Fix Texture Overlapping

shader_cache: Only lock covered instructions.

.gitmodules

@@ -13,9 +13,6 @@
 [submodule "dynarmic"]
     path = externals/dynarmic
     url = https://github.com/MerryMage/dynarmic.git
-[submodule "xbyak"]
-    path = externals/xbyak
-    url = https://github.com/herumi/xbyak.git
 [submodule "fmt"]
     path = externals/fmt
     url = https://github.com/fmtlib/fmt.git

externals/CMakeLists.txt

@@ -9,7 +9,6 @@ target_include_directories(catch-single-include INTERFACE catch/single_include)
 
 # Dynarmic
 if (ARCHITECTURE_x86_64)
-    add_library(xbyak INTERFACE)
     set(DYNARMIC_TESTS OFF)
     set(DYNARMIC_NO_BUNDLED_FMT ON)
     add_subdirectory(dynarmic)
@@ -53,14 +52,6 @@ target_include_directories(unicorn-headers INTERFACE ./unicorn/include)
 # SoundTouch
 add_subdirectory(soundtouch)
 
-# Xbyak
-if (ARCHITECTURE_x86_64)
-    # Defined before "dynarmic" above
-    # add_library(xbyak INTERFACE)
-    target_include_directories(xbyak INTERFACE ./xbyak/xbyak)
-    target_compile_definitions(xbyak INTERFACE XBYAK_NO_OP_NAMES)
-endif()
-
 # Opus
 add_subdirectory(opus)
 target_include_directories(opus INTERFACE ./opus/include)

src/common/CMakeLists.txt

@@ -44,7 +44,6 @@ add_library(common STATIC
     detached_tasks.cpp
     detached_tasks.h
     bit_field.h
-    bit_set.h
     cityhash.cpp
     cityhash.h
     color.h
@@ -95,14 +94,9 @@ if(ARCHITECTURE_x86_64)
         PRIVATE
             x64/cpu_detect.cpp
             x64/cpu_detect.h
-            x64/xbyak_abi.h
-            x64/xbyak_util.h
     )
 endif()
 
 create_target_directory_groups(common)
 
 target_link_libraries(common PUBLIC Boost::boost fmt microprofile)
-if (ARCHITECTURE_x86_64)
-    target_link_libraries(common PRIVATE xbyak)
-endif()

src/common/bit_set.h

@@ -1,244 +0,0 @@
-// This file is under the public domain.
-
-#pragma once
-
-#include <cstddef>
-#ifdef _WIN32
-#include <intrin.h>
-#endif
-#include <initializer_list>
-#include <new>
-#include <type_traits>
-#include "common/common_types.h"
-
-// namespace avoids conflict with OS X Carbon; don't use BitSet<T> directly
-namespace Common {
-
-// Helper functions:
-
-#ifdef _MSC_VER
-template <typename T>
-static inline int CountSetBits(T v) {
-    // from https://graphics.stanford.edu/~seander/bithacks.html
-    // GCC has this built in, but MSVC's intrinsic will only emit the actual
-    // POPCNT instruction, which we're not depending on
-    v = v - ((v >> 1) & (T) ~(T)0 / 3);
-    v = (v & (T) ~(T)0 / 15 * 3) + ((v >> 2) & (T) ~(T)0 / 15 * 3);
-    v = (v + (v >> 4)) & (T) ~(T)0 / 255 * 15;
-    return (T)(v * ((T) ~(T)0 / 255)) >> (sizeof(T) - 1) * 8;
-}
-static inline int LeastSignificantSetBit(u8 val) {
-    unsigned long index;
-    _BitScanForward(&index, val);
-    return (int)index;
-}
-static inline int LeastSignificantSetBit(u16 val) {
-    unsigned long index;
-    _BitScanForward(&index, val);
-    return (int)index;
-}
-static inline int LeastSignificantSetBit(u32 val) {
-    unsigned long index;
-    _BitScanForward(&index, val);
-    return (int)index;
-}
-static inline int LeastSignificantSetBit(u64 val) {
-    unsigned long index;
-    _BitScanForward64(&index, val);
-    return (int)index;
-}
-#else
-static inline int CountSetBits(u8 val) {
-    return __builtin_popcount(val);
-}
-static inline int CountSetBits(u16 val) {
-    return __builtin_popcount(val);
-}
-static inline int CountSetBits(u32 val) {
-    return __builtin_popcount(val);
-}
-static inline int CountSetBits(u64 val) {
-    return __builtin_popcountll(val);
-}
-static inline int LeastSignificantSetBit(u8 val) {
-    return __builtin_ctz(val);
-}
-static inline int LeastSignificantSetBit(u16 val) {
-    return __builtin_ctz(val);
-}
-static inline int LeastSignificantSetBit(u32 val) {
-    return __builtin_ctz(val);
-}
-static inline int LeastSignificantSetBit(u64 val) {
-    return __builtin_ctzll(val);
-}
-#endif
-
-// Similar to std::bitset, this is a class which encapsulates a bitset, i.e.
-// using the set bits of an integer to represent a set of integers.  Like that
-// class, it acts like an array of bools:
-//     BitSet32 bs;
-//     bs[1] = true;
-// but also like the underlying integer ([0] = least significant bit):
-//     BitSet32 bs2 = ...;
-//     bs = (bs ^ bs2) & BitSet32(0xffff);
-// The following additional functionality is provided:
-// - Construction using an initializer list.
-//     BitSet bs { 1, 2, 4, 8 };
-// - Efficiently iterating through the set bits:
-//     for (int i : bs)
-//         [i is the *index* of a set bit]
-//   (This uses the appropriate CPU instruction to find the next set bit in one
-//   operation.)
-// - Counting set bits using .Count() - see comment on that method.
-
-// TODO: use constexpr when MSVC gets out of the Dark Ages
-
-template <typename IntTy>
-class BitSet {
-    static_assert(!std::is_signed_v<IntTy>, "BitSet should not be used with signed types");
-
-public:
-    // A reference to a particular bit, returned from operator[].
-    class Ref {
-    public:
-        Ref(Ref&& other) : m_bs(other.m_bs), m_mask(other.m_mask) {}
-        Ref(BitSet* bs, IntTy mask) : m_bs(bs), m_mask(mask) {}
-        operator bool() const {
-            return (m_bs->m_val & m_mask) != 0;
-        }
-        bool operator=(bool set) {
-            m_bs->m_val = (m_bs->m_val & ~m_mask) | (set ? m_mask : 0);
-            return set;
-        }
-
-    private:
-        BitSet* m_bs;
-        IntTy m_mask;
-    };
-
-    // A STL-like iterator is required to be able to use range-based for loops.
-    class Iterator {
-    public:
-        Iterator(const Iterator& other) : m_val(other.m_val), m_bit(other.m_bit) {}
-        Iterator(IntTy val) : m_val(val), m_bit(0) {}
-        Iterator& operator=(Iterator other) {
-            new (this) Iterator(other);
-            return *this;
-        }
-        int operator*() {
-            return m_bit + ComputeLsb();
-        }
-        Iterator& operator++() {
-            int lsb = ComputeLsb();
-            m_val >>= lsb + 1;
-            m_bit += lsb + 1;
-            m_has_lsb = false;
-            return *this;
-        }
-        Iterator operator++(int _) {
-            Iterator other(*this);
-            ++*this;
-            return other;
-        }
-        bool operator==(Iterator other) const {
-            return m_val == other.m_val;
-        }
-        bool operator!=(Iterator other) const {
-            return m_val != other.m_val;
-        }
-
-    private:
-        int ComputeLsb() {
-            if (!m_has_lsb) {
-                m_lsb = LeastSignificantSetBit(m_val);
-                m_has_lsb = true;
-            }
-            return m_lsb;
-        }
-        IntTy m_val;
-        int m_bit;
-        int m_lsb = -1;
-        bool m_has_lsb = false;
-    };
-
-    BitSet() : m_val(0) {}
-    explicit BitSet(IntTy val) : m_val(val) {}
-    BitSet(std::initializer_list<int> init) {
-        m_val = 0;
-        for (int bit : init)
-            m_val |= (IntTy)1 << bit;
-    }
-
-    static BitSet AllTrue(std::size_t count) {
-        return BitSet(count == sizeof(IntTy) * 8 ? ~(IntTy)0 : (((IntTy)1 << count) - 1));
-    }
-
-    Ref operator[](std::size_t bit) {
-        return Ref(this, (IntTy)1 << bit);
-    }
-    const Ref operator[](std::size_t bit) const {
-        return (*const_cast<BitSet*>(this))[bit];
-    }
-    bool operator==(BitSet other) const {
-        return m_val == other.m_val;
-    }
-    bool operator!=(BitSet other) const {
-        return m_val != other.m_val;
-    }
-    bool operator<(BitSet other) const {
-        return m_val < other.m_val;
-    }
-    bool operator>(BitSet other) const {
-        return m_val > other.m_val;
-    }
-    BitSet operator|(BitSet other) const {
-        return BitSet(m_val | other.m_val);
-    }
-    BitSet operator&(BitSet other) const {
-        return BitSet(m_val & other.m_val);
-    }
-    BitSet operator^(BitSet other) const {
-        return BitSet(m_val ^ other.m_val);
-    }
-    BitSet operator~() const {
-        return BitSet(~m_val);
-    }
-    BitSet& operator|=(BitSet other) {
-        return *this = *this | other;
-    }
-    BitSet& operator&=(BitSet other) {
-        return *this = *this & other;
-    }
-    BitSet& operator^=(BitSet other) {
-        return *this = *this ^ other;
-    }
-    operator u32() = delete;
-    operator bool() {
-        return m_val != 0;
-    }
-
-    // Warning: Even though on modern CPUs this is a single fast instruction,
-    // Dolphin's official builds do not currently assume POPCNT support on x86,
-    // so slower explicit bit twiddling is generated.  Still should generally
-    // be faster than a loop.
-    unsigned int Count() const {
-        return CountSetBits(m_val);
-    }
-
-    Iterator begin() const {
-        return Iterator(m_val);
-    }
-    Iterator end() const {
-        return Iterator(0);
-    }
-
-    IntTy m_val;
-};
-
-} // namespace Common
-
-typedef Common::BitSet<u8> BitSet8;
-typedef Common::BitSet<u16> BitSet16;
-typedef Common::BitSet<u32> BitSet32;
-typedef Common::BitSet<u64> BitSet64;

src/common/math_util.h

@@ -4,18 +4,12 @@
 
 #pragma once
 
-#include <algorithm>
 #include <cstdlib>
 #include <type_traits>
 
 namespace MathUtil {
 
-static constexpr float PI = 3.14159265f;
-
-inline bool IntervalsIntersect(unsigned start0, unsigned length0, unsigned start1,
-                               unsigned length1) {
-    return (std::max(start0, start1) < std::min(start0 + length0, start1 + length1));
-}
+constexpr float PI = 3.14159265f;
 
 template <class T>
 struct Rectangle {
@@ -24,16 +18,16 @@ struct Rectangle {
     T right{};
     T bottom{};
 
-    Rectangle() = default;
+    constexpr Rectangle() = default;
 
-    Rectangle(T left, T top, T right, T bottom)
+    constexpr Rectangle(T left, T top, T right, T bottom)
         : left(left), top(top), right(right), bottom(bottom) {}
 
     T GetWidth() const {
-        return std::abs(static_cast<typename std::make_signed<T>::type>(right - left));
+        return std::abs(static_cast<std::make_signed_t<T>>(right - left));
     }
     T GetHeight() const {
-        return std::abs(static_cast<typename std::make_signed<T>::type>(bottom - top));
+        return std::abs(static_cast<std::make_signed_t<T>>(bottom - top));
     }
     Rectangle<T> TranslateX(const T x) const {
         return Rectangle{left + x, top, right + x, bottom};

src/common/thread.cpp

@@ -25,23 +25,6 @@
 
 namespace Common {
 
-int CurrentThreadId() {
-#ifdef _MSC_VER
-    return GetCurrentThreadId();
-#elif defined __APPLE__
-    return mach_thread_self();
-#else
-    return 0;
-#endif
-}
-
-#ifdef _WIN32
-// Supporting functions
-void SleepCurrentThread(int ms) {
-    Sleep(ms);
-}
-#endif
-
 #ifdef _MSC_VER
 
 void SetThreadAffinity(std::thread::native_handle_type thread, u32 mask) {
@@ -62,7 +45,7 @@ void SwitchCurrentThread() {
 
 // This is implemented much nicer in upcoming msvc++, see:
 // http://msdn.microsoft.com/en-us/library/xcb2z8hs(VS.100).aspx
-void SetCurrentThreadName(const char* szThreadName) {
+void SetCurrentThreadName(const char* name) {
     static const DWORD MS_VC_EXCEPTION = 0x406D1388;
 
 #pragma pack(push, 8)
@@ -75,7 +58,7 @@ void SetCurrentThreadName(const char* szThreadName) {
 #pragma pack(pop)
 
     info.dwType = 0x1000;
-    info.szName = szThreadName;
+    info.szName = name;
     info.dwThreadID = -1; // dwThreadID;
     info.dwFlags = 0;
 
@@ -107,10 +90,6 @@ void SetCurrentThreadAffinity(u32 mask) {
 }
 
 #ifndef _WIN32
-void SleepCurrentThread(int ms) {
-    usleep(1000 * ms);
-}
-
 void SwitchCurrentThread() {
     usleep(1000 * 1);
 }
@@ -118,15 +97,15 @@ void SwitchCurrentThread() {
 
 // MinGW with the POSIX threading model does not support pthread_setname_np
 #if !defined(_WIN32) || defined(_MSC_VER)
-void SetCurrentThreadName(const char* szThreadName) {
+void SetCurrentThreadName(const char* name) {
 #ifdef __APPLE__
-    pthread_setname_np(szThreadName);
+    pthread_setname_np(name);
 #elif defined(__Bitrig__) || defined(__DragonFly__) || defined(__FreeBSD__) || defined(__OpenBSD__)
-    pthread_set_name_np(pthread_self(), szThreadName);
+    pthread_set_name_np(pthread_self(), name);
 #elif defined(__NetBSD__)
-    pthread_setname_np(pthread_self(), "%s", (void*)szThreadName);
+    pthread_setname_np(pthread_self(), "%s", (void*)name);
 #else
-    pthread_setname_np(pthread_self(), szThreadName);
+    pthread_setname_np(pthread_self(), name);
 #endif
 }
 #endif

src/common/thread.h

@@ -13,15 +13,8 @@
 
 namespace Common {
 
-int CurrentThreadId();
-
-void SetThreadAffinity(std::thread::native_handle_type thread, u32 mask);
-void SetCurrentThreadAffinity(u32 mask);
-
 class Event {
 public:
-    Event() : is_set(false) {}
-
     void Set() {
         std::lock_guard<std::mutex> lk(mutex);
         if (!is_set) {
@@ -53,14 +46,14 @@ public:
     }
 
 private:
-    bool is_set;
+    bool is_set = false;
     std::condition_variable condvar;
     std::mutex mutex;
 };
 
 class Barrier {
 public:
-    explicit Barrier(std::size_t count_) : count(count_), waiting(0), generation(0) {}
+    explicit Barrier(std::size_t count_) : count(count_) {}
 
     /// Blocks until all "count" threads have called Sync()
     void Sync() {
@@ -80,12 +73,13 @@ public:
 private:
     std::condition_variable condvar;
     std::mutex mutex;
-    const std::size_t count;
-    std::size_t waiting;
-    std::size_t generation; // Incremented once each time the barrier is used
+    std::size_t count;
+    std::size_t waiting = 0;
+    std::size_t generation = 0; // Incremented once each time the barrier is used
 };
 
-void SleepCurrentThread(int ms);
+void SetThreadAffinity(std::thread::native_handle_type thread, u32 mask);
+void SetCurrentThreadAffinity(u32 mask);
 void SwitchCurrentThread(); // On Linux, this is equal to sleep 1ms
 void SetCurrentThreadName(const char* name);
 

src/common/x64/xbyak_abi.h

@@ -1,222 +0,0 @@
-// Copyright 2016 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <initializer_list>
-#include <xbyak.h>
-#include "common/assert.h"
-#include "common/bit_set.h"
-
-namespace Common::X64 {
-
-inline int RegToIndex(const Xbyak::Reg& reg) {
-    using Kind = Xbyak::Reg::Kind;
-    ASSERT_MSG((reg.getKind() & (Kind::REG | Kind::XMM)) != 0,
-               "RegSet only support GPRs and XMM registers.");
-    ASSERT_MSG(reg.getIdx() < 16, "RegSet only supports XXM0-15.");
-    return reg.getIdx() + (reg.getKind() == Kind::REG ? 0 : 16);
-}
-
-inline Xbyak::Reg64 IndexToReg64(int reg_index) {
-    ASSERT(reg_index < 16);
-    return Xbyak::Reg64(reg_index);
-}
-
-inline Xbyak::Xmm IndexToXmm(int reg_index) {
-    ASSERT(reg_index >= 16 && reg_index < 32);
-    return Xbyak::Xmm(reg_index - 16);
-}
-
-inline Xbyak::Reg IndexToReg(int reg_index) {
-    if (reg_index < 16) {
-        return IndexToReg64(reg_index);
-    } else {
-        return IndexToXmm(reg_index);
-    }
-}
-
-inline BitSet32 BuildRegSet(std::initializer_list<Xbyak::Reg> regs) {
-    BitSet32 bits;
-    for (const Xbyak::Reg& reg : regs) {
-        bits[RegToIndex(reg)] = true;
-    }
-    return bits;
-}
-
-const BitSet32 ABI_ALL_GPRS(0x0000FFFF);
-const BitSet32 ABI_ALL_XMMS(0xFFFF0000);
-
-#ifdef _WIN32
-
-// Microsoft x64 ABI
-const Xbyak::Reg ABI_RETURN = Xbyak::util::rax;
-const Xbyak::Reg ABI_PARAM1 = Xbyak::util::rcx;
-const Xbyak::Reg ABI_PARAM2 = Xbyak::util::rdx;
-const Xbyak::Reg ABI_PARAM3 = Xbyak::util::r8;
-const Xbyak::Reg ABI_PARAM4 = Xbyak::util::r9;
-
-const BitSet32 ABI_ALL_CALLER_SAVED = BuildRegSet({
-    // GPRs
-    Xbyak::util::rcx,
-    Xbyak::util::rdx,
-    Xbyak::util::r8,
-    Xbyak::util::r9,
-    Xbyak::util::r10,
-    Xbyak::util::r11,
-    // XMMs
-    Xbyak::util::xmm0,
-    Xbyak::util::xmm1,
-    Xbyak::util::xmm2,
-    Xbyak::util::xmm3,
-    Xbyak::util::xmm4,
-    Xbyak::util::xmm5,
-});
-
-const BitSet32 ABI_ALL_CALLEE_SAVED = BuildRegSet({
-    // GPRs
-    Xbyak::util::rbx,
-    Xbyak::util::rsi,
-    Xbyak::util::rdi,
-    Xbyak::util::rbp,
-    Xbyak::util::r12,
-    Xbyak::util::r13,
-    Xbyak::util::r14,
-    Xbyak::util::r15,
-    // XMMs
-    Xbyak::util::xmm6,
-    Xbyak::util::xmm7,
-    Xbyak::util::xmm8,
-    Xbyak::util::xmm9,
-    Xbyak::util::xmm10,
-    Xbyak::util::xmm11,
-    Xbyak::util::xmm12,
-    Xbyak::util::xmm13,
-    Xbyak::util::xmm14,
-    Xbyak::util::xmm15,
-});
-
-constexpr std::size_t ABI_SHADOW_SPACE = 0x20;
-
-#else
-
-// System V x86-64 ABI
-const Xbyak::Reg ABI_RETURN = Xbyak::util::rax;
-const Xbyak::Reg ABI_PARAM1 = Xbyak::util::rdi;
-const Xbyak::Reg ABI_PARAM2 = Xbyak::util::rsi;
-const Xbyak::Reg ABI_PARAM3 = Xbyak::util::rdx;
-const Xbyak::Reg ABI_PARAM4 = Xbyak::util::rcx;
-
-const BitSet32 ABI_ALL_CALLER_SAVED = BuildRegSet({
-    // GPRs
-    Xbyak::util::rcx,
-    Xbyak::util::rdx,
-    Xbyak::util::rdi,
-    Xbyak::util::rsi,
-    Xbyak::util::r8,
-    Xbyak::util::r9,
-    Xbyak::util::r10,
-    Xbyak::util::r11,
-    // XMMs
-    Xbyak::util::xmm0,
-    Xbyak::util::xmm1,
-    Xbyak::util::xmm2,
-    Xbyak::util::xmm3,
-    Xbyak::util::xmm4,
-    Xbyak::util::xmm5,
-    Xbyak::util::xmm6,
-    Xbyak::util::xmm7,
-    Xbyak::util::xmm8,
-    Xbyak::util::xmm9,
-    Xbyak::util::xmm10,
-    Xbyak::util::xmm11,
-    Xbyak::util::xmm12,
-    Xbyak::util::xmm13,
-    Xbyak::util::xmm14,
-    Xbyak::util::xmm15,
-});
-
-const BitSet32 ABI_ALL_CALLEE_SAVED = BuildRegSet({
-    // GPRs
-    Xbyak::util::rbx,
-    Xbyak::util::rbp,
-    Xbyak::util::r12,
-    Xbyak::util::r13,
-    Xbyak::util::r14,
-    Xbyak::util::r15,
-});
-
-constexpr std::size_t ABI_SHADOW_SPACE = 0;
-
-#endif
-
-inline void ABI_CalculateFrameSize(BitSet32 regs, std::size_t rsp_alignment,
-                                   std::size_t needed_frame_size, s32* out_subtraction,
-                                   s32* out_xmm_offset) {
-    int count = (regs & ABI_ALL_GPRS).Count();
-    rsp_alignment -= count * 8;
-    std::size_t subtraction = 0;
-    int xmm_count = (regs & ABI_ALL_XMMS).Count();
-    if (xmm_count) {
-        // If we have any XMMs to save, we must align the stack here.
-        subtraction = rsp_alignment & 0xF;
-    }
-    subtraction += 0x10 * xmm_count;
-    std::size_t xmm_base_subtraction = subtraction;
-    subtraction += needed_frame_size;
-    subtraction += ABI_SHADOW_SPACE;
-    // Final alignment.
-    rsp_alignment -= subtraction;
-    subtraction += rsp_alignment & 0xF;
-
-    *out_subtraction = (s32)subtraction;
-    *out_xmm_offset = (s32)(subtraction - xmm_base_subtraction);
-}
-
-inline std::size_t ABI_PushRegistersAndAdjustStack(Xbyak::CodeGenerator& code, BitSet32 regs,
-                                                   std::size_t rsp_alignment,
-                                                   std::size_t needed_frame_size = 0) {
-    s32 subtraction, xmm_offset;
-    ABI_CalculateFrameSize(regs, rsp_alignment, needed_frame_size, &subtraction, &xmm_offset);
-
-    for (int reg_index : (regs & ABI_ALL_GPRS)) {
-        code.push(IndexToReg64(reg_index));
-    }
-
-    if (subtraction != 0) {
-        code.sub(code.rsp, subtraction);
-    }
-
-    for (int reg_index : (regs & ABI_ALL_XMMS)) {
-        code.movaps(code.xword[code.rsp + xmm_offset], IndexToXmm(reg_index));
-        xmm_offset += 0x10;
-    }
-
-    return ABI_SHADOW_SPACE;
-}
-
-inline void ABI_PopRegistersAndAdjustStack(Xbyak::CodeGenerator& code, BitSet32 regs,
-                                           std::size_t rsp_alignment,
-                                           std::size_t needed_frame_size = 0) {
-    s32 subtraction, xmm_offset;
-    ABI_CalculateFrameSize(regs, rsp_alignment, needed_frame_size, &subtraction, &xmm_offset);
-
-    for (int reg_index : (regs & ABI_ALL_XMMS)) {
-        code.movaps(IndexToXmm(reg_index), code.xword[code.rsp + xmm_offset]);
-        xmm_offset += 0x10;
-    }
-
-    if (subtraction != 0) {
-        code.add(code.rsp, subtraction);
-    }
-
-    // GPRs need to be popped in reverse order
-    for (int reg_index = 15; reg_index >= 0; reg_index--) {
-        if (regs[reg_index]) {
-            code.pop(IndexToReg64(reg_index));
-        }
-    }
-}
-
-} // namespace Common::X64

src/common/x64/xbyak_util.h

@@ -1,47 +0,0 @@
-// Copyright 2016 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <type_traits>
-#include <xbyak.h>
-#include "common/x64/xbyak_abi.h"
-
-namespace Common::X64 {
-
-// Constants for use with cmpps/cmpss
-enum {
-    CMP_EQ = 0,
-    CMP_LT = 1,
-    CMP_LE = 2,
-    CMP_UNORD = 3,
-    CMP_NEQ = 4,
-    CMP_NLT = 5,
-    CMP_NLE = 6,
-    CMP_ORD = 7,
-};
-
-inline bool IsWithin2G(uintptr_t ref, uintptr_t target) {
-    u64 distance = target - (ref + 5);
-    return !(distance >= 0x8000'0000ULL && distance <= ~0x8000'0000ULL);
-}
-
-inline bool IsWithin2G(const Xbyak::CodeGenerator& code, uintptr_t target) {
-    return IsWithin2G(reinterpret_cast<uintptr_t>(code.getCurr()), target);
-}
-
-template <typename T>
-inline void CallFarFunction(Xbyak::CodeGenerator& code, const T f) {
-    static_assert(std::is_pointer_v<T>, "Argument must be a (function) pointer.");
-    std::size_t addr = reinterpret_cast<std::size_t>(f);
-    if (IsWithin2G(code, addr)) {
-        code.call(f);
-    } else {
-        // ABI_RETURN is a safe temp register to use before a call
-        code.mov(ABI_RETURN, addr);
-        code.call(ABI_RETURN);
-    }
-}
-
-} // namespace Common::X64

src/core/CMakeLists.txt

@@ -12,6 +12,8 @@ add_library(core STATIC
     core_timing.h
     core_timing_util.cpp
     core_timing_util.h
+    cpu_core_manager.cpp
+    cpu_core_manager.h
     crypto/aes_util.cpp
     crypto/aes_util.h
     crypto/encryption_layer.cpp
@@ -156,6 +158,8 @@ add_library(core STATIC
     hle/service/am/applets/applets.h
     hle/service/am/applets/software_keyboard.cpp
     hle/service/am/applets/software_keyboard.h
+    hle/service/am/applets/stub_applet.cpp
+    hle/service/am/applets/stub_applet.h
     hle/service/am/idle.cpp
     hle/service/am/idle.h
     hle/service/am/omm.cpp

src/core/core.cpp

@@ -14,6 +14,7 @@
 #include "core/core.h"
 #include "core/core_cpu.h"
 #include "core/core_timing.h"
+#include "core/cpu_core_manager.h"
 #include "core/file_sys/mode.h"
 #include "core/file_sys/vfs_concat.h"
 #include "core/file_sys/vfs_real.h"
@@ -28,7 +29,6 @@
 #include "core/hle/service/sm/sm.h"
 #include "core/loader/loader.h"
 #include "core/perf_stats.h"
-#include "core/settings.h"
 #include "core/telemetry_session.h"
 #include "frontend/applets/software_keyboard.h"
 #include "video_core/debug_utils/debug_utils.h"
@@ -71,64 +71,22 @@ FileSys::VirtualFile GetGameFileFromPath(const FileSys::VirtualFilesystem& vfs,
 
     return vfs->OpenFile(path, FileSys::Mode::Read);
 }
-
-/// Runs a CPU core while the system is powered on
-void RunCpuCore(Cpu& cpu_state) {
-    while (Core::System::GetInstance().IsPoweredOn()) {
-        cpu_state.RunLoop(true);
-    }
-}
 } // Anonymous namespace
 
 struct System::Impl {
     Cpu& CurrentCpuCore() {
-        if (Settings::values.use_multi_core) {
-            const auto& search = thread_to_cpu.find(std::this_thread::get_id());
-            ASSERT(search != thread_to_cpu.end());
-            ASSERT(search->second);
-            return *search->second;
-        }
-
-        // Otherwise, use single-threaded mode active_core variable
-        return *cpu_cores[active_core];
+        return cpu_core_manager.GetCurrentCore();
     }
 
     ResultStatus RunLoop(bool tight_loop) {
         status = ResultStatus::Success;
 
-        // Update thread_to_cpu in case Core 0 is run from a different host thread
-        thread_to_cpu[std::this_thread::get_id()] = cpu_cores[0].get();
-
-        if (GDBStub::IsServerEnabled()) {
-            GDBStub::HandlePacket();
-
-            // If the loop is halted and we want to step, use a tiny (1) number of instructions to
-            // execute. Otherwise, get out of the loop function.
-            if (GDBStub::GetCpuHaltFlag()) {
-                if (GDBStub::GetCpuStepFlag()) {
-                    tight_loop = false;
-                } else {
-                    return ResultStatus::Success;
-                }
-            }
-        }
-
-        for (active_core = 0; active_core < NUM_CPU_CORES; ++active_core) {
-            cpu_cores[active_core]->RunLoop(tight_loop);
-            if (Settings::values.use_multi_core) {
-                // Cores 1-3 are run on other threads in this mode
-                break;
-            }
-        }
-
-        if (GDBStub::IsServerEnabled()) {
-            GDBStub::SetCpuStepFlag(false);
-        }
+        cpu_core_manager.RunLoop(tight_loop);
 
         return status;
     }
 
-    ResultStatus Init(Frontend::EmuWindow& emu_window) {
+    ResultStatus Init(System& system, Frontend::EmuWindow& emu_window) {
         LOG_DEBUG(HW_Memory, "initialized OK");
 
         CoreTiming::Init();
@@ -145,12 +103,6 @@ struct System::Impl {
         auto main_process = Kernel::Process::Create(kernel, "main");
         kernel.MakeCurrentProcess(main_process.get());
 
-        cpu_barrier = std::make_unique<CpuBarrier>();
-        cpu_exclusive_monitor = Cpu::MakeExclusiveMonitor(cpu_cores.size());
-        for (std::size_t index = 0; index < cpu_cores.size(); ++index) {
-            cpu_cores[index] = std::make_unique<Cpu>(*cpu_exclusive_monitor, *cpu_barrier, index);
-        }
-
         telemetry_session = std::make_unique<Core::TelemetrySession>();
         service_manager = std::make_shared<Service::SM::ServiceManager>();
 
@@ -164,17 +116,8 @@ struct System::Impl {
 
         gpu_core = std::make_unique<Tegra::GPU>(renderer->Rasterizer());
 
-        // Create threads for CPU cores 1-3, and build thread_to_cpu map
-        // CPU core 0 is run on the main thread
-        thread_to_cpu[std::this_thread::get_id()] = cpu_cores[0].get();
-        if (Settings::values.use_multi_core) {
-            for (std::size_t index = 0; index < cpu_core_threads.size(); ++index) {
-                cpu_core_threads[index] =
-                    std::make_unique<std::thread>(RunCpuCore, std::ref(*cpu_cores[index + 1]));
-                thread_to_cpu[cpu_core_threads[index]->get_id()] = cpu_cores[index + 1].get();
-            }
-        }
-
+        cpu_core_manager.Initialize(system);
+        is_powered_on = true;
         LOG_DEBUG(Core, "Initialized OK");
 
         // Reset counters and set time origin to current frame
@@ -184,7 +127,8 @@ struct System::Impl {
         return ResultStatus::Success;
     }
 
-    ResultStatus Load(Frontend::EmuWindow& emu_window, const std::string& filepath) {
+    ResultStatus Load(System& system, Frontend::EmuWindow& emu_window,
+                      const std::string& filepath) {
         app_loader = Loader::GetLoader(GetGameFileFromPath(virtual_filesystem, filepath));
 
         if (!app_loader) {
@@ -201,7 +145,7 @@ struct System::Impl {
             return ResultStatus::ErrorSystemMode;
         }
 
-        ResultStatus init_result{Init(emu_window)};
+        ResultStatus init_result{Init(system, emu_window)};
         if (init_result != ResultStatus::Success) {
             LOG_CRITICAL(Core, "Failed to initialize system (Error {})!",
                          static_cast<int>(init_result));
@@ -231,6 +175,8 @@ struct System::Impl {
         Telemetry().AddField(Telemetry::FieldType::Performance, "Shutdown_Frametime",
                              perf_results.frametime * 1000.0);
 
+        is_powered_on = false;
+
         // Shutdown emulation session
         renderer.reset();
         GDBStub::Shutdown();
@@ -240,19 +186,7 @@ struct System::Impl {
         gpu_core.reset();
 
         // Close all CPU/threading state
-        cpu_barrier->NotifyEnd();
-        if (Settings::values.use_multi_core) {
-            for (auto& thread : cpu_core_threads) {
-                thread->join();
-                thread.reset();
-            }
-        }
-        thread_to_cpu.clear();
-        for (auto& cpu_core : cpu_cores) {
-            cpu_core.reset();
-        }
-        cpu_exclusive_monitor.reset();
-        cpu_barrier.reset();
+        cpu_core_manager.Shutdown();
 
         // Shutdown kernel and core timing
         kernel.Shutdown();
@@ -289,11 +223,8 @@ struct System::Impl {
     std::unique_ptr<VideoCore::RendererBase> renderer;
     std::unique_ptr<Tegra::GPU> gpu_core;
     std::shared_ptr<Tegra::DebugContext> debug_context;
-    std::unique_ptr<ExclusiveMonitor> cpu_exclusive_monitor;
-    std::unique_ptr<CpuBarrier> cpu_barrier;
-    std::array<std::unique_ptr<Cpu>, NUM_CPU_CORES> cpu_cores;
-    std::array<std::unique_ptr<std::thread>, NUM_CPU_CORES - 1> cpu_core_threads;
-    std::size_t active_core{}; ///< Active core, only used in single thread mode
+    CpuCoreManager cpu_core_manager;
+    bool is_powered_on = false;
 
     /// Frontend applets
     std::unique_ptr<Core::Frontend::SoftwareKeyboardApplet> software_keyboard;
@@ -307,9 +238,6 @@ struct System::Impl {
     ResultStatus status = ResultStatus::Success;
     std::string status_details = "";
 
-    /// Map of guest threads to CPU cores
-    std::map<std::thread::id, Cpu*> thread_to_cpu;
-
     Core::PerfStats perf_stats;
     Core::FrameLimiter frame_limiter;
 };
@@ -334,17 +262,15 @@ System::ResultStatus System::SingleStep() {
 }
 
 void System::InvalidateCpuInstructionCaches() {
-    for (auto& cpu : impl->cpu_cores) {
-        cpu->ArmInterface().ClearInstructionCache();
-    }
+    impl->cpu_core_manager.InvalidateAllInstructionCaches();
 }
 
 System::ResultStatus System::Load(Frontend::EmuWindow& emu_window, const std::string& filepath) {
-    return impl->Load(emu_window, filepath);
+    return impl->Load(*this, emu_window, filepath);
 }
 
 bool System::IsPoweredOn() const {
-    return impl->cpu_barrier && impl->cpu_barrier->IsAlive();
+    return impl->is_powered_on;
 }
 
 void System::PrepareReschedule() {
@@ -408,21 +334,20 @@ const ARM_Interface& System::ArmInterface(std::size_t core_index) const {
 }
 
 Cpu& System::CpuCore(std::size_t core_index) {
-    ASSERT(core_index < NUM_CPU_CORES);
-    return *impl->cpu_cores[core_index];
+    return impl->cpu_core_manager.GetCore(core_index);
 }
 
 const Cpu& System::CpuCore(std::size_t core_index) const {
     ASSERT(core_index < NUM_CPU_CORES);
-    return *impl->cpu_cores[core_index];
+    return impl->cpu_core_manager.GetCore(core_index);
 }
 
 ExclusiveMonitor& System::Monitor() {
-    return *impl->cpu_exclusive_monitor;
+    return impl->cpu_core_manager.GetExclusiveMonitor();
 }
 
 const ExclusiveMonitor& System::Monitor() const {
-    return *impl->cpu_exclusive_monitor;
+    return impl->cpu_core_manager.GetExclusiveMonitor();
 }
 
 Tegra::GPU& System::GPU() {
@@ -506,7 +431,7 @@ const Core::Frontend::SoftwareKeyboardApplet& System::GetSoftwareKeyboard() cons
 }
 
 System::ResultStatus System::Init(Frontend::EmuWindow& emu_window) {
-    return impl->Init(emu_window);
+    return impl->Init(*this, emu_window);
 }
 
 void System::Shutdown() {

src/core/cpu_core_manager.cpp

@@ -0,0 +1,142 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "common/assert.h"
+#include "core/arm/exclusive_monitor.h"
+#include "core/core.h"
+#include "core/core_cpu.h"
+#include "core/cpu_core_manager.h"
+#include "core/gdbstub/gdbstub.h"
+#include "core/settings.h"
+
+namespace Core {
+namespace {
+void RunCpuCore(const System& system, Cpu& cpu_state) {
+    while (system.IsPoweredOn()) {
+        cpu_state.RunLoop(true);
+    }
+}
+} // Anonymous namespace
+
+CpuCoreManager::CpuCoreManager() = default;
+CpuCoreManager::~CpuCoreManager() = default;
+
+void CpuCoreManager::Initialize(System& system) {
+    barrier = std::make_unique<CpuBarrier>();
+    exclusive_monitor = Cpu::MakeExclusiveMonitor(cores.size());
+
+    for (std::size_t index = 0; index < cores.size(); ++index) {
+        cores[index] = std::make_unique<Cpu>(*exclusive_monitor, *barrier, index);
+    }
+
+    // Create threads for CPU cores 1-3, and build thread_to_cpu map
+    // CPU core 0 is run on the main thread
+    thread_to_cpu[std::this_thread::get_id()] = cores[0].get();
+    if (!Settings::values.use_multi_core) {
+        return;
+    }
+
+    for (std::size_t index = 0; index < core_threads.size(); ++index) {
+        core_threads[index] = std::make_unique<std::thread>(RunCpuCore, std::cref(system),
+                                                            std::ref(*cores[index + 1]));
+        thread_to_cpu[core_threads[index]->get_id()] = cores[index + 1].get();
+    }
+}
+
+void CpuCoreManager::Shutdown() {
+    barrier->NotifyEnd();
+    if (Settings::values.use_multi_core) {
+        for (auto& thread : core_threads) {
+            thread->join();
+            thread.reset();
+        }
+    }
+
+    thread_to_cpu.clear();
+    for (auto& cpu_core : cores) {
+        cpu_core.reset();
+    }
+
+    exclusive_monitor.reset();
+    barrier.reset();
+}
+
+Cpu& CpuCoreManager::GetCore(std::size_t index) {
+    return *cores.at(index);
+}
+
+const Cpu& CpuCoreManager::GetCore(std::size_t index) const {
+    return *cores.at(index);
+}
+
+ExclusiveMonitor& CpuCoreManager::GetExclusiveMonitor() {
+    return *exclusive_monitor;
+}
+
+const ExclusiveMonitor& CpuCoreManager::GetExclusiveMonitor() const {
+    return *exclusive_monitor;
+}
+
+Cpu& CpuCoreManager::GetCurrentCore() {
+    if (Settings::values.use_multi_core) {
+        const auto& search = thread_to_cpu.find(std::this_thread::get_id());
+        ASSERT(search != thread_to_cpu.end());
+        ASSERT(search->second);
+        return *search->second;
+    }
+
+    // Otherwise, use single-threaded mode active_core variable
+    return *cores[active_core];
+}
+
+const Cpu& CpuCoreManager::GetCurrentCore() const {
+    if (Settings::values.use_multi_core) {
+        const auto& search = thread_to_cpu.find(std::this_thread::get_id());
+        ASSERT(search != thread_to_cpu.end());
+        ASSERT(search->second);
+        return *search->second;
+    }
+
+    // Otherwise, use single-threaded mode active_core variable
+    return *cores[active_core];
+}
+
+void CpuCoreManager::RunLoop(bool tight_loop) {
+    // Update thread_to_cpu in case Core 0 is run from a different host thread
+    thread_to_cpu[std::this_thread::get_id()] = cores[0].get();
+
+    if (GDBStub::IsServerEnabled()) {
+        GDBStub::HandlePacket();
+
+        // If the loop is halted and we want to step, use a tiny (1) number of instructions to
+        // execute. Otherwise, get out of the loop function.
+        if (GDBStub::GetCpuHaltFlag()) {
+            if (GDBStub::GetCpuStepFlag()) {
+                tight_loop = false;
+            } else {
+                return;
+            }
+        }
+    }
+
+    for (active_core = 0; active_core < NUM_CPU_CORES; ++active_core) {
+        cores[active_core]->RunLoop(tight_loop);
+        if (Settings::values.use_multi_core) {
+            // Cores 1-3 are run on other threads in this mode
+            break;
+        }
+    }
+
+    if (GDBStub::IsServerEnabled()) {
+        GDBStub::SetCpuStepFlag(false);
+    }
+}
+
+void CpuCoreManager::InvalidateAllInstructionCaches() {
+    for (auto& cpu : cores) {
+        cpu->ArmInterface().ClearInstructionCache();
+    }
+}
+
+} // namespace Core

src/core/cpu_core_manager.h

@@ -0,0 +1,59 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include <map>
+#include <memory>
+#include <thread>
+
+namespace Core {
+
+class Cpu;
+class CpuBarrier;
+class ExclusiveMonitor;
+class System;
+
+class CpuCoreManager {
+public:
+    CpuCoreManager();
+    CpuCoreManager(const CpuCoreManager&) = delete;
+    CpuCoreManager(CpuCoreManager&&) = delete;
+
+    ~CpuCoreManager();
+
+    CpuCoreManager& operator=(const CpuCoreManager&) = delete;
+    CpuCoreManager& operator=(CpuCoreManager&&) = delete;
+
+    void Initialize(System& system);
+    void Shutdown();
+
+    Cpu& GetCore(std::size_t index);
+    const Cpu& GetCore(std::size_t index) const;
+
+    Cpu& GetCurrentCore();
+    const Cpu& GetCurrentCore() const;
+
+    ExclusiveMonitor& GetExclusiveMonitor();
+    const ExclusiveMonitor& GetExclusiveMonitor() const;
+
+    void RunLoop(bool tight_loop);
+
+    void InvalidateAllInstructionCaches();
+
+private:
+    static constexpr std::size_t NUM_CPU_CORES = 4;
+
+    std::unique_ptr<ExclusiveMonitor> exclusive_monitor;
+    std::unique_ptr<CpuBarrier> barrier;
+    std::array<std::unique_ptr<Cpu>, NUM_CPU_CORES> cores;
+    std::array<std::unique_ptr<std::thread>, NUM_CPU_CORES - 1> core_threads;
+    std::size_t active_core{}; ///< Active core, only used in single thread mode
+
+    /// Map of guest threads to CPU cores
+    std::map<std::thread::id, Cpu*> thread_to_cpu;
+};
+
+} // namespace Core

src/core/hle/kernel/handle_table.cpp

@@ -12,12 +12,23 @@
 #include "core/hle/kernel/thread.h"
 
 namespace Kernel {
+namespace {
+constexpr u16 GetSlot(Handle handle) {
+    return handle >> 15;
+}
+
+constexpr u16 GetGeneration(Handle handle) {
+    return handle & 0x7FFF;
+}
+} // Anonymous namespace
 
 HandleTable::HandleTable() {
     next_generation = 1;
     Clear();
 }
 
+HandleTable::~HandleTable() = default;
+
 ResultVal<Handle> HandleTable::Create(SharedPtr<Object> obj) {
     DEBUG_ASSERT(obj != nullptr);
 

src/core/hle/kernel/handle_table.h

@@ -43,6 +43,7 @@ enum KernelHandle : Handle {
 class HandleTable final : NonCopyable {
 public:
     HandleTable();
+    ~HandleTable();
 
     /**
      * Allocates a handle for the given object.
@@ -89,18 +90,8 @@ public:
     void Clear();
 
 private:
-    /**
-     * This is the maximum limit of handles allowed per process in CTR-OS. It can be further
-     * reduced by ExHeader values, but this is not emulated here.
-     */
-    static const std::size_t MAX_COUNT = 4096;
-
-    static u16 GetSlot(Handle handle) {
-        return handle >> 15;
-    }
-    static u16 GetGeneration(Handle handle) {
-        return handle & 0x7FFF;
-    }
+    /// This is the maximum limit of handles allowed per process in Horizon
+    static constexpr std::size_t MAX_COUNT = 1024;
 
     /// Stores the Object referenced by the handle or null if the slot is empty.
     std::array<SharedPtr<Object>, MAX_COUNT> objects;

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

@@ -6,8 +6,6 @@
 #include <cinttypes>
 #include <cstring>
 #include <stack>
-#include "applets/applets.h"
-#include "applets/software_keyboard.h"
 #include "audio_core/audio_renderer.h"
 #include "core/core.h"
 #include "core/hle/ipc_helpers.h"
@@ -18,6 +16,9 @@
 #include "core/hle/service/am/am.h"
 #include "core/hle/service/am/applet_ae.h"
 #include "core/hle/service/am/applet_oe.h"
+#include "core/hle/service/am/applets/applets.h"
+#include "core/hle/service/am/applets/software_keyboard.h"
+#include "core/hle/service/am/applets/stub_applet.h"
 #include "core/hle/service/am/idle.h"
 #include "core/hle/service/am/omm.h"
 #include "core/hle/service/am/spsm.h"
@@ -532,8 +533,7 @@ void ICommonStateGetter::GetPerformanceMode(Kernel::HLERequestContext& ctx) {
 class ILibraryAppletAccessor final : public ServiceFramework<ILibraryAppletAccessor> {
 public:
     explicit ILibraryAppletAccessor(std::shared_ptr<Applets::Applet> applet)
-        : ServiceFramework("ILibraryAppletAccessor"), applet(std::move(applet)),
-          broker(std::make_shared<Applets::AppletDataBroker>()) {
+        : ServiceFramework("ILibraryAppletAccessor"), applet(std::move(applet)) {
         // clang-format off
         static const FunctionInfo functions[] = {
             {0, &ILibraryAppletAccessor::GetAppletStateChangedEvent, "GetAppletStateChangedEvent"},
@@ -562,7 +562,7 @@ public:
 
 private:
     void GetAppletStateChangedEvent(Kernel::HLERequestContext& ctx) {
-        const auto event = broker->GetStateChangedEvent();
+        const auto event = applet->GetBroker().GetStateChangedEvent();
         event->Signal();
 
         IPC::ResponseBuilder rb{ctx, 2, 1};
@@ -590,7 +590,7 @@ private:
     void Start(Kernel::HLERequestContext& ctx) {
         ASSERT(applet != nullptr);
 
-        applet->Initialize(broker);
+        applet->Initialize();
         applet->Execute();
 
         IPC::ResponseBuilder rb{ctx, 2};
@@ -601,7 +601,7 @@ private:
 
     void PushInData(Kernel::HLERequestContext& ctx) {
         IPC::RequestParser rp{ctx};
-        broker->PushNormalDataFromGame(*rp.PopIpcInterface<IStorage>());
+        applet->GetBroker().PushNormalDataFromGame(*rp.PopIpcInterface<IStorage>());
 
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
@@ -612,7 +612,7 @@ private:
     void PopOutData(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
 
-        const auto storage = broker->PopNormalDataToGame();
+        const auto storage = applet->GetBroker().PopNormalDataToGame();
         if (storage == nullptr) {
             rb.Push(ERR_NO_DATA_IN_CHANNEL);
             return;
@@ -626,7 +626,7 @@ private:
 
     void PushInteractiveInData(Kernel::HLERequestContext& ctx) {
         IPC::RequestParser rp{ctx};
-        broker->PushInteractiveDataFromGame(*rp.PopIpcInterface<IStorage>());
+        applet->GetBroker().PushInteractiveDataFromGame(*rp.PopIpcInterface<IStorage>());
 
         ASSERT(applet->IsInitialized());
         applet->ExecuteInteractive();
@@ -641,7 +641,7 @@ private:
     void PopInteractiveOutData(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
 
-        const auto storage = broker->PopInteractiveDataToGame();
+        const auto storage = applet->GetBroker().PopInteractiveDataToGame();
         if (storage == nullptr) {
             rb.Push(ERR_NO_DATA_IN_CHANNEL);
             return;
@@ -656,7 +656,7 @@ private:
     void GetPopOutDataEvent(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 1};
         rb.Push(RESULT_SUCCESS);
-        rb.PushCopyObjects(broker->GetNormalDataEvent());
+        rb.PushCopyObjects(applet->GetBroker().GetNormalDataEvent());
 
         LOG_DEBUG(Service_AM, "called");
     }
@@ -664,13 +664,12 @@ private:
     void GetPopInteractiveOutDataEvent(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 1};
         rb.Push(RESULT_SUCCESS);
-        rb.PushCopyObjects(broker->GetInteractiveDataEvent());
+        rb.PushCopyObjects(applet->GetBroker().GetInteractiveDataEvent());
 
         LOG_DEBUG(Service_AM, "called");
     }
 
     std::shared_ptr<Applets::Applet> applet;
-    std::shared_ptr<Applets::AppletDataBroker> broker;
 };
 
 void IStorage::Open(Kernel::HLERequestContext& ctx) {
@@ -763,8 +762,9 @@ static std::shared_ptr<Applets::Applet> GetAppletFromId(AppletId id) {
     case AppletId::SoftwareKeyboard:
         return std::make_shared<Applets::SoftwareKeyboard>();
     default:
-        UNREACHABLE_MSG("Unimplemented AppletId [{:08X}]!", static_cast<u32>(id));
-        return nullptr;
+        LOG_ERROR(Service_AM, "Unimplemented AppletId [{:08X}]! -- Falling back to stub!",
+                  static_cast<u32>(id));
+        return std::make_shared<Applets::StubApplet>();
     }
 }
 

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

@@ -98,10 +98,8 @@ Applet::Applet() = default;
 
 Applet::~Applet() = default;
 
-void Applet::Initialize(std::shared_ptr<AppletDataBroker> broker_) {
-    broker = std::move(broker_);
-
-    const auto common = broker->PopNormalDataToApplet();
+void Applet::Initialize() {
+    const auto common = broker.PopNormalDataToApplet();
     ASSERT(common != nullptr);
 
     const auto common_data = common->GetData();

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

@@ -4,14 +4,17 @@
 
 #pragma once
 
-#include <functional>
 #include <memory>
 #include <queue>
 #include "common/swap.h"
-#include "core/hle/kernel/event.h"
+#include "core/hle/kernel/kernel.h"
 
 union ResultCode;
 
+namespace Kernel {
+class Event;
+}
+
 namespace Service::AM {
 
 class IStorage;
@@ -43,19 +46,26 @@ public:
 
 private:
     // Queues are named from applet's perspective
-    std::queue<std::unique_ptr<IStorage>>
-        in_channel; // PopNormalDataToApplet and PushNormalDataFromGame
-    std::queue<std::unique_ptr<IStorage>>
-        out_channel; // PopNormalDataToGame and PushNormalDataFromApplet
-    std::queue<std::unique_ptr<IStorage>>
-        in_interactive_channel; // PopInteractiveDataToApplet and PushInteractiveDataFromGame
-    std::queue<std::unique_ptr<IStorage>>
-        out_interactive_channel; // PopInteractiveDataToGame and PushInteractiveDataFromApplet
+
+    // PopNormalDataToApplet and PushNormalDataFromGame
+    std::queue<std::unique_ptr<IStorage>> in_channel;
+
+    // PopNormalDataToGame and PushNormalDataFromApplet
+    std::queue<std::unique_ptr<IStorage>> out_channel;
+
+    // PopInteractiveDataToApplet and PushInteractiveDataFromGame
+    std::queue<std::unique_ptr<IStorage>> in_interactive_channel;
+
+    // PopInteractiveDataToGame and PushInteractiveDataFromApplet
+    std::queue<std::unique_ptr<IStorage>> out_interactive_channel;
 
     Kernel::SharedPtr<Kernel::Event> state_changed_event;
-    Kernel::SharedPtr<Kernel::Event> pop_out_data_event; // Signaled on PushNormalDataFromApplet
-    Kernel::SharedPtr<Kernel::Event>
-        pop_interactive_out_data_event; // Signaled on PushInteractiveDataFromApplet
+
+    // Signaled on PushNormalDataFromApplet
+    Kernel::SharedPtr<Kernel::Event> pop_out_data_event;
+
+    // Signaled on PushInteractiveDataFromApplet
+    Kernel::SharedPtr<Kernel::Event> pop_interactive_out_data_event;
 };
 
 class Applet {
@@ -63,7 +73,7 @@ public:
     Applet();
     virtual ~Applet();
 
-    virtual void Initialize(std::shared_ptr<AppletDataBroker> broker);
+    virtual void Initialize();
 
     virtual bool TransactionComplete() const = 0;
     virtual ResultCode GetStatus() const = 0;
@@ -74,6 +84,14 @@ public:
         return initialized;
     }
 
+    AppletDataBroker& GetBroker() {
+        return broker;
+    }
+
+    const AppletDataBroker& GetBroker() const {
+        return broker;
+    }
+
 protected:
     struct CommonArguments {
         u32_le arguments_version;
@@ -85,8 +103,8 @@ protected:
     };
     static_assert(sizeof(CommonArguments) == 0x20, "CommonArguments has incorrect size.");
 
-    CommonArguments common_args;
-    std::shared_ptr<AppletDataBroker> broker;
+    CommonArguments common_args{};
+    AppletDataBroker broker;
     bool initialized = false;
 };
 

src/core/hle/service/am/applets/software_keyboard.cpp

@@ -42,21 +42,21 @@ SoftwareKeyboard::SoftwareKeyboard() = default;
 
 SoftwareKeyboard::~SoftwareKeyboard() = default;
 
-void SoftwareKeyboard::Initialize(std::shared_ptr<AppletDataBroker> broker_) {
+void SoftwareKeyboard::Initialize() {
     complete = false;
     initial_text.clear();
     final_data.clear();
 
-    Applet::Initialize(std::move(broker_));
+    Applet::Initialize();
 
-    const auto keyboard_config_storage = broker->PopNormalDataToApplet();
+    const auto keyboard_config_storage = broker.PopNormalDataToApplet();
     ASSERT(keyboard_config_storage != nullptr);
     const auto& keyboard_config = keyboard_config_storage->GetData();
 
     ASSERT(keyboard_config.size() >= sizeof(KeyboardConfig));
     std::memcpy(&config, keyboard_config.data(), sizeof(KeyboardConfig));
 
-    const auto work_buffer_storage = broker->PopNormalDataToApplet();
+    const auto work_buffer_storage = broker.PopNormalDataToApplet();
     ASSERT(work_buffer_storage != nullptr);
     const auto& work_buffer = work_buffer_storage->GetData();
 
@@ -81,7 +81,7 @@ void SoftwareKeyboard::ExecuteInteractive() {
     if (complete)
         return;
 
-    const auto storage = broker->PopInteractiveDataToApplet();
+    const auto storage = broker.PopInteractiveDataToApplet();
     ASSERT(storage != nullptr);
     const auto data = storage->GetData();
     const auto status = static_cast<bool>(data[0]);
@@ -95,13 +95,13 @@ void SoftwareKeyboard::ExecuteInteractive() {
         std::memcpy(string.data(), data.data() + 4, string.size() * 2);
         frontend.SendTextCheckDialog(
             Common::UTF16StringFromFixedZeroTerminatedBuffer(string.data(), string.size()),
-            [this] { broker->SignalStateChanged(); });
+            [this] { broker.SignalStateChanged(); });
     }
 }
 
 void SoftwareKeyboard::Execute() {
     if (complete) {
-        broker->PushNormalDataFromApplet(IStorage{final_data});
+        broker.PushNormalDataFromApplet(IStorage{final_data});
         return;
     }
 
@@ -145,17 +145,17 @@ void SoftwareKeyboard::WriteText(std::optional<std::u16string> text) {
         final_data = output_main;
 
         if (complete) {
-            broker->PushNormalDataFromApplet(IStorage{output_main});
+            broker.PushNormalDataFromApplet(IStorage{output_main});
         } else {
-            broker->PushInteractiveDataFromApplet(IStorage{output_sub});
+            broker.PushInteractiveDataFromApplet(IStorage{output_sub});
         }
 
-        broker->SignalStateChanged();
+        broker.SignalStateChanged();
     } else {
         output_main[0] = 1;
         complete = true;
-        broker->PushNormalDataFromApplet(IStorage{output_main});
-        broker->SignalStateChanged();
+        broker.PushNormalDataFromApplet(IStorage{output_main});
+        broker.SignalStateChanged();
     }
 }
 } // namespace Service::AM::Applets

src/core/hle/service/am/applets/software_keyboard.h

@@ -4,7 +4,12 @@
 
 #pragma once
 
+#include <array>
+#include <string>
+#include <vector>
+
 #include "common/common_funcs.h"
+#include "common/swap.h"
 #include "core/hle/service/am/am.h"
 #include "core/hle/service/am/applets/applets.h"
 
@@ -50,7 +55,7 @@ public:
     SoftwareKeyboard();
     ~SoftwareKeyboard() override;
 
-    void Initialize(std::shared_ptr<AppletDataBroker> broker) override;
+    void Initialize() override;
 
     bool TransactionComplete() const override;
     ResultCode GetStatus() const override;

src/core/hle/service/am/applets/stub_applet.cpp

@@ -0,0 +1,70 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <string>
+
+#include "common/hex_util.h"
+#include "common/logging/log.h"
+#include "core/hle/result.h"
+#include "core/hle/service/am/am.h"
+#include "core/hle/service/am/applets/stub_applet.h"
+
+namespace Service::AM::Applets {
+
+static void LogCurrentStorage(AppletDataBroker& broker, std::string prefix) {
+    std::unique_ptr<IStorage> storage = broker.PopNormalDataToApplet();
+    for (; storage != nullptr; storage = broker.PopNormalDataToApplet()) {
+        const auto data = storage->GetData();
+        LOG_INFO(Service_AM,
+                 "called (STUBBED), during {} recieved normal data with size={:08X}, data={}",
+                 prefix, data.size(), Common::HexVectorToString(data));
+    }
+
+    storage = broker.PopInteractiveDataToApplet();
+    for (; storage != nullptr; storage = broker.PopInteractiveDataToApplet()) {
+        const auto data = storage->GetData();
+        LOG_INFO(Service_AM,
+                 "called (STUBBED), during {} recieved interactive data with size={:08X}, data={}",
+                 prefix, data.size(), Common::HexVectorToString(data));
+    }
+}
+
+StubApplet::StubApplet() = default;
+
+StubApplet::~StubApplet() = default;
+
+void StubApplet::Initialize() {
+    LOG_WARNING(Service_AM, "called (STUBBED)");
+    Applet::Initialize();
+    LogCurrentStorage(broker, "Initialize");
+}
+
+bool StubApplet::TransactionComplete() const {
+    LOG_WARNING(Service_AM, "called (STUBBED)");
+    return true;
+}
+
+ResultCode StubApplet::GetStatus() const {
+    LOG_WARNING(Service_AM, "called (STUBBED)");
+    return RESULT_SUCCESS;
+}
+
+void StubApplet::ExecuteInteractive() {
+    LOG_WARNING(Service_AM, "called (STUBBED)");
+    LogCurrentStorage(broker, "ExecuteInteractive");
+
+    broker.PushNormalDataFromApplet(IStorage{std::vector<u8>(0x1000)});
+    broker.PushInteractiveDataFromApplet(IStorage{std::vector<u8>(0x1000)});
+    broker.SignalStateChanged();
+}
+
+void StubApplet::Execute() {
+    LOG_WARNING(Service_AM, "called (STUBBED)");
+    LogCurrentStorage(broker, "Execute");
+
+    broker.PushNormalDataFromApplet(IStorage{std::vector<u8>(0x1000)});
+    broker.PushInteractiveDataFromApplet(IStorage{std::vector<u8>(0x1000)});
+    broker.SignalStateChanged();
+}
+} // namespace Service::AM::Applets

src/core/hle/service/am/applets/stub_applet.h

@@ -0,0 +1,24 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "core/hle/service/am/applets/applets.h"
+
+namespace Service::AM::Applets {
+
+class StubApplet final : public Applet {
+public:
+    StubApplet();
+    ~StubApplet() override;
+
+    void Initialize() override;
+
+    bool TransactionComplete() const override;
+    ResultCode GetStatus() const override;
+    void ExecuteInteractive() override;
+    void Execute() override;
+};
+
+} // namespace Service::AM::Applets

src/core/hle/service/audio/audout_u.cpp

@@ -44,8 +44,10 @@ enum class AudioState : u32 {
 
 class IAudioOut final : public ServiceFramework<IAudioOut> {
 public:
-    IAudioOut(AudoutParams audio_params, AudioCore::AudioOut& audio_core)
-        : ServiceFramework("IAudioOut"), audio_core(audio_core), audio_params(audio_params) {
+    IAudioOut(AudoutParams audio_params, AudioCore::AudioOut& audio_core, std::string&& device_name,
+              std::string&& unique_name)
+        : ServiceFramework("IAudioOut"), audio_core(audio_core), audio_params(audio_params),
+          device_name(std::move(device_name)) {
 
         static const FunctionInfo functions[] = {
             {0, &IAudioOut::GetAudioOutState, "GetAudioOutState"},
@@ -69,7 +71,7 @@ public:
             Kernel::Event::Create(kernel, Kernel::ResetType::Sticky, "IAudioOutBufferReleased");
 
         stream = audio_core.OpenStream(audio_params.sample_rate, audio_params.channel_count,
-                                       "IAudioOut", [=]() { buffer_event->Signal(); });
+                                       std::move(unique_name), [=]() { buffer_event->Signal(); });
     }
 
 private:
@@ -177,6 +179,7 @@ private:
 
     AudioCore::AudioOut& audio_core;
     AudioCore::StreamPtr stream;
+    std::string device_name;
 
     AudoutParams audio_params{};
 
@@ -199,7 +202,15 @@ void AudOutU::ListAudioOutsImpl(Kernel::HLERequestContext& ctx) {
 void AudOutU::OpenAudioOutImpl(Kernel::HLERequestContext& ctx) {
     LOG_DEBUG(Service_Audio, "called");
 
-    ctx.WriteBuffer(DefaultDevice);
+    const auto device_name_data{ctx.ReadBuffer()};
+    std::string device_name;
+    if (device_name_data[0] != '\0') {
+        device_name.assign(device_name_data.begin(), device_name_data.end());
+    } else {
+        device_name.assign(DefaultDevice.begin(), DefaultDevice.end());
+    }
+    ctx.WriteBuffer(device_name);
+
     IPC::RequestParser rp{ctx};
     auto params{rp.PopRaw<AudoutParams>()};
     if (params.channel_count <= 2) {
@@ -212,10 +223,9 @@ void AudOutU::OpenAudioOutImpl(Kernel::HLERequestContext& ctx) {
         params.sample_rate = DefaultSampleRate;
     }
 
-    // TODO(bunnei): Support more than one IAudioOut interface. When we add this, ListAudioOutsImpl
-    // will likely need to be updated as well.
-    ASSERT_MSG(!audio_out_interface, "Unimplemented");
-    audio_out_interface = std::make_shared<IAudioOut>(params, *audio_core);
+    std::string unique_name{fmt::format("{}-{}", device_name, audio_out_interfaces.size())};
+    auto audio_out_interface = std::make_shared<IAudioOut>(
+        params, *audio_core, std::move(device_name), std::move(unique_name));
 
     IPC::ResponseBuilder rb{ctx, 6, 0, 1};
     rb.Push(RESULT_SUCCESS);
@@ -224,6 +234,8 @@ void AudOutU::OpenAudioOutImpl(Kernel::HLERequestContext& ctx) {
     rb.Push<u32>(static_cast<u32>(AudioCore::Codec::PcmFormat::Int16));
     rb.Push<u32>(static_cast<u32>(AudioState::Stopped));
     rb.PushIpcInterface<Audio::IAudioOut>(audio_out_interface);
+
+    audio_out_interfaces.push_back(std::move(audio_out_interface));
 }
 
 AudOutU::AudOutU() : ServiceFramework("audout:u") {

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

@@ -4,6 +4,7 @@
 
 #pragma once
 
+#include <vector>
 #include "core/hle/service/service.h"
 
 namespace AudioCore {
@@ -24,7 +25,7 @@ public:
     ~AudOutU() override;
 
 private:
-    std::shared_ptr<IAudioOut> audio_out_interface;
+    std::vector<std::shared_ptr<IAudioOut>> audio_out_interfaces;
     std::unique_ptr<AudioCore::AudioOut> audio_core;
 
     void ListAudioOutsImpl(Kernel::HLERequestContext& ctx);

src/core/hle/service/hid/controllers/debug_pad.cpp

@@ -71,8 +71,9 @@ void Controller_DebugPad::OnUpdate(u8* data, std::size_t size) {
 
 void Controller_DebugPad::OnLoadInputDevices() {
     std::transform(Settings::values.debug_pad_buttons.begin(),
-                   Settings::values.debug_pad_buttons.end(), buttons.begin(),
-                   Input::CreateDevice<Input::ButtonDevice>);
+                   Settings::values.debug_pad_buttons.begin() +
+                       Settings::NativeButton::NUM_BUTTONS_HID,
+                   buttons.begin(), Input::CreateDevice<Input::ButtonDevice>);
     std::transform(Settings::values.debug_pad_analogs.begin(),
                    Settings::values.debug_pad_analogs.end(), analogs.begin(),
                    Input::CreateDevice<Input::AnalogDevice>);

src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.cpp

@@ -5,6 +5,8 @@
 #include <cstring>
 #include "common/assert.h"
 #include "common/logging/log.h"
+#include "core/core_timing.h"
+#include "core/core_timing_util.h"
 #include "core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.h"
 
 namespace Service::Nvidia::Devices {
@@ -33,6 +35,8 @@ u32 nvhost_ctrl_gpu::ioctl(Ioctl command, const std::vector<u8>& input, std::vec
         return ZBCQueryTable(input, output);
     case IoctlCommand::IocFlushL2:
         return FlushL2(input, output);
+    case IoctlCommand::IocGetGpuTime:
+        return GetGpuTime(input, output);
     }
     UNIMPLEMENTED_MSG("Unimplemented ioctl");
     return 0;
@@ -169,4 +173,13 @@ u32 nvhost_ctrl_gpu::FlushL2(const std::vector<u8>& input, std::vector<u8>& outp
     return 0;
 }
 
+u32 nvhost_ctrl_gpu::GetGpuTime(const std::vector<u8>& input, std::vector<u8>& output) {
+    LOG_DEBUG(Service_NVDRV, "called");
+    IoctlGetGpuTime params{};
+    std::memcpy(&params, input.data(), input.size());
+    params.gpu_time = CoreTiming::cyclesToNs(CoreTiming::GetTicks());
+    std::memcpy(output.data(), &params, output.size());
+    return 0;
+}
+
 } // namespace Service::Nvidia::Devices

src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.h

@@ -156,6 +156,11 @@ private:
     };
     static_assert(sizeof(IoctlFlushL2) == 8, "IoctlFlushL2 is incorrect size");
 
+    struct IoctlGetGpuTime {
+        u64_le gpu_time;
+    };
+    static_assert(sizeof(IoctlGetGpuTime) == 8, "IoctlGetGpuTime is incorrect size");
+
     u32 GetCharacteristics(const std::vector<u8>& input, std::vector<u8>& output);
     u32 GetTPCMasks(const std::vector<u8>& input, std::vector<u8>& output);
     u32 GetActiveSlotMask(const std::vector<u8>& input, std::vector<u8>& output);
@@ -164,6 +169,7 @@ private:
     u32 ZBCSetTable(const std::vector<u8>& input, std::vector<u8>& output);
     u32 ZBCQueryTable(const std::vector<u8>& input, std::vector<u8>& output);
     u32 FlushL2(const std::vector<u8>& input, std::vector<u8>& output);
+    u32 GetGpuTime(const std::vector<u8>& input, std::vector<u8>& output);
 };
 
 } // namespace Service::Nvidia::Devices

src/video_core/engines/maxwell_3d.cpp

@@ -69,6 +69,15 @@ void Maxwell3D::InitializeRegisterDefaults() {
     // TODO(Rodrigo): Most games do not set a point size. I think this is a case of a
     // register carrying a default value. Assume it's OpenGL's default (1).
     regs.point_size = 1.0f;
+
+    // TODO(bunnei): Some games do not initialize the color masks (e.g. Sonic Mania). Assuming a
+    // default of enabled fixes rendering here.
+    for (std::size_t color_mask = 0; color_mask < Regs::NumRenderTargets; color_mask++) {
+        regs.color_mask[color_mask].R.Assign(1);
+        regs.color_mask[color_mask].G.Assign(1);
+        regs.color_mask[color_mask].B.Assign(1);
+        regs.color_mask[color_mask].A.Assign(1);
+    }
 }
 
 void Maxwell3D::CallMacroMethod(u32 method, std::vector<u32> parameters) {

src/video_core/engines/maxwell_3d.h

@@ -389,6 +389,13 @@ public:
                 ReverseSubtract = 3,
                 Min = 4,
                 Max = 5,
+
+                // These values are used by Nouveau and some games.
+                AddGL = 0x8006,
+                SubtractGL = 0x8007,
+                ReverseSubtractGL = 0x8008,
+                MinGL = 0x800a,
+                MaxGL = 0x800b
             };
 
             enum class Factor : u32 {

src/video_core/engines/shader_bytecode.h

@@ -153,6 +153,7 @@ enum class PredCondition : u64 {
     NotEqual = 5,
     GreaterEqual = 6,
     LessThanWithNan = 9,
+    LessEqualWithNan = 11,
     GreaterThanWithNan = 12,
     NotEqualWithNan = 13,
     GreaterEqualWithNan = 14,

src/video_core/gpu.cpp

@@ -17,6 +17,8 @@ u32 FramebufferConfig::BytesPerPixel(PixelFormat format) {
     switch (format) {
     case PixelFormat::ABGR8:
         return 4;
+    default:
+        return 4;
     }
 
     UNREACHABLE();

src/video_core/macro_interpreter.cpp

@@ -35,6 +35,7 @@ void MacroInterpreter::Reset() {
     // The next parameter index starts at 1, because $r1 already has the value of the first
     // parameter.
     next_parameter_index = 1;
+    carry_flag = false;
 }
 
 bool MacroInterpreter::Step(u32 offset, bool is_delay_slot) {
@@ -135,14 +136,28 @@ MacroInterpreter::Opcode MacroInterpreter::GetOpcode(u32 offset) const {
     return {macro_memory[offset + pc / sizeof(u32)]};
 }
 
-u32 MacroInterpreter::GetALUResult(ALUOperation operation, u32 src_a, u32 src_b) const {
+u32 MacroInterpreter::GetALUResult(ALUOperation operation, u32 src_a, u32 src_b) {
     switch (operation) {
-    case ALUOperation::Add:
-        return src_a + src_b;
-    // TODO(Subv): Implement AddWithCarry
-    case ALUOperation::Subtract:
-        return src_a - src_b;
-    // TODO(Subv): Implement SubtractWithBorrow
+    case ALUOperation::Add: {
+        const u64 result{static_cast<u64>(src_a) + src_b};
+        carry_flag = result > 0xffffffff;
+        return static_cast<u32>(result);
+    }
+    case ALUOperation::AddWithCarry: {
+        const u64 result{static_cast<u64>(src_a) + src_b + (carry_flag ? 1ULL : 0ULL)};
+        carry_flag = result > 0xffffffff;
+        return static_cast<u32>(result);
+    }
+    case ALUOperation::Subtract: {
+        const u64 result{static_cast<u64>(src_a) - src_b};
+        carry_flag = result < 0x100000000;
+        return static_cast<u32>(result);
+    }
+    case ALUOperation::SubtractWithBorrow: {
+        const u64 result{static_cast<u64>(src_a) - src_b - (carry_flag ? 0ULL : 1ULL)};
+        carry_flag = result < 0x100000000;
+        return static_cast<u32>(result);
+    }
     case ALUOperation::Xor:
         return src_a ^ src_b;
     case ALUOperation::Or:

src/video_core/macro_interpreter.h

@@ -117,7 +117,7 @@ private:
     bool Step(u32 offset, bool is_delay_slot);
 
     /// Calculates the result of an ALU operation. src_a OP src_b;
-    u32 GetALUResult(ALUOperation operation, u32 src_a, u32 src_b) const;
+    u32 GetALUResult(ALUOperation operation, u32 src_a, u32 src_b);
 
     /// Performs the result operation on the input result and stores it in the specified register
     /// (if necessary).
@@ -165,5 +165,7 @@ private:
     std::vector<u32> parameters;
     /// Index of the next parameter that will be fetched by the 'parm' instruction.
     u32 next_parameter_index = 0;
+
+    bool carry_flag{};
 };
 } // namespace Tegra

src/video_core/renderer_opengl/gl_rasterizer_cache.cpp

@@ -1275,6 +1275,31 @@ Surface RasterizerCacheOpenGL::GetUncachedSurface(const SurfaceParams& params) {
     return surface;
 }
 
+void RasterizerCacheOpenGL::FastLayeredCopySurface(const Surface& src_surface,
+                                                   const Surface& dst_surface) {
+    const auto& init_params{src_surface->GetSurfaceParams()};
+    const auto& dst_params{dst_surface->GetSurfaceParams()};
+    VAddr address = init_params.addr;
+    const std::size_t layer_size = dst_params.LayerMemorySize();
+    for (u32 layer = 0; layer < dst_params.depth; layer++) {
+        for (u32 mipmap = 0; mipmap < dst_params.max_mip_level; mipmap++) {
+            const VAddr sub_address = address + dst_params.GetMipmapLevelOffset(mipmap);
+            const Surface& copy = TryGet(sub_address);
+            if (!copy)
+                continue;
+            const auto& src_params{copy->GetSurfaceParams()};
+            const u32 width{std::min(src_params.width, dst_params.MipWidth(mipmap))};
+            const u32 height{std::min(src_params.height, dst_params.MipHeight(mipmap))};
+
+            glCopyImageSubData(copy->Texture().handle, SurfaceTargetToGL(src_params.target), 0, 0,
+                               0, 0, dst_surface->Texture().handle,
+                               SurfaceTargetToGL(dst_params.target), mipmap, 0, 0, layer, width,
+                               height, 1);
+        }
+        address += layer_size;
+    }
+}
+
 void RasterizerCacheOpenGL::FermiCopySurface(
     const Tegra::Engines::Fermi2D::Regs::Surface& src_config,
     const Tegra::Engines::Fermi2D::Regs::Surface& dst_config) {
@@ -1340,11 +1365,13 @@ Surface RasterizerCacheOpenGL::RecreateSurface(const Surface& old_surface,
             CopySurface(old_surface, new_surface, copy_pbo.handle);
         }
         break;
-    case SurfaceTarget::TextureCubemap:
     case SurfaceTarget::Texture3D:
+        AccurateCopySurface(old_surface, new_surface);
+        break;
+    case SurfaceTarget::TextureCubemap:
     case SurfaceTarget::Texture2DArray:
     case SurfaceTarget::TextureCubeArray:
-        AccurateCopySurface(old_surface, new_surface);
+        FastLayeredCopySurface(old_surface, new_surface);
         break;
     default:
         LOG_CRITICAL(Render_OpenGL, "Unimplemented surface target={}",

src/video_core/renderer_opengl/gl_rasterizer_cache.h

@@ -350,6 +350,7 @@ private:
 
     /// Performs a slow but accurate surface copy, flushing to RAM and reinterpreting the data
     void AccurateCopySurface(const Surface& src_surface, const Surface& dst_surface);
+    void FastLayeredCopySurface(const Surface& src_surface, const Surface& dst_surface);
 
     /// The surface reserve is a "backup" cache, this is where we put unique surfaces that have
     /// previously been used. This is to prevent surfaces from being constantly created and

src/video_core/renderer_opengl/gl_shader_cache.cpp

@@ -84,6 +84,7 @@ CachedShader::CachedShader(VAddr addr, Maxwell::ShaderProgram program_type)
     }
 
     entries = program_result.second;
+    shader_length = entries.shader_length;
 
     if (program_type != Maxwell::ShaderProgram::Geometry) {
         OGLShader shader;

src/video_core/renderer_opengl/gl_shader_cache.h

@@ -30,7 +30,7 @@ public:
     }
 
     std::size_t GetSizeInBytes() const override {
-        return GLShader::MAX_PROGRAM_CODE_LENGTH * sizeof(u64);
+        return shader_length;
     }
 
     // We do not have to flush this cache as things in it are never modified by us.
@@ -82,6 +82,7 @@ private:
                                u32 max_vertices, const std::string& debug_name);
 
     VAddr addr;
+    std::size_t shader_length;
     Maxwell::ShaderProgram program_type;
     GLShader::ShaderSetup setup;
     GLShader::ShaderEntries entries;

src/video_core/renderer_opengl/gl_shader_gen.h

@@ -163,6 +163,7 @@ private:
 struct ShaderEntries {
     std::vector<ConstBufferEntry> const_buffer_entries;
     std::vector<SamplerEntry> texture_samplers;
+    std::size_t shader_length;
 };
 
 using ProgramResult = std::pair<std::string, ShaderEntries>;

src/video_core/renderer_opengl/maxwell_to_gl.h

@@ -218,14 +218,19 @@ inline GLenum DepthCompareFunc(Tegra::Texture::DepthCompareFunc func) {
 inline GLenum BlendEquation(Maxwell::Blend::Equation equation) {
     switch (equation) {
     case Maxwell::Blend::Equation::Add:
+    case Maxwell::Blend::Equation::AddGL:
         return GL_FUNC_ADD;
     case Maxwell::Blend::Equation::Subtract:
+    case Maxwell::Blend::Equation::SubtractGL:
         return GL_FUNC_SUBTRACT;
     case Maxwell::Blend::Equation::ReverseSubtract:
+    case Maxwell::Blend::Equation::ReverseSubtractGL:
         return GL_FUNC_REVERSE_SUBTRACT;
     case Maxwell::Blend::Equation::Min:
+    case Maxwell::Blend::Equation::MinGL:
         return GL_MIN;
     case Maxwell::Blend::Equation::Max:
+    case Maxwell::Blend::Equation::MaxGL:
         return GL_MAX;
     }
     LOG_ERROR(Render_OpenGL, "Unimplemented blend equation={}", static_cast<u32>(equation));

src/video_core/renderer_opengl/renderer_opengl.cpp

@@ -304,6 +304,12 @@ void RendererOpenGL::ConfigureFramebufferTexture(TextureInfo& texture,
         gl_framebuffer_data.resize(texture.width * texture.height * 4);
         break;
     default:
+        internal_format = GL_RGBA;
+        texture.gl_format = GL_RGBA;
+        texture.gl_type = GL_UNSIGNED_INT_8_8_8_8_REV;
+        gl_framebuffer_data.resize(texture.width * texture.height * 4);
+        LOG_CRITICAL(Render_OpenGL, "Unknown framebuffer pixel format: {}",
+                     static_cast<u32>(framebuffer.pixel_format));
         UNREACHABLE();
     }
 

src/yuzu/applets/software_keyboard.cpp

@@ -82,8 +82,8 @@ QtSoftwareKeyboardDialog::QtSoftwareKeyboardDialog(
                            : QString::fromStdU16String(parameters.submit_text),
                        QDialogButtonBox::AcceptRole);
 
-    connect(buttons, &QDialogButtonBox::accepted, this, &QtSoftwareKeyboardDialog::Submit);
-    connect(buttons, &QDialogButtonBox::rejected, this, &QtSoftwareKeyboardDialog::Reject);
+    connect(buttons, &QDialogButtonBox::accepted, this, &QtSoftwareKeyboardDialog::accept);
+    connect(buttons, &QDialogButtonBox::rejected, this, &QtSoftwareKeyboardDialog::reject);
     layout->addWidget(header_label);
     layout->addWidget(sub_label);
     layout->addWidget(guide_label);
@@ -96,16 +96,16 @@ QtSoftwareKeyboardDialog::QtSoftwareKeyboardDialog(
 
 QtSoftwareKeyboardDialog::~QtSoftwareKeyboardDialog() = default;
 
-void QtSoftwareKeyboardDialog::Submit() {
+void QtSoftwareKeyboardDialog::accept() {
     ok = true;
     text = line_edit->text().toStdU16String();
-    accept();
+    QDialog::accept();
 }
 
-void QtSoftwareKeyboardDialog::Reject() {
+void QtSoftwareKeyboardDialog::reject() {
     ok = false;
     text.clear();
-    accept();
+    QDialog::reject();
 }
 
 std::u16string QtSoftwareKeyboardDialog::GetText() const {
@@ -129,13 +129,13 @@ QtSoftwareKeyboard::~QtSoftwareKeyboard() = default;
 
 void QtSoftwareKeyboard::RequestText(std::function<void(std::optional<std::u16string>)> out,
                                      Core::Frontend::SoftwareKeyboardParameters parameters) const {
-    text_output = out;
+    text_output = std::move(out);
     emit MainWindowGetText(parameters);
 }
 
 void QtSoftwareKeyboard::SendTextCheckDialog(std::u16string error_message,
                                              std::function<void()> finished_check) const {
-    this->finished_check = finished_check;
+    this->finished_check = std::move(finished_check);
     emit MainWindowTextCheckDialog(error_message);
 }
 

src/yuzu/applets/software_keyboard.h

@@ -33,8 +33,8 @@ public:
                              Core::Frontend::SoftwareKeyboardParameters parameters);
     ~QtSoftwareKeyboardDialog() override;
 
-    void Submit();
-    void Reject();
+    void accept() override;
+    void reject() override;
 
     std::u16string GetText() const;
     bool GetStatus() const;
@@ -70,11 +70,10 @@ signals:
     void MainWindowGetText(Core::Frontend::SoftwareKeyboardParameters parameters) const;
     void MainWindowTextCheckDialog(std::u16string error_message) const;
 
-public slots:
+private:
     void MainWindowFinishedText(std::optional<std::u16string> text);
     void MainWindowFinishedCheckDialog();
 
-private:
     mutable std::function<void(std::optional<std::u16string>)> text_output;
     mutable std::function<void()> finished_check;
 };