diff options
| author | bunnei <bunneidev@gmail.com> | 2018-01-02 20:40:30 -0500 | 
|---|---|---|
| committer | bunnei <bunneidev@gmail.com> | 2018-01-02 20:40:30 -0500 | 
| commit | 480906fe1b31a8830aec80fbea04ec941894003f (patch) | |
| tree | cef091503185cde98d5a8a6cd64bda83d6b8ff38 /src/core/hle | |
| parent | 72f671fd7a72d75d6fd697673bee1be18a6dda8b (diff) | |
hle: Move SVC code to kernel namespace.
Diffstat (limited to 'src/core/hle')
| -rw-r--r-- | src/core/hle/kernel/svc.cpp (renamed from src/core/hle/svc.cpp) | 193 | ||||
| -rw-r--r-- | src/core/hle/kernel/svc.h (renamed from src/core/hle/svc.h) | 12 | ||||
| -rw-r--r-- | src/core/hle/kernel/svc_wrap.h (renamed from src/core/hle/function_wrappers.h) | 44 | 
3 files changed, 118 insertions, 131 deletions
| diff --git a/src/core/hle/svc.cpp b/src/core/hle/kernel/svc.cpp index 6ec151d94..3dae8b38b 100644 --- a/src/core/hle/svc.cpp +++ b/src/core/hle/kernel/svc.cpp @@ -5,7 +5,6 @@  #include "common/logging/log.h"  #include "common/microprofile.h"  #include "core/core_timing.h" -#include "core/hle/function_wrappers.h"  #include "core/hle/kernel/client_port.h"  #include "core/hle/kernel/client_session.h"  #include "core/hle/kernel/handle_table.h" @@ -13,27 +12,22 @@  #include "core/hle/kernel/object_address_table.h"  #include "core/hle/kernel/process.h"  #include "core/hle/kernel/resource_limit.h" +#include "core/hle/kernel/svc.h" +#include "core/hle/kernel/svc_wrap.h"  #include "core/hle/kernel/sync_object.h"  #include "core/hle/kernel/thread.h"  #include "core/hle/lock.h"  #include "core/hle/result.h"  #include "core/hle/service/service.h" -//////////////////////////////////////////////////////////////////////////////////////////////////// -// Namespace SVC - -using Kernel::ERR_INVALID_HANDLE; -using Kernel::Handle; -using Kernel::SharedPtr; - -namespace SVC { +namespace Kernel {  /// Set the process heap to a given Size. It can both extend and shrink the heap.  static ResultCode SetHeapSize(VAddr* heap_addr, u64 heap_size) {      LOG_TRACE(Kernel_SVC, "called, heap_size=0x%llx", heap_size); -    auto& process = *Kernel::g_current_process; +    auto& process = *g_current_process;      CASCADE_RESULT(*heap_addr, process.HeapAllocate(Memory::HEAP_VADDR, heap_size, -                                                    Kernel::VMAPermission::ReadWrite)); +                                                    VMAPermission::ReadWrite));      return RESULT_SUCCESS;  } @@ -41,48 +35,48 @@ static ResultCode SetHeapSize(VAddr* heap_addr, u64 heap_size) {  static ResultCode MapMemory(VAddr dst_addr, VAddr src_addr, u64 size) {      LOG_TRACE(Kernel_SVC, "called, dst_addr=0x%llx, src_addr=0x%llx, size=0x%llx", dst_addr,                src_addr, size); -    return Kernel::g_current_process->MirrorMemory(dst_addr, src_addr, size); +    return g_current_process->MirrorMemory(dst_addr, src_addr, size);  }  /// Unmaps a region that was previously mapped with svcMapMemory  static ResultCode UnmapMemory(VAddr dst_addr, VAddr src_addr, u64 size) {      LOG_TRACE(Kernel_SVC, "called, dst_addr=0x%llx, src_addr=0x%llx, size=0x%llx", dst_addr,                src_addr, size); -    return Kernel::g_current_process->UnmapMemory(dst_addr, src_addr, size); +    return g_current_process->UnmapMemory(dst_addr, src_addr, size);  }  /// Connect to an OS service given the port name, returns the handle to the port to out -static ResultCode ConnectToPort(Kernel::Handle* out_handle, VAddr port_name_address) { +static ResultCode ConnectToPort(Handle* out_handle, VAddr port_name_address) {      if (!Memory::IsValidVirtualAddress(port_name_address)) -        return Kernel::ERR_NOT_FOUND; +        return ERR_NOT_FOUND;      static constexpr std::size_t PortNameMaxLength = 11;      // Read 1 char beyond the max allowed port name to detect names that are too long.      std::string port_name = Memory::ReadCString(port_name_address, PortNameMaxLength + 1);      if (port_name.size() > PortNameMaxLength) -        return Kernel::ERR_PORT_NAME_TOO_LONG; +        return ERR_PORT_NAME_TOO_LONG;      LOG_TRACE(Kernel_SVC, "called port_name=%s", port_name.c_str());      auto it = Service::g_kernel_named_ports.find(port_name);      if (it == Service::g_kernel_named_ports.end()) {          LOG_WARNING(Kernel_SVC, "tried to connect to unknown port: %s", port_name.c_str()); -        return Kernel::ERR_NOT_FOUND; +        return ERR_NOT_FOUND;      }      auto client_port = it->second; -    SharedPtr<Kernel::ClientSession> client_session; +    SharedPtr<ClientSession> client_session;      CASCADE_RESULT(client_session, client_port->Connect());      // Return the client session -    CASCADE_RESULT(*out_handle, Kernel::g_handle_table.Create(client_session)); +    CASCADE_RESULT(*out_handle, g_handle_table.Create(client_session));      return RESULT_SUCCESS;  }  /// Makes a blocking IPC call to an OS service. -static ResultCode SendSyncRequest(Kernel::Handle handle) { -    SharedPtr<Kernel::SyncObject> session = Kernel::g_handle_table.Get<Kernel::SyncObject>(handle); +static ResultCode SendSyncRequest(Handle handle) { +    SharedPtr<SyncObject> session = g_handle_table.Get<SyncObject>(handle);      if (!session) {          LOG_ERROR(Kernel_SVC, "called with invalid handle=0x%08X", handle);          return ERR_INVALID_HANDLE; @@ -94,15 +88,15 @@ static ResultCode SendSyncRequest(Kernel::Handle handle) {      // TODO(Subv): svcSendSyncRequest should put the caller thread to sleep while the server      // responds and cause a reschedule. -    return session->SendSyncRequest(Kernel::GetCurrentThread()); +    return session->SendSyncRequest(GetCurrentThread());  }  /// Get the ID for the specified thread. -static ResultCode GetThreadId(u32* thread_id, Kernel::Handle thread_handle) { +static ResultCode GetThreadId(u32* thread_id, Handle thread_handle) {      LOG_TRACE(Kernel_SVC, "called thread=0x%08X", thread_handle); -    const SharedPtr<Kernel::Thread> thread = -        Kernel::g_handle_table.Get<Kernel::Thread>(thread_handle); +    const SharedPtr<Thread> thread = +        g_handle_table.Get<Thread>(thread_handle);      if (!thread) {          return ERR_INVALID_HANDLE;      } @@ -112,11 +106,11 @@ static ResultCode GetThreadId(u32* thread_id, Kernel::Handle thread_handle) {  }  /// Get the ID of the specified process -static ResultCode GetProcessId(u32* process_id, Kernel::Handle process_handle) { +static ResultCode GetProcessId(u32* process_id, Handle process_handle) {      LOG_TRACE(Kernel_SVC, "called process=0x%08X", process_handle); -    const SharedPtr<Kernel::Process> process = -        Kernel::g_handle_table.Get<Kernel::Process>(process_handle); +    const SharedPtr<Process> process = +        g_handle_table.Get<Process>(process_handle);      if (!process) {          return ERR_INVALID_HANDLE;      } @@ -141,18 +135,18 @@ static ResultCode LockMutex(Handle holding_thread_handle, VAddr mutex_addr,                "requesting_current_thread_handle=0x%08X",                holding_thread_handle, mutex_addr, requesting_thread_handle); -    SharedPtr<Kernel::Thread> holding_thread = -        Kernel::g_handle_table.Get<Kernel::Thread>(holding_thread_handle); -    SharedPtr<Kernel::Thread> requesting_thread = -        Kernel::g_handle_table.Get<Kernel::Thread>(requesting_thread_handle); +    SharedPtr<Thread> holding_thread = +        g_handle_table.Get<Thread>(holding_thread_handle); +    SharedPtr<Thread> requesting_thread = +        g_handle_table.Get<Thread>(requesting_thread_handle);      ASSERT(holding_thread);      ASSERT(requesting_thread); -    SharedPtr<Kernel::Mutex> mutex = Kernel::g_object_address_table.Get<Kernel::Mutex>(mutex_addr); +    SharedPtr<Mutex> mutex = g_object_address_table.Get<Mutex>(mutex_addr);      if (!mutex) {          // Create a new mutex for the specified address if one does not already exist -        mutex = Kernel::Mutex::Create(holding_thread, mutex_addr); +        mutex = Mutex::Create(holding_thread, mutex_addr);          mutex->name = Common::StringFromFormat("mutex-%llx", mutex_addr);      } @@ -175,10 +169,10 @@ static ResultCode LockMutex(Handle holding_thread_handle, VAddr mutex_addr,  static ResultCode UnlockMutex(VAddr mutex_addr) {      LOG_TRACE(Kernel_SVC, "called mutex_addr=0x%llx", mutex_addr); -    SharedPtr<Kernel::Mutex> mutex = Kernel::g_object_address_table.Get<Kernel::Mutex>(mutex_addr); +    SharedPtr<Mutex> mutex = g_object_address_table.Get<Mutex>(mutex_addr);      ASSERT(mutex); -    return mutex->Release(Kernel::GetCurrentThread()); +    return mutex->Release(GetCurrentThread());  }  /// Break program execution @@ -199,7 +193,7 @@ static ResultCode GetInfo(u64* result, u64 info_id, u64 handle, u64 info_sub_id)      LOG_TRACE(Kernel_SVC, "called info_id=0x%X, info_sub_id=0x%X, handle=0x%08X", info_id,                info_sub_id, handle); -    auto& vm_manager = Kernel::g_current_process->vm_manager; +    auto& vm_manager = g_current_process->vm_manager;      switch (static_cast<GetInfoType>(info_id)) {      case GetInfoType::TotalMemoryUsage:          *result = vm_manager.GetTotalMemoryUsage(); @@ -231,7 +225,7 @@ static ResultCode GetInfo(u64* result, u64 info_id, u64 handle, u64 info_sub_id)  /// Gets the priority for the specified thread  static ResultCode GetThreadPriority(u32* priority, Handle handle) { -    const SharedPtr<Kernel::Thread> thread = Kernel::g_handle_table.Get<Kernel::Thread>(handle); +    const SharedPtr<Thread> thread = g_handle_table.Get<Thread>(handle);      if (!thread)          return ERR_INVALID_HANDLE; @@ -242,18 +236,18 @@ static ResultCode GetThreadPriority(u32* priority, Handle handle) {  /// Sets the priority for the specified thread  static ResultCode SetThreadPriority(Handle handle, u32 priority) {      if (priority > THREADPRIO_LOWEST) { -        return Kernel::ERR_OUT_OF_RANGE; +        return ERR_OUT_OF_RANGE;      } -    SharedPtr<Kernel::Thread> thread = Kernel::g_handle_table.Get<Kernel::Thread>(handle); +    SharedPtr<Thread> thread = g_handle_table.Get<Thread>(handle);      if (!thread) -        return Kernel::ERR_INVALID_HANDLE; +        return ERR_INVALID_HANDLE;      // Note: The kernel uses the current process's resource limit instead of      // the one from the thread owner's resource limit. -    SharedPtr<Kernel::ResourceLimit>& resource_limit = Kernel::g_current_process->resource_limit; -    if (resource_limit->GetMaxResourceValue(Kernel::ResourceTypes::PRIORITY) > priority) { -        return Kernel::ERR_NOT_AUTHORIZED; +    SharedPtr<ResourceLimit>& resource_limit = g_current_process->resource_limit; +    if (resource_limit->GetMaxResourceValue(ResourceTypes::PRIORITY) > priority) { +        return ERR_NOT_AUTHORIZED;      }      thread->SetPriority(priority); @@ -275,19 +269,18 @@ static u32 GetCurrentProcessorNumber() {  /// Query process memory  static ResultCode QueryProcessMemory(MemoryInfo* memory_info, PageInfo* /*page_info*/, -                                     Kernel::Handle process_handle, u64 addr) { -    using Kernel::Process; -    Kernel::SharedPtr<Process> process = Kernel::g_handle_table.Get<Process>(process_handle); +                                     Handle process_handle, u64 addr) { +    SharedPtr<Process> process = g_handle_table.Get<Process>(process_handle);      if (!process) {          return ERR_INVALID_HANDLE;      }      auto vma = process->vm_manager.FindVMA(addr);      memory_info->attributes = 0; -    if (vma == Kernel::g_current_process->vm_manager.vma_map.end()) { +    if (vma == g_current_process->vm_manager.vma_map.end()) {          memory_info->base_address = 0; -        memory_info->permission = static_cast<u32>(Kernel::VMAPermission::None); +        memory_info->permission = static_cast<u32>(VMAPermission::None);          memory_info->size = 0; -        memory_info->type = static_cast<u32>(Kernel::MemoryState::Free); +        memory_info->type = static_cast<u32>(MemoryState::Free);      } else {          memory_info->base_address = vma->second.base;          memory_info->permission = static_cast<u32>(vma->second.permissions); @@ -302,25 +295,25 @@ static ResultCode QueryProcessMemory(MemoryInfo* memory_info, PageInfo* /*page_i  /// Query memory  static ResultCode QueryMemory(MemoryInfo* memory_info, PageInfo* page_info, VAddr addr) {      LOG_TRACE(Kernel_SVC, "called, addr=%llx", addr); -    return QueryProcessMemory(memory_info, page_info, Kernel::CurrentProcess, addr); +    return QueryProcessMemory(memory_info, page_info, CurrentProcess, addr);  }  /// Exits the current process  static void ExitProcess() { -    LOG_INFO(Kernel_SVC, "Process %u exiting", Kernel::g_current_process->process_id); +    LOG_INFO(Kernel_SVC, "Process %u exiting", g_current_process->process_id); -    ASSERT_MSG(Kernel::g_current_process->status == Kernel::ProcessStatus::Running, +    ASSERT_MSG(g_current_process->status == ProcessStatus::Running,                 "Process has already exited"); -    Kernel::g_current_process->status = Kernel::ProcessStatus::Exited; +    g_current_process->status = ProcessStatus::Exited;      // Stop all the process threads that are currently waiting for objects. -    auto& thread_list = Kernel::GetThreadList(); +    auto& thread_list = GetThreadList();      for (auto& thread : thread_list) { -        if (thread->owner_process != Kernel::g_current_process) +        if (thread->owner_process != g_current_process)              continue; -        if (thread == Kernel::GetCurrentThread()) +        if (thread == GetCurrentThread())              continue;          // TODO(Subv): When are the other running/ready threads terminated? @@ -332,7 +325,7 @@ static void ExitProcess() {      }      // Kill the current thread -    Kernel::GetCurrentThread()->Stop(); +    GetCurrentThread()->Stop();      Core::System::GetInstance().PrepareReschedule();  } @@ -343,17 +336,17 @@ static ResultCode CreateThread(Handle* out_handle, VAddr entry_point, u64 arg, V      std::string name = Common::StringFromFormat("unknown-%llx", entry_point);      if (priority > THREADPRIO_LOWEST) { -        return Kernel::ERR_OUT_OF_RANGE; +        return ERR_OUT_OF_RANGE;      } -    SharedPtr<Kernel::ResourceLimit>& resource_limit = Kernel::g_current_process->resource_limit; -    if (resource_limit->GetMaxResourceValue(Kernel::ResourceTypes::PRIORITY) > priority) { -        return Kernel::ERR_NOT_AUTHORIZED; +    SharedPtr<ResourceLimit>& resource_limit = g_current_process->resource_limit; +    if (resource_limit->GetMaxResourceValue(ResourceTypes::PRIORITY) > priority) { +        return ERR_NOT_AUTHORIZED;      }      if (processor_id == THREADPROCESSORID_DEFAULT) {          // Set the target CPU to the one specified in the process' exheader. -        processor_id = Kernel::g_current_process->ideal_processor; +        processor_id = g_current_process->ideal_processor;          ASSERT(processor_id != THREADPROCESSORID_DEFAULT);      } @@ -374,14 +367,14 @@ static ResultCode CreateThread(Handle* out_handle, VAddr entry_point, u64 arg, V          break;      } -    CASCADE_RESULT(SharedPtr<Kernel::Thread> thread, -                   Kernel::Thread::Create(name, entry_point, priority, arg, processor_id, stack_top, -                                          Kernel::g_current_process)); +    CASCADE_RESULT(SharedPtr<Thread> thread, +                   Thread::Create(name, entry_point, priority, arg, processor_id, stack_top, +                                          g_current_process));      thread->context.fpscr =          FPSCR_DEFAULT_NAN | FPSCR_FLUSH_TO_ZERO | FPSCR_ROUND_TOZERO; // 0x03C00000 -    CASCADE_RESULT(thread->guest_handle, Kernel::g_handle_table.Create(thread)); +    CASCADE_RESULT(thread->guest_handle, g_handle_table.Create(thread));      *out_handle = thread->guest_handle;      Core::System::GetInstance().PrepareReschedule(); @@ -398,8 +391,8 @@ static ResultCode CreateThread(Handle* out_handle, VAddr entry_point, u64 arg, V  static ResultCode StartThread(Handle thread_handle) {      LOG_TRACE(Kernel_SVC, "called thread=0x%08X", thread_handle); -    const SharedPtr<Kernel::Thread> thread = -        Kernel::g_handle_table.Get<Kernel::Thread>(thread_handle); +    const SharedPtr<Thread> thread = +        g_handle_table.Get<Thread>(thread_handle);      if (!thread) {          return ERR_INVALID_HANDLE;      } @@ -413,7 +406,7 @@ static ResultCode StartThread(Handle thread_handle) {  static void ExitThread() {      LOG_TRACE(Kernel_SVC, "called, pc=0x%08X", Core::CPU().GetPC()); -    Kernel::ExitCurrentThread(); +    ExitCurrentThread();      Core::System::GetInstance().PrepareReschedule();  } @@ -423,14 +416,14 @@ static void SleepThread(s64 nanoseconds) {      // Don't attempt to yield execution if there are no available threads to run,      // this way we avoid a useless reschedule to the idle thread. -    if (nanoseconds == 0 && !Kernel::HaveReadyThreads()) +    if (nanoseconds == 0 && !HaveReadyThreads())          return;      // Sleep current thread and check for next thread to schedule -    Kernel::WaitCurrentThread_Sleep(); +    WaitCurrentThread_Sleep();      // Create an event to wake the thread up after the specified nanosecond delay has passed -    Kernel::GetCurrentThread()->WakeAfterDelay(nanoseconds); +    GetCurrentThread()->WakeAfterDelay(nanoseconds);      Core::System::GetInstance().PrepareReschedule();  } @@ -442,9 +435,9 @@ static ResultCode SignalProcessWideKey(VAddr addr, u32 target) {  }  /// Close a handle -static ResultCode CloseHandle(Kernel::Handle handle) { +static ResultCode CloseHandle(Handle handle) {      LOG_TRACE(Kernel_SVC, "Closing handle 0x%08X", handle); -    return Kernel::g_handle_table.Close(handle); +    return g_handle_table.Close(handle);  }  namespace { @@ -459,47 +452,47 @@ struct FunctionDef {  static const FunctionDef SVC_Table[] = {      {0x00, nullptr, "Unknown"}, -    {0x01, HLE::Wrap<SetHeapSize>, "svcSetHeapSize"}, +    {0x01, SvcWrap<SetHeapSize>, "svcSetHeapSize"},      {0x02, nullptr, "svcSetMemoryPermission"},      {0x03, nullptr, "svcSetMemoryAttribute"}, -    {0x04, HLE::Wrap<MapMemory>, "svcMapMemory"}, -    {0x05, HLE::Wrap<UnmapMemory>, "svcUnmapMemory"}, -    {0x06, HLE::Wrap<QueryMemory>, "svcQueryMemory"}, -    {0x07, HLE::Wrap<ExitProcess>, "svcExitProcess"}, -    {0x08, HLE::Wrap<CreateThread>, "svcCreateThread"}, -    {0x09, HLE::Wrap<StartThread>, "svcStartThread"}, -    {0x0A, HLE::Wrap<ExitThread>, "svcExitThread"}, -    {0x0B, HLE::Wrap<SleepThread>, "svcSleepThread"}, -    {0x0C, HLE::Wrap<GetThreadPriority>, "svcGetThreadPriority"}, -    {0x0D, HLE::Wrap<SetThreadPriority>, "svcSetThreadPriority"}, +    {0x04, SvcWrap<MapMemory>, "svcMapMemory"}, +    {0x05, SvcWrap<UnmapMemory>, "svcUnmapMemory"}, +    {0x06, SvcWrap<QueryMemory>, "svcQueryMemory"}, +    {0x07, SvcWrap<ExitProcess>, "svcExitProcess"}, +    {0x08, SvcWrap<CreateThread>, "svcCreateThread"}, +    {0x09, SvcWrap<StartThread>, "svcStartThread"}, +    {0x0A, SvcWrap<ExitThread>, "svcExitThread"}, +    {0x0B, SvcWrap<SleepThread>, "svcSleepThread"}, +    {0x0C, SvcWrap<GetThreadPriority>, "svcGetThreadPriority"}, +    {0x0D, SvcWrap<SetThreadPriority>, "svcSetThreadPriority"},      {0x0E, nullptr, "svcGetThreadCoreMask"},      {0x0F, nullptr, "svcSetThreadCoreMask"}, -    {0x10, HLE::Wrap<GetCurrentProcessorNumber>, "svcGetCurrentProcessorNumber"}, +    {0x10, SvcWrap<GetCurrentProcessorNumber>, "svcGetCurrentProcessorNumber"},      {0x11, nullptr, "svcSignalEvent"},      {0x12, nullptr, "svcClearEvent"},      {0x13, nullptr, "svcMapSharedMemory"},      {0x14, nullptr, "svcUnmapSharedMemory"},      {0x15, nullptr, "svcCreateTransferMemory"}, -    {0x16, HLE::Wrap<CloseHandle>, "svcCloseHandle"}, +    {0x16, SvcWrap<CloseHandle>, "svcCloseHandle"},      {0x17, nullptr, "svcResetSignal"}, -    {0x18, HLE::Wrap<WaitSynchronization>, "svcWaitSynchronization"}, +    {0x18, SvcWrap<WaitSynchronization>, "svcWaitSynchronization"},      {0x19, nullptr, "svcCancelSynchronization"}, -    {0x1A, HLE::Wrap<LockMutex>, "svcLockMutex"}, -    {0x1B, HLE::Wrap<UnlockMutex>, "svcUnlockMutex"}, +    {0x1A, SvcWrap<LockMutex>, "svcLockMutex"}, +    {0x1B, SvcWrap<UnlockMutex>, "svcUnlockMutex"},      {0x1C, nullptr, "svcWaitProcessWideKeyAtomic"}, -    {0x1D, HLE::Wrap<SignalProcessWideKey>, "svcSignalProcessWideKey"}, +    {0x1D, SvcWrap<SignalProcessWideKey>, "svcSignalProcessWideKey"},      {0x1E, nullptr, "svcGetSystemTick"}, -    {0x1F, HLE::Wrap<ConnectToPort>, "svcConnectToPort"}, +    {0x1F, SvcWrap<ConnectToPort>, "svcConnectToPort"},      {0x20, nullptr, "svcSendSyncRequestLight"}, -    {0x21, HLE::Wrap<SendSyncRequest>, "svcSendSyncRequest"}, +    {0x21, SvcWrap<SendSyncRequest>, "svcSendSyncRequest"},      {0x22, nullptr, "svcSendSyncRequestWithUserBuffer"},      {0x23, nullptr, "svcSendAsyncRequestWithUserBuffer"}, -    {0x24, HLE::Wrap<GetProcessId>, "svcGetProcessId"}, -    {0x25, HLE::Wrap<GetThreadId>, "svcGetThreadId"}, -    {0x26, HLE::Wrap<Break>, "svcBreak"}, -    {0x27, HLE::Wrap<OutputDebugString>, "svcOutputDebugString"}, +    {0x24, SvcWrap<GetProcessId>, "svcGetProcessId"}, +    {0x25, SvcWrap<GetThreadId>, "svcGetThreadId"}, +    {0x26, SvcWrap<Break>, "svcBreak"}, +    {0x27, SvcWrap<OutputDebugString>, "svcOutputDebugString"},      {0x28, nullptr, "svcReturnFromException"}, -    {0x29, HLE::Wrap<GetInfo>, "svcGetInfo"}, +    {0x29, SvcWrap<GetInfo>, "svcGetInfo"},      {0x2A, nullptr, "svcFlushEntireDataCache"},      {0x2B, nullptr, "svcFlushDataCache"},      {0x2C, nullptr, "svcMapPhysicalMemory"}, @@ -616,4 +609,4 @@ void CallSVC(u32 immediate) {      }  } -} // namespace SVC +} // namespace Kernel diff --git a/src/core/hle/svc.h b/src/core/hle/kernel/svc.h index fd001a38e..b0265b6c8 100644 --- a/src/core/hle/svc.h +++ b/src/core/hle/kernel/svc.h @@ -1,4 +1,4 @@ -// Copyright 2014 Citra Emulator Project +// Copyright 2018 Citra Emulator Project  // Licensed under GPLv2 or any later version  // Refer to the license.txt file included. @@ -6,8 +6,7 @@  #include "common/common_types.h" -//////////////////////////////////////////////////////////////////////////////////////////////////// -// SVC types +namespace Kernel {  struct MemoryInfo {      u64 base_address; @@ -21,11 +20,6 @@ struct PageInfo {      u64 flags;  }; -//////////////////////////////////////////////////////////////////////////////////////////////////// -// Namespace SVC - -namespace SVC { -  /// Values accepted by svcGetInfo  enum class GetInfoType : u64 {      // 1.0.0+ @@ -41,4 +35,4 @@ enum class GetInfoType : u64 {  void CallSVC(u32 immediate); -} // namespace SVC +} // namespace Kernel diff --git a/src/core/hle/function_wrappers.h b/src/core/hle/kernel/svc_wrap.h index df2c3e843..65b64cb3c 100644 --- a/src/core/hle/function_wrappers.h +++ b/src/core/hle/kernel/svc_wrap.h @@ -1,4 +1,4 @@ -// Copyright 2017 Citra Emulator Project +// Copyright 2018 Citra Emulator Project  // Licensed under GPLv2 or any later version  // Refer to the license.txt file included. @@ -7,11 +7,11 @@  #include "common/common_types.h"  #include "core/arm/arm_interface.h"  #include "core/core.h" +#include "core/hle/kernel/svc.h"  #include "core/hle/result.h" -#include "core/hle/svc.h"  #include "core/memory.h" -namespace HLE { +namespace Kernel {  #define PARAM(n) Core::CPU().GetReg(n) @@ -27,22 +27,22 @@ static inline void FuncReturn(u64 res) {  // Function wrappers that return type ResultCode  template <ResultCode func(u64)> -void Wrap() { +void SvcWrap() {      FuncReturn(func(PARAM(0)).raw);  }  template <ResultCode func(u32)> -void Wrap() { +void SvcWrap() {      FuncReturn(func((u32)PARAM(0)).raw);  }  template <ResultCode func(u32, u32)> -void Wrap() { +void SvcWrap() {      FuncReturn(func((u32)PARAM(0), (u32)PARAM(1)).raw);  }  template <ResultCode func(u32*, u32)> -void Wrap() { +void SvcWrap() {      u32 param_1 = 0;      u32 retval = func(¶m_1, (u32)PARAM(1)).raw;      Core::CPU().SetReg(1, param_1); @@ -50,7 +50,7 @@ void Wrap() {  }  template <ResultCode func(u32*, u64)> -void Wrap() { +void SvcWrap() {      u32 param_1 = 0;      u32 retval = func(¶m_1, PARAM(1)).raw;      Core::CPU().SetReg(1, param_1); @@ -58,12 +58,12 @@ void Wrap() {  }  template <ResultCode func(u64, u32)> -void Wrap() { +void SvcWrap() {      FuncReturn(func(PARAM(0), (u32)PARAM(1)).raw);  }  template <ResultCode func(u64*, u64)> -void Wrap() { +void SvcWrap() {      u64 param_1 = 0;      u32 retval = func(¶m_1, PARAM(1)).raw;      Core::CPU().SetReg(1, param_1); @@ -71,22 +71,22 @@ void Wrap() {  }  template <ResultCode func(u32, u64, u32)> -void Wrap() { +void SvcWrap() {      FuncReturn(func((u32)PARAM(0), PARAM(1), (u32)PARAM(2)).raw);  }  template <ResultCode func(u64, u64, u64)> -void Wrap() { +void SvcWrap() {      FuncReturn(func(PARAM(0), PARAM(1), PARAM(2)).raw);  }  template <ResultCode func(u64, u64, s64)> -void Wrap() { +void SvcWrap() {      FuncReturn(func(PARAM(1), PARAM(2), (s64)PARAM(3)).raw);  }  template <ResultCode func(u64*, u64, u64, u64)> -void Wrap() { +void SvcWrap() {      u64 param_1 = 0;      u32 retval = func(¶m_1, PARAM(1), PARAM(2), PARAM(3)).raw;      Core::CPU().SetReg(1, param_1); @@ -94,7 +94,7 @@ void Wrap() {  }  template <ResultCode func(u32*, u64, u64, u64, u32, s32)> -void Wrap() { +void SvcWrap() {      u32 param_1 = 0;      u32 retval =          func(¶m_1, PARAM(1), PARAM(2), PARAM(3), (u32)PARAM(4), (s32)(PARAM(5) & 0xFFFFFFFF)) @@ -104,7 +104,7 @@ void Wrap() {  }  template <ResultCode func(MemoryInfo*, PageInfo*, u64)> -void Wrap() { +void SvcWrap() {      MemoryInfo memory_info = {};      PageInfo page_info = {};      u32 retval = func(&memory_info, &page_info, PARAM(2)).raw; @@ -122,7 +122,7 @@ void Wrap() {  // Function wrappers that return type u32  template <u32 func()> -void Wrap() { +void SvcWrap() {      FuncReturn(func());  } @@ -130,26 +130,26 @@ void Wrap() {  /// Function wrappers that return type void  template <void func()> -void Wrap() { +void SvcWrap() {      func();  }  template <void func(s64)> -void Wrap() { +void SvcWrap() {      func((s64)PARAM(0));  }  template <void func(u64, s32 len)> -void Wrap() { +void SvcWrap() {      func(PARAM(0), (s32)(PARAM(1) & 0xFFFFFFFF));  }  template <void func(u64, u64, u64)> -void Wrap() { +void SvcWrap() {      func(PARAM(0), PARAM(1), PARAM(2));  }  #undef PARAM  #undef FuncReturn -} // namespace HLE +} // namespace Kernel | 
