2026-09-18 06:06:10

Signed-off-by: lizzie <lizzie@eden-emu.dev>

.patch/ffmpeg/0001-steamdeck.patch

@@ -1,34 +0,0 @@
-diff --git a/libavutil/hwcontext_vulkan.c b/libavutil/hwcontext_vulkan.c
-index 2859ee1..766961f 100644
---- a/libavutil/hwcontext_vulkan.c
-+++ b/libavutil/hwcontext_vulkan.c
-@@ -1688,10 +1688,12 @@ static int setup_queue_families(AVHWDeviceContext *ctx, VkDeviceCreateInfo *cd)
- 
-     hwctx->nb_qf = 0;
-     hwctx->queue_flags = 0;
-+/* SD gamemode vkroots crash
- #ifdef VK_KHR_internally_synchronized_queues
-     if (p->vkctx.extensions & FF_VK_EXT_INTERNAL_QUEUE_SYNC)
-         hwctx->queue_flags |= VK_DEVICE_QUEUE_CREATE_INTERNALLY_SYNCHRONIZED_BIT_KHR;
- #endif
-+*/
- 
-     /* Pick each queue family to use. */
- #define PICK_QF(type, vid_op)                                            \
-diff --git a/libavutil/vulkan.c b/libavutil/vulkan.c
-index b2e575c..5b3c9cf 100644
---- a/libavutil/vulkan.c
-+++ b/libavutil/vulkan.c
-@@ -574,10 +574,12 @@ int ff_vk_exec_pool_init(FFVulkanContext *s, AVVulkanDeviceQueueFamily *qf,
-         e->qf = qf->idx;
-         VkDeviceQueueInfo2 qinfo = {
-             .sType            = VK_STRUCTURE_TYPE_DEVICE_QUEUE_INFO_2,
-+/* SD gamemode vkroots crash
- #ifdef VK_KHR_internally_synchronized_queues
-             .flags            = internal_queue_sync ?
-                                 VK_DEVICE_QUEUE_CREATE_INTERNALLY_SYNCHRONIZED_BIT_KHR : 0,
- #endif
-+*/
-             .queueFamilyIndex = qf->idx,
-             .queueIndex       = e->qi,
-         };

cpmfile.json

@@ -70,9 +70,9 @@
         "version": "v1.3.18"
     },
     "ffmpeg": {
-        "hash": "68f46abf0d5dc47ab95083d59e273131a8df7948b82e62db00f00fa416e1d0bd24b799671b5a60752209bb2ac0f75295055ed0085c89a98113323cde23b69710",
+        "hash": "ed177621176b3961bdcaa339187d3a7688c1c8b060b79c4bb0257cbc67ad7021ae5d5adca5303b45625abbbe3d9aafdd87ce777b8690ac295290d744c875489a",
         "repo": "FFmpeg/FFmpeg",
-        "version": "6efe500d2e"
+        "version": "c7b5f1537d"
     },
     "ffmpeg-ci": {
         "ci": true,

src/common/cpu_features.cpp

@@ -244,7 +244,8 @@ WallClock::WallClock(bool invariant_, u64 rdtsc_frequency_) noexcept
     , ns_rdtsc_factor{invariant_ ? GetFixedPoint64Factor(NsRatio::den, rdtsc_frequency_) : 0}
     , us_rdtsc_factor{invariant_ ? GetFixedPoint64Factor(UsRatio::den, rdtsc_frequency_) : 0}
     , ms_rdtsc_factor{invariant_ ? GetFixedPoint64Factor(MsRatio::den, rdtsc_frequency_) : 0}
-    , rdtsc_ns_factor{invariant_ ? GetFixedPoint64Factor(rdtsc_frequency_, NsRatio::den) : 1}
+    , rdtsc_ns_integer{invariant_ ? rdtsc_frequency_ / NsRatio::den : 1}
+    , rdtsc_ns_factor{invariant_ ? GetFixedPoint64Factor(rdtsc_frequency_ % NsRatio::den, NsRatio::den) : 0}
     , cntpct_rdtsc_factor{invariant_ ? GetFixedPoint64Factor(CNTFRQ, rdtsc_frequency_) : 0}
     , gputick_rdtsc_factor{invariant_ ? GetFixedPoint64Factor(GPUTickFreq, rdtsc_frequency_) : 0}
     , invariant{invariant_}
@@ -291,7 +292,7 @@ bool WallClock::IsNative() const {
 }
 
 u64 WallClock::NsToTicks(std::chrono::nanoseconds ns) const {
-    return invariant ? MultiplyHigh(ns.count(), rdtsc_ns_factor) : ns.count();
+    return ns.count() * rdtsc_ns_integer + MultiplyHigh(ns.count(), rdtsc_ns_factor);
 }
 #elif defined(HAS_NCE)
 namespace {
@@ -416,7 +417,7 @@ u64 WallClock::NsToTicks(std::chrono::nanoseconds ns) const {
 const WallClock g_wall_clock = [] {
 #if defined(ARCHITECTURE_x86_64)
     auto const& caps = Common::g_cpu_caps;
-    return WallClock(caps.invariant_tsc && caps.tsc_frequency >= std::nano::den, caps.tsc_frequency);
+    return WallClock(caps.invariant_tsc && caps.tsc_frequency > std::nano::den, caps.tsc_frequency);
 #elif defined(HAS_NCE)
     return WallClock(false, 1);
 #else

src/common/cpu_features.h

@@ -96,6 +96,7 @@ public:
     u64 ns_rdtsc_factor;
     u64 us_rdtsc_factor;
     u64 ms_rdtsc_factor;
+    u64 rdtsc_ns_integer;
     u64 rdtsc_ns_factor;
     u64 cntpct_rdtsc_factor;
     u64 gputick_rdtsc_factor;

src/core/arm/dynarmic/arm_dynarmic_32.cpp

@@ -22,21 +22,15 @@ DynarmicCallbacks32::DynarmicCallbacks32(ArmDynarmic32& parent, Kernel::KProcess
     , m_check_memory_access{m_debugger_enabled || !Settings::values.cpuopt_ignore_memory_aborts.GetValue()}
 {}
 
-u8 DynarmicCallbacks32::MemoryRead8(u32 vaddr) {
-    CheckMemoryAccess(vaddr, 1, Kernel::DebugWatchpointType::Read);
-    return m_memory.Read8(vaddr);
-}
-u16 DynarmicCallbacks32::MemoryRead16(u32 vaddr) {
-    CheckMemoryAccess(vaddr, 2, Kernel::DebugWatchpointType::Read);
-    return m_memory.Read16(vaddr);
-}
-u32 DynarmicCallbacks32::MemoryRead32(u32 vaddr) {
-    CheckMemoryAccess(vaddr, 4, Kernel::DebugWatchpointType::Read);
-    return m_memory.Read32(vaddr);
-}
-u64 DynarmicCallbacks32::MemoryRead64(u32 vaddr) {
-    CheckMemoryAccess(vaddr, 8, Kernel::DebugWatchpointType::Read);
-    return m_memory.Read64(vaddr);
+u64 DynarmicCallbacks32::MemoryRead(u32 vaddr, size_t size) {
+    CheckMemoryAccess(vaddr, size, Kernel::DebugWatchpointType::Read);
+    switch (size) {
+    case sizeof(u64): return m_memory.Read64(vaddr);
+    case sizeof(u32): return m_memory.Read32(vaddr);
+    case sizeof(u16): return m_memory.Read16(vaddr);
+    case sizeof(u8): return m_memory.Read8(vaddr);
+    default: UNREACHABLE();
+    }
 }
 
 std::optional<u32> DynarmicCallbacks32::MemoryReadCode(u32 vaddr) {
@@ -50,27 +44,17 @@ std::optional<u32> DynarmicCallbacks32::MemoryReadCode(u32 vaddr) {
     return cached_code_page.inst[(vaddr & Core::Memory::YUZU_PAGEMASK) / sizeof(u32)];
 }
 
-void DynarmicCallbacks32::MemoryWrite8(u32 vaddr, u8 value) {
-    if (CheckMemoryAccess(vaddr, 1, Kernel::DebugWatchpointType::Write)) {
-        m_memory.Write8(vaddr, value);
+void DynarmicCallbacks32::MemoryWrite(Dynarmic::A32::VAddr vaddr, u64 value, size_t size) {
+    if (CheckMemoryAccess(vaddr, size, Kernel::DebugWatchpointType::Write)) {
+        switch (size) {
+        case sizeof(u64): return m_memory.Write64(vaddr, value);
+        case sizeof(u32): return m_memory.Write32(vaddr, u32(value));
+        case sizeof(u16): return m_memory.Write16(vaddr, u16(value));
+        case sizeof(u8): return m_memory.Write8(vaddr, u8(value));
+        default: UNREACHABLE();
+        }
     }
 }
-void DynarmicCallbacks32::MemoryWrite16(u32 vaddr, u16 value) {
-    if (CheckMemoryAccess(vaddr, 2, Kernel::DebugWatchpointType::Write)) {
-        m_memory.Write16(vaddr, value);
-    }
-}
-void DynarmicCallbacks32::MemoryWrite32(u32 vaddr, u32 value) {
-    if (CheckMemoryAccess(vaddr, 4, Kernel::DebugWatchpointType::Write)) {
-        m_memory.Write32(vaddr, value);
-    }
-}
-void DynarmicCallbacks32::MemoryWrite64(u32 vaddr, u64 value) {
-    if (CheckMemoryAccess(vaddr, 8, Kernel::DebugWatchpointType::Write)) {
-        m_memory.Write64(vaddr, value);
-    }
-}
-
 bool DynarmicCallbacks32::MemoryWriteExclusive8(u32 vaddr, u8 value, u8 expected) {
     return CheckMemoryAccess(vaddr, 1, Kernel::DebugWatchpointType::Write) &&
             m_memory.WriteExclusive8(vaddr, value, expected);

src/core/arm/dynarmic/arm_dynarmic_32.h

@@ -30,18 +30,12 @@ class System;
 class DynarmicCallbacks32 : public Dynarmic::A32::UserCallbacks {
 public:
     explicit DynarmicCallbacks32(ArmDynarmic32& parent, Kernel::KProcess* process);
-    u8 MemoryRead8(u32 vaddr) override;
-    u16 MemoryRead16(u32 vaddr) override;
-    u32 MemoryRead32(u32 vaddr) override;
-    u64 MemoryRead64(u32 vaddr) override;
+    u64 MemoryRead(u32 vaddr, size_t size) override;
     std::optional<u32> MemoryReadCode(u32 vaddr) override;
     void InstructionSynchronizationBarrierRaised() override {
         last_code_addr = u64(-1); //reset back, force refetch
     }
-    void MemoryWrite8(u32 vaddr, u8 value) override;
-    void MemoryWrite16(u32 vaddr, u16 value) override;
-    void MemoryWrite32(u32 vaddr, u32 value) override;
-    void MemoryWrite64(u32 vaddr, u64 value) override;
+    void MemoryWrite(Dynarmic::A32::VAddr vaddr, u64 value, size_t size) override;
     bool MemoryWriteExclusive8(u32 vaddr, u8 value, u8 expected) override;
     bool MemoryWriteExclusive16(u32 vaddr, u16 value, u16 expected) override;
     bool MemoryWriteExclusive32(u32 vaddr, u32 value, u32 expected) override;

src/core/arm/dynarmic/arm_dynarmic_64.cpp

@@ -10,6 +10,7 @@
 #include "core/arm/dynarmic/dynarmic_exclusive_monitor.h"
 #include "core/core_timing.h"
 #include "core/hle/kernel/k_process.h"
+#include "dynarmic/interface/A64/config.h"
 
 namespace Core {
 
@@ -21,21 +22,15 @@ DynarmicCallbacks64::DynarmicCallbacks64(ArmDynarmic64& parent, Kernel::KProcess
     , m_check_memory_access{m_debugger_enabled || !Settings::values.cpuopt_ignore_memory_aborts.GetValue()}
 {}
 
-u8 DynarmicCallbacks64::MemoryRead8(u64 vaddr) {
-    CheckMemoryAccess(vaddr, 1, Kernel::DebugWatchpointType::Read);
-    return m_memory.Read8(vaddr);
-}
-u16 DynarmicCallbacks64::MemoryRead16(u64 vaddr) {
-    CheckMemoryAccess(vaddr, 2, Kernel::DebugWatchpointType::Read);
-    return m_memory.Read16(vaddr);
-}
-u32 DynarmicCallbacks64::MemoryRead32(u64 vaddr) {
-    CheckMemoryAccess(vaddr, 4, Kernel::DebugWatchpointType::Read);
-    return m_memory.Read32(vaddr);
-}
-u64 DynarmicCallbacks64::MemoryRead64(u64 vaddr) {
-    CheckMemoryAccess(vaddr, 8, Kernel::DebugWatchpointType::Read);
-    return m_memory.Read64(vaddr);
+u64 DynarmicCallbacks64::MemoryRead(u64 vaddr, size_t size) {
+    CheckMemoryAccess(vaddr, size, Kernel::DebugWatchpointType::Read);
+    switch (size) {
+    case sizeof(u64): return m_memory.Read64(vaddr);
+    case sizeof(u32): return m_memory.Read32(vaddr);
+    case sizeof(u16): return m_memory.Read16(vaddr);
+    case sizeof(u8): return m_memory.Read8(vaddr);
+    default: UNREACHABLE();
+    }
 }
 Dynarmic::A64::Vector DynarmicCallbacks64::MemoryRead128(u64 vaddr) {
     CheckMemoryAccess(vaddr, 16, Kernel::DebugWatchpointType::Read);
@@ -53,24 +48,15 @@ std::optional<u32> DynarmicCallbacks64::MemoryReadCode(u64 vaddr) {
     return cached_code_page.inst[(vaddr & Core::Memory::YUZU_PAGEMASK) / sizeof(u32)];
 }
 
-void DynarmicCallbacks64::MemoryWrite8(u64 vaddr, u8 value) {
-    if (CheckMemoryAccess(vaddr, 1, Kernel::DebugWatchpointType::Write)) {
-        m_memory.Write8(vaddr, value);
-    }
-}
-void DynarmicCallbacks64::MemoryWrite16(u64 vaddr, u16 value) {
-    if (CheckMemoryAccess(vaddr, 2, Kernel::DebugWatchpointType::Write)) {
-        m_memory.Write16(vaddr, value);
-    }
-}
-void DynarmicCallbacks64::MemoryWrite32(u64 vaddr, u32 value) {
-    if (CheckMemoryAccess(vaddr, 4, Kernel::DebugWatchpointType::Write)) {
-        m_memory.Write32(vaddr, value);
-    }
-}
-void DynarmicCallbacks64::MemoryWrite64(u64 vaddr, u64 value) {
-    if (CheckMemoryAccess(vaddr, 8, Kernel::DebugWatchpointType::Write)) {
-        m_memory.Write64(vaddr, value);
+void DynarmicCallbacks64::MemoryWrite(Dynarmic::A64::VAddr vaddr, u64 value, std::size_t size) {
+    if (CheckMemoryAccess(vaddr, size, Kernel::DebugWatchpointType::Write)) {
+        switch (size) {
+        case sizeof(u64): return m_memory.Write64(vaddr, u64(value));
+        case sizeof(u32): return m_memory.Write32(vaddr, u32(value));
+        case sizeof(u16): return m_memory.Write16(vaddr, u16(value));
+        case sizeof(u8): return m_memory.Write8(vaddr, u8(value));
+        default: UNREACHABLE();
+        }
     }
 }
 void DynarmicCallbacks64::MemoryWrite128(u64 vaddr, Dynarmic::A64::Vector value) {

src/core/arm/dynarmic/arm_dynarmic_64.h

@@ -16,6 +16,7 @@
 #include "common/hash.h"
 #include "core/arm/arm_interface.h"
 #include "core/arm/dynarmic/dynarmic_exclusive_monitor.h"
+#include "dynarmic/interface/A64/config.h"
 
 namespace Core::Memory {
 class Memory;
@@ -36,19 +37,13 @@ class DynarmicCallbacks64 : public Dynarmic::A64::UserCallbacks {
 public:
     explicit DynarmicCallbacks64(ArmDynarmic64& parent, Kernel::KProcess* process);
 
-    u8 MemoryRead8(u64 vaddr) override;
-    u16 MemoryRead16(u64 vaddr) override;
-    u32 MemoryRead32(u64 vaddr) override;
-    u64 MemoryRead64(u64 vaddr) override;
+    u64 MemoryRead(u64 vaddr, size_t size) override;
     Dynarmic::A64::Vector MemoryRead128(u64 vaddr) override;
     std::optional<u32> MemoryReadCode(u64 vaddr) override;
     void InstructionSynchronizationBarrierRaised() override {
         last_code_addr = u64(-1); //reset back, force refetch
     }
-    void MemoryWrite8(u64 vaddr, u8 value) override;
-    void MemoryWrite16(u64 vaddr, u16 value) override;
-    void MemoryWrite32(u64 vaddr, u32 value) override;
-    void MemoryWrite64(u64 vaddr, u64 value) override;
+    void MemoryWrite(Dynarmic::A64::VAddr vaddr, u64 value, std::size_t size) override;
     void MemoryWrite128(u64 vaddr, Dynarmic::A64::Vector value) override;
     bool MemoryWriteExclusive8(u64 vaddr, std::uint8_t value, std::uint8_t expected) override;
     bool MemoryWriteExclusive16(u64 vaddr, std::uint16_t value, std::uint16_t expected) override;

src/core/file_sys/patch_manager.cpp

@@ -9,6 +9,7 @@
 #include <cstddef>
 #include <cstring>
 
+#include "common/assert.h"
 #include "common/hex_util.h"
 #include "common/logging.h"
 #include "common/settings.h"
@@ -156,12 +157,15 @@ std::string GetUpdateVersionStringFromSlot(const ContentProvider* provider, u64
     NACP nacp{nacp_file};
     return nacp.GetVersionString();
 }
+YUZU_NO_INLINE std::optional<u64> GetParentApplicationId(const ContentProvider* provider, u64 title_id) {
+    return provider == nullptr ? std::nullopt : provider->GetParentApplicationId(title_id);
+}
 } // Anonymous namespace
 
 PatchManager::PatchManager(u64 title_id_,
                            const Service::FileSystem::FileSystemController& fs_controller_,
                            const ContentProvider& content_provider_)
-    : title_id{title_id_}, fs_controller{fs_controller_}, content_provider{content_provider_} {}
+    : title_id{title_id_}, parent_title_id{GetParentApplicationId(std::addressof(content_provider_), title_id_)}, fs_controller{fs_controller_}, content_provider{content_provider_} {}
 
 PatchManager::~PatchManager() = default;
 
@@ -169,13 +173,35 @@ u64 PatchManager::GetTitleID() const {
     return title_id;
 }
 
+VirtualDir PatchManager::GetModificationLoadRoot(bool sdmc) const {
+    const auto get_root = [&](u64 id) {
+        return sdmc ? fs_controller.GetSDMCModificationLoadRoot(id) : fs_controller.GetModificationLoadRoot(id);
+    };
+    auto root = get_root(title_id);
+    if (!parent_title_id) {
+        return root;
+    }
+    std::vector<VirtualDir> roots{std::move(root), get_root(*parent_title_id)};
+    std::erase(roots, nullptr);
+    return LayeredVfsDirectory::MakeLayeredDirectory(std::move(roots));
+}
+
+std::vector<std::string> PatchManager::GetDisabledAddons() const {
+    auto disabled = Settings::values.disabled_addons[title_id];
+    if (parent_title_id) {
+        const auto& shared = Settings::values.disabled_addons[*parent_title_id];
+        disabled.insert(disabled.end(), shared.begin(), shared.end());
+    }
+    return disabled;
+}
+
 VirtualDir PatchManager::PatchExeFS(VirtualDir exefs) const {
     LOG_INFO(Loader, "Patching ExeFS for title_id={:016X}", title_id);
 
     if (exefs == nullptr)
         return exefs;
 
-    const auto& disabled = Settings::values.disabled_addons[title_id];
+    const auto disabled = GetDisabledAddons();
 
     bool update_disabled = true;
     std::optional<u32> enabled_version;
@@ -303,8 +329,8 @@ VirtualDir PatchManager::PatchExeFS(VirtualDir exefs) const {
     }
 
     // LayeredExeFS
-    const auto load_dir = fs_controller.GetModificationLoadRoot(title_id);
-    const auto sdmc_load_dir = fs_controller.GetSDMCModificationLoadRoot(title_id);
+    const auto load_dir = GetModificationLoadRoot();
+    const auto sdmc_load_dir = GetModificationLoadRoot(true);
 
     std::vector<VirtualDir> patch_dirs = {sdmc_load_dir};
     if (load_dir != nullptr) {
@@ -346,7 +372,7 @@ VirtualDir PatchManager::PatchExeFS(VirtualDir exefs) const {
 }
 
 std::vector<VirtualFile> PatchManager::CollectPatches(const std::vector<VirtualDir>& patch_dirs, const std::string& build_id) const {
-    const auto& disabled = Settings::values.disabled_addons[title_id];
+    const auto disabled = GetDisabledAddons();
     const auto nso_build_id = fmt::format("{:0<64}", build_id);
 
     std::vector<VirtualFile> out;
@@ -405,7 +431,7 @@ std::vector<u8> PatchManager::PatchNSO(const std::vector<u8>& nso, const std::st
 
     LOG_INFO(Loader, "Patching NSO for name={}, build_id={}", name, build_id);
 
-    const auto load_dir = fs_controller.GetModificationLoadRoot(title_id);
+    const auto load_dir = GetModificationLoadRoot();
     if (load_dir == nullptr) {
         LOG_ERROR(Loader, "Cannot load mods for invalid title_id={:016X}", title_id);
         return nso;
@@ -448,7 +474,7 @@ bool PatchManager::HasNSOPatch(const BuildID& build_id_, std::string_view name)
 
     LOG_INFO(Loader, "Querying NSO patch existence for build_id={}, name={}", build_id, name);
 
-    const auto load_dir = fs_controller.GetModificationLoadRoot(title_id);
+    const auto load_dir = GetModificationLoadRoot();
     if (load_dir == nullptr) {
         LOG_ERROR(Loader, "Cannot load mods for invalid title_id={:016X}", title_id);
         return false;
@@ -462,13 +488,13 @@ bool PatchManager::HasNSOPatch(const BuildID& build_id_, std::string_view name)
 }
 
 std::vector<Core::Memory::CheatEntry> PatchManager::CreateCheatList(const BuildID& build_id_) const {
-    const auto load_dir = fs_controller.GetModificationLoadRoot(title_id);
+    const auto load_dir = GetModificationLoadRoot();
     if (load_dir == nullptr) {
         LOG_ERROR(Loader, "Cannot load mods for invalid title_id={:016X}", title_id);
         return {};
     }
 
-    const auto& disabled = Settings::values.disabled_addons[title_id];
+    const auto disabled = GetDisabledAddons();
     auto patch_dirs = load_dir->GetSubdirectories();
     std::sort(patch_dirs.begin(), patch_dirs.end(), [](auto const& l, auto const& r) { return l->GetName() < r->GetName(); });
 
@@ -502,17 +528,16 @@ std::vector<Core::Memory::CheatEntry> PatchManager::CreateCheatList(const BuildI
     return out;
 }
 
-static void ApplyLayeredFS(VirtualFile& romfs, u64 title_id, ContentRecordType type,
-                           const Service::FileSystem::FileSystemController& fs_controller) {
-    const auto load_dir = fs_controller.GetModificationLoadRoot(title_id);
-    const auto sdmc_load_dir = fs_controller.GetSDMCModificationLoadRoot(title_id);
+void PatchManager::ApplyLayeredFS(VirtualFile& romfs, ContentRecordType type) const {
+    const auto load_dir = GetModificationLoadRoot();
+    const auto sdmc_load_dir = GetModificationLoadRoot(true);
     if ((type != ContentRecordType::Program && type != ContentRecordType::Data &&
          type != ContentRecordType::HtmlDocument) ||
         (load_dir == nullptr && sdmc_load_dir == nullptr)) {
         return;
     }
 
-    const auto& disabled = Settings::values.disabled_addons[title_id];
+    const auto disabled = GetDisabledAddons();
     std::vector<VirtualDir> patch_dirs = load_dir->GetSubdirectories();
     if (std::find(disabled.cbegin(), disabled.cend(), "SDMC") == disabled.cend()) {
         patch_dirs.push_back(sdmc_load_dir);
@@ -591,7 +616,7 @@ VirtualFile PatchManager::PatchRomFS(const NCA* base_nca, VirtualFile base_romfs
 
     // Game Updates
     const auto update_tid = GetUpdateTitleID(title_id);
-    const auto& disabled = Settings::values.disabled_addons[title_id];
+    const auto disabled = GetDisabledAddons();
 
     bool update_disabled = true;
     std::optional<u32> enabled_version;
@@ -698,7 +723,7 @@ VirtualFile PatchManager::PatchRomFS(const NCA* base_nca, VirtualFile base_romfs
 
     // LayeredFS
     if (apply_layeredfs) {
-        ApplyLayeredFS(romfs, title_id, type, fs_controller);
+        ApplyLayeredFS(romfs, type);
     }
 
     return romfs;
@@ -1072,15 +1097,22 @@ std::vector<Patch> PatchManager::GetPatches(VirtualFile update_raw) const {
 std::optional<u32> PatchManager::GetGameVersion() const {
     const auto update_tid = GetUpdateTitleID(title_id);
     if (content_provider.HasEntry(update_tid, ContentRecordType::Program)) {
-        return content_provider.GetEntryVersion(update_tid);
+        const auto version = content_provider.GetEntryVersion(update_tid);
+        return version || !parent_title_id ? version : content_provider.GetEntryVersion(GetUpdateTitleID(*parent_title_id));
     }
 
-    return content_provider.GetEntryVersion(title_id);
+    const auto version = content_provider.GetEntryVersion(title_id);
+    return version || !parent_title_id ? version : content_provider.GetEntryVersion(*parent_title_id);
 }
 
 PatchManager::Metadata PatchManager::GetControlMetadata() const {
     const auto base_control_nca = content_provider.GetEntry(title_id, ContentRecordType::Control);
     if (base_control_nca == nullptr) {
+        if (parent_title_id) {
+            const auto control_id = content_provider.HasEntry(*parent_title_id, ContentRecordType::Control) ? *parent_title_id : GetUpdateTitleID(*parent_title_id);
+            const PatchManager parent{control_id, fs_controller, content_provider};
+            return parent.GetControlMetadata();
+        }
         return {};
     }
 

src/core/file_sys/patch_manager.h

@@ -109,10 +109,14 @@ public:
     [[nodiscard]] static PatchManager::Metadata GetMetadataFromBaseOrUpdate(Core::System& system, u64 application_id) noexcept;
 
 private:
+    [[nodiscard]] VirtualDir GetModificationLoadRoot(bool sdmc = false) const;
+    [[nodiscard]] std::vector<std::string> GetDisabledAddons() const;
+    void ApplyLayeredFS(VirtualFile& romfs, ContentRecordType type) const;
     [[nodiscard]] std::vector<VirtualFile> CollectPatches(const std::vector<VirtualDir>& patch_dirs,
                                                           const std::string& build_id) const;
 
     u64 title_id;
+    std::optional<u64> parent_title_id;
     const Service::FileSystem::FileSystemController& fs_controller;
     const ContentProvider& content_provider;
 };

src/core/file_sys/registered_cache.cpp

@@ -5,6 +5,7 @@
 // SPDX-License-Identifier: GPL-2.0-or-later
 
 #include <algorithm>
+#include <limits>
 #include <random>
 #include <regex>
 #include <openssl/evp.h>
@@ -348,6 +349,22 @@ std::vector<ContentProviderEntry> ContentProvider::ListEntries() const {
     return ListEntriesFilter(std::nullopt, std::nullopt, std::nullopt);
 }
 
+std::optional<u64> ContentProvider::GetParentApplicationId(u64 program_id) const {
+    const auto application_id = GetBaseTitleID(program_id);
+    const auto program_index = program_id - application_id;
+    if (program_index == 0 || program_index > std::numeric_limits<u8>::max()) {
+        return std::nullopt;
+    }
+    if (!ListEntriesFilter(TitleType::Application, ContentRecordType::Meta, program_id).empty()) {
+        return std::nullopt;
+    }
+    if ((!ListEntriesFilter(TitleType::Application, ContentRecordType::Meta, application_id).empty() && HasEntry(program_id, ContentRecordType::Program))
+     || (!ListEntriesFilter(TitleType::Update, ContentRecordType::Meta, GetUpdateTitleID(application_id)).empty() && HasEntry(GetUpdateTitleID(program_id), ContentRecordType::Program))) {
+        return application_id;
+    }
+    return std::nullopt;
+}
+
 PlaceholderCache::PlaceholderCache(VirtualDir dir_) : dir(std::move(dir_)) {}
 
 bool PlaceholderCache::Create(const NcaID& id, u64 size) const {

src/core/file_sys/registered_cache.h

@@ -100,6 +100,8 @@ public:
         std::optional<TitleType> title_type = {}, std::optional<ContentRecordType> record_type = {},
         std::optional<u64> title_id = {}) const = 0;
 
+    [[nodiscard]] std::optional<u64> GetParentApplicationId(u64 program_id) const;
+
 protected:
     // A single instance of KeyManager to be used by GetEntry()
     Core::Crypto::KeyManager& keys = Core::Crypto::KeyManager::Instance();

src/core/file_sys/savedata_factory.cpp

@@ -9,6 +9,7 @@
 #include "common/logging.h"
 #include "common/uuid.h"
 #include "core/core.h"
+#include "core/file_sys/registered_cache.h"
 #include "core/file_sys/savedata_factory.h"
 #include "core/file_sys/vfs/vfs.h"
 
@@ -61,17 +62,24 @@ SaveDataFactory::SaveDataFactory(Core::System& system_, ProgramId program_id_,
 
 SaveDataFactory::~SaveDataFactory() = default;
 
+std::string SaveDataFactory::GetSaveDataPath(SaveDataSpaceId space, SaveDataType type, u64 title_id, u128 user_id, u64 save_id) const {
+    if (type == SaveDataType::Account || type == SaveDataType::Device) {
+        const auto requested_id = title_id != 0 ? title_id : program_id;
+        const auto parent_id = system.GetContentProvider().GetParentApplicationId(requested_id);
+        title_id = parent_id.value_or(requested_id);
+    }
+    return GetFullPath(program_id, dir, space, type, title_id, user_id, save_id);
+}
+
 VirtualDir SaveDataFactory::Create(SaveDataSpaceId space, const SaveDataAttribute& meta) const {
-    const auto save_directory = GetFullPath(program_id, dir, space, meta.type, meta.program_id,
-                                            meta.user_id, meta.system_save_data_id);
+    const auto save_directory = GetSaveDataPath(space, meta.type, meta.program_id, meta.user_id, meta.system_save_data_id);
 
     return dir->CreateDirectoryRelative(save_directory);
 }
 
 VirtualDir SaveDataFactory::Open(SaveDataSpaceId space, const SaveDataAttribute& meta) const {
 
-    const auto save_directory = GetFullPath(program_id, dir, space, meta.type, meta.program_id,
-                                            meta.user_id, meta.system_save_data_id);
+    const auto save_directory = GetSaveDataPath(space, meta.type, meta.program_id, meta.user_id, meta.system_save_data_id);
 
     auto out = dir->GetDirectoryRelative(save_directory);
 
@@ -154,8 +162,7 @@ std::string SaveDataFactory::GetUserGameSaveDataRoot(u128 user_id, bool future)
 
 SaveDataSize SaveDataFactory::ReadSaveDataSize(SaveDataType type, u64 title_id,
                                                u128 user_id) const {
-    const auto path =
-        GetFullPath(program_id, dir, SaveDataSpaceId::User, type, title_id, user_id, 0);
+    const auto path = GetSaveDataPath(SaveDataSpaceId::User, type, title_id, user_id, 0);
     const auto relative_dir = GetOrCreateDirectoryRelative(dir, path);
 
     const auto size_file = relative_dir->GetFile(GetSaveDataSizeFileName());
@@ -173,8 +180,7 @@ SaveDataSize SaveDataFactory::ReadSaveDataSize(SaveDataType type, u64 title_id,
 
 void SaveDataFactory::WriteSaveDataSize(SaveDataType type, u64 title_id, u128 user_id,
                                         SaveDataSize new_value) const {
-    const auto path =
-        GetFullPath(program_id, dir, SaveDataSpaceId::User, type, title_id, user_id, 0);
+    const auto path = GetSaveDataPath(SaveDataSpaceId::User, type, title_id, user_id, 0);
     const auto relative_dir = GetOrCreateDirectoryRelative(dir, path);
 
     const auto size_file = relative_dir->CreateFile(GetSaveDataSizeFileName());

src/core/file_sys/savedata_factory.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
 
@@ -47,6 +50,7 @@ public:
     void SetAutoCreate(bool state);
 
 private:
+    std::string GetSaveDataPath(SaveDataSpaceId space, SaveDataType type, u64 title_id, u128 user_id, u64 save_id) const;
     Core::System& system;
     ProgramId program_id;
     VirtualDir dir;

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

@@ -92,6 +92,13 @@ public:
         }
     }
 
+    void RequestFocusStateChangedNotification(Kernel::KernelCore& kernel) {
+        if (m_focus_state_changed_notification_enabled) {
+            m_has_focus_state_changed = true;
+            this->SignalSystemEventIfNeeded(kernel);
+        }
+    }
+
     void OnOperationAndPerformanceModeChanged(Kernel::KernelCore& kernel);
 
 public:

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

@@ -164,14 +164,14 @@ Result ILibraryAppletAccessor::PushInData(SharedPointer<IStorage> storage) {
 Result ILibraryAppletAccessor::PopOutData(Out<SharedPointer<IStorage>> out_storage) {
     LOG_DEBUG(Service_AM, "called");
 
-    if (auto caller_applet = m_applet->caller_applet.lock(); caller_applet) {
-        caller_applet->lifecycle_manager.GetSystemEvent().Signal(system.Kernel());
-        caller_applet->lifecycle_manager.RequestResumeNotification();
-        caller_applet->lifecycle_manager.GetSystemEvent().Clear(system.Kernel());
-        caller_applet->lifecycle_manager.UpdateRequestedFocusState();
-    }
-
     R_TRY(m_broker->GetOutData().Pop(system.Kernel(), out_storage.Get()));
+    if (auto caller_applet = m_applet->caller_applet.lock(); caller_applet) {
+        std::scoped_lock lk{caller_applet->lock};
+        const bool focus_state_changed = caller_applet->lifecycle_manager.UpdateRequestedFocusState();
+        const bool is_front_app = m_applet->frontend && caller_applet->lifecycle_manager.IsApplication();
+        if (focus_state_changed) caller_applet->lifecycle_manager.SignalSystemEventIfNeeded(system.Kernel());
+        else if (is_front_app) caller_applet->lifecycle_manager.RequestFocusStateChangedNotification(system.Kernel());
+    }
 
     if (m_applet->applet_id == AppletId::ProfileSelect && *out_storage) {
         auto impl = (*out_storage)->GetImpl();

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

@@ -121,7 +121,7 @@ IHidSystemServer::IHidSystemServer(Core::System& system_, std::shared_ptr<Resour
         {547, nullptr, "GetAllowedBluetoothLinksCount"},
         {548, &IHidSystemServer::GetRegisteredDevices, "GetRegisteredDevices"},
         {549, nullptr, "GetConnectableRegisteredDevices"},
-        {551, nullptr, "GetRegisteredDevicesForControllerSupport"}, //20.0.0+
+        {551, &IHidSystemServer::GetRegisteredDevices, "GetRegisteredDevicesForControllerSupport"}, //20.0.0+ //mocked via 548 for Diablo 3 (at least)
         {700, nullptr, "ActivateUniquePad"},
         {702, &IHidSystemServer::AcquireUniquePadConnectionEventHandle, "AcquireUniquePadConnectionEventHandle"},
         {703, &IHidSystemServer::GetUniquePadIds, "GetUniquePadIds"},
@@ -758,7 +758,7 @@ void IHidSystemServer::AcquireDeviceRegisteredEventForControllerSupport(HLEReque
 }
 
 void IHidSystemServer::GetRegisteredDevices(HLERequestContext& ctx) {
-    LOG_WARNING(Service_HID, "(STUBBED) called");
+    LOG_WARNING(Service_HID, "(STUBBED) called, command={}", ctx.GetCommand()); //548 or 551
 
     struct RegisterData {
         std::array<u8, 0x68> data;

src/core/hle/service/jit/jit_context.cpp

@@ -71,17 +71,14 @@ public:
     void InstructionSynchronizationBarrierRaised() override {
         last_code_addr = u64(-1); //reset back, force refetch
     }
-    u8 MemoryRead8(u64 vaddr) override {
-        return ReadMemory<u8>(vaddr);
-    }
-    u16 MemoryRead16(u64 vaddr) override {
-        return ReadMemory<u16>(vaddr);
-    }
-    u32 MemoryRead32(u64 vaddr) override {
-        return ReadMemory<u32>(vaddr);
-    }
-    u64 MemoryRead64(u64 vaddr) override {
-        return ReadMemory<u64>(vaddr);
+    u64 MemoryRead(u64 vaddr, size_t size) override {
+        switch (size) {
+        case sizeof(u64): return ReadMemory<u64>(vaddr);
+        case sizeof(u32): return ReadMemory<u32>(vaddr);
+        case sizeof(u16): return ReadMemory<u16>(vaddr);
+        case sizeof(u8): return ReadMemory<u8>(vaddr);
+        default: UNREACHABLE();
+        }
     }
     u128 MemoryRead128(u64 vaddr) override {
         return ReadMemory<u128>(vaddr);
@@ -89,22 +86,19 @@ public:
     std::string MemoryReadCString(u64 vaddr) {
         std::string result{};
         u8 next;
-        while ((next = MemoryRead8(vaddr++)) != 0)
+        while ((next = u8(MemoryRead(vaddr++, sizeof(u8)))) != 0)
             result += char(next);
         return result;
     }
 
-    void MemoryWrite8(u64 vaddr, u8 value) override {
-        WriteMemory<u8>(vaddr, value);
-    }
-    void MemoryWrite16(u64 vaddr, u16 value) override {
-        WriteMemory<u16>(vaddr, value);
-    }
-    void MemoryWrite32(u64 vaddr, u32 value) override {
-        WriteMemory<u32>(vaddr, value);
-    }
-    void MemoryWrite64(u64 vaddr, u64 value) override {
-        WriteMemory<u64>(vaddr, value);
+    void MemoryWrite(u64 vaddr, u64 value, size_t size) override {
+        switch (size) {
+        case sizeof(u64): WriteMemory<u64>(vaddr, u64(value)); break;
+        case sizeof(u32): WriteMemory<u32>(vaddr, u32(value)); break;
+        case sizeof(u16): WriteMemory<u16>(vaddr, u16(value)); break;
+        case sizeof(u8): WriteMemory<u8>(vaddr, u8(value)); break;
+        default: UNREACHABLE();
+        }
     }
     void MemoryWrite128(u64 vaddr, u128 value) override {
         WriteMemory<u128>(vaddr, value);
@@ -193,14 +187,14 @@ public:
         // The loaded NRO file has ELF relocations that must be processed before it can run.
         // Normally this would be processed by RTLD, but in HLE context, we don't have
         // the linker available, so we have to do it ourselves.
-        const VAddr mod_offset{callbacks->MemoryRead32(4)};
-        if (callbacks->MemoryRead32(mod_offset) != Common::MakeMagic('M', 'O', 'D', '0'))
+        const VAddr mod_offset{callbacks->MemoryRead(4, sizeof(u32))};
+        if (callbacks->MemoryRead(mod_offset, sizeof(u32)) != Common::MakeMagic('M', 'O', 'D', '0'))
             return false;
 
         // For more info about dynamic entries, see the ELF ABI specification:
         // https://refspecs.linuxbase.org/elf/gabi4+/ch5.dynamic.html
         // https://refspecs.linuxbase.org/elf/gabi4+/ch4.reloc.html
-        VAddr dynamic_offset{mod_offset + callbacks->MemoryRead32(mod_offset + 4)};
+        VAddr dynamic_offset{mod_offset + callbacks->MemoryRead(mod_offset + 4, sizeof(u32))};
         VAddr rela_dyn = 0, relr_dyn = 0;
         size_t num_rela = 0, num_relr = 0;
         while (true) {
@@ -222,8 +216,8 @@ public:
         for (size_t i = 0; i < num_rela; i++) {
             const auto rela{callbacks->ReadMemory<Elf64_Rela>(rela_dyn + i * sizeof(Elf64_Rela))};
             if (Elf64RelType(rela.r_info) == ElfAArch64Relative) {
-                const VAddr contents{callbacks->MemoryRead64(rela.r_offset)};
-                callbacks->MemoryWrite64(rela.r_offset, contents + rela.r_addend);
+                const VAddr contents{callbacks->MemoryRead(rela.r_offset, sizeof(u64))};
+                callbacks->MemoryWrite(rela.r_offset, contents + rela.r_addend, sizeof(u64));
             }
         }
 
@@ -231,7 +225,7 @@ public:
         for (size_t i = 0; i < num_relr; i++) {
             const auto relr = callbacks->ReadMemory<Elf64_Relr>(relr_dyn + i * sizeof(Elf64_Relr));
             const auto incr = [&](VAddr where) {
-                callbacks->MemoryWrite64(where, callbacks->MemoryRead64(where) + relocbase);
+                callbacks->MemoryWrite(where, callbacks->MemoryRead(where, sizeof(u64)) + relocbase, sizeof(u64));
             };
             if ((relr & 1) == 0) {
                 // where pointer
@@ -294,7 +288,7 @@ public:
         if (argument_stack.size() > 8) {
             const VAddr new_sp = Common::AlignDown(top_of_stack - (argument_stack.size() - 8) * sizeof(u64), STACK_ALIGN);
             for (size_t i = 8; i < argument_stack.size(); i++)
-                callbacks->MemoryWrite64(new_sp + (i - 8) * sizeof(u64), argument_stack[i]);
+                callbacks->MemoryWrite(new_sp + (i - 8) * sizeof(u64), argument_stack[i], sizeof(u64));
             jit->SetSP(new_sp);
         }
         // Reset the call state for the next invocation
@@ -385,17 +379,17 @@ void DynarmicCallbacks64::CallSVC(u32 swi) {
 
         if (dest < src) {
             for (size_t i = 0; i < n; i++)
-                MemoryWrite8(dest + i, MemoryRead8(src + i));
+                MemoryWrite(dest + i, u8(MemoryRead(src + i, sizeof(u8))), sizeof(u8));
         } else {
             for (size_t i = n; i > 0; i--)
-                MemoryWrite8(dest + i - 1, MemoryRead8(src + i - 1));
+                MemoryWrite(dest + i - 1, u8(MemoryRead(src + i - 1, sizeof(u8))), sizeof(u8));
         }
     } else if (pc == parent.helpers[size_t(HelperFn::Memset)]) {
         const VAddr dest{parent.jit->GetRegister(0)};
         const u64 c{parent.jit->GetRegister(1)};
         const size_t n{parent.jit->GetRegister(2)};
         for (size_t i = 0; i < n; i++)
-            MemoryWrite8(dest + i, u8(c));
+            MemoryWrite(dest + i, u8(c), sizeof(u8));
     } else if (pc == parent.helpers[size_t(HelperFn::Resolve)]) {
         // X0 contains a char* for a symbol to resolve
         const auto name{MemoryReadCString(parent.jit->GetRegister(0))};
@@ -422,7 +416,7 @@ void DynarmicCallbacks64::CallSVC(u32 swi) {
 }
 
 void DynarmicCallbacks64::ExceptionRaised(u64 pc, Dynarmic::A64::Exception exception) {
-    auto const inst = MemoryRead32(pc);
+    auto const inst = MemoryRead(pc, sizeof(u32));
     LOG_CRITICAL(Service_JIT, "{} PC @ {:08x}, data = {:08x}", exception, pc, inst);
     parent.jit->HaltExecution();
 }

src/core/tools/freezer.cpp

@@ -33,18 +33,10 @@ u64 MemoryReadWidth(Core::Memory::Memory& memory, u32 width, VAddr addr) {
 
 void MemoryWriteWidth(Core::Memory::Memory& memory, u32 width, VAddr addr, u64 value) {
     switch (width) {
-    case 1:
-        memory.Write8(addr, static_cast<u8>(value));
-        break;
-    case 2:
-        memory.Write16(addr, static_cast<u16>(value));
-        break;
-    case 4:
-        memory.Write32(addr, static_cast<u32>(value));
-        break;
-    case 8:
-        memory.Write64(addr, value);
-        break;
+    case sizeof(u64): return memory.Write64(addr, value);
+    case sizeof(u32): return memory.Write32(addr, u32(value));
+    case sizeof(u16): return memory.Write16(addr, u16(value));
+    case sizeof(u8): return memory.Write8(addr, u8(value));
     default:
         UNREACHABLE();
     }

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

@@ -82,7 +82,7 @@ static void* EmitExclusiveReadCallTrampoline(oaknut::CodeGenerator& code, const
 
     auto fn = [](const A32::UserConfig& conf, A32::VAddr vaddr) -> T {
         return conf.global_monitor->ReadAndMark<T>(conf.processor_id, vaddr, [&]() -> T {
-            return (conf.callbacks->*callback)(vaddr);
+            return (conf.callbacks->*callback)(vaddr, sizeof(T));
         });
     };
 
@@ -136,12 +136,9 @@ static void* EmitExclusiveWriteCallTrampoline(oaknut::CodeGenerator& code, const
     oaknut::Label l_addr, l_this;
 
     auto fn = [](const A32::UserConfig& conf, A32::VAddr vaddr, T value) -> u32 {
-        return conf.global_monitor->DoExclusiveOperation<T>(conf.processor_id, vaddr,
-                                                            [&](T expected) -> bool {
-                                                                return (conf.callbacks->*callback)(vaddr, value, expected);
-                                                            })
-                 ? 0
-                 : 1;
+        return conf.global_monitor->DoExclusiveOperation<T>(conf.processor_id, vaddr, [&](T expected) -> bool {
+            return (conf.callbacks->*callback)(vaddr, value, expected);
+        }) ? 0 : 1;
     };
 
     void* target = code.xptr<void*>();
@@ -179,26 +176,14 @@ void A32AddressSpace::EmitPrelude() {
 
     UnprotectCodeMemory();
 
-    prelude_info.read_memory_8 = EmitCallTrampoline<&A32::UserCallbacks::MemoryRead8>(code, conf.callbacks);
-    prelude_info.read_memory_16 = EmitCallTrampoline<&A32::UserCallbacks::MemoryRead16>(code, conf.callbacks);
-    prelude_info.read_memory_32 = EmitCallTrampoline<&A32::UserCallbacks::MemoryRead32>(code, conf.callbacks);
-    prelude_info.read_memory_64 = EmitCallTrampoline<&A32::UserCallbacks::MemoryRead64>(code, conf.callbacks);
-    prelude_info.wrapped_read_memory_8 = EmitWrappedReadCallTrampoline<&A32::UserCallbacks::MemoryRead8>(code, conf.callbacks);
-    prelude_info.wrapped_read_memory_16 = EmitWrappedReadCallTrampoline<&A32::UserCallbacks::MemoryRead16>(code, conf.callbacks);
-    prelude_info.wrapped_read_memory_32 = EmitWrappedReadCallTrampoline<&A32::UserCallbacks::MemoryRead32>(code, conf.callbacks);
-    prelude_info.wrapped_read_memory_64 = EmitWrappedReadCallTrampoline<&A32::UserCallbacks::MemoryRead64>(code, conf.callbacks);
-    prelude_info.exclusive_read_memory_8 = EmitExclusiveReadCallTrampoline<&A32::UserCallbacks::MemoryRead8, u8>(code, conf);
-    prelude_info.exclusive_read_memory_16 = EmitExclusiveReadCallTrampoline<&A32::UserCallbacks::MemoryRead16, u16>(code, conf);
-    prelude_info.exclusive_read_memory_32 = EmitExclusiveReadCallTrampoline<&A32::UserCallbacks::MemoryRead32, u32>(code, conf);
-    prelude_info.exclusive_read_memory_64 = EmitExclusiveReadCallTrampoline<&A32::UserCallbacks::MemoryRead64, u64>(code, conf);
-    prelude_info.write_memory_8 = EmitCallTrampoline<&A32::UserCallbacks::MemoryWrite8>(code, conf.callbacks);
-    prelude_info.write_memory_16 = EmitCallTrampoline<&A32::UserCallbacks::MemoryWrite16>(code, conf.callbacks);
-    prelude_info.write_memory_32 = EmitCallTrampoline<&A32::UserCallbacks::MemoryWrite32>(code, conf.callbacks);
-    prelude_info.write_memory_64 = EmitCallTrampoline<&A32::UserCallbacks::MemoryWrite64>(code, conf.callbacks);
-    prelude_info.wrapped_write_memory_8 = EmitWrappedWriteCallTrampoline<&A32::UserCallbacks::MemoryWrite8>(code, conf.callbacks);
-    prelude_info.wrapped_write_memory_16 = EmitWrappedWriteCallTrampoline<&A32::UserCallbacks::MemoryWrite16>(code, conf.callbacks);
-    prelude_info.wrapped_write_memory_32 = EmitWrappedWriteCallTrampoline<&A32::UserCallbacks::MemoryWrite32>(code, conf.callbacks);
-    prelude_info.wrapped_write_memory_64 = EmitWrappedWriteCallTrampoline<&A32::UserCallbacks::MemoryWrite64>(code, conf.callbacks);
+    prelude_info.read_memory = EmitCallTrampoline<&A32::UserCallbacks::MemoryRead>(code, conf.callbacks);
+    prelude_info.wrapped_read_memory = EmitWrappedReadCallTrampoline<&A32::UserCallbacks::MemoryRead>(code, conf.callbacks);
+    prelude_info.exclusive_read_memory_8 = EmitExclusiveReadCallTrampoline<&A32::UserCallbacks::MemoryRead, u8>(code, conf);
+    prelude_info.exclusive_read_memory_16 = EmitExclusiveReadCallTrampoline<&A32::UserCallbacks::MemoryRead, u16>(code, conf);
+    prelude_info.exclusive_read_memory_32 = EmitExclusiveReadCallTrampoline<&A32::UserCallbacks::MemoryRead, u32>(code, conf);
+    prelude_info.exclusive_read_memory_64 = EmitExclusiveReadCallTrampoline<&A32::UserCallbacks::MemoryRead, u64>(code, conf);
+    prelude_info.write_memory = EmitCallTrampoline<&A32::UserCallbacks::MemoryWrite>(code, conf.callbacks);
+    prelude_info.wrapped_write_memory = EmitWrappedWriteCallTrampoline<&A32::UserCallbacks::MemoryWrite>(code, conf.callbacks);
     prelude_info.exclusive_write_memory_8 = EmitExclusiveWriteCallTrampoline<&A32::UserCallbacks::MemoryWriteExclusive8, u8>(code, conf);
     prelude_info.exclusive_write_memory_16 = EmitExclusiveWriteCallTrampoline<&A32::UserCallbacks::MemoryWriteExclusive16, u16>(code, conf);
     prelude_info.exclusive_write_memory_32 = EmitExclusiveWriteCallTrampoline<&A32::UserCallbacks::MemoryWriteExclusive32, u32>(code, conf);

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

@@ -81,7 +81,7 @@ static void* EmitExclusiveReadCallTrampoline(oaknut::CodeGenerator& code, const
 
     auto fn = [](const A64::UserConfig& conf, A64::VAddr vaddr) -> T {
         return conf.global_monitor->ReadAndMark<T>(conf.processor_id, vaddr, [&]() -> T {
-            return (conf.callbacks->*callback)(vaddr);
+            return (conf.callbacks->*callback)(vaddr, sizeof(T));
         });
     };
 
@@ -133,14 +133,10 @@ static void* EmitExclusiveWriteCallTrampoline(oaknut::CodeGenerator& code, const
     using namespace oaknut::util;
 
     oaknut::Label l_addr, l_this;
-
     auto fn = [](const A64::UserConfig& conf, A64::VAddr vaddr, T value) -> u32 {
-        return conf.global_monitor->DoExclusiveOperation<T>(conf.processor_id, vaddr,
-                                                            [&](T expected) -> bool {
-                                                                return (conf.callbacks->*callback)(vaddr, value, expected);
-                                                            })
-                 ? 0
-                 : 1;
+        return conf.global_monitor->DoExclusiveOperation<T>(conf.processor_id, vaddr, [&](T expected) -> bool {
+            return (conf.callbacks->*callback)(vaddr, value, expected);
+        }) ? 0 : 1;
     };
 
     void* target = code.xptr<void*>();
@@ -346,30 +342,18 @@ void A64AddressSpace::EmitPrelude() {
 
     UnprotectCodeMemory();
 
-    prelude_info.read_memory_8 = EmitCallTrampoline<&A64::UserCallbacks::MemoryRead8>(code, conf.callbacks);
-    prelude_info.read_memory_16 = EmitCallTrampoline<&A64::UserCallbacks::MemoryRead16>(code, conf.callbacks);
-    prelude_info.read_memory_32 = EmitCallTrampoline<&A64::UserCallbacks::MemoryRead32>(code, conf.callbacks);
-    prelude_info.read_memory_64 = EmitCallTrampoline<&A64::UserCallbacks::MemoryRead64>(code, conf.callbacks);
+    prelude_info.read_memory = EmitCallTrampoline<&A64::UserCallbacks::MemoryRead>(code, conf.callbacks);
     prelude_info.read_memory_128 = EmitRead128CallTrampoline(code, conf.callbacks);
-    prelude_info.wrapped_read_memory_8 = EmitWrappedReadCallTrampoline<&A64::UserCallbacks::MemoryRead8>(code, conf.callbacks);
-    prelude_info.wrapped_read_memory_16 = EmitWrappedReadCallTrampoline<&A64::UserCallbacks::MemoryRead16>(code, conf.callbacks);
-    prelude_info.wrapped_read_memory_32 = EmitWrappedReadCallTrampoline<&A64::UserCallbacks::MemoryRead32>(code, conf.callbacks);
-    prelude_info.wrapped_read_memory_64 = EmitWrappedReadCallTrampoline<&A64::UserCallbacks::MemoryRead64>(code, conf.callbacks);
+    prelude_info.wrapped_read_memory = EmitWrappedReadCallTrampoline<&A64::UserCallbacks::MemoryRead>(code, conf.callbacks);
     prelude_info.wrapped_read_memory_128 = EmitWrappedRead128CallTrampoline(code, conf.callbacks);
-    prelude_info.exclusive_read_memory_8 = EmitExclusiveReadCallTrampoline<&A64::UserCallbacks::MemoryRead8, u8>(code, conf);
-    prelude_info.exclusive_read_memory_16 = EmitExclusiveReadCallTrampoline<&A64::UserCallbacks::MemoryRead16, u16>(code, conf);
-    prelude_info.exclusive_read_memory_32 = EmitExclusiveReadCallTrampoline<&A64::UserCallbacks::MemoryRead32, u32>(code, conf);
-    prelude_info.exclusive_read_memory_64 = EmitExclusiveReadCallTrampoline<&A64::UserCallbacks::MemoryRead64, u64>(code, conf);
+    prelude_info.exclusive_read_memory_8 = EmitExclusiveReadCallTrampoline<&A64::UserCallbacks::MemoryRead, u8>(code, conf);
+    prelude_info.exclusive_read_memory_16 = EmitExclusiveReadCallTrampoline<&A64::UserCallbacks::MemoryRead, u16>(code, conf);
+    prelude_info.exclusive_read_memory_32 = EmitExclusiveReadCallTrampoline<&A64::UserCallbacks::MemoryRead, u32>(code, conf);
+    prelude_info.exclusive_read_memory_64 = EmitExclusiveReadCallTrampoline<&A64::UserCallbacks::MemoryRead, u64>(code, conf);
     prelude_info.exclusive_read_memory_128 = EmitExclusiveRead128CallTrampoline(code, conf);
-    prelude_info.write_memory_8 = EmitCallTrampoline<&A64::UserCallbacks::MemoryWrite8>(code, conf.callbacks);
-    prelude_info.write_memory_16 = EmitCallTrampoline<&A64::UserCallbacks::MemoryWrite16>(code, conf.callbacks);
-    prelude_info.write_memory_32 = EmitCallTrampoline<&A64::UserCallbacks::MemoryWrite32>(code, conf.callbacks);
-    prelude_info.write_memory_64 = EmitCallTrampoline<&A64::UserCallbacks::MemoryWrite64>(code, conf.callbacks);
+    prelude_info.write_memory = EmitCallTrampoline<&A64::UserCallbacks::MemoryWrite>(code, conf.callbacks);
     prelude_info.write_memory_128 = EmitWrite128CallTrampoline(code, conf.callbacks);
-    prelude_info.wrapped_write_memory_8 = EmitWrappedWriteCallTrampoline<&A64::UserCallbacks::MemoryWrite8>(code, conf.callbacks);
-    prelude_info.wrapped_write_memory_16 = EmitWrappedWriteCallTrampoline<&A64::UserCallbacks::MemoryWrite16>(code, conf.callbacks);
-    prelude_info.wrapped_write_memory_32 = EmitWrappedWriteCallTrampoline<&A64::UserCallbacks::MemoryWrite32>(code, conf.callbacks);
-    prelude_info.wrapped_write_memory_64 = EmitWrappedWriteCallTrampoline<&A64::UserCallbacks::MemoryWrite64>(code, conf.callbacks);
+    prelude_info.wrapped_write_memory = EmitWrappedWriteCallTrampoline<&A64::UserCallbacks::MemoryWrite>(code, conf.callbacks);
     prelude_info.wrapped_write_memory_128 = EmitWrappedWrite128CallTrampoline(code, conf.callbacks);
     prelude_info.exclusive_write_memory_8 = EmitExclusiveWriteCallTrampoline<&A64::UserCallbacks::MemoryWriteExclusive8, u8>(code, conf);
     prelude_info.exclusive_write_memory_16 = EmitExclusiveWriteCallTrampoline<&A64::UserCallbacks::MemoryWriteExclusive16, u16>(code, conf);

src/dynarmic/src/dynarmic/backend/arm64/address_space.cpp

@@ -145,32 +145,42 @@ void AddressSpace::Link(EmittedBlockInfo& block_info) {
         case LinkTarget::ReturnFromRunCode:
             c.B(prelude_info.return_from_run_code);
             break;
+        // { this, vaddr, size }
         case LinkTarget::ReadMemory8:
-            c.BL(prelude_info.read_memory_8);
+            c.LDR(X2, 1);
+            c.BL(prelude_info.read_memory);
             break;
         case LinkTarget::ReadMemory16:
-            c.BL(prelude_info.read_memory_16);
+            c.LDR(X2, 2);
+            c.BL(prelude_info.read_memory);
             break;
         case LinkTarget::ReadMemory32:
-            c.BL(prelude_info.read_memory_32);
+            c.LDR(X2, 4);
+            c.BL(prelude_info.read_memory);
             break;
         case LinkTarget::ReadMemory64:
-            c.BL(prelude_info.read_memory_64);
+            c.LDR(X2, 8);
+            c.BL(prelude_info.read_memory);
             break;
         case LinkTarget::ReadMemory128:
             c.BL(prelude_info.read_memory_128);
             break;
+        // { this, vaddr, size }
         case LinkTarget::WrappedReadMemory8:
-            c.BL(prelude_info.wrapped_read_memory_8);
+            c.LDR(X2, 1);
+            c.BL(prelude_info.wrapped_read_memory);
             break;
         case LinkTarget::WrappedReadMemory16:
-            c.BL(prelude_info.wrapped_read_memory_16);
+            c.LDR(X2, 2);
+            c.BL(prelude_info.wrapped_read_memory);
             break;
         case LinkTarget::WrappedReadMemory32:
-            c.BL(prelude_info.wrapped_read_memory_32);
+            c.LDR(X2, 4);
+            c.BL(prelude_info.wrapped_read_memory);
             break;
         case LinkTarget::WrappedReadMemory64:
-            c.BL(prelude_info.wrapped_read_memory_64);
+            c.LDR(X2, 8);
+            c.BL(prelude_info.wrapped_read_memory);
             break;
         case LinkTarget::WrappedReadMemory128:
             c.BL(prelude_info.wrapped_read_memory_128);
@@ -190,32 +200,41 @@ void AddressSpace::Link(EmittedBlockInfo& block_info) {
         case LinkTarget::ExclusiveReadMemory128:
             c.BL(prelude_info.exclusive_read_memory_128);
             break;
+        // { this, vaddr, value, size }
         case LinkTarget::WriteMemory8:
-            c.BL(prelude_info.write_memory_8);
+            c.LDR(X3, 1);
+            c.BL(prelude_info.write_memory);
             break;
         case LinkTarget::WriteMemory16:
-            c.BL(prelude_info.write_memory_16);
+            c.LDR(X3, 2);
+            c.BL(prelude_info.write_memory);
             break;
         case LinkTarget::WriteMemory32:
-            c.BL(prelude_info.write_memory_32);
+            c.LDR(X3, 4);
+            c.BL(prelude_info.write_memory);
             break;
         case LinkTarget::WriteMemory64:
-            c.BL(prelude_info.write_memory_64);
+            c.LDR(X3, 8);
+            c.BL(prelude_info.write_memory);
             break;
         case LinkTarget::WriteMemory128:
             c.BL(prelude_info.write_memory_128);
             break;
         case LinkTarget::WrappedWriteMemory8:
-            c.BL(prelude_info.wrapped_write_memory_8);
+            c.LDR(X3, 1);
+            c.BL(prelude_info.wrapped_write_memory);
             break;
         case LinkTarget::WrappedWriteMemory16:
-            c.BL(prelude_info.wrapped_write_memory_16);
+            c.LDR(X3, 2);
+            c.BL(prelude_info.wrapped_write_memory);
             break;
         case LinkTarget::WrappedWriteMemory32:
-            c.BL(prelude_info.wrapped_write_memory_32);
+            c.LDR(X3, 4);
+            c.BL(prelude_info.wrapped_write_memory);
             break;
         case LinkTarget::WrappedWriteMemory64:
-            c.BL(prelude_info.wrapped_write_memory_64);
+            c.LDR(X3, 8);
+            c.BL(prelude_info.wrapped_write_memory);
             break;
         case LinkTarget::WrappedWriteMemory128:
             c.BL(prelude_info.wrapped_write_memory_128);

src/dynarmic/src/dynarmic/backend/arm64/address_space.h

@@ -93,30 +93,18 @@ protected:
         void* return_to_dispatcher;
         void* return_from_run_code;
 
-        void* read_memory_8;
-        void* read_memory_16;
-        void* read_memory_32;
-        void* read_memory_64;
+        void* read_memory;
         void* read_memory_128;
-        void* wrapped_read_memory_8;
-        void* wrapped_read_memory_16;
-        void* wrapped_read_memory_32;
-        void* wrapped_read_memory_64;
+        void* wrapped_read_memory;
         void* wrapped_read_memory_128;
         void* exclusive_read_memory_8;
         void* exclusive_read_memory_16;
         void* exclusive_read_memory_32;
         void* exclusive_read_memory_64;
         void* exclusive_read_memory_128;
-        void* write_memory_8;
-        void* write_memory_16;
-        void* write_memory_32;
-        void* write_memory_64;
+        void* write_memory;
         void* write_memory_128;
-        void* wrapped_write_memory_8;
-        void* wrapped_write_memory_16;
-        void* wrapped_write_memory_32;
-        void* wrapped_write_memory_64;
+        void* wrapped_write_memory;
         void* wrapped_write_memory_128;
         void* exclusive_write_memory_8;
         void* exclusive_write_memory_16;

src/dynarmic/src/dynarmic/backend/x64/a32_emit_x64_memory.cpp

@@ -11,8 +11,7 @@
 #include <tuple>
 #include <utility>
 
-#include <fmt/format.h>
-#include <fmt/ostream.h>
+#include "common/logging.h"
 #include "dynarmic/mcl/integer_of_size.hpp"
 #include "dynarmic/backend/x64/xbyak.h"
 
@@ -31,16 +30,16 @@ using namespace Xbyak::util;
 void A32EmitX64::GenFastmemFallbacks() {
     const std::initializer_list<int> idxes{0, 1, 2, 3, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14};
     const std::array<std::pair<size_t, ArgCallback>, 4> read_callbacks{{
-        {8, Devirtualize<&A32::UserCallbacks::MemoryRead8>(conf.callbacks)},
-        {16, Devirtualize<&A32::UserCallbacks::MemoryRead16>(conf.callbacks)},
-        {32, Devirtualize<&A32::UserCallbacks::MemoryRead32>(conf.callbacks)},
-        {64, Devirtualize<&A32::UserCallbacks::MemoryRead64>(conf.callbacks)},
+        {8, Devirtualize<&A32::UserCallbacks::MemoryRead>(conf.callbacks)},
+        {16, Devirtualize<&A32::UserCallbacks::MemoryRead>(conf.callbacks)},
+        {32, Devirtualize<&A32::UserCallbacks::MemoryRead>(conf.callbacks)},
+        {64, Devirtualize<&A32::UserCallbacks::MemoryRead>(conf.callbacks)},
     }};
     const std::array<std::pair<size_t, ArgCallback>, 4> write_callbacks{{
-        {8, Devirtualize<&A32::UserCallbacks::MemoryWrite8>(conf.callbacks)},
-        {16, Devirtualize<&A32::UserCallbacks::MemoryWrite16>(conf.callbacks)},
-        {32, Devirtualize<&A32::UserCallbacks::MemoryWrite32>(conf.callbacks)},
-        {64, Devirtualize<&A32::UserCallbacks::MemoryWrite64>(conf.callbacks)},
+        {8, Devirtualize<&A32::UserCallbacks::MemoryWrite>(conf.callbacks)},
+        {16, Devirtualize<&A32::UserCallbacks::MemoryWrite>(conf.callbacks)},
+        {32, Devirtualize<&A32::UserCallbacks::MemoryWrite>(conf.callbacks)},
+        {64, Devirtualize<&A32::UserCallbacks::MemoryWrite>(conf.callbacks)},
     }};
     const std::array<std::pair<size_t, ArgCallback>, 4> exclusive_write_callbacks{{
         {8, Devirtualize<&A32::UserCallbacks::MemoryWriteExclusive8>(conf.callbacks)},
@@ -56,12 +55,12 @@ void A32EmitX64::GenFastmemFallbacks() {
                     code.align();
                     read_fallbacks[std::make_tuple(ordered, bitsize, vaddr_idx, value_idx)] = code.getCurr<void (*)()>();
                     ABI_PushCallerSaveRegistersAndAdjustStackExcept(code, HostLocRegIdx(value_idx));
+                    // params = { this, vaddr, size }
                     if (vaddr_idx != code.ABI_PARAM2.getIdx()) {
                         code.mov(code.ABI_PARAM2, Xbyak::Reg64{vaddr_idx});
                     }
-                    if (ordered) {
-                        code.mfence();
-                    }
+                    code.mov(code.ABI_PARAM3, bitsize / CHAR_BIT);
+                    if (ordered) code.mfence();
                     callback.EmitCall(code);
                     if (value_idx != code.ABI_RETURN.getIdx()) {
                         code.mov(Xbyak::Reg64{value_idx}, code.ABI_RETURN);
@@ -76,6 +75,7 @@ void A32EmitX64::GenFastmemFallbacks() {
                     code.align();
                     write_fallbacks[std::make_tuple(ordered, bitsize, vaddr_idx, value_idx)] = code.getCurr<void (*)()>();
                     ABI_PushCallerSaveRegistersAndAdjustStack(code);
+                    // params = { this, vaddr, value, size }
                     if (vaddr_idx == code.ABI_PARAM3.getIdx() && value_idx == code.ABI_PARAM2.getIdx()) {
                         code.xchg(code.ABI_PARAM2, code.ABI_PARAM3);
                     } else if (vaddr_idx == code.ABI_PARAM3.getIdx()) {
@@ -92,10 +92,9 @@ void A32EmitX64::GenFastmemFallbacks() {
                         }
                     }
                     code.ZeroExtendFrom(bitsize, code.ABI_PARAM3);
+                    code.mov(code.ABI_PARAM4, bitsize / CHAR_BIT);
                     callback.EmitCall(code);
-                    if (ordered) {
-                        code.mfence();
-                    }
+                    if (ordered) code.mfence();
                     ABI_PopCallerSaveRegistersAndAdjustStack(code);
                     code.ret();
                     PerfMapRegister(write_fallbacks[std::make_tuple(ordered, bitsize, vaddr_idx, value_idx)], code.getCurr(), fmt::format("a32_write_fallback_{}", bitsize));
@@ -138,35 +137,35 @@ void A32EmitX64::GenFastmemFallbacks() {
 #undef Axx
 
 void A32EmitX64::EmitA32ReadMemory8(A32EmitContext& ctx, IR::Inst* inst) {
-    EmitMemoryRead<8, &A32::UserCallbacks::MemoryRead8>(ctx, inst);
+    EmitMemoryRead<8, &A32::UserCallbacks::MemoryRead>(ctx, inst);
 }
 
 void A32EmitX64::EmitA32ReadMemory16(A32EmitContext& ctx, IR::Inst* inst) {
-    EmitMemoryRead<16, &A32::UserCallbacks::MemoryRead16>(ctx, inst);
+    EmitMemoryRead<16, &A32::UserCallbacks::MemoryRead>(ctx, inst);
 }
 
 void A32EmitX64::EmitA32ReadMemory32(A32EmitContext& ctx, IR::Inst* inst) {
-    EmitMemoryRead<32, &A32::UserCallbacks::MemoryRead32>(ctx, inst);
+    EmitMemoryRead<32, &A32::UserCallbacks::MemoryRead>(ctx, inst);
 }
 
 void A32EmitX64::EmitA32ReadMemory64(A32EmitContext& ctx, IR::Inst* inst) {
-    EmitMemoryRead<64, &A32::UserCallbacks::MemoryRead64>(ctx, inst);
+    EmitMemoryRead<64, &A32::UserCallbacks::MemoryRead>(ctx, inst);
 }
 
 void A32EmitX64::EmitA32WriteMemory8(A32EmitContext& ctx, IR::Inst* inst) {
-    EmitMemoryWrite<8, &A32::UserCallbacks::MemoryWrite8>(ctx, inst);
+    EmitMemoryWrite<8, &A32::UserCallbacks::MemoryWrite>(ctx, inst);
 }
 
 void A32EmitX64::EmitA32WriteMemory16(A32EmitContext& ctx, IR::Inst* inst) {
-    EmitMemoryWrite<16, &A32::UserCallbacks::MemoryWrite16>(ctx, inst);
+    EmitMemoryWrite<16, &A32::UserCallbacks::MemoryWrite>(ctx, inst);
 }
 
 void A32EmitX64::EmitA32WriteMemory32(A32EmitContext& ctx, IR::Inst* inst) {
-    EmitMemoryWrite<32, &A32::UserCallbacks::MemoryWrite32>(ctx, inst);
+    EmitMemoryWrite<32, &A32::UserCallbacks::MemoryWrite>(ctx, inst);
 }
 
 void A32EmitX64::EmitA32WriteMemory64(A32EmitContext& ctx, IR::Inst* inst) {
-    EmitMemoryWrite<64, &A32::UserCallbacks::MemoryWrite64>(ctx, inst);
+    EmitMemoryWrite<64, &A32::UserCallbacks::MemoryWrite>(ctx, inst);
 }
 
 void A32EmitX64::EmitA32ClearExclusive(A32EmitContext&, IR::Inst*) {
@@ -175,33 +174,33 @@ void A32EmitX64::EmitA32ClearExclusive(A32EmitContext&, IR::Inst*) {
 
 void A32EmitX64::EmitA32ExclusiveReadMemory8(A32EmitContext& ctx, IR::Inst* inst) {
     if (conf.fastmem_exclusive_access) {
-        EmitExclusiveReadMemoryInline<8, &A32::UserCallbacks::MemoryRead8>(ctx, inst);
+        EmitExclusiveReadMemoryInline<8, &A32::UserCallbacks::MemoryRead>(ctx, inst);
     } else {
-        EmitExclusiveReadMemory<8, &A32::UserCallbacks::MemoryRead8>(ctx, inst);
+        EmitExclusiveReadMemory<8, &A32::UserCallbacks::MemoryRead>(ctx, inst);
     }
 }
 
 void A32EmitX64::EmitA32ExclusiveReadMemory16(A32EmitContext& ctx, IR::Inst* inst) {
     if (conf.fastmem_exclusive_access) {
-        EmitExclusiveReadMemoryInline<16, &A32::UserCallbacks::MemoryRead16>(ctx, inst);
+        EmitExclusiveReadMemoryInline<16, &A32::UserCallbacks::MemoryRead>(ctx, inst);
     } else {
-        EmitExclusiveReadMemory<16, &A32::UserCallbacks::MemoryRead16>(ctx, inst);
+        EmitExclusiveReadMemory<16, &A32::UserCallbacks::MemoryRead>(ctx, inst);
     }
 }
 
 void A32EmitX64::EmitA32ExclusiveReadMemory32(A32EmitContext& ctx, IR::Inst* inst) {
     if (conf.fastmem_exclusive_access) {
-        EmitExclusiveReadMemoryInline<32, &A32::UserCallbacks::MemoryRead32>(ctx, inst);
+        EmitExclusiveReadMemoryInline<32, &A32::UserCallbacks::MemoryRead>(ctx, inst);
     } else {
-        EmitExclusiveReadMemory<32, &A32::UserCallbacks::MemoryRead32>(ctx, inst);
+        EmitExclusiveReadMemory<32, &A32::UserCallbacks::MemoryRead>(ctx, inst);
     }
 }
 
 void A32EmitX64::EmitA32ExclusiveReadMemory64(A32EmitContext& ctx, IR::Inst* inst) {
     if (conf.fastmem_exclusive_access) {
-        EmitExclusiveReadMemoryInline<64, &A32::UserCallbacks::MemoryRead64>(ctx, inst);
+        EmitExclusiveReadMemoryInline<64, &A32::UserCallbacks::MemoryRead>(ctx, inst);
     } else {
-        EmitExclusiveReadMemory<64, &A32::UserCallbacks::MemoryRead64>(ctx, inst);
+        EmitExclusiveReadMemory<64, &A32::UserCallbacks::MemoryRead>(ctx, inst);
     }
 }
 

src/dynarmic/src/dynarmic/backend/x64/a64_emit_x64_memory.cpp

@@ -11,8 +11,7 @@
 #include <tuple>
 #include <utility>
 
-#include <fmt/format.h>
-#include <fmt/ostream.h>
+#include "common/logging.h"
 #include "dynarmic/mcl/integer_of_size.hpp"
 #include "dynarmic/backend/x64/xbyak.h"
 
@@ -115,16 +114,16 @@ void A64EmitX64::GenMemory128Accessors() {
 void A64EmitX64::GenFastmemFallbacks() {
     const std::initializer_list<int> idxes{0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15};
     const std::array<std::pair<size_t, ArgCallback>, 4> read_callbacks{{
-        {8, Devirtualize<&A64::UserCallbacks::MemoryRead8>(conf.callbacks)},
-        {16, Devirtualize<&A64::UserCallbacks::MemoryRead16>(conf.callbacks)},
-        {32, Devirtualize<&A64::UserCallbacks::MemoryRead32>(conf.callbacks)},
-        {64, Devirtualize<&A64::UserCallbacks::MemoryRead64>(conf.callbacks)},
+        {8, Devirtualize<&A64::UserCallbacks::MemoryRead>(conf.callbacks)},
+        {16, Devirtualize<&A64::UserCallbacks::MemoryRead>(conf.callbacks)},
+        {32, Devirtualize<&A64::UserCallbacks::MemoryRead>(conf.callbacks)},
+        {64, Devirtualize<&A64::UserCallbacks::MemoryRead>(conf.callbacks)},
     }};
     const std::array<std::pair<size_t, ArgCallback>, 4> write_callbacks{{
-        {8, Devirtualize<&A64::UserCallbacks::MemoryWrite8>(conf.callbacks)},
-        {16, Devirtualize<&A64::UserCallbacks::MemoryWrite16>(conf.callbacks)},
-        {32, Devirtualize<&A64::UserCallbacks::MemoryWrite32>(conf.callbacks)},
-        {64, Devirtualize<&A64::UserCallbacks::MemoryWrite64>(conf.callbacks)},
+        {8, Devirtualize<&A64::UserCallbacks::MemoryWrite>(conf.callbacks)},
+        {16, Devirtualize<&A64::UserCallbacks::MemoryWrite>(conf.callbacks)},
+        {32, Devirtualize<&A64::UserCallbacks::MemoryWrite>(conf.callbacks)},
+        {64, Devirtualize<&A64::UserCallbacks::MemoryWrite>(conf.callbacks)},
     }};
     const std::array<std::pair<size_t, ArgCallback>, 4> exclusive_write_callbacks{{
         {8, Devirtualize<&A64::UserCallbacks::MemoryWriteExclusive8>(conf.callbacks)},
@@ -204,12 +203,12 @@ void A64EmitX64::GenFastmemFallbacks() {
                     code.align();
                     read_fallbacks[std::make_tuple(ordered, bitsize, vaddr_idx, value_idx)] = code.getCurr<void (*)()>();
                     ABI_PushCallerSaveRegistersAndAdjustStackExcept(code, HostLocRegIdx(value_idx));
+                    // params = { this, vaddr, size }
                     if (vaddr_idx != code.ABI_PARAM2.getIdx()) {
                         code.mov(code.ABI_PARAM2, Xbyak::Reg64{vaddr_idx});
                     }
-                    if (ordered) {
-                        code.mfence();
-                    }
+                    code.mov(code.ABI_PARAM3, bitsize / CHAR_BIT);
+                    if (ordered) code.mfence();
                     callback.EmitCall(code);
                     if (value_idx != code.ABI_RETURN.getIdx()) {
                         code.mov(Xbyak::Reg64{value_idx}, code.ABI_RETURN);
@@ -224,6 +223,7 @@ void A64EmitX64::GenFastmemFallbacks() {
                     code.align();
                     write_fallbacks[std::make_tuple(ordered, bitsize, vaddr_idx, value_idx)] = code.getCurr<void (*)()>();
                     ABI_PushCallerSaveRegistersAndAdjustStack(code);
+                    // params = { this, vaddr, value, size }
                     if (vaddr_idx == code.ABI_PARAM3.getIdx() && value_idx == code.ABI_PARAM2.getIdx()) {
                         code.xchg(code.ABI_PARAM2, code.ABI_PARAM3);
                     } else if (vaddr_idx == code.ABI_PARAM3.getIdx()) {
@@ -240,10 +240,9 @@ void A64EmitX64::GenFastmemFallbacks() {
                         }
                     }
                     code.ZeroExtendFrom(bitsize, code.ABI_PARAM3);
+                    code.mov(code.ABI_PARAM4, bitsize / CHAR_BIT);
                     callback.EmitCall(code);
-                    if (ordered) {
-                        code.mfence();
-                    }
+                    if (ordered) code.mfence();
                     ABI_PopCallerSaveRegistersAndAdjustStack(code);
                     code.ret();
                     PerfMapRegister(write_fallbacks[std::make_tuple(ordered, bitsize, vaddr_idx, value_idx)], code.getCurr(), fmt::format("a64_write_fallback_{}", bitsize));
@@ -286,19 +285,19 @@ void A64EmitX64::GenFastmemFallbacks() {
 #undef Axx
 
 void A64EmitX64::EmitA64ReadMemory8(A64EmitContext& ctx, IR::Inst* inst) {
-    EmitMemoryRead<8, &A64::UserCallbacks::MemoryRead8>(ctx, inst);
+    EmitMemoryRead<8, &A64::UserCallbacks::MemoryRead>(ctx, inst);
 }
 
 void A64EmitX64::EmitA64ReadMemory16(A64EmitContext& ctx, IR::Inst* inst) {
-    EmitMemoryRead<16, &A64::UserCallbacks::MemoryRead16>(ctx, inst);
+    EmitMemoryRead<16, &A64::UserCallbacks::MemoryRead>(ctx, inst);
 }
 
 void A64EmitX64::EmitA64ReadMemory32(A64EmitContext& ctx, IR::Inst* inst) {
-    EmitMemoryRead<32, &A64::UserCallbacks::MemoryRead32>(ctx, inst);
+    EmitMemoryRead<32, &A64::UserCallbacks::MemoryRead>(ctx, inst);
 }
 
 void A64EmitX64::EmitA64ReadMemory64(A64EmitContext& ctx, IR::Inst* inst) {
-    EmitMemoryRead<64, &A64::UserCallbacks::MemoryRead64>(ctx, inst);
+    EmitMemoryRead<64, &A64::UserCallbacks::MemoryRead>(ctx, inst);
 }
 
 void A64EmitX64::EmitA64ReadMemory128(A64EmitContext& ctx, IR::Inst* inst) {
@@ -306,23 +305,23 @@ void A64EmitX64::EmitA64ReadMemory128(A64EmitContext& ctx, IR::Inst* inst) {
 }
 
 void A64EmitX64::EmitA64WriteMemory8(A64EmitContext& ctx, IR::Inst* inst) {
-    EmitMemoryWrite<8, &A64::UserCallbacks::MemoryWrite8>(ctx, inst);
+    EmitMemoryWrite<8, &A64::UserCallbacks::MemoryWrite>(ctx, inst);
 }
 
 void A64EmitX64::EmitA64WriteMemory16(A64EmitContext& ctx, IR::Inst* inst) {
-    EmitMemoryWrite<16, &A64::UserCallbacks::MemoryWrite16>(ctx, inst);
+    EmitMemoryWrite<16, &A64::UserCallbacks::MemoryWrite>(ctx, inst);
 }
 
 void A64EmitX64::EmitA64WriteMemory32(A64EmitContext& ctx, IR::Inst* inst) {
-    EmitMemoryWrite<32, &A64::UserCallbacks::MemoryWrite32>(ctx, inst);
+    EmitMemoryWrite<32, &A64::UserCallbacks::MemoryWrite>(ctx, inst);
 }
 
 void A64EmitX64::EmitA64WriteMemory64(A64EmitContext& ctx, IR::Inst* inst) {
-    EmitMemoryWrite<64, &A64::UserCallbacks::MemoryWrite64>(ctx, inst);
+    EmitMemoryWrite<64, &A64::UserCallbacks::MemoryWrite>(ctx, inst);
 }
 
 void A64EmitX64::EmitA64WriteMemory128(A64EmitContext& ctx, IR::Inst* inst) {
-    EmitMemoryWrite<128, &A64::UserCallbacks::MemoryWrite64>(ctx, inst);
+    EmitMemoryWrite<128, &A64::UserCallbacks::MemoryWrite>(ctx, inst);
 }
 
 void A64EmitX64::EmitA64ClearExclusive(A64EmitContext&, IR::Inst*) {
@@ -331,33 +330,33 @@ void A64EmitX64::EmitA64ClearExclusive(A64EmitContext&, IR::Inst*) {
 
 void A64EmitX64::EmitA64ExclusiveReadMemory8(A64EmitContext& ctx, IR::Inst* inst) {
     if (conf.fastmem_exclusive_access) {
-        EmitExclusiveReadMemoryInline<8, &A64::UserCallbacks::MemoryRead8>(ctx, inst);
+        EmitExclusiveReadMemoryInline<8, &A64::UserCallbacks::MemoryRead>(ctx, inst);
     } else {
-        EmitExclusiveReadMemory<8, &A64::UserCallbacks::MemoryRead8>(ctx, inst);
+        EmitExclusiveReadMemory<8, &A64::UserCallbacks::MemoryRead>(ctx, inst);
     }
 }
 
 void A64EmitX64::EmitA64ExclusiveReadMemory16(A64EmitContext& ctx, IR::Inst* inst) {
     if (conf.fastmem_exclusive_access) {
-        EmitExclusiveReadMemoryInline<16, &A64::UserCallbacks::MemoryRead16>(ctx, inst);
+        EmitExclusiveReadMemoryInline<16, &A64::UserCallbacks::MemoryRead>(ctx, inst);
     } else {
-        EmitExclusiveReadMemory<16, &A64::UserCallbacks::MemoryRead16>(ctx, inst);
+        EmitExclusiveReadMemory<16, &A64::UserCallbacks::MemoryRead>(ctx, inst);
     }
 }
 
 void A64EmitX64::EmitA64ExclusiveReadMemory32(A64EmitContext& ctx, IR::Inst* inst) {
     if (conf.fastmem_exclusive_access) {
-        EmitExclusiveReadMemoryInline<32, &A64::UserCallbacks::MemoryRead32>(ctx, inst);
+        EmitExclusiveReadMemoryInline<32, &A64::UserCallbacks::MemoryRead>(ctx, inst);
     } else {
-        EmitExclusiveReadMemory<32, &A64::UserCallbacks::MemoryRead32>(ctx, inst);
+        EmitExclusiveReadMemory<32, &A64::UserCallbacks::MemoryRead>(ctx, inst);
     }
 }
 
 void A64EmitX64::EmitA64ExclusiveReadMemory64(A64EmitContext& ctx, IR::Inst* inst) {
     if (conf.fastmem_exclusive_access) {
-        EmitExclusiveReadMemoryInline<64, &A64::UserCallbacks::MemoryRead64>(ctx, inst);
+        EmitExclusiveReadMemoryInline<64, &A64::UserCallbacks::MemoryRead>(ctx, inst);
     } else {
-        EmitExclusiveReadMemory<64, &A64::UserCallbacks::MemoryRead64>(ctx, inst);
+        EmitExclusiveReadMemory<64, &A64::UserCallbacks::MemoryRead>(ctx, inst);
     }
 }
 

src/dynarmic/src/dynarmic/backend/x64/emit_x64_memory.cpp.inc

@@ -1,3 +1,6 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
 /* This file is part of the dynarmic project.
  * Copyright (c) 2022 MerryMage
  * SPDX-License-Identifier: 0BSD
@@ -11,10 +14,6 @@
 #define AxxJitState CONCATENATE_TOKENS(Axx, JitState)
 #define AxxUserConfig Axx::UserConfig
 
-namespace {
-using Vector = std::array<u64, 2>;
-}
-
 std::optional<AxxEmitX64::DoNotFastmemMarker> AxxEmitX64::ShouldFastmem(AxxEmitContext& ctx, IR::Inst* inst) const {
     if (!conf.fastmem_pointer || !exception_handler.SupportsFastmem()) {
         return std::nullopt;
@@ -28,22 +27,19 @@ std::optional<AxxEmitX64::DoNotFastmemMarker> AxxEmitX64::ShouldFastmem(AxxEmitC
 }
 
 FakeCall AxxEmitX64::FastmemCallback(u64 rip_) {
-    const auto iter = fastmem_patch_info.find(rip_);
-    if (iter != fastmem_patch_info.end()) {
+    if (auto const it = fastmem_patch_info.find(rip_); it != fastmem_patch_info.end()) {
         FakeCall result{
-            .call_rip = iter->second.callback,
-            .ret_rip = iter->second.resume_rip,
+            .call_rip = it->second.callback,
+            .ret_rip = it->second.resume_rip,
         };
-        if (iter->second.recompile) {
-            const auto marker = iter->second.marker;
+        if (it->second.recompile) {
+            const auto marker = it->second.marker;
             do_not_fastmem.insert(marker);
             InvalidateBasicBlocks({std::get<0>(marker)});
         }
         return result;
     }
-    fmt::print("dynarmic: Segfault happened within JITted code at rip = {:016x}\n"
-        "Segfault wasn't at a fastmem patch location!\n", rip_);
-    UNREACHABLE(); //("iter != fastmem_patch_info.end()");
+    UNREACHABLE_MSG("SIGSEGV @ JIT, rip={:#016x}", rip_); //("iter != fastmem_patch_info.end()");
 }
 
 template<std::size_t bitsize, auto callback>
@@ -56,16 +52,13 @@ void AxxEmitX64::EmitMemoryRead(AxxEmitContext& ctx, IR::Inst* inst) {
         // Neither fastmem nor page table: Use callbacks
         if constexpr (bitsize == 128) {
             ctx.reg_alloc.HostCall(code, nullptr, {}, args[1]);
-            if (ordered) {
-                code.mfence();
-            }
+            if (ordered) code.mfence();
             code.CallFunction(memory_read_128);
             ctx.reg_alloc.DefineValue(code, inst, xmm1);
         } else {
             ctx.reg_alloc.HostCall(code, inst, {}, args[1]);
-            if (ordered) {
-                code.mfence();
-            }
+            code.mov(code.ABI_PARAM3.cvt32(), bitsize / CHAR_BIT);
+            if (ordered) code.mfence();
             Devirtualize<callback>(conf.callbacks).EmitCall(code);
             code.ZeroExtendFrom(bitsize, code.ABI_RETURN);
         }
@@ -83,22 +76,17 @@ void AxxEmitX64::EmitMemoryRead(AxxEmitContext& ctx, IR::Inst* inst) {
 
     const Xbyak::Reg64 vaddr = ctx.reg_alloc.UseGpr(code, args[1]);
     const int value_idx = bitsize == 128 ? ctx.reg_alloc.ScratchXmm(code).getIdx() : ctx.reg_alloc.ScratchGpr(code).getIdx();
-
     const auto wrapped_fn = read_fallbacks[std::make_tuple(ordered, bitsize, vaddr.getIdx(), value_idx)];
 
     SharedLabel abort = ctx.GenSharedLabel(), end = ctx.GenSharedLabel();
-
     if (fastmem_marker) {
         // Use fastmem
         bool require_abort_handling = false;
         const auto src_ptr = EmitFastmemVAddr(code, ctx, *abort, vaddr, require_abort_handling);
-
         const auto location = EmitReadMemoryMov<bitsize>(code, value_idx, src_ptr, ordered);
-
         ctx.deferred_emits.emplace_back([=, this, &ctx] {
             code.L(*abort);
             code.call(wrapped_fn);
-
             fastmem_patch_info.emplace(
                 std::bit_cast<u64>(location),
                 FastmemPatchInfo{
@@ -107,25 +95,23 @@ void AxxEmitX64::EmitMemoryRead(AxxEmitContext& ctx, IR::Inst* inst) {
                     *fastmem_marker,
                     conf.recompile_on_fastmem_failure,
                 });
-
             EmitCheckMemoryAbort(ctx, inst, end);
             code.jmp(*end, code.T_NEAR);
         });
-    } else {
+    } else if (conf.page_table) {
         // Use page table
-        ASSERT(conf.page_table);
         const auto src_ptr = EmitVAddrLookup(code, ctx, bitsize, *abort, vaddr);
         EmitReadMemoryMov<bitsize>(code, value_idx, src_ptr, ordered);
-
         ctx.deferred_emits.emplace_back([=, this, &ctx] {
             code.L(*abort);
             code.call(wrapped_fn);
             EmitCheckMemoryAbort(ctx, inst, end);
             code.jmp(*end, code.T_NEAR);
         });
+    } else {
+        UNREACHABLE();
     }
     code.L(*end);
-
     if constexpr (bitsize == 128) {
         ctx.reg_alloc.DefineValue(code, inst, Xbyak::Xmm{value_idx});
     } else {
@@ -148,12 +134,12 @@ void AxxEmitX64::EmitMemoryWrite(AxxEmitContext& ctx, IR::Inst* inst) {
             ctx.reg_alloc.HostCall(code, nullptr);
             code.CallFunction(memory_write_128);
         } else {
+            // { this, vaddr, value, size }
             ctx.reg_alloc.HostCall(code, nullptr, {}, args[1], args[2]);
+            code.mov(code.ABI_PARAM4.cvt32(), bitsize / CHAR_BIT);
             Devirtualize<callback>(conf.callbacks).EmitCall(code);
         }
-        if (ordered) {
-            code.mfence();
-        }
+        if (ordered) code.mfence();
         EmitCheckMemoryAbort(ctx, inst);
         return;
     }
@@ -168,18 +154,16 @@ void AxxEmitX64::EmitMemoryWrite(AxxEmitContext& ctx, IR::Inst* inst) {
 
     const Xbyak::Reg64 vaddr = ctx.reg_alloc.UseGpr(code, args[1]);
     const int value_idx = bitsize == 128
-                            ? ctx.reg_alloc.UseXmm(code, args[2]).getIdx()
-                            : (ordered ? ctx.reg_alloc.UseScratchGpr(code, args[2]).getIdx() : ctx.reg_alloc.UseGpr(code, args[2]).getIdx());
+        ? ctx.reg_alloc.UseXmm(code, args[2]).getIdx()
+        : (ordered ? ctx.reg_alloc.UseScratchGpr(code, args[2]).getIdx() : ctx.reg_alloc.UseGpr(code, args[2]).getIdx());
 
     const auto wrapped_fn = write_fallbacks[std::make_tuple(ordered, bitsize, vaddr.getIdx(), value_idx)];
 
     SharedLabel abort = ctx.GenSharedLabel(), end = ctx.GenSharedLabel();
-
     if (fastmem_marker) {
         // Use fastmem
         bool require_abort_handling = false;
         const auto dest_ptr = EmitFastmemVAddr(code, ctx, *abort, vaddr, require_abort_handling);
-
         const auto location = EmitWriteMemoryMov<bitsize>(code, dest_ptr, value_idx, ordered);
 
         ctx.deferred_emits.emplace_back([=, this, &ctx] {
@@ -198,9 +182,8 @@ void AxxEmitX64::EmitMemoryWrite(AxxEmitContext& ctx, IR::Inst* inst) {
             EmitCheckMemoryAbort(ctx, inst, end);
             code.jmp(*end, code.T_NEAR);
         });
-    } else {
+    } else if (conf.page_table) {
         // Use page table
-        ASSERT(conf.page_table);
         const auto dest_ptr = EmitVAddrLookup(code, ctx, bitsize, *abort, vaddr);
         EmitWriteMemoryMov<bitsize>(code, dest_ptr, value_idx, ordered);
 
@@ -210,6 +193,8 @@ void AxxEmitX64::EmitMemoryWrite(AxxEmitContext& ctx, IR::Inst* inst) {
             EmitCheckMemoryAbort(ctx, inst, end);
             code.jmp(*end, code.T_NEAR);
         });
+    } else {
+        UNREACHABLE();
     }
     code.L(*end);
 }
@@ -230,12 +215,11 @@ void AxxEmitX64::EmitExclusiveReadMemory(AxxEmitContext& ctx, IR::Inst* inst) {
         if (ordered) {
             code.mfence();
         }
-        code.CallLambda(
-            [](AxxUserConfig& conf, Axx::VAddr vaddr) -> T {
-                return conf.global_monitor->ReadAndMark<T>(conf.processor_id, vaddr, [&]() -> T {
-                    return (conf.callbacks->*callback)(vaddr);
-                });
+        code.CallLambda([](AxxUserConfig& conf, Axx::VAddr vaddr) -> T {
+            return conf.global_monitor->ReadAndMark<T>(conf.processor_id, vaddr, [&]() -> T {
+                return (conf.callbacks->*callback)(vaddr, bitsize / CHAR_BIT);
             });
+        });
         code.ZeroExtendFrom(bitsize, code.ABI_RETURN);
     } else {
         const Xbyak::Xmm result = ctx.reg_alloc.ScratchXmm(code);
@@ -250,12 +234,11 @@ void AxxEmitX64::EmitExclusiveReadMemory(AxxEmitContext& ctx, IR::Inst* inst) {
         if (ordered) {
             code.mfence();
         }
-        code.CallLambda(
-            [](AxxUserConfig& conf, Axx::VAddr vaddr, Vector& ret) {
-                ret = conf.global_monitor->ReadAndMark<Vector>(conf.processor_id, vaddr, [&]() -> Vector {
-                    return (conf.callbacks->*callback)(vaddr);
-                });
+        code.CallLambda([](AxxUserConfig& conf, Axx::VAddr vaddr, u128& ret) {
+            ret = conf.global_monitor->ReadAndMark<u128>(conf.processor_id, vaddr, [&]() -> u128 {
+                return (conf.callbacks->*callback)(vaddr);
             });
+        });
         code.movups(result, xword[rsp + ABI_SHADOW_SPACE]);
         ctx.reg_alloc.ReleaseStackSpace(code, 16 + ABI_SHADOW_SPACE);
 
@@ -303,8 +286,8 @@ void AxxEmitX64::EmitExclusiveWriteMemory(AxxEmitContext& ctx, IR::Inst* inst) {
         ctx.reg_alloc.AllocStackSpace(code, 16 + ABI_SHADOW_SPACE);
         code.lea(code.ABI_PARAM3, ptr[rsp + ABI_SHADOW_SPACE]);
         code.movaps(xword[code.ABI_PARAM3], xmm1);
-        code.CallLambda([](AxxUserConfig& conf, Axx::VAddr vaddr, Vector& value) -> u32 {
-            return conf.global_monitor->DoExclusiveOperation<Vector>(conf.processor_id, vaddr, [&](Vector expected) -> bool {
+        code.CallLambda([](AxxUserConfig& conf, Axx::VAddr vaddr, u128& value) -> u32 {
+            return conf.global_monitor->DoExclusiveOperation<u128>(conf.processor_id, vaddr, [&](u128 expected) -> bool {
                 return (conf.callbacks->*callback)(vaddr, value, expected);
             }) ? 0 : 1;
         });

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

@@ -135,10 +135,16 @@ template<>
     if (unused_top_bits == 0) {
         code.mov(tmp, vaddr);
         code.shr(tmp, int(page_table_const_bits));
+        if (ctx.conf.page_table_log2_stride > 3) {
+            code.shl(tmp, int(ctx.conf.page_table_log2_stride));
+            code.mov(page, qword[r14 + tmp.cvt64()]);
+        } else {
+            code.mov(page, qword[r14 + tmp.cvt64() * int(1 << ctx.conf.page_table_log2_stride)]);
+        }
     } else if (ctx.conf.silently_mirror_page_table) {
         if (valid_page_index_bits >= 32) {
             if (code.HasHostFeature(HostFeature::BMI2)) {
-                const Xbyak::Reg64 bit_count = ctx.reg_alloc.ScratchGpr(code);
+                auto const bit_count = ctx.reg_alloc.ScratchGpr(code);
                 code.mov(bit_count, unused_top_bits);
                 code.bzhi(tmp, vaddr, bit_count);
                 code.shr(tmp, int(page_table_const_bits));
@@ -153,19 +159,31 @@ template<>
             code.shr(tmp, int(page_table_const_bits));
             code.and_(tmp, u32((1 << valid_page_index_bits) - 1));
         }
+        if (ctx.conf.page_table_log2_stride > 3) {
+            code.shl(tmp, int(ctx.conf.page_table_log2_stride));
+            code.mov(page, qword[r14 + tmp.cvt64()]);
+        } else {
+            code.mov(page, qword[r14 + tmp.cvt64() * int(1 << ctx.conf.page_table_log2_stride)]);
+        }
     } else {
+        // Common VA sizes: 39 - 12 => 27, 42 - 12 => 30
+        // Check if bits outside of VA space are non-zero
         ASSERT(valid_page_index_bits < 32);
         code.mov(tmp, vaddr);
+        if (ctx.conf.check_halt_on_memory_access) {
+            auto const tmp2 = ctx.reg_alloc.ScratchGpr(code);
+            code.mov(tmp2, u64(-(1ull << valid_page_index_bits) << page_table_const_bits));
+            code.test(tmp, tmp2);
+            code.jnz(abort, code.T_NEAR);
+            ctx.reg_alloc.Release(tmp2);
+        }
         code.shr(tmp, int(page_table_const_bits));
-        code.test(tmp, u32(-(1 << valid_page_index_bits)));
-        code.jnz(abort, code.T_NEAR);
-    }
-
-    if (ctx.conf.page_table_log2_stride > 3) {
-        code.shl(tmp, int(ctx.conf.page_table_log2_stride));
-        code.mov(page, qword[r14 + tmp.cvt64()]);
-    } else {
-        code.mov(page, qword[r14 + tmp.cvt64() * int(1 << ctx.conf.page_table_log2_stride)]);
+        if (ctx.conf.page_table_log2_stride > 3) {
+            code.shl(tmp, int(ctx.conf.page_table_log2_stride));
+            code.mov(page, qword[r14 + tmp.cvt64()]);
+        } else {
+            code.mov(page, qword[r14 + tmp.cvt64() * int(1 << ctx.conf.page_table_log2_stride)]);
+        }
     }
 
     // check for marked bit, use as unmapped if marked
@@ -193,7 +211,7 @@ template<>
         return page + vaddr;
     }
     code.mov(tmp, vaddr);
-    code.and_(tmp, static_cast<u32>(page_table_const_mask));
+    code.and_(tmp, u32(page_table_const_mask));
     return page + tmp;
 }
 

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

@@ -66,7 +66,9 @@ struct UserCallbacks : public TranslateCallbacks {
 
     // All reads through this callback are 4-byte aligned.
     // Memory must be interpreted as little endian.
-    std::optional<std::uint32_t> MemoryReadCode(VAddr vaddr) override { return MemoryRead32(vaddr); }
+    std::optional<std::uint32_t> MemoryReadCode(VAddr vaddr) override {
+        return std::uint32_t(MemoryRead(vaddr, sizeof(std::uint32_t)));
+    }
 
     // This function is called before the instruction at pc is read.
     // IR code can be emitted by the callee prior to instruction handling.
@@ -80,16 +82,10 @@ struct UserCallbacks : public TranslateCallbacks {
 
     // Reads through these callbacks may not be aligned.
     // Memory must be interpreted as if ENDIANSTATE == 0, endianness will be corrected by the JIT.
-    virtual std::uint8_t MemoryRead8(VAddr vaddr) = 0;
-    virtual std::uint16_t MemoryRead16(VAddr vaddr) = 0;
-    virtual std::uint32_t MemoryRead32(VAddr vaddr) = 0;
-    virtual std::uint64_t MemoryRead64(VAddr vaddr) = 0;
+    virtual std::uint64_t MemoryRead(VAddr vaddr, std::size_t size) = 0;
 
     // Writes through these callbacks may not be aligned.
-    virtual void MemoryWrite8(VAddr vaddr, std::uint8_t value) = 0;
-    virtual void MemoryWrite16(VAddr vaddr, std::uint16_t value) = 0;
-    virtual void MemoryWrite32(VAddr vaddr, std::uint32_t value) = 0;
-    virtual void MemoryWrite64(VAddr vaddr, std::uint64_t value) = 0;
+    virtual void MemoryWrite(VAddr vaddr, std::uint64_t value, std::size_t size) = 0;
 
     // Writes through these callbacks may not be aligned.
     virtual bool MemoryWriteExclusive8(VAddr /*vaddr*/, std::uint8_t /*value*/, std::uint8_t /*expected*/) { return false; }

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

@@ -89,20 +89,16 @@ struct UserCallbacks {
 
     // All reads through this callback are 4-byte aligned.
     // Memory must be interpreted as little endian.
-    virtual std::optional<std::uint32_t> MemoryReadCode(VAddr vaddr) { return MemoryRead32(vaddr); }
+    virtual std::optional<std::uint32_t> MemoryReadCode(VAddr vaddr) {
+        return std::uint32_t(MemoryRead(vaddr, sizeof(std::uint32_t)));
+    }
 
     // Reads through these callbacks may not be aligned.
-    virtual std::uint8_t MemoryRead8(VAddr vaddr) = 0;
-    virtual std::uint16_t MemoryRead16(VAddr vaddr) = 0;
-    virtual std::uint32_t MemoryRead32(VAddr vaddr) = 0;
-    virtual std::uint64_t MemoryRead64(VAddr vaddr) = 0;
+    virtual std::uint64_t MemoryRead(VAddr vaddr, std::size_t size) = 0;
     virtual Vector MemoryRead128(VAddr vaddr) = 0;
 
     // Writes through these callbacks may not be aligned.
-    virtual void MemoryWrite8(VAddr vaddr, std::uint8_t value) = 0;
-    virtual void MemoryWrite16(VAddr vaddr, std::uint16_t value) = 0;
-    virtual void MemoryWrite32(VAddr vaddr, std::uint32_t value) = 0;
-    virtual void MemoryWrite64(VAddr vaddr, std::uint64_t value) = 0;
+    virtual void MemoryWrite(VAddr vaddr, std::uint64_t value, std::size_t size) = 0;
     virtual void MemoryWrite128(VAddr vaddr, Vector value) = 0;
 
     // Writes through these callbacks may not be aligned.

src/dynarmic/src/dynarmic/ir/opt_passes.cpp

@@ -40,7 +40,7 @@ static void ConstantMemoryReads(IR::Block& block, A32::UserCallbacks* cb) {
             if (inst.AreAllArgsImmediates()) {
                 const u32 vaddr = inst.GetArg(1).GetU32();
                 if (cb->IsReadOnlyMemory(vaddr)) {
-                    const u8 value_from_memory = cb->MemoryRead8(vaddr);
+                    const u8 value_from_memory = u8(cb->MemoryRead(vaddr, sizeof(u8)));
                     inst.ReplaceUsesWith(IR::Value{value_from_memory});
                 }
             }
@@ -51,7 +51,7 @@ static void ConstantMemoryReads(IR::Block& block, A32::UserCallbacks* cb) {
             if (inst.AreAllArgsImmediates()) {
                 const u32 vaddr = inst.GetArg(1).GetU32();
                 if (cb->IsReadOnlyMemory(vaddr)) {
-                    const u16 value_from_memory = cb->MemoryRead16(vaddr);
+                    const u16 value_from_memory = u16(cb->MemoryRead(vaddr, sizeof(u16)));
                     inst.ReplaceUsesWith(IR::Value{value_from_memory});
                 }
             }
@@ -62,7 +62,7 @@ static void ConstantMemoryReads(IR::Block& block, A32::UserCallbacks* cb) {
             if (inst.AreAllArgsImmediates()) {
                 const u32 vaddr = inst.GetArg(1).GetU32();
                 if (cb->IsReadOnlyMemory(vaddr)) {
-                    const u32 value_from_memory = cb->MemoryRead32(vaddr);
+                    const u32 value_from_memory = u32(cb->MemoryRead(vaddr, sizeof(u32)));
                     inst.ReplaceUsesWith(IR::Value{value_from_memory});
                 }
             }
@@ -73,7 +73,7 @@ static void ConstantMemoryReads(IR::Block& block, A32::UserCallbacks* cb) {
             if (inst.AreAllArgsImmediates()) {
                 const u32 vaddr = inst.GetArg(1).GetU32();
                 if (cb->IsReadOnlyMemory(vaddr)) {
-                    const u64 value_from_memory = cb->MemoryRead64(vaddr);
+                    const u64 value_from_memory = u64(cb->MemoryRead(vaddr, sizeof(u64)));
                     inst.ReplaceUsesWith(IR::Value{value_from_memory});
                 }
             }

src/dynarmic/tests/A32/fuzz_thumb.cpp

@@ -503,7 +503,7 @@ TEST_CASE("Fuzz Thumb32 instructions set", "[JitX64][Thumb][Thumb32]") {
     }
 }
 
-TEST_CASE("Verify fix for off by one error in MemoryRead32 worked", "[Thumb][Thumb16]") {
+TEST_CASE("Verify fix for off by one error in MemoryRead<32> worked", "[Thumb][Thumb16]") {
     ThumbTestEnv test_env;
 
     // Prepare test subjects

src/dynarmic/tests/A32/test_arm_instructions.cpp

@@ -553,9 +553,9 @@ TEST_CASE("arm: Memory access (fastmem)", "[arm][A32]") {
     memset(backing_memory, 0, memory_size);
     memcpy(backing_memory + 0x100, "Lorem ipsum dolor sit amet, consectetur adipiscing elit.", 57);
 
-    env.MemoryWrite32(0, 0xE5904000);  // LDR R4, [R0]
-    env.MemoryWrite32(4, 0xE5814000);  // STR R4, [R1]
-    env.MemoryWrite32(8, 0xEAFFFFFE);  // B .
+    env.MemoryWrite(0, 0xE5904000, sizeof(u32));  // LDR R4, [R0]
+    env.MemoryWrite(4, 0xE5814000, sizeof(u32));  // STR R4, [R1]
+    env.MemoryWrite(8, 0xEAFFFFFE, sizeof(u32));  // B .
     jit.Regs()[0] = 0x100;
     jit.Regs()[1] = 0x1F0;
     jit.Regs()[15] = 0;       // PC = 0

src/dynarmic/tests/A32/testenv.h

@@ -16,6 +16,7 @@
 
 #include "common/assert.h"
 #include "common/common_types.h"
+#include "dynarmic/frontend/A32/translate/translate_callbacks.h"
 #include "dynarmic/interface/A32/a32.h"
 
 template<typename InstructionType_, u32 infinite_loop_u32>
@@ -59,42 +60,51 @@ public:
         return infinite_loop_u32;  // B .
     }
 
-    std::uint8_t MemoryRead8(u32 vaddr) override {
-        if (IsInCodeMem(vaddr)) {
-            return reinterpret_cast<u8*>(code_mem.data())[vaddr];
+    u64 MemoryRead(u32 vaddr, size_t size) override {
+        switch (size) {
+        case sizeof(u64):
+            return MemoryRead(vaddr, sizeof(u32))
+                | MemoryRead(vaddr + sizeof(u32), sizeof(u32)) << 32;
+        case sizeof(u32):
+            return MemoryRead(vaddr, sizeof(u16))
+                | MemoryRead(vaddr + sizeof(u16), sizeof(u16)) << 16;
+        case sizeof(u16):
+            return MemoryRead(vaddr, sizeof(u8))
+                | MemoryRead(vaddr + sizeof(u8), sizeof(u8)) << 8;
+        case sizeof(u8): {
+            if (IsInCodeMem(vaddr))
+                return reinterpret_cast<u8*>(code_mem.data())[vaddr];
+            if (auto iter = modified_memory.find(vaddr); iter != modified_memory.end())
+                return iter->second;
+            return u8(vaddr);
         }
-        if (auto iter = modified_memory.find(vaddr); iter != modified_memory.end()) {
-            return iter->second;
+        default:
+            std::abort();
         }
-        return static_cast<u8>(vaddr);
-    }
-    std::uint16_t MemoryRead16(u32 vaddr) override {
-        return u16(MemoryRead8(vaddr)) | u16(MemoryRead8(vaddr + 1)) << 8;
-    }
-    std::uint32_t MemoryRead32(u32 vaddr) override {
-        return u32(MemoryRead16(vaddr)) | u32(MemoryRead16(vaddr + 2)) << 16;
-    }
-    std::uint64_t MemoryRead64(u32 vaddr) override {
-        return u64(MemoryRead32(vaddr)) | u64(MemoryRead32(vaddr + 4)) << 32;
     }
 
-    void MemoryWrite8(u32 vaddr, std::uint8_t value) override {
-        if (vaddr < code_mem.size() * sizeof(u32)) {
-            code_mem_modified_by_guest = true;
+    void MemoryWrite(Dynarmic::A32::VAddr vaddr, u64 value, size_t size) override {
+        switch (size) {
+        case sizeof(u64):
+            MemoryWrite(vaddr, u32(value), sizeof(u32));
+            MemoryWrite(vaddr + 4, u32(value >> 32), sizeof(u32));
+            break;
+        case sizeof(u32):
+            MemoryWrite(vaddr, u16(value), sizeof(u16));
+            MemoryWrite(vaddr + 2, u16(value >> 16), sizeof(u16));
+            break;
+        case sizeof(u16):
+            MemoryWrite(vaddr, u8(value), sizeof(u8));
+            MemoryWrite(vaddr + 1, u8(value >> 8), sizeof(u8));
+            break;
+        case sizeof(u8):
+            if (vaddr < code_mem.size() * sizeof(u32))
+                code_mem_modified_by_guest = true;
+            modified_memory[vaddr] = value;
+            break;
+        default:
+            std::abort();
         }
-        modified_memory[vaddr] = value;
-    }
-    void MemoryWrite16(u32 vaddr, std::uint16_t value) override {
-        MemoryWrite8(vaddr, static_cast<u8>(value));
-        MemoryWrite8(vaddr + 1, static_cast<u8>(value >> 8));
-    }
-    void MemoryWrite32(u32 vaddr, std::uint32_t value) override {
-        MemoryWrite16(vaddr, static_cast<u16>(value));
-        MemoryWrite16(vaddr + 2, static_cast<u16>(value >> 16));
-    }
-    void MemoryWrite64(u32 vaddr, std::uint64_t value) override {
-        MemoryWrite32(vaddr, static_cast<u32>(value));
-        MemoryWrite32(vaddr + 4, static_cast<u32>(value >> 32));
     }
 
     void CallSVC(std::uint32_t swi) override {
@@ -143,46 +153,41 @@ public:
         return read<std::uint32_t>(vaddr);
     }
 
-    std::uint8_t MemoryRead8(std::uint32_t vaddr) override {
-        return read<std::uint8_t>(vaddr);
-    }
-    std::uint16_t MemoryRead16(std::uint32_t vaddr) override {
-        return read<std::uint16_t>(vaddr);
-    }
-    std::uint32_t MemoryRead32(std::uint32_t vaddr) override {
-        return read<std::uint32_t>(vaddr);
-    }
-    std::uint64_t MemoryRead64(std::uint32_t vaddr) override {
-        return read<std::uint64_t>(vaddr);
+    u64 MemoryRead(u32 vaddr, size_t size) override {
+        switch (size) {
+        case sizeof(u64): return read<u64>(vaddr);
+        case sizeof(u32): return read<u32>(vaddr);
+        case sizeof(u16): return read<u16>(vaddr);
+        case sizeof(u8): return read<u8>(vaddr);
+        default:
+            std::abort();
+        }
     }
 
-    void MemoryWrite8(std::uint32_t vaddr, std::uint8_t value) override {
-        write(vaddr, value);
-    }
-    void MemoryWrite16(std::uint32_t vaddr, std::uint16_t value) override {
-        write(vaddr, value);
-    }
-    void MemoryWrite32(std::uint32_t vaddr, std::uint32_t value) override {
-        write(vaddr, value);
-    }
-    void MemoryWrite64(std::uint32_t vaddr, std::uint64_t value) override {
-        write(vaddr, value);
+    void MemoryWrite(Dynarmic::A32::VAddr vaddr, std::uint64_t value, size_t size) override {
+        switch (size) {
+        case sizeof(u64): return write<u64>(vaddr, u64(value));
+        case sizeof(u32): return write<u32>(vaddr, u32(value));
+        case sizeof(u16): return write<u16>(vaddr, u16(value));
+        case sizeof(u8): return write<u8>(vaddr, u8(value));
+        default: std::abort();
+        }
     }
 
-    bool MemoryWriteExclusive8(std::uint32_t vaddr, std::uint8_t value, [[maybe_unused]] std::uint8_t expected) override {
-        MemoryWrite8(vaddr, value);
+    bool MemoryWriteExclusive8(Dynarmic::A32::VAddr vaddr, std::uint8_t value, [[maybe_unused]] std::uint8_t expected) override {
+        MemoryWrite(vaddr, value, sizeof(u8));
         return true;
     }
-    bool MemoryWriteExclusive16(std::uint32_t vaddr, std::uint16_t value, [[maybe_unused]] std::uint16_t expected) override {
-        MemoryWrite16(vaddr, value);
+    bool MemoryWriteExclusive16(Dynarmic::A32::VAddr vaddr, std::uint16_t value, [[maybe_unused]] std::uint16_t expected) override {
+        MemoryWrite(vaddr, value, sizeof(u16));
         return true;
     }
-    bool MemoryWriteExclusive32(std::uint32_t vaddr, std::uint32_t value, [[maybe_unused]] std::uint32_t expected) override {
-        MemoryWrite32(vaddr, value);
+    bool MemoryWriteExclusive32(Dynarmic::A32::VAddr vaddr, std::uint32_t value, [[maybe_unused]] std::uint32_t expected) override {
+        MemoryWrite(vaddr, value, sizeof(u32));
         return true;
     }
-    bool MemoryWriteExclusive64(std::uint32_t vaddr, std::uint64_t value, [[maybe_unused]] std::uint64_t expected) override {
-        MemoryWrite64(vaddr, value);
+    bool MemoryWriteExclusive64(Dynarmic::A32::VAddr vaddr, std::uint64_t value, [[maybe_unused]] std::uint64_t expected) override {
+        MemoryWrite(vaddr, value, sizeof(u64));
         return true;
     }
 

src/dynarmic/tests/A64/fibonacci.cpp

@@ -25,48 +25,55 @@ public:
     u64 ticks_left = 0;
     ::Common::unordered_map<u64, u8> memory{};
 
-    u8 MemoryRead8(u64 vaddr) override {
-        return memory[vaddr];
-    }
-
-    u16 MemoryRead16(u64 vaddr) override {
-        return u16(MemoryRead8(vaddr)) | u16(MemoryRead8(vaddr + 1)) << 8;
-    }
-
-    u32 MemoryRead32(u64 vaddr) override {
-        return u32(MemoryRead16(vaddr)) | u32(MemoryRead16(vaddr + 2)) << 16;
-    }
-
-    u64 MemoryRead64(u64 vaddr) override {
-        return u64(MemoryRead32(vaddr)) | u64(MemoryRead32(vaddr + 4)) << 32;
+    u64 MemoryRead(u64 vaddr, size_t size) override {
+        switch (size) {
+        case sizeof(u64):
+            return MemoryRead(vaddr, sizeof(u32))
+                | MemoryRead(vaddr + sizeof(u32), sizeof(u32)) << 32;
+        case sizeof(u32):
+            return MemoryRead(vaddr, sizeof(u16))
+                | MemoryRead(vaddr + sizeof(u16), sizeof(u16)) << 16;
+        case sizeof(u16):
+            return MemoryRead(vaddr, sizeof(u8))
+                | MemoryRead(vaddr + sizeof(u8), sizeof(u8)) << 8;
+        case sizeof(u8):
+            return memory[vaddr];
+        default:
+            std::abort();
+        }
     }
 
     std::array<u64, 2> MemoryRead128(u64 vaddr) override {
-        return {MemoryRead64(vaddr), MemoryRead64(vaddr + 8)};
+        return {
+            MemoryRead(vaddr, sizeof(u64)),
+            MemoryRead(vaddr + sizeof(u64), sizeof(u64))
+        };
     }
 
-    void MemoryWrite8(u64 vaddr, u8 value) override {
-        memory[vaddr] = value;
+    void MemoryWrite(Dynarmic::A64::VAddr vaddr, u64 value, size_t size) override {
+        switch (size) {
+        case sizeof(u64):
+            MemoryWrite(vaddr, u32(value), sizeof(u32));
+            MemoryWrite(vaddr + 4, u32(value >> 32), sizeof(u32));
+            break;
+        case sizeof(u32):
+            MemoryWrite(vaddr, u16(value), sizeof(u16));
+            MemoryWrite(vaddr + 2, u16(value >> 16), sizeof(u16));
+            break;
+        case sizeof(u16):
+            MemoryWrite(vaddr, u8(value), sizeof(u8));
+            MemoryWrite(vaddr + 1, u8(value >> 8), sizeof(u8));
+            break;
+        case sizeof(u8):
+            memory[vaddr] = value;
+            break;
+        default:
+            std::abort();
+        }
     }
-
-    void MemoryWrite16(u64 vaddr, u16 value) override {
-        MemoryWrite8(vaddr, u8(value));
-        MemoryWrite8(vaddr + 1, u8(value >> 8));
-    }
-
-    void MemoryWrite32(u64 vaddr, u32 value) override {
-        MemoryWrite16(vaddr, u16(value));
-        MemoryWrite16(vaddr + 2, u16(value >> 16));
-    }
-
-    void MemoryWrite64(u64 vaddr, u64 value) override {
-        MemoryWrite32(vaddr, u32(value));
-        MemoryWrite32(vaddr + 4, u32(value >> 32));
-    }
-
     void MemoryWrite128(u64 vaddr, std::array<u64, 2> value) override {
-        MemoryWrite64(vaddr, value[0]);
-        MemoryWrite64(vaddr + 8, value[1]);
+        MemoryWrite(vaddr, value[0], sizeof(u64));
+        MemoryWrite(vaddr + 8, value[1], sizeof(u64));
     }
 
     void CallSVC(u32) override {
@@ -135,9 +142,9 @@ TEST_CASE("A64: fibonacci", "[a64]") {
     code.RET();
 
     for (size_t i = 0; i < 1024; i++) {
-        env.MemoryWrite32(i * 4, instructions[i]);
+        env.MemoryWrite(i * 4, instructions[i], sizeof(u32));
     }
-    env.MemoryWrite32(8888, 0xd4200000);
+    env.MemoryWrite(8888, 0xd4200000, sizeof(u32));
     cpu.SetRegister(30, 8888);
 
     cpu.SetRegister(0, 10);

src/dynarmic/tests/A64/testenv.h

@@ -12,6 +12,7 @@
 #include "common/assert.h"
 #include "common/common_types.h"
 #include "dynarmic/interface/A64/a64.h"
+#include "dynarmic/interface/A64/config.h"
 
 using Vector = Dynarmic::A64::Vector;
 
@@ -34,64 +35,79 @@ public:
         return code_mem[index];
     }
 
-    std::uint8_t MemoryRead8(u64 vaddr) override {
-        if (IsInCodeMem(vaddr)) {
-            return reinterpret_cast<u8*>(code_mem.data())[vaddr - code_mem_start_address];
+    u64 MemoryRead(u64 vaddr, size_t size) override {
+        switch (size) {
+        case sizeof(u64):
+            return MemoryRead(vaddr, sizeof(u32))
+                | MemoryRead(vaddr + sizeof(u32), sizeof(u32)) << 32;
+        case sizeof(u32):
+            return MemoryRead(vaddr, sizeof(u16))
+                | MemoryRead(vaddr + sizeof(u16), sizeof(u16)) << 16;
+        case sizeof(u16):
+            return MemoryRead(vaddr, sizeof(u8))
+                | MemoryRead(vaddr + sizeof(u8), sizeof(u8)) << 8;
+        case sizeof(u8): {
+            if (IsInCodeMem(vaddr))
+                return reinterpret_cast<u8*>(code_mem.data())[vaddr - code_mem_start_address];
+            if (auto const it = modified_memory.find(vaddr); it != modified_memory.end())
+                return it->second;
+            return u8(vaddr);
+        }
+        default:
+            std::abort();
         }
-        if (auto const it = modified_memory.find(vaddr); it != modified_memory.end())
-            return it->second;
-        return u8(vaddr);
-    }
-    std::uint16_t MemoryRead16(u64 vaddr) override {
-        return u16(MemoryRead8(vaddr)) | u16(MemoryRead8(vaddr + 1)) << 8;
-    }
-    std::uint32_t MemoryRead32(u64 vaddr) override {
-        return u32(MemoryRead16(vaddr)) | u32(MemoryRead16(vaddr + 2)) << 16;
-    }
-    std::uint64_t MemoryRead64(u64 vaddr) override {
-        return u64(MemoryRead32(vaddr)) | u64(MemoryRead32(vaddr + 4)) << 32;
-    }
-    Vector MemoryRead128(u64 vaddr) override {
-        return {MemoryRead64(vaddr), MemoryRead64(vaddr + 8)};
     }
 
-    void MemoryWrite8(u64 vaddr, std::uint8_t value) override {
-        if (IsInCodeMem(vaddr)) {
-            code_mem_modified_by_guest = true;
+    Vector MemoryRead128(u64 vaddr) override {
+        return {
+            MemoryRead(vaddr, sizeof(u64)),
+            MemoryRead(vaddr + 8, sizeof(u64))
+        };
+    }
+
+    void MemoryWrite(Dynarmic::A64::VAddr vaddr, u64 value, size_t size) override {
+        switch (size) {
+        case sizeof(u64):
+            MemoryWrite(vaddr, u32(value), sizeof(u32));
+            MemoryWrite(vaddr + 4, u32(value >> 32), sizeof(u32));
+            break;
+        case sizeof(u32):
+            MemoryWrite(vaddr, u16(value), sizeof(u16));
+            MemoryWrite(vaddr + 2, u16(value >> 16), sizeof(u16));
+            break;
+        case sizeof(u16):
+            MemoryWrite(vaddr, u8(value), sizeof(u8));
+            MemoryWrite(vaddr + 1, u8(value >> 8), sizeof(u8));
+            break;
+        case sizeof(u8):
+            if (IsInCodeMem(vaddr)) {
+                code_mem_modified_by_guest = true;
+            }
+            modified_memory[vaddr] = value;
+            break;
+        default:
+            std::abort();
         }
-        modified_memory[vaddr] = value;
-    }
-    void MemoryWrite16(u64 vaddr, std::uint16_t value) override {
-        MemoryWrite8(vaddr, u8(value));
-        MemoryWrite8(vaddr + 1, u8(value >> 8));
-    }
-    void MemoryWrite32(u64 vaddr, std::uint32_t value) override {
-        MemoryWrite16(vaddr, u16(value));
-        MemoryWrite16(vaddr + 2, u16(value >> 16));
-    }
-    void MemoryWrite64(u64 vaddr, std::uint64_t value) override {
-        MemoryWrite32(vaddr, u32(value));
-        MemoryWrite32(vaddr + 4, u32(value >> 32));
     }
     void MemoryWrite128(u64 vaddr, Vector value) override {
-        MemoryWrite64(vaddr, value[0]);
-        MemoryWrite64(vaddr + 8, value[1]);
+        MemoryWrite(vaddr, value[0], sizeof(u64));
+        MemoryWrite(vaddr + 8, value[1], sizeof(u64));
     }
 
     bool MemoryWriteExclusive8(u64 vaddr, std::uint8_t value, [[maybe_unused]] std::uint8_t expected) override {
-        MemoryWrite8(vaddr, value);
+        MemoryWrite(vaddr, value, sizeof(u8));
         return true;
     }
     bool MemoryWriteExclusive16(u64 vaddr, std::uint16_t value, [[maybe_unused]] std::uint16_t expected) override {
-        MemoryWrite16(vaddr, value);
+        MemoryWrite(vaddr, value, sizeof(u16));
         return true;
     }
     bool MemoryWriteExclusive32(u64 vaddr, std::uint32_t value, [[maybe_unused]] std::uint32_t expected) override {
-        MemoryWrite32(vaddr, value);
+        MemoryWrite(vaddr, value, sizeof(u32));
         return true;
     }
     bool MemoryWriteExclusive64(u64 vaddr, std::uint64_t value, [[maybe_unused]] std::uint64_t expected) override {
-        MemoryWrite64(vaddr, value);
+        MemoryWrite(vaddr, value, sizeof(u64));
         return true;
     }
     bool MemoryWriteExclusive128(u64 vaddr, Vector value, [[maybe_unused]] Vector expected) override {
@@ -145,52 +161,46 @@ public:
         return read<std::uint32_t>(vaddr);
     }
 
-    std::uint8_t MemoryRead8(u64 vaddr) override {
-        return read<std::uint8_t>(vaddr);
-    }
-    std::uint16_t MemoryRead16(u64 vaddr) override {
-        return read<std::uint16_t>(vaddr);
-    }
-    std::uint32_t MemoryRead32(u64 vaddr) override {
-        return read<std::uint32_t>(vaddr);
-    }
-    std::uint64_t MemoryRead64(u64 vaddr) override {
-        return read<std::uint64_t>(vaddr);
+    u64 MemoryRead(u64 vaddr, size_t size) override {
+        switch (size) {
+        case sizeof(u64): return read<u64>(vaddr);
+        case sizeof(u32): return read<u32>(vaddr);
+        case sizeof(u16): return read<u16>(vaddr);
+        case sizeof(u8): return read<u8>(vaddr);
+        default: std::abort();
+        }
     }
     Vector MemoryRead128(u64 vaddr) override {
         return read<Vector>(vaddr);
     }
 
-    void MemoryWrite8(u64 vaddr, std::uint8_t value) override {
-        write(vaddr, value);
-    }
-    void MemoryWrite16(u64 vaddr, std::uint16_t value) override {
-        write(vaddr, value);
-    }
-    void MemoryWrite32(u64 vaddr, std::uint32_t value) override {
-        write(vaddr, value);
-    }
-    void MemoryWrite64(u64 vaddr, std::uint64_t value) override {
-        write(vaddr, value);
+    void MemoryWrite(u64 vaddr, std::uint64_t value, size_t size) override {
+        switch (size) {
+        case sizeof(u64): return write<u64>(vaddr, u64(value));
+        case sizeof(u32): return write<u32>(vaddr, u32(value));
+        case sizeof(u16): return write<u16>(vaddr, u16(value));
+        case sizeof(u8): return write<u8>(vaddr, u8(value));
+        default: std::abort();
+        }
     }
     void MemoryWrite128(u64 vaddr, Vector value) override {
         write(vaddr, value);
     }
 
     bool MemoryWriteExclusive8(u64 vaddr, std::uint8_t value, [[maybe_unused]] std::uint8_t expected) override {
-        MemoryWrite8(vaddr, value);
+        MemoryWrite(vaddr, value, sizeof(u8));
         return true;
     }
     bool MemoryWriteExclusive16(u64 vaddr, std::uint16_t value, [[maybe_unused]] std::uint16_t expected) override {
-        MemoryWrite16(vaddr, value);
+        MemoryWrite(vaddr, value, sizeof(u16));
         return true;
     }
     bool MemoryWriteExclusive32(u64 vaddr, std::uint32_t value, [[maybe_unused]] std::uint32_t expected) override {
-        MemoryWrite32(vaddr, value);
+        MemoryWrite(vaddr, value, sizeof(u32));
         return true;
     }
     bool MemoryWriteExclusive64(u64 vaddr, std::uint64_t value, [[maybe_unused]] std::uint64_t expected) override {
-        MemoryWrite64(vaddr, value);
+        MemoryWrite(vaddr, value, sizeof(u64));
         return true;
     }
     bool MemoryWriteExclusive128(u64 vaddr, Vector value, [[maybe_unused]] Vector expected) override {

src/dynarmic/tests/print_info.cpp

@@ -18,6 +18,7 @@
 #include <fmt/format.h>
 #include <fmt/ostream.h>
 #include <fmt/ranges.h>
+#include "dynarmic/frontend/A32/translate/translate_callbacks.h"
 #include "dynarmic/mcl/bit.hpp"
 #include "common/common_types.h"
 
@@ -118,36 +119,47 @@ public:
     u64 ticks_left = 0;
     std::map<u32, u8> memory;
 
-    std::uint8_t MemoryRead8(u32 vaddr) override {
-        if (auto iter = memory.find(vaddr); iter != memory.end()) {
-            return iter->second;
+    u64 MemoryRead(Dynarmic::A32::VAddr vaddr, size_t size) override {
+        switch (size) {
+        case sizeof(u64):
+            return MemoryRead(vaddr, sizeof(u32))
+                | MemoryRead(vaddr + sizeof(u32), sizeof(u32)) << 32;
+        case sizeof(u32):
+            return MemoryRead(vaddr, sizeof(u16))
+                | MemoryRead(vaddr + sizeof(u16), sizeof(u16)) << 16;
+        case sizeof(u16):
+            return MemoryRead(vaddr, sizeof(u8))
+                | MemoryRead(vaddr + sizeof(u8), sizeof(u8)) << 8;
+        case sizeof(u8): {
+            if (auto const it = memory.find(vaddr); it != memory.end())
+                return it->second;
+            return 0;
+        }
+        default:
+            std::abort();
         }
-        return 0;
-    }
-    std::uint16_t MemoryRead16(u32 vaddr) override {
-        return u16(MemoryRead8(vaddr)) | u16(MemoryRead8(vaddr + 1)) << 8;
-    }
-    std::uint32_t MemoryRead32(u32 vaddr) override {
-        return u32(MemoryRead16(vaddr)) | u32(MemoryRead16(vaddr + 2)) << 16;
-    }
-    std::uint64_t MemoryRead64(u32 vaddr) override {
-        return u64(MemoryRead32(vaddr)) | u64(MemoryRead32(vaddr + 4)) << 32;
     }
 
-    void MemoryWrite8(u32 vaddr, std::uint8_t value) override {
-        memory[vaddr] = value;
-    }
-    void MemoryWrite16(u32 vaddr, std::uint16_t value) override {
-        MemoryWrite8(vaddr, static_cast<u8>(value));
-        MemoryWrite8(vaddr + 1, static_cast<u8>(value >> 8));
-    }
-    void MemoryWrite32(u32 vaddr, std::uint32_t value) override {
-        MemoryWrite16(vaddr, static_cast<u16>(value));
-        MemoryWrite16(vaddr + 2, static_cast<u16>(value >> 16));
-    }
-    void MemoryWrite64(u32 vaddr, std::uint64_t value) override {
-        MemoryWrite32(vaddr, static_cast<u32>(value));
-        MemoryWrite32(vaddr + 4, static_cast<u32>(value >> 32));
+    void MemoryWrite(Dynarmic::A32::VAddr vaddr, u64 value, size_t size) override {
+        switch (size) {
+        case sizeof(u64):
+            MemoryWrite(vaddr, u32(value), sizeof(u32));
+            MemoryWrite(vaddr + 4, u32(value >> 32), sizeof(u32));
+            break;
+        case sizeof(u32):
+            MemoryWrite(vaddr, u16(value), sizeof(u16));
+            MemoryWrite(vaddr + 2, u16(value >> 16), sizeof(u16));
+            break;
+        case sizeof(u16):
+            MemoryWrite(vaddr, u8(value), sizeof(u8));
+            MemoryWrite(vaddr + 1, u8(value >> 8), sizeof(u8));
+            break;
+        case sizeof(u8):
+            memory[vaddr] = value;
+            break;
+        default:
+            std::abort();
+        }
     }
 
     void CallSVC(std::uint32_t swi) override {
@@ -231,7 +243,7 @@ void ExecuteA32Instruction(u32 instruction) {
             if (const auto address = get_value()) {
                 fmt::print("value: ");
                 if (const auto value = get_value()) {
-                    env.MemoryWrite32(*address, *value);
+                    env.MemoryWrite(*address, *value, sizeof(u32));
                     fmt::print("> mem[{:#08x}] = {:#08x}\n", *address, *value);
                 }
             }
@@ -247,8 +259,8 @@ void ExecuteA32Instruction(u32 instruction) {
     cpu.SetFpscr(fpscr);
 
     const u32 initial_pc = regs[15];
-    env.MemoryWrite32(initial_pc + 0, instruction);
-    env.MemoryWrite32(initial_pc + 4, 0xEAFFFFFE);  // B +0
+    env.MemoryWrite(initial_pc + 0, instruction, sizeof(u32));
+    env.MemoryWrite(initial_pc + 4, 0xEAFFFFFE, sizeof(u32));  // B +0
 
     cpu.Run();
     fmt::print("{}", fmt::join(cpu.Disassemble(), "\n"));

src/dynarmic/tests/unicorn_emu/a32_unicorn.cpp

@@ -290,7 +290,7 @@ bool A32Unicorn<TestEnvironment>::UnmappedMemoryHook(uc_engine* uc, uc_mem_type
         auto page = std::make_unique<Page>();
         page->address = base_address;
         for (size_t i = 0; i < page->data.size(); ++i)
-            page->data[i] = this_->testenv.MemoryRead8(static_cast<u32>(base_address + i));
+            page->data[i] = u8(this_->testenv.MemoryRead(u32(base_address + i), sizeof(u8)));
 
         uc_err err = uc_mem_map_ptr(uc, base_address, page->data.size(), permissions, page->data.data());
         if (err == UC_ERR_MAP)
@@ -321,24 +321,7 @@ bool A32Unicorn<TestEnvironment>::UnmappedMemoryHook(uc_engine* uc, uc_mem_type
 template<class TestEnvironment>
 bool A32Unicorn<TestEnvironment>::MemoryWriteHook(uc_engine* /*uc*/, uc_mem_type /*type*/, u32 start_address, int size, u64 value, void* user_data) {
     auto* this_ = static_cast<A32Unicorn*>(user_data);
-
-    switch (size) {
-    case 1:
-        this_->testenv.MemoryWrite8(start_address, static_cast<u8>(value));
-        break;
-    case 2:
-        this_->testenv.MemoryWrite16(start_address, static_cast<u16>(value));
-        break;
-    case 4:
-        this_->testenv.MemoryWrite32(start_address, static_cast<u32>(value));
-        break;
-    case 8:
-        this_->testenv.MemoryWrite64(start_address, value);
-        break;
-    default:
-        UNREACHABLE();
-    }
-
+    this_->testenv.MemoryWrite(start_address, value, size);
     return true;
 }
 

src/dynarmic/tests/unicorn_emu/a64_unicorn.cpp

@@ -197,7 +197,7 @@ bool A64Unicorn::UnmappedMemoryHook(uc_engine* uc, uc_mem_type /*type*/, u64 sta
         auto page = std::make_unique<Page>();
         page->address = base_address;
         for (size_t i = 0; i < page->data.size(); ++i)
-            page->data[i] = this_->testenv.MemoryRead8(base_address + i);
+            page->data[i] = u8(this_->testenv.MemoryRead(base_address + i, sizeof(u8)));
 
         uc_err err = uc_mem_map_ptr(uc, base_address, page->data.size(), permissions, page->data.data());
         if (err == UC_ERR_MAP)
@@ -227,23 +227,6 @@ bool A64Unicorn::UnmappedMemoryHook(uc_engine* uc, uc_mem_type /*type*/, u64 sta
 
 bool A64Unicorn::MemoryWriteHook(uc_engine* /*uc*/, uc_mem_type /*type*/, u64 start_address, int size, u64 value, void* user_data) {
     auto* this_ = static_cast<A64Unicorn*>(user_data);
-
-    switch (size) {
-    case 1:
-        this_->testenv.MemoryWrite8(start_address, static_cast<u8>(value));
-        break;
-    case 2:
-        this_->testenv.MemoryWrite16(start_address, static_cast<u16>(value));
-        break;
-    case 4:
-        this_->testenv.MemoryWrite32(start_address, static_cast<u32>(value));
-        break;
-    case 8:
-        this_->testenv.MemoryWrite64(start_address, value);
-        break;
-    default:
-        UNREACHABLE();
-    }
-
+    this_->testenv.MemoryWrite(start_address, value, size);
     return true;
 }

src/hid_core/resources/npad/npad.cpp

@@ -398,21 +398,21 @@ void NPad::InitNewlyAddedController(Kernel::KernelCore& kernel, u64 aruid, Core:
 void NPad::WriteEmptyEntry(NpadInternalState* npad) {
     NPadGenericState dummy_pad_state{};
     NpadGcTriggerState dummy_gc_state{};
-    dummy_pad_state.sampling_number = npad->fullkey_lifo.ReadCurrentEntry().sampling_number + 1;
+    dummy_pad_state.sampling_number = npad->fullkey_lifo.ReadCurrentEntry().state.sampling_number + 1;
     npad->fullkey_lifo.WriteNextEntry(dummy_pad_state);
-    dummy_pad_state.sampling_number = npad->handheld_lifo.ReadCurrentEntry().sampling_number + 1;
+    dummy_pad_state.sampling_number = npad->handheld_lifo.ReadCurrentEntry().state.sampling_number + 1;
     npad->handheld_lifo.WriteNextEntry(dummy_pad_state);
-    dummy_pad_state.sampling_number = npad->joy_dual_lifo.ReadCurrentEntry().sampling_number + 1;
+    dummy_pad_state.sampling_number = npad->joy_dual_lifo.ReadCurrentEntry().state.sampling_number + 1;
     npad->joy_dual_lifo.WriteNextEntry(dummy_pad_state);
-    dummy_pad_state.sampling_number = npad->joy_left_lifo.ReadCurrentEntry().sampling_number + 1;
+    dummy_pad_state.sampling_number = npad->joy_left_lifo.ReadCurrentEntry().state.sampling_number + 1;
     npad->joy_left_lifo.WriteNextEntry(dummy_pad_state);
-    dummy_pad_state.sampling_number = npad->joy_right_lifo.ReadCurrentEntry().sampling_number + 1;
+    dummy_pad_state.sampling_number = npad->joy_right_lifo.ReadCurrentEntry().state.sampling_number + 1;
     npad->joy_right_lifo.WriteNextEntry(dummy_pad_state);
-    dummy_pad_state.sampling_number = npad->palma_lifo.ReadCurrentEntry().sampling_number + 1;
+    dummy_pad_state.sampling_number = npad->palma_lifo.ReadCurrentEntry().state.sampling_number + 1;
     npad->palma_lifo.WriteNextEntry(dummy_pad_state);
-    dummy_pad_state.sampling_number = npad->system_ext_lifo.ReadCurrentEntry().sampling_number + 1;
+    dummy_pad_state.sampling_number = npad->system_ext_lifo.ReadCurrentEntry().state.sampling_number + 1;
     npad->system_ext_lifo.WriteNextEntry(dummy_pad_state);
-    dummy_gc_state.sampling_number = npad->gc_trigger_lifo.ReadCurrentEntry().sampling_number + 1;
+    dummy_gc_state.sampling_number = npad->gc_trigger_lifo.ReadCurrentEntry().state.sampling_number + 1;
     npad->gc_trigger_lifo.WriteNextEntry(dummy_gc_state);
 }