Learn more</a>]]></string>
+    <string name="app_disclaimer">This software will run games for the Nintendo Switch game console. No game titles or keys are included.&lt;br /&gt;&lt;br /&gt;Before you begin, please locate your <![CDATA[<b> prod.keys </b>]]> file on your device storage.&lt;br /&gt;&lt;br /&gt;<![CDATA[<a href=\"https://git.eden-emu.dev/eden-emu/eden/src/branch/master/docs/user/QuickStart.md\">Learn more</a>]]></string>
     <string name="notice_notification_channel_name">Notices and errors</string>
     <string name="notice_notification_channel_id" translatable="false">noticesAndErrors</string>
     <string name="notice_notification_channel_description">Shows notifications when something goes wrong.</string>
@@ -933,7 +933,7 @@
     <string name="loader_error_file_not_found">ROM file does not exist</string>
 
     <string name="loader_requires_firmware">Game Requires Firmware</string>
-    <string name="loader_requires_firmware_description"><![CDATA[The game you are trying to launch requires firmware to boot or to get past the opening menu. Please <a href=\"https://yuzu-mirror.github.io/help/quickstart\"> dump and install firmware</a>, or press \"OK\" to launch anyways.]]></string>
+    <string name="loader_requires_firmware_description"><![CDATA[This game may require firmware to function properly, and you don\'t have any installed. Please install firmware before launching, or press \"OK\" to launch anyways.]]></string>
 
     <!-- Intent Launch strings -->
     <string name="searching_for_game">Searching for game...</string>

src/audio_core/opus/decoder.cpp

@@ -8,6 +8,7 @@
 #include "audio_core/opus/hardware_opus.h"
 #include "audio_core/opus/parameters.h"
 #include "common/alignment.h"
+#include "common/scope_exit.h"
 #include "common/swap.h"
 #include "core/core.h"
 
@@ -28,10 +29,18 @@ OpusDecoder::OpusDecoder(Core::System& system_, HardwareOpus& hardware_opus_)
 OpusDecoder::~OpusDecoder() {
     if (decode_object_initialized) {
         hardware_opus.ShutdownDecodeObject(shared_buffer.data(), shared_buffer.size());
+        hardware_opus.UnregisterDecoder(this);
     }
 }
 
 Result OpusDecoder::Initialize(const OpusParametersEx& params, Kernel::KTransferMemory* transfer_memory, u64 transfer_memory_size) {
+    R_TRY(hardware_opus.RegisterDecoder(this));
+    SCOPE_EXIT {
+        if (!decode_object_initialized) {
+            hardware_opus.UnregisterDecoder(this);
+        }
+    };
+
     auto frame_size{params.use_large_frame_size ? 5760 : 1920};
     shared_buffer.resize(transfer_memory_size);
     shared_memory_mapped = true;
@@ -61,6 +70,13 @@ Result OpusDecoder::Initialize(const OpusParametersEx& params, Kernel::KTransfer
 }
 
 Result OpusDecoder::Initialize(const OpusMultiStreamParametersEx& params, Kernel::KTransferMemory* transfer_memory, u64 transfer_memory_size) {
+    R_TRY(hardware_opus.RegisterDecoder(this));
+    SCOPE_EXIT {
+        if (!decode_object_initialized) {
+            hardware_opus.UnregisterDecoder(this);
+        }
+    };
+
     auto frame_size{params.use_large_frame_size ? 5760 : 1920};
     shared_buffer.resize(transfer_memory_size, 0);
     shared_memory_mapped = true;

src/audio_core/opus/hardware_opus.cpp

@@ -4,6 +4,7 @@
 // SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
+#include <algorithm>
 #include <array>
 
 #include "audio_core/audio_core.h"
@@ -45,6 +46,25 @@ HardwareOpus::HardwareOpus(Core::System& system_)
     opus_decoder.SetSharedMemory(shared_memory);
 }
 
+Result HardwareOpus::RegisterDecoder(OpusDecoder* decoder) {
+    std::scoped_lock l{mutex};
+    const auto slot = std::ranges::find(decoders, nullptr);
+    if (slot == decoders.end()) {
+        R_THROW(ResultOutOfOpusDecoders);
+    }
+    *slot = decoder;
+    R_SUCCEED();
+}
+
+void HardwareOpus::UnregisterDecoder(OpusDecoder* decoder) {
+    std::scoped_lock l{mutex};
+    const auto slot = std::ranges::find(decoders, decoder);
+    if (slot == decoders.end()) {
+        return;
+    }
+    *slot = nullptr;
+}
+
 u32 HardwareOpus::GetWorkBufferSize(u32 channel) {
     if (!opus_decoder.IsRunning()) {
         return 0;

src/audio_core/opus/hardware_opus.h

@@ -3,6 +3,7 @@
 
 #pragma once
 
+#include <array>
 #include <mutex>
 #include <opus.h>
 
@@ -12,9 +13,12 @@
 #include "core/hle/service/audio/errors.h"
 
 namespace AudioCore::OpusDecoder {
+class OpusDecoder;
 class HardwareOpus {
 public:
     HardwareOpus(Core::System& system);
+    Result RegisterDecoder(OpusDecoder* decoder);
+    void UnregisterDecoder(OpusDecoder* decoder);
 
     u32 GetWorkBufferSize(u32 channel);
     u32 GetWorkBufferSizeForMultiStream(u32 total_stream_count, u32 stereo_stream_count);
@@ -39,6 +43,7 @@ public:
 private:
     Core::System& system;
     std::mutex mutex;
+    std::array<OpusDecoder*, 24> decoders{};
     ADSP::OpusDecoder::OpusDecoder& opus_decoder;
     ADSP::OpusDecoder::SharedMemory shared_memory;
 };

src/common/CMakeLists.txt

@@ -89,7 +89,6 @@ add_library(
     param_package.h
     parent_of_member.h
     point.h
-    quaternion.h
     range_map.h
     range_mutex.h
     range_sets.h
@@ -109,6 +108,8 @@ add_library(
     settings_setting.h
     slot_vector.h
     socket_types.h
+    sparse_large_vector.cpp
+    sparse_large_vector.h
     spin_lock.h
     stb.cpp
     stb.h
@@ -138,8 +139,6 @@ add_library(
     uuid.cpp
     uuid.h
     vector_math.h
-    virtual_buffer.cpp
-    virtual_buffer.h
     zstd_compression.cpp
     zstd_compression.h
     fs/ryujinx_compat.h fs/ryujinx_compat.cpp

src/common/fiber.cpp

@@ -9,7 +9,6 @@
 
 #include "common/assert.h"
 #include "common/fiber.h"
-#include "common/virtual_buffer.h"
 
 #include <boost/context/detail/fcontext.hpp>
 

src/common/host_memory.cpp

@@ -117,7 +117,6 @@ template <typename T>
 static void GetFuncAddress(Common::DynamicLibrary& dll, const char* name, T& pfn) {
     if (!dll.GetSymbol(name, &pfn)) {
         LOG_CRITICAL(HW_Memory, "Failed to load {}", name);
-        throw std::bad_alloc{};
     }
 }
 
@@ -179,6 +178,14 @@ public:
         Release();
     }
 
+    void* Allocate(size_t size) {
+        auto* ptr = VirtualAlloc(nullptr, size, MEM_RESERVE | MEM_COMMIT, PAGE_READWRITE);
+        if (ptr == nullptr) {
+            LOG_CRITICAL(HW_Memory, "Failed to allocate fallback buffer with size {:#x}, error {}", size, GetLastError());
+        }
+        return ptr;
+    }
+
     void Map(size_t virtual_offset, size_t host_offset, size_t length, MemoryPermission perms) {
         std::unique_lock lock{placeholder_mutex};
         if (!IsNiechePlaceholder(virtual_offset, length)) {
@@ -399,6 +406,10 @@ private:
 // For managarm: see https://github.com/managarm/managarm/issues/1370
 #else // ^^^ Windows ^^^ vvv POSIX vvv
 
+#ifndef MAP_NOCORE
+#define MAP_NOCORE 0
+#endif
+
 #ifdef ARCHITECTURE_arm64
 
 static void* ChooseVirtualBase(size_t virtual_size) {
@@ -423,7 +434,7 @@ static void* ChooseVirtualBase(size_t virtual_size) {
         // Note: we may be able to take advantage of MAP_FIXED_NOREPLACE here.
         void* map_pointer =
             mmap(reinterpret_cast<void*>(hint_address), virtual_size, PROT_READ | PROT_WRITE,
-                 MAP_PRIVATE | MAP_ANONYMOUS | MAP_NORESERVE, -1, 0);
+                 MAP_PRIVATE | MAP_ANONYMOUS | MAP_NORESERVE | MAP_NOCORE, -1, 0);
 
         // If we successfully mapped, we're done.
         if (reinterpret_cast<uintptr_t>(map_pointer) == hint_address) {
@@ -443,11 +454,11 @@ static void* ChooseVirtualBase(size_t virtual_size) {
 
 static void* ChooseVirtualBase(size_t virtual_size) {
 #if defined(__FreeBSD__) || defined(__DragonFly__) || defined(__OpenBSD__) || defined(__sun__) || defined(__HAIKU__) || defined(__managarm__) || defined(__AIX__)
-    void* virtual_base = mmap(nullptr, virtual_size, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS | MAP_NORESERVE | MAP_ALIGNED_SUPER, -1, 0);
+    void* virtual_base = mmap(nullptr, virtual_size, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS | MAP_NORESERVE | MAP_ALIGNED_SUPER | MAP_NOCORE, -1, 0);
     if (virtual_base != MAP_FAILED)
         return virtual_base;
 #endif
-    return mmap(nullptr, virtual_size, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS | MAP_NORESERVE, -1, 0);
+    return mmap(nullptr, virtual_size, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS | MAP_NORESERVE | MAP_NOCORE, -1, 0);
 }
 
 #endif
@@ -541,13 +552,13 @@ public:
         }
         if (use_anon) {
             LOG_WARNING(Common_Memory, "Using private mappings instead of shared ones");
-            backing_base = static_cast<u8*>(mmap(nullptr, backing_size, PROT_READ | PROT_WRITE, MAP_ANONYMOUS | MAP_PRIVATE, -1, 0));
+            backing_base = static_cast<u8*>(mmap(nullptr, backing_size, PROT_READ | PROT_WRITE, MAP_ANONYMOUS | MAP_PRIVATE | MAP_NOCORE, -1, 0));
             if (fd > 0) {
                 fd = -1;
                 close(fd);
             }
         } else {
-            backing_base = static_cast<u8*>(mmap(nullptr, backing_size, PROT_READ | PROT_WRITE, MAP_SHARED, fd, 0));
+            backing_base = static_cast<u8*>(mmap(nullptr, backing_size, PROT_READ | PROT_WRITE, MAP_SHARED | MAP_NOCORE, fd, 0));
         }
         if (backing_base == MAP_FAILED) {
             LOG_CRITICAL(HW_Memory, "mmap failed: {}", strerror(errno));
@@ -571,6 +582,14 @@ public:
         Release();
     }
 
+    void* Allocate(size_t size) {
+        auto* ptr = mmap(nullptr, size, PROT_READ | PROT_WRITE, MAP_ANONYMOUS | MAP_PRIVATE, -1, 0);
+        if (ptr == MAP_FAILED) {
+            LOG_CRITICAL(HW_Memory, "Failed to allocate fallback buffer with size {:#x}, {}", size, strerror(errno));
+        }
+        return ptr;
+    }
+
     void Map(size_t virtual_offset, size_t host_offset, size_t length, MemoryPermission perms) {
         // Intersect the range with our address space.
         AdjustMap(&virtual_offset, &length);
@@ -691,12 +710,10 @@ HostMemory::HostMemory(size_t backing_size_, size_t virtual_size_)
 {
 #if defined(__OPENORBIS__) || defined(__managarm__)
     LOG_WARNING(HW_Memory, "Platform doesn't support fastmem");
-    fallback_buffer.emplace(backing_size);
-    backing_base = fallback_buffer->data();
+    backing_base = static_cast<u8*>(mmap(nullptr, backing_size, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS, -1, 0));
     virtual_base = nullptr;
 #else
     // Try to allocate a fastmem arena.
-    // The implementation will fail with std::bad_alloc on errors.
     impl = std::make_unique<HostMemory::Impl>(AlignUp(backing_size, PageAlignment), AlignUp(virtual_size, PageAlignment) + HugePageSize);
     if (impl->Init()) {
         backing_base = impl->backing_base;
@@ -707,16 +724,26 @@ HostMemory::HostMemory(size_t backing_size_, size_t virtual_size_)
             virtual_base_offset = virtual_base - impl->virtual_base;
         }
     } else {
-        impl.reset();
         LOG_WARNING(HW_Memory, "Platform can support fastmem, but can't create it");
-        fallback_buffer.emplace(backing_size);
-        backing_base = fallback_buffer->data();
+        fallback_buffer = true;
+        backing_base = static_cast<u8*>(impl->Allocate(backing_size));
         virtual_base = nullptr;
+        impl.reset();
     }
 #endif
 }
 
-HostMemory::~HostMemory() = default;
+HostMemory::~HostMemory() {
+#ifdef _WIN32
+    if (fallback_buffer) {
+        VirtualFree(backing_base, backing_size, MEM_RELEASE);
+    }
+#else
+    if (fallback_buffer) {
+        munmap(backing_base, backing_size);
+    }
+#endif
+}
 
 HostMemory::HostMemory(HostMemory&&) noexcept = default;
 

src/common/host_memory.h

@@ -10,7 +10,6 @@
 #include <optional>
 #include "common/common_funcs.h"
 #include "common/common_types.h"
-#include "common/virtual_buffer.h"
 
 namespace Common {
 
@@ -86,7 +85,7 @@ private:
     u8* virtual_base{};
     size_t virtual_base_offset{};
     // Windows requires it for kernels whom lack proper support for some functions!
-    std::optional<Common::VirtualBuffer<u8>> fallback_buffer;
+    bool fallback_buffer{false};
 };
 
 } // namespace Common

src/common/page_table.cpp

@@ -1,4 +1,4 @@
-// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
 // SPDX-License-Identifier: GPL-3.0-or-later
 
 // SPDX-FileCopyrightText: Copyright 2019 yuzu Emulator Project
@@ -13,39 +13,11 @@ PageTable::PageTable() = default;
 
 PageTable::~PageTable() noexcept = default;
 
-bool PageTable::BeginTraversal(TraversalEntry* out_entry, TraversalContext* out_context,
-                               Common::ProcessAddress address) const {
-    out_context->next_offset = GetInteger(address);
-    out_context->next_page = address / page_size;
-
-    return this->ContinueTraversal(out_entry, out_context);
-}
-
-bool PageTable::ContinueTraversal(TraversalEntry* out_entry, TraversalContext* context) const {
-    // Setup invalid defaults.
-    out_entry->phys_addr = 0;
-    out_entry->block_size = page_size;
-    // Validate that we can read the actual entry.
-    if (auto const page = context->next_page; page < entries.size()) {
-        // Validate that the entry is mapped.
-        if (auto const paddr = entries[page].addr; paddr != 0) {
-            // Populate the results.
-            out_entry->phys_addr = paddr + context->next_offset;
-            context->next_page += 1;
-            context->next_offset += page_size;
-            return true;
-        }
-    }
-    context->next_page += 1;
-    context->next_offset += page_size;
-    return false;
-}
-
-void PageTable::Resize(std::size_t address_space_width_in_bits, std::size_t page_size_in_bits) {
-    auto const num_page_table_entries = 1ULL << (address_space_width_in_bits - page_size_in_bits);
-    entries.resize(num_page_table_entries);
+void PageTable::Resize(std::size_t address_space_width_in_bits, std::size_t page_bits) {
+    auto const num_page_table_entries = 1ULL << (address_space_width_in_bits - page_bits);
+    entries.ResizeAndClear(num_page_table_entries);
     current_address_space_width_in_bits = address_space_width_in_bits;
-    page_size = 1ULL << page_size_in_bits;
+    current_page_bits = page_bits;
 }
 
 } // namespace Common

src/common/page_table.h

@@ -9,22 +9,22 @@
 #include <atomic>
 
 #include "common/common_types.h"
+#include "common/sparse_large_vector.h"
 #include "common/typed_address.h"
-#include "common/virtual_buffer.h"
 
 namespace Common {
 
 enum class PageType : u8 {
     /// Page is unmapped and should cause an access error.
-    Unmapped,
+    Unmapped = 0b00,
     /// Page is mapped to regular memory. This is the only type you can get pointers to.
-    Memory,
+    Memory = 0b01,
     /// Page is mapped to regular memory, but inaccessible from CPU fastmem and must use
     /// the callbacks.
-    DebugMemory,
+    DebugMemory = 0b10,
     /// Page is mapped to regular memory, but also needs to check for rasterizer cache flushing and
     /// invalidation
-    RasterizerCachedMemory,
+    RasterizerCachedMemory = 0b11,
 };
 
 /**
@@ -42,57 +42,86 @@ struct PageTable {
         u64 next_offset{};
     };
 
-    /// Number of bits reserved for attribute tagging.
-    /// This can be at most the guaranteed alignment of the pointers in the page table.
-    static constexpr int ATTRIBUTE_BITS = 2;
+    /// Masks out bits reserved for attribute tagging.
+    static constexpr u64 ATTRIBUTE_MASK = ((1ULL << 44) - 1) << 12;
+
+    /// Specifies sign bit for page table entries.
+    static constexpr u64 SIGN_BIT = 45 + 12; // 44 bits of data + page offset
 
     /**
-     * Pair of host pointer and page type attribute.
-     * This uses the lower bits of a given pointer to store the attribute tag.
+     * Atomic tuple of host pointer, page type, and block id.
+     * This uses the lower bits of a given pointer to store the attributes.
      * Writing and reading the pointer attribute pair is guaranteed to be atomic for the same method
      * call. In other words, they are guaranteed to be synchronized at all times.
      */
-    class PageInfo {
+    class PageEntryData {
     public:
+        struct Data {
+            Data(bool marked_, PageType type_, u16 block_, u64 page_)
+                : marked(static_cast<u64>(marked_)       & 0b1)
+                , type(static_cast<u64>(type_)           & ((1ULL << 2) - 1))
+                , block(static_cast<u64>(block_)         & ((1ULL << 9) - 1))
+                , page((page_ >> 12)                     & ((1ULL << 45) - 1))
+                , block2((static_cast<u64>(block_) >> 9) & ((1ULL << 7) - 1)) {}
+            u64 marked : 1;
+            u64 type   : 2;
+            u64 block  : 9;
+            u64 page   : 45; // 44 bits of actual data (64 - page offset (12) - reserved (8)) + a sign bit
+            u64 block2 : 7;
+        };
+
+        [[nodiscard]] Data Raw() const noexcept {
+            return std::bit_cast<Data>(data_raw.load(std::memory_order_relaxed));
+        }
+
         /// Returns the page pointer
-        [[nodiscard]] uintptr_t Pointer() const noexcept {
-            return ExtractPointer(raw.load(std::memory_order_relaxed));
+        [[nodiscard]] uintptr_t Pointer(bool ignored_marked = false) const noexcept {
+            return ExtractPointer(std::bit_cast<Data>(data_raw.load(std::memory_order_relaxed)), ignored_marked);
         }
 
         /// Returns the page type attribute
         [[nodiscard]] PageType Type() const noexcept {
-            return ExtractType(raw.load(std::memory_order_relaxed));
+            return static_cast<PageType>(std::bit_cast<Data>(data_raw.load(std::memory_order_relaxed)).type);
+        }
+
+        /// Returns the block identifier.
+        [[nodiscard]] u16 Block() const noexcept {
+            return ExtractBlock(std::bit_cast<Data>(data_raw.load(std::memory_order_relaxed)));
         }
 
         /// Returns the page pointer and attribute pair, extracted from the same atomic read
-        [[nodiscard]] std::pair<uintptr_t, PageType> PointerType() const noexcept {
-            const uintptr_t non_atomic_raw = raw.load(std::memory_order_relaxed);
-            return {ExtractPointer(non_atomic_raw), ExtractType(non_atomic_raw)};
+        [[nodiscard]] std::tuple<uintptr_t, PageType, u16> PointerTypeBlock(bool ignore_marked = false) const noexcept {
+            const auto non_atomic_raw = std::bit_cast<Data>(data_raw.load(std::memory_order_relaxed));
+            return {ExtractPointer(non_atomic_raw, ignore_marked), static_cast<PageType>(non_atomic_raw.type), ExtractBlock(non_atomic_raw)};
         }
 
-        /// Returns the raw representation of the page information.
-        /// Use ExtractPointer and ExtractType to unpack the value.
-        [[nodiscard]] uintptr_t Raw() const noexcept {
-            return raw.load(std::memory_order_relaxed);
+        /// Write page info atomically
+        constexpr void Store(bool marked, PageType type, u16 block, uintptr_t pointer) noexcept {
+            data_raw.store(std::bit_cast<u64>(Data{marked, type, block, pointer}));
         }
 
-        /// Write a page pointer and type pair atomically
-        void Store(uintptr_t pointer, PageType type) noexcept {
-            raw.store(pointer | uintptr_t(type));
+        constexpr void MarkRasterizerCached() noexcept {
+            data_raw.fetch_or(0b111);
+        }
+
+        constexpr void MarkDebug(u64 ptr, u16 block) noexcept {
+            Store(true, PageType::DebugMemory, block, ptr);
         }
 
         /// Unpack a pointer from a page info raw representation
-        [[nodiscard]] static uintptr_t ExtractPointer(uintptr_t raw) noexcept {
-            return raw & (~uintptr_t{0} << ATTRIBUTE_BITS);
+        [[nodiscard]] static uintptr_t ExtractPointer(Data raw, bool ignore_marked = false) noexcept {
+            return raw.marked && !ignore_marked ? 0
+                // shift raw.page's fake sign bit to the actual sign bit, then sign extend
+                : ((s64)(raw.page << (64 - 44))) >> (64 - 44 - 12);
         }
 
-        /// Unpack a page type from a page info raw representation
-        [[nodiscard]] static PageType ExtractType(uintptr_t raw) noexcept {
-            return static_cast<PageType>(raw & ((uintptr_t{1} << ATTRIBUTE_BITS) - 1));
+        [[nodiscard]] static u16 ExtractBlock(Data raw) noexcept {
+            return static_cast<u16>(raw.block | (raw.block2 << 9));
         }
 
     private:
-        std::atomic<uintptr_t> raw;
+        std::atomic<u64> data_raw;
+        static_assert(sizeof(Data) == sizeof(std::atomic<u64>));
     };
 
     PageTable();
@@ -100,13 +129,8 @@ struct PageTable {
 
     PageTable(const PageTable&) = delete;
     PageTable& operator=(const PageTable&) = delete;
-
-    PageTable(PageTable&&) noexcept = default;
-    PageTable& operator=(PageTable&&) noexcept = default;
-
-    bool BeginTraversal(TraversalEntry* out_entry, TraversalContext* out_context,
-                        Common::ProcessAddress address) const;
-    bool ContinueTraversal(TraversalEntry* out_entry, TraversalContext* context) const;
+    PageTable(PageTable&&) noexcept = delete;
+    PageTable& operator=(PageTable&&) noexcept = delete;
 
     /**
      * Resizes the page table to be able to accommodate enough pages within
@@ -121,30 +145,14 @@ struct PageTable {
         return current_address_space_width_in_bits;
     }
 
-    bool GetPhysicalAddress(Common::PhysicalAddress* out_phys_addr,
-                            Common::ProcessAddress virt_addr) const {
-        if (virt_addr > (1ULL << this->GetAddressSpaceBits())) {
-            return false;
-        }
-
-        *out_phys_addr = entries[virt_addr / page_size].addr + GetInteger(virt_addr);
-        return true;
-    }
-
     /// Vector of memory pointers backing each page. An entry can only be non-null if the
     /// corresponding attribute element is of type `Memory`.
-    struct PageEntryData {
-        PageInfo ptr;
-        u64 block;
-        u64 addr;
-        u64 padding;
-    };
-    VirtualBuffer<PageEntryData> entries;
-    static_assert(sizeof(PageEntryData) == 32);
+    SparseLargeVector<PageEntryData> entries;
+    static_assert(sizeof(PageEntryData) == 8);
 
     u8* fastmem_arena{};
     std::size_t current_address_space_width_in_bits{};
-    std::size_t page_size{};
+    std::size_t current_page_bits{};
 };
 
 } // namespace Common

src/common/quaternion.h

@@ -1,79 +0,0 @@
-// SPDX-FileCopyrightText: 2016 Citra Emulator Project
-// SPDX-License-Identifier: GPL-2.0-or-later
-
-#pragma once
-
-#include "common/vector_math.h"
-
-namespace Common {
-
-template <typename T>
-class Quaternion {
-public:
-    Vec3<T> xyz;
-    T w{};
-
-    [[nodiscard]] Quaternion<decltype(-T{})> Inverse() const {
-        return {-xyz, w};
-    }
-
-    [[nodiscard]] Quaternion<decltype(T{} + T{})> operator+(const Quaternion& other) const {
-        return {xyz + other.xyz, w + other.w};
-    }
-
-    [[nodiscard]] Quaternion<decltype(T{} - T{})> operator-(const Quaternion& other) const {
-        return {xyz - other.xyz, w - other.w};
-    }
-
-    [[nodiscard]] Quaternion<decltype(T{} * T{} - T{} * T{})> operator*(
-        const Quaternion& other) const {
-        return {xyz * other.w + other.xyz * w + Cross(xyz, other.xyz),
-                w * other.w - Dot(xyz, other.xyz)};
-    }
-
-    [[nodiscard]] Quaternion<T> Normalized() const {
-        T length = std::sqrt(xyz.Length2() + w * w);
-        return {xyz / length, w / length};
-    }
-
-    [[nodiscard]] std::array<decltype(-T{}), 16> ToMatrix() const {
-        const T x2 = xyz[0] * xyz[0];
-        const T y2 = xyz[1] * xyz[1];
-        const T z2 = xyz[2] * xyz[2];
-
-        const T xy = xyz[0] * xyz[1];
-        const T wz = w * xyz[2];
-        const T xz = xyz[0] * xyz[2];
-        const T wy = w * xyz[1];
-        const T yz = xyz[1] * xyz[2];
-        const T wx = w * xyz[0];
-
-        return {1.0f - 2.0f * (y2 + z2),
-                2.0f * (xy + wz),
-                2.0f * (xz - wy),
-                0.0f,
-                2.0f * (xy - wz),
-                1.0f - 2.0f * (x2 + z2),
-                2.0f * (yz + wx),
-                0.0f,
-                2.0f * (xz + wy),
-                2.0f * (yz - wx),
-                1.0f - 2.0f * (x2 + y2),
-                0.0f,
-                0.0f,
-                0.0f,
-                0.0f,
-                1.0f};
-    }
-};
-
-template <typename T>
-[[nodiscard]] auto QuaternionRotate(const Quaternion<T>& q, const Vec3<T>& v) {
-    return v + 2 * Cross(q.xyz, Cross(q.xyz, v) + v * q.w);
-}
-
-[[nodiscard]] inline Quaternion<float> MakeQuaternion(const Vec3<float>& axis, float angle) {
-    return {axis * std::sin(angle / 2), std::cos(angle / 2)};
-}
-
-} // namespace Common

src/common/settings.h

@@ -858,7 +858,7 @@ struct Values {
     SwitchableSetting<std::string> program_args{linkage,
                                                 std::string(),
                                                 "program_args",
-                                                Category::System,
+                                                Category::Debugging,
                                                 Specialization::Default,
                                                 true,    // save_ - persist in config file
                                                 false};  // runtime_modifiable_ - startup-only
@@ -904,7 +904,7 @@ struct Values {
                                            0,
                                            65535,
                                            "debug_knobs",
-                                           Category::System,
+                                           Category::Debugging,
                                            Specialization::Countable,
                                            true,
                                            true};

src/common/sparse_large_vector.cpp

@@ -0,0 +1,143 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
+/* virtual_buffer.cpp */
+// SPDX-FileCopyrightText: Copyright 2020 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#ifdef _WIN32
+#include <windows.h>
+#include <mutex>
+#else
+#include <sys/mman.h>
+#endif
+
+#include "common/alignment.h"
+#include "common/assert.h"
+#include "common/sparse_large_vector.h"
+
+namespace Common {
+
+#ifdef _WIN32
+static std::vector<std::pair<u64, u64>> vector_regions {};
+
+// Workaround for handling non-commited memory accessed by Dynarmic; usually result of an error
+static LONG WINAPI FakePageFaultHandler(PEXCEPTION_POINTERS info) {
+    DWORD code = info->ExceptionRecord->ExceptionCode;
+    u64 exception_addr = reinterpret_cast<u64>(info->ExceptionRecord->ExceptionAddress);
+
+    if (code != EXCEPTION_ACCESS_VIOLATION) {
+        // Not our problem
+        return EXCEPTION_CONTINUE_SEARCH;
+    }
+
+    u64 addr = 0, addr2 = 0;
+
+    for (auto region: vector_regions) {
+        auto addr_shifted = exception_addr >> HostPageBits;
+        if (region.first <= addr_shifted && addr_shifted <= region.second) {
+            addr = addr_shifted;
+        }
+
+        // Page-boundary accesses
+        if (auto addr_ = (exception_addr + 0x40) >> HostPageBits; addr_ != addr_shifted && region.first <= addr_ && addr_ <= region.second) {
+            addr2 = addr_;
+        }
+
+        if (addr != 0 || addr2 != 0) {
+            break;
+        }
+    }
+
+    if (addr == 0 && addr2 == 0) {
+        // Not our problem
+        return EXCEPTION_CONTINUE_SEARCH;
+    }
+
+    LOG_ERROR(HW_Memory, "Accessing an unallocated region of a SparseLargeVector at {:#x}; this shouldn't happen and is likely a Dynarmic error!", exception_addr);
+
+    // Commit this region
+    if (addr != 0) {
+        if (!CommitVectorPage(addr << HostPageBits, false)) {
+            return EXCEPTION_CONTINUE_SEARCH;
+        }
+    }
+    // Commit next region if needed
+    if (addr2 != 0) {
+        if (!CommitVectorPage(addr2 << HostPageBits, false)) {
+            return EXCEPTION_CONTINUE_SEARCH;
+        }
+    }
+
+    return EXCEPTION_CONTINUE_EXECUTION;
+}
+
+bool CommitVectorPage(uintptr_t addr, bool write) noexcept {
+    MEMORY_BASIC_INFORMATION info {};
+    auto res = VirtualQuery(reinterpret_cast<void*>(addr), &info, sizeof(info));
+    if (res == 0) {
+        LOG_CRITICAL(HW_Memory, "Failed to query large buffer region at {:#x} with error {}, will try committing anyway", addr, GetLastError());
+    } else if (info.State != MEM_RESERVE) {
+        LOG_ERROR(HW_Memory, "Tried to commit an unreserved large buffer region at {:#x} that is not mapped or is already committed (state {:#x})", addr, info.State);
+        return false;
+    }
+
+    auto perm = write ? PAGE_READWRITE : PAGE_READONLY;
+    void* res2 = VirtualAlloc(reinterpret_cast<LPVOID>(addr), HostPageSize, MEM_COMMIT, perm);
+    if (res2 == nullptr) {
+        LOG_ERROR(HW_Memory, "Failed to commit large buffer region at {:#x}, error {}", addr, GetLastError());
+        return false;
+    }
+
+    return true;
+}
+#endif
+
+#ifndef MAP_NOCORE
+#define MAP_NOCORE 0
+#endif
+
+void* AllocateMemoryPages(std::size_t size) noexcept {
+    if (auto page = HostPageSize; size % page != 0) {
+        LOG_WARNING(HW_Memory, "Allocating unaligned large vector with size {:#x}; aligning to {} page size", size, page);
+        size = AlignUp(size, page);
+    }
+
+#ifdef _WIN32
+    // We will never use this memory entirely so instead of committing it up front let's just reserve it and commit each page individually
+    void* base = VirtualAlloc(nullptr, size, MEM_RESERVE, PAGE_READWRITE);
+
+    if (base != nullptr) {
+        vector_regions.emplace_back(reinterpret_cast<u64>(base), reinterpret_cast<u64>(base) + size);
+
+        static std::once_flag flag;
+        std::call_once(flag, []() { AddVectoredExceptionHandler(1, FakePageFaultHandler); });
+    } else {
+        // Try committing everything instead??
+        LOG_WARNING(HW_Memory, "Failed to reserve large vector region with error {}, trying to commit instead..", GetLastError());
+        base = VirtualAlloc(nullptr, size, MEM_COMMIT, PAGE_READWRITE);
+    }
+    ASSERT_MSG(base, "Failed to reserve {:#x} sized region with error {}", size, GetLastError());
+#else
+    void* base = mmap(nullptr, size, PROT_READ, MAP_ANON | MAP_PRIVATE | MAP_NOCORE, -1, 0);
+    if (base == MAP_FAILED)
+        base = nullptr;
+    ASSERT_MSG(base, "Failed to allocate {:#x} sized region with error {}", size, strerror(errno));
+#endif
+    return base;
+}
+
+void FreeMemoryPages(void* base, [[maybe_unused]] std::size_t size) noexcept {
+    if (auto page = HostPageSize; size % page != 0) {
+        size = AlignUp(size, page);
+    }
+    if (!base)
+        return;
+#ifdef _WIN32
+    ASSERT(VirtualFree(base, 0, MEM_RELEASE));
+#else
+    ASSERT(munmap(base, size) == 0);
+#endif
+}
+
+} // namespace Common

src/common/sparse_large_vector.h

@@ -0,0 +1,194 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
+/* virtual_buffer.h */
+// SPDX-FileCopyrightText: Copyright 2020 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <atomic>
+#include <bit>
+#include <utility>
+#include <vector>
+
+#ifndef _WIN32
+#include <unistd.h>
+#include <sys/mman.h>
+#endif
+
+#include "common/alignment.h"
+#include "common/assert.h"
+
+namespace Common {
+
+#ifdef _WIN32
+constexpr u64 HostPageSize = 0x1000;
+constexpr u64 HostPageBits = 12;
+constexpr u64 HostPageMask = ~(HostPageSize - 1);
+bool CommitVectorPage(uintptr_t addr, bool write) noexcept;
+#else
+const u64 HostPageSize = sysconf(_SC_PAGESIZE);
+const u64 HostPageBits = std::countr_zero(HostPageSize);
+const u64 HostPageMask = ~(HostPageSize - 1);
+#endif
+
+void* AllocateMemoryPages(std::size_t size) noexcept;
+void FreeMemoryPages(void* base, std::size_t size) noexcept;
+
+/// A large page-aligned buffer that has optimized memory usage for zero-writes.
+template <typename T>
+requires std::is_trivially_copyable_v<T>
+class SparseLargeVector final {
+public:
+    constexpr SparseLargeVector() = default;
+
+    explicit SparseLargeVector(std::size_t count) noexcept
+        : alloc_size{count * sizeof(T)}
+    {
+        base_ptr = static_cast<T*>(AllocateMemoryPages(alloc_size));
+
+        // each item in vector holds information for 64 pages
+        auto denom = HostPageSize * 64;
+        committed_pages = std::vector<std::atomic<u64>>((alloc_size + denom - 1) / denom);
+    }
+
+    ~SparseLargeVector() noexcept {
+        FreeMemoryPages(base_ptr, alloc_size);
+    }
+
+    SparseLargeVector(const SparseLargeVector&) = delete;
+    SparseLargeVector& operator=(const SparseLargeVector&) = delete;
+    SparseLargeVector(SparseLargeVector&& other) = delete;
+    SparseLargeVector& operator=(SparseLargeVector&& other) = delete;
+
+    void ResizeAndClear(std::size_t count) noexcept {
+        if (auto const new_size = count * sizeof(T); new_size != alloc_size) {
+            FreeMemoryPages(base_ptr, alloc_size);
+            alloc_size = new_size;
+            base_ptr = static_cast<T*>(AllocateMemoryPages(alloc_size));
+
+            auto denom = HostPageSize * 64;
+            committed_pages = std::vector<std::atomic<u64>>((alloc_size + denom - 1) / denom);
+        }
+    }
+
+    /// Returns a reference to the value of the requested index and allocates memory if needed.
+    T& GetAndFault(std::size_t index) noexcept {
+        if (index > alloc_size / sizeof(T)) {
+            UNREACHABLE_MSG("Out of bounds RW access on SparseLargeVector @ {}", index);
+        }
+
+        if (!IsCommittedPage(index)) {
+            CommitPage(index);
+        }
+        return base_ptr[index];
+    }
+
+    /// Returns a reference to the value of the requested index if initialized, or will otherwise return a zero-initialized object.
+    const T& GetOrDefault(std::size_t index) const {
+#ifdef _WIN32
+        if (!IsCommittedPage(index)) {
+            return *reinterpret_cast<const T*>(&default_val);
+        }
+#endif
+        // On non-Windows, OS page table should optimize this by pointing to a zero page if unallocated.
+        return base_ptr[index];
+    }
+
+    void Set(std::size_t index, const T& value) noexcept {
+        if (index > alloc_size / sizeof(T)) {
+            LOG_CRITICAL(Common_Memory, "Out of bounds write on SparseLargeVector @ {}", index);
+            return;
+        }
+        if (!IsCommittedPage(index))
+            CommitPage(index);
+        base_ptr[index] = value;
+    }
+
+    void ZeroRegion(std::size_t start, std::size_t end_) noexcept {
+        u64 base = reinterpret_cast<u64>(&base_ptr[start]);
+        const u64 end = reinterpret_cast<u64>(&base_ptr[end_]);
+
+        const u64 end_page = AlignUp(base, HostPageSize);
+        const u64 first_size = (std::min)(end_page, end) - base;
+
+        if (IsCommittedPage(start / sizeof(T))) {
+            std::memset(reinterpret_cast<void*>(base), 0, first_size);
+        }
+
+        if (end <= end_page)
+            return;
+
+        base = end_page;
+
+        for (u64 page = base; page < end; page += HostPageSize) {
+            if (!IsCommittedPage((page - reinterpret_cast<u64>(base_ptr)) / sizeof(T))) {
+                continue;
+            }
+
+            std::memset(reinterpret_cast<void*>(page), 0, (std::min)( HostPageSize, end - page));
+        }
+    }
+
+    constexpr void CommitRegion(size_t index, size_t end_) {
+        const u64 base = static_cast<u64>(index) * sizeof(T);
+        const u64 end = static_cast<u64>(end_) * sizeof(T);
+
+        for (u64 page = AlignDown(base, HostPageSize); page < end; page += HostPageSize) {
+            if (!IsCommittedPage(page / sizeof(T))) {
+                CommitPage(page / sizeof(T));
+            }
+        }
+    }
+
+    constexpr T& GetUnchecked(size_t index) {
+        return base_ptr[index];
+    }
+
+    [[nodiscard]] constexpr const T& operator[](std::size_t index) const noexcept {
+        return GetOrDefault(index);
+    }
+
+    [[nodiscard]] constexpr const T* data() const noexcept {
+        return base_ptr;
+    }
+
+    [[nodiscard]] constexpr std::size_t size() const noexcept {
+        return alloc_size / sizeof(T);
+    }
+
+private:
+    [[nodiscard]] constexpr bool IsCommittedPage(std::size_t index) const noexcept {
+        if (index > alloc_size / sizeof(T)) {
+            LOG_CRITICAL(Common_Memory, "Out of bounds access on large vector @ {}", index);
+            return false;
+        }
+
+        auto page = (index * sizeof(T)) >> HostPageBits;
+        auto val = committed_pages[page >> 6].load(std::memory_order_acquire);
+        return (val >> (page & 63)) & 1;
+    }
+
+    constexpr void CommitPage(std::size_t index) noexcept {
+        auto page_index = (index * sizeof(T)) >> HostPageBits;
+        auto page = reinterpret_cast<uintptr_t>(base_ptr + index) & HostPageMask;
+#if defined(_WIN32)
+        CommitVectorPage(page, true);
+#else
+        mprotect(reinterpret_cast<void*>(page), HostPageSize, PROT_READ | PROT_WRITE);
+#endif
+
+        committed_pages[page_index >> 6].fetch_or(1ULL << (page_index & 63), std::memory_order_release);
+    }
+
+    std::size_t alloc_size{};
+    T* base_ptr{};
+
+    std::vector<std::atomic<u64>> committed_pages{};
+#ifdef _WIN32
+    const std::array<u8, sizeof(T)> default_val{};
+#endif
+};
+
+} // namespace Common

src/common/vector_math.h

@@ -1,4 +1,4 @@
-// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
 // SPDX-License-Identifier: GPL-3.0-or-later
 
 // SPDX-FileCopyrightText: 2014 Tony Wasserka
@@ -7,752 +7,128 @@
 
 #pragma once
 
-#ifdef __ARM_NEON
-#include <arm_neon.h>
-#endif
-
 #include <cmath>
 #include <type_traits>
 
 namespace Common {
 
-template <typename T>
-class Vec2;
-template <typename T>
-class Vec3;
-template <typename T>
-class Vec4;
-
-template <typename T>
-class Vec2 {
+template <typename T, size_t N>
+class Vec {
 public:
-    T x{};
-    T y{};
+    std::array<T, N> elems{};
 
-    constexpr Vec2() = default;
-    constexpr Vec2(const T& x_, const T& y_) : x(x_), y(y_) {}
+    constexpr Vec() = default;
+    constexpr Vec(T e0) noexcept : elems{e0} {}
+    constexpr Vec(T e0, T e1) noexcept : elems{e0, e1} {}
+    constexpr Vec(T e0, T e1, T e2) noexcept : elems{e0, e1, e2} {}
+    constexpr Vec(T e0, T e1, T e2, T e4) noexcept : elems{e0, e1, e2, e4} {}
+    //explicit constexpr Vec(const std::initializer_list<T> elems_) noexcept : elems{elems_} {}
 
-    template <typename T2>
-    [[nodiscard]] constexpr Vec2<T2> Cast() const {
-        return Vec2<T2>(static_cast<T2>(x), static_cast<T2>(y));
+    [[nodiscard]] constexpr Vec<decltype(T{} + T{}), N> operator+(const Vec o) const noexcept {
+        Vec<decltype(T{} + T{}), N> r{};
+        for (size_t i = 0; i < N; ++i)
+            r.elems[i] = elems[i] + o.elems[i];
+        return r;
     }
+    constexpr Vec<T, N> operator+=(const Vec<T, N> o) noexcept { return *this = *this + o; }
 
-    [[nodiscard]] static constexpr Vec2 AssignToAll(const T& f) {
-        return Vec2{f, f};
-    }
-
-    [[nodiscard]] constexpr Vec2<decltype(T{} + T{})> operator+(const Vec2& other) const {
-        return {x + other.x, y + other.y};
-    }
-    constexpr Vec2& operator+=(const Vec2& other) {
-        x += other.x;
-        y += other.y;
-        return *this;
-    }
-    [[nodiscard]] constexpr Vec2<decltype(T{} - T{})> operator-(const Vec2& other) const {
-        return {x - other.x, y - other.y};
-    }
-    constexpr Vec2& operator-=(const Vec2& other) {
-        x -= other.x;
-        y -= other.y;
-        return *this;
+    [[nodiscard]] constexpr Vec<decltype(T{} - T{}), N> operator-(const Vec o) const noexcept {
+        Vec<decltype(T{} - T{}), N> r{};
+        for (size_t i = 0; i < N; ++i)
+            r.elems[i] = elems[i] - o.elems[i];
+        return r;
     }
+    constexpr Vec<T, N> operator-=(const Vec<T, N> o) noexcept { return *this = *this - o; }
 
     template <typename U = T>
-    [[nodiscard]] constexpr Vec2<std::enable_if_t<std::is_signed_v<U>, U>> operator-() const {
-        return {-x, -y};
-    }
-    [[nodiscard]] constexpr Vec2<decltype(T{} * T{})> operator*(const Vec2& other) const {
-        return {x * other.x, y * other.y};
+    [[nodiscard]] constexpr Vec<std::enable_if_t<std::is_signed_v<U>, U>, N> operator-() const noexcept {
+        Vec<U, N> r{};
+        for (size_t i = 0; i < N; ++i)
+            r.elems[i] = -elems[i];
+        return r;
     }
 
+    [[nodiscard]] constexpr Vec<decltype(T{} * T{}), N> operator*(const Vec o) const noexcept {
+        Vec<decltype(T{} * T{}), N> r{};
+        for (size_t i = 0; i < N; ++i)
+            r.elems[i] = elems[i] * o.elems[i];
+        return r;
+    }
     template <typename V>
-    [[nodiscard]] constexpr Vec2<decltype(T{} * V{})> operator*(const V& f) const {
+    [[nodiscard]] constexpr Vec<decltype(T{} * V{}), N> operator*(const V f) const noexcept {
         using TV = decltype(T{} * V{});
         using C = std::common_type_t<T, V>;
-
-        return {
-            static_cast<TV>(static_cast<C>(x) * static_cast<C>(f)),
-            static_cast<TV>(static_cast<C>(y) * static_cast<C>(f)),
-        };
+        Vec<TV, N> r{};
+        for (size_t i = 0; i < N; ++i)
+            r.elems[i] = TV(C(elems[i]) * C(f));
+        return r;
     }
+    template <typename V>
+    constexpr Vec<T, N> operator*=(const V f) noexcept { return *this = *this * f; }
 
     template <typename V>
-    constexpr Vec2& operator*=(const V& f) {
-        *this = *this * f;
-        return *this;
-    }
-
-    template <typename V>
-    [[nodiscard]] constexpr Vec2<decltype(T{} / V{})> operator/(const V& f) const {
+    [[nodiscard]] constexpr Vec<decltype(T{} / V{}), N> operator/(const V f) const noexcept {
         using TV = decltype(T{} / V{});
         using C = std::common_type_t<T, V>;
-
-        return {
-            static_cast<TV>(static_cast<C>(x) / static_cast<C>(f)),
-            static_cast<TV>(static_cast<C>(y) / static_cast<C>(f)),
-        };
+        Vec<TV, N> r{};
+        for (size_t i = 0; i < N; ++i)
+            r.elems[i] = TV(C(elems[i]) / C(f));
+        return r;
     }
-
     template <typename V>
-    constexpr Vec2& operator/=(const V& f) {
-        *this = *this / f;
-        return *this;
-    }
+    constexpr Vec<T, N> operator/=(const V f) noexcept { return *this = *this / f; }
 
-    [[nodiscard]] constexpr T Length2() const {
-        return x * x + y * y;
+    [[nodiscard]] constexpr T Length2() const noexcept {
+        T r{};
+        for (size_t i = 0; i < N; ++i)
+            r += elems[i] * elems[i];
+        return r;
     }
-
     // Only implemented for T=float
-    [[nodiscard]] float Length() const;
-    [[nodiscard]] float Normalize(); // returns the previous length, which is often useful
+    [[nodiscard]] T Length() const { return T(std::sqrt(float(Length2()))); }
+    [[nodiscard]] Vec<T, N> Normalized() const { return *this / Length(); }
+    [[nodiscard]] constexpr T& operator[](std::size_t i) noexcept { return elems[i]; }
+    [[nodiscard]] constexpr const T& operator[](std::size_t i) const noexcept { return elems[i]; }
 
-    [[nodiscard]] constexpr T& operator[](std::size_t i) {
-        return *((&x) + i);
-    }
-    [[nodiscard]] constexpr const T& operator[](std::size_t i) const {
-        return *((&x) + i);
-    }
+    [[nodiscard]] std::array<decltype(-T{}), 16> ToMatrix() const {
+        const T x2 = elems[0] * elems[0];
+        const T y2 = elems[1] * elems[1];
+        const T z2 = elems[2] * elems[2];
 
-    constexpr void SetZero() {
-        x = 0;
-        y = 0;
-    }
-
-    // Common aliases: UV (texel coordinates), ST (texture coordinates)
-    [[nodiscard]] constexpr T& u() {
-        return x;
-    }
-    [[nodiscard]] constexpr T& v() {
-        return y;
-    }
-    [[nodiscard]] constexpr T& s() {
-        return x;
-    }
-    [[nodiscard]] constexpr T& t() {
-        return y;
-    }
-
-    [[nodiscard]] constexpr const T& u() const {
-        return x;
-    }
-    [[nodiscard]] constexpr const T& v() const {
-        return y;
-    }
-    [[nodiscard]] constexpr const T& s() const {
-        return x;
-    }
-    [[nodiscard]] constexpr const T& t() const {
-        return y;
-    }
-
-    // swizzlers - create a subvector of specific components
-    [[nodiscard]] constexpr Vec2 yx() const {
-        return Vec2(y, x);
-    }
-    [[nodiscard]] constexpr Vec2 vu() const {
-        return Vec2(y, x);
-    }
-    [[nodiscard]] constexpr Vec2 ts() const {
-        return Vec2(y, x);
+        const T xy = elems[0] * elems[1];
+        const T wz = elems[3] * elems[2];
+        const T xz = elems[0] * elems[2];
+        const T wy = elems[3] * elems[1];
+        const T yz = elems[1] * elems[2];
+        const T wx = elems[3] * elems[0];
+        return {
+            1.0f - 2.0f * (y2 + z2),
+            2.0f * (xy + wz),
+            2.0f * (xz - wy),
+            0.0f,
+            2.0f * (xy - wz),
+            1.0f - 2.0f * (x2 + z2),
+            2.0f * (yz + wx),
+            0.0f,
+            2.0f * (xz + wy),
+            2.0f * (yz - wx),
+            1.0f - 2.0f * (x2 + y2),
+            0.0f,
+            0.0f,
+            0.0f,
+            0.0f,
+            1.0f
+        };
     }
 };
 
-template <typename T, typename V>
-[[nodiscard]] constexpr Vec2<T> operator*(const V& f, const Vec2<T>& vec) {
+template <typename T, size_t N, typename V>
+[[nodiscard]] constexpr Vec<T, N> operator*(const V f, const Vec<T, N> v) noexcept {
     using C = std::common_type_t<T, V>;
-
-    return Vec2<T>(static_cast<T>(static_cast<C>(f) * static_cast<C>(vec.x)),
-                   static_cast<T>(static_cast<C>(f) * static_cast<C>(vec.y)));
-}
-
-using Vec2f = Vec2<float>;
-
-template <>
-inline float Vec2<float>::Length() const {
-    return std::sqrt(x * x + y * y);
-}
-
-template <>
-inline float Vec2<float>::Normalize() {
-    float length = Length();
-    *this /= length;
-    return length;
-}
-
-template <typename T>
-class Vec3 {
-public:
-    T x{};
-    T y{};
-    T z{};
-
-    constexpr Vec3() = default;
-    constexpr Vec3(const T& x_, const T& y_, const T& z_) : x(x_), y(y_), z(z_) {}
-
-    template <typename T2>
-    [[nodiscard]] constexpr Vec3<T2> Cast() const {
-        return Vec3<T2>(static_cast<T2>(x), static_cast<T2>(y), static_cast<T2>(z));
-    }
-
-    [[nodiscard]] static constexpr Vec3 AssignToAll(const T& f) {
-        return Vec3(f, f, f);
-    }
-
-    [[nodiscard]] constexpr Vec3<decltype(T{} + T{})> operator+(const Vec3& other) const {
-        return {x + other.x, y + other.y, z + other.z};
-    }
-
-    constexpr Vec3& operator+=(const Vec3& other) {
-        x += other.x;
-        y += other.y;
-        z += other.z;
-        return *this;
-    }
-
-    [[nodiscard]] constexpr Vec3<decltype(T{} - T{})> operator-(const Vec3& other) const {
-        return {x - other.x, y - other.y, z - other.z};
-    }
-
-    constexpr Vec3& operator-=(const Vec3& other) {
-        x -= other.x;
-        y -= other.y;
-        z -= other.z;
-        return *this;
-    }
-
-    template <typename U = T>
-    [[nodiscard]] constexpr Vec3<std::enable_if_t<std::is_signed_v<U>, U>> operator-() const {
-        return {-x, -y, -z};
-    }
-
-    [[nodiscard]] constexpr Vec3<decltype(T{} * T{})> operator*(const Vec3& other) const {
-        return {x * other.x, y * other.y, z * other.z};
-    }
-
-    template <typename V>
-    [[nodiscard]] constexpr Vec3<decltype(T{} * V{})> operator*(const V& f) const {
-        using TV = decltype(T{} * V{});
-        using C = std::common_type_t<T, V>;
-
-        return {
-            static_cast<TV>(static_cast<C>(x) * static_cast<C>(f)),
-            static_cast<TV>(static_cast<C>(y) * static_cast<C>(f)),
-            static_cast<TV>(static_cast<C>(z) * static_cast<C>(f)),
-        };
-    }
-
-    template <typename V>
-    constexpr Vec3& operator*=(const V& f) {
-        *this = *this * f;
-        return *this;
-    }
-    template <typename V>
-    [[nodiscard]] constexpr Vec3<decltype(T{} / V{})> operator/(const V& f) const {
-        using TV = decltype(T{} / V{});
-        using C = std::common_type_t<T, V>;
-
-        return {
-            static_cast<TV>(static_cast<C>(x) / static_cast<C>(f)),
-            static_cast<TV>(static_cast<C>(y) / static_cast<C>(f)),
-            static_cast<TV>(static_cast<C>(z) / static_cast<C>(f)),
-        };
-    }
-
-    template <typename V>
-    constexpr Vec3& operator/=(const V& f) {
-        *this = *this / f;
-        return *this;
-    }
-
-    void RotateFromOrigin(float roll, float pitch, float yaw) {
-        float temp = y;
-        y = std::cos(roll) * y - std::sin(roll) * z;
-        z = std::sin(roll) * temp + std::cos(roll) * z;
-
-        temp = x;
-        x = std::cos(pitch) * x + std::sin(pitch) * z;
-        z = -std::sin(pitch) * temp + std::cos(pitch) * z;
-
-        temp = x;
-        x = std::cos(yaw) * x - std::sin(yaw) * y;
-        y = std::sin(yaw) * temp + std::cos(yaw) * y;
-    }
-
-    [[nodiscard]] constexpr T Length2() const {
-        return x * x + y * y + z * z;
-    }
-
-    // Only implemented for T=float
-    [[nodiscard]] float Length() const;
-    [[nodiscard]] Vec3 Normalized() const;
-    [[nodiscard]] float Normalize(); // returns the previous length, which is often useful
-
-    [[nodiscard]] constexpr T& operator[](std::size_t i) {
-        return *((&x) + i);
-    }
-
-    [[nodiscard]] constexpr const T& operator[](std::size_t i) const {
-        return *((&x) + i);
-    }
-
-    constexpr void SetZero() {
-        x = 0;
-        y = 0;
-        z = 0;
-    }
-
-    // Common aliases: UVW (texel coordinates), RGB (colors), STQ (texture coordinates)
-    [[nodiscard]] constexpr T& u() {
-        return x;
-    }
-    [[nodiscard]] constexpr T& v() {
-        return y;
-    }
-    [[nodiscard]] constexpr T& w() {
-        return z;
-    }
-
-    [[nodiscard]] constexpr T& r() {
-        return x;
-    }
-    [[nodiscard]] constexpr T& g() {
-        return y;
-    }
-    [[nodiscard]] constexpr T& b() {
-        return z;
-    }
-
-    [[nodiscard]] constexpr T& s() {
-        return x;
-    }
-    [[nodiscard]] constexpr T& t() {
-        return y;
-    }
-    [[nodiscard]] constexpr T& q() {
-        return z;
-    }
-
-    [[nodiscard]] constexpr const T& u() const {
-        return x;
-    }
-    [[nodiscard]] constexpr const T& v() const {
-        return y;
-    }
-    [[nodiscard]] constexpr const T& w() const {
-        return z;
-    }
-
-    [[nodiscard]] constexpr const T& r() const {
-        return x;
-    }
-    [[nodiscard]] constexpr const T& g() const {
-        return y;
-    }
-    [[nodiscard]] constexpr const T& b() const {
-        return z;
-    }
-
-    [[nodiscard]] constexpr const T& s() const {
-        return x;
-    }
-    [[nodiscard]] constexpr const T& t() const {
-        return y;
-    }
-    [[nodiscard]] constexpr const T& q() const {
-        return z;
-    }
-
-// swizzlers - create a subvector of specific components
-// e.g. Vec2 uv() { return Vec2(x,y); }
-// _DEFINE_SWIZZLER2 defines a single such function, DEFINE_SWIZZLER2 defines all of them for all
-// component names (x<->r) and permutations (xy<->yx)
-#define _DEFINE_SWIZZLER2(a, b, name)                                                              \
-    [[nodiscard]] constexpr Vec2<T> name() const { return Vec2<T>(a, b); }
-#define DEFINE_SWIZZLER2(a, b, a2, b2, a3, b3, a4, b4)                                             \
-    _DEFINE_SWIZZLER2(a, b, a##b);                                                                 \
-    _DEFINE_SWIZZLER2(a, b, a2##b2);                                                               \
-    _DEFINE_SWIZZLER2(a, b, a3##b3);                                                               \
-    _DEFINE_SWIZZLER2(a, b, a4##b4);                                                               \
-    _DEFINE_SWIZZLER2(b, a, b##a);                                                                 \
-    _DEFINE_SWIZZLER2(b, a, b2##a2);                                                               \
-    _DEFINE_SWIZZLER2(b, a, b3##a3);                                                               \
-    _DEFINE_SWIZZLER2(b, a, b4##a4)
-
-    DEFINE_SWIZZLER2(x, y, r, g, u, v, s, t);
-    DEFINE_SWIZZLER2(x, z, r, b, u, w, s, q);
-    DEFINE_SWIZZLER2(y, z, g, b, v, w, t, q);
-#undef DEFINE_SWIZZLER2
-#undef _DEFINE_SWIZZLER2
-};
-
-template <typename T, typename V>
-[[nodiscard]] constexpr Vec3<T> operator*(const V& f, const Vec3<T>& vec) {
-    using C = std::common_type_t<T, V>;
-
-    return Vec3<T>(static_cast<T>(static_cast<C>(f) * static_cast<C>(vec.x)),
-                   static_cast<T>(static_cast<C>(f) * static_cast<C>(vec.y)),
-                   static_cast<T>(static_cast<C>(f) * static_cast<C>(vec.z)));
-}
-
-template <>
-inline float Vec3<float>::Length() const {
-    return std::sqrt(x * x + y * y + z * z);
-}
-
-template <>
-inline Vec3<float> Vec3<float>::Normalized() const {
-    return *this / Length();
-}
-
-template <>
-inline float Vec3<float>::Normalize() {
-    float length = Length();
-    *this /= length;
-    return length;
-}
-
-using Vec3f = Vec3<float>;
-
-template <typename T>
-class Vec4 {
-public:
-    T x{};
-    T y{};
-    T z{};
-    T w{};
-
-    constexpr Vec4() = default;
-    constexpr Vec4(const T& x_, const T& y_, const T& z_, const T& w_)
-        : x(x_), y(y_), z(z_), w(w_) {}
-
-    template <typename T2>
-    [[nodiscard]] constexpr Vec4<T2> Cast() const {
-        return Vec4<T2>(static_cast<T2>(x), static_cast<T2>(y), static_cast<T2>(z),
-                        static_cast<T2>(w));
-    }
-
-    [[nodiscard]] static constexpr Vec4 AssignToAll(const T& f) {
-        return Vec4(f, f, f, f);
-    }
-
-    [[nodiscard]] constexpr Vec4<decltype(T{} + T{})> operator+(const Vec4& other) const {
-        return {x + other.x, y + other.y, z + other.z, w + other.w};
-    }
-
-    constexpr Vec4& operator+=(const Vec4& other) {
-        x += other.x;
-        y += other.y;
-        z += other.z;
-        w += other.w;
-        return *this;
-    }
-
-    [[nodiscard]] constexpr Vec4<decltype(T{} - T{})> operator-(const Vec4& other) const {
-        return {x - other.x, y - other.y, z - other.z, w - other.w};
-    }
-
-    constexpr Vec4& operator-=(const Vec4& other) {
-        x -= other.x;
-        y -= other.y;
-        z -= other.z;
-        w -= other.w;
-        return *this;
-    }
-
-    template <typename U = T>
-    [[nodiscard]] constexpr Vec4<std::enable_if_t<std::is_signed_v<U>, U>> operator-() const {
-        return {-x, -y, -z, -w};
-    }
-
-    [[nodiscard]] constexpr Vec4<decltype(T{} * T{})> operator*(const Vec4& other) const {
-        return {x * other.x, y * other.y, z * other.z, w * other.w};
-    }
-
-    template <typename V>
-    [[nodiscard]] constexpr Vec4<decltype(T{} * V{})> operator*(const V& f) const {
-        using TV = decltype(T{} * V{});
-        using C = std::common_type_t<T, V>;
-
-        return {
-            static_cast<TV>(static_cast<C>(x) * static_cast<C>(f)),
-            static_cast<TV>(static_cast<C>(y) * static_cast<C>(f)),
-            static_cast<TV>(static_cast<C>(z) * static_cast<C>(f)),
-            static_cast<TV>(static_cast<C>(w) * static_cast<C>(f)),
-        };
-    }
-
-    template <typename V>
-    constexpr Vec4& operator*=(const V& f) {
-        *this = *this * f;
-        return *this;
-    }
-
-    template <typename V>
-    [[nodiscard]] constexpr Vec4<decltype(T{} / V{})> operator/(const V& f) const {
-        using TV = decltype(T{} / V{});
-        using C = std::common_type_t<T, V>;
-
-        return {
-            static_cast<TV>(static_cast<C>(x) / static_cast<C>(f)),
-            static_cast<TV>(static_cast<C>(y) / static_cast<C>(f)),
-            static_cast<TV>(static_cast<C>(z) / static_cast<C>(f)),
-            static_cast<TV>(static_cast<C>(w) / static_cast<C>(f)),
-        };
-    }
-
-    template <typename V>
-    constexpr Vec4& operator/=(const V& f) {
-        *this = *this / f;
-        return *this;
-    }
-
-    [[nodiscard]] constexpr T Length2() const {
-        return x * x + y * y + z * z + w * w;
-    }
-
-    [[nodiscard]] constexpr T& operator[](std::size_t i) {
-        return *((&x) + i);
-    }
-
-    [[nodiscard]] constexpr const T& operator[](std::size_t i) const {
-        return *((&x) + i);
-    }
-
-    constexpr void SetZero() {
-        x = 0;
-        y = 0;
-        z = 0;
-        w = 0;
-    }
-
-    // Common alias: RGBA (colors)
-    [[nodiscard]] constexpr T& r() {
-        return x;
-    }
-    [[nodiscard]] constexpr T& g() {
-        return y;
-    }
-    [[nodiscard]] constexpr T& b() {
-        return z;
-    }
-    [[nodiscard]] constexpr T& a() {
-        return w;
-    }
-
-    [[nodiscard]] constexpr const T& r() const {
-        return x;
-    }
-    [[nodiscard]] constexpr const T& g() const {
-        return y;
-    }
-    [[nodiscard]] constexpr const T& b() const {
-        return z;
-    }
-    [[nodiscard]] constexpr const T& a() const {
-        return w;
-    }
-
-// Swizzlers - Create a subvector of specific components
-// e.g. Vec2 uv() { return Vec2(x,y); }
-
-// _DEFINE_SWIZZLER2 defines a single such function
-// DEFINE_SWIZZLER2_COMP1 defines one-component functions for all component names (x<->r)
-// DEFINE_SWIZZLER2_COMP2 defines two component functions for all component names (x<->r) and
-// permutations (xy<->yx)
-#define _DEFINE_SWIZZLER2(a, b, name)                                                              \
-    [[nodiscard]] constexpr Vec2<T> name() const { return Vec2<T>(a, b); }
-#define DEFINE_SWIZZLER2_COMP1(a, a2)                                                              \
-    _DEFINE_SWIZZLER2(a, a, a##a);                                                                 \
-    _DEFINE_SWIZZLER2(a, a, a2##a2)
-#define DEFINE_SWIZZLER2_COMP2(a, b, a2, b2)                                                       \
-    _DEFINE_SWIZZLER2(a, b, a##b);                                                                 \
-    _DEFINE_SWIZZLER2(a, b, a2##b2);                                                               \
-    _DEFINE_SWIZZLER2(b, a, b##a);                                                                 \
-    _DEFINE_SWIZZLER2(b, a, b2##a2)
-
-    DEFINE_SWIZZLER2_COMP2(x, y, r, g);
-    DEFINE_SWIZZLER2_COMP2(x, z, r, b);
-    DEFINE_SWIZZLER2_COMP2(x, w, r, a);
-    DEFINE_SWIZZLER2_COMP2(y, z, g, b);
-    DEFINE_SWIZZLER2_COMP2(y, w, g, a);
-    DEFINE_SWIZZLER2_COMP2(z, w, b, a);
-    DEFINE_SWIZZLER2_COMP1(x, r);
-    DEFINE_SWIZZLER2_COMP1(y, g);
-    DEFINE_SWIZZLER2_COMP1(z, b);
-    DEFINE_SWIZZLER2_COMP1(w, a);
-#undef DEFINE_SWIZZLER2_COMP1
-#undef DEFINE_SWIZZLER2_COMP2
-#undef _DEFINE_SWIZZLER2
-
-#define _DEFINE_SWIZZLER3(a, b, c, name)                                                           \
-    [[nodiscard]] constexpr Vec3<T> name() const { return Vec3<T>(a, b, c); }
-#define DEFINE_SWIZZLER3_COMP1(a, a2)                                                              \
-    _DEFINE_SWIZZLER3(a, a, a, a##a##a);                                                           \
-    _DEFINE_SWIZZLER3(a, a, a, a2##a2##a2)
-#define DEFINE_SWIZZLER3_COMP3(a, b, c, a2, b2, c2)                                                \
-    _DEFINE_SWIZZLER3(a, b, c, a##b##c);                                                           \
-    _DEFINE_SWIZZLER3(a, c, b, a##c##b);                                                           \
-    _DEFINE_SWIZZLER3(b, a, c, b##a##c);                                                           \
-    _DEFINE_SWIZZLER3(b, c, a, b##c##a);                                                           \
-    _DEFINE_SWIZZLER3(c, a, b, c##a##b);                                                           \
-    _DEFINE_SWIZZLER3(c, b, a, c##b##a);                                                           \
-    _DEFINE_SWIZZLER3(a, b, c, a2##b2##c2);                                                        \
-    _DEFINE_SWIZZLER3(a, c, b, a2##c2##b2);                                                        \
-    _DEFINE_SWIZZLER3(b, a, c, b2##a2##c2);                                                        \
-    _DEFINE_SWIZZLER3(b, c, a, b2##c2##a2);                                                        \
-    _DEFINE_SWIZZLER3(c, a, b, c2##a2##b2);                                                        \
-    _DEFINE_SWIZZLER3(c, b, a, c2##b2##a2)
-
-    DEFINE_SWIZZLER3_COMP3(x, y, z, r, g, b);
-    DEFINE_SWIZZLER3_COMP3(x, y, w, r, g, a);
-    DEFINE_SWIZZLER3_COMP3(x, z, w, r, b, a);
-    DEFINE_SWIZZLER3_COMP3(y, z, w, g, b, a);
-    DEFINE_SWIZZLER3_COMP1(x, r);
-    DEFINE_SWIZZLER3_COMP1(y, g);
-    DEFINE_SWIZZLER3_COMP1(z, b);
-    DEFINE_SWIZZLER3_COMP1(w, a);
-#undef DEFINE_SWIZZLER3_COMP1
-#undef DEFINE_SWIZZLER3_COMP3
-#undef _DEFINE_SWIZZLER3
-};
-
-template <typename T, typename V>
-[[nodiscard]] constexpr Vec4<decltype(V{} * T{})> operator*(const V& f, const Vec4<T>& vec) {
-    using TV = decltype(V{} * T{});
-    using C = std::common_type_t<T, V>;
-
-    return {
-        static_cast<TV>(static_cast<C>(f) * static_cast<C>(vec.x)),
-        static_cast<TV>(static_cast<C>(f) * static_cast<C>(vec.y)),
-        static_cast<TV>(static_cast<C>(f) * static_cast<C>(vec.z)),
-        static_cast<TV>(static_cast<C>(f) * static_cast<C>(vec.w)),
-    };
-}
-
-using Vec4f = Vec4<float>;
-
-template <typename T>
-constexpr decltype(T{} * T{} + T{} * T{}) Dot(const Vec2<T>& a, const Vec2<T>& b) {
-    return a.x * b.x + a.y * b.y;
-}
-
-template <typename T>
-[[nodiscard]] constexpr decltype(T{} * T{} + T{} * T{}) Dot(const Vec3<T>& a, const Vec3<T>& b) {
-    return a.x * b.x + a.y * b.y + a.z * b.z;
-}
-
-template <typename T>
-[[nodiscard]] constexpr decltype(T{} * T{} + T{} * T{}) Dot(const Vec4<T>& a, const Vec4<T>& b) {
-    return a.x * b.x + a.y * b.y + a.z * b.z + a.w * b.w;
-}
-
-template <>
-[[nodiscard]] inline float Dot(const Vec4<float>& a, const Vec4<float>& b) {
-#ifdef __ARM_NEON
-    float32x4_t va = vld1q_f32(&a.x);
-    float32x4_t vb = vld1q_f32(&b.x);
-    float32x4_t result = vmulq_f32(va, vb);
-#if defined(__aarch64__) // Use vaddvq_f32 in ARMv8 architectures
-    return vaddvq_f32(result);
-#else // Use manual addition for older architectures
-    float32x2_t sum2 = vadd_f32(vget_high_f32(result), vget_low_f32(result));
-    return vget_lane_f32(vpadd_f32(sum2, sum2), 0);
-#endif
-#else
-    return a.x * b.x + a.y * b.y + a.z * b.z + a.w * b.w;
-#endif
-}
-
-template <typename T>
-[[nodiscard]] constexpr Vec3<decltype(T{} * T{} - T{} * T{})> Cross(const Vec3<T>& a,
-                                                                    const Vec3<T>& b) {
-    return {a.y * b.z - a.z * b.y, a.z * b.x - a.x * b.z, a.x * b.y - a.y * b.x};
-}
-
-// linear interpolation via float: 0.0=begin, 1.0=end
-template <typename X>
-[[nodiscard]] constexpr decltype(X{} * float{} + X{} * float{}) Lerp(const X& begin, const X& end,
-                                                                     const float t) {
-    return begin * (1.f - t) + end * t;
-}
-
-// linear interpolation via int: 0=begin, base=end
-template <typename X, int base>
-[[nodiscard]] constexpr decltype((X{} * int{} + X{} * int{}) / base) LerpInt(const X& begin,
-                                                                             const X& end,
-                                                                             const int t) {
-    return (begin * (base - t) + end * t) / base;
-}
-
-// bilinear interpolation. s is for interpolating x00-x01 and x10-x11, and t is for the second
-// interpolation.
-template <typename X>
-[[nodiscard]] constexpr auto BilinearInterp(const X& x00, const X& x01, const X& x10, const X& x11,
-                                            const float s, const float t) {
-    auto y0 = Lerp(x00, x01, s);
-    auto y1 = Lerp(x10, x11, s);
-    return Lerp(y0, y1, t);
-}
-
-// Utility vector factories
-template <typename T>
-[[nodiscard]] constexpr Vec2<T> MakeVec(const T& x, const T& y) {
-    return Vec2<T>{x, y};
-}
-
-template <typename T>
-[[nodiscard]] constexpr Vec3<T> MakeVec(const T& x, const T& y, const T& z) {
-    return Vec3<T>{x, y, z};
-}
-
-template <typename T>
-[[nodiscard]] constexpr Vec4<T> MakeVec(const T& x, const T& y, const Vec2<T>& zw) {
-    return MakeVec(x, y, zw[0], zw[1]);
-}
-
-template <typename T>
-[[nodiscard]] constexpr Vec3<T> MakeVec(const Vec2<T>& xy, const T& z) {
-    return MakeVec(xy[0], xy[1], z);
-}
-
-template <typename T>
-[[nodiscard]] constexpr Vec3<T> MakeVec(const T& x, const Vec2<T>& yz) {
-    return MakeVec(x, yz[0], yz[1]);
-}
-
-template <typename T>
-[[nodiscard]] constexpr Vec4<T> MakeVec(const T& x, const T& y, const T& z, const T& w) {
-    return Vec4<T>{x, y, z, w};
-}
-
-template <typename T>
-[[nodiscard]] constexpr Vec4<T> MakeVec(const Vec2<T>& xy, const T& z, const T& w) {
-    return MakeVec(xy[0], xy[1], z, w);
-}
-
-template <typename T>
-[[nodiscard]] constexpr Vec4<T> MakeVec(const T& x, const Vec2<T>& yz, const T& w) {
-    return MakeVec(x, yz[0], yz[1], w);
-}
-
-// NOTE: This has priority over "Vec2<Vec2<T>> MakeVec(const Vec2<T>& x, const Vec2<T>& y)".
-//       Even if someone wanted to use an odd object like Vec2<Vec2<T>>, the compiler would error
-//       out soon enough due to misuse of the returned structure.
-template <typename T>
-[[nodiscard]] constexpr Vec4<T> MakeVec(const Vec2<T>& xy, const Vec2<T>& zw) {
-    return MakeVec(xy[0], xy[1], zw[0], zw[1]);
-}
-
-template <typename T>
-[[nodiscard]] constexpr Vec4<T> MakeVec(const Vec3<T>& xyz, const T& w) {
-    return MakeVec(xyz[0], xyz[1], xyz[2], w);
-}
-
-template <typename T>
-[[nodiscard]] constexpr Vec4<T> MakeVec(const T& x, const Vec3<T>& yzw) {
-    return MakeVec(x, yzw[0], yzw[1], yzw[2]);
+    Vec<T, N> r{};
+    for (size_t i = 0; i < N; ++i)
+        r.elems[i] = T(C(f) * C(v.elems[i]));
+    return r;
 }
 
 } // namespace Common

src/common/virtual_buffer.cpp

@@ -1,44 +0,0 @@
-// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
-// SPDX-License-Identifier: GPL-3.0-or-later
-
-// SPDX-FileCopyrightText: Copyright 2020 yuzu Emulator Project
-// SPDX-License-Identifier: GPL-2.0-or-later
-
-#ifdef _WIN32
-#include <windows.h>
-#else
-#include <sys/mman.h>
-#endif
-
-#include "common/assert.h"
-#include "common/virtual_buffer.h"
-
-namespace Common {
-
-void* AllocateMemoryPages(std::size_t size) noexcept {
-#ifdef _WIN32
-    void* base = VirtualAlloc(nullptr, size, MEM_COMMIT | MEM_RESERVE, PAGE_READWRITE);
-    if (base == nullptr) {
-        // Probably failing to reserve is less likely than failing to commit
-        base = VirtualAlloc(nullptr, size, MEM_COMMIT, PAGE_READWRITE);
-    }
-#else
-    void* base = mmap(nullptr, size, PROT_READ | PROT_WRITE, MAP_ANON | MAP_PRIVATE, -1, 0);
-    if (base == MAP_FAILED)
-        base = nullptr;
-#endif
-    ASSERT(base);
-    return base;
-}
-
-void FreeMemoryPages(void* base, [[maybe_unused]] std::size_t size) noexcept {
-    if (!base)
-        return;
-#ifdef _WIN32
-    ASSERT(VirtualFree(base, 0, MEM_RELEASE));
-#else
-    ASSERT(munmap(base, size) == 0);
-#endif
-}
-
-} // namespace Common

src/common/virtual_buffer.h

@@ -1,84 +0,0 @@
-// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
-// SPDX-License-Identifier: GPL-3.0-or-later
-
-// SPDX-FileCopyrightText: Copyright 2020 yuzu Emulator Project
-// SPDX-License-Identifier: GPL-2.0-or-later
-
-#pragma once
-
-#include <utility>
-
-namespace Common {
-
-void* AllocateMemoryPages(std::size_t size) noexcept;
-void FreeMemoryPages(void* base, std::size_t size) noexcept;
-
-template <typename T>
-class VirtualBuffer final {
-public:
-    // TODO: Uncomment this and change Common::PageTable::PageInfo to be trivially constructible
-    // using std::atomic_ref once libc++ has support for it
-    // static_assert(
-    //     std::is_trivially_constructible_v<T>,
-    //     "T must be trivially constructible, as non-trivial constructors will not be executed "
-    //     "with the current allocator");
-
-    constexpr VirtualBuffer() = default;
-    explicit VirtualBuffer(std::size_t count) noexcept
-        : alloc_size{count * sizeof(T)}
-    {
-        base_ptr = reinterpret_cast<T*>(AllocateMemoryPages(alloc_size));
-    }
-
-    ~VirtualBuffer() noexcept {
-        FreeMemoryPages(base_ptr, alloc_size);
-    }
-
-    VirtualBuffer(const VirtualBuffer&) = delete;
-    VirtualBuffer& operator=(const VirtualBuffer&) = delete;
-
-    VirtualBuffer(VirtualBuffer&& other) noexcept
-        : alloc_size{std::exchange(other.alloc_size, 0)}
-        , base_ptr{std::exchange(other.base_ptr, nullptr)}
-    {}
-
-    VirtualBuffer& operator=(VirtualBuffer&& other) noexcept {
-        alloc_size = std::exchange(other.alloc_size, 0);
-        base_ptr = std::exchange(other.base_ptr, nullptr);
-        return *this;
-    }
-
-    void resize(std::size_t count) noexcept {
-        if (auto const new_size = count * sizeof(T); new_size != alloc_size) {
-            FreeMemoryPages(base_ptr, alloc_size);
-            alloc_size = new_size;
-            base_ptr = reinterpret_cast<T*>(AllocateMemoryPages(alloc_size));
-        }
-    }
-
-    [[nodiscard]] constexpr const T& operator[](std::size_t index) const noexcept {
-        return base_ptr[index];
-    }
-
-    [[nodiscard]] constexpr T& operator[](std::size_t index) noexcept {
-        return base_ptr[index];
-    }
-
-    [[nodiscard]] constexpr T* data() noexcept {
-        return base_ptr;
-    }
-
-    [[nodiscard]] constexpr const T* data() const noexcept {
-        return base_ptr;
-    }
-
-    [[nodiscard]] constexpr std::size_t size() const noexcept {
-        return alloc_size / sizeof(T);
-    }
-
-private:
-    std::size_t alloc_size{};
-    T* base_ptr{};
-};
-
-} // namespace Common

src/core/CMakeLists.txt

@@ -711,8 +711,6 @@ add_library(core STATIC
     hle/service/kernel_helpers.h
     hle/service/lbl/lbl.cpp
     hle/service/lbl/lbl.h
-    hle/service/ldn/client_process_monitor.cpp
-    hle/service/ldn/client_process_monitor.h
     hle/service/ldn/lan_discovery.cpp
     hle/service/ldn/lan_discovery.h
     hle/service/ldn/ldn.cpp
@@ -832,8 +830,6 @@ add_library(core STATIC
     hle/service/ns/system_update_control.h
     hle/service/ns/system_update_interface.cpp
     hle/service/ns/system_update_interface.h
-    hle/service/ns/vulnerability_manager_interface.cpp
-    hle/service/ns/vulnerability_manager_interface.h
     hle/service/nvdrv/core/container.cpp
     hle/service/nvdrv/core/container.h
     hle/service/nvdrv/core/heap_mapper.cpp
@@ -1067,8 +1063,6 @@ add_library(core STATIC
     hle/service/sockets/sockets.h
     hle/service/sockets/sockets_translate.cpp
     hle/service/sockets/sockets_translate.h
-    hle/service/spl/csrng.cpp
-    hle/service/spl/csrng.h
     hle/service/spl/spl.cpp
     hle/service/spl/spl.h
     hle/service/spl/spl_module.cpp

src/core/arm/dynarmic/arm_dynarmic_32.cpp

@@ -172,12 +172,12 @@ void ArmDynarmic32::MakeJit(Common::PageTable* page_table) {
     if (page_table) {
         constexpr size_t PageBits = 12;
         constexpr size_t NumPageTableEntries = 1 << (32 - PageBits);
-        constexpr size_t PageLog2Stride = 5;
-        static_assert(1 << PageLog2Stride == sizeof(Common::PageTable::PageEntryData));
 
-        config.page_table = reinterpret_cast<std::array<std::uint8_t*, NumPageTableEntries>*>(page_table->entries.data());
-        config.page_table_pointer_mask_bits = Common::PageTable::ATTRIBUTE_BITS;
-        config.page_table_log2_stride = PageLog2Stride;
+        // Dynarmic will not write to the page table, const_cast is safe here
+        config.page_table = reinterpret_cast<std::array<std::uint8_t*, NumPageTableEntries>*>(
+            const_cast<Common::PageTable::PageEntryData*>(page_table->entries.data()));
+        config.page_table_pointer_mask = Common::PageTable::ATTRIBUTE_MASK;
+        config.page_table_marked_bit = 0;
         config.absolute_offset_page_table = true;
         config.detect_misaligned_access_via_page_table = 16 | 32 | 64 | 128;
         config.only_detect_misalignment_via_page_table_on_page_boundary = true;
@@ -188,6 +188,13 @@ void ArmDynarmic32::MakeJit(Common::PageTable* page_table) {
 
         config.fastmem_exclusive_access = config.fastmem_pointer  != std::nullopt;
         config.recompile_on_exclusive_fastmem_failure = true;
+
+        if (reinterpret_cast<u64>(m_system.DeviceMemory().buffer.BackingBasePointer() +
+            Kernel::Board::Nintendo::Nx::KSystemControl::Init::GetIntendedMemorySize()) < (1ULL << 39)) {
+            // Systems like FreeBSD allocate memory really low by default, and since we pack our page table entries,
+            // we have to manually sign extend when our actual pointer is negative.
+            config.page_table_sign_extension = Common::PageTable::SIGN_BIT;
+        }
     }
 
     // Multi-process state
@@ -420,6 +427,7 @@ void ArmDynarmic32::SignalInterrupt(Kernel::KThread* thread) {
 }
 
 void ArmDynarmic32::ClearInstructionCache() {
+    m_cb->last_code_addr = u64(-1);
     m_jit->ClearCache();
 }
 

src/core/arm/dynarmic/arm_dynarmic_64.cpp

@@ -211,13 +211,12 @@ void ArmDynarmic64::MakeJit(Common::PageTable* page_table, std::size_t address_s
 
     // Memory
     if (page_table) {
-        constexpr size_t PageLog2Stride = 5;
-        static_assert(1 << PageLog2Stride == sizeof(Common::PageTable::PageEntryData));
-
-        config.page_table = reinterpret_cast<void**>(page_table->entries.data());
+        // Dynarmic will not write to the page table, const_cast is safe here
+        config.page_table = reinterpret_cast<void**>(
+            const_cast<Common::PageTable::PageEntryData*>(page_table->entries.data()));
         config.page_table_address_space_bits = std::uint32_t(address_space_bits);
-        config.page_table_pointer_mask_bits = Common::PageTable::ATTRIBUTE_BITS;
-        config.page_table_log2_stride = PageLog2Stride;
+        config.page_table_pointer_mask = Common::PageTable::ATTRIBUTE_MASK;
+        config.page_table_marked_bit = 0;
         config.silently_mirror_page_table = false;
         config.absolute_offset_page_table = true;
         config.detect_misaligned_access_via_page_table = 16 | 32 | 64 | 128;
@@ -231,6 +230,13 @@ void ArmDynarmic64::MakeJit(Common::PageTable* page_table, std::size_t address_s
 
         config.fastmem_exclusive_access = config.fastmem_pointer != std::nullopt;
         config.recompile_on_exclusive_fastmem_failure = true;
+
+        if (reinterpret_cast<u64>(m_system.DeviceMemory().buffer.BackingBasePointer() +
+            Kernel::Board::Nintendo::Nx::KSystemControl::Init::GetIntendedMemorySize()) < (1ULL << 39)) {
+            // Systems like FreeBSD allocate memory really low by default, and since we pack our page table entries,
+            // we have to manually sign extend when our actual pointer is negative.
+            config.page_table_sign_extension = Common::PageTable::SIGN_BIT;
+        }
     }
 
     // Multi-process state
@@ -447,6 +453,7 @@ void ArmDynarmic64::SignalInterrupt(Kernel::KThread* thread) {
 }
 
 void ArmDynarmic64::ClearInstructionCache() {
+    m_cb->last_code_addr = u64(-1);
     m_jit->ClearCache();
 }
 

src/core/arm/nce/instructions.h

@@ -121,7 +121,10 @@ union Exclusive {
     constexpr explicit Exclusive(u32 raw_) : raw{raw_} {}
 
     constexpr bool Verify() {
-        return this->GetSig() == 0x10;
+        if (this->GetSig() != 0x10) return false;
+        const bool pair = decltype(l)::ExtractValue(raw) & 1;
+        const bool fixed_rt2 = decltype(rt2)::ExtractValue(raw) == 0b11111;
+        return pair || fixed_rt2;
     }
 
     constexpr u32 GetSig() {

src/core/device_memory.h

@@ -1,3 +1,6 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
 // SPDX-FileCopyrightText: Copyright 2020 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
@@ -26,8 +29,11 @@ public:
 
     template <typename T>
     Common::PhysicalAddress GetPhysicalAddr(const T* ptr) const {
-        return (reinterpret_cast<uintptr_t>(ptr) -
-                reinterpret_cast<uintptr_t>(buffer.BackingBasePointer())) +
+        return GetPhysicalAddr(reinterpret_cast<uintptr_t>(ptr));
+    }
+
+    Common::PhysicalAddress GetPhysicalAddr(uintptr_t ptr) const {
+        return (ptr - reinterpret_cast<uintptr_t>(buffer.BackingBasePointer())) +
                DramMemoryMap::Base;
     }
 

src/core/device_memory_manager.h

@@ -18,7 +18,7 @@
 #include "common/common_types.h"
 #include "common/range_mutex.h"
 #include "common/scratch_buffer.h"
-#include "common/virtual_buffer.h"
+#include "common/sparse_large_vector.h"
 
 namespace Core {
 
@@ -178,8 +178,8 @@ private:
         u32 continuity_tracker;
         u32 compressed_physical_ptr;
     };
-    Common::VirtualBuffer<u32> compressed_device_addr;
-    Common::VirtualBuffer<TrackedEntry> tracked_entries;
+    Common::SparseLargeVector<u32> compressed_device_addr;
+    Common::SparseLargeVector<TrackedEntry> tracked_entries;
 
     // Process memory interfaces
 
@@ -200,8 +200,8 @@ private:
         return std::make_pair(asid, address);
     }
 
-    void InsertCPUBacking(size_t page_index, VAddr address, Asid asid) {
-        tracked_entries[page_index].cpu_backing_address = address | (asid.id << asid_start_bit);
+    constexpr void InsertCPUBacking(size_t page_index, VAddr address, Asid asid) {
+        tracked_entries.GetUnchecked(page_index).cpu_backing_address = address | (asid.id << asid_start_bit);
     }
 
     std::array<TranslationEntry, 4> t_slot{};

src/core/device_memory_manager.inc

@@ -177,17 +177,6 @@ DeviceMemoryManager<Traits>::DeviceMemoryManager(const DeviceMemory& device_memo
 {
     impl = std::make_unique<DeviceMemoryManagerAllocator<Traits>>();
     cached_pages = std::make_unique<CachedPages>();
-
-    const size_t total_virtual = device_as_size >> Memory::YUZU_PAGEBITS;
-    for (size_t i = 0; i < total_virtual; i++) {
-        tracked_entries[i].compressed_physical_ptr = 0;
-        tracked_entries[i].continuity_tracker = 1;
-        tracked_entries[i].cpu_backing_address = 0;
-    }
-    const size_t total_phys = 1ULL << ((Settings::values.memory_layout_mode.GetValue() == Settings::MemoryLayout::Memory_4Gb ? physical_min_bits : physical_max_bits) - Memory::YUZU_PAGEBITS);
-    for (size_t i = 0; i < total_phys; i++) {
-        compressed_device_addr[i] = 0;
-    }
 }
 
 template <typename Traits>
@@ -220,26 +209,28 @@ void DeviceMemoryManager<Traits>::Map(DAddr address, VAddr virtual_address, size
     size_t start_page_d = address >> Memory::YUZU_PAGEBITS;
     size_t num_pages = Common::AlignUp(size, Memory::YUZU_PAGESIZE) >> Memory::YUZU_PAGEBITS;
     std::scoped_lock lk(mapping_guard);
+
+    tracked_entries.CommitRegion(start_page_d, start_page_d + num_pages);
     for (size_t i = 0; i < num_pages; i++) {
         const VAddr new_vaddress = virtual_address + i * Memory::YUZU_PAGESIZE;
         auto* ptr = process_memory->GetPointerSilent(Common::ProcessAddress(new_vaddress));
         if (ptr == nullptr) [[unlikely]] {
-            tracked_entries[start_page_d + i].compressed_physical_ptr = 0;
+            tracked_entries.GetUnchecked(start_page_d + i).compressed_physical_ptr = 0;
             continue;
         }
         auto phys_addr = static_cast<u32>(GetRawPhysicalAddr(ptr) >> Memory::YUZU_PAGEBITS) + 1U;
-        tracked_entries[start_page_d + i].compressed_physical_ptr = phys_addr;
+        tracked_entries.GetUnchecked(start_page_d + i).compressed_physical_ptr = phys_addr;
         InsertCPUBacking(start_page_d + i, new_vaddress, asid);
         const u32 base_dev = compressed_device_addr[phys_addr - 1U];
         const u32 new_dev = static_cast<u32>(start_page_d + i);
         if (base_dev == 0) [[likely]] {
-            compressed_device_addr[phys_addr - 1U] = new_dev;
+            compressed_device_addr.GetAndFault(phys_addr - 1U) = new_dev;
             continue;
         }
         u32 start_id = base_dev & MULTI_MASK;
         if ((base_dev >> MULTI_FLAG_BITS) == 0) {
             start_id = impl->multi_dev_address.Register(base_dev);
-            compressed_device_addr[phys_addr - 1U] = MULTI_FLAG | start_id;
+            compressed_device_addr.GetAndFault(phys_addr - 1U) = MULTI_FLAG | start_id;
         }
         impl->multi_dev_address.Register(new_dev, start_id);
     }
@@ -255,24 +246,26 @@ void DeviceMemoryManager<Traits>::Unmap(DAddr address, size_t size) {
     size_t num_pages = Common::AlignUp(size, Memory::YUZU_PAGESIZE) >> Memory::YUZU_PAGEBITS;
     device_inter->InvalidateRegion(address, size);
     std::scoped_lock lk(mapping_guard);
+
+    tracked_entries.CommitRegion(start_page_d, start_page_d + num_pages); // should already be committed, but just in case
     for (size_t i = 0; i < num_pages; i++) {
-        auto phys_addr = tracked_entries[start_page_d + i].compressed_physical_ptr;
-        tracked_entries[start_page_d + i].compressed_physical_ptr = 0;
-        tracked_entries[start_page_d + i].cpu_backing_address = 0;
+        auto& entry = tracked_entries.GetUnchecked(start_page_d + i);
+        auto phys_addr = entry.compressed_physical_ptr;
+        entry.compressed_physical_ptr = 0;
+        entry.cpu_backing_address = 0;
         if (phys_addr != 0) [[likely]] {
-            const u32 base_dev = compressed_device_addr[phys_addr - 1U];
+            u32& base_dev = compressed_device_addr.GetAndFault(phys_addr - 1U);
             if ((base_dev >> MULTI_FLAG_BITS) == 0) [[likely]] {
-                compressed_device_addr[phys_addr - 1] = 0;
+                base_dev = 0;
                 continue;
             }
             const auto [more_entries, new_start] = impl->multi_dev_address.Unregister(
                 static_cast<u32>(start_page_d + i), base_dev & MULTI_MASK);
             if (!more_entries) {
-                compressed_device_addr[phys_addr - 1] =
-                    impl->multi_dev_address.ReleaseEntry(new_start);
+                base_dev = impl->multi_dev_address.ReleaseEntry(new_start);
                 continue;
             }
-            compressed_device_addr[phys_addr - 1] = new_start | MULTI_FLAG;
+            base_dev = new_start | MULTI_FLAG;
         }
     }
     t_slot = {};
@@ -285,6 +278,8 @@ void DeviceMemoryManager<Traits>::TrackContinuityImpl(DAddr address, VAddr virtu
     size_t num_pages = Common::AlignUp(size, Memory::YUZU_PAGESIZE) >> Memory::YUZU_PAGEBITS;
     uintptr_t last_ptr = 0;
     size_t page_count = 1;
+
+    tracked_entries.CommitRegion(start_page_d, start_page_d + num_pages);
     for (size_t i = num_pages; i > 0; i--) {
         size_t index = i - 1;
         const VAddr new_vaddress = virtual_address + index * Memory::YUZU_PAGESIZE;
@@ -296,14 +291,14 @@ void DeviceMemoryManager<Traits>::TrackContinuityImpl(DAddr address, VAddr virtu
             page_count = 1;
         }
         last_ptr = new_ptr;
-        tracked_entries[start_page_d + index].continuity_tracker = static_cast<u32>(page_count);
+        tracked_entries.GetUnchecked(start_page_d + index).continuity_tracker = static_cast<u32>(page_count) - 1;
     }
 }
 template <typename Traits>
 u8* DeviceMemoryManager<Traits>::GetSpan(const DAddr src_addr, const std::size_t size) {
     size_t page_index = src_addr >> page_bits;
     size_t subbits = src_addr & page_mask;
-    if ((static_cast<size_t>(tracked_entries[page_index].continuity_tracker) << page_bits) >= size + subbits) {
+    if ((static_cast<size_t>(tracked_entries[page_index].continuity_tracker+1) << page_bits) >= size + subbits) {
         return GetPointer<u8>(src_addr);
     }
     return nullptr;
@@ -313,7 +308,7 @@ template <typename Traits>
 const u8* DeviceMemoryManager<Traits>::GetSpan(const DAddr src_addr, const std::size_t size) const {
     size_t page_index = src_addr >> page_bits;
     size_t subbits = src_addr & page_mask;
-    if ((static_cast<size_t>(tracked_entries[page_index].continuity_tracker) << page_bits) >= size + subbits) {
+    if ((static_cast<size_t>(tracked_entries[page_index].continuity_tracker+1) << page_bits) >= size + subbits) {
         return GetPointer<u8>(src_addr);
     }
     return nullptr;
@@ -383,7 +378,7 @@ void DeviceMemoryManager<Traits>::WalkBlock(DAddr addr, std::size_t size, auto o
     std::size_t page_index = addr >> Memory::YUZU_PAGEBITS;
     std::size_t page_offset = addr & Memory::YUZU_PAGEMASK;
     while (remaining_size) {
-        const size_t next_pages = std::size_t(tracked_entries[page_index].continuity_tracker);
+        const size_t next_pages = std::size_t(tracked_entries[page_index].continuity_tracker+1);
         const std::size_t copy_amount = (std::min)((next_pages << Memory::YUZU_PAGEBITS) - page_offset, remaining_size);
         const auto current_vaddr = u64((page_index << Memory::YUZU_PAGEBITS) + page_offset);
         SCOPE_EXIT{

src/core/file_sys/fssystem/fssystem_hierarchical_integrity_verification_storage.cpp

@@ -1,3 +1,6 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
 // SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
@@ -121,6 +124,7 @@ size_t HierarchicalIntegrityVerificationStorage::Read(u8* buffer, size_t size,
 }
 
 size_t HierarchicalIntegrityVerificationStorage::GetSize() const {
+    ASSERT(m_data_size >= 0);
     return m_data_size;
 }
 

src/core/file_sys/ips_layer.cpp

@@ -120,7 +120,7 @@ std::array<u8, 32> IPSwitchCompiler::GetBuildID() const {
 
 static IPSwitchRecord EscapeStringSequences(std::string_view sv) {
     IPSwitchRecord r{};
-    for (auto it = sv.cbegin(); it != sv.cend(); ) {
+    for (auto it = sv.cbegin(); it < sv.cend(); ) {
         if (*it == '\\' && it + 1 < sv.cend()) {
             switch (it[1]) {
             case 'a': r.data[r.count] = '\a'; break;
@@ -198,6 +198,8 @@ void IPSwitchCompiler::Parse(std::span<u8 const> bytes) {
                 LOG_WARNING(Loader, "Unknown flag {}", line);
                 break;
             }
+        } else if (patches.empty()) {
+            LOG_WARNING(Loader, "Invalid line not in a patch {}", line);
         } else {
             size_t offset = size_t(std::strtoul(line.data(), nullptr, 16));
             offset += size_t(offset_shift);
@@ -253,25 +255,30 @@ void IPSwitchCompiler::Parse(std::span<u8 const> bytes) {
             // now make a nominal preprocessed line: remove comments
             char quote = '\0';
             auto const sline_start = p;
+            auto last_char = p;
             for (; p < sline.cend(); ) {
                 // we dont check for "//", IPS checks for '/' only...
-                if ((!quote && p[0] == '/')
+                if (std::isspace(*p)) {
+                    ++p;
+                } else if ((!quote && p[0] == '/')
                 || (!quote && p[0] == '#')) {
                     break;
                 } else if (p[0] == '\"' || p[0] == '\'') {
                     quote = (p[0] == quote) ? '\0' : p[0];
                     ++p;
+                    last_char = p;
                 } else if (p + 1 < sline.cend() && p[0] == '\\') {
                     p += 2;
+                    last_char = p;
                 } else {
                     ++p;
+                    last_char = p;
                 }
             }
             // now we have the preprocessed string ;)
-            std::string_view pp_str(sline_start, p);
-            if (pp_str.size() > 0 && !parse_line(pp_str)) {
+            std::string_view const pp_str(sline_start, last_char);
+            if (pp_str.size() > 0 && !parse_line(pp_str))
                 break;
-            }
         }
     }
 }

src/core/hle/kernel/k_page_table_base.cpp

@@ -635,6 +635,36 @@ Result KPageTableBase::CheckMemoryState(const KMemoryInfo& info, KMemoryState st
     R_SUCCEED();
 }
 
+bool KPageTableBase::BeginTraversal(const Common::PageTable &impl, TraversalEntry *out_entry, TraversalContext *out_context,
+    Common::ProcessAddress address) const {
+    out_context->next_offset = GetInteger(address);
+    out_context->next_page = GetInteger(address) >> PageBits;
+
+    return ContinueTraversal(impl, out_entry, out_context);
+}
+
+bool KPageTableBase::ContinueTraversal(const Common::PageTable &impl, TraversalEntry *out_entry,
+    TraversalContext *context) const {
+    // Setup invalid defaults.
+    out_entry->phys_addr = 0;
+    out_entry->block_size = PageSize;
+    // Validate that we can read the actual entry.
+    if (auto const page = context->next_page; page < impl.entries.size()) {
+        // Validate that the entry is mapped.
+        if (auto const paddr = impl.entries[page].Pointer(true); paddr != 0) {
+            // Populate the results and return true
+            out_entry->phys_addr = GetInteger(m_system.DeviceMemory().GetPhysicalAddr(paddr + context->next_offset));
+            context->next_page += 1;
+            context->next_offset += PageSize;
+            return true;
+        }
+    }
+    context->next_page += 1;
+    context->next_offset += PageSize;
+    // Otherwise return false
+    return false;
+}
+
 Result KPageTableBase::CheckMemoryStateContiguous(size_t* out_blocks_needed, KProcessAddress addr,
                                                   size_t size, KMemoryState state_mask,
                                                   KMemoryState state, KMemoryPermission perm_mask,
@@ -940,7 +970,7 @@ Result KPageTableBase::QueryMappingImpl(KProcessAddress* out, KPhysicalAddress a
     size_t tot_size = 0;
 
     next_valid =
-        impl.BeginTraversal(std::addressof(next_entry), std::addressof(context), region_start);
+        BeginTraversal(impl, std::addressof(next_entry), std::addressof(context), region_start);
     next_entry.block_size =
         (next_entry.block_size - (GetInteger(region_start) & (next_entry.block_size - 1)));
 
@@ -976,7 +1006,7 @@ Result KPageTableBase::QueryMappingImpl(KProcessAddress* out, KPhysicalAddress a
             break;
         }
 
-        next_valid = impl.ContinueTraversal(std::addressof(next_entry), std::addressof(context));
+        next_valid = ContinueTraversal(impl, std::addressof(next_entry), std::addressof(context));
     }
 
     // Check the last entry.
@@ -1754,7 +1784,7 @@ Result KPageTableBase::MakePageGroup(KPageGroup& pg, KProcessAddress addr, size_
     // Begin traversal.
     TraversalContext context;
     TraversalEntry next_entry;
-    R_UNLESS(impl.BeginTraversal(std::addressof(next_entry), std::addressof(context), addr),
+    R_UNLESS(BeginTraversal(impl, std::addressof(next_entry), std::addressof(context), addr),
              ResultInvalidCurrentMemory);
 
     // Prepare tracking variables.
@@ -1764,7 +1794,7 @@ Result KPageTableBase::MakePageGroup(KPageGroup& pg, KProcessAddress addr, size_
 
     // Iterate, adding to group as we go.
     while (tot_size < size) {
-        R_UNLESS(impl.ContinueTraversal(std::addressof(next_entry), std::addressof(context)),
+        R_UNLESS(ContinueTraversal(impl, std::addressof(next_entry), std::addressof(context)),
                  ResultInvalidCurrentMemory);
 
         if (next_entry.phys_addr != (cur_addr + cur_size)) {
@@ -1828,7 +1858,7 @@ bool KPageTableBase::IsValidPageGroup(const KPageGroup& pg, KProcessAddress addr
     // Begin traversal.
     TraversalContext context;
     TraversalEntry next_entry;
-    if (!impl.BeginTraversal(std::addressof(next_entry), std::addressof(context), addr)) {
+    if (!BeginTraversal(impl, std::addressof(next_entry), std::addressof(context), addr)) {
         return false;
     }
 
@@ -1839,7 +1869,7 @@ bool KPageTableBase::IsValidPageGroup(const KPageGroup& pg, KProcessAddress addr
 
     // Iterate, comparing expected to actual.
     while (tot_size < size) {
-        if (!impl.ContinueTraversal(std::addressof(next_entry), std::addressof(context))) {
+        if (!ContinueTraversal(impl, std::addressof(next_entry), std::addressof(context))) {
             return false;
         }
 
@@ -1896,7 +1926,7 @@ Result KPageTableBase::GetContiguousMemoryRangeWithState(
     // Begin a traversal.
     TraversalContext context;
     TraversalEntry cur_entry = {.phys_addr = 0, .block_size = 0};
-    R_UNLESS(impl.BeginTraversal(std::addressof(cur_entry), std::addressof(context), address),
+    R_UNLESS(BeginTraversal(impl, std::addressof(cur_entry), std::addressof(context), address),
              ResultInvalidCurrentMemory);
 
     // Traverse until we have enough size or we aren't contiguous any more.
@@ -1905,7 +1935,7 @@ Result KPageTableBase::GetContiguousMemoryRangeWithState(
     for (contig_size =
              cur_entry.block_size - (GetInteger(phys_address) & (cur_entry.block_size - 1));
          contig_size < size; contig_size += cur_entry.block_size) {
-        if (!impl.ContinueTraversal(std::addressof(cur_entry), std::addressof(context))) {
+        if (!ContinueTraversal(impl, std::addressof(cur_entry), std::addressof(context))) {
             break;
         }
         if (cur_entry.phys_addr != phys_address + contig_size) {
@@ -2334,7 +2364,7 @@ Result KPageTableBase::QueryPhysicalAddress(Svc::lp64::PhysicalMemoryInfo* out,
         TraversalContext context;
         TraversalEntry next_entry;
         bool traverse_valid =
-            m_impl.BeginTraversal(std::addressof(next_entry), std::addressof(context), virt_addr);
+            BeginTraversal(m_impl, std::addressof(next_entry), std::addressof(context), virt_addr);
         R_UNLESS(traverse_valid, ResultInvalidCurrentMemory);
 
         // Set tracking variables.
@@ -2345,7 +2375,7 @@ Result KPageTableBase::QueryPhysicalAddress(Svc::lp64::PhysicalMemoryInfo* out,
         while (true) {
             // Continue the traversal.
             traverse_valid =
-                m_impl.ContinueTraversal(std::addressof(next_entry), std::addressof(context));
+                ContinueTraversal(m_impl, std::addressof(next_entry), std::addressof(context));
             if (!traverse_valid) {
                 break;
             }
@@ -2567,7 +2597,7 @@ Result KPageTableBase::UnmapIoRegion(KProcessAddress dst_address, KPhysicalAddre
         TraversalContext context;
         TraversalEntry next_entry;
         ASSERT(
-            impl.BeginTraversal(std::addressof(next_entry), std::addressof(context), dst_address));
+            BeginTraversal(impl, std::addressof(next_entry), std::addressof(context), dst_address));
 
         // Check that the physical region matches.
         R_UNLESS(next_entry.phys_addr == phys_addr, ResultInvalidMemoryRegion);
@@ -2577,7 +2607,7 @@ Result KPageTableBase::UnmapIoRegion(KProcessAddress dst_address, KPhysicalAddre
                  next_entry.block_size - (GetInteger(phys_addr) & (next_entry.block_size - 1));
              checked_size < size; checked_size += next_entry.block_size) {
             // Continue the traversal.
-            ASSERT(impl.ContinueTraversal(std::addressof(next_entry), std::addressof(context)));
+            ASSERT(ContinueTraversal(impl, std::addressof(next_entry), std::addressof(context)));
 
             // Check that the physical region matches.
             R_UNLESS(next_entry.phys_addr == phys_addr + checked_size, ResultInvalidMemoryRegion);
@@ -3029,7 +3059,7 @@ Result KPageTableBase::InvalidateProcessDataCache(KProcessAddress address, size_
     TraversalContext context;
     TraversalEntry next_entry;
     bool traverse_valid =
-        impl.BeginTraversal(std::addressof(next_entry), std::addressof(context), address);
+        BeginTraversal(impl, std::addressof(next_entry), std::addressof(context), address);
     R_UNLESS(traverse_valid, ResultInvalidCurrentMemory);
 
     // Prepare tracking variables.
@@ -3041,7 +3071,7 @@ Result KPageTableBase::InvalidateProcessDataCache(KProcessAddress address, size_
     while (tot_size < size) {
         // Continue the traversal.
         traverse_valid =
-            impl.ContinueTraversal(std::addressof(next_entry), std::addressof(context));
+            ContinueTraversal(impl, std::addressof(next_entry), std::addressof(context));
         R_UNLESS(traverse_valid, ResultInvalidCurrentMemory);
 
         if (next_entry.phys_addr != (cur_addr + cur_size)) {
@@ -3129,7 +3159,7 @@ Result KPageTableBase::ReadDebugMemory(KProcessAddress dst_address, KProcessAddr
     TraversalContext context;
     TraversalEntry next_entry;
     bool traverse_valid =
-        impl.BeginTraversal(std::addressof(next_entry), std::addressof(context), src_address);
+        BeginTraversal(impl, std::addressof(next_entry), std::addressof(context), src_address);
     R_UNLESS(traverse_valid, ResultInvalidCurrentMemory);
 
     // Prepare tracking variables.
@@ -3167,7 +3197,7 @@ Result KPageTableBase::ReadDebugMemory(KProcessAddress dst_address, KProcessAddr
     while (tot_size < size) {
         // Continue the traversal.
         traverse_valid =
-            impl.ContinueTraversal(std::addressof(next_entry), std::addressof(context));
+            ContinueTraversal(impl, std::addressof(next_entry), std::addressof(context));
         ASSERT(traverse_valid);
 
         if (next_entry.phys_addr != (cur_addr + cur_size)) {
@@ -3225,7 +3255,7 @@ Result KPageTableBase::WriteDebugMemory(KProcessAddress dst_address, KProcessAdd
     TraversalContext context;
     TraversalEntry next_entry;
     bool traverse_valid =
-        impl.BeginTraversal(std::addressof(next_entry), std::addressof(context), dst_address);
+        BeginTraversal(impl, std::addressof(next_entry), std::addressof(context), dst_address);
     R_UNLESS(traverse_valid, ResultInvalidCurrentMemory);
 
     // Prepare tracking variables.
@@ -3267,7 +3297,7 @@ Result KPageTableBase::WriteDebugMemory(KProcessAddress dst_address, KProcessAdd
     while (tot_size < size) {
         // Continue the traversal.
         traverse_valid =
-            impl.ContinueTraversal(std::addressof(next_entry), std::addressof(context));
+            ContinueTraversal(impl, std::addressof(next_entry), std::addressof(context));
         ASSERT(traverse_valid);
 
         if (next_entry.phys_addr != (cur_addr + cur_size)) {
@@ -3728,7 +3758,7 @@ Result KPageTableBase::CopyMemoryFromLinearToUser(
         TraversalContext context;
         TraversalEntry next_entry;
         bool traverse_valid =
-            impl.BeginTraversal(std::addressof(next_entry), std::addressof(context), src_addr);
+            BeginTraversal(impl, std::addressof(next_entry), std::addressof(context), src_addr);
         ASSERT(traverse_valid);
 
         // Prepare tracking variables.
@@ -3768,7 +3798,7 @@ Result KPageTableBase::CopyMemoryFromLinearToUser(
         while (tot_size < size) {
             // Continue the traversal.
             traverse_valid =
-                impl.ContinueTraversal(std::addressof(next_entry), std::addressof(context));
+                ContinueTraversal(impl, std::addressof(next_entry), std::addressof(context));
             ASSERT(traverse_valid);
 
             if (next_entry.phys_addr != (cur_addr + cur_size)) {
@@ -3822,7 +3852,7 @@ Result KPageTableBase::CopyMemoryFromLinearToKernel(
         TraversalContext context;
         TraversalEntry next_entry;
         bool traverse_valid =
-            impl.BeginTraversal(std::addressof(next_entry), std::addressof(context), src_addr);
+            BeginTraversal(impl, std::addressof(next_entry), std::addressof(context), src_addr);
         ASSERT(traverse_valid);
 
         // Prepare tracking variables.
@@ -3845,7 +3875,7 @@ Result KPageTableBase::CopyMemoryFromLinearToKernel(
         while (tot_size < size) {
             // Continue the traversal.
             traverse_valid =
-                impl.ContinueTraversal(std::addressof(next_entry), std::addressof(context));
+                ContinueTraversal(impl, std::addressof(next_entry), std::addressof(context));
             ASSERT(traverse_valid);
 
             if (next_entry.phys_addr != (cur_addr + cur_size)) {
@@ -3902,7 +3932,7 @@ Result KPageTableBase::CopyMemoryFromUserToLinear(
         TraversalContext context;
         TraversalEntry next_entry;
         bool traverse_valid =
-            impl.BeginTraversal(std::addressof(next_entry), std::addressof(context), dst_addr);
+            BeginTraversal(impl, std::addressof(next_entry), std::addressof(context), dst_addr);
         ASSERT(traverse_valid);
 
         // Prepare tracking variables.
@@ -3941,7 +3971,7 @@ Result KPageTableBase::CopyMemoryFromUserToLinear(
         while (tot_size < size) {
             // Continue the traversal.
             traverse_valid =
-                impl.ContinueTraversal(std::addressof(next_entry), std::addressof(context));
+                ContinueTraversal(impl, std::addressof(next_entry), std::addressof(context));
             ASSERT(traverse_valid);
 
             if (next_entry.phys_addr != (cur_addr + cur_size)) {
@@ -3997,7 +4027,7 @@ Result KPageTableBase::CopyMemoryFromKernelToLinear(KProcessAddress dst_addr, si
         TraversalContext context;
         TraversalEntry next_entry;
         bool traverse_valid =
-            impl.BeginTraversal(std::addressof(next_entry), std::addressof(context), dst_addr);
+            BeginTraversal(impl, std::addressof(next_entry), std::addressof(context), dst_addr);
         ASSERT(traverse_valid);
 
         // Prepare tracking variables.
@@ -4020,7 +4050,7 @@ Result KPageTableBase::CopyMemoryFromKernelToLinear(KProcessAddress dst_addr, si
         while (tot_size < size) {
             // Continue the traversal.
             traverse_valid =
-                impl.ContinueTraversal(std::addressof(next_entry), std::addressof(context));
+                ContinueTraversal(impl, std::addressof(next_entry), std::addressof(context));
             ASSERT(traverse_valid);
 
             if (next_entry.phys_addr != (cur_addr + cur_size)) {
@@ -4089,10 +4119,10 @@ Result KPageTableBase::CopyMemoryFromHeapToHeap(
         bool traverse_valid;
 
         // Begin traversal.
-        traverse_valid = src_impl.BeginTraversal(std::addressof(src_next_entry),
+        traverse_valid = BeginTraversal(src_impl, std::addressof(src_next_entry),
                                                  std::addressof(src_context), src_addr);
         ASSERT(traverse_valid);
-        traverse_valid = dst_impl.BeginTraversal(std::addressof(dst_next_entry),
+        traverse_valid = BeginTraversal(dst_impl, std::addressof(dst_next_entry),
                                                  std::addressof(dst_context), dst_addr);
         ASSERT(traverse_valid);
 
@@ -4127,7 +4157,7 @@ Result KPageTableBase::CopyMemoryFromHeapToHeap(
             if (ofs + cur_copy_size != size) {
                 if (cur_src_addr + cur_min_size == cur_src_block_addr + cur_src_size) {
                     // Continue the src traversal.
-                    traverse_valid = src_impl.ContinueTraversal(std::addressof(src_next_entry),
+                    traverse_valid = ContinueTraversal(src_impl, std::addressof(src_next_entry),
                                                                 std::addressof(src_context));
                     ASSERT(traverse_valid);
 
@@ -4138,7 +4168,7 @@ Result KPageTableBase::CopyMemoryFromHeapToHeap(
                 if (cur_dst_addr + cur_min_size ==
                     dst_next_entry.phys_addr + dst_next_entry.block_size) {
                     // Continue the dst traversal.
-                    traverse_valid = dst_impl.ContinueTraversal(std::addressof(dst_next_entry),
+                    traverse_valid = ContinueTraversal(dst_impl, std::addressof(dst_next_entry),
                                                                 std::addressof(dst_context));
                     ASSERT(traverse_valid);
 
@@ -4223,10 +4253,10 @@ Result KPageTableBase::CopyMemoryFromHeapToHeapWithoutCheckDestination(
         bool traverse_valid;
 
         // Begin traversal.
-        traverse_valid = src_impl.BeginTraversal(std::addressof(src_next_entry),
+        traverse_valid = BeginTraversal(src_impl, std::addressof(src_next_entry),
                                                  std::addressof(src_context), src_addr);
         ASSERT(traverse_valid);
-        traverse_valid = dst_impl.BeginTraversal(std::addressof(dst_next_entry),
+        traverse_valid = BeginTraversal(dst_impl, std::addressof(dst_next_entry),
                                                  std::addressof(dst_context), dst_addr);
         ASSERT(traverse_valid);
 
@@ -4261,7 +4291,7 @@ Result KPageTableBase::CopyMemoryFromHeapToHeapWithoutCheckDestination(
             if (ofs + cur_copy_size != size) {
                 if (cur_src_addr + cur_min_size == cur_src_block_addr + cur_src_size) {
                     // Continue the src traversal.
-                    traverse_valid = src_impl.ContinueTraversal(std::addressof(src_next_entry),
+                    traverse_valid = ContinueTraversal(src_impl, std::addressof(src_next_entry),
                                                                 std::addressof(src_context));
                     ASSERT(traverse_valid);
 
@@ -4272,7 +4302,7 @@ Result KPageTableBase::CopyMemoryFromHeapToHeapWithoutCheckDestination(
                 if (cur_dst_addr + cur_min_size ==
                     dst_next_entry.phys_addr + dst_next_entry.block_size) {
                     // Continue the dst traversal.
-                    traverse_valid = dst_impl.ContinueTraversal(std::addressof(dst_next_entry),
+                    traverse_valid = ContinueTraversal(dst_impl, std::addressof(dst_next_entry),
                                                                 std::addressof(dst_context));
                     ASSERT(traverse_valid);
 
@@ -4547,7 +4577,7 @@ Result KPageTableBase::SetupForIpcServer(KProcessAddress* out_addr, size_t size,
     // Begin traversal.
     TraversalContext context;
     TraversalEntry next_entry;
-    bool traverse_valid = src_impl.BeginTraversal(std::addressof(next_entry),
+    bool traverse_valid = BeginTraversal(src_impl, std::addressof(next_entry),
                                                   std::addressof(context), aligned_src_start);
     ASSERT(traverse_valid);
 
@@ -4597,7 +4627,7 @@ Result KPageTableBase::SetupForIpcServer(KProcessAddress* out_addr, size_t size,
         // If the block's size was one page, we may need to continue traversal.
         if (cur_block_size == 0 && aligned_src_size > PageSize) {
             traverse_valid =
-                src_impl.ContinueTraversal(std::addressof(next_entry), std::addressof(context));
+                ContinueTraversal(src_impl, std::addressof(next_entry), std::addressof(context));
             ASSERT(traverse_valid);
 
             cur_block_addr = next_entry.phys_addr;
@@ -4610,7 +4640,7 @@ Result KPageTableBase::SetupForIpcServer(KProcessAddress* out_addr, size_t size,
     while (aligned_src_start + tot_block_size < mapping_src_end) {
         // Continue the traversal.
         traverse_valid =
-            src_impl.ContinueTraversal(std::addressof(next_entry), std::addressof(context));
+            ContinueTraversal(src_impl, std::addressof(next_entry), std::addressof(context));
         ASSERT(traverse_valid);
 
         // Process the block.
@@ -4653,7 +4683,7 @@ Result KPageTableBase::SetupForIpcServer(KProcessAddress* out_addr, size_t size,
         if (mapped_block_end + cur_block_size < aligned_src_end &&
             cur_block_size == last_block_size) {
             traverse_valid =
-                src_impl.ContinueTraversal(std::addressof(next_entry), std::addressof(context));
+                ContinueTraversal(src_impl, std::addressof(next_entry), std::addressof(context));
             ASSERT(traverse_valid);
 
             cur_block_addr = next_entry.phys_addr;
@@ -5601,7 +5631,7 @@ Result KPageTableBase::UnmapProcessMemory(KProcessAddress dst_address, size_t si
             ContiguousRangeInfo(KPageTableBase& pt, KProcessAddress address, size_t size)
                 : m_pt(pt), m_remaining_size(size) {
                 // Begin a traversal.
-                ASSERT(m_pt.GetImpl().BeginTraversal(std::addressof(m_entry),
+                ASSERT(m_pt.BeginTraversal(m_pt.GetImpl(), std::addressof(m_entry),
                                                      std::addressof(m_context), address));
 
                 // Setup tracking fields.
@@ -5632,7 +5662,7 @@ Result KPageTableBase::UnmapProcessMemory(KProcessAddress dst_address, size_t si
             void DetermineContiguousBlockExtents() {
                 // Continue traversing until we're not contiguous, or we have enough.
                 while (m_cur_size < m_remaining_size) {
-                    ASSERT(m_pt.GetImpl().ContinueTraversal(std::addressof(m_entry),
+                    ASSERT(m_pt.ContinueTraversal(m_pt.GetImpl(), std::addressof(m_entry),
                                                             std::addressof(m_context)));
 
                     // If we're not contiguous, we're done.

src/core/hle/kernel/k_page_table_base.h

@@ -370,6 +370,10 @@ private:
                                  size_t num_pages, size_t alignment, size_t offset,
                                  size_t guard_pages) const;
 
+    bool BeginTraversal(const Common::PageTable& impl, TraversalEntry* out_entry, TraversalContext* out_context,
+                        Common::ProcessAddress address) const;
+    bool ContinueTraversal(const Common::PageTable& impl, TraversalEntry* out_entry, TraversalContext* context) const;
+
     Result CheckMemoryStateContiguous(size_t* out_blocks_needed, KProcessAddress addr, size_t size,
                                       KMemoryState state_mask, KMemoryState state,
                                       KMemoryPermission perm_mask, KMemoryPermission perm,
@@ -474,7 +478,14 @@ private:
         // Validate pre-conditions.
         ASSERT(this->IsLockedByCurrentThread());
 
-        return this->GetImpl().GetPhysicalAddress(out, virt_addr);
+        if (virt_addr > (1ULL << m_address_space_width)) {
+            return false;
+        }
+
+        *out = m_system.DeviceMemory().GetPhysicalAddr(
+            this->GetImpl().entries[GetInteger(virt_addr) >> PageBits].Pointer(true) + GetInteger(virt_addr));
+
+        return true;
     }
 
 public:

src/core/hle/kernel/svc.cpp

@@ -1,4 +1,4 @@
-// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
 // SPDX-License-Identifier: GPL-3.0-or-later
 // SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-late
@@ -2501,14 +2501,15 @@ void Call(Core::System& system, u32 imm) {
     auto& process = GetCurrentProcess(kernel);
     std::array<uint64_t, 8> args;
     kernel.CurrentPhysicalCore().SaveSvcArguments(process, args);
-    LOG_TRACE(Kernel_SVC, "{} [0]={:#x} [1]={:#x} [2]={:#x} [3]={:#x} [4]={:#x} [5]={:#x} [6]={:#x}",
-        imm, GetArg32(args, 0), GetArg32(args, 1), GetArg32(args, 2),
-        GetArg32(args, 3), GetArg32(args, 4), GetArg32(args, 5), GetArg32(args, 6));
     //kernel.EnterSVCProfile();
-    if (process.Is64Bit())
+    LOG_TRACE(Kernel_SVC, "{} [0]={:#x} [1]={:#x} [2]={:#x} [3]={:#x} [4]={:#x} [5]={:#x} [6]={:#x}",
+        imm, GetArg64(args, 0), GetArg64(args, 1), GetArg64(args, 2),
+        GetArg64(args, 3), GetArg64(args, 4), GetArg64(args, 5), GetArg64(args, 6));
+    if (process.Is64Bit()) {
         Call64(system, imm, args);
-    else
+    } else {
         Call32(system, imm, args);
+    }
     //kernel.ExitSVCProfile();
     kernel.CurrentPhysicalCore().LoadSvcArguments(process, args);
 }

src/core/hle/kernel/svc/svc_info.cpp

@@ -10,17 +10,15 @@
 #include "core/hle/kernel/k_process.h"
 #include "core/hle/kernel/k_resource_limit.h"
 #include "core/hle/kernel/svc.h"
+#include "core/hle/kernel/svc_results.h"
+#include "core/hle/kernel/svc_version.h"
 
 namespace Kernel::Svc {
 
 /// Gets system/memory information for the current process
-Result GetInfo(Core::System& system, u64* result, InfoType info_id_type, Handle handle,
-               u64 info_sub_id) {
-    LOG_TRACE(Kernel_SVC, "called info_id={:#x}, info_sub_id={:#x}, handle={:#08x}",
-              info_id_type, info_sub_id, handle);
-
-    u32 info_id = static_cast<u32>(info_id_type);
-
+Result GetInfo(Core::System& system, u64* result, InfoType info_id_type, Handle handle, u64 info_sub_id) {
+    LOG_TRACE(Kernel_SVC, "called info_id={:#x}, info_sub_id={:#x}, handle={:#08x}", info_id_type, info_sub_id, handle);
+    u32 info_id = u32(info_id_type);
     switch (info_id_type) {
     case InfoType::CoreMask:
     case InfoType::PriorityMask:
@@ -53,152 +51,123 @@ Result GetInfo(Core::System& system, u64* result, InfoType info_id_type, Handle
         case InfoType::CoreMask:
             *result = process->GetCoreMask();
             R_SUCCEED();
-
         case InfoType::PriorityMask:
             *result = process->GetPriorityMask();
             R_SUCCEED();
-
         case InfoType::AliasRegionAddress:
             *result = GetInteger(process->GetPageTable().GetAliasRegionStart());
             R_SUCCEED();
-
         case InfoType::AliasRegionSize:
             *result = process->GetPageTable().GetAliasRegionSize();
             R_SUCCEED();
         case InfoType::HeapRegionAddress:
             *result = GetInteger(process->GetPageTable().GetHeapRegionStart());
             R_SUCCEED();
-
         case InfoType::HeapRegionSize:
             *result = process->GetPageTable().GetHeapRegionSize();
             R_SUCCEED();
-
         case InfoType::AslrRegionAddress:
             *result = GetInteger(process->GetPageTable().GetAliasCodeRegionStart());
             R_SUCCEED();
-
         case InfoType::AslrRegionSize:
             *result = process->GetPageTable().GetAliasCodeRegionSize();
             R_SUCCEED();
-
         case InfoType::StackRegionAddress:
             *result = GetInteger(process->GetPageTable().GetStackRegionStart());
             R_SUCCEED();
-
         case InfoType::StackRegionSize:
             *result = process->GetPageTable().GetStackRegionSize();
             R_SUCCEED();
-
         case InfoType::TotalMemorySize:
             *result = process->GetTotalUserPhysicalMemorySize(system.Kernel());
             R_SUCCEED();
-
         case InfoType::UsedMemorySize:
             *result = process->GetUsedUserPhysicalMemorySize(system.Kernel());
             R_SUCCEED();
-
         case InfoType::SystemResourceSizeTotal:
             *result = process->GetTotalSystemResourceSize();
             R_SUCCEED();
-
         case InfoType::SystemResourceSizeUsed:
             *result = process->GetUsedSystemResourceSize();
             R_SUCCEED();
-
         case InfoType::ProgramId:
             *result = process->GetProgramId();
             R_SUCCEED();
-
         case InfoType::UserExceptionContextAddress:
             *result = GetInteger(process->GetProcessLocalRegionAddress());
             R_SUCCEED();
-
         case InfoType::TotalNonSystemMemorySize:
             *result = process->GetTotalNonSystemUserPhysicalMemorySize(system.Kernel());
             R_SUCCEED();
-
         case InfoType::UsedNonSystemMemorySize:
             *result = process->GetUsedNonSystemUserPhysicalMemorySize(system.Kernel());
             R_SUCCEED();
-
         case InfoType::IsApplication:
             *result = process->IsApplication();
             R_SUCCEED();
-
         case InfoType::FreeThreadCount:
             if (KResourceLimit* resource_limit = process->GetResourceLimit();
                 resource_limit != nullptr) {
-                const auto current_value =
-                    resource_limit->GetCurrentValue(Svc::LimitableResource::ThreadCountMax);
-                const auto limit_value =
-                    resource_limit->GetLimitValue(Svc::LimitableResource::ThreadCountMax);
+                const auto current_value = resource_limit->GetCurrentValue(Svc::LimitableResource::ThreadCountMax);
+                const auto limit_value = resource_limit->GetLimitValue(Svc::LimitableResource::ThreadCountMax);
                 *result = limit_value - current_value;
             } else {
                 *result = 0;
             }
             R_SUCCEED();
-
         case InfoType::AliasRegionExtraSize: {
-            if (info_sub_id != 0) {
-                return ResultInvalidCombination;
-            }
-
+            R_UNLESS(info_sub_id == 0, ResultInvalidCombination);
             KProcess* current_process = GetCurrentProcessPointer(system.Kernel());
             *result = current_process->GetPageTable().GetAliasRegionExtraSize();
-
             R_SUCCEED();
         }
-
         default:
             break;
         }
-
         LOG_ERROR(Kernel_SVC, "Unimplemented svcGetInfo id={:#016x}", info_id);
         R_THROW(ResultInvalidEnumValue);
     }
-
     case InfoType::DebuggerAttached:
         *result = 0;
         R_SUCCEED();
-
     case InfoType::ResourceLimit: {
         R_UNLESS(handle == 0, ResultInvalidHandle);
         R_UNLESS(info_sub_id == 0, ResultInvalidCombination);
-
         KProcess* const current_process = GetCurrentProcessPointer(system.Kernel());
         KHandleTable& handle_table = current_process->GetHandleTable();
-        const auto resource_limit = current_process->GetResourceLimit();
-        if (!resource_limit) {
+        if (auto const resource_limit = current_process->GetResourceLimit(); resource_limit) {
+            Handle resource_handle{};
+            R_TRY(handle_table.Add(system.Kernel(), std::addressof(resource_handle), resource_limit));
+            *result = resource_handle;
+        } else {
             *result = Svc::InvalidHandle;
-            // Yes, the kernel considers this a successful operation.
-            R_SUCCEED();
         }
-
-        Handle resource_handle{};
-        R_TRY(handle_table.Add(system.Kernel(), std::addressof(resource_handle), resource_limit));
-
-        *result = resource_handle;
+        // Yes, the kernel considers this a successful operation either way.
         R_SUCCEED();
     }
-
     case InfoType::RandomEntropy:
         R_UNLESS(handle == 0, ResultInvalidHandle);
         R_UNLESS(info_sub_id < 4, ResultInvalidCombination);
-
         *result = GetCurrentProcess(system.Kernel()).GetRandomEntropy(info_sub_id);
         R_SUCCEED();
-
-    case InfoType::InitialProcessIdRange:
-        LOG_WARNING(Kernel_SVC,
-                    "(STUBBED) Attempted to query privileged process id bounds, returned 0");
-        *result = 0;
-        R_SUCCEED();
-
+    case InfoType::InitialProcessIdRange: {
+        LOG_WARNING(Kernel_SVC, "(STUBBED) Attempted to query privileged process id bounds, returned 0/64");
+        R_UNLESS(handle == InvalidHandle, ResultInvalidHandle);
+        switch (InitialProcessIdRangeInfo(info_sub_id)) {
+        case InitialProcessIdRangeInfo::Minimum:
+            *result = 0; //todo
+            R_SUCCEED();
+        case InitialProcessIdRangeInfo::Maximum:
+            *result = 64; //todo
+            R_SUCCEED();
+        default:
+            R_THROW(ResultInvalidCombination);
+        }
+    }
     case InfoType::ThreadTickCount: {
         constexpr u64 num_cpus = 4;
         if (info_sub_id != 0xFFFFFFFFFFFFFFFF && info_sub_id >= num_cpus) {
-            LOG_ERROR(Kernel_SVC, "Core count is out of range, expected {} but got {}", num_cpus,
-                      info_sub_id);
+            LOG_ERROR(Kernel_SVC, "Core count is out of range, expected {} but got {}", num_cpus, info_sub_id);
             R_THROW(ResultInvalidCombination);
         }
 
@@ -206,8 +175,7 @@ Result GetInfo(Core::System& system, u64* result, InfoType info_id_type, Handle
             .GetHandleTable()
             .GetObject<KThread>(system.Kernel(), Handle(handle));
         if (thread.IsNull()) {
-            LOG_ERROR(Kernel_SVC, "Thread handle does not exist, handle={:#08x}",
-                      static_cast<Handle>(handle));
+            LOG_ERROR(Kernel_SVC, "Thread handle does not exist, handle={:#08x}", Handle(handle));
             R_THROW(ResultInvalidHandle);
         }
 
@@ -220,7 +188,6 @@ Result GetInfo(Core::System& system, u64* result, InfoType info_id_type, Handle
         u64 out_ticks = 0;
         if (same_thread && info_sub_id == 0xFFFFFFFFFFFFFFFF) {
             const u64 thread_ticks = current_thread->GetCpuTime();
-
             out_ticks = thread_ticks + (core_timing.GetClockTicks() - prev_ctx_ticks);
         } else if (same_thread && info_sub_id == system.Kernel().CurrentPhysicalCoreIndex()) {
             out_ticks = core_timing.GetClockTicks() - prev_ctx_ticks;
@@ -230,19 +197,40 @@ Result GetInfo(Core::System& system, u64* result, InfoType info_id_type, Handle
         R_SUCCEED();
     }
     case InfoType::IdleTickCount: {
-        // Verify the input handle is invalid.
-        R_UNLESS(handle == InvalidHandle, ResultInvalidHandle);
-
         // Verify the requested core is valid.
         const bool core_valid =
-            (info_sub_id == 0xFFFFFFFFFFFFFFFF) ||
-            (info_sub_id == static_cast<u64>(system.Kernel().CurrentPhysicalCoreIndex()));
+            (info_sub_id == 0xFFFFFFFFFFFFFFFF)
+            || (info_sub_id == u64(system.Kernel().CurrentPhysicalCoreIndex()));
         R_UNLESS(core_valid, ResultInvalidCombination);
 
+        // Verify the input handle is invalid.
+        R_UNLESS(handle == InvalidHandle, ResultInvalidHandle);
+
         // Get the idle tick count.
         *result = system.Kernel().CurrentScheduler()->GetIdleThread()->GetCpuTime();
         R_SUCCEED();
     }
+    case InfoType::MesosphereMeta: {
+        enum MesosphereMetaInfo : u64 {
+            KernelVersion = 0,
+            IsKTraceEnabled = 1,
+            IsSingleStepEnabled = 2,
+        };
+        R_UNLESS(handle == InvalidHandle, ResultInvalidHandle);
+        switch (MesosphereMetaInfo(info_sub_id)) {
+        case MesosphereMetaInfo::KernelVersion:
+            *result = Kernel::Svc::SupportedKernelVersion;
+            R_SUCCEED();
+        case MesosphereMetaInfo::IsKTraceEnabled:
+            *result = 0;
+            R_SUCCEED();
+        case MesosphereMetaInfo::IsSingleStepEnabled:
+            *result = 0;
+            R_SUCCEED();
+        default:
+            R_THROW(ResultInvalidCombination);
+        }
+    }
     case InfoType::MesosphereCurrentProcess: {
         // Verify the input handle is invalid.
         R_UNLESS(handle == InvalidHandle, ResultInvalidHandle);
@@ -255,13 +243,11 @@ Result GetInfo(Core::System& system, u64* result, InfoType info_id_type, Handle
         KHandleTable& handle_table = current_process->GetHandleTable();
 
         // Get a new handle for the current process.
-        Handle tmp;
+        Handle tmp{};
         R_TRY(handle_table.Add(system.Kernel(), std::addressof(tmp), current_process));
 
         // Set the output.
         *result = tmp;
-
-        // We succeeded.
         R_SUCCEED();
     }
     default:

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

@@ -1387,7 +1387,8 @@ public:
             {5, &ACC_U1::GetProfile, "GetProfile"},
             {6, nullptr, "GetProfileDigest"},
             {50, &ACC_U1::IsUserRegistrationRequestPermitted, "IsUserRegistrationRequestPermitted"},
-            {51, &ACC_U1::TrySelectUserWithoutInteraction, "TrySelectUserWithoutInteraction"},
+            {51, &ACC_U1::TrySelectUserWithoutInteractionDeprecated, "TrySelectUserWithoutInteractionDeprecated"},
+            {52, &ACC_U1::TrySelectUserWithoutInteraction, "TrySelectUserWithoutInteraction"},
             {60, &ACC_U1::ListOpenContextStoredUsers, "ListOpenContextStoredUsers"},
             {99, nullptr, "DebugActivateOpenContextRetention"},
             {100, nullptr, "GetUserRegistrationNotifier"},

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

@@ -154,6 +154,7 @@ enum class AppletMessage : u32 {
     DetectLongPressingCaptureButton = 91,
     AlbumScreenShotTaken = 92,
     AlbumRecordingSaved = 93,
+    StartupLogoDisappeared = 95,
 };
 
 enum class LibraryAppletMode : u32 {

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

@@ -91,6 +91,7 @@ struct Applet {
     // Common state
     bool sleep_lock_enabled{};
     bool vr_mode_enabled{};
+    bool vr_mode_enabled_3d{};
     bool lcd_backlight_off_enabled{};
     APM::CpuBoostMode boost_mode{};
     bool request_exit_to_library_applet_at_execute_next_program_enabled{};

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

@@ -73,20 +73,16 @@ Result DisplayLayerManager::CreateManagedDisplayLayer(u64* out_layer_id) {
     R_TRY(m_manager_display_service->CreateManagedLayer(
         out_layer_id, 0, display_id, Service::AppletResourceUserId{m_process->GetProcessId()}));
 
-    m_manager_display_service->SetLayerVisibility(m_visible, *out_layer_id);
-    (void)m_display_service->GetContainer()->SetLayerStackMask(*out_layer_id,
-                                                              this->GetLayerStackMask());
-
     if (m_applet_id != AppletId::Application) {
         (void)m_manager_display_service->SetLayerBlending(m_blending_enabled, *out_layer_id);
         if (m_applet_id == AppletId::OverlayDisplay) {
-            (void)m_manager_display_service->SetLayerZIndex(Overlay, *out_layer_id);
+            (void)m_manager_display_service->SetLayerZIndex(-1, *out_layer_id);
             (void)m_display_service->GetContainer()->SetLayerIsOverlay(*out_layer_id, true);
         } else {
-            (void)m_manager_display_service->SetLayerZIndex(Foreground, *out_layer_id);
+            (void)m_manager_display_service->SetLayerZIndex(1, *out_layer_id);
         }
     }
-
+    (void)m_display_service->GetContainer()->SetLayerZIndex(*out_layer_id, true);
     m_managed_display_layers.emplace(*out_layer_id);
 
     R_SUCCEED();
@@ -126,12 +122,11 @@ Result DisplayLayerManager::IsSystemBufferSharingEnabled() {
 
     // Ensure the overlay layer is visible
     m_manager_display_service->SetLayerVisibility(m_visible, m_system_shared_layer_id);
-    (void)m_display_service->GetContainer()->SetLayerStackMask(m_system_shared_layer_id,
-                                                              this->GetLayerStackMask());
     m_manager_display_service->SetLayerBlending(m_blending_enabled, m_system_shared_layer_id);
-    s32 initial_z = Foreground;
+    s32 initial_z = 1;
+    (void)m_display_service->GetContainer()->SetLayerZIndex(m_system_shared_layer_id, true);
     if (m_applet_id == AppletId::OverlayDisplay) {
-        initial_z = Overlay;
+        initial_z = -1;
         (void)m_display_service->GetContainer()->SetLayerIsOverlay(m_system_shared_layer_id, true);
     }
     m_manager_display_service->SetLayerZIndex(initial_z, m_system_shared_layer_id);

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

@@ -1,4 +1,4 @@
-// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
 // SPDX-License-Identifier: GPL-3.0-or-later
 
 // SPDX-FileCopyrightText: Copyright 2018 yuzu Emulator Project
@@ -73,8 +73,7 @@ Result IAllSystemAppletProxiesService::OpenLibraryAppletProxy(
 
 Result IAllSystemAppletProxiesService::OpenOverlayAppletProxy(
     Out<SharedPointer<IOverlayAppletProxy>> out_overlay_applet_proxy, ClientProcessId pid,
-    InCopyHandle<Kernel::KProcess> process_handle,
-    InLargeData<AppletAttribute, BufferAttr_HipcMapAlias> attribute) {
+    InCopyHandle<Kernel::KProcess> process_handle) {
     LOG_WARNING(Service_AM, "called");
 
     if (const auto applet = this->GetAppletFromProcessId(pid); applet) {
@@ -89,8 +88,7 @@ Result IAllSystemAppletProxiesService::OpenOverlayAppletProxy(
 
 Result IAllSystemAppletProxiesService::OpenSystemApplicationProxy(
     Out<SharedPointer<IApplicationProxy>> out_system_application_proxy, ClientProcessId pid,
-    InCopyHandle<Kernel::KProcess> process_handle,
-    InLargeData<AppletAttribute, BufferAttr_HipcMapAlias> attribute) {
+    InCopyHandle<Kernel::KProcess> process_handle) {
     LOG_DEBUG(Service_AM, "called");
 
     if (const auto applet = this->GetAppletFromProcessId(pid); applet) {

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

@@ -1,4 +1,4 @@
-// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
 // SPDX-License-Identifier: GPL-3.0-or-later
 
 // SPDX-FileCopyrightText: Copyright 2018 yuzu Emulator Project
@@ -36,15 +36,13 @@ private:
                                   InCopyHandle<Kernel::KProcess> process_handle,
                                   InLargeData<AppletAttribute, BufferAttr_HipcMapAlias> attribute);
     Result OpenOverlayAppletProxy(Out<SharedPointer<IOverlayAppletProxy>> out_overlay_applet_proxy,
-                           ClientProcessId pid, InCopyHandle<Kernel::KProcess> process_handle,
-                           InLargeData<AppletAttribute, BufferAttr_HipcMapAlias> attribute);
+                           ClientProcessId pid, InCopyHandle<Kernel::KProcess> process_handle);
     Result OpenLibraryAppletProxyOld(
         Out<SharedPointer<ILibraryAppletProxy>> out_library_applet_proxy, ClientProcessId pid,
         InCopyHandle<Kernel::KProcess> process_handle);
     Result OpenSystemApplicationProxy(
         Out<SharedPointer<IApplicationProxy>> out_system_application_proxy, ClientProcessId pid,
-        InCopyHandle<Kernel::KProcess> process_handle,
-        InLargeData<AppletAttribute, BufferAttr_HipcMapAlias> attribute);
+        InCopyHandle<Kernel::KProcess> process_handle);
     Result GetSystemProcessCommonFunctions();
     Result GetAppletAlternativeFunctions();
 

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

@@ -39,7 +39,7 @@ ICommonStateGetter::ICommonStateGetter(Core::System& system_, std::shared_ptr<Ap
         {14, nullptr, "GetWakeupCount"}, //11.0.0+
         {15, nullptr, "Unknown15"}, //19.0.0+
         {20, D<&ICommonStateGetter::PushToGeneralChannel>, "PushToGeneralChannel"},
-        {30, nullptr, "GetHomeButtonReaderLockAccessor"},
+        {30, D<&ICommonStateGetter::GetHomeButtonReaderLockAccessor>, "GetHomeButtonReaderLockAccessor"},
         {31, D<&ICommonStateGetter::GetReaderLockAccessorEx>, "GetReaderLockAccessorEx"}, //2.0.0+
         {32, D<&ICommonStateGetter::GetWriterLockAccessorEx>, "GetWriterLockAccessorEx"}, //7.0.0+
         {40, nullptr, "GetCradleFwVersion"}, //2.0.0+
@@ -65,7 +65,7 @@ ICommonStateGetter::ICommonStateGetter(Core::System& system_, std::shared_ptr<Ap
         {100, D<&ICommonStateGetter::SetHandlingHomeButtonShortPressedEnabled>, "SetHandlingHomeButtonShortPressedEnabled"},
         {110, nullptr, "OpenMyGpuErrorHandler"},
         {120, D<&ICommonStateGetter::GetAppletLaunchedHistory>, "GetAppletLaunchedHistory"}, //13.0.0+
-        {130, nullptr, "Unknown130"}, //21.0.0+
+        {130, D<&ICommonStateGetter::EnableStartupLogoDisappearedMessage>, "EnableStartupLogoDisappearedMessage"}, //21.0.0+
         {200, D<&ICommonStateGetter::GetOperationModeSystemInfo>, "GetOperationModeSystemInfo"},
         {300, D<&ICommonStateGetter::GetSettingsPlatformRegion>, "GetSettingsPlatformRegion"},
         {400, nullptr, "ActivateMigrationService"},
@@ -74,14 +74,17 @@ ICommonStateGetter::ICommonStateGetter(Core::System& system_, std::shared_ptr<Ap
         {501, nullptr, "SuppressDisablingSleepTemporarily"},
         {502, nullptr, "IsSleepEnabled"},
         {503, nullptr, "IsDisablingSleepSuppressed"},
-        {600, nullptr, "Unknown600"}, //20.0.0+
+        {600, nullptr, "SetHidInputMagnificationForApplication"}, //20.0.0+
         {610, D<&ICommonStateGetter::Unknown610>, "Unknown610"}, //21.0.0+
         {611, D<&ICommonStateGetter::Unknown611>, "Unknown611"}, //22.0.0+
         {900, D<&ICommonStateGetter::SetRequestExitToLibraryAppletAtExecuteNextProgramEnabled>, "SetRequestExitToLibraryAppletAtExecuteNextProgramEnabled"}, //11.0.0+
         {910, nullptr, "GetLaunchRequiredTick"}, //17.0.0+
-        {1000, nullptr, "BeginVrMode3d"}, //19.0.0+
-        {1001, nullptr, "EndVrMode3d"}, //19.0.0+
-        {1002, nullptr, "IsVrModeEnabled3d"}, //19.0.0+
+        {1000, D<&ICommonStateGetter::BeginVrMode3d>, "BeginVrMode3d"}, //19.0.0+
+        {1001, D<&ICommonStateGetter::EndVrMode3d>, "EndVrMode3d"}, //19.0.0+
+        {1002, D<&ICommonStateGetter::IsVrModeEnabled3d>, "IsVrModeEnabled3d"}, //19.0.0+
+        {1003, D<&ICommonStateGetter::GetVrLaboGoggleViewport>, "GetVrLaboGoggleViewport"}, //21.0.0+
+        {1004, D<&ICommonStateGetter::GetPanelPhysicalSizeForSpecificTitle>, "GetPanelPhysicalSizeForSpecificTitle"}, //21.0.0+
+        {1005, D<&ICommonStateGetter::GetPanelResolutionForSpecificTitle>, "GetPanelResolutionForSpecificTitle"}, //21.0.0+
     };
     // clang-format on
 
@@ -311,6 +314,12 @@ Result ICommonStateGetter::PerformSystemButtonPressingIfInFocus(SystemButtonType
     R_SUCCEED();
 }
 
+Result ICommonStateGetter::EnableStartupLogoDisappearedMessage() {
+    LOG_WARNING(Service_AM, "(STUBBED) called");
+    m_applet->lifecycle_manager.PushUnorderedMessage(system.Kernel(), AppletMessage::StartupLogoDisappeared);
+    R_SUCCEED();
+}
+
 Result ICommonStateGetter::GetOperationModeSystemInfo(Out<u32> out_operation_mode_system_info) {
     LOG_WARNING(Service_AM, "(STUBBED) called");
     *out_operation_mode_system_info = 0;
@@ -355,6 +364,12 @@ Result ICommonStateGetter::PushToGeneralChannel(SharedPointer<IStorage> storage)
     R_SUCCEED();
 }
 
+Result ICommonStateGetter::GetHomeButtonReaderLockAccessor(Out<SharedPointer<ILockAccessor>> out_lock_accessor) {
+    LOG_DEBUG(Service_AM, "called");
+    *out_lock_accessor = std::make_shared<ILockAccessor>(system);
+    R_SUCCEED();
+}
+
 Result ICommonStateGetter::SetHandlingHomeButtonShortPressedEnabled(bool enabled) {
     LOG_DEBUG(Service_AM, "called, enabled={} applet_id={}", enabled, m_applet->applet_id);
 
@@ -363,14 +378,58 @@ Result ICommonStateGetter::SetHandlingHomeButtonShortPressedEnabled(bool enabled
     R_SUCCEED();
 }
 
-Result ICommonStateGetter::Unknown610() {
+Result ICommonStateGetter::Unknown610(u64 unk) {
     LOG_WARNING(Service_AM, "(STUBBED) called");
     R_SUCCEED();
 }
 
-Result ICommonStateGetter::Unknown611() {
+Result ICommonStateGetter::Unknown611(u8 unk) {
     LOG_WARNING(Service_AM, "(STUBBED) called");
     R_SUCCEED();
 }
 
+Result ICommonStateGetter::BeginVrMode3d() {
+    std::scoped_lock lk{m_applet->lock};
+    m_applet->vr_mode_enabled_3d = true;
+    LOG_WARNING(Service_AM, "VR Mode is {}", m_applet->vr_mode_enabled_3d ? "on" : "off");
+    R_SUCCEED();
+}
+
+Result ICommonStateGetter::EndVrMode3d() {
+    std::scoped_lock lk{m_applet->lock};
+    m_applet->vr_mode_enabled_3d = false;
+    LOG_WARNING(Service_AM, "VR Mode is {}", m_applet->vr_mode_enabled_3d ? "on" : "off");
+    R_SUCCEED();
+}
+
+Result ICommonStateGetter::IsVrModeEnabled3d(Out<bool> out_is_vr_mode_enabled_3d) {
+    LOG_WARNING(Service_AM, "(STUBBED) called");
+    std::scoped_lock lk{m_applet->lock};
+    *out_is_vr_mode_enabled_3d = m_applet->vr_mode_enabled_3d;
+    R_SUCCEED();
+}
+
+Result ICommonStateGetter::GetVrLaboGoggleViewport(Out<s32> out_x, Out<s32> out_y, Out<s32> out_width, Out<s32> out_height) {
+    LOG_WARNING(Service_AM, "(STUBBED) called");
+    *out_x = 0;
+    *out_y = 0;
+    *out_width = 1280;
+    *out_height = 720;
+    R_SUCCEED();
+}
+
+Result ICommonStateGetter::GetPanelPhysicalSizeForSpecificTitle(Out<f32> out_width, Out<f32> out_height) {
+    LOG_WARNING(Service_AM, "(STUBBED) called");
+    *out_width = 137250.0f / 1000.0f;
+    *out_height = 77200.0f / 1000.0f;
+    R_SUCCEED();
+}
+
+Result ICommonStateGetter::GetPanelResolutionForSpecificTitle(Out<s32> out_width, Out<s32> out_height) {
+    LOG_WARNING(Service_AM, "(STUBBED) called");
+    *out_width = 1280;
+    *out_height = 720;
+    R_SUCCEED();
+}
+
 } // namespace Service::AM

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

@@ -56,15 +56,23 @@ private:
     Result GetDefaultDisplayResolution(Out<s32> out_width, Out<s32> out_height);
     Result GetBuiltInDisplayType(Out<s32> out_display_type);
     Result PerformSystemButtonPressingIfInFocus(SystemButtonType type);
+    Result EnableStartupLogoDisappearedMessage();
     Result GetOperationModeSystemInfo(Out<u32> out_operation_mode_system_info);
     Result GetAppletLaunchedHistory(Out<s32> out_count,
                                     OutArray<AppletId, BufferAttr_HipcMapAlias> out_applet_ids);
     Result GetSettingsPlatformRegion(Out<Set::PlatformRegion> out_settings_platform_region);
     Result SetRequestExitToLibraryAppletAtExecuteNextProgramEnabled();
     Result PushToGeneralChannel(SharedPointer<IStorage> storage); // cmd 20
+    Result GetHomeButtonReaderLockAccessor(Out<SharedPointer<ILockAccessor>> out_lock_accessor);
     Result SetHandlingHomeButtonShortPressedEnabled(bool enabled);
-    Result Unknown610();
-    Result Unknown611();
+    Result Unknown610(u64 unk);
+    Result Unknown611(u8 unk);
+    Result BeginVrMode3d();
+    Result EndVrMode3d();
+    Result IsVrModeEnabled3d(Out<bool> out_is_vr_mode_enabled_3d);
+    Result GetVrLaboGoggleViewport(Out<s32> out_x, Out<s32> out_y, Out<s32> out_width, Out<s32> out_height);
+    Result GetPanelPhysicalSizeForSpecificTitle(Out<f32> out_width, Out<f32> out_height);
+    Result GetPanelResolutionForSpecificTitle(Out<s32> out_width, Out<s32> out_height);
 
     void SetCpuBoostMode(HLERequestContext& ctx);
 

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

@@ -1,3 +1,6 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
 // SPDX-FileCopyrightText: Copyright 2024 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
@@ -26,17 +29,14 @@ IStorage::~IStorage() = default;
 
 Result IStorage::Open(Out<SharedPointer<IStorageAccessor>> out_storage_accessor) {
     LOG_DEBUG(Service_AM, "called");
-
     R_UNLESS(m_impl->GetHandle() == nullptr, AM::ResultInvalidStorageType);
-
     *out_storage_accessor = std::make_shared<IStorageAccessor>(system, m_impl);
     R_SUCCEED();
 }
 
-Result IStorage::OpenTransferStorage(
-    Out<SharedPointer<ITransferStorageAccessor>> out_transfer_storage_accessor) {
+Result IStorage::OpenTransferStorage(Out<SharedPointer<ITransferStorageAccessor>> out_transfer_storage_accessor) {
+    LOG_DEBUG(Service_AM, "called");
     R_UNLESS(m_impl->GetHandle() != nullptr, AM::ResultInvalidStorageType);
-
     *out_transfer_storage_accessor = std::make_shared<ITransferStorageAccessor>(system, m_impl);
     R_SUCCEED();
 }

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

@@ -33,6 +33,7 @@ constexpr Result ResultLibOpusInternalError{ErrorModule::HwOpus, 4};
 constexpr Result ResultBufferTooSmall{ErrorModule::HwOpus, 3};
 constexpr Result ResultLibOpusBadArg{ErrorModule::HwOpus, 2};
 constexpr Result ResultInvalidOpusDSPReturnCode{ErrorModule::HwOpus, 259};
+constexpr Result ResultOutOfOpusDecoders{ErrorModule::HwOpus, 385};
 constexpr Result ResultInvalidOpusSampleRate{ErrorModule::HwOpus, 1001};
 constexpr Result ResultInvalidOpusChannelCount{ErrorModule::HwOpus, 1002};
 

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

@@ -798,7 +798,7 @@ void LoopProcess(Core::System& system) {
     const auto FileSystemProxyFactory = [&] { return std::make_shared<FSP_SRV>(system); };
 
     server_manager->RegisterNamedService("fsp-ldr", std::make_shared<FSP_LDR>(system));
-    server_manager->RegisterNamedService("fsp:pr", std::make_shared<FSP_PR>(system));
+    server_manager->RegisterNamedService("fsp-pr", std::make_shared<FSP_PR>(system));
     server_manager->RegisterNamedService("fsp-srv", std::move(FileSystemProxyFactory));
     ServerManager::RunServer(std::move(server_manager));
 }

src/core/hle/service/i2c/i2c.cpp

@@ -32,7 +32,6 @@ public:
         RegisterHandlers(functions);
     }
     ~I2CSession() override = default;
-
     Result Send(InBuffer<BufferAttr_HipcMapAlias> in_data, u32 transaction_option) {
         LOG_WARNING(Service, "(stubbed) topt={}", transaction_option);
         R_THROW(ResultUnknown);
@@ -50,21 +49,40 @@ public:
         : ServiceFramework{system_, "i2c"}
     {
         static const FunctionInfo functions[] = {
-            {0, nullptr, "OpenSessionForDev"},
+            {0, C<&I2C::OpenSessionForDev>, "OpenSessionForDev"},
             {1, C<&I2C::OpenSession>, "OpenSession"},
-            {2, nullptr, "HasDevice"},
-            {3, nullptr, "HasDeviceForDev"},
-            {4, nullptr, "OpenSession2"},
+            {2, C<&I2C::HasDevice>, "HasDevice"},
+            {3, C<&I2C::HasDeviceForDev>, "HasDeviceForDev"},
+            {4, C<&I2C::OpenSession2>, "OpenSession2"},
         };
         RegisterHandlers(functions);
     }
     ~I2C() override = default;
-
-    Result OpenSession(I2CDevice device, OutInterface<I2CSession> out_session) {
+    Result OpenSessionForDev(OutInterface<I2CSession> out_session, s32 bus_idx, u32 slave_address, u32 addressing_mode, u32 speed_mode) {
         LOG_DEBUG(Service, "(stubbed)");
         *out_session = std::make_shared<I2CSession>(system);
         R_SUCCEED();
     }
+    Result OpenSession(OutInterface<I2CSession> out_session, I2CDevice device) {
+        LOG_DEBUG(Service, "(stubbed)");
+        *out_session = std::make_shared<I2CSession>(system);
+        R_SUCCEED();
+    }
+    Result HasDevice(Out<bool> out_has_device, I2CDevice device) {
+        LOG_DEBUG(Service, "(stubbed)");
+        *out_has_device = false;
+        R_SUCCEED();
+    }
+    Result HasDeviceForDev(Out<bool> out_has_device, I2CDevice device) {
+        LOG_DEBUG(Service, "(stubbed)");
+        *out_has_device = false;
+        R_SUCCEED();
+    }
+    Result OpenSession2(OutInterface<I2CSession> out_session, u32 device_code) {
+        LOG_DEBUG(Service, "(stubbed) device_code={}", device_code);
+        *out_session = std::make_shared<I2CSession>(system);
+        R_SUCCEED();
+    }
 };
 
 void LoopProcess(Core::System& system) {

src/core/hle/service/ldn/client_process_monitor.cpp

@@ -1,27 +0,0 @@
-// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
-// SPDX-License-Identifier: GPL-3.0-or-later
-
-#include "core/hle/service/cmif_serialization.h"
-#include "core/hle/service/ldn/ldn_results.h"
-#include "core/hle/service/ldn/client_process_monitor.h"
-#include "core/hle/service/ipc_helpers.h"
-
-namespace Service::LDN {
-
-IClientProcessMonitor::IClientProcessMonitor(Core::System& system_)
-    : ServiceFramework{system_, "IClientProcessMonitor"} {
-    static const FunctionInfo functions[] = {
-        {0, D<&IClientProcessMonitor::RegisterClient>, "RegisterClient"},
-    };
-    RegisterHandlers(functions);
-}
-
-IClientProcessMonitor::~IClientProcessMonitor() = default;
-
-Result IClientProcessMonitor::RegisterClient() {
-    LOG_WARNING(Service_LDN, "(STUBBED) called");
-
-    R_SUCCEED();
-}
-
-} // namespace Service::LDN

src/core/hle/service/ldn/client_process_monitor.h

@@ -1,25 +0,0 @@
-// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
-// SPDX-License-Identifier: GPL-3.0-or-later
-
-#pragma once
-
-#include "core/hle/service/cmif_types.h"
-#include "core/hle/service/service.h"
-
-namespace Core {
-class System;
-}
-
-namespace Service::LDN {
-
-class IClientProcessMonitor final
-    : public ServiceFramework<IClientProcessMonitor> {
-public:
-    explicit IClientProcessMonitor(Core::System& system_);
-    ~IClientProcessMonitor() override;
-
-private:
-    Result RegisterClient();
-};
-
-} // namespace Service::LDN

src/core/hle/service/ldn/ldn.cpp

@@ -1,4 +1,4 @@
-// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
 // SPDX-License-Identifier: GPL-3.0-or-later
 
 // SPDX-FileCopyrightText: Copyright 2018 yuzu Emulator Project
@@ -7,7 +7,6 @@
 #include "core/core.h"
 #include "core/hle/service/cmif_serialization.h"
 #include "core/hle/service/ldn/ldn.h"
-#include "core/hle/service/ldn/client_process_monitor.h"
 #include "core/hle/service/ldn/monitor_service.h"
 #include "core/hle/service/ldn/sf_monitor_service.h"
 #include "core/hle/service/ldn/sf_service.h"
@@ -17,6 +16,26 @@
 
 namespace Service::LDN {
 
+class IClientProcessMonitor final
+    : public ServiceFramework<IClientProcessMonitor> {
+public:
+    explicit IClientProcessMonitor(Core::System& system_)
+        : ServiceFramework{system_, "IClientProcessMonitor"} {
+        // clang-format off
+        static const FunctionInfo functions[] = {
+            {0, D<&IClientProcessMonitor::RegisterClient>, "RegisterClient"},
+        };
+        // clang-format on
+        RegisterHandlers(functions);
+    }
+    ~IClientProcessMonitor() override = default;
+private:
+    Result RegisterClient(ClientProcessId pid) {
+        LOG_WARNING(Service_LDN, "(STUBBED) called");
+        R_SUCCEED();
+    }
+};
+
 class IMonitorServiceCreator final : public ServiceFramework<IMonitorServiceCreator> {
 public:
     explicit IMonitorServiceCreator(Core::System& system_) : ServiceFramework{system_, "ldn:m"} {
@@ -25,7 +44,6 @@ public:
             {0, C<&IMonitorServiceCreator::CreateMonitorService>, "CreateMonitorService"}
         };
         // clang-format on
-
         RegisterHandlers(functions);
     }
 

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

@@ -139,7 +139,7 @@ IApplicationManagerInterface::IApplicationManagerInterface(Core::System& system_
         {405, nullptr, "ListApplicationControlCacheEntryInfo"},
         {406, nullptr, "GetApplicationControlProperty"},
         {407, &IApplicationManagerInterface::ListApplicationTitle, "ListApplicationTitle"},
-        {408, nullptr, "ListApplicationIcon"},
+        {408, &IApplicationManagerInterface::ListApplicationIcon, "ListApplicationIcon"},
         {411, nullptr, "Unknown411"}, //19.0.0+
         {412, nullptr, "Unknown412"}, //19.0.0+
         {413, nullptr, "Unknown413"}, //19.0.0+
@@ -894,4 +894,9 @@ void IApplicationManagerInterface::ListApplicationTitle(HLERequestContext& ctx)
     IReadOnlyApplicationControlDataInterface(system).ListApplicationTitle(ctx);
 }
 
+void IApplicationManagerInterface::ListApplicationIcon(HLERequestContext& ctx) {
+    LOG_DEBUG(Service_NS, "called");
+    IReadOnlyApplicationControlDataInterface(system).ListApplicationIcon(ctx);
+}
+
 } // namespace Service::NS

src/core/hle/service/ns/application_manager_interface.h

@@ -78,6 +78,7 @@ public:
 
     void PushApplicationRecord(HLERequestContext& ctx);
     void ListApplicationTitle(HLERequestContext& ctx);
+    void ListApplicationIcon(HLERequestContext& ctx);
 
 private:
     KernelHelpers::ServiceContext service_context;

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

@@ -10,7 +10,9 @@
 #include "core/hle/service/ns/query_service.h"
 #include "core/hle/service/ns/service_getter_interface.h"
 #include "core/hle/service/ns/system_update_interface.h"
-#include "core/hle/service/ns/vulnerability_manager_interface.h"
+#include "core/hle/service/ipc_helpers.h"
+#include "core/hle/service/cmif_serialization.h"
+#include "core/hle/service/cmif_types.h"
 #include "core/hle/service/server_manager.h"
 
 namespace Service::NS {
@@ -39,6 +41,43 @@ public:
     }
 };
 
+class IVulnerabilityManagerInterface final
+    : public ServiceFramework<IVulnerabilityManagerInterface> {
+public:
+    explicit IVulnerabilityManagerInterface(Core::System& system_)
+        : ServiceFramework{system_, "ns:vm"} {
+        // clang-format off
+        static const FunctionInfo functions[] = {
+            {1200, D<&IVulnerabilityManagerInterface::NeedsUpdateVulnerability>, "NeedsUpdateVulnerability"},
+            {1201, nullptr, "UpdateSafeSystemVersionForDebug"},
+            {1202, nullptr, "GetSafeSystemVersion"},
+            {3100, D<&IVulnerabilityManagerInterface::GetSafeSystemVersionCheckInfo>, "GetSafeSystemVersionCheckInfo"},
+            {3101, nullptr, "RequestUpdateSafeSystemVersionCheckInfo"},
+            {3102, D<&IVulnerabilityManagerInterface::ResetSafeSystemVersionCheckInfo>, "ResetSafeSystemVersionCheckInfo"},
+        };
+        // clang-format on
+        RegisterHandlers(functions);
+    }
+    ~IVulnerabilityManagerInterface() override = default;
+
+    Result NeedsUpdateVulnerability(Out<bool> out_needs_update_vulnerability) {
+        LOG_WARNING(Service_NS, "(STUBBED)");
+        *out_needs_update_vulnerability = false;
+        R_SUCCEED();
+    }
+
+    Result GetSafeSystemVersionCheckInfo(Out<std::array<u8, 0x20>> out_system_version_check_info) {
+        LOG_WARNING(Service_NS, "(STUBBED)");
+        *out_system_version_check_info = {};
+        R_SUCCEED();
+    }
+
+    Result ResetSafeSystemVersionCheckInfo() {
+        LOG_WARNING(Service_NS, "(STUBBED)");
+        R_SUCCEED();
+    }
+};
+
 void LoopProcess(Core::System& system) {
     auto server_manager = std::make_unique<ServerManager>(system);
 

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

@@ -10,13 +10,17 @@
 #include <string>
 
 #include "common/stb.h"
+#include "common/logging.h"
 #include "common/settings.h"
 #include "core/file_sys/control_metadata.h"
 #include "core/file_sys/patch_manager.h"
 #include "core/file_sys/vfs/vfs.h"
 #include "core/hle/kernel/k_process.h"
 #include "core/hle/kernel/k_transfer_memory.h"
+#include "core/hle/result.h"
 #include "core/hle/service/cmif_serialization.h"
+#include "core/hle/service/cmif_types.h"
+#include "core/hle/service/hle_ipc.h"
 #include "core/hle/service/ns/language.h"
 #include "core/hle/service/ns/ns_types.h"
 #include "core/hle/service/ns/ns_results.h"
@@ -72,24 +76,26 @@ void SanitizeJPEGImageSize(std::vector<u8>& image) {
 
 // IAsyncValue implementation for ListApplicationTitle
 // https://switchbrew.org/wiki/NS_services#ListApplicationTitle
-class IAsyncValueForListApplicationTitle final : public ServiceFramework<IAsyncValueForListApplicationTitle> {
+class IAsyncValue final : public ServiceFramework<IAsyncValue> {
 public:
-    explicit IAsyncValueForListApplicationTitle(Core::System& system_, s32 offset, s32 size)
-        : ServiceFramework{system_, "IAsyncValue"}, service_context{system_, "IAsyncValue"},
-          data_offset{offset}, data_size{size} {
+    explicit IAsyncValue(Core::System& system_, s32 offset, s32 size)
+        : ServiceFramework{system_, "IAsyncValue"}
+        , service_context{system_, "IAsyncValue"}
+        , data_offset{offset}
+        , data_size{size}
+    {
         static const FunctionInfo functions[] = {
-            {0, &IAsyncValueForListApplicationTitle::GetSize, "GetSize"},
-            {1, &IAsyncValueForListApplicationTitle::Get, "Get"},
-            {2, &IAsyncValueForListApplicationTitle::Cancel, "Cancel"},
-            {3, &IAsyncValueForListApplicationTitle::GetErrorContext, "GetErrorContext"},
+            {0, D<&IAsyncValue::GetSize>, "GetSize"},
+            {1, D<&IAsyncValue::Get>, "Get"},
+            {2, D<&IAsyncValue::Cancel>, "Cancel"},
+            {3, D<&IAsyncValue::GetErrorContext>, "GetErrorContext"},
         };
         RegisterHandlers(functions);
-
         completion_event = service_context.CreateEvent("IAsyncValue:Completion");
         completion_event->GetReadableEvent().Signal(system.Kernel());
     }
 
-    ~IAsyncValueForListApplicationTitle() override {
+    ~IAsyncValue() override {
         service_context.CloseEvent(completion_event);
     }
 
@@ -98,35 +104,24 @@ public:
     }
 
 private:
-    void GetSize(HLERequestContext& ctx) {
+    Result GetSize(Out<s64> out_data_size) {
         LOG_DEBUG(Service_NS, "called");
-        IPC::ResponseBuilder rb{ctx, 4};
-        rb.Push(ResultSuccess);
-        rb.Push<s64>(data_size);
+        *out_data_size = data_size;
+        R_SUCCEED();
     }
-
-    void Get(HLERequestContext& ctx) {
+    Result Get(OutBuffer<BufferAttr_HipcMapAlias> out_data_offset) {
         LOG_DEBUG(Service_NS, "called");
-        std::vector<u8> buffer(sizeof(s32));
-        std::memcpy(buffer.data(), &data_offset, sizeof(s32));
-        ctx.WriteBuffer(buffer);
-
-        IPC::ResponseBuilder rb{ctx, 2};
-        rb.Push(ResultSuccess);
+        std::memcpy(out_data_offset.data(), &data_offset, sizeof(s32));
+        R_SUCCEED();
     }
-
-    void Cancel(HLERequestContext& ctx) {
+    Result Cancel() {
         LOG_DEBUG(Service_NS, "called");
-        IPC::ResponseBuilder rb{ctx, 2};
-        rb.Push(ResultSuccess);
+        R_SUCCEED();
     }
-
-    void GetErrorContext(HLERequestContext& ctx) {
+    Result GetErrorContext() {
         LOG_DEBUG(Service_NS, "called");
-        IPC::ResponseBuilder rb{ctx, 2};
-        rb.Push(ResultSuccess);
+        R_SUCCEED();
     }
-
     KernelHelpers::ServiceContext service_context;
     Kernel::KEvent* completion_event{};
     s32 data_offset;
@@ -144,6 +139,7 @@ IReadOnlyApplicationControlDataInterface::IReadOnlyApplicationControlDataInterfa
         {3, nullptr, "ConvertLanguageCodeToApplicationLanguage"},
         {4, nullptr, "SelectApplicationDesiredLanguage"},
         {5, D<&IReadOnlyApplicationControlDataInterface::GetApplicationControlData2>, "GetApplicationControlData"},
+        {10, &IReadOnlyApplicationControlDataInterface::ListApplicationIcon, "ListApplicationIcon"},
         {13, &IReadOnlyApplicationControlDataInterface::ListApplicationTitle, "ListApplicationTitle"},
         {19, D<&IReadOnlyApplicationControlDataInterface::GetApplicationControlData3>, "GetApplicationControlData"},
     };
@@ -160,8 +156,7 @@ Result IReadOnlyApplicationControlDataInterface::GetApplicationControlData(
     LOG_INFO(Service_NS, "called with control_source={}, application_id={:016X}",
              application_control_source, application_id);
 
-    const FileSys::PatchManager pm{application_id, system.GetFileSystemController(),
-                                   system.GetContentProvider()};
+    const FileSys::PatchManager pm{application_id, system.GetFileSystemController(), system.GetContentProvider()};
     const auto control = pm.GetControlMetadata();
     const auto size = out_buffer.size();
 
@@ -169,8 +164,7 @@ Result IReadOnlyApplicationControlDataInterface::GetApplicationControlData(
     const auto total_size = sizeof(FileSys::RawNACP) + icon_size;
 
     if (size < total_size) {
-        LOG_ERROR(Service_NS, "output buffer is too small! (actual={:016X}, expected_min=0x4000)",
-                  size);
+        LOG_ERROR(Service_NS, "output buffer is too small! (actual={:016X}, expected_min=0x4000)", size);
         R_THROW(ResultUnknown);
     }
 
@@ -178,8 +172,7 @@ Result IReadOnlyApplicationControlDataInterface::GetApplicationControlData(
         const auto bytes = control.first->GetRawBytes();
         std::memcpy(out_buffer.data(), bytes.data(), bytes.size());
     } else {
-        LOG_WARNING(Service_NS, "missing NACP data for application_id={:016X}, defaulting to zero",
-                    application_id);
+        LOG_WARNING(Service_NS, "missing NACP data for application_id={:016X}, defaulting to zero", application_id);
         std::memset(out_buffer.data(), 0, sizeof(FileSys::RawNACP));
     }
 
@@ -204,15 +197,12 @@ Result IReadOnlyApplicationControlDataInterface::GetApplicationDesiredLanguage(
     // Convert to application language, get priority list
     const auto application_language = ConvertToApplicationLanguage(language_code);
     if (application_language == std::nullopt) {
-        LOG_ERROR(Service_NS, "Could not convert application language! language_code={}",
-                  language_code);
+        LOG_ERROR(Service_NS, "Could not convert application language! language_code={}", language_code);
         R_THROW(Service::NS::ResultApplicationLanguageNotFound);
     }
     const auto priority_list = GetApplicationLanguagePriorityList(*application_language);
     if (!priority_list) {
-        LOG_ERROR(Service_NS,
-                  "Could not find application language priorities! application_language={}",
-                  *application_language);
+        LOG_ERROR(Service_NS, "Could not find application language priorities! application_language={}", *application_language);
         R_THROW(Service::NS::ResultApplicationLanguageNotFound);
     }
 
@@ -256,8 +246,7 @@ Result IReadOnlyApplicationControlDataInterface::GetApplicationControlData2(
     const auto nacp_size = sizeof(FileSys::RawNACP);
 
     if (size < nacp_size) {
-        LOG_ERROR(Service_NS, "output buffer is too small! (actual={:016X}, expected_min={:08X})",
-                  size, nacp_size);
+        LOG_ERROR(Service_NS, "output buffer is too small! (actual={:016X}, expected_min={:08X})", size, nacp_size);
         R_THROW(ResultUnknown);
     }
 
@@ -308,63 +297,81 @@ Result IReadOnlyApplicationControlDataInterface::GetApplicationControlData2(
     R_SUCCEED();
 }
 
-
-void IReadOnlyApplicationControlDataInterface::ListApplicationTitle(HLERequestContext& ctx) {
-    /*
-    IPC::RequestParser rp{ctx};
-    auto control_source = rp.PopRaw<u8>();
-    rp.Skip(7, false);
-    auto transfer_memory_size = rp.Pop<u64>();
-    */
-
+void IReadOnlyApplicationControlDataInterface::ListApplicationIcon(HLERequestContext& ctx) {
+    LOG_WARNING(Service_NS, "(stubbed)");
     const auto app_ids_buffer = ctx.ReadBuffer();
-    const size_t app_count = app_ids_buffer.size() / sizeof(u64);
-
-    std::vector<u64> application_ids(app_count);
-    if (app_count > 0) {
-        std::memcpy(application_ids.data(), app_ids_buffer.data(), app_count * sizeof(u64));
-    }
-
+    const u64 app_count = app_ids_buffer.size() / sizeof(u64);
     auto t_mem_obj = ctx.GetObjectFromHandle<Kernel::KTransferMemory>(ctx.GetCopyHandle(0));
     auto* t_mem = t_mem_obj.GetPointerUnsafe();
-
-    constexpr size_t title_entry_size = sizeof(FileSys::LanguageEntry);
-    const size_t total_data_size = app_count * title_entry_size;
-
-    constexpr s32 data_offset = 0;
-
+    size_t out_length = 0;
     if (t_mem != nullptr && t_mem->GetOwner() != nullptr && app_count > 0) {
         auto& memory = t_mem->GetOwner()->GetMemory();
         const auto t_mem_address = t_mem->GetSourceAddress();
-
+        // u64 - app count
+        memory.WriteBlock(t_mem_address + out_length, &app_count, sizeof(u64));
+        out_length += sizeof(u64);
+        // [list of u64] - size of icons
         for (size_t i = 0; i < app_count; ++i) {
-            const u64 app_id = application_ids[i];
-            const FileSys::PatchManager pm{app_id, system.GetFileSystemController(),
-                                           system.GetContentProvider()};
+            const u64 app_id = app_ids_buffer[i];
+            const FileSys::PatchManager pm{app_id, system.GetFileSystemController(), system.GetContentProvider()};
             const auto control = pm.GetControlMetadata();
-
-            FileSys::LanguageEntry entry{};
-            if (control.first != nullptr) {
-                entry = control.first->GetLanguageEntry();
+            u64 full_size = control.second->GetSize();
+            memory.WriteBlock(t_mem_address + out_length, &full_size, sizeof(u64));
+            out_length += sizeof(u64);
+        }
+        // [list of raw icon data]
+        std::vector<u8> full_icon_data;
+        for (size_t i = 0; i < app_count; ++i) {
+            const u64 app_id = app_ids_buffer[i];
+            const FileSys::PatchManager pm{app_id, system.GetFileSystemController(), system.GetContentProvider()};
+            const auto control = pm.GetControlMetadata();
+            auto const full_size = control.second->GetSize();
+            if (full_size > 0) {
+                full_icon_data.resize(full_size);
+                control.second->Read(full_icon_data.data(), full_size, 0);
+                memory.WriteBlock(t_mem_address + out_length, full_icon_data.data(), full_size);
+                out_length += full_size;
             }
-
-            const size_t offset = i * title_entry_size;
-            memory.WriteBlock(t_mem_address + offset, &entry, title_entry_size);
         }
     }
-
-    auto async_value = std::make_shared<IAsyncValueForListApplicationTitle>(
-        system, data_offset, static_cast<s32>(total_data_size));
-
+    auto async_value = std::make_shared<IAsyncValue>(system, 0, s32(out_length));
     IPC::ResponseBuilder rb{ctx, 2, 1, 1};
     rb.Push(ResultSuccess);
     rb.PushCopyObjects(ctx, async_value->ReadableEvent());
     rb.PushIpcInterface(ctx, std::move(async_value));
 }
 
-Result IReadOnlyApplicationControlDataInterface::GetApplicationControlData3(
-   OutBuffer<BufferAttr_HipcMapAlias> out_buffer, Out<u32> out_flags_a, Out<u32> out_flags_b,
-   Out<u32> out_actual_size, ApplicationControlSource application_control_source, u8 flag1, u8 flag2, u64 application_id) {
+void IReadOnlyApplicationControlDataInterface::ListApplicationTitle(HLERequestContext& ctx) {
+    const auto app_ids_buffer = ctx.ReadBuffer();
+    const size_t app_count = app_ids_buffer.size() / sizeof(u64);
+    auto t_mem_obj = ctx.GetObjectFromHandle<Kernel::KTransferMemory>(ctx.GetCopyHandle(0));
+    auto* t_mem = t_mem_obj.GetPointerUnsafe();
+    constexpr size_t title_entry_size = sizeof(FileSys::LanguageEntry);
+    const size_t total_data_size = app_count * title_entry_size;
+    constexpr s32 data_offset = 0;
+    if (t_mem != nullptr && app_count > 0) {
+        auto& memory = system.ApplicationMemory();
+        const auto t_mem_address = t_mem->GetSourceAddress();
+        for (size_t i = 0; i < app_count; ++i) {
+            const u64 app_id = app_ids_buffer[i];
+            const FileSys::PatchManager pm{app_id, system.GetFileSystemController(), system.GetContentProvider()};
+            const auto control = pm.GetControlMetadata();
+            FileSys::LanguageEntry entry{};
+            if (control.first != nullptr) {
+                entry = control.first->GetLanguageEntry();
+            }
+            const size_t offset = i * title_entry_size;
+            memory.WriteBlock(t_mem_address + offset, &entry, title_entry_size);
+        }
+    }
+    auto async_value = std::make_shared<IAsyncValue>(system, data_offset, s32(total_data_size));
+    IPC::ResponseBuilder rb{ctx, 2, 1, 1};
+    rb.Push(ResultSuccess);
+    rb.PushCopyObjects(ctx, async_value->ReadableEvent());
+    rb.PushIpcInterface(ctx, std::move(async_value));
+}
+
+Result IReadOnlyApplicationControlDataInterface::GetApplicationControlData3(OutBuffer<BufferAttr_HipcMapAlias> out_buffer, Out<u32> out_flags_a, Out<u32> out_flags_b, Out<u32> out_actual_size, ApplicationControlSource application_control_source, u8 flag1, u8 flag2, u64 application_id) {
     LOG_INFO(Service_NS, "called with control_source={}, flags=({:02X},{:02X}), application_id={:016X}",
              application_control_source, flag1, flag2, application_id);
 

src/core/hle/service/ns/read_only_application_control_data_interface.h

@@ -34,6 +34,7 @@ public:
         u8 flag1,
         u8 flag2,
         u64 application_id);
+    void ListApplicationIcon(HLERequestContext& ctx);
     void ListApplicationTitle(HLERequestContext& ctx);
     Result GetApplicationControlData3(
         OutBuffer<BufferAttr_HipcMapAlias> out_buffer,

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

@@ -1,4 +1,4 @@
-// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
 // SPDX-License-Identifier: GPL-3.0-or-later
 
 // SPDX-FileCopyrightText: Copyright 2024 yuzu Emulator Project
@@ -16,10 +16,8 @@ IReadOnlyApplicationRecordInterface::IReadOnlyApplicationRecordInterface(Core::S
     static const FunctionInfo functions[] = {
         {0, D<&IReadOnlyApplicationRecordInterface::HasApplicationRecord>, "HasApplicationRecord"},
         {1, nullptr, "NotifyApplicationFailure"},
-        {2, D<&IReadOnlyApplicationRecordInterface::IsDataCorruptedResult>,
-         "IsDataCorruptedResult"},
-        {3, D<&IReadOnlyApplicationRecordInterface::ListApplicationRecord>,
-         "ListApplicationRecord"},
+        {2, D<&IReadOnlyApplicationRecordInterface::IsDataCorruptedResult>, "IsDataCorruptedResult"},
+        {3, D<&IReadOnlyApplicationRecordInterface::ListApplicationRecord>, "ListApplicationRecord"},
     };
     // clang-format on
 

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

@@ -1,31 +0,0 @@
-// SPDX-FileCopyrightText: Copyright 2024 yuzu Emulator Project
-// SPDX-License-Identifier: GPL-2.0-or-later
-
-#include "core/hle/service/cmif_serialization.h"
-#include "core/hle/service/ns/vulnerability_manager_interface.h"
-
-namespace Service::NS {
-
-IVulnerabilityManagerInterface::IVulnerabilityManagerInterface(Core::System& system_)
-    : ServiceFramework{system_, "ns:vm"} {
-    // clang-format off
-    static const FunctionInfo functions[] = {
-        {1200, D<&IVulnerabilityManagerInterface::NeedsUpdateVulnerability>, "NeedsUpdateVulnerability"},
-        {1201, nullptr, "UpdateSafeSystemVersionForDebug"},
-        {1202, nullptr, "GetSafeSystemVersion"},
-    };
-    // clang-format on
-
-    RegisterHandlers(functions);
-}
-
-IVulnerabilityManagerInterface::~IVulnerabilityManagerInterface() = default;
-
-Result IVulnerabilityManagerInterface::NeedsUpdateVulnerability(
-    Out<bool> out_needs_update_vulnerability) {
-    LOG_WARNING(Service_NS, "(STUBBED) called");
-    *out_needs_update_vulnerability = false;
-    R_SUCCEED();
-}
-
-} // namespace Service::NS

src/core/hle/service/ns/vulnerability_manager_interface.h

@@ -1,21 +0,0 @@
-// SPDX-FileCopyrightText: Copyright 2024 yuzu Emulator Project
-// SPDX-License-Identifier: GPL-2.0-or-later
-
-#pragma once
-
-#include "core/hle/service/cmif_types.h"
-#include "core/hle/service/service.h"
-
-namespace Service::NS {
-
-class IVulnerabilityManagerInterface final
-    : public ServiceFramework<IVulnerabilityManagerInterface> {
-public:
-    explicit IVulnerabilityManagerInterface(Core::System& system_);
-    ~IVulnerabilityManagerInterface() override;
-
-private:
-    Result NeedsUpdateVulnerability(Out<bool> out_needs_update_vulnerability);
-};
-
-} // namespace Service::NS

src/core/hle/service/set/setting_formats/device_settings.h

@@ -1,3 +1,6 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
 // SPDX-FileCopyrightText: Copyright 2018 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
@@ -30,15 +33,15 @@ struct DeviceSettings {
     INSERT_PADDING_BYTES(0x20); // Reserved
 
     // nn::settings::system::ConsoleSixAxisSensorAccelerationBias
-    Common::Vec3<f32> console_six_axis_sensor_acceleration_bias;
+    Common::Vec<f32, 3> console_six_axis_sensor_acceleration_bias;
     // nn::settings::system::ConsoleSixAxisSensorAngularVelocityBias
-    Common::Vec3<f32> console_six_axis_sensor_angular_velocity_bias;
+    Common::Vec<f32, 3> console_six_axis_sensor_angular_velocity_bias;
     // nn::settings::system::ConsoleSixAxisSensorAccelerationGain
     std::array<u8, 0x24> console_six_axis_sensor_acceleration_gain;
     // nn::settings::system::ConsoleSixAxisSensorAngularVelocityGain
     std::array<u8, 0x24> console_six_axis_sensor_angular_velocity_gain;
     // nn::settings::system::ConsoleSixAxisSensorAngularVelocityTimeBias
-    Common::Vec3<f32> console_six_axis_sensor_angular_velocity_time_bias;
+    Common::Vec<f32, 3> console_six_axis_sensor_angular_velocity_time_bias;
     // nn::settings::system::ConsoleSixAxisSensorAngularAcceleration
     std::array<u8, 0x24> console_six_axis_sensor_angular_acceleration;
 };

src/core/hle/service/set/setting_formats/system_settings.h

@@ -1,4 +1,4 @@
-// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
 // SPDX-License-Identifier: GPL-3.0-or-later
 
 // SPDX-FileCopyrightText: Copyright 2018 yuzu Emulator Project
@@ -153,15 +153,15 @@ struct SystemSettings {
     INSERT_PADDING_BYTES(0x7FF8); // Reserved
 
     // nn::settings::system::ConsoleSixAxisSensorAccelerationBias
-    Common::Vec3<f32> console_six_axis_sensor_acceleration_bias;
+    Common::Vec<f32, 3> console_six_axis_sensor_acceleration_bias;
     // nn::settings::system::ConsoleSixAxisSensorAngularVelocityBias
-    Common::Vec3<f32> console_six_axis_sensor_angular_velocity_bias;
+    Common::Vec<f32, 3> console_six_axis_sensor_angular_velocity_bias;
     // nn::settings::system::ConsoleSixAxisSensorAccelerationGain
     std::array<u8, 0x24> console_six_axis_sensor_acceleration_gain;
     // nn::settings::system::ConsoleSixAxisSensorAngularVelocityGain
     std::array<u8, 0x24> console_six_axis_sensor_angular_velocity_gain;
     // nn::settings::system::ConsoleSixAxisSensorAngularVelocityTimeBias
-    Common::Vec3<f32> console_six_axis_sensor_angular_velocity_time_bias;
+    Common::Vec<f32, 3> console_six_axis_sensor_angular_velocity_time_bias;
     // nn::settings::system::ConsoleSixAxisSensorAngularAcceleration
     std::array<u8, 0x24> console_six_axis_sensor_angular_velocity_acceleration;
     INSERT_PADDING_BYTES(0x70); // Reserved

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

@@ -1,18 +0,0 @@
-// SPDX-FileCopyrightText: Copyright 2018 yuzu Emulator Project
-// SPDX-License-Identifier: GPL-2.0-or-later
-
-#include "core/hle/service/spl/csrng.h"
-
-namespace Service::SPL {
-
-CSRNG::CSRNG(Core::System& system_, std::shared_ptr<Module> module_)
-    : Interface(system_, std::move(module_), "csrng") {
-    static const FunctionInfo functions[] = {
-        {0, &CSRNG::GenerateRandomBytes, "GenerateRandomBytes"},
-    };
-    RegisterHandlers(functions);
-}
-
-CSRNG::~CSRNG() = default;
-
-} // namespace Service::SPL

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

@@ -1,20 +0,0 @@
-// SPDX-FileCopyrightText: Copyright 2018 yuzu Emulator Project
-// SPDX-License-Identifier: GPL-2.0-or-later
-
-#pragma once
-
-#include "core/hle/service/spl/spl_module.h"
-
-namespace Core {
-class System;
-}
-
-namespace Service::SPL {
-
-class CSRNG final : public Module::Interface {
-public:
-    explicit CSRNG(Core::System& system_, std::shared_ptr<Module> module_);
-    ~CSRNG() override;
-};
-
-} // namespace Service::SPL

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

@@ -13,7 +13,6 @@
 #include "core/hle/api_version.h"
 #include "core/hle/service/ipc_helpers.h"
 #include "core/hle/service/server_manager.h"
-#include "core/hle/service/spl/csrng.h"
 #include "core/hle/service/spl/spl.h"
 #include "core/hle/service/spl/spl_module.h"
 
@@ -203,6 +202,18 @@ Result Module::Interface::GetConfigImpl(u64* out_config, ConfigItem config_item)
     }
 }
 
+class CSRNG final : public Module::Interface {
+public:
+    explicit CSRNG(Core::System& system_, std::shared_ptr<Module> module_)
+        : Interface(system_, std::move(module_), "csrng") {
+        static const FunctionInfo functions[] = {
+            {0, &CSRNG::GenerateRandomBytes, "GenerateRandomBytes"},
+        };
+        RegisterHandlers(functions);
+    }
+    ~CSRNG() override = default;
+};
+
 void LoopProcess(Core::System& system) {
     auto server_manager = std::make_unique<ServerManager>(system);
     auto module = std::make_shared<Module>();

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

@@ -13,6 +13,7 @@
 #include "core/hle/service/vi/manager_display_service.h"
 #include "core/hle/service/vi/system_display_service.h"
 #include "core/hle/service/vi/vi_results.h"
+#include "service_creator.h"
 
 namespace Service::VI {
 
@@ -38,6 +39,7 @@ IApplicationDisplayService::IApplicationDisplayService(Core::System& system_,
         {2031, C<&IApplicationDisplayService::DestroyStrayLayer>, "DestroyStrayLayer"},
         {2101, C<&IApplicationDisplayService::SetLayerScalingMode>, "SetLayerScalingMode"},
         {2102, C<&IApplicationDisplayService::ConvertScalingMode>, "ConvertScalingMode"},
+        {2103, C<&IApplicationDisplayService::Cmd2103>, "Cmd2103"},
         {2450, C<&IApplicationDisplayService::GetIndirectLayerImageMap>, "GetIndirectLayerImageMap"},
         {2451, nullptr, "GetIndirectLayerImageCropMap"},
         {2460, C<&IApplicationDisplayService::GetIndirectLayerImageRequiredMemoryInfo>, "GetIndirectLayerImageRequiredMemoryInfo"},
@@ -289,6 +291,11 @@ Result IApplicationDisplayService::ConvertScalingMode(Out<ConvertedScaleMode> ou
     }
 }
 
+Result IApplicationDisplayService::Cmd2103(Out<std::array<u8, 0x18>> out_unk18) {
+    LOG_WARNING(Service_VI, "(stubbed)");
+    R_SUCCEED();
+}
+
 Result IApplicationDisplayService::GetIndirectLayerImageMap(
     Out<u64> out_size, Out<u64> out_stride,
     OutBuffer<BufferAttr_HipcMapTransferAllowsNonSecure | BufferAttr_HipcMapAlias> out_buffer,

src/core/hle/service/vi/application_display_service.h

@@ -64,6 +64,7 @@ public:
     Result GetDisplayVsyncEvent(OutCopyHandle<Kernel::KReadableEvent> out_vsync_event,
                                 u64 display_id);
     Result ConvertScalingMode(Out<ConvertedScaleMode> out_scaling_mode, NintendoScaleMode mode);
+    Result Cmd2103(Out<std::array<u8, 0x18>> out_unk18);
     Result GetIndirectLayerImageMap(
         Out<u64> out_size, Out<u64> out_stride,
         OutBuffer<BufferAttr_HipcMapTransferAllowsNonSecure | BufferAttr_HipcMapAlias> out_buffer,

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

@@ -300,28 +300,6 @@ Result SharedBufferManager::CreateSession(Kernel::KProcess* owner_process, u64*
             }
         }
 
-        // Claim a presentation slot range.
-        u32 slot_base = 0;
-
-        std::array<bool, SharedBufferMaxSessions> in_use{};
-        for (const auto& [existing_aruid, existing] : m_sessions) {
-            const u32 index = existing.presentation_slot_base / SharedBufferSlotsPerSession;
-
-            if (index < in_use.size())
-                in_use[index] = true;
-        }
-
-        u32 index = 0;
-        while (index < in_use.size() && in_use[index])
-            index++;
-
-        if (index >= in_use.size()) {
-            LOG_ERROR(Service_VI, "Out of shared buffer presentation slots ({} sessions)", SharedBufferMaxSessions);
-            R_THROW(VI::ResultOperationFailed);
-        }
-
-        slot_base = index * SharedBufferSlotsPerSession;
-
         // Map into process.
         Common::ProcessAddress map_address{};
         R_TRY(MapSharedBufferIntoProcessAddressSpace(std::addressof(map_address), m_buffer_page_group,
@@ -330,7 +308,6 @@ Result SharedBufferManager::CreateSession(Kernel::KProcess* owner_process, u64*
         // Create new session.
         auto [it, was_emplaced] = m_sessions.emplace(aruid, SharedBufferSession{});
         auto& session = it->second;
-        session.presentation_slot_base = slot_base;
 
         auto& container = m_nvdrv->GetContainer();
         session.session_id = container.OpenSession(owner_process);

src/core/memory.cpp

@@ -101,8 +101,10 @@ struct Memory::Impl {
         }
 
         u64 protect_bytes = 0, protect_begin = 0;
+
+        current_page_table->entries.CommitRegion(vaddr >> YUZU_PAGEBITS, (vaddr + size) >> YUZU_PAGEBITS);
         for (u64 addr = vaddr; addr < vaddr + size; addr += YUZU_PAGESIZE) {
-            const Common::PageType page_type = current_page_table->entries[addr >> YUZU_PAGEBITS].ptr.Type();
+            const Common::PageType page_type = current_page_table->entries.GetUnchecked(addr >> YUZU_PAGEBITS).Type();
             switch (page_type) {
             case Common::PageType::RasterizerCachedMemory:
                 if (protect_bytes > 0) {
@@ -123,16 +125,14 @@ struct Memory::Impl {
     }
 
     [[nodiscard]] u8* GetPointerFromRasterizerCachedMemory(u64 vaddr) const {
-        Common::PhysicalAddress const paddr = current_page_table->entries[vaddr >> YUZU_PAGEBITS].addr;
-        if (paddr)
-            return system.DeviceMemory().GetPointer<u8>(paddr + vaddr);
+        if (u64 paddr = current_page_table->entries[vaddr >> YUZU_PAGEBITS].Pointer(true); paddr)
+            return reinterpret_cast<u8*>(paddr) + vaddr;
         return {};
     }
 
     [[nodiscard]] u8* GetPointerFromDebugMemory(u64 vaddr) const {
-        const Common::PhysicalAddress paddr = current_page_table->entries[vaddr >> YUZU_PAGEBITS].addr;
-        if (paddr != 0)
-            return system.DeviceMemory().GetPointer<u8>(paddr + vaddr);
+        if (u64 paddr = current_page_table->entries[vaddr >> YUZU_PAGEBITS].Pointer(true); paddr)
+            return reinterpret_cast<u8*>(paddr) + vaddr;
         return {};
     }
 
@@ -243,10 +243,12 @@ struct Memory::Impl {
         std::size_t page_index = addr >> YUZU_PAGEBITS;
         std::size_t page_offset = addr & YUZU_PAGEMASK;
         bool user_accessible = true;
+
+        current_page_table->entries.CommitRegion(page_index, page_index + (size >> YUZU_PAGEBITS) + 1);
         while (remaining_size != 0) {
             const std::size_t copy_amount = (std::min)(std::size_t(YUZU_PAGESIZE) - page_offset, remaining_size);
             const auto current_vaddr = u64((page_index << YUZU_PAGEBITS) + page_offset);
-            const auto [pointer, type] = current_page_table->entries[page_index].ptr.PointerType();
+            const auto [pointer, type, _] = current_page_table->entries.GetUnchecked(page_index).PointerTypeBlock();
             switch (type) {
             case Common::PageType::Unmapped: {
                 user_accessible = false;
@@ -297,10 +299,10 @@ struct Memory::Impl {
     }
 
     [[nodiscard]] inline const u8* GetSpan(const VAddr addr, const std::size_t size) const noexcept {
-        return (current_page_table->entries[addr >> YUZU_PAGEBITS].block == current_page_table->entries[(addr + size) >> YUZU_PAGEBITS].block) ? GetPointerSilent(addr) : nullptr;
+        return (current_page_table->entries[addr >> YUZU_PAGEBITS].Block() == current_page_table->entries[(addr + size) >> YUZU_PAGEBITS].Block()) ? GetPointerSilent(addr) : nullptr;
     }
     [[nodiscard]] inline u8* GetSpan(const VAddr addr, const std::size_t size) noexcept {
-        return (current_page_table->entries[addr >> YUZU_PAGEBITS].block == current_page_table->entries[(addr + size) >> YUZU_PAGEBITS].block) ? GetPointerSilent(addr) : nullptr;
+        return (current_page_table->entries[addr >> YUZU_PAGEBITS].Block() == current_page_table->entries[(addr + size) >> YUZU_PAGEBITS].Block()) ? GetPointerSilent(addr) : nullptr;
     }
 
     bool WriteBlockImpl(const Common::ProcessAddress addr, const void* buffer, const std::size_t size, bool unsafe) {
@@ -404,11 +406,14 @@ struct Memory::Impl {
         // The region is at a granularity of CPU pages.
 
         const u64 num_pages = ((vaddr + size - 1) >> YUZU_PAGEBITS) - (vaddr >> YUZU_PAGEBITS) + 1;
+
+        current_page_table->entries.CommitRegion(vaddr >> YUZU_PAGEBITS, (vaddr >> YUZU_PAGEBITS) + num_pages);
         for (u64 i = 0; i < num_pages; ++i, vaddr += YUZU_PAGESIZE) {
-            const Common::PageType page_type = current_page_table->entries[vaddr >> YUZU_PAGEBITS].ptr.Type();
+            auto& entry = current_page_table->entries.GetUnchecked(vaddr >> YUZU_PAGEBITS);
+            const auto [pointer, type, block] = entry.PointerTypeBlock(true);
             if (debug) {
                 // Switch page type to debug if now debug
-                switch (page_type) {
+                switch (type) {
                 case Common::PageType::Unmapped:
                     ASSERT(false && "Attempted to mark unmapped pages as debug");
                     break;
@@ -417,14 +422,14 @@ struct Memory::Impl {
                     // Page is already marked.
                     break;
                 case Common::PageType::Memory:
-                    current_page_table->entries[vaddr >> YUZU_PAGEBITS].ptr.Store(0, Common::PageType::DebugMemory);
+                    entry.MarkDebug(pointer, block);
                     break;
                 default:
                     UNREACHABLE();
                 }
             } else {
                 // Switch page type to non-debug if now non-debug
-                switch (page_type) {
+                switch (type) {
                 case Common::PageType::Unmapped:
                     ASSERT(false && "Attempted to mark unmapped pages as non-debug");
                     break;
@@ -433,8 +438,7 @@ struct Memory::Impl {
                     // Don't mess with already non-debug or rasterizer memory.
                     break;
                 case Common::PageType::DebugMemory: {
-                    u8* const pointer = GetPointerFromDebugMemory(vaddr & ~YUZU_PAGEMASK);
-                    current_page_table->entries[vaddr >> YUZU_PAGEBITS].ptr.Store(uintptr_t(pointer) - (vaddr & ~YUZU_PAGEMASK), Common::PageType::Memory);
+                    entry.Store(false, Common::PageType::Memory, block, pointer);
                     break;
                 }
                 default:
@@ -466,8 +470,10 @@ struct Memory::Impl {
         // is different). This assumes the specified GPU address region is contiguous as well.
 
         const u64 num_pages = ((vaddr + size - 1) >> YUZU_PAGEBITS) - (vaddr >> YUZU_PAGEBITS) + 1;
+        current_page_table->entries.CommitRegion(vaddr >> YUZU_PAGEBITS, (vaddr >> YUZU_PAGEBITS) + num_pages);
         for (u64 i = 0; i < num_pages; ++i, vaddr += YUZU_PAGESIZE) {
-            const Common::PageType page_type= current_page_table->entries[vaddr >> YUZU_PAGEBITS].ptr.Type();
+            auto& entry = current_page_table->entries.GetUnchecked(vaddr >> YUZU_PAGEBITS);
+            const Common::PageType page_type = entry.Type();
             if (cached) {
                 // Switch page type to cached if now cached
                 switch (page_type) {
@@ -477,7 +483,7 @@ struct Memory::Impl {
                     break;
                 case Common::PageType::DebugMemory:
                 case Common::PageType::Memory:
-                    current_page_table->entries[vaddr >> YUZU_PAGEBITS].ptr.Store(0, Common::PageType::RasterizerCachedMemory);
+                    entry.MarkRasterizerCached();
                     break;
                 case Common::PageType::RasterizerCachedMemory:
                     // There can be more than one GPU region mapped per CPU region, so it's common
@@ -499,13 +505,13 @@ struct Memory::Impl {
                     // that this area is already unmarked as cached.
                     break;
                 case Common::PageType::RasterizerCachedMemory: {
-                    if (u8* const pointer = GetPointerFromRasterizerCachedMemory(vaddr & ~YUZU_PAGEMASK); pointer == nullptr) {
+                    if (auto [ptr, _, block] = entry.PointerTypeBlock(true); ptr == 0) {
                         // It's possible that this function has been called while updating the
                         // pagetable after unmapping a VMA. In that case the underlying VMA will no
                         // longer exist, and we should just leave the pagetable entry blank.
-                        current_page_table->entries[vaddr >> YUZU_PAGEBITS].ptr.Store(0, Common::PageType::Unmapped);
+                        entry.Store(false, Common::PageType::Unmapped, block, 0);
                     } else {
-                        current_page_table->entries[vaddr >> YUZU_PAGEBITS].ptr.Store(uintptr_t(pointer) - (vaddr & ~YUZU_PAGEMASK), Common::PageType::Memory);
+                        entry.Store(false, Common::PageType::Memory, block, ptr);
                     }
                     break;
                 }
@@ -539,22 +545,18 @@ struct Memory::Impl {
             ASSERT_MSG(type != Common::PageType::Memory,
                        "Mapping memory page without a pointer @ {:016x}", base * YUZU_PAGESIZE);
 
-            while (base != end) {
-                page_table.entries[base].ptr.Store(0, type);
-                page_table.entries[base].addr = 0;
-                page_table.entries[base].block = 0;
-                base += 1;
-            }
+            page_table.entries.ZeroRegion(base, end);
         } else {
-            auto orig_base = base;
-            while (base != end) {
-                auto host_ptr = uintptr_t(system.DeviceMemory().GetPointer<u8>(target)) - (base << YUZU_PAGEBITS);
-                auto backing = GetInteger(target) - (base << YUZU_PAGEBITS);
-                page_table.entries[base].ptr.Store(host_ptr, type);
-                page_table.entries[base].addr = backing;
-                page_table.entries[base].block = orig_base << YUZU_PAGEBITS;
+            auto current_block = block_count.fetch_add(1, std::memory_order_relaxed);
+            ASSERT(current_block != 65535);
 
-                ASSERT_MSG(page_table.entries[base].ptr.Pointer(),
+            page_table.entries.CommitRegion(base, end);
+            while (base != end) {
+                auto host_ptr = reinterpret_cast<u64>(system.DeviceMemory().GetPointer<u8>(target)) - (base << YUZU_PAGEBITS);;
+                auto& entry = page_table.entries.GetUnchecked(base);
+
+                entry.Store(false, type, current_block, host_ptr);
+                ASSERT_MSG(page_table.entries[base].Pointer(),
                            "memory mapping base yield a nullptr within the table");
 
                 base += 1;
@@ -569,11 +571,11 @@ struct Memory::Impl {
         vaddr &= 0xffffffffffffULL;
         if (AddressSpaceContains(*current_page_table, vaddr, 1)) [[likely]] {
             // Avoid adding any extra logic to this fast-path block
-            const uintptr_t raw_pointer = current_page_table->entries[vaddr >> YUZU_PAGEBITS].ptr.Raw();
-            if (const uintptr_t pointer = Common::PageTable::PageInfo::ExtractPointer(raw_pointer)) [[likely]] {
+            const auto raw = current_page_table->entries[vaddr >> YUZU_PAGEBITS].Raw();
+            if (auto pointer = Common::PageTable::PageEntryData::ExtractPointer(raw); pointer) [[likely]] {
                 return reinterpret_cast<u8*>(pointer + vaddr);
             } else {
-                switch (Common::PageTable::PageInfo::ExtractType(raw_pointer)) {
+                switch (static_cast<Common::PageType>(raw.type)) {
                 case Common::PageType::Memory:
                     ASSERT_MSG(false, "Mapped memory page without a pointer @ {:#016x}", vaddr);
                     return nullptr;
@@ -773,6 +775,7 @@ struct Memory::Impl {
 #else
     Common::HostMemory* host_buffer{};
 #endif
+    std::atomic<u16> block_count = 0;
 };
 
 Memory::Memory(Core::System& system_) : system{system_} {
@@ -811,7 +814,7 @@ bool Memory::IsValidVirtualAddress(const Common::ProcessAddress vaddr) const {
     if (page >= page_table.entries.size()) {
         return false;
     }
-    const auto [pointer, type] = page_table.entries[page].ptr.PointerType();
+    const auto [pointer, type, _] = page_table.entries[page].PointerTypeBlock();
     return pointer != 0 || type == Common::PageType::RasterizerCachedMemory ||
            type == Common::PageType::DebugMemory;
 }

src/dynarmic/src/dynarmic/backend/arm64/a32_address_space.cpp

@@ -371,8 +371,10 @@ EmitConfig A32AddressSpace::GetEmitConfig() {
 
         .page_table_pointer = std::bit_cast<u64>(conf.page_table),
         .page_table_address_space_bits = 32,
-        .page_table_pointer_mask_bits = conf.page_table_pointer_mask_bits,
+        .page_table_pointer_mask = conf.page_table_pointer_mask,
         .page_table_log2_stride = conf.page_table_log2_stride,
+        .page_table_marked_bit = conf.page_table_marked_bit,
+        .page_table_sign_extension = conf.page_table_sign_extension,
         .silently_mirror_page_table = true,
         .absolute_offset_page_table = conf.absolute_offset_page_table,
         .detect_misaligned_access_via_page_table = conf.detect_misaligned_access_via_page_table,

src/dynarmic/src/dynarmic/backend/arm64/a64_address_space.cpp

@@ -545,8 +545,10 @@ EmitConfig A64AddressSpace::GetEmitConfig() {
 
         .page_table_pointer = std::bit_cast<u64>(conf.page_table),
         .page_table_address_space_bits = conf.page_table_address_space_bits,
-        .page_table_pointer_mask_bits = conf.page_table_pointer_mask_bits,
+        .page_table_pointer_mask = conf.page_table_pointer_mask,
         .page_table_log2_stride = conf.page_table_log2_stride,
+        .page_table_marked_bit = conf.page_table_marked_bit,
+        .page_table_sign_extension = conf.page_table_sign_extension,
         .silently_mirror_page_table = conf.silently_mirror_page_table,
         .absolute_offset_page_table = conf.absolute_offset_page_table,
         .detect_misaligned_access_via_page_table = conf.detect_misaligned_access_via_page_table,

src/dynarmic/src/dynarmic/backend/arm64/emit_arm64.h

@@ -128,8 +128,10 @@ struct EmitConfig {
     // Page table
     u64 page_table_pointer;
     std::size_t page_table_address_space_bits;
-    int page_table_pointer_mask_bits;
+    u64 page_table_pointer_mask;
     std::size_t page_table_log2_stride;
+    std::optional<std::uint8_t> page_table_marked_bit;
+    std::optional<std::uint8_t> page_table_sign_extension;
     bool silently_mirror_page_table;
     bool absolute_offset_page_table;
     u8 detect_misaligned_access_via_page_table;

src/dynarmic/src/dynarmic/backend/arm64/emit_arm64_memory.cpp

@@ -273,9 +273,18 @@ std::pair<oaknut::XReg, oaknut::XReg> InlinePageTableEmitVAddrLookup(oaknut::Cod
     // load x0 = *<(u8*)pagetable + index>
     code.LDR(Xscratch0, Xpagetable, Xscratch0);
 
-    if (ctx.conf.page_table_pointer_mask_bits != 0) {
-        const u64 mask = u64(~u64(0)) << ctx.conf.page_table_pointer_mask_bits;
-        code.AND(Xscratch0, Xscratch0, mask);
+    if (ctx.conf.page_table_marked_bit) {
+        // check for marked bit
+        code.TBNZ(Xscratch0, *ctx.conf.page_table_marked_bit, *fallback);
+    }
+
+    if (ctx.conf.page_table_pointer_mask != 0) {
+        code.AND(Xscratch0, Xscratch0, ctx.conf.page_table_pointer_mask);
+    }
+
+    // TODO: combine this with page_table_pointer_mask
+    if (ctx.conf.page_table_sign_extension) {
+        code.SBFM(Xscratch0, Xscratch0, 0, *ctx.conf.page_table_sign_extension);
     }
 
     code.CBZ(Xscratch0, *fallback);

src/dynarmic/src/dynarmic/backend/x64/emit_x64_memory.h

@@ -9,6 +9,7 @@
 #pragma once
 
 #include <bit>
+#include <utility>
 #include "dynarmic/backend/x64/xbyak.h"
 
 #include "dynarmic/backend/x64/a32_emit_x64.h"
@@ -78,27 +79,40 @@ Xbyak::RegExp EmitVAddrLookup(BlockOfCode& code, EmitContext& ctx, size_t bitsiz
 template<>
 [[maybe_unused]] Xbyak::RegExp EmitVAddrLookup<A32EmitContext>(BlockOfCode& code, A32EmitContext& ctx, size_t bitsize, Xbyak::Label& abort, Xbyak::Reg64 vaddr) {
     const Xbyak::Reg64 page = ctx.reg_alloc.ScratchGpr(code);
-    const Xbyak::Reg32 tmp = ctx.conf.absolute_offset_page_table ? page.cvt32() : ctx.reg_alloc.ScratchGpr(code).cvt32();
+    const Xbyak::Reg64 tmp = ctx.conf.absolute_offset_page_table && ctx.conf.page_table_pointer_mask == 0 ? page : ctx.reg_alloc.ScratchGpr(code);
 
-    EmitDetectMisalignedVAddr(code, ctx, bitsize, abort, vaddr, tmp.cvt64());
+    EmitDetectMisalignedVAddr(code, ctx, bitsize, abort, vaddr, tmp);
 
-    // TODO: This code assumes vaddr has been zext from 32-bits to 64-bits.
-
-    code.mov(tmp, vaddr.cvt32());
+    code.mov(tmp, vaddr);
     code.shr(tmp, int(page_table_const_bits));
     code.shl(tmp, int(ctx.conf.page_table_log2_stride));
     code.mov(page, qword[r14 + tmp.cvt64()]);
-    if (ctx.conf.page_table_pointer_mask_bits == 0) {
+
+    // check for marked bit, use as unmapped if marked
+	if (ctx.conf.page_table_marked_bit) {
+        code.bt(page, *ctx.conf.page_table_marked_bit);
+        code.jc(abort, code.T_NEAR);
+	}
+    // mask away attributes
+    if (ctx.conf.page_table_pointer_mask == 0) {
         code.test(page, page);
+    } else if (std::in_range<s32>(ctx.conf.page_table_pointer_mask)) {
+        code.and_(page, ctx.conf.page_table_pointer_mask);
     } else {
-        code.and_(page, ~u32(0) << ctx.conf.page_table_pointer_mask_bits);
+        code.mov(tmp, ctx.conf.page_table_pointer_mask);
+        code.and_(page, tmp);
     }
+    if (ctx.conf.page_table_sign_extension) {
+        code.shl(page, *ctx.conf.page_table_sign_extension);
+        code.sar(page, *ctx.conf.page_table_sign_extension);
+    }
+
     code.jz(abort, code.T_NEAR);
     if (ctx.conf.absolute_offset_page_table) {
         return page + vaddr;
     }
-    code.mov(tmp, vaddr.cvt32());
-    code.and_(tmp, static_cast<u32>(page_table_const_mask));
+    code.mov(tmp, vaddr);
+    code.and_(tmp, u32(page_table_const_mask));
     return page + tmp.cvt64();
 }
 
@@ -108,7 +122,7 @@ template<>
     const size_t unused_top_bits = 64 - ctx.conf.page_table_address_space_bits;
 
     const Xbyak::Reg64 page = ctx.reg_alloc.ScratchGpr(code);
-    const Xbyak::Reg64 tmp = ctx.conf.absolute_offset_page_table ? page : ctx.reg_alloc.ScratchGpr(code);
+    const Xbyak::Reg64 tmp = ctx.conf.absolute_offset_page_table && ctx.conf.page_table_pointer_mask == 0 ? page : ctx.reg_alloc.ScratchGpr(code);
 
     EmitDetectMisalignedVAddr(code, ctx, bitsize, abort, vaddr, tmp);
 
@@ -143,11 +157,26 @@ template<>
 
     code.shl(tmp, int(ctx.conf.page_table_log2_stride));
     code.mov(page, qword[r14 + tmp]);
-    if (ctx.conf.page_table_pointer_mask_bits == 0) {
-        code.test(page, page);
-    } else {
-        code.and_(page, ~u32(0) << ctx.conf.page_table_pointer_mask_bits);
+
+    // check for marked bit, use as unmapped if marked
+    if (ctx.conf.page_table_marked_bit) {
+        code.bt(page, *ctx.conf.page_table_marked_bit);
+        code.jc(abort, code.T_NEAR);
     }
+    // mask away attributes
+    if (ctx.conf.page_table_pointer_mask == 0) {
+        code.test(page, page);
+    } else if (std::in_range<s32>(ctx.conf.page_table_pointer_mask)) {
+        code.and_(page, ctx.conf.page_table_pointer_mask);
+    } else {
+        code.mov(tmp, ctx.conf.page_table_pointer_mask);
+        code.and_(page, tmp);
+    }
+    if (ctx.conf.page_table_sign_extension) {
+        code.shl(page, *ctx.conf.page_table_sign_extension);
+        code.sar(page, *ctx.conf.page_table_sign_extension);
+    }
+
     code.jz(abort, code.T_NEAR);
     if (ctx.conf.absolute_offset_page_table) {
         return page + vaddr;

src/dynarmic/src/dynarmic/interface/A32/config.h

@@ -159,14 +159,23 @@ struct UserConfig {
     /// Maximum size is limited by the maximum length of a x86_64 / arm64 jump.
     std::uint32_t code_cache_size = 128 * 1024 * 1024;  // bytes
 
-    /// Masks out the first N bits in host pointers from the page table.
+    /// Applies a bit mask to the bits in host pointers from the page table.
     /// The intention behind this is to allow users of Dynarmic to pack attributes in the
     /// same integer and update the pointer attribute pair atomically.
-    /// If the configured value is 3, all pointers will be forcefully aligned to 8 bytes.
-    std::int32_t page_table_pointer_mask_bits = 0;
+    /// If the configured value is ~(0b111ULL), all pointers will be forcefully aligned to 8 bytes.
+    std::uint64_t page_table_pointer_mask = 0;
 
-    // Log2 of the size per page entry, value should be either 3 or 4
-    std::size_t page_table_log2_stride = 3;
+    /// Log2 of the size per page entry, value should be either 3 or 4
+    std::uint32_t page_table_log2_stride = 3;
+
+    /// Setting this value has Dynarmic check the specified bit of the page pointer provided by page table.
+    /// If the bit is set to 1, Dynarmic will treat it as unmapped.
+    /// This bit should be included as part of `page_table_pointer_mask_bits`.
+    std::optional<std::uint8_t> page_table_marked_bit = std::nullopt;
+
+    /// If this value is set, Dynarmic will sign extend the page table pointer by this bit.
+    /// Useful for compacting bits into the page table and should be used as part of `page_table_pointer_mask`.
+    std::optional<std::uint8_t> page_table_sign_extension = std::nullopt;
 
     /// Select the architecture version to use.
     /// There are minor behavioural differences between versions.

src/dynarmic/src/dynarmic/interface/A64/config.h

@@ -173,14 +173,23 @@ struct UserConfig {
     /// This is only used if page_table is not nullptr.
     std::uint32_t page_table_address_space_bits = 36;
 
-    /// Masks out the first N bits in host pointers from the page table.
+    /// Applies a bit mask to the bits in host pointers from the page table.
     /// The intention behind this is to allow users of Dynarmic to pack attributes in the
     /// same integer and update the pointer attribute pair atomically.
-    /// If the configured value is 3, all pointers will be forcefully aligned to 8 bytes.
-    std::int32_t page_table_pointer_mask_bits = 0;
+    /// If the configured value is ~(0b111ULL), all pointers will be forcefully aligned to 8 bytes.
+    std::uint64_t page_table_pointer_mask = 0;
 
-    // Log2 of the size per page entry, value should be either 3 or 4
-    std::size_t page_table_log2_stride = 3;
+    /// Log2 of the size per page entry, value should be either 3 or 4
+    std::uint32_t page_table_log2_stride = 3;
+
+    /// Setting this value has Dynarmic check the specified bit of the page pointer provided by page table.
+    /// If the bit is set to 1, Dynarmic will treat it as unmapped.
+    /// This bit should be included as part of `page_table_pointer_mask`.
+    std::optional<std::uint8_t> page_table_marked_bit = std::nullopt;
+
+    /// If this value is set, Dynarmic will sign extend the page table pointer by this bit.
+    /// Useful for compacting bits into the page table and should be used as part of `page_table_pointer_mask`.
+    std::optional<std::uint8_t> page_table_sign_extension = std::nullopt;
 
     /// Counter-timer frequency register. The value of the register is not interpreted by
     /// dynarmic.

src/dynarmic/tests/A32/fuzz_arm.cpp

@@ -321,8 +321,6 @@ static void RunTestInstance(Dynarmic::A32::Jit& jit,
     uni_env.PadCodeMem();
     jit_env.modified_memory.clear();
     uni_env.modified_memory.clear();
-    jit_env.interrupts.clear();
-    uni_env.interrupts.clear();
 
     jit.Regs() = regs;
     jit.ExtRegs() = vecs;

src/dynarmic/tests/A64/fuzz_with_unicorn.cpp

@@ -176,8 +176,6 @@ static void RunTestInstance(Dynarmic::A64::Jit& jit, A64Unicorn& uni, A64TestEnv
     uni_env.code_mem_start_address = instructions_start;
     jit_env.modified_memory.clear();
     uni_env.modified_memory.clear();
-    jit_env.interrupts.clear();
-    uni_env.interrupts.clear();
 
     const u64 initial_sp = RandInt<u64>(0x30'0000'0000, 0x40'0000'0000) * 4;
 

src/dynarmic/tests/print_info.cpp

@@ -282,20 +282,17 @@ int main(int argc, char** argv) {
         return 1;
     }
 
-    const char* const hex_instruction = [argv] {
-        if (strlen(argv[2]) > 2 && argv[2][0] == '0' && argv[2][1] == 'x') {
-            return argv[2] + 2;
-        }
-        return argv[2];
+    const char* const hex_instruction = [s = argv[2]] {
+        if (strlen(s) > 2 && s[0] == '0' && s[1] == 'x')
+            return s + 2;
+        return s;
     }();
-
     if (strlen(hex_instruction) > 8) {
         fmt::print("hex string too long\n");
         return 1;
     }
 
     const u32 instruction = strtol(hex_instruction, nullptr, 16);
-
     if (strcmp(argv[1], "a32") == 0) {
         PrintA32Instruction(instruction);
     } else if (strcmp(argv[1], "a64") == 0) {
@@ -306,13 +303,11 @@ int main(int argc, char** argv) {
         fmt::print("Invalid mode: {}\nValid values: a32, a64, thumb\n", argv[1]);
         return 1;
     }
-
     if (argc == 4) {
         if (strcmp(argv[3], "-exec") != 0) {
             fmt::print("Invalid option {}\n", argv[3]);
             return 1;
         }
-
         if (strcmp(argv[1], "a32") == 0) {
             ExecuteA32Instruction(instruction);
         } else {
@@ -320,6 +315,5 @@ int main(int argc, char** argv) {
             return 1;
         }
     }
-
     return 0;
 }

src/dynarmic/tests/test_generator.cpp

@@ -354,6 +354,8 @@ u32 GenRandomA64Inst(u64 pc, bool is_last_inst) {
             "MSR_reg",
             "MSR_imm",
             "MRS",
+            // generates F16
+            "FCADD_vec",
         };
 
         for (const auto& [fn, bitstring] : list) {
@@ -440,7 +442,6 @@ void RunTestInstance(Dynarmic::A32::Jit& jit,
         std::copy(instructions.begin(), instructions.end(), jit_env.code_mem.begin() + num_words);
         jit_env.PadCodeMem();
         jit_env.modified_memory.clear();
-        jit_env.interrupts.clear();
 
         jit.Regs() = regs;
         jit.ExtRegs() = vecs;
@@ -510,7 +511,6 @@ void RunTestInstance(Dynarmic::A64::Jit& jit,
         jit_env.code_mem.emplace_back(0x14000000);  // B .
         jit_env.code_mem_start_address = start_address;
         jit_env.modified_memory.clear();
-        jit_env.interrupts.clear();
 
         jit.SetRegisters(regs);
         jit.SetVectors(vecs);
@@ -536,36 +536,52 @@ void RunTestInstance(Dynarmic::A64::Jit& jit,
         fmt::print("initial_regs:");
         for (u64 i : regs)
             fmt::print(" {:016x}", i);
-        fmt::print("\n");
-        fmt::print("initial_vecs:");
+        fmt::print(
+            "\n"
+            "initial_vecs:"
+        );
         for (auto i : vecs)
             fmt::print(" {:016x}:{:016x}", i[0], i[1]);
-        fmt::print("\n");
-        fmt::print("initial_sp: {:016x}\n", initial_sp);
-        fmt::print("initial_pstate: {:08x}\n", pstate);
-        fmt::print("initial_fpcr: {:08x}\n", fpcr);
-        fmt::print("final_regs:");
+        fmt::print(
+            "\n"
+            "initial_sp: {:016x}\n"
+            "initial_pstate: {:08x}\n"
+            "initial_fpcr: {:08x}\n"
+            "final_regs:"
+            , initial_sp
+            , pstate
+            , fpcr
+        );
         for (u64 i : jit.GetRegisters())
             fmt::print(" {:016x}", i);
-        fmt::print("\n");
-        fmt::print("final_vecs:");
+        fmt::print(
+            "\n"
+            "final_vecs:"
+        );
         for (auto i : jit.GetVectors())
             fmt::print(" {:016x}:{:016x}", i[0], i[1]);
-        fmt::print("\n");
-        fmt::print("final_sp: {:016x}\n", jit.GetSP());
-        fmt::print("final_pc: {:016x}\n", jit.GetPC());
-        fmt::print("final_pstate: {:08x}\n", jit.GetPstate());
-        fmt::print("final_fpcr: {:08x}\n", jit.GetFpcr());
-        fmt::print("final_qc : {}\n", FP::FPSR{jit.GetFpsr()}.QC());
-        fmt::print("mod_mem:");
+        fmt::print(
+            "\n"
+            "final_sp: {:016x}\n"
+            "final_pc: {:016x}\n"
+            "final_pstate: {:08x}\n"
+            "final_fpcr: {:08x}\n"
+            "final_qc : {}\n"
+            "mod_mem:"
+            , jit.GetSP()
+            , jit.GetPC()
+            , jit.GetPstate()
+            , jit.GetFpcr()
+            , FP::FPSR{jit.GetFpsr()}.QC()
+        );
         for (auto [addr, value] : jit_env.modified_memory)
             fmt::print(" {:08x}:{:02x}", addr, value);
-        fmt::print("\n");
-        fmt::print("interrupts:\n");
-        for (const auto& i : jit_env.interrupts)
-            std::puts(i.c_str());
-        fmt::print("===\n");
-        fmt::print("{}", jit.Disassemble());
+        fmt::print(
+            "\n"
+            "===\n"
+            "{}"
+            , jit.Disassemble()
+        );
     }
 }
 

src/dynarmic/tests/test_reader.cpp

@@ -99,7 +99,6 @@ void RunTestInstance(Dynarmic::A32::Jit& jit,
         std::copy(instructions.begin(), instructions.end(), jit_env.code_mem.begin() + num_words);
         jit_env.PadCodeMem();
         jit_env.modified_memory.clear();
-        jit_env.interrupts.clear();
 
         jit.Regs() = regs;
         jit.ExtRegs() = vecs;
@@ -192,7 +191,6 @@ void RunTestInstance(A64::Jit& jit,
         jit_env.code_mem.emplace_back(0x14000000);  // B .
         jit_env.code_mem_start_address = start_address;
         jit_env.modified_memory.clear();
-        jit_env.interrupts.clear();
 
         jit.SetRegisters(regs);
         jit.SetVectors(vecs);

src/frontend_common/config.cpp

@@ -299,12 +299,17 @@ void Config::ReadDataStorageValues() {
 void Config::ReadDebuggingValues() {
     BeginGroup(Settings::TranslateCategory(Settings::Category::Debugging));
 
-    // Intentionally not using the QT default setting as this is intended to be changed in the ini
-    Settings::values.record_frame_times =
-        ReadBooleanSetting(std::string("record_frame_times"), std::make_optional(false));
+    if (global) {
+        // Intentionally not using the QT default setting as this is intended to be changed in the ini
+        Settings::values.record_frame_times =
+            ReadBooleanSetting(std::string("record_frame_times"), std::make_optional(false));
 
-    ReadCategory(Settings::Category::Debugging);
-    ReadCategory(Settings::Category::DebuggingGraphics);
+        ReadCategory(Settings::Category::Debugging);
+        ReadCategory(Settings::Category::DebuggingGraphics);
+    } else {
+        ReadSettingGeneric(&Settings::values.program_args);
+        ReadSettingGeneric(&Settings::values.debug_knobs);
+    }
 
     EndGroup();
 }
@@ -415,12 +420,12 @@ void Config::ReadLibraryAppletValues() {
 void Config::ReadValues() {
     if (global) {
         ReadDataStorageValues();
-        ReadDebuggingValues();
         ReadDisabledAddOnValues();
         ReadServiceValues();
         ReadWebServiceValues();
         ReadMiscellaneousValues();
     }
+    ReadDebuggingValues();
     ReadLibraryAppletValues();
     ReadNetworkValues();
     ReadControlValues();
@@ -511,13 +516,13 @@ void Config::SaveValues() {
     if (global) {
         LOG_DEBUG(Config, "Saving global generic configuration values");
         SaveDataStorageValues();
-        SaveDebuggingValues();
         SaveDisabledAddOnValues();
         SaveWebServiceValues();
         SaveMiscellaneousValues();
     } else {
         LOG_DEBUG(Config, "Saving only generic configuration values");
     }
+    SaveDebuggingValues();
     SaveLibraryAppletValues();
     SaveNetworkValues();
     SaveControlValues();
@@ -600,11 +605,16 @@ void Config::SaveDataStorageValues() {
 void Config::SaveDebuggingValues() {
     BeginGroup(Settings::TranslateCategory(Settings::Category::Debugging));
 
-    // Intentionally not using the QT default setting as this is intended to be changed in the ini
-    WriteBooleanSetting(std::string("record_frame_times"), Settings::values.record_frame_times);
+    if (global) {
+        // Intentionally not using the QT default setting as this is intended to be changed in the ini
+        WriteBooleanSetting(std::string("record_frame_times"), Settings::values.record_frame_times);
 
-    WriteCategory(Settings::Category::Debugging);
-    WriteCategory(Settings::Category::DebuggingGraphics);
+        WriteCategory(Settings::Category::Debugging);
+        WriteCategory(Settings::Category::DebuggingGraphics);
+    } else {
+        WriteSettingGeneric(&Settings::values.program_args);
+        WriteSettingGeneric(&Settings::values.debug_knobs);
+    }
 
     EndGroup();
 }

src/hid_core/frontend/emulated_console.cpp

@@ -170,12 +170,12 @@ void EmulatedConsole::SetMotion(const Common::Input::CallbackStatus& callback) {
     auto& emulated = console.motion_values.emulated;
 
     raw_status = TransformToMotion(callback);
-    emulated.SetAcceleration(Common::Vec3f{
+    emulated.SetAcceleration(Common::Vec<f32, 3>{
         raw_status.accel.x.value,
         raw_status.accel.y.value,
         raw_status.accel.z.value,
     });
-    emulated.SetGyroscope(Common::Vec3f{
+    emulated.SetGyroscope(Common::Vec<f32, 3>{
         raw_status.gyro.x.value,
         raw_status.gyro.y.value,
         raw_status.gyro.z.value,

src/hid_core/frontend/emulated_console.h

@@ -18,7 +18,6 @@
 #include "common/input.h"
 #include "common/param_package.h"
 #include "common/point.h"
-#include "common/quaternion.h"
 #include "common/vector_math.h"
 #include "hid_core/frontend/motion_input.h"
 #include "hid_core/hid_types.h"
@@ -43,12 +42,12 @@ using TouchValues = std::array<Common::Input::TouchStatus, MaxTouchDevices>;
 
 // Contains all motion related data that is used on the services
 struct ConsoleMotion {
-    Common::Vec3f accel{};
-    Common::Vec3f gyro{};
-    Common::Vec3f rotation{};
-    std::array<Common::Vec3f, 3> orientation{};
-    Common::Quaternion<f32> quaternion{};
-    Common::Vec3f gyro_bias{};
+    Common::Vec<f32, 3> accel{};
+    Common::Vec<f32, 3> gyro{};
+    Common::Vec<f32, 3> rotation{};
+    std::array<Common::Vec<f32, 3>, 3> orientation{};
+    Common::Vec<f32, 4> quaternion{};
+    Common::Vec<f32, 3> gyro_bias{};
     f32 verticalization_error{};
     bool is_at_rest{};
 };

src/hid_core/frontend/emulated_controller.cpp

@@ -1051,12 +1051,12 @@ void EmulatedController::SetMotion(const Common::Input::CallbackStatus& callback
     auto& emulated = controller.motion_values[index].emulated;
 
     raw_status = TransformToMotion(callback);
-    emulated.SetAcceleration(Common::Vec3f{
+    emulated.SetAcceleration(Common::Vec<f32, 3>{
         raw_status.accel.x.value,
         raw_status.accel.y.value,
         raw_status.accel.z.value,
     });
-    emulated.SetGyroscope(Common::Vec3f{
+    emulated.SetGyroscope(Common::Vec<f32, 3>{
         raw_status.gyro.x.value,
         raw_status.gyro.y.value,
         raw_status.gyro.z.value,

src/hid_core/frontend/emulated_controller.h

@@ -107,11 +107,11 @@ struct RingSensorForce {
 using NfcState = Common::Input::NfcStatus;
 
 struct ControllerMotion {
-    Common::Vec3f accel{};
-    Common::Vec3f gyro{};
-    Common::Vec3f rotation{};
-    Common::Vec3f euler{};
-    std::array<Common::Vec3f, 3> orientation{};
+    Common::Vec<f32, 3> accel{};
+    Common::Vec<f32, 3> gyro{};
+    Common::Vec<f32, 3> rotation{};
+    Common::Vec<f32, 3> euler{};
+    std::array<Common::Vec<f32, 3>, 3> orientation{};
     bool is_at_rest{};
 };
 

src/hid_core/frontend/motion_input.cpp

@@ -26,20 +26,19 @@ void MotionInput::SetPID(f32 new_kp, f32 new_ki, f32 new_kd) {
     kd = new_kd;
 }
 
-void MotionInput::SetAcceleration(const Common::Vec3f& acceleration) {
+void MotionInput::SetAcceleration(const Common::Vec<f32, 3>& acceleration) {
     accel = acceleration;
-
-    accel.x = std::clamp(accel.x, -AccelMaxValue, AccelMaxValue);
-    accel.y = std::clamp(accel.y, -AccelMaxValue, AccelMaxValue);
-    accel.z = std::clamp(accel.z, -AccelMaxValue, AccelMaxValue);
+    accel[0] = std::clamp(accel[0], -AccelMaxValue, AccelMaxValue);
+    accel[1] = std::clamp(accel[1], -AccelMaxValue, AccelMaxValue);
+    accel[2] = std::clamp(accel[2], -AccelMaxValue, AccelMaxValue);
 }
 
-void MotionInput::SetGyroscope(const Common::Vec3f& gyroscope) {
+void MotionInput::SetGyroscope(const Common::Vec<f32, 3>& gyroscope) {
     gyro = gyroscope - gyro_bias;
 
-    gyro.x = std::clamp(gyro.x, -GyroMaxValue, GyroMaxValue);
-    gyro.y = std::clamp(gyro.y, -GyroMaxValue, GyroMaxValue);
-    gyro.z = std::clamp(gyro.z, -GyroMaxValue, GyroMaxValue);
+    gyro[0] = std::clamp(gyro[0], -GyroMaxValue, GyroMaxValue);
+    gyro[1] = std::clamp(gyro[1], -GyroMaxValue, GyroMaxValue);
+    gyro[2] = std::clamp(gyro[2], -GyroMaxValue, GyroMaxValue);
 
     // Auto adjust gyro_bias to minimize drift
     if (!IsMoving(IsAtRestRelaxed)) {
@@ -59,25 +58,25 @@ void MotionInput::SetGyroscope(const Common::Vec3f& gyroscope) {
     }
 }
 
-void MotionInput::SetQuaternion(const Common::Quaternion<f32>& quaternion) {
+void MotionInput::SetQuaternion(const Common::Vec<f32, 4>& quaternion) {
     quat = quaternion;
 }
 
-void MotionInput::SetEulerAngles(const Common::Vec3f& euler_angles) {
-    const float cr = std::cos(euler_angles.x * 0.5f);
-    const float sr = std::sin(euler_angles.x * 0.5f);
-    const float cp = std::cos(euler_angles.y * 0.5f);
-    const float sp = std::sin(euler_angles.y * 0.5f);
-    const float cy = std::cos(euler_angles.z * 0.5f);
-    const float sy = std::sin(euler_angles.z * 0.5f);
+void MotionInput::SetEulerAngles(const Common::Vec<f32, 3>& euler_angles) {
+    const float cr = std::cos(euler_angles[0] * 0.5f);
+    const float sr = std::sin(euler_angles[0] * 0.5f);
+    const float cp = std::cos(euler_angles[1] * 0.5f);
+    const float sp = std::sin(euler_angles[1] * 0.5f);
+    const float cy = std::cos(euler_angles[2] * 0.5f);
+    const float sy = std::sin(euler_angles[2] * 0.5f);
 
-    quat.w = cr * cp * cy + sr * sp * sy;
-    quat.xyz.x = sr * cp * cy - cr * sp * sy;
-    quat.xyz.y = cr * sp * cy + sr * cp * sy;
-    quat.xyz.z = cr * cp * sy - sr * sp * cy;
+    quat[3] = cr * cp * cy + sr * sp * sy;
+    quat[0] = sr * cp * cy - cr * sp * sy;
+    quat[1] = cr * sp * cy + sr * cp * sy;
+    quat[2] = cr * cp * sy - sr * sp * cy;
 }
 
-void MotionInput::SetGyroBias(const Common::Vec3f& bias) {
+void MotionInput::SetGyroBias(const Common::Vec<f32, 3>& bias) {
     gyro_bias = bias;
 }
 
@@ -98,7 +97,7 @@ void MotionInput::ResetRotations() {
 }
 
 void MotionInput::ResetQuaternion() {
-    quat = {{0.0f, 0.0f, -1.0f}, 0.0f};
+    quat = Common::Vec<f32, 4>{0.0f, 0.0f, -1.0f, 0.0f};
 }
 
 bool MotionInput::IsMoving(f32 sensitivity) const {
@@ -137,10 +136,10 @@ void MotionInput::UpdateOrientation(u64 elapsed_time) {
         ResetOrientation();
     }
     // Short name local variable for readability
-    f32 q1 = quat.w;
-    f32 q2 = quat.xyz[0];
-    f32 q3 = quat.xyz[1];
-    f32 q4 = quat.xyz[2];
+    f32 q1 = quat[3];
+    f32 q2 = quat[0];
+    f32 q3 = quat[1];
+    f32 q4 = quat[2];
     const auto sample_period = static_cast<f32>(elapsed_time) / 1000000.0f;
 
     // Ignore invalid elapsed time
@@ -150,23 +149,23 @@ void MotionInput::UpdateOrientation(u64 elapsed_time) {
 
     const auto normal_accel = accel.Normalized();
     auto rad_gyro = gyro * std::numbers::pi_v<float> * 2.f;
-    const f32 swap = rad_gyro.x;
-    rad_gyro.x = rad_gyro.y;
-    rad_gyro.y = -swap;
-    rad_gyro.z = -rad_gyro.z;
+    const f32 swap = rad_gyro[0];
+    rad_gyro[0] = rad_gyro[1];
+    rad_gyro[1] = -swap;
+    rad_gyro[2] = -rad_gyro[2];
 
     // Clear gyro values if there is no gyro present
     if (only_accelerometer) {
-        rad_gyro.x = 0;
-        rad_gyro.y = 0;
-        rad_gyro.z = 0;
+        rad_gyro[0] = 0;
+        rad_gyro[1] = 0;
+        rad_gyro[2] = 0;
     }
 
     // Ignore drift correction if acceleration is not reliable
     if (accel.Length() >= 0.75f && accel.Length() <= 1.25f) {
-        const f32 ax = -normal_accel.x;
-        const f32 ay = normal_accel.y;
-        const f32 az = -normal_accel.z;
+        const f32 ax = -normal_accel[0];
+        const f32 ay = normal_accel[1];
+        const f32 az = -normal_accel[2];
 
         // Estimated direction of gravity
         const f32 vx = 2.0f * (q2 * q4 - q1 * q3);
@@ -174,7 +173,7 @@ void MotionInput::UpdateOrientation(u64 elapsed_time) {
         const f32 vz = q1 * q1 - q2 * q2 - q3 * q3 + q4 * q4;
 
         // Error is cross product between estimated direction and measured direction of gravity
-        const Common::Vec3f new_real_error = {
+        const Common::Vec<f32, 3> new_real_error{
             az * vx - ax * vz,
             ay * vz - az * vy,
             ax * vy - ay * vx,
@@ -202,16 +201,16 @@ void MotionInput::UpdateOrientation(u64 elapsed_time) {
             rad_gyro += 10.0f * kd * derivative_error;
 
             // Emulate gyro values for games that need them
-            gyro.x = -rad_gyro.y;
-            gyro.y = rad_gyro.x;
-            gyro.z = -rad_gyro.z;
+            gyro[0] = -rad_gyro[1];
+            gyro[1] = rad_gyro[0];
+            gyro[2] = -rad_gyro[2];
             UpdateRotation(elapsed_time);
         }
     }
 
-    const f32 gx = rad_gyro.y;
-    const f32 gy = rad_gyro.x;
-    const f32 gz = rad_gyro.z;
+    const f32 gx = rad_gyro[1];
+    const f32 gy = rad_gyro[0];
+    const f32 gz = rad_gyro[2];
 
     // Integrate rate of change of quaternion
     const f32 pa = q2;
@@ -222,57 +221,58 @@ void MotionInput::UpdateOrientation(u64 elapsed_time) {
     q3 = pb + (q1 * gy - pa * gz + pc * gx) * (0.5f * sample_period);
     q4 = pc + (q1 * gz + pa * gy - pb * gx) * (0.5f * sample_period);
 
-    quat.w = q1;
-    quat.xyz[0] = q2;
-    quat.xyz[1] = q3;
-    quat.xyz[2] = q4;
+    quat[3] = q1;
+    quat[0] = q2;
+    quat[1] = q3;
+    quat[2] = q4;
     quat = quat.Normalized();
 }
 
-std::array<Common::Vec3f, 3> MotionInput::GetOrientation() const {
-    const Common::Quaternion<float> quad{
-        .xyz = {-quat.xyz[1], -quat.xyz[0], -quat.w},
-        .w = -quat.xyz[2],
+std::array<Common::Vec<f32, 3>, 3> MotionInput::GetOrientation() const {
+    const Common::Vec<f32, 4> quad{
+        -quat[1],
+        -quat[0],
+        -quat[3],
+        -quat[2],
     };
-    const std::array<float, 16> matrix4x4 = quad.ToMatrix();
-
-    return {Common::Vec3f(matrix4x4[0], matrix4x4[1], -matrix4x4[2]),
-            Common::Vec3f(matrix4x4[4], matrix4x4[5], -matrix4x4[6]),
-            Common::Vec3f(-matrix4x4[8], -matrix4x4[9], matrix4x4[10])};
+    const std::array<f32, 16> matrix4x4 = quad.ToMatrix();
+    return {Common::Vec<f32, 3>(matrix4x4[0], matrix4x4[1], -matrix4x4[2]),
+            Common::Vec<f32, 3>(matrix4x4[4], matrix4x4[5], -matrix4x4[6]),
+            Common::Vec<f32, 3>(-matrix4x4[8], -matrix4x4[9], matrix4x4[10])};
 }
 
-Common::Vec3f MotionInput::GetAcceleration() const {
+Common::Vec<f32, 3> MotionInput::GetAcceleration() const {
     return accel;
 }
 
-Common::Vec3f MotionInput::GetGyroscope() const {
+Common::Vec<f32, 3> MotionInput::GetGyroscope() const {
     return gyro;
 }
 
-Common::Vec3f MotionInput::GetGyroBias() const {
+Common::Vec<f32, 3> MotionInput::GetGyroBias() const {
     return gyro_bias;
 }
 
-Common::Quaternion<f32> MotionInput::GetQuaternion() const {
+Common::Vec<f32, 4> MotionInput::GetQuaternion() const {
     return quat;
 }
 
-Common::Vec3f MotionInput::GetRotations() const {
+Common::Vec<f32, 3> MotionInput::GetRotations() const {
     return rotations;
 }
 
-Common::Vec3f MotionInput::GetEulerAngles() const {
+Common::Vec<f32, 3> MotionInput::GetEulerAngles() const {
     // roll (x-axis rotation)
-    const float sinr_cosp = 2 * (quat.w * quat.xyz.x + quat.xyz.y * quat.xyz.z);
-    const float cosr_cosp = 1 - 2 * (quat.xyz.x * quat.xyz.x + quat.xyz.y * quat.xyz.y);
+    const float sinr_cosp = 2 * (quat[3] * quat[0] + quat[1] * quat[2]);
+    const float cosr_cosp = 1 - 2 * (quat[0] * quat[0] + quat[1] * quat[1]);
 
     // pitch (y-axis rotation)
-    const float sinp = std::sqrt(1 + 2 * (quat.w * quat.xyz.y - quat.xyz.x * quat.xyz.z));
-    const float cosp = std::sqrt(1 - 2 * (quat.w * quat.xyz.y - quat.xyz.x * quat.xyz.z));
+    const float sinp = std::sqrt(1 + 2 * (quat[3] * quat[1] - quat[0] * quat[2]));
+    const float cosp = std::sqrt(1 - 2 * (quat[3] * quat[1] - quat[0] * quat[2]));
 
     // yaw (z-axis rotation)
-    const float siny_cosp = 2 * (quat.w * quat.xyz.z + quat.xyz.x * quat.xyz.y);
-    const float cosy_cosp = 1 - 2 * (quat.xyz.y * quat.xyz.y + quat.xyz.z * quat.xyz.z);
+    const float siny_cosp = 2 * (quat[3] * quat[2] + quat[0] * quat[1]);
+    const float cosy_cosp = 1 - 2 * (quat[1] * quat[1] + quat[2] * quat[2]);
 
     return {
         std::atan2(sinr_cosp, cosr_cosp),
@@ -285,13 +285,13 @@ void MotionInput::ResetOrientation() {
     if (!reset_enabled || only_accelerometer) {
         return;
     }
-    if (!IsMoving(IsAtRestRelaxed) && accel.z <= -0.9f) {
+    if (!IsMoving(IsAtRestRelaxed) && accel[2] <= -0.9f) {
         ++reset_counter;
         if (reset_counter > 900) {
-            quat.w = 0;
-            quat.xyz[0] = 0;
-            quat.xyz[1] = 0;
-            quat.xyz[2] = -1;
+            quat[3] = 0;
+            quat[0] = 0;
+            quat[1] = 0;
+            quat[2] = -1;
             SetOrientationFromAccelerometer();
             integral_error = {};
             reset_counter = 0;
@@ -309,15 +309,15 @@ void MotionInput::SetOrientationFromAccelerometer() {
 
     while (!IsCalibrated(0.01f) && ++iterations < 100) {
         // Short name local variable for readability
-        f32 q1 = quat.w;
-        f32 q2 = quat.xyz[0];
-        f32 q3 = quat.xyz[1];
-        f32 q4 = quat.xyz[2];
+        f32 q1 = quat[3];
+        f32 q2 = quat[0];
+        f32 q3 = quat[1];
+        f32 q4 = quat[2];
 
-        Common::Vec3f rad_gyro;
-        const f32 ax = -normal_accel.x;
-        const f32 ay = normal_accel.y;
-        const f32 az = -normal_accel.z;
+        Common::Vec<f32, 3> rad_gyro;
+        const f32 ax = -normal_accel[0];
+        const f32 ay = normal_accel[1];
+        const f32 az = -normal_accel[2];
 
         // Estimated direction of gravity
         const f32 vx = 2.0f * (q2 * q4 - q1 * q3);
@@ -325,7 +325,7 @@ void MotionInput::SetOrientationFromAccelerometer() {
         const f32 vz = q1 * q1 - q2 * q2 - q3 * q3 + q4 * q4;
 
         // Error is cross product between estimated direction and measured direction of gravity
-        const Common::Vec3f new_real_error = {
+        const Common::Vec<f32, 3> new_real_error = {
             az * vx - ax * vz,
             ay * vz - az * vy,
             ax * vy - ay * vx,
@@ -338,9 +338,9 @@ void MotionInput::SetOrientationFromAccelerometer() {
         rad_gyro += 5.0f * ki * integral_error;
         rad_gyro += 10.0f * kd * derivative_error;
 
-        const f32 gx = rad_gyro.y;
-        const f32 gy = rad_gyro.x;
-        const f32 gz = rad_gyro.z;
+        const f32 gx = rad_gyro[1];
+        const f32 gy = rad_gyro[0];
+        const f32 gz = rad_gyro[2];
 
         // Integrate rate of change of quaternion
         const f32 pa = q2;
@@ -351,10 +351,10 @@ void MotionInput::SetOrientationFromAccelerometer() {
         q3 = pb + (q1 * gy - pa * gz + pc * gx) * (0.5f * sample_period);
         q4 = pc + (q1 * gz + pa * gy - pb * gx) * (0.5f * sample_period);
 
-        quat.w = q1;
-        quat.xyz[0] = q2;
-        quat.xyz[1] = q3;
-        quat.xyz[2] = q4;
+        quat[3] = q1;
+        quat[0] = q2;
+        quat[1] = q3;
+        quat[2] = q4;
         quat = quat.Normalized();
     }
 }

src/hid_core/frontend/motion_input.h

@@ -1,10 +1,12 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
 // SPDX-FileCopyrightText: Copyright 2020 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
 #pragma once
 
 #include "common/common_types.h"
-#include "common/quaternion.h"
 #include "common/vector_math.h"
 
 namespace Core::HID {
@@ -34,11 +36,11 @@ public:
     MotionInput& operator=(MotionInput&&) = default;
 
     void SetPID(f32 new_kp, f32 new_ki, f32 new_kd);
-    void SetAcceleration(const Common::Vec3f& acceleration);
-    void SetGyroscope(const Common::Vec3f& gyroscope);
-    void SetQuaternion(const Common::Quaternion<f32>& quaternion);
-    void SetEulerAngles(const Common::Vec3f& euler_angles);
-    void SetGyroBias(const Common::Vec3f& bias);
+    void SetAcceleration(const Common::Vec<f32, 3>& acceleration);
+    void SetGyroscope(const Common::Vec<f32, 3>& gyroscope);
+    void SetQuaternion(const Common::Vec<f32, 4>& quaternion);
+    void SetEulerAngles(const Common::Vec<f32, 3>& euler_angles);
+    void SetGyroBias(const Common::Vec<f32, 3>& bias);
     void SetGyroThreshold(f32 threshold);
 
     /// Applies a modifier on top of the normal gyro threshold
@@ -53,13 +55,13 @@ public:
 
     void Calibrate();
 
-    [[nodiscard]] std::array<Common::Vec3f, 3> GetOrientation() const;
-    [[nodiscard]] Common::Vec3f GetAcceleration() const;
-    [[nodiscard]] Common::Vec3f GetGyroscope() const;
-    [[nodiscard]] Common::Vec3f GetGyroBias() const;
-    [[nodiscard]] Common::Vec3f GetRotations() const;
-    [[nodiscard]] Common::Quaternion<f32> GetQuaternion() const;
-    [[nodiscard]] Common::Vec3f GetEulerAngles() const;
+    [[nodiscard]] std::array<Common::Vec<f32, 3>, 3> GetOrientation() const;
+    [[nodiscard]] Common::Vec<f32, 3> GetAcceleration() const;
+    [[nodiscard]] Common::Vec<f32, 3> GetGyroscope() const;
+    [[nodiscard]] Common::Vec<f32, 3> GetGyroBias() const;
+    [[nodiscard]] Common::Vec<f32, 3> GetRotations() const;
+    [[nodiscard]] Common::Vec<f32, 4> GetQuaternion() const;
+    [[nodiscard]] Common::Vec<f32, 3> GetEulerAngles() const;
 
     [[nodiscard]] bool IsMoving(f32 sensitivity) const;
     [[nodiscard]] bool IsCalibrated(f32 sensitivity) const;
@@ -75,24 +77,24 @@ private:
     f32 kd;
 
     // PID errors
-    Common::Vec3f real_error;
-    Common::Vec3f integral_error;
-    Common::Vec3f derivative_error;
+    Common::Vec<f32, 3> real_error;
+    Common::Vec<f32, 3> integral_error;
+    Common::Vec<f32, 3> derivative_error;
 
     // Quaternion containing the device orientation
-    Common::Quaternion<f32> quat;
+    Common::Vec<f32, 4> quat;
 
     // Number of full rotations in each axis
-    Common::Vec3f rotations;
+    Common::Vec<f32, 3> rotations;
 
     // Acceleration vector measurement in G force
-    Common::Vec3f accel;
+    Common::Vec<f32, 3> accel;
 
     // Gyroscope vector measurement in radians/s.
-    Common::Vec3f gyro;
+    Common::Vec<f32, 3> gyro;
 
     // Vector to be subtracted from gyro measurements
-    Common::Vec3f gyro_bias;
+    Common::Vec<f32, 3> gyro_bias;
 
     // Minimum gyro amplitude to detect if the device is moving
     f32 gyro_threshold = 0.0f;

src/hid_core/hid_types.h

@@ -609,10 +609,10 @@ static_assert(sizeof(SixAxisSensorAttribute) == 4, "SixAxisSensorAttribute is an
 struct SixAxisSensorState {
     s64 delta_time{};
     s64 sampling_number{};
-    Common::Vec3f accel{};
-    Common::Vec3f gyro{};
-    Common::Vec3f rotation{};
-    std::array<Common::Vec3f, 3> orientation{};
+    Common::Vec<f32, 3> accel{};
+    Common::Vec<f32, 3> gyro{};
+    Common::Vec<f32, 3> rotation{};
+    std::array<Common::Vec<f32, 3>, 3> orientation{};
     SixAxisSensorAttribute attribute{};
     INSERT_PADDING_BYTES(4); // Reserved
 };

src/hid_core/resources/shared_memory_format.h

@@ -1,4 +1,4 @@
-// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
 // SPDX-License-Identifier: GPL-3.0-or-later
 
 // SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
@@ -196,7 +196,7 @@ struct ConsoleSixAxisSensorSharedMemoryFormat {
     bool is_seven_six_axis_sensor_at_rest{};
     INSERT_PADDING_BYTES(3); // padding
     f32 verticalization_error{};
-    Common::Vec3f gyro_bias{};
+    Common::Vec<f32, 3> gyro_bias{};
     INSERT_PADDING_BYTES(4); // padding
 };
 static_assert(sizeof(ConsoleSixAxisSensorSharedMemoryFormat) == 0x20,

src/hid_core/resources/six_axis/seven_six_axis.cpp

@@ -46,14 +46,11 @@ void SevenSixAxis::OnUpdate(const Core::Timing::CoreTiming& core_timing) {
     next_seven_sixaxis_state.accel = motion_status.accel;
     next_seven_sixaxis_state.gyro = motion_status.gyro;
     next_seven_sixaxis_state.quaternion = {
-        {
-            motion_status.quaternion.xyz.y,
-            motion_status.quaternion.xyz.x,
-            -motion_status.quaternion.w,
-        },
-        -motion_status.quaternion.xyz.z,
+        motion_status.quaternion[1],
+        motion_status.quaternion[0],
+        -motion_status.quaternion[3],
+        -motion_status.quaternion[2],
     };
-
     seven_sixaxis_lifo.WriteNextEntry(next_seven_sixaxis_state);
     transfer_memory_owner->GetMemory().WriteBlock(transfer_memory, &seven_sixaxis_lifo,
                                                   sizeof(seven_sixaxis_lifo));

src/hid_core/resources/six_axis/seven_six_axis.h

@@ -7,7 +7,7 @@
 #pragma once
 
 #include "common/common_types.h"
-#include "common/quaternion.h"
+#include "common/vector_math.h"
 #include "common/typed_address.h"
 #include "hid_core/resources/controller_base.h"
 #include "hid_core/resources/ring_lifo.h"
@@ -51,9 +51,9 @@ private:
         u64 timestamp{};
         u64 sampling_number{};
         u64 unknown{};
-        Common::Vec3f accel{};
-        Common::Vec3f gyro{};
-        Common::Quaternion<f32> quaternion{};
+        Common::Vec<f32, 3> accel{};
+        Common::Vec<f32, 3> gyro{};
+        Common::Vec<f32, 4> quaternion{};
     };
     static_assert(sizeof(SevenSixAxisState) == 0x48, "SevenSixAxisState is an invalid size");
 

src/hid_core/resources/six_axis/six_axis.cpp

@@ -1,3 +1,6 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
 // SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-3.0-or-later
 
@@ -93,9 +96,9 @@ void SixAxis::OnUpdate(const Core::Timing::CoreTiming& core_timing) {
                     .accel = {0, 0, -1.0f},
                     .orientation =
                         {
-                            Common::Vec3f{1.0f, 0, 0},
-                            Common::Vec3f{0, 1.0f, 0},
-                            Common::Vec3f{0, 0, 1.0f},
+                            Common::Vec<f32, 3>{1.0f, 0, 0},
+                            Common::Vec<f32, 3>{0, 1.0f, 0},
+                            Common::Vec<f32, 3>{0, 0, 1.0f},
                         },
                     .attribute = {1},
                 };

src/input_common/drivers/mouse.cpp

@@ -88,8 +88,8 @@ void Mouse::UpdateStickInput() {
         last_mouse_change *= maximum_stick_range;
     }
 
-    SetAxis(identifier, mouse_axis_x, last_mouse_change.x);
-    SetAxis(identifier, mouse_axis_y, -last_mouse_change.y);
+    SetAxis(identifier, mouse_axis_x, last_mouse_change[0]);
+    SetAxis(identifier, mouse_axis_y, -last_mouse_change[1]);
 
     // Decay input over time
     const float clamped_length = (std::min)(1.0f, length);
@@ -104,20 +104,20 @@ void Mouse::UpdateMotionInput() {
     const float sensitivity =
         IsMousePanningEnabled() ? default_motion_panning_sensitivity : default_motion_sensitivity;
 
-    const float rotation_velocity = std::sqrt(last_motion_change.x * last_motion_change.x +
-                                              last_motion_change.y * last_motion_change.y);
+    const float rotation_velocity = std::sqrt(last_motion_change[0] * last_motion_change[0] +
+                                              last_motion_change[1] * last_motion_change[1]);
 
     // Clamp rotation speed
     if (rotation_velocity > maximum_rotation_speed / sensitivity) {
         const float multiplier = maximum_rotation_speed / rotation_velocity / sensitivity;
-        last_motion_change.x = last_motion_change.x * multiplier;
-        last_motion_change.y = last_motion_change.y * multiplier;
+        last_motion_change[0] = last_motion_change[0] * multiplier;
+        last_motion_change[1] = last_motion_change[1] * multiplier;
     }
 
     const BasicMotion motion_data{
-        .gyro_x = last_motion_change.x * sensitivity,
-        .gyro_y = last_motion_change.y * sensitivity,
-        .gyro_z = last_motion_change.z * sensitivity,
+        .gyro_x = last_motion_change[0] * sensitivity,
+        .gyro_y = last_motion_change[1] * sensitivity,
+        .gyro_z = last_motion_change[2] * sensitivity,
         .accel_x = 0,
         .accel_y = 0,
         .accel_z = 0,
@@ -125,53 +125,46 @@ void Mouse::UpdateMotionInput() {
     };
 
     if (IsMousePanningEnabled()) {
-        last_motion_change.x = 0;
-        last_motion_change.y = 0;
+        last_motion_change[0] = 0;
+        last_motion_change[1] = 0;
     }
-    last_motion_change.z = 0;
+    last_motion_change[2] = 0;
 
     SetMotion(motion_identifier, 0, motion_data);
 }
 
 void Mouse::Move(int x, int y, int center_x, int center_y) {
     if (IsMousePanningEnabled()) {
-        const auto mouse_change =
-            (Common::MakeVec(x, y) - Common::MakeVec(center_x, center_y)).Cast<float>();
-        const float x_sensitivity =
-            Settings::values.mouse_panning_x_sensitivity.GetValue() * default_panning_sensitivity;
-        const float y_sensitivity =
-            Settings::values.mouse_panning_y_sensitivity.GetValue() * default_panning_sensitivity;
-        const float deadzone_counterweight =
-            Settings::values.mouse_panning_deadzone_counterweight.GetValue() *
-            default_deadzone_counterweight;
-
-        last_motion_change += {-mouse_change.y * x_sensitivity, -mouse_change.x * y_sensitivity, 0};
-        last_mouse_change.x += mouse_change.x * x_sensitivity;
-        last_mouse_change.y += mouse_change.y * y_sensitivity;
-
-        // Bind the mouse change to [0 <= deadzone_counterweight <= 1.0]
+        auto const mouse_change_int = Common::Vec<int, 2>(x, y) - Common::Vec<int, 2>(center_x, center_y);
+        auto const mouse_change = Common::Vec<float, 2>(float(mouse_change_int[0]), float(mouse_change_int[1]));
+        auto const x_sensitivity = Settings::values.mouse_panning_x_sensitivity.GetValue() * default_panning_sensitivity;
+        auto const y_sensitivity = Settings::values.mouse_panning_y_sensitivity.GetValue() * default_panning_sensitivity;
+        auto const deadzone_cw = Settings::values.mouse_panning_deadzone_counterweight.GetValue() * default_deadzone_counterweight;
+        last_motion_change += {-mouse_change[1] * x_sensitivity, -mouse_change[0] * y_sensitivity, 0};
+        last_mouse_change[0] += mouse_change[0] * x_sensitivity;
+        last_mouse_change[1] += mouse_change[1] * y_sensitivity;
+        // Bind the mouse change to [0 <= deadzone_cw <= 1.0]
         const float length = last_mouse_change.Length();
-        if (length < deadzone_counterweight && length != 0.0f) {
+        if (length < deadzone_cw && length != 0.0f) {
             last_mouse_change /= length;
-            last_mouse_change *= deadzone_counterweight;
+            last_mouse_change *= deadzone_cw;
         }
-
         return;
     }
 
     if (button_pressed) {
-        const auto mouse_move = Common::MakeVec<int>(x, y) - mouse_origin;
+        const auto mouse_move = Common::Vec<int, 2>(x, y) - mouse_origin;
         const float x_sensitivity =
             Settings::values.mouse_panning_x_sensitivity.GetValue() * default_stick_sensitivity;
         const float y_sensitivity =
             Settings::values.mouse_panning_y_sensitivity.GetValue() * default_stick_sensitivity;
-        SetAxis(identifier, mouse_axis_x, static_cast<float>(mouse_move.x) * x_sensitivity);
-        SetAxis(identifier, mouse_axis_y, static_cast<float>(-mouse_move.y) * y_sensitivity);
+        SetAxis(identifier, mouse_axis_x, float(mouse_move[0]) * x_sensitivity);
+        SetAxis(identifier, mouse_axis_y, float(-mouse_move[1]) * y_sensitivity);
 
         last_motion_change = {
-            static_cast<float>(-mouse_move.y) * x_sensitivity,
-            static_cast<float>(-mouse_move.x) * y_sensitivity,
-            last_motion_change.z,
+            float(-mouse_move[1]) * x_sensitivity,
+            float(-mouse_move[0]) * y_sensitivity,
+            last_motion_change[2],
         };
     }
 }
@@ -220,18 +213,18 @@ void Mouse::ReleaseButton(MouseButton button) {
         SetAxis(identifier, mouse_axis_y, 0);
     }
 
-    last_motion_change.x = 0;
-    last_motion_change.y = 0;
+    last_motion_change[0] = 0;
+    last_motion_change[1] = 0;
 
     button_pressed = false;
 }
 
 void Mouse::MouseWheelChange(int x, int y) {
-    wheel_position.x += x;
-    wheel_position.y += y;
-    last_motion_change.z += static_cast<f32>(y);
-    SetAxis(identifier, wheel_axis_x, static_cast<f32>(wheel_position.x));
-    SetAxis(identifier, wheel_axis_y, static_cast<f32>(wheel_position.y));
+    wheel_position[0] += x;
+    wheel_position[1] += y;
+    last_motion_change[2] += static_cast<f32>(y);
+    SetAxis(identifier, wheel_axis_x, static_cast<f32>(wheel_position[0]));
+    SetAxis(identifier, wheel_axis_y, static_cast<f32>(wheel_position[1]));
 }
 
 void Mouse::ReleaseAllButtons() {

src/input_common/drivers/mouse.h

@@ -107,11 +107,11 @@ private:
 
     Common::Input::ButtonNames GetUIButtonName(const Common::ParamPackage& params) const;
 
-    Common::Vec2<int> mouse_origin;
-    Common::Vec2<int> last_mouse_position;
-    Common::Vec2<float> last_mouse_change;
-    Common::Vec3<float> last_motion_change;
-    Common::Vec2<int> wheel_position;
+    Common::Vec<int, 2> mouse_origin;
+    Common::Vec<int, 2> last_mouse_position;
+    Common::Vec<float, 2> last_mouse_change;
+    Common::Vec<float, 3> last_motion_change;
+    Common::Vec<int, 2> wheel_position;
     bool button_pressed = false;
 };
 

src/qt_common/util/content.cpp

@@ -28,9 +28,8 @@ bool CheckGameFirmware(u64 program_id) {
         !FirmwareManager::CheckFirmwarePresence(*system)) {
         auto result = QtCommon::Frontend::Warning(
             tr("Game Requires Firmware"),
-            tr("The game you are trying to launch requires firmware to boot or to get past the "
-               "opening menu. Please <a href='https://yuzu-mirror.github.io/help/quickstart'>"
-               "dump and install firmware</a>, or press \"OK\" to launch anyways."),
+            tr("This game may require firmware to function properly, and you don't have any installed. "
+                "Please install firmware before launching, or press \"OK\" to launch anyways."),
             QtCommon::Frontend::Ok | QtCommon::Frontend::Cancel);
 
         return result == QtCommon::Frontend::Ok;

src/video_core/CMakeLists.txt

@@ -20,6 +20,7 @@ add_library(video_core STATIC
     buffer_cache/buffer_cache.h
     buffer_cache/memory_tracker_base.h
     buffer_cache/usage_tracker.h
+    buffer_cache/virtual_range_cache.h
     buffer_cache/word_manager.h
     cache_types.h
     capture.h
@@ -166,6 +167,8 @@ add_library(video_core STATIC
     renderer_vulkan/vk_fence_manager.h
     renderer_vulkan/vk_graphics_pipeline.cpp
     renderer_vulkan/vk_graphics_pipeline.h
+    renderer_vulkan/vk_multi_range_buffer.cpp
+    renderer_vulkan/vk_multi_range_buffer.h
     renderer_vulkan/vk_master_semaphore.cpp
     renderer_vulkan/vk_master_semaphore.h
     renderer_vulkan/vk_pipeline_cache.cpp

src/video_core/buffer_cache/buffer_cache.h

@@ -112,6 +112,13 @@ void BufferCache<P>::TickFrame() {
     async_buffers_death_ring.clear();
 }
 
+template <class P>
+void BufferCache<P>::UnmapGPUMemory(size_t as_id, GPUVAddr gpu_addr, size_t size) {
+    if constexpr (requires { runtime.BindMultiRangeStorageBuffer(u64{}, bool{}); }) {
+        virtual_ranges.Unmap(as_id, gpu_addr, size);
+    }
+}
+
 template <class P>
 void BufferCache<P>::WriteMemory(DAddr device_addr, u64 size) {
     if (memory_tracker.IsRegionGpuModified(device_addr, size)) {
@@ -208,8 +215,8 @@ bool BufferCache<P>::DMACopy(GPUVAddr src_address, GPUVAddr dest_address, u64 am
     BufferId buffer_b;
     do {
         channel_state->has_deleted_buffers = false;
-        buffer_a = FindBuffer(*cpu_src_address, static_cast<u32>(amount));
-        buffer_b = FindBuffer(*cpu_dest_address, static_cast<u32>(amount));
+        buffer_a = FindBuffer(*cpu_src_address, static_cast<u32>(amount), false);
+        buffer_b = FindBuffer(*cpu_dest_address, static_cast<u32>(amount), false);
     } while (channel_state->has_deleted_buffers);
     auto& src_buffer = slot_buffers[buffer_a];
     auto& dest_buffer = slot_buffers[buffer_b];
@@ -265,7 +272,7 @@ bool BufferCache<P>::DMAClear(GPUVAddr dst_address, u64 amount, u32 value) {
     ClearDownload(*cpu_dst_address, size);
     gpu_modified_ranges.Subtract(*cpu_dst_address, size);
 
-    const BufferId buffer = FindBuffer(*cpu_dst_address, static_cast<u32>(size));
+    const BufferId buffer = FindBuffer(*cpu_dst_address, static_cast<u32>(size), false);
     Buffer& dest_buffer = slot_buffers[buffer];
     const u32 offset = dest_buffer.Offset(*cpu_dst_address);
     runtime.ClearBuffer(dest_buffer, offset, size, value);
@@ -287,7 +294,7 @@ std::pair<typename P::Buffer*, u32> BufferCache<P>::ObtainBuffer(GPUVAddr gpu_ad
 template <class P>
 std::pair<typename P::Buffer*, u32> BufferCache<P>::ObtainCPUBuffer(
     DAddr device_addr, u32 size, ObtainBufferSynchronize sync_info, ObtainBufferOperation post_op) {
-    const BufferId buffer_id = FindBuffer(device_addr, size);
+    const BufferId buffer_id = FindBuffer(device_addr, size, false);
     Buffer& buffer = slot_buffers[buffer_id];
 
     // synchronize op
@@ -998,11 +1005,85 @@ void BufferCache<P>::BindHostGraphicsUniformBuffer(size_t stage, u32 index, u32
     channel_state->fast_bound_uniform_buffers[stage] &= ~(1u << binding_index);
 }
 
+template <class P>
+void BufferCache<P>::ResolveMultiRangeStorage(Binding& binding, bool is_written,
+                                              std::vector<MultiRangeSegment>& pool) {
+    binding.segment_first = 0;
+    binding.segment_count = 0;
+    if constexpr (requires { runtime.BindMultiRangeStorageBuffer(u64{}, bool{}); }) {
+        if (binding.gpu_addr == 0 || binding.size == 0) {
+            return;
+        }
+        if (is_written && !runtime.PrefersSparseSources()) {
+            return;
+        }
+        const VirtualSegments* found =
+            virtual_ranges.Query(*gpu_memory, binding.gpu_addr, binding.size);
+        if (!found || found->size() < 2) {
+            return;
+        }
+        const VirtualSegments segments = *found;
+        const u32 first = static_cast<u32>(pool.size());
+        const bool prefer_sparse = runtime.PrefersSparseSources();
+        for (const VirtualSegment& segment : segments) {
+            const BufferId buffer_id =
+                FindBuffer(segment.device_addr, segment.size, prefer_sparse);
+            if (!buffer_id) {
+                pool.resize(first);
+                return;
+            }
+            pool.push_back(MultiRangeSegment{
+                .buffer_id = buffer_id,
+                .device_addr = segment.device_addr,
+                .size = segment.size,
+            });
+        }
+        binding.segment_first = first;
+        binding.segment_count = static_cast<u32>(segments.size());
+    }
+}
+
+template <class P>
+bool BufferCache<P>::BindMultiRangeStorage(const Binding& binding, bool is_written,
+                                           std::span<const MultiRangeSegment> pool) {
+    if constexpr (requires { runtime.BindMultiRangeStorageBuffer(u64{}, bool{}); }) {
+        if (binding.segment_count < 2) {
+            return false;
+        }
+        if (binding.segment_first + binding.segment_count > pool.size()) {
+            return false;
+        }
+        const u64 key = (static_cast<u64>(gpu_memory->GetID()) << 48) ^ binding.gpu_addr;
+        runtime.ResetMultiRange();
+        for (u32 index = 0; index < binding.segment_count; ++index) {
+            const MultiRangeSegment& segment = pool[binding.segment_first + index];
+            Buffer& buffer = slot_buffers[segment.buffer_id];
+            TouchBuffer(buffer, segment.buffer_id);
+            if (SynchronizeBuffer(buffer, segment.device_addr, segment.size)) {
+                runtime.InvalidateMultiRange(key);
+            }
+            const u32 offset = buffer.Offset(segment.device_addr);
+            buffer.MarkUsage(offset, segment.size);
+            if (is_written) {
+                MarkWrittenBuffer(segment.buffer_id, segment.device_addr, segment.size);
+            }
+            runtime.PushMultiRangeSource(buffer, offset, segment.size);
+        }
+        return runtime.BindMultiRangeStorageBuffer(key, is_written);
+    } else {
+        return false;
+    }
+}
+
 template <class P>
 void BufferCache<P>::BindHostGraphicsStorageBuffers(size_t stage) {
     u32 binding_index = 0;
     ForEachEnabledBit(channel_state->enabled_storage_buffers[stage], [&](u32 index) {
         const Binding& binding = channel_state->storage_buffers[stage][index];
+        const bool is_written = ((channel_state->written_storage_buffers[stage] >> index) & 1) != 0;
+        if (BindMultiRangeStorage(binding, is_written, graphics_segments)) {
+            return;
+        }
         Buffer& buffer = slot_buffers[binding.buffer_id];
         TouchBuffer(buffer, binding.buffer_id);
         const u32 size = binding.size;
@@ -1010,7 +1091,6 @@ void BufferCache<P>::BindHostGraphicsStorageBuffers(size_t stage) {
 
         const u32 offset = buffer.Offset(binding.device_addr);
         buffer.MarkUsage(offset, size);
-        const bool is_written = ((channel_state->written_storage_buffers[stage] >> index) & 1) != 0;
 
         if (is_written) {
             MarkWrittenBuffer(binding.buffer_id, binding.device_addr, size);
@@ -1139,6 +1219,11 @@ void BufferCache<P>::BindHostComputeStorageBuffers() {
     u32 binding_index = 0;
     ForEachEnabledBit(channel_state->enabled_compute_storage_buffers, [&](u32 index) {
         const Binding& binding = channel_state->compute_storage_buffers[index];
+        const bool is_written =
+            ((channel_state->written_compute_storage_buffers >> index) & 1) != 0;
+        if (BindMultiRangeStorage(binding, is_written, compute_segments)) {
+            return;
+        }
         Buffer& buffer = slot_buffers[binding.buffer_id];
         TouchBuffer(buffer, binding.buffer_id);
         const u32 size = binding.size;
@@ -1146,8 +1231,6 @@ void BufferCache<P>::BindHostComputeStorageBuffers() {
 
         const u32 offset = buffer.Offset(binding.device_addr);
         buffer.MarkUsage(offset, size);
-        const bool is_written =
-            ((channel_state->written_compute_storage_buffers >> index) & 1) != 0;
 
         if (is_written) {
             MarkWrittenBuffer(binding.buffer_id, binding.device_addr, size);
@@ -1193,6 +1276,7 @@ void BufferCache<P>::BindHostComputeTextureBuffers() {
 
 template <class P>
 void BufferCache<P>::DoUpdateGraphicsBuffers(bool is_indexed) {
+    graphics_segments.clear();
     BufferOperations([&]() {
         if (is_indexed) {
             UpdateIndexBuffer();
@@ -1212,6 +1296,7 @@ void BufferCache<P>::DoUpdateGraphicsBuffers(bool is_indexed) {
 
 template <class P>
 void BufferCache<P>::DoUpdateComputeBuffers() {
+    compute_segments.clear();
     BufferOperations([&]() {
         UpdateComputeUniformBuffers();
         UpdateComputeStorageBuffers();
@@ -1234,11 +1319,11 @@ void BufferCache<P>::UpdateIndexBuffer() {
         auto inline_index_size = static_cast<u32>(draw_state.inline_index_draw_indexes.size());
         u32 buffer_size = Common::AlignUp(inline_index_size, CACHING_PAGESIZE);
         if (inline_buffer_id == NULL_BUFFER_ID) [[unlikely]] {
-            inline_buffer_id = CreateBuffer(0, buffer_size);
+            inline_buffer_id = CreateBuffer(0, buffer_size, false);
         }
         if (slot_buffers[inline_buffer_id].SizeBytes() < buffer_size) [[unlikely]] {
             slot_buffers.erase(inline_buffer_id);
-            inline_buffer_id = CreateBuffer(0, buffer_size);
+            inline_buffer_id = CreateBuffer(0, buffer_size, false);
         }
         channel_state->index_buffer = Binding{
             .device_addr = 0,
@@ -1261,7 +1346,7 @@ void BufferCache<P>::UpdateIndexBuffer() {
     channel_state->index_buffer = Binding{
         .device_addr = *device_addr,
         .size = size,
-        .buffer_id = FindBuffer(*device_addr, size),
+        .buffer_id = FindBuffer(*device_addr, size, false),
     };
 }
 
@@ -1298,7 +1383,7 @@ void BufferCache<P>::UpdateVertexBuffer(u32 index) {
     if (!gpu_memory->IsWithinGPUAddressRange(gpu_addr_end) || size >= 64_MiB) {
         size = static_cast<u32>(gpu_memory->MaxContinuousRange(gpu_addr_begin, size));
     }
-    const BufferId buffer_id = FindBuffer(*device_addr, size);
+    const BufferId buffer_id = FindBuffer(*device_addr, size, false);
     const Binding binding{
         .device_addr = *device_addr,
         .size = size,
@@ -1319,7 +1404,7 @@ void BufferCache<P>::UpdateDrawIndirect() {
         binding = Binding{
             .device_addr = *device_addr,
             .size = static_cast<u32>(size),
-            .buffer_id = FindBuffer(*device_addr, static_cast<u32>(size)),
+            .buffer_id = FindBuffer(*device_addr, static_cast<u32>(size), false),
         };
     };
     if (current_draw_indirect->include_count) {
@@ -1343,7 +1428,7 @@ void BufferCache<P>::UpdateUniformBuffers(size_t stage) {
             channel_state->dirty_uniform_buffers[stage] |= 1U << index;
         }
         // Resolve buffer
-        binding.buffer_id = FindBuffer(binding.device_addr, binding.size);
+        binding.buffer_id = FindBuffer(binding.device_addr, binding.size, false);
     });
 }
 
@@ -1352,8 +1437,10 @@ void BufferCache<P>::UpdateStorageBuffers(size_t stage) {
     ForEachEnabledBit(channel_state->enabled_storage_buffers[stage], [&](u32 index) {
         // Resolve buffer
         Binding& binding = channel_state->storage_buffers[stage][index];
-        const BufferId buffer_id = FindBuffer(binding.device_addr, binding.size);
+        const BufferId buffer_id = FindBuffer(binding.device_addr, binding.size, false);
         binding.buffer_id = buffer_id;
+        const bool is_written = ((channel_state->written_storage_buffers[stage] >> index) & 1) != 0;
+        ResolveMultiRangeStorage(binding, is_written, graphics_segments);
     });
 }
 
@@ -1361,7 +1448,7 @@ template <class P>
 void BufferCache<P>::UpdateTextureBuffers(size_t stage) {
     ForEachEnabledBit(channel_state->enabled_texture_buffers[stage], [&](u32 index) {
         Binding& binding = channel_state->texture_buffers[stage][index];
-        binding.buffer_id = FindBuffer(binding.device_addr, binding.size);
+        binding.buffer_id = FindBuffer(binding.device_addr, binding.size, false);
     });
 }
 
@@ -1385,7 +1472,7 @@ void BufferCache<P>::UpdateTransformFeedbackBuffer(u32 index) {
         channel_state->transform_feedback_buffers[index] = NULL_BINDING;
         return;
     }
-    const BufferId buffer_id = FindBuffer(*device_addr, size);
+    const BufferId buffer_id = FindBuffer(*device_addr, size, false);
     channel_state->transform_feedback_buffers[index] = Binding{
         .device_addr = *device_addr,
         .size = size,
@@ -1407,7 +1494,7 @@ void BufferCache<P>::UpdateComputeUniformBuffers() {
                 binding.size = cbuf.size;
             }
         }
-        binding.buffer_id = FindBuffer(binding.device_addr, binding.size);
+        binding.buffer_id = FindBuffer(binding.device_addr, binding.size, false);
     });
 }
 
@@ -1416,7 +1503,10 @@ void BufferCache<P>::UpdateComputeStorageBuffers() {
     ForEachEnabledBit(channel_state->enabled_compute_storage_buffers, [&](u32 index) {
         // Resolve buffer
         Binding& binding = channel_state->compute_storage_buffers[index];
-        binding.buffer_id = FindBuffer(binding.device_addr, binding.size);
+        binding.buffer_id = FindBuffer(binding.device_addr, binding.size, false);
+        const bool is_written =
+            ((channel_state->written_compute_storage_buffers >> index) & 1) != 0;
+        ResolveMultiRangeStorage(binding, is_written, compute_segments);
     });
 }
 
@@ -1424,7 +1514,7 @@ template <class P>
 void BufferCache<P>::UpdateComputeTextureBuffers() {
     ForEachEnabledBit(channel_state->enabled_compute_texture_buffers, [&](u32 index) {
         Binding& binding = channel_state->compute_texture_buffers[index];
-        binding.buffer_id = FindBuffer(binding.device_addr, binding.size);
+        binding.buffer_id = FindBuffer(binding.device_addr, binding.size, false);
     });
 }
 
@@ -1440,7 +1530,7 @@ void BufferCache<P>::MarkWrittenBuffer(BufferId buffer_id, DAddr device_addr, u3
 }
 
 template <class P>
-BufferId BufferCache<P>::FindBuffer(DAddr device_addr, u32 size) {
+BufferId BufferCache<P>::FindBuffer(DAddr device_addr, u32 size, bool sparse_compatible) {
     if (device_addr == 0) {
         return NULL_BUFFER_ID;
     }
@@ -1450,10 +1540,18 @@ BufferId BufferCache<P>::FindBuffer(DAddr device_addr, u32 size) {
         Buffer& buffer = slot_buffers[buffer_id];
         WaitForGpuFenceIfNeeded(buffer);
         if (buffer.IsInBounds(device_addr, size)) {
-            return buffer_id;
+            bool usable = true;
+            if constexpr (requires { buffer.IsSparseCompatible(); }) {
+                if (sparse_compatible && !buffer.IsSparseCompatible()) {
+                    usable = false;
+                }
+            }
+            if (usable) {
+                return buffer_id;
+            }
         }
     }
-    return CreateBuffer(device_addr, size);
+    return CreateBuffer(device_addr, size, sparse_compatible);
 }
 
 template <class P>
@@ -1575,13 +1673,15 @@ void BufferCache<P>::JoinOverlap(BufferId new_buffer_id, BufferId overlap_id,
 }
 
 template <class P>
-BufferId BufferCache<P>::CreateBuffer(DAddr device_addr, u32 wanted_size) {
+BufferId BufferCache<P>::CreateBuffer(DAddr device_addr, u32 wanted_size,
+                                      bool sparse_compatible) {
     DAddr device_addr_end = Common::AlignUp(device_addr + wanted_size, CACHING_PAGESIZE);
     device_addr = Common::AlignDown(device_addr, CACHING_PAGESIZE);
     wanted_size = static_cast<u32>(device_addr_end - device_addr);
     const OverlapResult overlap = ResolveOverlaps(device_addr, wanted_size);
     const u32 size = static_cast<u32>(overlap.end - overlap.begin);
-    const BufferId new_buffer_id = slot_buffers.insert(runtime, overlap.begin, size);
+    const BufferId new_buffer_id =
+        slot_buffers.insert(runtime, overlap.begin, size, sparse_compatible);
     auto& new_buffer = slot_buffers[new_buffer_id];
     const size_t size_bytes = new_buffer.SizeBytes();
     runtime.ClearBuffer(new_buffer, 0, size_bytes, 0);
@@ -1745,7 +1845,7 @@ void BufferCache<P>::InlineMemoryImplementation(DAddr dest_address, size_t copy_
     ClearDownload(dest_address, copy_size);
     gpu_modified_ranges.Subtract(dest_address, copy_size);
 
-    BufferId buffer_id = FindBuffer(dest_address, static_cast<u32>(copy_size));
+    BufferId buffer_id = FindBuffer(dest_address, static_cast<u32>(copy_size), false);
     auto& buffer = slot_buffers[buffer_id];
     SynchronizeBuffer(buffer, dest_address, static_cast<u32>(copy_size));
 
@@ -1831,6 +1931,9 @@ void BufferCache<P>::DownloadBufferMemory(Buffer& buffer, DAddr device_addr, u64
 
 template <class P>
 void BufferCache<P>::DeleteBuffer(BufferId buffer_id, bool do_not_mark) {
+    if constexpr (requires { runtime.OnBufferDeleted(slot_buffers[buffer_id]); }) {
+        runtime.OnBufferDeleted(slot_buffers[buffer_id]);
+    }
     bool dirty_index{false};
     boost::container::small_vector<u64, NUM_VERTEX_BUFFERS> dirty_vertex_buffers;
     const auto scalar_replace = [buffer_id](Binding& binding) {
@@ -1934,9 +2037,14 @@ Binding BufferCache<P>::StorageBufferBinding(GPUVAddr ssbo_addr, u32 cbuf_index,
     // The end address used for size calculation does not need to be aligned
     const DAddr cpu_end = Common::AlignUp(*device_addr + size, Core::DEVICE_PAGESIZE);
 
+    u32 binding_size = static_cast<u32>(cpu_end - *aligned_device_addr);
+    if (is_written) {
+        binding_size = aligned_size;
+    }
     const Binding binding{
         .device_addr = *aligned_device_addr,
-        .size = is_written ? aligned_size : static_cast<u32>(cpu_end - *aligned_device_addr),
+        .gpu_addr = aligned_gpu_addr,
+        .size = binding_size,
         .buffer_id = BufferId{},
     };
     return binding;

src/video_core/buffer_cache/buffer_cache_base.h

@@ -29,6 +29,7 @@
 #include "common/settings.h"
 #include "common/slot_vector.h"
 #include "video_core/buffer_cache/buffer_base.h"
+#include "video_core/buffer_cache/virtual_range_cache.h"
 #include "video_core/control/channel_state_cache.h"
 #include "video_core/delayed_destruction_ring.h"
 #include "video_core/dirty_flags.h"
@@ -81,8 +82,17 @@ static constexpr u32 DEFAULT_SKIP_CACHE_SIZE = static_cast<u32>(4_KiB);
 
 struct Binding {
     DAddr device_addr{};
+    GPUVAddr gpu_addr{};
     u32 size{};
     BufferId buffer_id;
+    u32 segment_first{};
+    u32 segment_count{};
+};
+
+struct MultiRangeSegment {
+    BufferId buffer_id;
+    DAddr device_addr{};
+    u32 size{};
 };
 
 struct TextureBufferBinding : Binding {
@@ -215,6 +225,14 @@ public:
 
     void TickFrame();
 
+    bool BindMultiRangeStorage(const Binding& binding, bool is_written,
+                               std::span<const MultiRangeSegment> pool);
+
+    void ResolveMultiRangeStorage(Binding& binding, bool is_written,
+                                  std::vector<MultiRangeSegment>& pool);
+
+    void UnmapGPUMemory(size_t as_id, GPUVAddr gpu_addr, size_t size);
+
     void WriteMemory(DAddr device_addr, u64 size);
 
     void CachedWriteMemory(DAddr device_addr, u64 size);
@@ -414,7 +432,7 @@ private:
 
     void MarkWrittenBuffer(BufferId buffer_id, DAddr device_addr, u32 size);
 
-    [[nodiscard]] BufferId FindBuffer(DAddr device_addr, u32 size);
+    [[nodiscard]] BufferId FindBuffer(DAddr device_addr, u32 size, bool sparse_compatible);
 
     void WaitForGpuFenceIfNeeded(Buffer& buffer);
 
@@ -422,7 +440,8 @@ private:
 
     void JoinOverlap(BufferId new_buffer_id, BufferId overlap_id, bool accumulate_stream_score);
 
-    [[nodiscard]] BufferId CreateBuffer(DAddr device_addr, u32 wanted_size);
+    [[nodiscard]] BufferId CreateBuffer(DAddr device_addr, u32 wanted_size,
+                                        bool sparse_compatible);
 
     void Register(BufferId buffer_id);
 
@@ -513,6 +532,9 @@ private:
         using TickType = u64;
     };
     Common::LeastRecentlyUsedCache<LRUItemParams> lru_cache;
+    VirtualRangeCache virtual_ranges;
+    std::vector<MultiRangeSegment> graphics_segments;
+    std::vector<MultiRangeSegment> compute_segments;
     u64 frame_tick = 0;
     u64 total_used_memory = 0;
     u64 minimum_memory = 0;

src/video_core/buffer_cache/virtual_range_cache.h

@@ -0,0 +1,173 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
+#pragma once
+
+#include <atomic>
+#include <limits>
+#include <mutex>
+#include <optional>
+#include <vector>
+
+#include <boost/container/small_vector.hpp>
+
+#include "common/common_types.h"
+#include "common/container/unordered_map.h"
+#include "video_core/memory_manager.h"
+
+namespace VideoCommon {
+
+struct VirtualSegment {
+    GPUVAddr gpu_addr;
+    DAddr device_addr;
+    u32 size;
+};
+
+using VirtualSegments = boost::container::small_vector<VirtualSegment, 8>;
+
+class VirtualRangeCache {
+public:
+    static constexpr size_t MAX_ENTRIES = 8192;
+    static constexpr size_t MAX_DEFERRED = 4096;
+
+    const VirtualSegments* Query(Tegra::MemoryManager& memory, GPUVAddr gpu_addr, u32 size) {
+        if (has_deferred.load(std::memory_order_acquire)) {
+            ApplyDeferred();
+        }
+        if (entries.size() > MAX_ENTRIES) {
+            entries.clear();
+        }
+        const size_t as_id = memory.GetID();
+        const u64 key = MakeKey(as_id, gpu_addr);
+        const auto it = entries.find(key);
+        if (it != entries.end() && it->second.as_id == as_id &&
+            it->second.gpu_addr == gpu_addr && it->second.size == size) {
+            return &it->second.segments;
+        }
+        Entry entry{};
+        entry.as_id = as_id;
+        entry.gpu_addr = gpu_addr;
+        entry.size = size;
+        const auto ranges = memory.GetSubmappedRange(gpu_addr, size);
+        GPUVAddr expected = gpu_addr;
+        bool contiguous = true;
+        for (const auto& [range_addr, range_size] : ranges) {
+            if (range_addr != expected || range_size == 0) {
+                contiguous = false;
+                break;
+            }
+            const std::optional<DAddr> device_addr = memory.GpuToCpuAddress(range_addr);
+            if (!device_addr || *device_addr == 0) {
+                contiguous = false;
+                break;
+            }
+            if (range_size > static_cast<size_t>((std::numeric_limits<u32>::max)())) {
+                contiguous = false;
+                break;
+            }
+            entry.segments.push_back(VirtualSegment{
+                .gpu_addr = range_addr,
+                .device_addr = *device_addr,
+                .size = static_cast<u32>(range_size),
+            });
+            expected += range_size;
+        }
+        if (!contiguous || expected != gpu_addr + size) {
+            entry.segments.clear();
+        }
+        const auto result = entries.insert_or_assign(key, std::move(entry));
+        return &result.first->second.segments;
+    }
+
+    void Unmap(size_t as_id, GPUVAddr gpu_addr, u64 size) {
+        if (size == 0) {
+            return;
+        }
+        {
+            std::scoped_lock lock{deferred_mutex};
+            if (!deferred.empty()) {
+                DeferredUnmap& last = deferred.back();
+                if (last.as_id == as_id && last.gpu_addr + last.size == gpu_addr) {
+                    last.size += size;
+                    has_deferred.store(true, std::memory_order_release);
+                    return;
+                }
+            }
+            if (deferred.size() >= MAX_DEFERRED) {
+                deferred.clear();
+                deferred_overflow = true;
+            } else {
+                deferred.push_back(DeferredUnmap{
+                    .as_id = as_id,
+                    .gpu_addr = gpu_addr,
+                    .size = size,
+                });
+            }
+        }
+        has_deferred.store(true, std::memory_order_release);
+    }
+
+private:
+    struct Entry {
+        VirtualSegments segments;
+        size_t as_id{};
+        GPUVAddr gpu_addr{};
+        u32 size{};
+    };
+
+    struct DeferredUnmap {
+        size_t as_id;
+        GPUVAddr gpu_addr;
+        u64 size;
+    };
+
+    static u64 MakeKey(size_t as_id, GPUVAddr gpu_addr) {
+        return (static_cast<u64>(as_id) << 48) ^ gpu_addr;
+    }
+
+    void ApplyDeferred() {
+        std::vector<DeferredUnmap> pending;
+        bool overflow = false;
+        {
+            std::scoped_lock lock{deferred_mutex};
+            has_deferred.store(false, std::memory_order_release);
+            pending.swap(deferred);
+            overflow = deferred_overflow;
+            deferred_overflow = false;
+        }
+        if (overflow) {
+            entries.clear();
+            return;
+        }
+        if (pending.empty() || entries.empty()) {
+            return;
+        }
+        for (auto it = entries.begin(); it != entries.end();) {
+            const Entry& entry = it->second;
+            const GPUVAddr entry_end = entry.gpu_addr + entry.size;
+            bool overlaps = false;
+            for (const DeferredUnmap& unmap : pending) {
+                if (unmap.as_id != entry.as_id) {
+                    continue;
+                }
+                if (entry.gpu_addr < unmap.gpu_addr + unmap.size && unmap.gpu_addr < entry_end) {
+                    overlaps = true;
+                    break;
+                }
+            }
+            if (overlaps) {
+                it = entries.erase(it);
+            } else {
+                ++it;
+            }
+        }
+    }
+
+    ::Common::unordered_map<u64, Entry> entries;
+    std::vector<DeferredUnmap> deferred;
+    std::mutex deferred_mutex;
+    std::atomic<bool> has_deferred{false};
+    bool deferred_overflow{};
+};
+
+} // namespace VideoCommon

src/video_core/gpu.cpp

@@ -55,11 +55,11 @@ constexpr u64 GpuClockMultiplier(Settings::GpuClock clock) {
 
 struct GPU::Impl {
     explicit Impl(Core::System& system_, bool is_async_, bool use_nvdec_)
-        : system{system_}
+        : gpu_thread{system_}
+        , system{system_}
         , use_nvdec{use_nvdec_}
         , shader_notify()
         , is_async{is_async_}
-        , gpu_thread{system_}
     {}
 
     ~Impl() = default;
@@ -301,6 +301,10 @@ struct GPU::Impl {
         return out;
     }
 
+    // Destruction of thread must be done before all (non trivial)
+    // previous members has been destroyed
+    VideoCommon::GPUThread::ThreadManager gpu_thread;
+
     Core::System& system;
 
     std::unique_ptr<VideoCore::RendererBase> renderer;
@@ -329,7 +333,6 @@ struct GPU::Impl {
 
     const bool is_async;
 
-    VideoCommon::GPUThread::ThreadManager gpu_thread;
     std::unique_ptr<Core::Frontend::GraphicsContext> cpu_context;
 
     Tegra::Control::Scheduler scheduler;

src/video_core/memory_manager.cpp

@@ -46,8 +46,8 @@ MemoryManager::MemoryManager(Core::System& system_, MaxwellDeviceMemoryManager&
     page_table_mask = page_table_size - 1;
     big_page_table_mask = big_page_table_size - 1;
 
+    big_page_table_dev.ResizeAndClear(big_page_table_size);
     big_entries.resize(big_page_table_size / 32, 0);
-    big_page_table_dev.resize(big_page_table_size);
     big_page_continuous.resize(big_page_table_size / continuous_bits, 0);
     entries.resize(page_table_size / 32, 0);
 }
@@ -143,7 +143,7 @@ GPUVAddr MemoryManager::BigPageTableOp(GPUVAddr gpu_addr, [[maybe_unused]] DAddr
             const DAddr current_dev_addr = dev_addr + offset;
             const auto index = PageEntryIndex(current_gpu_addr, true);
             const u32 sub_value = static_cast<u32>(current_dev_addr >> cpu_page_bits);
-            big_page_table_dev[index] = sub_value;
+            big_page_table_dev.Set(index, sub_value);
             const bool is_continuous = ([&] {
                 uintptr_t base_ptr{
                     reinterpret_cast<uintptr_t>(memory.GetPointer<u8>(current_dev_addr))};

src/video_core/memory_manager.h

@@ -17,7 +17,7 @@
 #include "common/multi_level_page_table.h"
 #include "common/range_map.h"
 #include "common/scratch_buffer.h"
-#include "common/virtual_buffer.h"
+#include "common/sparse_large_vector.h"
 #include "video_core/invalidation_accumulator.h"
 #include "video_core/cache_types.h"
 #include "video_core/host1x/gpu_device_memory_manager.h"
@@ -214,7 +214,7 @@ private:
 
     Common::MultiLevelPageTable<u32> page_table;
     Common::RangeMap<GPUVAddr, PTEKind> kind_map;
-    Common::VirtualBuffer<u32> big_page_table_dev;
+    Common::SparseLargeVector<u32> big_page_table_dev;
 
     std::vector<u64> big_page_continuous;
     boost::container::small_vector<std::pair<DAddr, std::size_t>, 32> page_stash{};

src/video_core/renderer_opengl/gl_buffer_cache.cpp

@@ -52,7 +52,7 @@ constexpr std::array PROGRAM_LUT{
 Buffer::Buffer(BufferCacheRuntime&, VideoCommon::NullBufferParams null_params)
     : VideoCommon::BufferBase(null_params) {}
 
-Buffer::Buffer(BufferCacheRuntime& runtime, DAddr cpu_addr_, u64 size_bytes_)
+Buffer::Buffer(BufferCacheRuntime& runtime, DAddr cpu_addr_, u64 size_bytes_, bool)
     : VideoCommon::BufferBase(cpu_addr_, size_bytes_) {
     buffer.Create();
     if (runtime.device.HasDebuggingToolAttached()) {

src/video_core/renderer_opengl/gl_buffer_cache.h

@@ -23,7 +23,8 @@ class BufferCacheRuntime;
 
 class Buffer : public VideoCommon::BufferBase {
 public:
-    explicit Buffer(BufferCacheRuntime&, DAddr cpu_addr, u64 size_bytes);
+    explicit Buffer(BufferCacheRuntime&, DAddr cpu_addr, u64 size_bytes,
+                    bool sparse_compatible);
     explicit Buffer(BufferCacheRuntime&, VideoCommon::NullBufferParams);
 
     void ImmediateUpload(size_t offset, std::span<const u8> data) noexcept;

src/video_core/renderer_vulkan/vk_buffer_cache.cpp

@@ -56,7 +56,8 @@ size_t BytesPerIndex(VkIndexType index_type) {
     }
 }
 
-vk::Buffer CreateBuffer(const Device& device, const MemoryAllocator& memory_allocator, u64 size) {
+vk::Buffer CreateBuffer(const Device& device, const MemoryAllocator& memory_allocator, u64 size,
+                        VkDeviceSize sparse_alignment) {
     VkBufferUsageFlags flags =
         VK_BUFFER_USAGE_TRANSFER_SRC_BIT | VK_BUFFER_USAGE_TRANSFER_DST_BIT |
         VK_BUFFER_USAGE_UNIFORM_TEXEL_BUFFER_BIT | VK_BUFFER_USAGE_STORAGE_TEXEL_BUFFER_BIT |
@@ -82,6 +83,9 @@ vk::Buffer CreateBuffer(const Device& device, const MemoryAllocator& memory_allo
         .queueFamilyIndexCount = 0,
         .pQueueFamilyIndices = nullptr,
     };
+    if (sparse_alignment > 1) {
+        return memory_allocator.CreateBuffer(buffer_ci, MemoryUsage::DeviceLocal, sparse_alignment);
+    }
     return memory_allocator.CreateBuffer(buffer_ci, MemoryUsage::DeviceLocal);
 }
 } // Anonymous namespace
@@ -99,10 +103,14 @@ Buffer::Buffer(BufferCacheRuntime& runtime, VideoCommon::NullBufferParams null_p
     }
 }
 
-Buffer::Buffer(BufferCacheRuntime& runtime, DAddr cpu_addr_, u64 size_bytes_)
+Buffer::Buffer(BufferCacheRuntime& runtime, DAddr cpu_addr_, u64 size_bytes_,
+               bool sparse_compatible_)
     : VideoCommon::BufferBase(cpu_addr_, size_bytes_), device{&runtime.device},
       scheduler{&runtime.scheduler},
-      buffer{CreateBuffer(*device, runtime.memory_allocator, SizeBytes())}, tracker{SizeBytes()} {
+      buffer{CreateBuffer(*device, runtime.memory_allocator, SizeBytes(),
+                          runtime.SparseAlignmentFor(sparse_compatible_))},
+      tracker{SizeBytes()} {
+    sparse_compatible = sparse_compatible_;
     if (runtime.device.HasDebuggingToolAttached()) {
         buffer.SetObjectNameEXT(fmt::format("Buffer {:#x}", CpuAddr()).c_str());
     }
@@ -348,7 +356,8 @@ BufferCacheRuntime::BufferCacheRuntime(const Device& device_, MemoryAllocator& m
     : device{device_}, memory_allocator{memory_allocator_}, scheduler{scheduler_},
       staging_pool{staging_pool_}, guest_descriptor_queue{guest_descriptor_queue_},
       quad_index_pass(device, scheduler, descriptor_pool, staging_pool,
-                      compute_pass_descriptor_queue) {
+                      compute_pass_descriptor_queue),
+      multi_range_buffers(device_) {
     const VkDriverIdKHR driver_id = device.GetDriverID();
     limit_dynamic_storage_buffers = driver_id == VK_DRIVER_ID_QUALCOMM_PROPRIETARY ||
                                     driver_id == VK_DRIVER_ID_ARM_PROPRIETARY;
@@ -536,6 +545,37 @@ void BufferCacheRuntime::ClearBuffer(VkBuffer dest_buffer, u32 offset, size_t si
     });
 }
 
+bool BufferCacheRuntime::BindMultiRangeStorageBuffer(u64 key, bool is_written) {
+    if (multi_range_sources.empty() || multi_range_total == 0) {
+        return false;
+    }
+    const MultiRangeRef ref = multi_range_buffers.Get(device, scheduler, memory_allocator, key,
+                                                     multi_range_sources, multi_range_total);
+    if (ref.handle == VK_NULL_HANDLE) {
+        return false;
+    }
+    if (is_written && !ref.sparse) {
+        return false;
+    }
+    if (ref.needs_gather) {
+        PreCopyBarrier();
+        VkDeviceSize dst_offset = 0;
+        for (const MultiRangeSource& source : multi_range_sources) {
+            const std::array<VideoCommon::BufferCopy, 1> copy{VideoCommon::BufferCopy{
+                .src_offset = u64(source.offset),
+                .dst_offset = u64(dst_offset),
+                .size = size_t(source.size),
+            }};
+            CopyBuffer(ref.handle, source.handle, copy, false);
+            dst_offset += source.size;
+        }
+        PostCopyBarrier();
+        multi_range_buffers.MarkGathered(key);
+    }
+    guest_descriptor_queue.AddBuffer(ref.handle, ref.address, 0, ref.size);
+    return true;
+}
+
 void BufferCacheRuntime::BindIndexBuffer(PrimitiveTopology topology, IndexFormat index_format,
                                          u32 base_vertex, u32 num_indices, VkBuffer buffer,
                                          u32 offset, [[maybe_unused]] u32 size) {

src/video_core/renderer_vulkan/vk_buffer_cache.h

@@ -8,11 +8,14 @@
 
 #include <limits>
 
+#include <boost/container/small_vector.hpp>
+
 #include "video_core/buffer_cache/buffer_cache_base.h"
 #include "video_core/buffer_cache/memory_tracker_base.h"
 #include "video_core/buffer_cache/usage_tracker.h"
 #include "video_core/engines/maxwell_3d.h"
 #include "video_core/renderer_vulkan/vk_compute_pass.h"
+#include "video_core/renderer_vulkan/vk_multi_range_buffer.h"
 #include "video_core/renderer_vulkan/vk_staging_buffer_pool.h"
 #include "video_core/renderer_vulkan/vk_update_descriptor.h"
 #include "video_core/surface.h"
@@ -31,7 +34,8 @@ class BufferCacheRuntime;
 class Buffer : public VideoCommon::BufferBase {
 public:
     explicit Buffer(BufferCacheRuntime&, VideoCommon::NullBufferParams null_params);
-    explicit Buffer(BufferCacheRuntime& runtime, VAddr cpu_addr_, u64 size_bytes_);
+    explicit Buffer(BufferCacheRuntime& runtime, VAddr cpu_addr_, u64 size_bytes_,
+                    bool sparse_compatible_);
 
     [[nodiscard]] VkBufferView View(u32 offset, u32 size, VideoCore::Surface::PixelFormat format);
 
@@ -43,6 +47,14 @@ public:
         return device_address;
     }
 
+    [[nodiscard]] bool IsSparseCompatible() const noexcept {
+        return sparse_compatible;
+    }
+
+    [[nodiscard]] vk::MemoryLocation Location() const noexcept {
+        return buffer.Location();
+    }
+
     [[nodiscard]] bool IsRegionUsed(u64 offset, u64 size) const noexcept {
         return tracker.IsUsed(offset, size);
     }
@@ -77,6 +89,7 @@ private:
     VkDeviceAddress device_address{};
     u64 last_usage_tick{};
     bool is_null{};
+    bool sparse_compatible{};
 };
 
 class QuadArrayIndexBuffer;
@@ -125,7 +138,7 @@ public:
 
     void PreCopyBarrier();
 
-    void CopyBuffer(VkBuffer src_buffer, VkBuffer dst_buffer,
+    void CopyBuffer(VkBuffer dst_buffer, VkBuffer src_buffer,
                     std::span<const VideoCommon::BufferCopy> copies, bool barrier,
                     bool can_reorder_upload = false);
 
@@ -155,6 +168,46 @@ public:
         return ref.mapped_span;
     }
 
+    [[nodiscard]] VkDeviceSize SparseAlignmentFor(bool sparse_compatible) const noexcept {
+        if (!sparse_compatible || !multi_range_buffers.use_sparse) {
+            return 0;
+        }
+        return multi_range_buffers.block_size;
+    }
+
+    [[nodiscard]] bool PrefersSparseSources() const noexcept {
+        return multi_range_buffers.use_sparse;
+    }
+
+    void ResetMultiRange() noexcept {
+        multi_range_sources.clear();
+        multi_range_total = 0;
+    }
+
+    void PushMultiRangeSource(const Buffer& buffer, u32 offset, u32 size) {
+        const vk::MemoryLocation location = buffer.Location();
+        multi_range_sources.push_back(MultiRangeSource{
+            .handle = buffer.Handle(),
+            .memory = location.memory,
+            .memory_offset = location.offset,
+            .offset = offset,
+            .size = size,
+            .write_tick = buffer.getWriteTick(),
+            .memory_type = location.memory_type,
+        });
+        multi_range_total += size;
+    }
+
+    bool BindMultiRangeStorageBuffer(u64 key, bool is_written);
+
+    void InvalidateMultiRange(u64 key) {
+        multi_range_buffers.Invalidate(key);
+    }
+
+    void OnBufferDeleted(const Buffer& buffer) {
+        multi_range_buffers.DropOwner(scheduler, buffer.Handle());
+    }
+
     void BindUniformBuffer(const Buffer& buffer, u32 offset, u32 size) {
         BindBuffer(buffer, offset, size);
     }
@@ -208,6 +261,10 @@ private:
     std::unique_ptr<Uint8Pass> uint8_pass;
     QuadIndexedPass quad_index_pass;
 
+    MultiRangeBufferCache multi_range_buffers;
+    boost::container::small_vector<MultiRangeSource, 16> multi_range_sources;
+    VkDeviceSize multi_range_total{};
+
     bool limit_dynamic_storage_buffers = false;
     u32 max_dynamic_storage_buffers = (std::numeric_limits<u32>::max)();
 };