early-access version 4100

CMakeLists.txt

@@ -322,6 +322,10 @@ if (ARCHITECTURE_x86 OR ARCHITECTURE_x86_64)
     find_package(xbyak 6 CONFIG)
 endif()
 
+if (ARCHITECTURE_arm64)
+    find_package(oaknut 2.0.1 CONFIG)
+endif()
+
 if (ARCHITECTURE_x86_64 OR ARCHITECTURE_arm64)
     find_package(dynarmic 6.4.0 CONFIG)
 endif()

README.md

@@ -1,7 +1,7 @@
 yuzu emulator early access
 =============
 
-This is the source code for early-access 4099.
+This is the source code for early-access 4100.
 
 ## Legal Notice
 

externals/CMakeLists.txt

@@ -14,16 +14,17 @@ set(BUILD_SHARED_LIBS OFF)
 # Skip install rules for all externals
 set_directory_properties(PROPERTIES EXCLUDE_FROM_ALL ON)
 
-# xbyak
+# Xbyak (also used by Dynarmic, so needs to be added first)
 if ((ARCHITECTURE_x86 OR ARCHITECTURE_x86_64) AND NOT TARGET xbyak::xbyak)
     add_subdirectory(xbyak)
 endif()
 
-# Dynarmic
+# Oaknut (also used by Dynarmic, so needs to be added first)
 if (ARCHITECTURE_arm64 AND NOT TARGET merry::oaknut)
     add_subdirectory(oaknut)
 endif()
 
+# Dynarmic
 if ((ARCHITECTURE_x86_64 OR ARCHITECTURE_arm64) AND NOT TARGET dynarmic::dynarmic)
     set(DYNARMIC_IGNORE_ASSERTS ON)
     add_subdirectory(dynarmic)

src/android/app/src/main/java/org/yuzu/yuzu_emu/NativeLibrary.kt

@@ -261,7 +261,7 @@ object NativeLibrary {
     /**
      * Begins emulation.
      */
-    external fun run(path: String?, programIndex: Int = 0)
+    external fun run(path: String?, programIndex: Int, frontendInitiated: Boolean)
 
     // Surface Handling
     external fun surfaceChanged(surf: Surface?)

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

@@ -927,7 +927,7 @@ class EmulationFragment : Fragment(), SurfaceHolder.Callback {
             emulationThread.join()
             emulationThread = Thread({
                 Log.debug("[EmulationFragment] Starting emulation thread.")
-                NativeLibrary.run(gamePath, programIndex)
+                NativeLibrary.run(gamePath, programIndex, false)
             }, "NativeEmulation")
             emulationThread.start()
         }
@@ -981,7 +981,7 @@ class EmulationFragment : Fragment(), SurfaceHolder.Callback {
                 State.STOPPED -> {
                     emulationThread = Thread({
                         Log.debug("[EmulationFragment] Starting emulation thread.")
-                        NativeLibrary.run(gamePath, programIndex)
+                        NativeLibrary.run(gamePath, programIndex, true)
                     }, "NativeEmulation")
                     emulationThread.start()
                 }

src/android/app/src/main/jni/native.cpp

@@ -219,7 +219,8 @@ void EmulationSession::SetAppletId(int applet_id) {
 }
 
 Core::SystemResultStatus EmulationSession::InitializeEmulation(const std::string& filepath,
-                                                               const std::size_t program_index) {
+                                                               const std::size_t program_index,
+                                                               const bool frontend_initiated) {
     std::scoped_lock lock(m_mutex);
 
     // Create the render window.
@@ -251,6 +252,8 @@ Core::SystemResultStatus EmulationSession::InitializeEmulation(const std::string
     // Load the ROM.
     Service::AM::FrontendAppletParameters params{
         .applet_id = static_cast<Service::AM::AppletId>(m_applet_id),
+        .launch_type = frontend_initiated ? Service::AM::LaunchType::FrontendInitiated
+                                          : Service::AM::LaunchType::ApplicationInitiated,
         .program_index = static_cast<s32>(program_index),
     };
     m_load_result = m_system.Load(EmulationSession::GetInstance().Window(), filepath, params);
@@ -447,7 +450,8 @@ u64 EmulationSession::GetProgramId(JNIEnv* env, jstring jprogramId) {
 }
 
 static Core::SystemResultStatus RunEmulation(const std::string& filepath,
-                                             const size_t program_index = 0) {
+                                             const size_t program_index,
+                                             const bool frontend_initiated) {
     MicroProfileOnThreadCreate("EmuThread");
     SCOPE_EXIT({ MicroProfileShutdown(); });
 
@@ -460,7 +464,8 @@ static Core::SystemResultStatus RunEmulation(const std::string& filepath,
 
     SCOPE_EXIT({ EmulationSession::GetInstance().ShutdownEmulation(); });
 
-    jconst result = EmulationSession::GetInstance().InitializeEmulation(filepath, program_index);
+    jconst result = EmulationSession::GetInstance().InitializeEmulation(filepath, program_index,
+                                                                        frontend_initiated);
     if (result != Core::SystemResultStatus::Success) {
         return result;
     }
@@ -757,10 +762,12 @@ void Java_org_yuzu_yuzu_1emu_NativeLibrary_logSettings(JNIEnv* env, jobject jobj
 }
 
 void Java_org_yuzu_yuzu_1emu_NativeLibrary_run(JNIEnv* env, jobject jobj, jstring j_path,
-                                               jint j_program_index) {
+                                               jint j_program_index,
+                                               jboolean j_frontend_initiated) {
     const std::string path = GetJString(env, j_path);
 
-    const Core::SystemResultStatus result{RunEmulation(path, j_program_index)};
+    const Core::SystemResultStatus result{
+        RunEmulation(path, j_program_index, j_frontend_initiated)};
     if (result != Core::SystemResultStatus::Success) {
         env->CallStaticVoidMethod(IDCache::GetNativeLibraryClass(),
                                   IDCache::GetExitEmulationActivity(), static_cast<int>(result));

src/android/app/src/main/jni/native.h

@@ -47,7 +47,8 @@ public:
     void InitializeSystem(bool reload);
     void SetAppletId(int applet_id);
     Core::SystemResultStatus InitializeEmulation(const std::string& filepath,
-                                                 const std::size_t program_index = 0);
+                                                 const std::size_t program_index,
+                                                 const bool frontend_initiated);
 
     bool IsHandheldOnly();
     void SetDeviceType([[maybe_unused]] int index, int type);

src/core/hle/service/cmif_serialization.h

@@ -122,14 +122,14 @@ struct RequestLayout {
     u32 domain_interface_count;
 };
 
-template <ArgumentType Type1, ArgumentType Type2, typename MethodArguments, size_t PrevAlign = 1, size_t DataOffset = 0, size_t ArgIndex = 0>
+template <typename MethodArguments, size_t PrevAlign = 1, size_t DataOffset = 0, size_t ArgIndex = 0>
-constexpr u32 GetArgumentRawDataSize() {
+constexpr u32 GetInRawDataSize() {
     if constexpr (ArgIndex >= std::tuple_size_v<MethodArguments>) {
         return static_cast<u32>(DataOffset);
     } else {
         using ArgType = std::tuple_element_t<ArgIndex, MethodArguments>;
 
-        if constexpr (ArgumentTraits<ArgType>::Type == Type1 || ArgumentTraits<ArgType>::Type == Type2) {
+        if constexpr (ArgumentTraits<ArgType>::Type == ArgumentType::InData || ArgumentTraits<ArgType>::Type == ArgumentType::InProcessId) {
             constexpr size_t ArgAlign = alignof(ArgType);
             constexpr size_t ArgSize = sizeof(ArgType);
 
@@ -138,9 +138,33 @@ constexpr u32 GetArgumentRawDataSize() {
             constexpr size_t ArgOffset = Common::AlignUp(DataOffset, ArgAlign);
             constexpr size_t ArgEnd = ArgOffset + ArgSize;
 
-            return GetArgumentRawDataSize<Type1, Type2, MethodArguments, ArgAlign, ArgEnd, ArgIndex + 1>();
+            return GetInRawDataSize<MethodArguments, ArgAlign, ArgEnd, ArgIndex + 1>();
         } else {
-            return GetArgumentRawDataSize<Type1, Type2, MethodArguments, PrevAlign, DataOffset, ArgIndex + 1>();
+            return GetInRawDataSize<MethodArguments, PrevAlign, DataOffset, ArgIndex + 1>();
+        }
+    }
+}
+
+template <typename MethodArguments, size_t PrevAlign = 1, size_t DataOffset = 0, size_t ArgIndex = 0>
+constexpr u32 GetOutRawDataSize() {
+    if constexpr (ArgIndex >= std::tuple_size_v<MethodArguments>) {
+        return static_cast<u32>(DataOffset);
+    } else {
+        using ArgType = std::tuple_element_t<ArgIndex, MethodArguments>;
+
+        if constexpr (ArgumentTraits<ArgType>::Type == ArgumentType::OutData) {
+            using RawArgType = typename ArgType::Type;
+            constexpr size_t ArgAlign = alignof(RawArgType);
+            constexpr size_t ArgSize = sizeof(RawArgType);
+
+            static_assert(PrevAlign <= ArgAlign, "Output argument is not ordered by alignment");
+
+            constexpr size_t ArgOffset = Common::AlignUp(DataOffset, ArgAlign);
+            constexpr size_t ArgEnd = ArgOffset + ArgSize;
+
+            return GetOutRawDataSize<MethodArguments, ArgAlign, ArgEnd, ArgIndex + 1>();
+        } else {
+            return GetOutRawDataSize<MethodArguments, PrevAlign, DataOffset, ArgIndex + 1>();
         }
     }
 }
@@ -165,7 +189,7 @@ constexpr RequestLayout GetNonDomainReplyInLayout() {
     return RequestLayout{
         .copy_handle_count = GetArgumentTypeCount<ArgumentType::InCopyHandle, MethodArguments>(),
         .move_handle_count = 0,
-        .cmif_raw_data_size = GetArgumentRawDataSize<ArgumentType::InData, ArgumentType::InProcessId, MethodArguments>(),
+        .cmif_raw_data_size = GetInRawDataSize<MethodArguments>(),
         .domain_interface_count = 0,
     };
 }
@@ -175,7 +199,7 @@ constexpr RequestLayout GetDomainReplyInLayout() {
     return RequestLayout{
         .copy_handle_count = GetArgumentTypeCount<ArgumentType::InCopyHandle, MethodArguments>(),
         .move_handle_count = 0,
-        .cmif_raw_data_size = GetArgumentRawDataSize<ArgumentType::InData, ArgumentType::InProcessId, MethodArguments>(),
+        .cmif_raw_data_size = GetInRawDataSize<MethodArguments>(),
         .domain_interface_count = GetArgumentTypeCount<ArgumentType::InInterface, MethodArguments>(),
     };
 }
@@ -185,7 +209,7 @@ constexpr RequestLayout GetNonDomainReplyOutLayout() {
     return RequestLayout{
         .copy_handle_count = GetArgumentTypeCount<ArgumentType::OutCopyHandle, MethodArguments>(),
         .move_handle_count = GetArgumentTypeCount<ArgumentType::OutMoveHandle, MethodArguments>() + GetArgumentTypeCount<ArgumentType::OutInterface, MethodArguments>(),
-        .cmif_raw_data_size = GetArgumentRawDataSize<ArgumentType::OutData, ArgumentType::OutData, MethodArguments>(),
+        .cmif_raw_data_size = GetOutRawDataSize<MethodArguments>(),
         .domain_interface_count = 0,
     };
 }
@@ -195,7 +219,7 @@ constexpr RequestLayout GetDomainReplyOutLayout() {
     return RequestLayout{
         .copy_handle_count = GetArgumentTypeCount<ArgumentType::OutCopyHandle, MethodArguments>(),
         .move_handle_count = GetArgumentTypeCount<ArgumentType::OutMoveHandle, MethodArguments>(),
-        .cmif_raw_data_size = GetArgumentRawDataSize<ArgumentType::OutData, ArgumentType::OutData, MethodArguments>(),
+        .cmif_raw_data_size = GetOutRawDataSize<MethodArguments>(),
         .domain_interface_count = GetArgumentTypeCount<ArgumentType::OutInterface, MethodArguments>(),
     };
 }
@@ -337,13 +361,15 @@ void WriteOutArgument(bool is_domain, CallArguments& args, u8* raw_data, HLERequ
         using ArgType = std::tuple_element_t<ArgIndex, MethodArguments>;
 
         if constexpr (ArgumentTraits<ArgType>::Type == ArgumentType::OutData) {
-            constexpr size_t ArgAlign = alignof(ArgType);
+            using RawArgType = decltype(std::get<ArgIndex>(args).raw);
-            constexpr size_t ArgSize = sizeof(ArgType);
+            constexpr size_t ArgAlign = alignof(RawArgType);
+            constexpr size_t ArgSize = sizeof(RawArgType);
 
             static_assert(PrevAlign <= ArgAlign, "Output argument is not ordered by alignment");
             static_assert(!RawDataFinished, "All output interface arguments must appear after raw data");
             static_assert(!std::is_pointer_v<ArgType>, "Output raw data must not be a pointer");
-            static_assert(std::is_trivially_copyable_v<decltype(std::get<ArgIndex>(args).raw)>, "Output raw data must be trivially copyable");
+            static_assert(!std::is_pointer_v<RawArgType>, "Output raw data must not be a pointer");
+            static_assert(std::is_trivially_copyable_v<RawArgType>, "Output raw data must be trivially copyable");
 
             constexpr size_t ArgOffset = Common::AlignUp(DataOffset, ArgAlign);
             constexpr size_t ArgEnd = ArgOffset + ArgSize;

src/core/hle/service/nvdrv/core/container.cpp

@@ -83,7 +83,9 @@ SessionId Container::OpenSession(Kernel::KProcess* process) {
 
             // Check if this memory block is heap.
             if (svc_mem_info.state == Kernel::Svc::MemoryState::Normal) {
-                if (svc_mem_info.size > region_size) {
+                if (region_start + region_size == svc_mem_info.base_address) {
+                    region_size += svc_mem_info.size;
+                } else if (svc_mem_info.size > region_size) {
                     region_size = svc_mem_info.size;
                     region_start = svc_mem_info.base_address;
                 }

src/core/hle/service/sockets/sockets.h

@@ -24,6 +24,7 @@ enum class Errno : u32 {
     CONNRESET = 104,
     NOTCONN = 107,
     TIMEDOUT = 110,
+    CONNREFUSED = 111,
     INPROGRESS = 115,
 };
 

src/core/hle/service/sockets/sockets_translate.cpp

@@ -25,6 +25,8 @@ Errno Translate(Network::Errno value) {
         return Errno::MFILE;
     case Network::Errno::PIPE:
         return Errno::PIPE;
+    case Network::Errno::CONNREFUSED:
+        return Errno::CONNREFUSED;
     case Network::Errno::NOTCONN:
         return Errno::NOTCONN;
     case Network::Errno::TIMEDOUT:

src/core/internal_network/network.cpp

@@ -693,20 +693,23 @@ std::pair<SocketBase::AcceptResult, Errno> Socket::Accept() {
     sockaddr_in addr;
     socklen_t addrlen = sizeof(addr);
 
-    std::vector<WSAPOLLFD> host_pollfds{
+    const bool wait_for_accept = !is_non_blocking;
-        WSAPOLLFD{fd, POLLIN, 0},
+    if (wait_for_accept) {
-        WSAPOLLFD{GetInterruptSocket(), POLLIN, 0},
+        std::vector<WSAPOLLFD> host_pollfds{
-    };
+            WSAPOLLFD{fd, POLLIN, 0},
+            WSAPOLLFD{GetInterruptSocket(), POLLIN, 0},
+        };
 
-    while (true) {
+        while (true) {
-        const int pollres =
+            const int pollres =
-            WSAPoll(host_pollfds.data(), static_cast<ULONG>(host_pollfds.size()), -1);
+                WSAPoll(host_pollfds.data(), static_cast<ULONG>(host_pollfds.size()), -1);
-        if (host_pollfds[1].revents != 0) {
+            if (host_pollfds[1].revents != 0) {
-            // Interrupt signaled before a client could be accepted, break
+                // Interrupt signaled before a client could be accepted, break
-            return {AcceptResult{}, Errno::AGAIN};
+                return {AcceptResult{}, Errno::AGAIN};
-        }
+            }
-        if (pollres > 0) {
+            if (pollres > 0) {
-            break;
+                break;
+            }
         }
     }
 
@@ -913,6 +916,7 @@ Errno Socket::SetRcvTimeo(u32 value) {
 
 Errno Socket::SetNonBlock(bool enable) {
     if (EnableNonBlock(fd, enable)) {
+        is_non_blocking = enable;
         return Errno::SUCCESS;
     }
     return GetAndLogLastError();

src/core/internal_network/sockets.h

@@ -166,6 +166,9 @@ public:
     bool IsOpened() const override;
 
     void HandleProxyPacket(const ProxyPacket& packet) override;
+
+private:
+    bool is_non_blocking = false;
 };
 
 std::pair<s32, Errno> Poll(std::vector<PollFD>& poll_fds, s32 timeout);

src/video_core/texture_cache/texture_cache.h

@@ -1431,7 +1431,8 @@ ImageId TextureCache<P>::JoinImages(const ImageInfo& info, GPUVAddr gpu_addr, DA
             }
         }
     };
-    ForEachSparseImageInRegion(gpu_addr, size_bytes, region_check_gpu);
+    ForEachSparseImageInRegion(channel_state->gpu_memory.GetID(), gpu_addr, size_bytes,
+                               region_check_gpu);
 
     bool can_rescale = info.rescaleable;
     bool any_rescaled = false;
@@ -1842,7 +1843,7 @@ void TextureCache<P>::ForEachImageInRegionGPU(size_t as_id, GPUVAddr gpu_addr, s
     if (!storage_id) {
         return;
     }
-    auto& gpu_page_table = gpu_page_table_storage[*storage_id];
+    auto& gpu_page_table = gpu_page_table_storage[*storage_id * 2];
     ForEachGPUPage(gpu_addr, size,
                    [this, &gpu_page_table, &images, gpu_addr, size, func](u64 page) {
                        const auto it = gpu_page_table.find(page);
@@ -1882,41 +1883,48 @@ void TextureCache<P>::ForEachImageInRegionGPU(size_t as_id, GPUVAddr gpu_addr, s
 
 template <class P>
 template <typename Func>
-void TextureCache<P>::ForEachSparseImageInRegion(GPUVAddr gpu_addr, size_t size, Func&& func) {
+void TextureCache<P>::ForEachSparseImageInRegion(size_t as_id, GPUVAddr gpu_addr, size_t size,
+                                                 Func&& func) {
     using FuncReturn = typename std::invoke_result<Func, ImageId, Image&>::type;
     static constexpr bool BOOL_BREAK = std::is_same_v<FuncReturn, bool>;
     boost::container::small_vector<ImageId, 8> images;
-    ForEachGPUPage(gpu_addr, size, [this, &images, gpu_addr, size, func](u64 page) {
+    auto storage_id = getStorageID(as_id);
-        const auto it = sparse_page_table.find(page);
+    if (!storage_id) {
-        if (it == sparse_page_table.end()) {
+        return;
-            if constexpr (BOOL_BREAK) {
+    }
-                return false;
+    auto& sparse_page_table = gpu_page_table_storage[*storage_id * 2 + 1];
-            } else {
+    ForEachGPUPage(gpu_addr, size,
-                return;
+                   [this, &sparse_page_table, &images, gpu_addr, size, func](u64 page) {
-            }
+                       const auto it = sparse_page_table.find(page);
-        }
+                       if (it == sparse_page_table.end()) {
-        for (const ImageId image_id : it->second) {
+                           if constexpr (BOOL_BREAK) {
-            Image& image = slot_images[image_id];
+                               return false;
-            if (True(image.flags & ImageFlagBits::Picked)) {
+                           } else {
-                continue;
+                               return;
-            }
+                           }
-            if (!image.OverlapsGPU(gpu_addr, size)) {
+                       }
-                continue;
+                       for (const ImageId image_id : it->second) {
-            }
+                           Image& image = slot_images[image_id];
-            image.flags |= ImageFlagBits::Picked;
+                           if (True(image.flags & ImageFlagBits::Picked)) {
-            images.push_back(image_id);
+                               continue;
-            if constexpr (BOOL_BREAK) {
+                           }
-                if (func(image_id, image)) {
+                           if (!image.OverlapsGPU(gpu_addr, size)) {
-                    return true;
+                               continue;
-                }
+                           }
-            } else {
+                           image.flags |= ImageFlagBits::Picked;
-                func(image_id, image);
+                           images.push_back(image_id);
-            }
+                           if constexpr (BOOL_BREAK) {
-        }
+                               if (func(image_id, image)) {
-        if constexpr (BOOL_BREAK) {
+                                   return true;
-            return false;
+                               }
-        }
+                           } else {
-    });
+                               func(image_id, image);
+                           }
+                       }
+                       if constexpr (BOOL_BREAK) {
+                           return false;
+                       }
+                   });
     for (const ImageId image_id : images) {
         slot_images[image_id].flags &= ~ImageFlagBits::Picked;
     }
@@ -1988,8 +1996,9 @@ void TextureCache<P>::RegisterImage(ImageId image_id) {
             sparse_maps.push_back(map_id);
         });
     sparse_views.emplace(image_id, std::move(sparse_maps));
-    ForEachGPUPage(image.gpu_addr, image.guest_size_bytes,
+    ForEachGPUPage(image.gpu_addr, image.guest_size_bytes, [this, image_id](u64 page) {
-                   [this, image_id](u64 page) { sparse_page_table[page].push_back(image_id); });
+        (*channel_state->sparse_page_table)[page].push_back(image_id);
+    });
 }
 
 template <class P>
@@ -2042,7 +2051,7 @@ void TextureCache<P>::UnregisterImage(ImageId image_id) {
         return;
     }
     ForEachGPUPage(image.gpu_addr, image.guest_size_bytes, [this, &clear_page_table](u64 page) {
-        clear_page_table(page, sparse_page_table);
+        clear_page_table(page, (*channel_state->sparse_page_table));
     });
     auto it = sparse_views.find(image_id);
     ASSERT(it != sparse_views.end());
@@ -2496,13 +2505,15 @@ void TextureCache<P>::CreateChannel(struct Tegra::Control::ChannelState& channel
     const auto it = channel_map.find(channel.bind_id);
     auto* this_state = &channel_storage[it->second];
     const auto& this_as_ref = address_spaces[channel.memory_manager->GetID()];
-    this_state->gpu_page_table = &gpu_page_table_storage[this_as_ref.storage_id];
+    this_state->gpu_page_table = &gpu_page_table_storage[this_as_ref.storage_id * 2];
+    this_state->sparse_page_table = &gpu_page_table_storage[this_as_ref.storage_id * 2 + 1];
 }
 
 /// Bind a channel for execution.
 template <class P>
 void TextureCache<P>::OnGPUASRegister([[maybe_unused]] size_t map_id) {
     gpu_page_table_storage.emplace_back();
+    gpu_page_table_storage.emplace_back();
 }
 
 } // namespace VideoCommon

src/video_core/texture_cache/texture_cache_base.h

@@ -86,6 +86,7 @@ public:
     std::unordered_map<TSCEntry, SamplerId> samplers;
 
     TextureCacheGPUMap* gpu_page_table;
+    TextureCacheGPUMap* sparse_page_table;
 };
 
 template <class P>
@@ -357,7 +358,7 @@ private:
     void ForEachImageInRegionGPU(size_t as_id, GPUVAddr gpu_addr, size_t size, Func&& func);
 
     template <typename Func>
-    void ForEachSparseImageInRegion(GPUVAddr gpu_addr, size_t size, Func&& func);
+    void ForEachSparseImageInRegion(size_t as_id, GPUVAddr gpu_addr, size_t size, Func&& func);
 
     /// Iterates over all the images in a region calling func
     template <typename Func>
@@ -431,7 +432,6 @@ private:
     std::unordered_map<RenderTargets, FramebufferId> framebuffers;
 
     std::unordered_map<u64, std::vector<ImageMapId>, Common::IdentityHash<u64>> page_table;
-    std::unordered_map<u64, std::vector<ImageId>, Common::IdentityHash<u64>> sparse_page_table;
     std::unordered_map<ImageId, boost::container::small_vector<ImageViewId, 16>> sparse_views;
 
     DAddr virtual_invalid_space{};