Merge pull request #4405 from wwylele/svc-global

SVC: new wrapper template & pass system reference across the SVC barrier

src/core/CMakeLists.txt

@@ -103,7 +103,6 @@ add_library(core STATIC
     hle/applets/mint.h
     hle/applets/swkbd.cpp
     hle/applets/swkbd.h
-    hle/function_wrappers.h
     hle/ipc.h
     hle/ipc_helpers.h
     hle/kernel/address_arbiter.cpp
@@ -148,6 +147,7 @@ add_library(core STATIC
     hle/kernel/shared_page.h
     hle/kernel/svc.cpp
     hle/kernel/svc.h
+    hle/kernel/svc_wrapper.h
     hle/kernel/thread.cpp
     hle/kernel/thread.h
     hle/kernel/timer.cpp

src/core/arm/dynarmic/arm_dynarmic.cpp

@@ -72,7 +72,7 @@ private:
 class DynarmicUserCallbacks final : public Dynarmic::A32::UserCallbacks {
 public:
     explicit DynarmicUserCallbacks(ARM_Dynarmic& parent)
-        : parent(parent), timing(parent.system.CoreTiming()) {}
+        : parent(parent), timing(parent.system.CoreTiming()), svc_context(parent.system) {}
     ~DynarmicUserCallbacks() = default;
 
     std::uint8_t MemoryRead8(VAddr vaddr) override {
@@ -123,7 +123,7 @@ public:
     }
 
     void CallSVC(std::uint32_t swi) override {
-        Kernel::CallSVC(swi);
+        svc_context.CallSVC(swi);
     }
 
     void ExceptionRaised(VAddr pc, Dynarmic::A32::Exception exception) override {
@@ -158,6 +158,7 @@ public:
 
     ARM_Dynarmic& parent;
     Core::Timing& timing;
+    Kernel::SVCContext svc_context;
 };
 
 ARM_Dynarmic::ARM_Dynarmic(Core::System& system, PrivilegeMode initial_mode)

src/core/arm/dyncom/arm_dyncom_interpreter.cpp

@@ -3864,7 +3864,7 @@ SWI_INST : {
         cpu->NumInstrsToExecute =
             num_instrs >= cpu->NumInstrsToExecute ? 0 : cpu->NumInstrsToExecute - num_instrs;
         num_instrs = 0;
-        Kernel::CallSVC(inst_cream->num & 0xFFFF);
+        Kernel::SVCContext{Core::System::GetInstance()}.CallSVC(inst_cream->num & 0xFFFF);
         // The kernel would call ERET to get here, which clears exclusive memory state.
         cpu->UnsetExclusiveMemoryAddress();
     }

src/core/core.cpp

@@ -172,6 +172,8 @@ System::ResultStatus System::Init(EmuWindow& emu_window, u32 system_mode) {
 
     timing = std::make_unique<Timing>();
 
+    kernel = std::make_unique<Kernel::KernelSystem>(system_mode);
+
     if (Settings::values.use_cpu_jit) {
 #ifdef ARCHITECTURE_x86_64
         cpu_core = std::make_unique<ARM_Dynarmic>(*this, USER32MODE);
@@ -197,7 +199,6 @@ System::ResultStatus System::Init(EmuWindow& emu_window, u32 system_mode) {
     archive_manager = std::make_unique<Service::FS::ArchiveManager>(*this);
 
     HW::Init();
-    kernel = std::make_unique<Kernel::KernelSystem>(system_mode);
     Service::Init(*this);
     GDBStub::Init();
 

src/core/hle/function_wrappers.h

@@ -1,265 +0,0 @@
-// Copyright 2014 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include "common/common_types.h"
-#include "core/arm/arm_interface.h"
-#include "core/core.h"
-#include "core/hle/kernel/kernel.h"
-#include "core/hle/kernel/svc.h"
-#include "core/hle/result.h"
-#include "core/memory.h"
-
-namespace HLE {
-
-static inline u32 Param(int n) {
-    return Core::CPU().GetReg(n);
-}
-
-/**
- * HLE a function return from the current ARM11 userland process
- * @param res Result to return
- */
-static inline void FuncReturn(u32 res) {
-    Core::CPU().SetReg(0, res);
-}
-
-/**
- * HLE a function return (64-bit) from the current ARM11 userland process
- * @param res Result to return (64-bit)
- * @todo Verify that this function is correct
- */
-static inline void FuncReturn64(u64 res) {
-    Core::CPU().SetReg(0, (u32)(res & 0xFFFFFFFF));
-    Core::CPU().SetReg(1, (u32)((res >> 32) & 0xFFFFFFFF));
-}
-
-////////////////////////////////////////////////////////////////////////////////////////////////////
-// Function wrappers that return type ResultCode
-
-template <ResultCode func(u32, u32, u32, u32)>
-void Wrap() {
-    FuncReturn(func(Param(0), Param(1), Param(2), Param(3)).raw);
-}
-
-template <ResultCode func(u32*, u32, u32, u32, u32, u32)>
-void Wrap() {
-    u32 param_1 = 0;
-    u32 retval = func(&param_1, Param(0), Param(1), Param(2), Param(3), Param(4)).raw;
-    Core::CPU().SetReg(1, param_1);
-    FuncReturn(retval);
-}
-
-template <ResultCode func(u32*, u32, u32, u32, u32, s32)>
-void Wrap() {
-    u32 param_1 = 0;
-    u32 retval = func(&param_1, Param(0), Param(1), Param(2), Param(3), Param(4)).raw;
-    Core::CPU().SetReg(1, param_1);
-    FuncReturn(retval);
-}
-
-template <ResultCode func(s32*, VAddr, s32, bool, s64)>
-void Wrap() {
-    s32 param_1 = 0;
-    s32 retval =
-        func(&param_1, Param(1), (s32)Param(2), (Param(3) != 0), (((s64)Param(4) << 32) | Param(0)))
-            .raw;
-
-    Core::CPU().SetReg(1, (u32)param_1);
-    FuncReturn(retval);
-}
-
-template <ResultCode func(s32*, VAddr, s32, u32)>
-void Wrap() {
-    s32 param_1 = 0;
-    u32 retval = func(&param_1, Param(1), (s32)Param(2), Param(3)).raw;
-
-    Core::CPU().SetReg(1, (u32)param_1);
-    FuncReturn(retval);
-}
-
-template <ResultCode func(u32, u32, u32, u32, s64)>
-void Wrap() {
-    FuncReturn(
-        func(Param(0), Param(1), Param(2), Param(3), (((s64)Param(5) << 32) | Param(4))).raw);
-}
-
-template <ResultCode func(u32*)>
-void Wrap() {
-    u32 param_1 = 0;
-    u32 retval = func(&param_1).raw;
-    Core::CPU().SetReg(1, param_1);
-    FuncReturn(retval);
-}
-
-template <ResultCode func(u32, s64)>
-void Wrap() {
-    s32 retval = func(Param(0), (((s64)Param(3) << 32) | Param(2))).raw;
-
-    FuncReturn(retval);
-}
-
-template <ResultCode func(Kernel::MemoryInfo*, Kernel::PageInfo*, u32)>
-void Wrap() {
-    Kernel::MemoryInfo memory_info = {};
-    Kernel::PageInfo page_info = {};
-    u32 retval = func(&memory_info, &page_info, Param(2)).raw;
-    Core::CPU().SetReg(1, memory_info.base_address);
-    Core::CPU().SetReg(2, memory_info.size);
-    Core::CPU().SetReg(3, memory_info.permission);
-    Core::CPU().SetReg(4, memory_info.state);
-    Core::CPU().SetReg(5, page_info.flags);
-    FuncReturn(retval);
-}
-
-template <ResultCode func(Kernel::MemoryInfo*, Kernel::PageInfo*, Kernel::Handle, u32)>
-void Wrap() {
-    Kernel::MemoryInfo memory_info = {};
-    Kernel::PageInfo page_info = {};
-    u32 retval = func(&memory_info, &page_info, Param(2), Param(3)).raw;
-    Core::CPU().SetReg(1, memory_info.base_address);
-    Core::CPU().SetReg(2, memory_info.size);
-    Core::CPU().SetReg(3, memory_info.permission);
-    Core::CPU().SetReg(4, memory_info.state);
-    Core::CPU().SetReg(5, page_info.flags);
-    FuncReturn(retval);
-}
-
-template <ResultCode func(s32*, u32)>
-void Wrap() {
-    s32 param_1 = 0;
-    u32 retval = func(&param_1, Param(1)).raw;
-    Core::CPU().SetReg(1, param_1);
-    FuncReturn(retval);
-}
-
-template <ResultCode func(u32, s32)>
-void Wrap() {
-    FuncReturn(func(Param(0), (s32)Param(1)).raw);
-}
-
-template <ResultCode func(u32*, u32)>
-void Wrap() {
-    u32 param_1 = 0;
-    u32 retval = func(&param_1, Param(1)).raw;
-    Core::CPU().SetReg(1, param_1);
-    FuncReturn(retval);
-}
-
-template <ResultCode func(u32)>
-void Wrap() {
-    FuncReturn(func(Param(0)).raw);
-}
-
-template <ResultCode func(u32*, s32, s32)>
-void Wrap() {
-    u32 param_1 = 0;
-    u32 retval = func(&param_1, Param(1), Param(2)).raw;
-    Core::CPU().SetReg(1, param_1);
-    FuncReturn(retval);
-}
-
-template <ResultCode func(s32*, u32, s32)>
-void Wrap() {
-    s32 param_1 = 0;
-    u32 retval = func(&param_1, Param(1), Param(2)).raw;
-    Core::CPU().SetReg(1, param_1);
-    FuncReturn(retval);
-}
-
-template <ResultCode func(s64*, u32, s32)>
-void Wrap() {
-    s64 param_1 = 0;
-    u32 retval = func(&param_1, Param(1), Param(2)).raw;
-    Core::CPU().SetReg(1, (u32)param_1);
-    Core::CPU().SetReg(2, (u32)(param_1 >> 32));
-    FuncReturn(retval);
-}
-
-template <ResultCode func(u32*, u32, u32, u32, u32)>
-void Wrap() {
-    u32 param_1 = 0;
-    // The last parameter is passed in R0 instead of R4
-    u32 retval = func(&param_1, Param(1), Param(2), Param(3), Param(0)).raw;
-    Core::CPU().SetReg(1, param_1);
-    FuncReturn(retval);
-}
-
-template <ResultCode func(u32, s64, s64)>
-void Wrap() {
-    s64 param1 = ((u64)Param(3) << 32) | Param(2);
-    s64 param2 = ((u64)Param(4) << 32) | Param(1);
-    FuncReturn(func(Param(0), param1, param2).raw);
-}
-
-template <ResultCode func(s64*, Kernel::Handle, u32)>
-void Wrap() {
-    s64 param_1 = 0;
-    u32 retval = func(&param_1, Param(1), Param(2)).raw;
-    Core::CPU().SetReg(1, (u32)param_1);
-    Core::CPU().SetReg(2, (u32)(param_1 >> 32));
-    FuncReturn(retval);
-}
-
-template <ResultCode func(Kernel::Handle, u32)>
-void Wrap() {
-    FuncReturn(func(Param(0), Param(1)).raw);
-}
-
-template <ResultCode func(Kernel::Handle*, Kernel::Handle*, VAddr, u32)>
-void Wrap() {
-    Kernel::Handle param_1 = 0;
-    Kernel::Handle param_2 = 0;
-    u32 retval = func(&param_1, &param_2, Param(2), Param(3)).raw;
-    Core::CPU().SetReg(1, param_1);
-    Core::CPU().SetReg(2, param_2);
-    FuncReturn(retval);
-}
-
-template <ResultCode func(Kernel::Handle*, Kernel::Handle*)>
-void Wrap() {
-    Kernel::Handle param_1 = 0;
-    Kernel::Handle param_2 = 0;
-    u32 retval = func(&param_1, &param_2).raw;
-    Core::CPU().SetReg(1, param_1);
-    Core::CPU().SetReg(2, param_2);
-    FuncReturn(retval);
-}
-
-////////////////////////////////////////////////////////////////////////////////////////////////////
-// Function wrappers that return type u32
-
-template <u32 func()>
-void Wrap() {
-    FuncReturn(func());
-}
-
-////////////////////////////////////////////////////////////////////////////////////////////////////
-// Function wrappers that return type s64
-
-template <s64 func()>
-void Wrap() {
-    FuncReturn64(func());
-}
-
-////////////////////////////////////////////////////////////////////////////////////////////////////
-/// Function wrappers that return type void
-
-template <void func(s64)>
-void Wrap() {
-    func(((s64)Param(1) << 32) | Param(0));
-}
-
-template <void func(VAddr, int len)>
-void Wrap() {
-    func(Param(0), Param(1));
-}
-
-template <void func(u8)>
-void Wrap() {
-    func((u8)Param(0));
-}
-
-} // namespace HLE

src/core/hle/kernel/svc.h

@@ -4,50 +4,25 @@
 
 #pragma once
 
+#include <memory>
 #include "common/common_types.h"
 
+namespace Core {
+class System;
+} // namespace Core
+
 namespace Kernel {
 
-struct MemoryInfo {
+class SVC;
-    u32 base_address;
-    u32 size;
-    u32 permission;
-    u32 state;
-};
-
-struct PageInfo {
-    u32 flags;
-};
-
-/// Values accepted by svcGetSystemInfo's type parameter.
-enum class SystemInfoType {
-    /**
-     * Reports total used memory for all regions or a specific one, according to the extra
-     * parameter. See `SystemInfoMemUsageRegion`.
-     */
-    REGION_MEMORY_USAGE = 0,
-    /**
-     * Returns the memory usage for certain allocations done internally by the kernel.
-     */
-    KERNEL_ALLOCATED_PAGES = 2,
-    /**
-     * "This returns the total number of processes which were launched directly by the kernel.
-     * For the ARM11 NATIVE_FIRM kernel, this is 5, for processes sm, fs, pm, loader, and pxi."
-     */
-    KERNEL_SPAWNED_PIDS = 26,
-};
-
-/**
- * Accepted by svcGetSystemInfo param with REGION_MEMORY_USAGE type. Selects a region to query
- * memory usage of.
- */
-enum class SystemInfoMemUsageRegion {
-    ALL = 0,
-    APPLICATION = 1,
-    SYSTEM = 2,
-    BASE = 3,
-};
 
+class SVCContext {
+public:
+    SVCContext(Core::System& system);
+    ~SVCContext();
     void CallSVC(u32 immediate);
 
+private:
+    std::unique_ptr<SVC> impl;
+};
+
 } // namespace Kernel

src/core/hle/kernel/svc_wrapper.h

@@ -0,0 +1,294 @@
+// Copyright 2014 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include <cstring>
+#include <type_traits>
+#include "common/common_types.h"
+
+namespace Kernel {
+
+/*
+ * This class defines the SVC ABI, and provides a wrapper template for translating between ARM
+ * registers and SVC parameters and return value. The SVC context class should inherit from this
+ * class using CRTP (`class SVC : public SVCWrapper<SVC> {...}`), and use the Wrap() template to
+ * convert a SVC function interface to a void()-type function that interacts with registers
+ * interface GetReg() and SetReg().
+ */
+template <typename Context>
+class SVCWrapper {
+protected:
+    template <auto F>
+    void Wrap() {
+        WrapHelper<decltype(F)>::Call(*static_cast<Context*>(this), F);
+    };
+
+private:
+    // A special param index represents the return value
+    static constexpr std::size_t INDEX_RETURN = ~(std::size_t)0;
+
+    struct ParamSlot {
+        // whether the slot is used for a param
+        bool used;
+
+        // index of a param in the function signature, starting from 0, or special value
+        // INDEX_RETURN
+        std::size_t param_index;
+
+        // index of a 32-bit word within the param
+        std::size_t word_index;
+    };
+
+    // Size in words of the given type in ARM ABI
+    template <typename T>
+    static constexpr std::size_t WordSize() {
+        static_assert(std::is_trivially_copyable_v<T>);
+        if constexpr (std::is_pointer_v<T>) {
+            return 1;
+        } else if constexpr (std::is_same_v<T, bool>) {
+            return 1;
+        } else {
+            return (sizeof(T) - 1) / 4 + 1;
+        }
+    }
+
+    // Size in words of the given type in ARM ABI with pointer removed
+    template <typename T>
+    static constexpr std::size_t OutputWordSize() {
+        if constexpr (std::is_pointer_v<T>) {
+            return WordSize<std::remove_pointer_t<T>>();
+        } else {
+            return 0;
+        }
+    }
+
+    // Describes the register assignments of a SVC
+    struct SVCABI {
+        static constexpr std::size_t RegCount = 8;
+
+        // Register assignments for input params.
+        // For example, in[0] records which input param and word r0 stores
+        std::array<ParamSlot, RegCount> in{};
+
+        // Register assignments for output params.
+        // For example, out[0] records which output param (with pointer removed) and word r0 stores
+        std::array<ParamSlot, RegCount> out{};
+
+        // Search the register assignments for a word of a param
+        constexpr std::size_t GetRegisterIndex(std::size_t param_index,
+                                               std::size_t word_index) const {
+            for (std::size_t slot = 0; slot < RegCount; ++slot) {
+                if (in[slot].used && in[slot].param_index == param_index &&
+                    in[slot].word_index == word_index) {
+                    return slot;
+                }
+                if (out[slot].used && out[slot].param_index == param_index &&
+                    out[slot].word_index == word_index) {
+                    return slot;
+                }
+            }
+            // should throw, but gcc bugs out here
+            return 0x12345678;
+        }
+    };
+
+    // Generates ABI info for a given SVC signature R(Context::)(T...)
+    template <typename R, typename... T>
+    static constexpr SVCABI GetSVCABI() {
+        constexpr std::size_t param_count = sizeof...(T);
+        std::array<bool, param_count> param_is_output{{std::is_pointer_v<T>...}};
+        std::array<std::size_t, param_count> param_size{{WordSize<T>()...}};
+        std::array<std::size_t, param_count> output_param_size{{OutputWordSize<T>()...}};
+        std::array<bool, param_count> param_need_align{
+            {(std::is_same_v<T, u64> || std::is_same_v<T, s64>)...}};
+
+        // derives ARM ABI, which assigns all params to r0~r3 and then stack
+        std::array<ParamSlot, 4> armabi_reg{};
+        std::array<ParamSlot, 4> armabi_stack{};
+        std::size_t reg_pos = 0;
+        std::size_t stack_pos = 0;
+        for (std::size_t i = 0; i < param_count; ++i) {
+            if (param_need_align[i]) {
+                reg_pos += reg_pos % 2;
+            }
+            for (std::size_t j = 0; j < param_size[i]; ++j) {
+                if (reg_pos == 4) {
+                    armabi_stack[stack_pos++] = ParamSlot{true, i, j};
+                } else {
+                    armabi_reg[reg_pos++] = ParamSlot{true, i, j};
+                }
+            }
+        }
+
+        // now derives SVC ABI which is a modified version of ARM ABI
+
+        SVCABI mod_abi{};
+        std::size_t out_pos = 0; // next available output register
+
+        // assign return value to output registers
+        if constexpr (!std::is_void_v<R>) {
+            for (std::size_t j = 0; j < WordSize<R>(); ++j) {
+                mod_abi.out[out_pos++] = {true, INDEX_RETURN, j};
+            }
+        }
+
+        // assign output params (pointer-type params) to output registers
+        for (std::size_t i = 0; i < param_count; ++i) {
+            if (param_is_output[i]) {
+                for (std::size_t j = 0; j < output_param_size[i]; ++j) {
+                    mod_abi.out[out_pos++] = ParamSlot{true, i, j};
+                }
+            }
+        }
+
+        // assign input params (non-pointer-type params) to input registers
+        stack_pos = 0; // next available stack param
+        for (std::size_t k = 0; k < mod_abi.in.size(); ++k) {
+            if (k < armabi_reg.size() && armabi_reg[k].used &&
+                !param_is_output[armabi_reg[k].param_index]) {
+                // If this is within the ARM ABI register range and it is a used input param, use
+                // the same register position
+                mod_abi.in[k] = armabi_reg[k];
+            } else {
+                // Otherwise, assign it with the next available stack input
+                // If all stack inputs have been allocated, this would do nothing
+                // and leaves the slot unused.
+                // Loop until an input stack param is found
+                while (stack_pos < armabi_stack.size() &&
+                       (!armabi_stack[stack_pos].used ||
+                        param_is_output[armabi_stack[stack_pos].param_index])) {
+                    ++stack_pos;
+                }
+                // Reassign the stack param to the free register
+                if (stack_pos < armabi_stack.size()) {
+                    mod_abi.in[k] = armabi_stack[stack_pos++];
+                }
+            }
+        }
+
+        return mod_abi;
+    }
+
+    template <std::size_t param_index, std::size_t word_size, std::size_t... indices>
+    static constexpr std::array<std::size_t, word_size> GetRegIndicesImpl(
+        SVCABI abi, std::index_sequence<indices...>) {
+        return {{abi.GetRegisterIndex(param_index, indices)...}};
+    }
+
+    // Gets assigned register index for the param_index-th param of word_size in a function with
+    // signature R(Context::)(Ts...)
+    template <std::size_t param_index, std::size_t word_size, typename R, typename... Ts>
+    static constexpr std::array<std::size_t, word_size> GetRegIndices() {
+        constexpr SVCABI abi = GetSVCABI<R, Ts...>();
+        return GetRegIndicesImpl<param_index, word_size>(abi,
+                                                         std::make_index_sequence<word_size>{});
+    }
+
+    // Gets the value for the param_index-th param of word_size in a function with signature
+    // R(Context::)(Ts...)
+    // T is the param type, which must be a non-pointer as it is an input param.
+    // Must hold: Ts[param_index] == T
+    template <std::size_t param_index, typename T, typename R, typename... Ts>
+    static void GetParam(Context& context, T& value) {
+        static_assert(!std::is_pointer_v<T>, "T should not be a pointer");
+        constexpr auto regi = GetRegIndices<param_index, WordSize<T>(), R, Ts...>();
+        if constexpr (std::is_class_v<T> || std::is_union_v<T>) {
+            std::array<u32, WordSize<T>()> buf;
+            for (std::size_t i = 0; i < WordSize<T>(); ++i) {
+                buf[i] = context.GetReg(regi[i]);
+            }
+            std::memcpy(&value, &buf, sizeof(T));
+        } else if constexpr (WordSize<T>() == 2) {
+            u64 l = context.GetReg(regi[0]);
+            u64 h = context.GetReg(regi[1]);
+            value = static_cast<T>(l | (h << 32));
+        } else if constexpr (std::is_same_v<T, bool>) {
+            // TODO: Is this correct or should only test the lowest byte?
+            value = context.GetReg(regi[0]) != 0;
+        } else {
+            value = static_cast<T>(context.GetReg(regi[0]));
+        }
+    }
+
+    // Sets the value for the param_index-th param of word_size in a function with signature
+    // R(Context::)(Ts...)
+    // T is the param type with pointer removed, which was originally a pointer-type output param
+    // Must hold: (Ts[param_index] == T*) || (R==T && param_index == INDEX_RETURN)
+    template <std::size_t param_index, typename T, typename R, typename... Ts>
+    static void SetParam(Context& context, const T& value) {
+        static_assert(!std::is_pointer_v<T>, "T should have pointer removed before passing in");
+        constexpr auto regi = GetRegIndices<param_index, WordSize<T>(), R, Ts...>();
+        if constexpr (std::is_class_v<T> || std::is_union_v<T>) {
+            std::array<u32, WordSize<T>()> buf;
+            std::memcpy(&buf, &value, sizeof(T));
+            for (std::size_t i = 0; i < WordSize<T>(); ++i) {
+                context.SetReg(regi[i], buf[i]);
+            }
+        } else if constexpr (WordSize<T>() == 2) {
+            u64 u = static_cast<u64>(value);
+            context.SetReg(regi[0], static_cast<u32>(u & 0xFFFFFFFF));
+            context.SetReg(regi[1], static_cast<u32>(u >> 32));
+        } else {
+            u32 u = static_cast<u32>(value);
+            context.SetReg(regi[0], u);
+        }
+    }
+
+    template <typename SVCT, typename R, typename... Ts>
+    struct WrapPass;
+
+    template <typename SVCT, typename R, typename T, typename... Ts>
+    struct WrapPass<SVCT, R, T, Ts...> {
+        // Do I/O for the param T in function R(Context::svc)(Us..., T, Ts...) and then move on to
+        // the next param.
+
+        // Us are params whose I/O is already handled.
+        // T is the current param to do I/O.
+        // Ts are params whose I/O is not handled yet.
+        template <typename... Us>
+        static void Call(Context& context, SVCT svc, Us... u) {
+            static_assert(std::is_same_v<SVCT, R (Context::*)(Us..., T, Ts...)>);
+            constexpr std::size_t current_param_index = sizeof...(Us);
+            if constexpr (std::is_pointer_v<T>) {
+                using OutputT = std::remove_pointer_t<T>;
+                OutputT output;
+                WrapPass<SVCT, R, Ts...>::Call(context, svc, u..., &output);
+                SetParam<current_param_index, OutputT, R, Us..., T, Ts...>(context, output);
+            } else {
+                T input;
+                GetParam<current_param_index, T, R, Us..., T, Ts...>(context, input);
+                WrapPass<SVCT, R, Ts...>::Call(context, svc, u..., input);
+            }
+        }
+    };
+
+    template <typename SVCT, typename R>
+    struct WrapPass<SVCT, R /*empty for T, Ts...*/> {
+        // Call function R(Context::svc)(Us...) and transfer the return value to registers
+        template <typename... Us>
+        static void Call(Context& context, SVCT svc, Us... u) {
+            static_assert(std::is_same_v<SVCT, R (Context::*)(Us...)>);
+            if constexpr (std::is_void_v<R>) {
+                (context.*svc)(u...);
+            } else {
+                R r = (context.*svc)(u...);
+                SetParam<INDEX_RETURN, R, R, Us...>(context, r);
+            }
+        }
+    };
+
+    template <typename T>
+    struct WrapHelper;
+
+    template <typename R, typename... T>
+    struct WrapHelper<R (Context::*)(T...)> {
+        static void Call(Context& context, R (Context::*svc)(T...)) {
+            WrapPass<decltype(svc), R, T...>::Call(context, svc /*Empty for Us*/);
+        }
+    };
+};
+
+} // namespace Kernel

src/core/hle/result.h

@@ -216,10 +216,7 @@ union ResultCode {
         : raw(description.FormatValue(description_) | module.FormatValue(module_) |
               summary.FormatValue(summary_) | level.FormatValue(level_)) {}
 
-    constexpr ResultCode& operator=(const ResultCode& o) {
+    constexpr ResultCode& operator=(const ResultCode& o) = default;
-        raw = o.raw;
-        return *this;
-    }
 
     constexpr bool IsSuccess() const {
         return is_error.ExtractValue(raw) == 0;