Ignore asserts - DO NOT CHECK IN

gl_rasterizer: Workaround for when exceeding max UBO size.

.travis.yml

@@ -42,3 +42,7 @@ notifications:
   webhooks:
     urls:
       - https://api.yuzu-emu.org/code/travis/notify
+
+cache:
+  directories:
+    - $HOME/.ccache

.travis/linux/build.sh

@@ -1,3 +1,3 @@
 #!/bin/bash -ex
 
-docker run -v $(pwd):/yuzu ubuntu:18.04 /bin/bash /yuzu/.travis/linux/docker.sh
+docker run -e CCACHE_DIR=/ccache -v $HOME/.ccache:/ccache -v $(pwd):/yuzu ubuntu:18.04 /bin/bash /yuzu/.travis/linux/docker.sh

.travis/linux/docker.sh

@@ -1,16 +1,18 @@
 #!/bin/bash -ex
 
 apt-get update
-apt-get install -y build-essential git libqt5opengl5-dev libsdl2-dev libssl-dev python qtbase5-dev wget
-
-# Get a recent version of CMake
-wget https://cmake.org/files/v3.10/cmake-3.10.1-Linux-x86_64.sh
-sh cmake-3.10.1-Linux-x86_64.sh --exclude-subdir --prefix=/ --skip-license
+apt-get install --no-install-recommends -y build-essential git libqt5opengl5-dev libsdl2-dev libssl-dev python qtbase5-dev wget cmake ninja-build ccache
 
 cd /yuzu
 
+export PATH=/usr/lib/ccache:$PATH
+ln -sf /usr/bin/ccache /usr/lib/ccache/cc
+ln -sf /usr/bin/ccache /usr/lib/ccache/c++
 mkdir build && cd build
-cmake .. -DYUZU_BUILD_UNICORN=ON -DCMAKE_BUILD_TYPE=Release
-make -j4
+ccache --show-stats > ccache_before
+cmake .. -DYUZU_BUILD_UNICORN=ON -DCMAKE_BUILD_TYPE=Release -G Ninja
+ninja
+ccache --show-stats > ccache_after
+diff -U100 ccache_before ccache_after || true
 
 ctest -VV -C Release

.travis/macos/build.sh

@@ -7,8 +7,12 @@ export Qt5_DIR=$(brew --prefix)/opt/qt5
 export UNICORNDIR=$(pwd)/externals/unicorn
 
 mkdir build && cd build
+export PATH=/usr/local/opt/ccache/libexec:$PATH
+ccache --show-stats > ccache_before
 cmake --version
 cmake .. -DYUZU_BUILD_UNICORN=ON -DCMAKE_BUILD_TYPE=Release
 make -j4
+ccache --show-stats > ccache_after
+diff -U100 ccache_before ccache_after || true
 
 ctest -VV -C Release

.travis/macos/deps.sh

@@ -1,5 +1,5 @@
 #!/bin/sh -ex
 
 brew update
-brew install dylibbundler p7zip qt5 sdl2
+brew install dylibbundler p7zip qt5 sdl2 ccache
 brew outdated cmake || brew upgrade cmake

src/common/assert.h

@@ -28,19 +28,12 @@ __declspec(noinline, noreturn)
 }
 
 #define ASSERT(_a_)                                                                                \
-    do                                                                                             \
-        if (!(_a_)) {                                                                              \
-            assert_noinline_call([] { NGLOG_CRITICAL(Debug, "Assertion Failed!"); });              \
-        }                                                                                          \
-    while (0)
+    if (!(_a_)) {                                                                                  \
+    }
 
 #define ASSERT_MSG(_a_, ...)                                                                       \
-    do                                                                                             \
-        if (!(_a_)) {                                                                              \
-            assert_noinline_call(                                                                  \
-                [&] { NGLOG_CRITICAL(Debug, "Assertion Failed!\n" __VA_ARGS__); });                \
-        }                                                                                          \
-    while (0)
+    if (!(_a_)) {                                                                                  \
+    }
 
 #define UNREACHABLE() ASSERT_MSG(false, "Unreachable code!")
 #define UNREACHABLE_MSG(...) ASSERT_MSG(false, __VA_ARGS__)

src/common/logging/backend.cpp

@@ -41,6 +41,7 @@ namespace Log {
     SUB(Service, FS)                                                                               \
     SUB(Service, HID)                                                                              \
     SUB(Service, LM)                                                                               \
+    SUB(Service, MM)                                                                               \
     SUB(Service, NFP)                                                                              \
     SUB(Service, NIFM)                                                                             \
     SUB(Service, NS)                                                                               \

src/common/logging/log.h

@@ -61,6 +61,7 @@ enum class Class : ClassType {
     Service_FS,        ///< The FS (Filesystem) service
     Service_HID,       ///< The HID (Human interface device) service
     Service_LM,        ///< The LM (Logger) service
+    Service_MM,        ///< The MM (Multimedia) service
     Service_NFP,       ///< The NFP service
     Service_NIFM,      ///< The NIFM (Network interface) service
     Service_NS,        ///< The NS services

src/common/string_util.cpp

@@ -64,6 +64,10 @@ std::string ArrayToString(const u8* data, size_t size, int line_len, bool spaces
     return oss.str();
 }
 
+std::string StringFromBuffer(const std::vector<u8>& data) {
+    return std::string(data.begin(), std::find(data.begin(), data.end(), '\0'));
+}
+
 // Turns "  hej " into "hej". Also handles tabs.
 std::string StripSpaces(const std::string& str) {
     const size_t s = str.find_first_not_of(" \t\r\n");

src/common/string_util.h

@@ -21,6 +21,8 @@ std::string ToUpper(std::string str);
 
 std::string ArrayToString(const u8* data, size_t size, int line_len = 20, bool spaces = true);
 
+std::string StringFromBuffer(const std::vector<u8>& data);
+
 std::string StripSpaces(const std::string& s);
 std::string StripQuotes(const std::string& s);
 

src/core/CMakeLists.txt

@@ -40,6 +40,8 @@ add_library(core STATIC
     hle/config_mem.h
     hle/ipc.h
     hle/ipc_helpers.h
+    hle/kernel/address_arbiter.cpp
+    hle/kernel/address_arbiter.h
     hle/kernel/client_port.cpp
     hle/kernel/client_port.h
     hle/kernel/client_session.cpp
@@ -148,6 +150,8 @@ add_library(core STATIC
     hle/service/hid/hid.h
     hle/service/lm/lm.cpp
     hle/service/lm/lm.h
+    hle/service/mm/mm_u.cpp
+    hle/service/mm/mm_u.h
     hle/service/nifm/nifm.cpp
     hle/service/nifm/nifm.h
     hle/service/nifm/nifm_a.cpp
@@ -255,6 +259,8 @@ add_library(core STATIC
     loader/linker.h
     loader/loader.cpp
     loader/loader.h
+    loader/nca.cpp
+    loader/nca.h
     loader/nro.cpp
     loader/nro.h
     loader/nso.cpp

src/core/file_sys/partition_filesystem.cpp

@@ -19,13 +19,20 @@ Loader::ResultStatus PartitionFilesystem::Load(const std::string& file_path, siz
     if (file.GetSize() < sizeof(Header))
         return Loader::ResultStatus::Error;
 
+    file.Seek(offset, SEEK_SET);
     // For cartridges, HFSs can get very large, so we need to calculate the size up to
     // the actual content itself instead of just blindly reading in the entire file.
     Header pfs_header;
     if (!file.ReadBytes(&pfs_header, sizeof(Header)))
         return Loader::ResultStatus::Error;
 
-    bool is_hfs = (memcmp(pfs_header.magic.data(), "HFS", 3) == 0);
+    if (pfs_header.magic != Common::MakeMagic('H', 'F', 'S', '0') &&
+        pfs_header.magic != Common::MakeMagic('P', 'F', 'S', '0')) {
+        return Loader::ResultStatus::ErrorInvalidFormat;
+    }
+
+    bool is_hfs = pfs_header.magic == Common::MakeMagic('H', 'F', 'S', '0');
+
     size_t entry_size = is_hfs ? sizeof(HFSEntry) : sizeof(PFSEntry);
     size_t metadata_size =
         sizeof(Header) + (pfs_header.num_entries * entry_size) + pfs_header.strtab_size;
@@ -50,7 +57,12 @@ Loader::ResultStatus PartitionFilesystem::Load(const std::vector<u8>& file_data,
         return Loader::ResultStatus::Error;
 
     memcpy(&pfs_header, &file_data[offset], sizeof(Header));
-    is_hfs = (memcmp(pfs_header.magic.data(), "HFS", 3) == 0);
+    if (pfs_header.magic != Common::MakeMagic('H', 'F', 'S', '0') &&
+        pfs_header.magic != Common::MakeMagic('P', 'F', 'S', '0')) {
+        return Loader::ResultStatus::ErrorInvalidFormat;
+    }
+
+    is_hfs = pfs_header.magic == Common::MakeMagic('H', 'F', 'S', '0');
 
     size_t entries_offset = offset + sizeof(Header);
     size_t entry_size = is_hfs ? sizeof(HFSEntry) : sizeof(PFSEntry);
@@ -73,21 +85,21 @@ u32 PartitionFilesystem::GetNumEntries() const {
     return pfs_header.num_entries;
 }
 
-u64 PartitionFilesystem::GetEntryOffset(int index) const {
+u64 PartitionFilesystem::GetEntryOffset(u32 index) const {
     if (index > GetNumEntries())
         return 0;
 
     return content_offset + pfs_entries[index].fs_entry.offset;
 }
 
-u64 PartitionFilesystem::GetEntrySize(int index) const {
+u64 PartitionFilesystem::GetEntrySize(u32 index) const {
     if (index > GetNumEntries())
         return 0;
 
     return pfs_entries[index].fs_entry.size;
 }
 
-std::string PartitionFilesystem::GetEntryName(int index) const {
+std::string PartitionFilesystem::GetEntryName(u32 index) const {
     if (index > GetNumEntries())
         return "";
 
@@ -113,7 +125,7 @@ u64 PartitionFilesystem::GetFileSize(const std::string& name) const {
 }
 
 void PartitionFilesystem::Print() const {
-    NGLOG_DEBUG(Service_FS, "Magic:                  {:.4}", pfs_header.magic.data());
+    NGLOG_DEBUG(Service_FS, "Magic:                  {}", pfs_header.magic);
     NGLOG_DEBUG(Service_FS, "Files:                  {}", pfs_header.num_entries);
     for (u32 i = 0; i < pfs_header.num_entries; i++) {
         NGLOG_DEBUG(Service_FS, " > File {}:              {} (0x{:X} bytes, at 0x{:X})", i,

src/core/file_sys/partition_filesystem.h

@@ -27,9 +27,9 @@ public:
     Loader::ResultStatus Load(const std::vector<u8>& file_data, size_t offset = 0);
 
     u32 GetNumEntries() const;
-    u64 GetEntryOffset(int index) const;
-    u64 GetEntrySize(int index) const;
-    std::string GetEntryName(int index) const;
+    u64 GetEntryOffset(u32 index) const;
+    u64 GetEntrySize(u32 index) const;
+    std::string GetEntryName(u32 index) const;
     u64 GetFileOffset(const std::string& name) const;
     u64 GetFileSize(const std::string& name) const;
 
@@ -37,7 +37,7 @@ public:
 
 private:
     struct Header {
-        std::array<char, 4> magic;
+        u32_le magic;
         u32_le num_entries;
         u32_le strtab_size;
         INSERT_PADDING_BYTES(0x4);

src/core/hle/kernel/address_arbiter.cpp

@@ -0,0 +1,173 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "common/assert.h"
+#include "common/common_funcs.h"
+#include "common/common_types.h"
+#include "core/core.h"
+#include "core/hle/kernel/errors.h"
+#include "core/hle/kernel/kernel.h"
+#include "core/hle/kernel/process.h"
+#include "core/hle/kernel/thread.h"
+#include "core/hle/lock.h"
+#include "core/memory.h"
+
+namespace Kernel {
+namespace AddressArbiter {
+
+// Performs actual address waiting logic.
+static ResultCode WaitForAddress(VAddr address, s64 timeout) {
+    SharedPtr<Thread> current_thread = GetCurrentThread();
+    current_thread->arb_wait_address = address;
+    current_thread->status = THREADSTATUS_WAIT_ARB;
+    current_thread->wakeup_callback = nullptr;
+
+    current_thread->WakeAfterDelay(timeout);
+
+    Core::System::GetInstance().CpuCore(current_thread->processor_id).PrepareReschedule();
+    return RESULT_TIMEOUT;
+}
+
+// Gets the threads waiting on an address.
+static void GetThreadsWaitingOnAddress(std::vector<SharedPtr<Thread>>& waiting_threads,
+                                       VAddr address) {
+    auto RetrieveWaitingThreads =
+        [](size_t core_index, std::vector<SharedPtr<Thread>>& waiting_threads, VAddr arb_addr) {
+            const auto& scheduler = Core::System::GetInstance().Scheduler(core_index);
+            auto& thread_list = scheduler->GetThreadList();
+
+            for (auto& thread : thread_list) {
+                if (thread->arb_wait_address == arb_addr)
+                    waiting_threads.push_back(thread);
+            }
+        };
+
+    // Retrieve a list of all threads that are waiting for this address.
+    RetrieveWaitingThreads(0, waiting_threads, address);
+    RetrieveWaitingThreads(1, waiting_threads, address);
+    RetrieveWaitingThreads(2, waiting_threads, address);
+    RetrieveWaitingThreads(3, waiting_threads, address);
+    // Sort them by priority, such that the highest priority ones come first.
+    std::sort(waiting_threads.begin(), waiting_threads.end(),
+              [](const SharedPtr<Thread>& lhs, const SharedPtr<Thread>& rhs) {
+                  return lhs->current_priority < rhs->current_priority;
+              });
+}
+
+// Wake up num_to_wake (or all) threads in a vector.
+static void WakeThreads(std::vector<SharedPtr<Thread>>& waiting_threads, s32 num_to_wake) {
+    // Only process up to 'target' threads, unless 'target' is <= 0, in which case process
+    // them all.
+    size_t last = waiting_threads.size();
+    if (num_to_wake > 0)
+        last = num_to_wake;
+
+    // Signal the waiting threads.
+    for (size_t i = 0; i < last; i++) {
+        ASSERT(waiting_threads[i]->status = THREADSTATUS_WAIT_ARB);
+        waiting_threads[i]->SetWaitSynchronizationResult(RESULT_SUCCESS);
+        waiting_threads[i]->arb_wait_address = 0;
+        waiting_threads[i]->ResumeFromWait();
+    }
+}
+
+// Signals an address being waited on.
+ResultCode SignalToAddress(VAddr address, s32 num_to_wake) {
+    // Get threads waiting on the address.
+    std::vector<SharedPtr<Thread>> waiting_threads;
+    GetThreadsWaitingOnAddress(waiting_threads, address);
+
+    WakeThreads(waiting_threads, num_to_wake);
+    return RESULT_SUCCESS;
+}
+
+// Signals an address being waited on and increments its value if equal to the value argument.
+ResultCode IncrementAndSignalToAddressIfEqual(VAddr address, s32 value, s32 num_to_wake) {
+    // Ensure that we can write to the address.
+    if (!Memory::IsValidVirtualAddress(address)) {
+        return ERR_INVALID_ADDRESS_STATE;
+    }
+
+    if (static_cast<s32>(Memory::Read32(address)) == value) {
+        Memory::Write32(address, static_cast<u32>(value + 1));
+    } else {
+        return ERR_INVALID_STATE;
+    }
+
+    return SignalToAddress(address, num_to_wake);
+}
+
+// Signals an address being waited on and modifies its value based on waiting thread count if equal
+// to the value argument.
+ResultCode ModifyByWaitingCountAndSignalToAddressIfEqual(VAddr address, s32 value,
+                                                         s32 num_to_wake) {
+    // Ensure that we can write to the address.
+    if (!Memory::IsValidVirtualAddress(address)) {
+        return ERR_INVALID_ADDRESS_STATE;
+    }
+
+    // Get threads waiting on the address.
+    std::vector<SharedPtr<Thread>> waiting_threads;
+    GetThreadsWaitingOnAddress(waiting_threads, address);
+
+    // Determine the modified value depending on the waiting count.
+    s32 updated_value;
+    if (waiting_threads.size() == 0) {
+        updated_value = value - 1;
+    } else if (num_to_wake <= 0 || waiting_threads.size() <= num_to_wake) {
+        updated_value = value + 1;
+    } else {
+        updated_value = value;
+    }
+
+    if (static_cast<s32>(Memory::Read32(address)) == value) {
+        Memory::Write32(address, static_cast<u32>(updated_value));
+    } else {
+        return ERR_INVALID_STATE;
+    }
+
+    WakeThreads(waiting_threads, num_to_wake);
+    return RESULT_SUCCESS;
+}
+
+// Waits on an address if the value passed is less than the argument value, optionally decrementing.
+ResultCode WaitForAddressIfLessThan(VAddr address, s32 value, s64 timeout, bool should_decrement) {
+    // Ensure that we can read the address.
+    if (!Memory::IsValidVirtualAddress(address)) {
+        return ERR_INVALID_ADDRESS_STATE;
+    }
+
+    s32 cur_value = static_cast<s32>(Memory::Read32(address));
+    if (cur_value < value) {
+        Memory::Write32(address, static_cast<u32>(cur_value - 1));
+    } else {
+        return ERR_INVALID_STATE;
+    }
+    // Short-circuit without rescheduling, if timeout is zero.
+    if (timeout == 0) {
+        return RESULT_TIMEOUT;
+    }
+
+    return WaitForAddress(address, timeout);
+}
+
+// Waits on an address if the value passed is equal to the argument value.
+ResultCode WaitForAddressIfEqual(VAddr address, s32 value, s64 timeout) {
+    // Ensure that we can read the address.
+    if (!Memory::IsValidVirtualAddress(address)) {
+        return ERR_INVALID_ADDRESS_STATE;
+    }
+    // Only wait for the address if equal.
+    if (static_cast<s32>(Memory::Read32(address)) != value) {
+        return ERR_INVALID_STATE;
+    }
+    // Short-circuit without rescheduling, if timeout is zero.
+    if (timeout == 0) {
+        return RESULT_TIMEOUT;
+    }
+
+    return WaitForAddress(address, timeout);
+}
+} // namespace AddressArbiter
+} // namespace Kernel

src/core/hle/kernel/address_arbiter.h

@@ -0,0 +1,32 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "core/hle/result.h"
+
+namespace Kernel {
+
+namespace AddressArbiter {
+enum class ArbitrationType {
+    WaitIfLessThan = 0,
+    DecrementAndWaitIfLessThan = 1,
+    WaitIfEqual = 2,
+};
+
+enum class SignalType {
+    Signal = 0,
+    IncrementAndSignalIfEqual = 1,
+    ModifyByWaitingCountAndSignalIfEqual = 2,
+};
+
+ResultCode SignalToAddress(VAddr address, s32 num_to_wake);
+ResultCode IncrementAndSignalToAddressIfEqual(VAddr address, s32 value, s32 num_to_wake);
+ResultCode ModifyByWaitingCountAndSignalToAddressIfEqual(VAddr address, s32 value, s32 num_to_wake);
+
+ResultCode WaitForAddressIfLessThan(VAddr address, s32 value, s64 timeout, bool should_decrement);
+ResultCode WaitForAddressIfEqual(VAddr address, s32 value, s64 timeout);
+} // namespace AddressArbiter
+
+} // namespace Kernel

src/core/hle/kernel/errors.h

@@ -20,13 +20,16 @@ enum {
     MaxConnectionsReached = 52,
 
     // Confirmed Switch OS error codes
-    MisalignedAddress = 102,
+    InvalidAddress = 102,
+    InvalidMemoryState = 106,
     InvalidProcessorId = 113,
     InvalidHandle = 114,
     InvalidCombination = 116,
     Timeout = 117,
     SynchronizationCanceled = 118,
     TooLarge = 119,
+    InvalidEnumValue = 120,
+    InvalidState = 125,
 };
 }
 
@@ -39,14 +42,15 @@ constexpr ResultCode ERR_SESSION_CLOSED_BY_REMOTE(-1);
 constexpr ResultCode ERR_PORT_NAME_TOO_LONG(-1);
 constexpr ResultCode ERR_WRONG_PERMISSION(-1);
 constexpr ResultCode ERR_MAX_CONNECTIONS_REACHED(-1);
-constexpr ResultCode ERR_INVALID_ENUM_VALUE(-1);
+constexpr ResultCode ERR_INVALID_ENUM_VALUE(ErrorModule::Kernel, ErrCodes::InvalidEnumValue);
 constexpr ResultCode ERR_INVALID_ENUM_VALUE_FND(-1);
 constexpr ResultCode ERR_INVALID_COMBINATION(-1);
 constexpr ResultCode ERR_INVALID_COMBINATION_KERNEL(-1);
 constexpr ResultCode ERR_OUT_OF_MEMORY(-1);
-constexpr ResultCode ERR_INVALID_ADDRESS(-1);
-constexpr ResultCode ERR_INVALID_ADDRESS_STATE(-1);
+constexpr ResultCode ERR_INVALID_ADDRESS(ErrorModule::Kernel, ErrCodes::InvalidAddress);
+constexpr ResultCode ERR_INVALID_ADDRESS_STATE(ErrorModule::Kernel, ErrCodes::InvalidMemoryState);
 constexpr ResultCode ERR_INVALID_HANDLE(ErrorModule::Kernel, ErrCodes::InvalidHandle);
+constexpr ResultCode ERR_INVALID_STATE(ErrorModule::Kernel, ErrCodes::InvalidState);
 constexpr ResultCode ERR_INVALID_POINTER(-1);
 constexpr ResultCode ERR_INVALID_OBJECT_ADDR(-1);
 constexpr ResultCode ERR_NOT_AUTHORIZED(-1);

src/core/hle/kernel/hle_ipc.cpp

@@ -271,6 +271,11 @@ std::vector<u8> HLERequestContext::ReadBuffer(int buffer_index) const {
 }
 
 size_t HLERequestContext::WriteBuffer(const void* buffer, size_t size, int buffer_index) const {
+    if (size == 0) {
+        NGLOG_WARNING(Core, "skip empty buffer write");
+        return 0;
+    }
+
     const bool is_buffer_b{BufferDescriptorB().size() && BufferDescriptorB()[buffer_index].Size()};
     const size_t buffer_size{GetWriteBufferSize(buffer_index)};
     if (size > buffer_size) {

src/core/hle/kernel/mutex.cpp

@@ -59,7 +59,7 @@ ResultCode Mutex::TryAcquire(VAddr address, Handle holding_thread_handle,
                              Handle requesting_thread_handle) {
     // The mutex address must be 4-byte aligned
     if ((address % sizeof(u32)) != 0) {
-        return ResultCode(ErrorModule::Kernel, ErrCodes::MisalignedAddress);
+        return ResultCode(ErrorModule::Kernel, ErrCodes::InvalidAddress);
     }
 
     SharedPtr<Thread> holding_thread = g_handle_table.Get<Thread>(holding_thread_handle);
@@ -97,7 +97,7 @@ ResultCode Mutex::TryAcquire(VAddr address, Handle holding_thread_handle,
 ResultCode Mutex::Release(VAddr address) {
     // The mutex address must be 4-byte aligned
     if ((address % sizeof(u32)) != 0) {
-        return ResultCode(ErrorModule::Kernel, ErrCodes::MisalignedAddress);
+        return ResultCode(ErrorModule::Kernel, ErrCodes::InvalidAddress);
     }
 
     auto [thread, num_waiters] = GetHighestPriorityMutexWaitingThread(GetCurrentThread(), address);

src/core/hle/kernel/svc.cpp

@@ -11,6 +11,7 @@
 #include "common/string_util.h"
 #include "core/core.h"
 #include "core/core_timing.h"
+#include "core/hle/kernel/address_arbiter.h"
 #include "core/hle/kernel/client_port.h"
 #include "core/hle/kernel/client_session.h"
 #include "core/hle/kernel/event.h"
@@ -316,6 +317,11 @@ static ResultCode GetInfo(u64* result, u64 info_id, u64 handle, u64 info_sub_id)
                       "(STUBBED) Attempted to query privileged process id bounds, returned 0");
         *result = 0;
         break;
+    case GetInfoType::UserExceptionContextAddr:
+        NGLOG_WARNING(Kernel_SVC,
+                      "(STUBBED) Attempted to query user exception context address, returned 0");
+        *result = 0;
+        break;
     default:
         UNIMPLEMENTED();
     }
@@ -575,7 +581,7 @@ static void SleepThread(s64 nanoseconds) {
     Core::System::GetInstance().PrepareReschedule();
 }
 
-/// Signal process wide key atomic
+/// Wait process wide key atomic
 static ResultCode WaitProcessWideKeyAtomic(VAddr mutex_addr, VAddr condition_variable_addr,
                                            Handle thread_handle, s64 nano_seconds) {
     NGLOG_TRACE(
@@ -684,6 +690,58 @@ static ResultCode SignalProcessWideKey(VAddr condition_variable_addr, s32 target
     return RESULT_SUCCESS;
 }
 
+// Wait for an address (via Address Arbiter)
+static ResultCode WaitForAddress(VAddr address, u32 type, s32 value, s64 timeout) {
+    NGLOG_WARNING(Kernel_SVC, "called, address=0x{:X}, type=0x{:X}, value=0x{:X}, timeout={}",
+                  address, type, value, timeout);
+    // If the passed address is a kernel virtual address, return invalid memory state.
+    if (Memory::IsKernelVirtualAddress(address)) {
+        return ERR_INVALID_ADDRESS_STATE;
+    }
+    // If the address is not properly aligned to 4 bytes, return invalid address.
+    if (address % sizeof(u32) != 0) {
+        return ERR_INVALID_ADDRESS;
+    }
+
+    switch (static_cast<AddressArbiter::ArbitrationType>(type)) {
+    case AddressArbiter::ArbitrationType::WaitIfLessThan:
+        return AddressArbiter::WaitForAddressIfLessThan(address, value, timeout, false);
+    case AddressArbiter::ArbitrationType::DecrementAndWaitIfLessThan:
+        return AddressArbiter::WaitForAddressIfLessThan(address, value, timeout, true);
+    case AddressArbiter::ArbitrationType::WaitIfEqual:
+        return AddressArbiter::WaitForAddressIfEqual(address, value, timeout);
+    default:
+        return ERR_INVALID_ENUM_VALUE;
+    }
+}
+
+// Signals to an address (via Address Arbiter)
+static ResultCode SignalToAddress(VAddr address, u32 type, s32 value, s32 num_to_wake) {
+    NGLOG_WARNING(Kernel_SVC,
+                  "called, address=0x{:X}, type=0x{:X}, value=0x{:X}, num_to_wake=0x{:X}", address,
+                  type, value, num_to_wake);
+    // If the passed address is a kernel virtual address, return invalid memory state.
+    if (Memory::IsKernelVirtualAddress(address)) {
+        return ERR_INVALID_ADDRESS_STATE;
+    }
+    // If the address is not properly aligned to 4 bytes, return invalid address.
+    if (address % sizeof(u32) != 0) {
+        return ERR_INVALID_ADDRESS;
+    }
+
+    switch (static_cast<AddressArbiter::SignalType>(type)) {
+    case AddressArbiter::SignalType::Signal:
+        return AddressArbiter::SignalToAddress(address, num_to_wake);
+    case AddressArbiter::SignalType::IncrementAndSignalIfEqual:
+        return AddressArbiter::IncrementAndSignalToAddressIfEqual(address, value, num_to_wake);
+    case AddressArbiter::SignalType::ModifyByWaitingCountAndSignalIfEqual:
+        return AddressArbiter::ModifyByWaitingCountAndSignalToAddressIfEqual(address, value,
+                                                                             num_to_wake);
+    default:
+        return ERR_INVALID_ENUM_VALUE;
+    }
+}
+
 /// This returns the total CPU ticks elapsed since the CPU was powered-on
 static u64 GetSystemTick() {
     const u64 result{CoreTiming::GetTicks()};
@@ -744,7 +802,7 @@ static ResultCode SetThreadCoreMask(Handle thread_handle, u32 core, u64 mask) {
         ASSERT(thread->owner_process->ideal_processor != THREADPROCESSORID_DEFAULT);
         // Set the target CPU to the one specified in the process' exheader.
         core = thread->owner_process->ideal_processor;
-        mask = 1 << core;
+        mask = 1ull << core;
     }
 
     if (mask == 0) {
@@ -761,7 +819,7 @@ static ResultCode SetThreadCoreMask(Handle thread_handle, u32 core, u64 mask) {
     }
 
     // Error out if the input core isn't enabled in the input mask.
-    if (core < Core::NUM_CPU_CORES && (mask & (1 << core)) == 0) {
+    if (core < Core::NUM_CPU_CORES && (mask & (1ull << core)) == 0) {
         return ResultCode(ErrorModule::Kernel, ErrCodes::InvalidCombination);
     }
 
@@ -856,8 +914,8 @@ static const FunctionDef SVC_Table[] = {
     {0x31, nullptr, "GetResourceLimitCurrentValue"},
     {0x32, SvcWrap<SetThreadActivity>, "SetThreadActivity"},
     {0x33, SvcWrap<GetThreadContext>, "GetThreadContext"},
-    {0x34, nullptr, "WaitForAddress"},
-    {0x35, nullptr, "SignalToAddress"},
+    {0x34, SvcWrap<WaitForAddress>, "WaitForAddress"},
+    {0x35, SvcWrap<SignalToAddress>, "SignalToAddress"},
     {0x36, nullptr, "Unknown"},
     {0x37, nullptr, "Unknown"},
     {0x38, nullptr, "Unknown"},

src/core/hle/kernel/svc_wrap.h

@@ -179,6 +179,20 @@ void SvcWrap() {
     FuncReturn(retval);
 }
 
+template <ResultCode func(u64, u32, s32, s64)>
+void SvcWrap() {
+    FuncReturn(
+        func(PARAM(0), (u32)(PARAM(1) & 0xFFFFFFFF), (s32)(PARAM(2) & 0xFFFFFFFF), (s64)PARAM(3))
+            .raw);
+}
+
+template <ResultCode func(u64, u32, s32, s32)>
+void SvcWrap() {
+    FuncReturn(func(PARAM(0), (u32)(PARAM(1) & 0xFFFFFFFF), (s32)(PARAM(2) & 0xFFFFFFFF),
+                    (s32)(PARAM(3) & 0xFFFFFFFF))
+                   .raw);
+}
+
 ////////////////////////////////////////////////////////////////////////////////////////////////////
 // Function wrappers that return type u32
 

src/core/hle/kernel/thread.cpp

@@ -140,6 +140,11 @@ static void ThreadWakeupCallback(u64 thread_handle, int cycles_late) {
         }
     }
 
+    if (thread->arb_wait_address != 0) {
+        ASSERT(thread->status == THREADSTATUS_WAIT_ARB);
+        thread->arb_wait_address = 0;
+    }
+
     if (resume)
         thread->ResumeFromWait();
 }
@@ -179,6 +184,7 @@ void Thread::ResumeFromWait() {
     case THREADSTATUS_WAIT_SLEEP:
     case THREADSTATUS_WAIT_IPC:
     case THREADSTATUS_WAIT_MUTEX:
+    case THREADSTATUS_WAIT_ARB:
         break;
 
     case THREADSTATUS_READY:

src/core/hle/kernel/thread.h

@@ -45,6 +45,7 @@ enum ThreadStatus {
     THREADSTATUS_WAIT_SYNCH_ANY, ///< Waiting due to WaitSynch1 or WaitSynchN with wait_all = false
     THREADSTATUS_WAIT_SYNCH_ALL, ///< Waiting due to WaitSynchronizationN with wait_all = true
     THREADSTATUS_WAIT_MUTEX,     ///< Waiting due to an ArbitrateLock/WaitProcessWideKey svc
+    THREADSTATUS_WAIT_ARB,       ///< Waiting due to a SignalToAddress/WaitForAddress svc
     THREADSTATUS_DORMANT,        ///< Created but not yet made ready
     THREADSTATUS_DEAD            ///< Run to completion, or forcefully terminated
 };
@@ -230,6 +231,9 @@ public:
     VAddr mutex_wait_address; ///< If waiting on a Mutex, this is the mutex address
     Handle wait_handle;       ///< The handle used to wait for the mutex.
 
+    // If waiting for an AddressArbiter, this is the address being waited on.
+    VAddr arb_wait_address{0};
+
     std::string name;
 
     /// Handle used by guest emulated application to access this thread

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

@@ -17,7 +17,8 @@ constexpr u64 audio_ticks{static_cast<u64>(CoreTiming::BASE_CLOCK_RATE / 200)};
 
 class IAudioRenderer final : public ServiceFramework<IAudioRenderer> {
 public:
-    IAudioRenderer() : ServiceFramework("IAudioRenderer") {
+    IAudioRenderer(AudioRendererParameter audren_params)
+        : ServiceFramework("IAudioRenderer"), worker_params(audren_params) {
         static const FunctionInfo functions[] = {
             {0, nullptr, "GetAudioRendererSampleRate"},
             {1, nullptr, "GetAudioRendererSampleCount"},
@@ -57,27 +58,37 @@ private:
     }
 
     void RequestUpdateAudioRenderer(Kernel::HLERequestContext& ctx) {
-        NGLOG_DEBUG(Service_Audio, "{}", ctx.Description());
-        AudioRendererResponseData response_data{};
+        UpdateDataHeader config{};
+        auto buf = ctx.ReadBuffer();
+        std::memcpy(&config, buf.data(), sizeof(UpdateDataHeader));
+        u32 memory_pool_count = worker_params.effect_count + (worker_params.voice_count * 4);
 
-        response_data.section_0_size =
-            static_cast<u32>(response_data.state_entries.size() * sizeof(AudioRendererStateEntry));
-        response_data.section_1_size = static_cast<u32>(response_data.section_1.size());
-        response_data.section_2_size = static_cast<u32>(response_data.section_2.size());
-        response_data.section_3_size = static_cast<u32>(response_data.section_3.size());
-        response_data.section_4_size = static_cast<u32>(response_data.section_4.size());
-        response_data.section_5_size = static_cast<u32>(response_data.section_5.size());
-        response_data.total_size = sizeof(AudioRendererResponseData);
+        std::vector<MemoryPoolInfo> mem_pool_info(memory_pool_count);
+        std::memcpy(mem_pool_info.data(),
+                    buf.data() + sizeof(UpdateDataHeader) + config.behavior_size,
+                    memory_pool_count * sizeof(MemoryPoolInfo));
 
-        for (unsigned i = 0; i < response_data.state_entries.size(); i++) {
-            // 4 = Busy and 5 = Ready?
-            response_data.state_entries[i].state = 5;
+        UpdateDataHeader response_data{worker_params};
+
+        ASSERT(ctx.GetWriteBufferSize() == response_data.total_size);
+
+        std::vector<u8> output(response_data.total_size);
+        std::memcpy(output.data(), &response_data, sizeof(UpdateDataHeader));
+        std::vector<MemoryPoolEntry> memory_pool(memory_pool_count);
+        for (unsigned i = 0; i < memory_pool.size(); i++) {
+            if (mem_pool_info[i].pool_state == MemoryPoolStates::RequestAttach)
+                memory_pool[i].state = MemoryPoolStates::Attached;
+            else if (mem_pool_info[i].pool_state == MemoryPoolStates::RequestDetach)
+                memory_pool[i].state = MemoryPoolStates::Detached;
+            else
+                memory_pool[i].state = mem_pool_info[i].pool_state;
         }
+        std::memcpy(output.data() + sizeof(UpdateDataHeader), memory_pool.data(),
+                    response_data.memory_pools_size);
 
-        ctx.WriteBuffer(&response_data, response_data.total_size);
+        ctx.WriteBuffer(output);
 
         IPC::ResponseBuilder rb{ctx, 2};
-
         rb.Push(RESULT_SUCCESS);
 
         NGLOG_WARNING(Service_Audio, "(STUBBED) called");
@@ -109,48 +120,66 @@ private:
         NGLOG_WARNING(Service_Audio, "(STUBBED) called");
     }
 
-    struct AudioRendererStateEntry {
-        u32_le state;
+    enum class MemoryPoolStates : u32 { // Should be LE
+        Invalid = 0x0,
+        Unknown = 0x1,
+        RequestDetach = 0x2,
+        Detached = 0x3,
+        RequestAttach = 0x4,
+        Attached = 0x5,
+        Released = 0x6,
+    };
+
+    struct MemoryPoolEntry {
+        MemoryPoolStates state;
         u32_le unknown_4;
         u32_le unknown_8;
         u32_le unknown_c;
     };
-    static_assert(sizeof(AudioRendererStateEntry) == 0x10,
-                  "AudioRendererStateEntry has wrong size");
+    static_assert(sizeof(MemoryPoolEntry) == 0x10, "MemoryPoolEntry has wrong size");
 
-    struct AudioRendererResponseData {
-        u32_le unknown_0;
-        u32_le section_5_size;
-        u32_le section_0_size;
-        u32_le section_1_size;
-        u32_le unknown_10;
-        u32_le section_2_size;
-        u32_le unknown_18;
-        u32_le section_3_size;
-        u32_le section_4_size;
-        u32_le unknown_24;
-        u32_le unknown_28;
-        u32_le unknown_2c;
-        u32_le unknown_30;
-        u32_le unknown_34;
-        u32_le unknown_38;
-        u32_le total_size;
-
-        std::array<AudioRendererStateEntry, 0x18e> state_entries;
-
-        std::array<u8, 0x600> section_1;
-        std::array<u8, 0xe0> section_2;
-        std::array<u8, 0x20> section_3;
-        std::array<u8, 0x10> section_4;
-        std::array<u8, 0xb0> section_5;
+    struct MemoryPoolInfo {
+        u64_le pool_address;
+        u64_le pool_size;
+        MemoryPoolStates pool_state;
+        INSERT_PADDING_WORDS(3); // Unknown
     };
-    static_assert(sizeof(AudioRendererResponseData) == 0x20e0,
-                  "AudioRendererResponseData has wrong size");
+    static_assert(sizeof(MemoryPoolInfo) == 0x20, "MemoryPoolInfo has wrong size");
+
+    struct UpdateDataHeader {
+        UpdateDataHeader() {}
+
+        UpdateDataHeader(const AudioRendererParameter& config) {
+            revision = Common::MakeMagic('R', 'E', 'V', '4'); // 5.1.0 Revision
+            behavior_size = 0xb0;
+            memory_pools_size = (config.effect_count + (config.voice_count * 4)) * 0x10;
+            voices_size = config.voice_count * 0x10;
+            effects_size = config.effect_count * 0x10;
+            sinks_size = config.sink_count * 0x20;
+            performance_manager_size = 0x10;
+            total_size = sizeof(UpdateDataHeader) + behavior_size + memory_pools_size +
+                         voices_size + effects_size + sinks_size + performance_manager_size;
+        }
+
+        u32_le revision;
+        u32_le behavior_size;
+        u32_le memory_pools_size;
+        u32_le voices_size;
+        u32_le voice_resource_size;
+        u32_le effects_size;
+        u32_le mixes_size;
+        u32_le sinks_size;
+        u32_le performance_manager_size;
+        INSERT_PADDING_WORDS(6);
+        u32_le total_size;
+    };
+    static_assert(sizeof(UpdateDataHeader) == 0x40, "UpdateDataHeader has wrong size");
 
     /// This is used to trigger the audio event callback.
     CoreTiming::EventType* audio_event;
 
     Kernel::SharedPtr<Kernel::Event> system_event;
+    AudioRendererParameter worker_params;
 };
 
 class IAudioDevice final : public ServiceFramework<IAudioDevice> {
@@ -248,31 +277,33 @@ AudRenU::AudRenU() : ServiceFramework("audren:u") {
 }
 
 void AudRenU::OpenAudioRenderer(Kernel::HLERequestContext& ctx) {
+    IPC::RequestParser rp{ctx};
+    auto params = rp.PopRaw<AudioRendererParameter>();
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
 
     rb.Push(RESULT_SUCCESS);
-    rb.PushIpcInterface<Audio::IAudioRenderer>();
+    rb.PushIpcInterface<Audio::IAudioRenderer>(std::move(params));
 
     NGLOG_DEBUG(Service_Audio, "called");
 }
 
 void AudRenU::GetAudioRendererWorkBufferSize(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
-    auto params = rp.PopRaw<WorkerBufferParameters>();
+    auto params = rp.PopRaw<AudioRendererParameter>();
 
-    u64 buffer_sz = Common::AlignUp(4 * params.unknown8, 0x40);
-    buffer_sz += params.unknownC * 1024;
-    buffer_sz += 0x940 * (params.unknownC + 1);
+    u64 buffer_sz = Common::AlignUp(4 * params.unknown_8, 0x40);
+    buffer_sz += params.unknown_c * 1024;
+    buffer_sz += 0x940 * (params.unknown_c + 1);
     buffer_sz += 0x3F0 * params.voice_count;
-    buffer_sz += Common::AlignUp(8 * (params.unknownC + 1), 0x10);
+    buffer_sz += Common::AlignUp(8 * (params.unknown_c + 1), 0x10);
     buffer_sz += Common::AlignUp(8 * params.voice_count, 0x10);
     buffer_sz +=
-        Common::AlignUp((0x3C0 * (params.sink_count + params.unknownC) + 4 * params.sample_count) *
-                            (params.unknown8 + 6),
+        Common::AlignUp((0x3C0 * (params.sink_count + params.unknown_c) + 4 * params.sample_count) *
+                            (params.unknown_8 + 6),
                         0x40);
 
-    if (IsFeatureSupported(AudioFeatures::Splitter, params.magic)) {
-        u32 count = params.unknownC + 1;
+    if (IsFeatureSupported(AudioFeatures::Splitter, params.revision)) {
+        u32 count = params.unknown_c + 1;
         u64 node_count = Common::AlignUp(count, 0x40);
         u64 node_state_buffer_sz =
             4 * (node_count * node_count) + 0xC * node_count + 2 * (node_count / 8);
@@ -287,20 +318,20 @@ void AudRenU::GetAudioRendererWorkBufferSize(Kernel::HLERequestContext& ctx) {
     }
 
     buffer_sz += 0x20 * (params.effect_count + 4 * params.voice_count) + 0x50;
-    if (IsFeatureSupported(AudioFeatures::Splitter, params.magic)) {
-        buffer_sz += 0xE0 * params.unknown2c;
+    if (IsFeatureSupported(AudioFeatures::Splitter, params.revision)) {
+        buffer_sz += 0xE0 * params.unknown_2c;
         buffer_sz += 0x20 * params.splitter_count;
-        buffer_sz += Common::AlignUp(4 * params.unknown2c, 0x10);
+        buffer_sz += Common::AlignUp(4 * params.unknown_2c, 0x10);
     }
     buffer_sz = Common::AlignUp(buffer_sz, 0x40) + 0x170 * params.sink_count;
     u64 output_sz = buffer_sz + 0x280 * params.sink_count + 0x4B0 * params.effect_count +
                     ((params.voice_count * 256) | 0x40);
 
-    if (params.unknown1c >= 1) {
+    if (params.unknown_1c >= 1) {
         output_sz = Common::AlignUp(((16 * params.sink_count + 16 * params.effect_count +
                                       16 * params.voice_count + 16) +
                                      0x658) *
-                                            (params.unknown1c + 1) +
+                                            (params.unknown_1c + 1) +
                                         0xc0,
                                     0x40) +
                     output_sz;
@@ -328,7 +359,7 @@ bool AudRenU::IsFeatureSupported(AudioFeatures feature, u32_le revision) const {
     u32_be version_num = (revision - Common::MakeMagic('R', 'E', 'V', '0')); // Byte swap
     switch (feature) {
     case AudioFeatures::Splitter:
-        return version_num >= 2;
+        return version_num >= 2u;
     default:
         return false;
     }

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

@@ -12,6 +12,24 @@ class HLERequestContext;
 
 namespace Service::Audio {
 
+struct AudioRendererParameter {
+    u32_le sample_rate;
+    u32_le sample_count;
+    u32_le unknown_8;
+    u32_le unknown_c;
+    u32_le voice_count;
+    u32_le sink_count;
+    u32_le effect_count;
+    u32_le unknown_1c;
+    u8 unknown_20;
+    INSERT_PADDING_BYTES(3);
+    u32_le splitter_count;
+    u32_le unknown_2c;
+    INSERT_PADDING_WORDS(1);
+    u32_le revision;
+};
+static_assert(sizeof(AudioRendererParameter) == 52, "AudioRendererParameter is an invalid size");
+
 class AudRenU final : public ServiceFramework<AudRenU> {
 public:
     explicit AudRenU();
@@ -22,25 +40,6 @@ private:
     void GetAudioRendererWorkBufferSize(Kernel::HLERequestContext& ctx);
     void GetAudioDevice(Kernel::HLERequestContext& ctx);
 
-    struct WorkerBufferParameters {
-        u32_le sample_rate;
-        u32_le sample_count;
-        u32_le unknown8;
-        u32_le unknownC;
-        u32_le voice_count;
-        u32_le sink_count;
-        u32_le effect_count;
-        u32_le unknown1c;
-        u8 unknown20;
-        u8 padding1[3];
-        u32_le splitter_count;
-        u32_le unknown2c;
-        u8 padding2[4];
-        u32_le magic;
-    };
-    static_assert(sizeof(WorkerBufferParameters) == 52,
-                  "WorkerBufferParameters is an invalid size");
-
     enum class AudioFeatures : u32 {
         Splitter,
     };

src/core/hle/service/filesystem/fsp_srv.cpp

@@ -4,6 +4,7 @@
 
 #include <cinttypes>
 #include "common/logging/log.h"
+#include "common/string_util.h"
 #include "core/core.h"
 #include "core/file_sys/directory.h"
 #include "core/file_sys/filesystem.h"
@@ -258,9 +259,7 @@ public:
         IPC::RequestParser rp{ctx};
 
         auto file_buffer = ctx.ReadBuffer();
-        auto end = std::find(file_buffer.begin(), file_buffer.end(), '\0');
-
-        std::string name(file_buffer.begin(), end);
+        std::string name = Common::StringFromBuffer(file_buffer);
 
         u64 mode = rp.Pop<u64>();
         u32 size = rp.Pop<u32>();
@@ -275,9 +274,7 @@ public:
         IPC::RequestParser rp{ctx};
 
         auto file_buffer = ctx.ReadBuffer();
-        auto end = std::find(file_buffer.begin(), file_buffer.end(), '\0');
-
-        std::string name(file_buffer.begin(), end);
+        std::string name = Common::StringFromBuffer(file_buffer);
 
         NGLOG_DEBUG(Service_FS, "called file {}", name);
 
@@ -289,9 +286,7 @@ public:
         IPC::RequestParser rp{ctx};
 
         auto file_buffer = ctx.ReadBuffer();
-        auto end = std::find(file_buffer.begin(), file_buffer.end(), '\0');
-
-        std::string name(file_buffer.begin(), end);
+        std::string name = Common::StringFromBuffer(file_buffer);
 
         NGLOG_DEBUG(Service_FS, "called directory {}", name);
 
@@ -305,13 +300,11 @@ public:
         std::vector<u8> buffer;
         buffer.resize(ctx.BufferDescriptorX()[0].Size());
         Memory::ReadBlock(ctx.BufferDescriptorX()[0].Address(), buffer.data(), buffer.size());
-        auto end = std::find(buffer.begin(), buffer.end(), '\0');
-        std::string src_name(buffer.begin(), end);
+        std::string src_name = Common::StringFromBuffer(buffer);
 
         buffer.resize(ctx.BufferDescriptorX()[1].Size());
         Memory::ReadBlock(ctx.BufferDescriptorX()[1].Address(), buffer.data(), buffer.size());
-        end = std::find(buffer.begin(), buffer.end(), '\0');
-        std::string dst_name(buffer.begin(), end);
+        std::string dst_name = Common::StringFromBuffer(buffer);
 
         NGLOG_DEBUG(Service_FS, "called file '{}' to file '{}'", src_name, dst_name);
 
@@ -323,9 +316,7 @@ public:
         IPC::RequestParser rp{ctx};
 
         auto file_buffer = ctx.ReadBuffer();
-        auto end = std::find(file_buffer.begin(), file_buffer.end(), '\0');
-
-        std::string name(file_buffer.begin(), end);
+        std::string name = Common::StringFromBuffer(file_buffer);
 
         auto mode = static_cast<FileSys::Mode>(rp.Pop<u32>());
 
@@ -349,9 +340,7 @@ public:
         IPC::RequestParser rp{ctx};
 
         auto file_buffer = ctx.ReadBuffer();
-        auto end = std::find(file_buffer.begin(), file_buffer.end(), '\0');
-
-        std::string name(file_buffer.begin(), end);
+        std::string name = Common::StringFromBuffer(file_buffer);
 
         // TODO(Subv): Implement this filter.
         u32 filter_flags = rp.Pop<u32>();
@@ -376,9 +365,7 @@ public:
         IPC::RequestParser rp{ctx};
 
         auto file_buffer = ctx.ReadBuffer();
-        auto end = std::find(file_buffer.begin(), file_buffer.end(), '\0');
-
-        std::string name(file_buffer.begin(), end);
+        std::string name = Common::StringFromBuffer(file_buffer);
 
         NGLOG_DEBUG(Service_FS, "called file {}", name);
 

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

@@ -84,6 +84,10 @@ private:
 
         for (size_t controller = 0; controller < mem.controllers.size(); controller++) {
             for (int index = 0; index < HID_NUM_LAYOUTS; index++) {
+                // TODO(DarkLordZach): Is this layout/controller config actually invalid?
+                if (controller == Controller_Handheld && index == Layout_Single)
+                    continue;
+
                 ControllerLayout& layout = mem.controllers[controller].layouts[index];
                 layout.header.num_entries = HID_NUM_ENTRIES;
                 layout.header.max_entry_index = HID_NUM_ENTRIES - 1;
@@ -94,7 +98,6 @@ private:
                 layout.header.latest_entry = (layout.header.latest_entry + 1) % HID_NUM_ENTRIES;
 
                 ControllerInputEntry& entry = layout.entries[layout.header.latest_entry];
-                entry.connection_state = ConnectionState_Connected | ConnectionState_Wired;
                 entry.timestamp++;
                 // TODO(shinyquagsire23): Is this always identical to timestamp?
                 entry.timestamp_2++;
@@ -103,6 +106,8 @@ private:
                 if (controller != Controller_Handheld)
                     continue;
 
+                entry.connection_state = ConnectionState_Connected | ConnectionState_Wired;
+
                 // TODO(shinyquagsire23): Set up some LUTs for each layout mapping in the future?
                 // For now everything is just the default handheld layout, but split Joy-Con will
                 // rotate the face buttons and directions for certain layouts.

src/core/hle/service/hid/hid.h

@@ -12,7 +12,7 @@ namespace Service::HID {
 // Begin enums and output structs
 
 constexpr u32 HID_NUM_ENTRIES = 17;
-constexpr u32 HID_NUM_LAYOUTS = 2;
+constexpr u32 HID_NUM_LAYOUTS = 7;
 constexpr s32 HID_JOYSTICK_MAX = 0x8000;
 constexpr s32 HID_JOYSTICK_MIN = -0x8000;
 

src/core/hle/service/mm/mm_u.cpp

@@ -0,0 +1,50 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "common/logging/log.h"
+#include "core/hle/ipc_helpers.h"
+#include "core/hle/kernel/client_session.h"
+#include "core/hle/service/mm/mm_u.h"
+
+namespace Service::MM {
+
+void InstallInterfaces(SM::ServiceManager& service_manager) {
+    std::make_shared<MM_U>()->InstallAsService(service_manager);
+}
+
+void MM_U::Initialize(Kernel::HLERequestContext& ctx) {
+    NGLOG_WARNING(Service_MM, "(STUBBED) called");
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(RESULT_SUCCESS);
+}
+
+void MM_U::SetAndWait(Kernel::HLERequestContext& ctx) {
+    IPC::RequestParser rp{ctx};
+    min = rp.Pop<u32>();
+    max = rp.Pop<u32>();
+    current = min;
+
+    NGLOG_WARNING(Service_MM, "(STUBBED) called, min=0x{:X}, max=0x{:X}", min, max);
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(RESULT_SUCCESS);
+}
+
+void MM_U::Get(Kernel::HLERequestContext& ctx) {
+    NGLOG_WARNING(Service_MM, "(STUBBED) called");
+    IPC::ResponseBuilder rb{ctx, 3};
+    rb.Push(RESULT_SUCCESS);
+    rb.Push(current);
+}
+
+MM_U::MM_U() : ServiceFramework("mm:u") {
+    static const FunctionInfo functions[] = {
+        {0, nullptr, "InitializeOld"},        {1, nullptr, "FinalizeOld"},
+        {2, nullptr, "SetAndWaitOld"},        {3, nullptr, "GetOld"},
+        {4, &MM_U::Initialize, "Initialize"}, {5, nullptr, "Finalize"},
+        {6, &MM_U::SetAndWait, "SetAndWait"}, {7, &MM_U::Get, "Get"},
+    };
+    RegisterHandlers(functions);
+}
+
+} // namespace Service::MM

src/core/hle/service/mm/mm_u.h

@@ -0,0 +1,29 @@
+// 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/service.h"
+
+namespace Service::MM {
+
+class MM_U final : public ServiceFramework<MM_U> {
+public:
+    MM_U();
+    ~MM_U() = default;
+
+private:
+    void Initialize(Kernel::HLERequestContext& ctx);
+    void SetAndWait(Kernel::HLERequestContext& ctx);
+    void Get(Kernel::HLERequestContext& ctx);
+
+    u32 min{0};
+    u32 max{0};
+    u32 current{0};
+};
+
+/// Registers all MM services with the specified service manager.
+void InstallInterfaces(SM::ServiceManager& service_manager);
+
+} // namespace Service::MM

src/core/hle/service/nfp/nfp.cpp

@@ -4,6 +4,8 @@
 
 #include "common/logging/log.h"
 #include "core/hle/ipc_helpers.h"
+#include "core/hle/kernel/event.h"
+#include "core/hle/service/hid/hid.h"
 #include "core/hle/service/nfp/nfp.h"
 #include "core/hle/service/nfp/nfp_user.h"
 
@@ -18,7 +20,7 @@ public:
         static const FunctionInfo functions[] = {
             {0, &IUser::Initialize, "Initialize"},
             {1, nullptr, "Finalize"},
-            {2, nullptr, "ListDevices"},
+            {2, &IUser::ListDevices, "ListDevices"},
             {3, nullptr, "StartDetection"},
             {4, nullptr, "StopDetection"},
             {5, nullptr, "Mount"},
@@ -33,24 +35,116 @@ public:
             {14, nullptr, "GetRegisterInfo"},
             {15, nullptr, "GetCommonInfo"},
             {16, nullptr, "GetModelInfo"},
-            {17, nullptr, "AttachActivateEvent"},
-            {18, nullptr, "AttachDeactivateEvent"},
-            {19, nullptr, "GetState"},
-            {20, nullptr, "GetDeviceState"},
-            {21, nullptr, "GetNpadId"},
+            {17, &IUser::AttachActivateEvent, "AttachActivateEvent"},
+            {18, &IUser::AttachDeactivateEvent, "AttachDeactivateEvent"},
+            {19, &IUser::GetState, "GetState"},
+            {20, &IUser::GetDeviceState, "GetDeviceState"},
+            {21, &IUser::GetNpadId, "GetNpadId"},
             {22, nullptr, "GetApplicationArea2"},
-            {23, nullptr, "AttachAvailabilityChangeEvent"},
+            {23, &IUser::AttachAvailabilityChangeEvent, "AttachAvailabilityChangeEvent"},
             {24, nullptr, "RecreateApplicationArea"},
         };
         RegisterHandlers(functions);
+
+        activate_event = Kernel::Event::Create(Kernel::ResetType::OneShot, "IUser:ActivateEvent");
+        deactivate_event =
+            Kernel::Event::Create(Kernel::ResetType::OneShot, "IUser:DeactivateEvent");
+        availability_change_event =
+            Kernel::Event::Create(Kernel::ResetType::OneShot, "IUser:AvailabilityChangeEvent");
     }
 
 private:
+    enum class State : u32 {
+        NonInitialized = 0,
+        Initialized = 1,
+    };
+
+    enum class DeviceState : u32 {
+        Initialized = 0,
+    };
+
     void Initialize(Kernel::HLERequestContext& ctx) {
         NGLOG_WARNING(Service_NFP, "(STUBBED) called");
+
+        state = State::Initialized;
+
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
     }
+
+    void ListDevices(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        const u32 array_size = rp.Pop<u32>();
+
+        ctx.WriteBuffer(&device_handle, sizeof(device_handle));
+
+        NGLOG_WARNING(Service_NFP, "(STUBBED) called, array_size={}", array_size);
+
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(RESULT_SUCCESS);
+        rb.Push<u32>(0);
+    }
+
+    void AttachActivateEvent(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        const u64 dev_handle = rp.Pop<u64>();
+        NGLOG_WARNING(Service_NFP, "(STUBBED) called, dev_handle=0x{:X}", dev_handle);
+
+        IPC::ResponseBuilder rb{ctx, 2, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushCopyObjects(activate_event);
+    }
+
+    void AttachDeactivateEvent(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        const u64 dev_handle = rp.Pop<u64>();
+        NGLOG_WARNING(Service_NFP, "(STUBBED) called, dev_handle=0x{:X}", dev_handle);
+
+        IPC::ResponseBuilder rb{ctx, 2, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushCopyObjects(deactivate_event);
+    }
+
+    void GetState(Kernel::HLERequestContext& ctx) {
+        NGLOG_WARNING(Service_NFP, "(STUBBED) called");
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(RESULT_SUCCESS);
+        rb.Push<u32>(static_cast<u32>(state));
+    }
+
+    void GetDeviceState(Kernel::HLERequestContext& ctx) {
+        NGLOG_WARNING(Service_NFP, "(STUBBED) called");
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(RESULT_SUCCESS);
+        rb.Push<u32>(static_cast<u32>(device_state));
+    }
+
+    void GetNpadId(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        const u64 dev_handle = rp.Pop<u64>();
+        NGLOG_WARNING(Service_NFP, "(STUBBED) called, dev_handle=0x{:X}", dev_handle);
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(RESULT_SUCCESS);
+        rb.Push<u32>(npad_id);
+    }
+
+    void AttachAvailabilityChangeEvent(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        const u64 dev_handle = rp.Pop<u64>();
+        NGLOG_WARNING(Service_NFP, "(STUBBED) called, dev_handle=0x{:X}", dev_handle);
+
+        IPC::ResponseBuilder rb{ctx, 2, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushCopyObjects(availability_change_event);
+    }
+
+    const u64 device_handle{0xDEAD};
+    const HID::ControllerID npad_id{HID::Controller_Player1};
+    State state{State::NonInitialized};
+    DeviceState device_state{DeviceState::Initialized};
+    Kernel::SharedPtr<Kernel::Event> activate_event;
+    Kernel::SharedPtr<Kernel::Event> deactivate_event;
+    Kernel::SharedPtr<Kernel::Event> availability_change_event;
 };
 
 void Module::Interface::CreateUserInterface(Kernel::HLERequestContext& ctx) {

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

@@ -121,8 +121,9 @@ u32 nvhost_gpu::AllocateObjectContext(const std::vector<u8>& input, std::vector<
 }
 
 u32 nvhost_gpu::SubmitGPFIFO(const std::vector<u8>& input, std::vector<u8>& output) {
-    if (input.size() < sizeof(IoctlSubmitGpfifo))
+    if (input.size() < sizeof(IoctlSubmitGpfifo)) {
         UNIMPLEMENTED();
+    }
     IoctlSubmitGpfifo params{};
     std::memcpy(&params, input.data(), sizeof(IoctlSubmitGpfifo));
     NGLOG_WARNING(Service_NVDRV, "(STUBBED) called, gpfifo={:X}, num_entries={:X}, flags={:X}",

src/core/hle/service/service.cpp

@@ -26,6 +26,7 @@
 #include "core/hle/service/friend/friend.h"
 #include "core/hle/service/hid/hid.h"
 #include "core/hle/service/lm/lm.h"
+#include "core/hle/service/mm/mm_u.h"
 #include "core/hle/service/nfp/nfp.h"
 #include "core/hle/service/nifm/nifm.h"
 #include "core/hle/service/ns/ns.h"
@@ -191,6 +192,7 @@ void Init(std::shared_ptr<SM::ServiceManager>& sm) {
     Friend::InstallInterfaces(*sm);
     HID::InstallInterfaces(*sm);
     LM::InstallInterfaces(*sm);
+    MM::InstallInterfaces(*sm);
     NFP::InstallInterfaces(*sm);
     NIFM::InstallInterfaces(*sm);
     NS::InstallInterfaces(*sm);

src/core/hle/service/set/set.cpp

@@ -12,9 +12,6 @@
 namespace Service::Set {
 
 void SET::GetAvailableLanguageCodes(Kernel::HLERequestContext& ctx) {
-    IPC::RequestParser rp{ctx};
-    u32 id = rp.Pop<u32>();
-
     static constexpr std::array<LanguageCode, 17> available_language_codes = {{
         LanguageCode::JA,
         LanguageCode::EN_US,
@@ -50,7 +47,7 @@ SET::SET() : ServiceFramework("set") {
         {2, nullptr, "MakeLanguageCode"},
         {3, nullptr, "GetAvailableLanguageCodeCount"},
         {4, nullptr, "GetRegionCode"},
-        {5, nullptr, "GetAvailableLanguageCodes2"},
+        {5, &SET::GetAvailableLanguageCodes, "GetAvailableLanguageCodes2"},
         {6, nullptr, "GetAvailableLanguageCodeCount2"},
         {7, nullptr, "GetKeyCodeMap"},
         {8, nullptr, "GetQuestFlag"},

src/core/loader/loader.cpp

@@ -9,6 +9,7 @@
 #include "core/hle/kernel/process.h"
 #include "core/loader/deconstructed_rom_directory.h"
 #include "core/loader/elf.h"
+#include "core/loader/nca.h"
 #include "core/loader/nro.h"
 #include "core/loader/nso.h"
 
@@ -32,6 +33,7 @@ FileType IdentifyFile(FileUtil::IOFile& file, const std::string& filepath) {
     CHECK_TYPE(ELF)
     CHECK_TYPE(NSO)
     CHECK_TYPE(NRO)
+    CHECK_TYPE(NCA)
 
 #undef CHECK_TYPE
 
@@ -57,6 +59,8 @@ FileType GuessFromExtension(const std::string& extension_) {
         return FileType::NRO;
     else if (extension == ".nso")
         return FileType::NSO;
+    else if (extension == ".nca")
+        return FileType::NCA;
 
     return FileType::Unknown;
 }
@@ -69,6 +73,8 @@ const char* GetFileTypeString(FileType type) {
         return "NRO";
     case FileType::NSO:
         return "NSO";
+    case FileType::NCA:
+        return "NCA";
     case FileType::DeconstructedRomDirectory:
         return "Directory";
     case FileType::Error:
@@ -104,6 +110,10 @@ static std::unique_ptr<AppLoader> GetFileLoader(FileUtil::IOFile&& file, FileTyp
     case FileType::NRO:
         return std::make_unique<AppLoader_NRO>(std::move(file), filepath);
 
+    // NX NCA file format.
+    case FileType::NCA:
+        return std::make_unique<AppLoader_NCA>(std::move(file), filepath);
+
     // NX deconstructed ROM directory.
     case FileType::DeconstructedRomDirectory:
         return std::make_unique<AppLoader_DeconstructedRomDirectory>(std::move(file), filepath);

src/core/loader/loader.h

@@ -29,6 +29,7 @@ enum class FileType {
     ELF,
     NSO,
     NRO,
+    NCA,
     DeconstructedRomDirectory,
 };
 

src/core/loader/nca.cpp

@@ -0,0 +1,303 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <vector>
+
+#include "common/common_funcs.h"
+#include "common/file_util.h"
+#include "common/logging/log.h"
+#include "common/swap.h"
+#include "core/core.h"
+#include "core/file_sys/program_metadata.h"
+#include "core/file_sys/romfs_factory.h"
+#include "core/hle/kernel/process.h"
+#include "core/hle/kernel/resource_limit.h"
+#include "core/hle/service/filesystem/filesystem.h"
+#include "core/loader/nca.h"
+#include "core/loader/nso.h"
+#include "core/memory.h"
+
+namespace Loader {
+
+// Media offsets in headers are stored divided by 512. Mult. by this to get real offset.
+constexpr u64 MEDIA_OFFSET_MULTIPLIER = 0x200;
+
+constexpr u64 SECTION_HEADER_SIZE = 0x200;
+constexpr u64 SECTION_HEADER_OFFSET = 0x400;
+
+enum class NcaContentType : u8 { Program = 0, Meta = 1, Control = 2, Manual = 3, Data = 4 };
+
+enum class NcaSectionFilesystemType : u8 { PFS0 = 0x2, ROMFS = 0x3 };
+
+struct NcaSectionTableEntry {
+    u32_le media_offset;
+    u32_le media_end_offset;
+    INSERT_PADDING_BYTES(0x8);
+};
+static_assert(sizeof(NcaSectionTableEntry) == 0x10, "NcaSectionTableEntry has incorrect size.");
+
+struct NcaHeader {
+    std::array<u8, 0x100> rsa_signature_1;
+    std::array<u8, 0x100> rsa_signature_2;
+    u32_le magic;
+    u8 is_system;
+    NcaContentType content_type;
+    u8 crypto_type;
+    u8 key_index;
+    u64_le size;
+    u64_le title_id;
+    INSERT_PADDING_BYTES(0x4);
+    u32_le sdk_version;
+    u8 crypto_type_2;
+    INSERT_PADDING_BYTES(15);
+    std::array<u8, 0x10> rights_id;
+    std::array<NcaSectionTableEntry, 0x4> section_tables;
+    std::array<std::array<u8, 0x20>, 0x4> hash_tables;
+    std::array<std::array<u8, 0x10>, 0x4> key_area;
+    INSERT_PADDING_BYTES(0xC0);
+};
+static_assert(sizeof(NcaHeader) == 0x400, "NcaHeader has incorrect size.");
+
+struct NcaSectionHeaderBlock {
+    INSERT_PADDING_BYTES(3);
+    NcaSectionFilesystemType filesystem_type;
+    u8 crypto_type;
+    INSERT_PADDING_BYTES(3);
+};
+static_assert(sizeof(NcaSectionHeaderBlock) == 0x8, "NcaSectionHeaderBlock has incorrect size.");
+
+struct Pfs0Superblock {
+    NcaSectionHeaderBlock header_block;
+    std::array<u8, 0x20> hash;
+    u32_le size;
+    INSERT_PADDING_BYTES(4);
+    u64_le hash_table_offset;
+    u64_le hash_table_size;
+    u64_le pfs0_header_offset;
+    u64_le pfs0_size;
+    INSERT_PADDING_BYTES(432);
+};
+static_assert(sizeof(Pfs0Superblock) == 0x200, "Pfs0Superblock has incorrect size.");
+
+static bool IsValidNca(const NcaHeader& header) {
+    return header.magic == Common::MakeMagic('N', 'C', 'A', '2') ||
+           header.magic == Common::MakeMagic('N', 'C', 'A', '3');
+}
+
+// TODO(DarkLordZach): Add support for encrypted.
+class Nca final {
+    std::vector<FileSys::PartitionFilesystem> pfs;
+    std::vector<u64> pfs_offset;
+
+    u64 romfs_offset = 0;
+    u64 romfs_size = 0;
+
+    boost::optional<u8> exefs_id = boost::none;
+
+    FileUtil::IOFile file;
+    std::string path;
+
+    u64 GetExeFsFileOffset(const std::string& file_name) const;
+    u64 GetExeFsFileSize(const std::string& file_name) const;
+
+public:
+    ResultStatus Load(FileUtil::IOFile&& file, std::string path);
+
+    FileSys::PartitionFilesystem GetPfs(u8 id) const;
+
+    u64 GetRomFsOffset() const;
+    u64 GetRomFsSize() const;
+
+    std::vector<u8> GetExeFsFile(const std::string& file_name);
+};
+
+static bool IsPfsExeFs(const FileSys::PartitionFilesystem& pfs) {
+    // According to switchbrew, an exefs must only contain these two files:
+    return pfs.GetFileSize("main") > 0 && pfs.GetFileSize("main.npdm") > 0;
+}
+
+ResultStatus Nca::Load(FileUtil::IOFile&& in_file, std::string in_path) {
+    file = std::move(in_file);
+    path = in_path;
+    file.Seek(0, SEEK_SET);
+    std::array<u8, sizeof(NcaHeader)> header_array{};
+    if (sizeof(NcaHeader) != file.ReadBytes(header_array.data(), sizeof(NcaHeader)))
+        NGLOG_CRITICAL(Loader, "File reader errored out during header read.");
+
+    NcaHeader header{};
+    std::memcpy(&header, header_array.data(), sizeof(NcaHeader));
+    if (!IsValidNca(header))
+        return ResultStatus::ErrorInvalidFormat;
+
+    int number_sections =
+        std::count_if(std::begin(header.section_tables), std::end(header.section_tables),
+                      [](NcaSectionTableEntry entry) { return entry.media_offset > 0; });
+
+    for (int i = 0; i < number_sections; ++i) {
+        // Seek to beginning of this section.
+        file.Seek(SECTION_HEADER_OFFSET + i * SECTION_HEADER_SIZE, SEEK_SET);
+        std::array<u8, sizeof(NcaSectionHeaderBlock)> array{};
+        if (sizeof(NcaSectionHeaderBlock) !=
+            file.ReadBytes(array.data(), sizeof(NcaSectionHeaderBlock)))
+            NGLOG_CRITICAL(Loader, "File reader errored out during header read.");
+
+        NcaSectionHeaderBlock block{};
+        std::memcpy(&block, array.data(), sizeof(NcaSectionHeaderBlock));
+
+        if (block.filesystem_type == NcaSectionFilesystemType::ROMFS) {
+            romfs_offset = header.section_tables[i].media_offset * MEDIA_OFFSET_MULTIPLIER;
+            romfs_size =
+                header.section_tables[i].media_end_offset * MEDIA_OFFSET_MULTIPLIER - romfs_offset;
+        } else if (block.filesystem_type == NcaSectionFilesystemType::PFS0) {
+            Pfs0Superblock sb{};
+            // Seek back to beginning of this section.
+            file.Seek(SECTION_HEADER_OFFSET + i * SECTION_HEADER_SIZE, SEEK_SET);
+            if (sizeof(Pfs0Superblock) != file.ReadBytes(&sb, sizeof(Pfs0Superblock)))
+                NGLOG_CRITICAL(Loader, "File reader errored out during header read.");
+
+            u64 offset = (static_cast<u64>(header.section_tables[i].media_offset) *
+                          MEDIA_OFFSET_MULTIPLIER) +
+                         sb.pfs0_header_offset;
+            FileSys::PartitionFilesystem npfs{};
+            ResultStatus status = npfs.Load(path, offset);
+
+            if (status == ResultStatus::Success) {
+                pfs.emplace_back(std::move(npfs));
+                pfs_offset.emplace_back(offset);
+            }
+        }
+    }
+
+    for (size_t i = 0; i < pfs.size(); ++i) {
+        if (IsPfsExeFs(pfs[i]))
+            exefs_id = i;
+    }
+
+    return ResultStatus::Success;
+}
+
+FileSys::PartitionFilesystem Nca::GetPfs(u8 id) const {
+    return pfs[id];
+}
+
+u64 Nca::GetExeFsFileOffset(const std::string& file_name) const {
+    if (exefs_id == boost::none)
+        return 0;
+    return pfs[*exefs_id].GetFileOffset(file_name) + pfs_offset[*exefs_id];
+}
+
+u64 Nca::GetExeFsFileSize(const std::string& file_name) const {
+    if (exefs_id == boost::none)
+        return 0;
+    return pfs[*exefs_id].GetFileSize(file_name);
+}
+
+u64 Nca::GetRomFsOffset() const {
+    return romfs_offset;
+}
+
+u64 Nca::GetRomFsSize() const {
+    return romfs_size;
+}
+
+std::vector<u8> Nca::GetExeFsFile(const std::string& file_name) {
+    std::vector<u8> out(GetExeFsFileSize(file_name));
+    file.Seek(GetExeFsFileOffset(file_name), SEEK_SET);
+    file.ReadBytes(out.data(), GetExeFsFileSize(file_name));
+    return out;
+}
+
+AppLoader_NCA::AppLoader_NCA(FileUtil::IOFile&& file, std::string filepath)
+    : AppLoader(std::move(file)), filepath(std::move(filepath)) {}
+
+FileType AppLoader_NCA::IdentifyType(FileUtil::IOFile& file, const std::string&) {
+    file.Seek(0, SEEK_SET);
+    std::array<u8, 0x400> header_enc_array{};
+    if (0x400 != file.ReadBytes(header_enc_array.data(), 0x400))
+        return FileType::Error;
+
+    // TODO(DarkLordZach): Assuming everything is decrypted. Add crypto support.
+    NcaHeader header{};
+    std::memcpy(&header, header_enc_array.data(), sizeof(NcaHeader));
+
+    if (IsValidNca(header) && header.content_type == NcaContentType::Program)
+        return FileType::NCA;
+
+    return FileType::Error;
+}
+
+ResultStatus AppLoader_NCA::Load(Kernel::SharedPtr<Kernel::Process>& process) {
+    if (is_loaded) {
+        return ResultStatus::ErrorAlreadyLoaded;
+    }
+    if (!file.IsOpen()) {
+        return ResultStatus::Error;
+    }
+
+    nca = std::make_unique<Nca>();
+    ResultStatus result = nca->Load(std::move(file), filepath);
+    if (result != ResultStatus::Success) {
+        return result;
+    }
+
+    result = metadata.Load(nca->GetExeFsFile("main.npdm"));
+    if (result != ResultStatus::Success) {
+        return result;
+    }
+    metadata.Print();
+
+    const FileSys::ProgramAddressSpaceType arch_bits{metadata.GetAddressSpaceType()};
+    if (arch_bits == FileSys::ProgramAddressSpaceType::Is32Bit) {
+        return ResultStatus::ErrorUnsupportedArch;
+    }
+
+    VAddr next_load_addr{Memory::PROCESS_IMAGE_VADDR};
+    for (const auto& module : {"rtld", "main", "subsdk0", "subsdk1", "subsdk2", "subsdk3",
+                               "subsdk4", "subsdk5", "subsdk6", "subsdk7", "sdk"}) {
+        const VAddr load_addr = next_load_addr;
+        next_load_addr = AppLoader_NSO::LoadModule(module, nca->GetExeFsFile(module), load_addr);
+        if (next_load_addr) {
+            NGLOG_DEBUG(Loader, "loaded module {} @ 0x{:X}", module, load_addr);
+        } else {
+            next_load_addr = load_addr;
+        }
+    }
+
+    process->program_id = metadata.GetTitleID();
+    process->svc_access_mask.set();
+    process->address_mappings = default_address_mappings;
+    process->resource_limit =
+        Kernel::ResourceLimit::GetForCategory(Kernel::ResourceLimitCategory::APPLICATION);
+    process->Run(Memory::PROCESS_IMAGE_VADDR, metadata.GetMainThreadPriority(),
+                 metadata.GetMainThreadStackSize());
+
+    if (nca->GetRomFsSize() > 0)
+        Service::FileSystem::RegisterFileSystem(std::make_unique<FileSys::RomFS_Factory>(*this),
+                                                Service::FileSystem::Type::RomFS);
+
+    is_loaded = true;
+    return ResultStatus::Success;
+}
+
+ResultStatus AppLoader_NCA::ReadRomFS(std::shared_ptr<FileUtil::IOFile>& romfs_file, u64& offset,
+                                      u64& size) {
+    if (nca->GetRomFsSize() == 0) {
+        NGLOG_DEBUG(Loader, "No RomFS available");
+        return ResultStatus::ErrorNotUsed;
+    }
+
+    romfs_file = std::make_shared<FileUtil::IOFile>(filepath, "rb");
+
+    offset = nca->GetRomFsOffset();
+    size = nca->GetRomFsSize();
+
+    NGLOG_DEBUG(Loader, "RomFS offset:           0x{:016X}", offset);
+    NGLOG_DEBUG(Loader, "RomFS size:             0x{:016X}", size);
+
+    return ResultStatus::Success;
+}
+
+AppLoader_NCA::~AppLoader_NCA() = default;
+
+} // namespace Loader

src/core/loader/nca.h

@@ -0,0 +1,49 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <string>
+#include "common/common_types.h"
+#include "core/file_sys/partition_filesystem.h"
+#include "core/file_sys/program_metadata.h"
+#include "core/hle/kernel/kernel.h"
+#include "core/loader/loader.h"
+
+namespace Loader {
+
+class Nca;
+
+/// Loads an NCA file
+class AppLoader_NCA final : public AppLoader {
+public:
+    AppLoader_NCA(FileUtil::IOFile&& file, std::string filepath);
+
+    /**
+     * Returns the type of the file
+     * @param file FileUtil::IOFile open file
+     * @param filepath Path of the file that we are opening.
+     * @return FileType found, or FileType::Error if this loader doesn't know it
+     */
+    static FileType IdentifyType(FileUtil::IOFile& file, const std::string& filepath);
+
+    FileType GetFileType() override {
+        return IdentifyType(file, filepath);
+    }
+
+    ResultStatus Load(Kernel::SharedPtr<Kernel::Process>& process) override;
+
+    ResultStatus ReadRomFS(std::shared_ptr<FileUtil::IOFile>& romfs_file, u64& offset,
+                           u64& size) override;
+
+    ~AppLoader_NCA();
+
+private:
+    std::string filepath;
+    FileSys::ProgramMetadata metadata;
+
+    std::unique_ptr<Nca> nca;
+};
+
+} // namespace Loader

src/core/loader/nso.cpp

@@ -66,8 +66,22 @@ FileType AppLoader_NSO::IdentifyType(FileUtil::IOFile& file, const std::string&)
     return FileType::Error;
 }
 
+static std::vector<u8> DecompressSegment(const std::vector<u8>& compressed_data,
+                                         const NsoSegmentHeader& header) {
+    std::vector<u8> uncompressed_data;
+    uncompressed_data.resize(header.size);
+    const int bytes_uncompressed = LZ4_decompress_safe(
+        reinterpret_cast<const char*>(compressed_data.data()),
+        reinterpret_cast<char*>(uncompressed_data.data()), compressed_data.size(), header.size);
+
+    ASSERT_MSG(bytes_uncompressed == header.size && bytes_uncompressed == uncompressed_data.size(),
+               "{} != {} != {}", bytes_uncompressed, header.size, uncompressed_data.size());
+
+    return uncompressed_data;
+}
+
 static std::vector<u8> ReadSegment(FileUtil::IOFile& file, const NsoSegmentHeader& header,
-                                   int compressed_size) {
+                                   size_t compressed_size) {
     std::vector<u8> compressed_data;
     compressed_data.resize(compressed_size);
 
@@ -77,22 +91,65 @@ static std::vector<u8> ReadSegment(FileUtil::IOFile& file, const NsoSegmentHeade
         return {};
     }
 
-    std::vector<u8> uncompressed_data;
-    uncompressed_data.resize(header.size);
-    const int bytes_uncompressed = LZ4_decompress_safe(
-        reinterpret_cast<const char*>(compressed_data.data()),
-        reinterpret_cast<char*>(uncompressed_data.data()), compressed_size, header.size);
-
-    ASSERT_MSG(bytes_uncompressed == header.size && bytes_uncompressed == uncompressed_data.size(),
-               "{} != {} != {}", bytes_uncompressed, header.size, uncompressed_data.size());
-
-    return uncompressed_data;
+    return DecompressSegment(compressed_data, header);
 }
 
 static constexpr u32 PageAlignSize(u32 size) {
     return (size + Memory::PAGE_MASK) & ~Memory::PAGE_MASK;
 }
 
+VAddr AppLoader_NSO::LoadModule(const std::string& name, const std::vector<u8>& file_data,
+                                VAddr load_base) {
+    if (file_data.size() < sizeof(NsoHeader))
+        return {};
+
+    NsoHeader nso_header;
+    std::memcpy(&nso_header, file_data.data(), sizeof(NsoHeader));
+
+    if (nso_header.magic != Common::MakeMagic('N', 'S', 'O', '0'))
+        return {};
+
+    // Build program image
+    Kernel::SharedPtr<Kernel::CodeSet> codeset = Kernel::CodeSet::Create("");
+    std::vector<u8> program_image;
+    for (int i = 0; i < nso_header.segments.size(); ++i) {
+        std::vector<u8> compressed_data(nso_header.segments_compressed_size[i]);
+        for (int j = 0; j < nso_header.segments_compressed_size[i]; ++j)
+            compressed_data[j] = file_data[nso_header.segments[i].offset + j];
+        std::vector<u8> data = DecompressSegment(compressed_data, nso_header.segments[i]);
+        program_image.resize(nso_header.segments[i].location);
+        program_image.insert(program_image.end(), data.begin(), data.end());
+        codeset->segments[i].addr = nso_header.segments[i].location;
+        codeset->segments[i].offset = nso_header.segments[i].location;
+        codeset->segments[i].size = PageAlignSize(static_cast<u32>(data.size()));
+    }
+
+    // MOD header pointer is at .text offset + 4
+    u32 module_offset;
+    std::memcpy(&module_offset, program_image.data() + 4, sizeof(u32));
+
+    // Read MOD header
+    ModHeader mod_header{};
+    // Default .bss to size in segment header if MOD0 section doesn't exist
+    u32 bss_size{PageAlignSize(nso_header.segments[2].bss_size)};
+    std::memcpy(&mod_header, program_image.data() + module_offset, sizeof(ModHeader));
+    const bool has_mod_header{mod_header.magic == Common::MakeMagic('M', 'O', 'D', '0')};
+    if (has_mod_header) {
+        // Resize program image to include .bss section and page align each section
+        bss_size = PageAlignSize(mod_header.bss_end_offset - mod_header.bss_start_offset);
+    }
+    codeset->data.size += bss_size;
+    const u32 image_size{PageAlignSize(static_cast<u32>(program_image.size()) + bss_size)};
+    program_image.resize(image_size);
+
+    // Load codeset for current process
+    codeset->name = name;
+    codeset->memory = std::make_shared<std::vector<u8>>(std::move(program_image));
+    Core::CurrentProcess()->LoadModule(codeset, load_base);
+
+    return load_base + image_size;
+}
+
 VAddr AppLoader_NSO::LoadModule(const std::string& path, VAddr load_base) {
     FileUtil::IOFile file(path, "rb");
     if (!file.IsOpen()) {

src/core/loader/nso.h

@@ -29,6 +29,9 @@ public:
         return IdentifyType(file, filepath);
     }
 
+    static VAddr LoadModule(const std::string& name, const std::vector<u8>& file_data,
+                            VAddr load_base);
+
     static VAddr LoadModule(const std::string& path, VAddr load_base);
 
     ResultStatus Load(Kernel::SharedPtr<Kernel::Process>& process) override;

src/core/memory.cpp

@@ -241,6 +241,10 @@ bool IsValidVirtualAddress(const VAddr vaddr) {
     return IsValidVirtualAddress(*Core::CurrentProcess(), vaddr);
 }
 
+bool IsKernelVirtualAddress(const VAddr vaddr) {
+    return KERNEL_REGION_VADDR <= vaddr && vaddr < KERNEL_REGION_END;
+}
+
 bool IsValidPhysicalAddress(const PAddr paddr) {
     return GetPhysicalPointer(paddr) != nullptr;
 }

src/core/memory.h

@@ -188,6 +188,11 @@ enum : VAddr {
     MAP_REGION_VADDR = NEW_MAP_REGION_VADDR_END,
     MAP_REGION_SIZE = 0x1000000000,
     MAP_REGION_VADDR_END = MAP_REGION_VADDR + MAP_REGION_SIZE,
+
+    /// Kernel Virtual Address Range
+    KERNEL_REGION_VADDR = 0xFFFFFF8000000000,
+    KERNEL_REGION_SIZE = 0x7FFFE00000,
+    KERNEL_REGION_END = KERNEL_REGION_VADDR + KERNEL_REGION_SIZE,
 };
 
 /// Currently active page table
@@ -197,6 +202,8 @@ PageTable* GetCurrentPageTable();
 /// Determines if the given VAddr is valid for the specified process.
 bool IsValidVirtualAddress(const Kernel::Process& process, const VAddr vaddr);
 bool IsValidVirtualAddress(const VAddr addr);
+/// Determines if the given VAddr is a kernel address
+bool IsKernelVirtualAddress(const VAddr addr);
 
 bool IsValidPhysicalAddress(const PAddr addr);
 

src/video_core/CMakeLists.txt

@@ -9,6 +9,8 @@ add_library(video_core STATIC
     engines/maxwell_3d.h
     engines/maxwell_compute.cpp
     engines/maxwell_compute.h
+    engines/maxwell_dma.cpp
+    engines/maxwell_dma.h
     engines/shader_bytecode.h
     gpu.cpp
     gpu.h
@@ -39,6 +41,8 @@ add_library(video_core STATIC
     renderer_opengl/maxwell_to_gl.h
     renderer_opengl/renderer_opengl.cpp
     renderer_opengl/renderer_opengl.h
+    textures/astc.cpp
+    textures/astc.h
     textures/decoders.cpp
     textures/decoders.h
     textures/texture.h

src/video_core/command_processor.cpp

@@ -16,6 +16,7 @@
 #include "video_core/engines/fermi_2d.h"
 #include "video_core/engines/maxwell_3d.h"
 #include "video_core/engines/maxwell_compute.h"
+#include "video_core/engines/maxwell_dma.h"
 #include "video_core/gpu.h"
 #include "video_core/renderer_base.h"
 #include "video_core/video_core.h"
@@ -60,8 +61,11 @@ void GPU::WriteReg(u32 method, u32 subchannel, u32 value, u32 remaining_params)
     case EngineID::MAXWELL_COMPUTE_B:
         maxwell_compute->WriteReg(method, value);
         break;
+    case EngineID::MAXWELL_DMA_COPY_A:
+        maxwell_dma->WriteReg(method, value);
+        break;
     default:
-        UNIMPLEMENTED();
+        UNIMPLEMENTED_MSG("Unimplemented engine");
     }
 }
 

src/video_core/debug_utils/debug_utils.h

@@ -55,8 +55,10 @@ public:
         virtual ~BreakPointObserver() {
             auto context = context_weak.lock();
             if (context) {
-                std::unique_lock<std::mutex> lock(context->breakpoint_mutex);
-                context->breakpoint_observers.remove(this);
+                {
+                    std::unique_lock<std::mutex> lock(context->breakpoint_mutex);
+                    context->breakpoint_observers.remove(this);
+                }
 
                 // If we are the last observer to be destroyed, tell the debugger context that
                 // it is free to continue. In particular, this is required for a proper yuzu

src/video_core/engines/fermi_2d.cpp

@@ -47,6 +47,7 @@ void Fermi2D::HandleSurfaceCopy() {
 
     if (regs.src.linear == regs.dst.linear) {
         // If the input layout and the output layout are the same, just perform a raw copy.
+        ASSERT(regs.src.BlockHeight() == regs.dst.BlockHeight());
         Memory::CopyBlock(dest_cpu, source_cpu,
                           src_bytes_per_pixel * regs.dst.width * regs.dst.height);
         return;

src/video_core/engines/maxwell_3d.cpp

@@ -328,8 +328,9 @@ std::vector<Texture::FullTextureInfo> Maxwell3D::GetStageTextures(Regs::ShaderSt
 
         Texture::FullTextureInfo tex_info{};
         // TODO(Subv): Use the shader to determine which textures are actually accessed.
-        tex_info.index = (current_texture - tex_info_buffer.address - TextureInfoOffset) /
-                         sizeof(Texture::TextureHandle);
+        tex_info.index =
+            static_cast<u32>(current_texture - tex_info_buffer.address - TextureInfoOffset) /
+            sizeof(Texture::TextureHandle);
 
         // Load the TIC data.
         if (tex_handle.tic_id != 0) {

src/video_core/engines/maxwell_3d.h

@@ -318,6 +318,7 @@ public:
             Equation equation_a;
             Factor factor_source_a;
             Factor factor_dest_a;
+            INSERT_PADDING_WORDS(1);
         };
 
         union {
@@ -432,7 +433,27 @@ public:
                     };
                 } rt_control;
 
-                INSERT_PADDING_WORDS(0xCF);
+                INSERT_PADDING_WORDS(0x31);
+
+                u32 independent_blend_enable;
+
+                INSERT_PADDING_WORDS(0x15);
+
+                struct {
+                    u32 separate_alpha;
+                    Blend::Equation equation_rgb;
+                    Blend::Factor factor_source_rgb;
+                    Blend::Factor factor_dest_rgb;
+                    Blend::Equation equation_a;
+                    Blend::Factor factor_source_a;
+                    INSERT_PADDING_WORDS(1);
+                    Blend::Factor factor_dest_a;
+
+                    u32 enable_common;
+                    u32 enable[NumRenderTargets];
+                } blend;
+
+                INSERT_PADDING_WORDS(0x77);
 
                 struct {
                     u32 tsc_address_high;
@@ -557,9 +578,7 @@ public:
 
                 } vertex_array[NumVertexArrays];
 
-                Blend blend;
-
-                INSERT_PADDING_WORDS(0x39);
+                Blend independent_blend[NumRenderTargets];
 
                 struct {
                     u32 limit_high;
@@ -722,6 +741,8 @@ ASSERT_REG_POSITION(vertex_buffer, 0x35D);
 ASSERT_REG_POSITION(zeta, 0x3F8);
 ASSERT_REG_POSITION(vertex_attrib_format[0], 0x458);
 ASSERT_REG_POSITION(rt_control, 0x487);
+ASSERT_REG_POSITION(independent_blend_enable, 0x4B9);
+ASSERT_REG_POSITION(blend, 0x4CF);
 ASSERT_REG_POSITION(tsc, 0x557);
 ASSERT_REG_POSITION(tic, 0x55D);
 ASSERT_REG_POSITION(code_address, 0x582);
@@ -729,7 +750,7 @@ ASSERT_REG_POSITION(draw, 0x585);
 ASSERT_REG_POSITION(index_array, 0x5F2);
 ASSERT_REG_POSITION(query, 0x6C0);
 ASSERT_REG_POSITION(vertex_array[0], 0x700);
-ASSERT_REG_POSITION(blend, 0x780);
+ASSERT_REG_POSITION(independent_blend, 0x780);
 ASSERT_REG_POSITION(vertex_array_limit[0], 0x7C0);
 ASSERT_REG_POSITION(shader_config[0], 0x800);
 ASSERT_REG_POSITION(const_buffer, 0x8E0);

src/video_core/engines/maxwell_dma.cpp

@@ -0,0 +1,69 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/memory.h"
+#include "video_core/engines/maxwell_dma.h"
+#include "video_core/textures/decoders.h"
+
+namespace Tegra {
+namespace Engines {
+
+MaxwellDMA::MaxwellDMA(MemoryManager& memory_manager) : memory_manager(memory_manager) {}
+
+void MaxwellDMA::WriteReg(u32 method, u32 value) {
+    ASSERT_MSG(method < Regs::NUM_REGS,
+               "Invalid MaxwellDMA register, increase the size of the Regs structure");
+
+    regs.reg_array[method] = value;
+
+#define MAXWELLDMA_REG_INDEX(field_name)                                                           \
+    (offsetof(Tegra::Engines::MaxwellDMA::Regs, field_name) / sizeof(u32))
+
+    switch (method) {
+    case MAXWELLDMA_REG_INDEX(exec): {
+        HandleCopy();
+        break;
+    }
+    }
+
+#undef MAXWELLDMA_REG_INDEX
+}
+
+void MaxwellDMA::HandleCopy() {
+    NGLOG_WARNING(HW_GPU, "Requested a DMA copy");
+
+    const GPUVAddr source = regs.src_address.Address();
+    const GPUVAddr dest = regs.dst_address.Address();
+
+    const VAddr source_cpu = *memory_manager.GpuToCpuAddress(source);
+    const VAddr dest_cpu = *memory_manager.GpuToCpuAddress(dest);
+
+    // TODO(Subv): Perform more research and implement all features of this engine.
+    ASSERT(regs.exec.enable_swizzle == 0);
+    ASSERT(regs.exec.enable_2d == 1);
+    ASSERT(regs.exec.query_mode == Regs::QueryMode::None);
+    ASSERT(regs.exec.query_intr == Regs::QueryIntr::None);
+    ASSERT(regs.exec.copy_mode == Regs::CopyMode::Unk2);
+    ASSERT(regs.src_params.pos_x == 0);
+    ASSERT(regs.src_params.pos_y == 0);
+    ASSERT(regs.dst_params.pos_x == 0);
+    ASSERT(regs.dst_params.pos_y == 0);
+    ASSERT(regs.exec.is_dst_linear != regs.exec.is_src_linear);
+
+    u8* src_buffer = Memory::GetPointer(source_cpu);
+    u8* dst_buffer = Memory::GetPointer(dest_cpu);
+
+    if (regs.exec.is_dst_linear && !regs.exec.is_src_linear) {
+        // If the input is tiled and the output is linear, deswizzle the input and copy it over.
+        Texture::CopySwizzledData(regs.src_params.size_x, regs.src_params.size_y, 1, 1, src_buffer,
+                                  dst_buffer, true, regs.src_params.BlockHeight());
+    } else {
+        // If the input is linear and the output is tiled, swizzle the input and copy it over.
+        Texture::CopySwizzledData(regs.dst_params.size_x, regs.dst_params.size_y, 1, 1, dst_buffer,
+                                  src_buffer, false, regs.dst_params.BlockHeight());
+    }
+}
+
+} // namespace Engines
+} // namespace Tegra

src/video_core/engines/maxwell_dma.h

@@ -0,0 +1,155 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include "common/assert.h"
+#include "common/bit_field.h"
+#include "common/common_funcs.h"
+#include "common/common_types.h"
+#include "video_core/gpu.h"
+#include "video_core/memory_manager.h"
+
+namespace Tegra {
+namespace Engines {
+
+class MaxwellDMA final {
+public:
+    explicit MaxwellDMA(MemoryManager& memory_manager);
+    ~MaxwellDMA() = default;
+
+    /// Write the value to the register identified by method.
+    void WriteReg(u32 method, u32 value);
+
+    struct Regs {
+        static constexpr size_t NUM_REGS = 0x1D6;
+
+        struct Parameters {
+            union {
+                BitField<0, 4, u32> block_depth;
+                BitField<4, 4, u32> block_height;
+                BitField<8, 4, u32> block_width;
+            };
+            u32 size_x;
+            u32 size_y;
+            u32 size_z;
+            u32 pos_z;
+            union {
+                BitField<0, 16, u32> pos_x;
+                BitField<16, 16, u32> pos_y;
+            };
+
+            u32 BlockHeight() const {
+                return 1 << block_height;
+            }
+        };
+
+        static_assert(sizeof(Parameters) == 24, "Parameters has wrong size");
+
+        enum class CopyMode : u32 {
+            None = 0,
+            Unk1 = 1,
+            Unk2 = 2,
+        };
+
+        enum class QueryMode : u32 {
+            None = 0,
+            Short = 1,
+            Long = 2,
+        };
+
+        enum class QueryIntr : u32 {
+            None = 0,
+            Block = 1,
+            NonBlock = 2,
+        };
+
+        union {
+            struct {
+                INSERT_PADDING_WORDS(0xC0);
+
+                struct {
+                    union {
+                        BitField<0, 2, CopyMode> copy_mode;
+                        BitField<2, 1, u32> flush;
+
+                        BitField<3, 2, QueryMode> query_mode;
+                        BitField<5, 2, QueryIntr> query_intr;
+
+                        BitField<7, 1, u32> is_src_linear;
+                        BitField<8, 1, u32> is_dst_linear;
+
+                        BitField<9, 1, u32> enable_2d;
+                        BitField<10, 1, u32> enable_swizzle;
+                    };
+                } exec;
+
+                INSERT_PADDING_WORDS(0x3F);
+
+                struct {
+                    u32 address_high;
+                    u32 address_low;
+
+                    GPUVAddr Address() const {
+                        return static_cast<GPUVAddr>((static_cast<GPUVAddr>(address_high) << 32) |
+                                                     address_low);
+                    }
+                } src_address;
+
+                struct {
+                    u32 address_high;
+                    u32 address_low;
+
+                    GPUVAddr Address() const {
+                        return static_cast<GPUVAddr>((static_cast<GPUVAddr>(address_high) << 32) |
+                                                     address_low);
+                    }
+                } dst_address;
+
+                u32 src_pitch;
+                u32 dst_pitch;
+                u32 x_count;
+                u32 y_count;
+
+                INSERT_PADDING_WORDS(0xBB);
+
+                Parameters dst_params;
+
+                INSERT_PADDING_WORDS(1);
+
+                Parameters src_params;
+
+                INSERT_PADDING_WORDS(0x13);
+            };
+            std::array<u32, NUM_REGS> reg_array;
+        };
+    } regs{};
+
+    MemoryManager& memory_manager;
+
+private:
+    /// Performs the copy from the source buffer to the destination buffer as configured in the
+    /// registers.
+    void HandleCopy();
+};
+
+#define ASSERT_REG_POSITION(field_name, position)                                                  \
+    static_assert(offsetof(MaxwellDMA::Regs, field_name) == position * 4,                          \
+                  "Field " #field_name " has invalid position")
+
+ASSERT_REG_POSITION(exec, 0xC0);
+ASSERT_REG_POSITION(src_address, 0x100);
+ASSERT_REG_POSITION(dst_address, 0x102);
+ASSERT_REG_POSITION(src_pitch, 0x104);
+ASSERT_REG_POSITION(dst_pitch, 0x105);
+ASSERT_REG_POSITION(x_count, 0x106);
+ASSERT_REG_POSITION(y_count, 0x107);
+ASSERT_REG_POSITION(dst_params, 0x1C3);
+ASSERT_REG_POSITION(src_params, 0x1CA);
+
+#undef ASSERT_REG_POSITION
+
+} // namespace Engines
+} // namespace Tegra

src/video_core/engines/shader_bytecode.h

@@ -168,13 +168,22 @@ enum class SubOp : u64 {
     Min = 0x8,
 };
 
-enum class FloatRoundingOp : u64 {
+enum class F2iRoundingOp : u64 {
     None = 0,
     Floor = 1,
     Ceil = 2,
     Trunc = 3,
 };
 
+enum class F2fRoundingOp : u64 {
+    None = 0,
+    Pass = 3,
+    Round = 8,
+    Floor = 9,
+    Ceil = 10,
+    Trunc = 11,
+};
+
 enum class UniformType : u64 {
     UnsignedByte = 0,
     SignedByte = 1,
@@ -207,12 +216,12 @@ union Instruction {
 
     union {
         BitField<20, 19, u64> imm20_19;
-        BitField<20, 32, u64> imm20_32;
+        BitField<20, 32, s64> imm20_32;
         BitField<45, 1, u64> negate_b;
         BitField<46, 1, u64> abs_a;
         BitField<48, 1, u64> negate_a;
         BitField<49, 1, u64> abs_b;
-        BitField<50, 1, u64> abs_d;
+        BitField<50, 1, u64> saturate_d;
         BitField<56, 1, u64> negate_imm;
 
         union {
@@ -220,11 +229,19 @@ union Instruction {
             BitField<42, 1, u64> negate_pred;
         } fmnmx;
 
+        union {
+            BitField<39, 1, u64> invert_a;
+            BitField<40, 1, u64> invert_b;
+            BitField<41, 2, LogicOperation> operation;
+            BitField<44, 2, u64> unk44;
+            BitField<48, 3, Pred> pred48;
+        } lop;
+
         union {
             BitField<53, 2, LogicOperation> operation;
             BitField<55, 1, u64> invert_a;
             BitField<56, 1, u64> invert_b;
-        } lop;
+        } lop32i;
 
         float GetImm20_19() const {
             float result{};
@@ -237,7 +254,7 @@ union Instruction {
 
         float GetImm20_32() const {
             float result{};
-            u32 imm{static_cast<u32>(imm20_32)};
+            s32 imm{static_cast<s32>(imm20_32)};
             std::memcpy(&result, &imm, sizeof(imm));
             return result;
         }
@@ -250,11 +267,31 @@ union Instruction {
         }
     } alu;
 
+    union {
+        BitField<48, 1, u64> is_signed;
+    } shift;
+
     union {
         BitField<39, 5, u64> shift_amount;
         BitField<48, 1, u64> negate_b;
         BitField<49, 1, u64> negate_a;
-    } iscadd;
+    } alu_integer;
+
+    union {
+        BitField<54, 1, u64> saturate;
+        BitField<56, 1, u64> negate_a;
+    } iadd32i;
+
+    union {
+        BitField<20, 8, u64> shift_position;
+        BitField<28, 8, u64> shift_length;
+        BitField<48, 1, u64> negate_b;
+        BitField<49, 1, u64> negate_a;
+
+        u64 GetLeftShiftValue() const {
+            return 32 - (shift_position + shift_length);
+        }
+    } bfe;
 
     union {
         BitField<48, 1, u64> negate_b;
@@ -305,20 +342,29 @@ union Instruction {
     } fset;
 
     union {
-        BitField<10, 2, Register::Size> size;
+        BitField<39, 3, u64> pred39;
+        BitField<42, 1, u64> neg_pred;
+        BitField<44, 1, u64> bf;
+        BitField<45, 2, PredOperation> op;
+        BitField<48, 1, u64> is_signed;
+        BitField<49, 3, PredCondition> cond;
+    } iset;
+
+    union {
+        BitField<8, 2, Register::Size> dest_size;
+        BitField<10, 2, Register::Size> src_size;
         BitField<12, 1, u64> is_output_signed;
         BitField<13, 1, u64> is_input_signed;
         BitField<41, 2, u64> selector;
         BitField<45, 1, u64> negate_a;
         BitField<49, 1, u64> abs_a;
-        BitField<50, 1, u64> saturate_a;
 
         union {
-            BitField<39, 2, FloatRoundingOp> rounding;
+            BitField<39, 2, F2iRoundingOp> rounding;
         } f2i;
 
         union {
-            BitField<39, 4, u64> rounding;
+            BitField<39, 4, F2fRoundingOp> rounding;
         } f2f;
     } conversion;
 
@@ -326,7 +372,7 @@ union Instruction {
         BitField<31, 4, u64> component_mask;
 
         bool IsComponentEnabled(size_t component) const {
-            return ((1 << component) & component_mask) != 0;
+            return ((1ull << component) & component_mask) != 0;
         }
     } tex;
 
@@ -345,7 +391,7 @@ union Instruction {
 
             ASSERT(component_mask_selector < mask.size());
 
-            return ((1 << component) & mask[component_mask_selector]) != 0;
+            return ((1ull << component) & mask[component_mask_selector]) != 0;
         }
     } texs;
 
@@ -390,6 +436,10 @@ class OpCode {
 public:
     enum class Id {
         KIL,
+        SSY,
+        BFE_C,
+        BFE_R,
+        BFE_IMM,
         BRA,
         LD_A,
         LD_C,
@@ -411,6 +461,10 @@ public:
         FMUL_R,
         FMUL_IMM,
         FMUL32_IMM,
+        IADD_C,
+        IADD_R,
+        IADD_IMM,
+        IADD32I,
         ISCADD_C, // Scale and Add
         ISCADD_R,
         ISCADD_IMM,
@@ -430,6 +484,9 @@ public:
         I2I_C,
         I2I_R,
         I2I_IMM,
+        LOP_C,
+        LOP_R,
+        LOP_IMM,
         LOP32I,
         MOV_C,
         MOV_R,
@@ -444,6 +501,9 @@ public:
         FMNMX_C,
         FMNMX_R,
         FMNMX_IMM,
+        IMNMX_C,
+        IMNMX_R,
+        IMNMX_IMM,
         FSETP_C, // Set Predicate
         FSETP_R,
         FSETP_IMM,
@@ -453,20 +513,30 @@ public:
         ISETP_C,
         ISETP_IMM,
         ISETP_R,
+        ISET_R,
+        ISET_C,
+        ISET_IMM,
         PSETP,
+        XMAD_IMM,
+        XMAD_CR,
+        XMAD_RC,
+        XMAD_RR,
     };
 
     enum class Type {
         Trivial,
         Arithmetic,
-        Logic,
+        ArithmeticImmediate,
+        ArithmeticInteger,
+        ArithmeticIntegerImmediate,
+        Bfe,
         Shift,
-        ScaledAdd,
         Ffma,
         Flow,
         Memory,
         FloatSet,
         FloatSetPredicate,
+        IntegerSet,
         IntegerSetPredicate,
         PredicateSetPredicate,
         Conversion,
@@ -565,6 +635,7 @@ private:
         std::vector<Matcher> table = {
 #define INST(bitstring, op, type, name) Detail::GetMatcher(bitstring, op, type, name)
             INST("111000110011----", Id::KIL, Type::Flow, "KIL"),
+            INST("111000101001----", Id::SSY, Type::Flow, "SSY"),
             INST("111000100100----", Id::BRA, Type::Flow, "BRA"),
             INST("1110111111011---", Id::LD_A, Type::Memory, "LD_A"),
             INST("1110111110010---", Id::LD_C, Type::Memory, "LD_C"),
@@ -585,10 +656,14 @@ private:
             INST("0100110001101---", Id::FMUL_C, Type::Arithmetic, "FMUL_C"),
             INST("0101110001101---", Id::FMUL_R, Type::Arithmetic, "FMUL_R"),
             INST("0011100-01101---", Id::FMUL_IMM, Type::Arithmetic, "FMUL_IMM"),
-            INST("00011110--------", Id::FMUL32_IMM, Type::Arithmetic, "FMUL32_IMM"),
-            INST("0100110000011---", Id::ISCADD_C, Type::ScaledAdd, "ISCADD_C"),
-            INST("0101110000011---", Id::ISCADD_R, Type::ScaledAdd, "ISCADD_R"),
-            INST("0011100-00011---", Id::ISCADD_IMM, Type::ScaledAdd, "ISCADD_IMM"),
+            INST("00011110--------", Id::FMUL32_IMM, Type::ArithmeticImmediate, "FMUL32_IMM"),
+            INST("0100110000010---", Id::IADD_C, Type::ArithmeticInteger, "IADD_C"),
+            INST("0101110000010---", Id::IADD_R, Type::ArithmeticInteger, "IADD_R"),
+            INST("0011100-00010---", Id::IADD_IMM, Type::ArithmeticInteger, "IADD_IMM"),
+            INST("0001110---------", Id::IADD32I, Type::ArithmeticIntegerImmediate, "IADD32I"),
+            INST("0100110000011---", Id::ISCADD_C, Type::ArithmeticInteger, "ISCADD_C"),
+            INST("0101110000011---", Id::ISCADD_R, Type::ArithmeticInteger, "ISCADD_R"),
+            INST("0011100-00011---", Id::ISCADD_IMM, Type::ArithmeticInteger, "ISCADD_IMM"),
             INST("0101000010000---", Id::MUFU, Type::Arithmetic, "MUFU"),
             INST("0100110010010---", Id::RRO_C, Type::Arithmetic, "RRO_C"),
             INST("0101110010010---", Id::RRO_R, Type::Arithmetic, "RRO_R"),
@@ -602,11 +677,20 @@ private:
             INST("0100110010011---", Id::MOV_C, Type::Arithmetic, "MOV_C"),
             INST("0101110010011---", Id::MOV_R, Type::Arithmetic, "MOV_R"),
             INST("0011100-10011---", Id::MOV_IMM, Type::Arithmetic, "MOV_IMM"),
-            INST("000000010000----", Id::MOV32_IMM, Type::Arithmetic, "MOV32_IMM"),
+            INST("000000010000----", Id::MOV32_IMM, Type::ArithmeticImmediate, "MOV32_IMM"),
             INST("0100110001100---", Id::FMNMX_C, Type::Arithmetic, "FMNMX_C"),
             INST("0101110001100---", Id::FMNMX_R, Type::Arithmetic, "FMNMX_R"),
             INST("0011100-01100---", Id::FMNMX_IMM, Type::Arithmetic, "FMNMX_IMM"),
-            INST("000001----------", Id::LOP32I, Type::Logic, "LOP32I"),
+            INST("0100110000100---", Id::IMNMX_C, Type::Arithmetic, "FMNMX_IMM"),
+            INST("0101110000100---", Id::IMNMX_R, Type::Arithmetic, "FMNMX_IMM"),
+            INST("0011100-00100---", Id::IMNMX_IMM, Type::Arithmetic, "FMNMX_IMM"),
+            INST("0100110000000---", Id::BFE_C, Type::Bfe, "BFE_C"),
+            INST("0101110000000---", Id::BFE_R, Type::Bfe, "BFE_R"),
+            INST("0011100-00000---", Id::BFE_IMM, Type::Bfe, "BFE_IMM"),
+            INST("0100110001000---", Id::LOP_C, Type::ArithmeticInteger, "LOP_C"),
+            INST("0101110001000---", Id::LOP_R, Type::ArithmeticInteger, "LOP_R"),
+            INST("0011100001000---", Id::LOP_IMM, Type::ArithmeticInteger, "LOP_IMM"),
+            INST("000001----------", Id::LOP32I, Type::ArithmeticIntegerImmediate, "LOP32I"),
             INST("0100110001001---", Id::SHL_C, Type::Shift, "SHL_C"),
             INST("0101110001001---", Id::SHL_R, Type::Shift, "SHL_R"),
             INST("0011100-01001---", Id::SHL_IMM, Type::Shift, "SHL_IMM"),
@@ -628,7 +712,14 @@ private:
             INST("010010110110----", Id::ISETP_C, Type::IntegerSetPredicate, "ISETP_C"),
             INST("010110110110----", Id::ISETP_R, Type::IntegerSetPredicate, "ISETP_R"),
             INST("0011011-0110----", Id::ISETP_IMM, Type::IntegerSetPredicate, "ISETP_IMM"),
+            INST("010110110101----", Id::ISET_R, Type::IntegerSet, "ISET_R"),
+            INST("010010110101----", Id::ISET_C, Type::IntegerSet, "ISET_C"),
+            INST("0011011-0101----", Id::ISET_IMM, Type::IntegerSet, "ISET_IMM"),
             INST("0101000010010---", Id::PSETP, Type::PredicateSetPredicate, "PSETP"),
+            INST("0011011-00------", Id::XMAD_IMM, Type::Arithmetic, "XMAD_IMM"),
+            INST("0100111---------", Id::XMAD_CR, Type::Arithmetic, "XMAD_CR"),
+            INST("010100010-------", Id::XMAD_RC, Type::Arithmetic, "XMAD_RC"),
+            INST("0101101100------", Id::XMAD_RR, Type::Arithmetic, "XMAD_RR"),
         };
 #undef INST
         std::stable_sort(table.begin(), table.end(), [](const auto& a, const auto& b) {

src/video_core/gpu.cpp

@@ -5,6 +5,7 @@
 #include "video_core/engines/fermi_2d.h"
 #include "video_core/engines/maxwell_3d.h"
 #include "video_core/engines/maxwell_compute.h"
+#include "video_core/engines/maxwell_dma.h"
 #include "video_core/gpu.h"
 
 namespace Tegra {
@@ -14,6 +15,7 @@ GPU::GPU() {
     maxwell_3d = std::make_unique<Engines::Maxwell3D>(*memory_manager);
     fermi_2d = std::make_unique<Engines::Fermi2D>(*memory_manager);
     maxwell_compute = std::make_unique<Engines::MaxwellCompute>();
+    maxwell_dma = std::make_unique<Engines::MaxwellDMA>(*memory_manager);
 }
 
 GPU::~GPU() = default;

src/video_core/gpu.h

@@ -63,6 +63,7 @@ namespace Engines {
 class Fermi2D;
 class Maxwell3D;
 class MaxwellCompute;
+class MaxwellDMA;
 } // namespace Engines
 
 enum class EngineID {
@@ -103,6 +104,8 @@ private:
     std::unique_ptr<Engines::Fermi2D> fermi_2d;
     /// Compute engine
     std::unique_ptr<Engines::MaxwellCompute> maxwell_compute;
+    /// DMA engine
+    std::unique_ptr<Engines::MaxwellDMA> maxwell_dma;
 };
 
 } // namespace Tegra

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -197,8 +197,8 @@ void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset) {
     ASSERT_MSG(!gpu.regs.shader_config[0].enable, "VertexA is unsupported!");
 
     // Next available bindpoints to use when uploading the const buffers and textures to the GLSL
-    // shaders.
-    u32 current_constbuffer_bindpoint = 0;
+    // shaders. The constbuffer bindpoint starts after the shader stage configuration bind points.
+    u32 current_constbuffer_bindpoint = uniform_buffers.size();
     u32 current_texture_bindpoint = 0;
 
     for (unsigned index = 1; index < Maxwell::MaxShaderProgram; ++index) {
@@ -214,13 +214,17 @@ void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset) {
             continue;
         }
 
+        GLShader::MaxwellUniformData ubo{};
+        ubo.SetFromRegs(gpu.state.shader_stages[stage]);
+        std::memcpy(buffer_ptr, &ubo, sizeof(ubo));
+
+        // Flush the buffer so that the GPU can see the data we just wrote.
+        glFlushMappedBufferRange(GL_ARRAY_BUFFER, buffer_offset, sizeof(ubo));
+
         // Upload uniform data as one UBO per stage
         const GLintptr ubo_offset = buffer_offset;
         copy_buffer(uniform_buffers[stage].handle, ubo_offset,
                     sizeof(GLShader::MaxwellUniformData));
-        GLShader::MaxwellUniformData* ub_ptr =
-            reinterpret_cast<GLShader::MaxwellUniformData*>(buffer_ptr);
-        ub_ptr->SetFromRegs(gpu.state.shader_stages[stage]);
 
         buffer_ptr += sizeof(GLShader::MaxwellUniformData);
         buffer_offset += sizeof(GLShader::MaxwellUniformData);
@@ -345,6 +349,9 @@ void RasterizerOpenGL::DrawArrays() {
     // Sync the viewport
     SyncViewport(surfaces_rect, res_scale);
 
+    // Sync the blend state registers
+    SyncBlendState();
+
     // TODO(bunnei): Sync framebuffer_scale uniform here
     // TODO(bunnei): Sync scissorbox uniform(s) here
 
@@ -430,7 +437,7 @@ void RasterizerOpenGL::DrawArrays() {
 
     // Unbind textures for potential future use as framebuffer attachments
     for (auto& texture_unit : state.texture_units) {
-        texture_unit.texture_2d = 0;
+        texture_unit.Unbind();
     }
     state.Apply();
 
@@ -451,32 +458,7 @@ void RasterizerOpenGL::DrawArrays() {
     }
 }
 
-void RasterizerOpenGL::NotifyMaxwellRegisterChanged(u32 method) {
-    const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
-    switch (method) {
-    case MAXWELL3D_REG_INDEX(blend.separate_alpha):
-        ASSERT_MSG(false, "unimplemented");
-        break;
-    case MAXWELL3D_REG_INDEX(blend.equation_rgb):
-        state.blend.rgb_equation = MaxwellToGL::BlendEquation(regs.blend.equation_rgb);
-        break;
-    case MAXWELL3D_REG_INDEX(blend.factor_source_rgb):
-        state.blend.src_rgb_func = MaxwellToGL::BlendFunc(regs.blend.factor_source_rgb);
-        break;
-    case MAXWELL3D_REG_INDEX(blend.factor_dest_rgb):
-        state.blend.dst_rgb_func = MaxwellToGL::BlendFunc(regs.blend.factor_dest_rgb);
-        break;
-    case MAXWELL3D_REG_INDEX(blend.equation_a):
-        state.blend.a_equation = MaxwellToGL::BlendEquation(regs.blend.equation_a);
-        break;
-    case MAXWELL3D_REG_INDEX(blend.factor_source_a):
-        state.blend.src_a_func = MaxwellToGL::BlendFunc(regs.blend.factor_source_a);
-        break;
-    case MAXWELL3D_REG_INDEX(blend.factor_dest_a):
-        state.blend.dst_a_func = MaxwellToGL::BlendFunc(regs.blend.factor_dest_a);
-        break;
-    }
-}
+void RasterizerOpenGL::NotifyMaxwellRegisterChanged(u32 method) {}
 
 void RasterizerOpenGL::FlushAll() {
     MICROPROFILE_SCOPE(OpenGL_CacheManagement);
@@ -626,32 +608,44 @@ u32 RasterizerOpenGL::SetupConstBuffers(Maxwell::ShaderStage stage, GLuint progr
 
         boost::optional<VAddr> addr = gpu.memory_manager->GpuToCpuAddress(buffer.address);
 
-        std::vector<u8> data;
+        size_t size = 0;
+
         if (used_buffer.IsIndirect()) {
             // Buffer is accessed indirectly, so upload the entire thing
-            data.resize(buffer.size * sizeof(float));
+            size = buffer.size * sizeof(float);
+
+            if (size > MaxConstbufferSize) {
+                NGLOG_ERROR(HW_GPU, "indirect constbuffer size {} exceeds maximum {}", size,
+                            MaxConstbufferSize);
+                size = MaxConstbufferSize;
+            }
         } else {
             // Buffer is accessed directly, upload just what we use
-            data.resize(used_buffer.GetSize() * sizeof(float));
+            size = used_buffer.GetSize() * sizeof(float);
         }
 
+        // Align the actual size so it ends up being a multiple of vec4 to meet the OpenGL std140
+        // UBO alignment requirements.
+        size = Common::AlignUp(size, sizeof(GLvec4));
+        ASSERT_MSG(size <= MaxConstbufferSize, "Constbuffer too big");
+
+        std::vector<u8> data(size);
         Memory::ReadBlock(*addr, data.data(), data.size());
 
-        glBindBuffer(GL_SHADER_STORAGE_BUFFER, buffer_draw_state.ssbo);
-        glBufferData(GL_SHADER_STORAGE_BUFFER, data.size(), data.data(), GL_DYNAMIC_DRAW);
-        glBindBuffer(GL_SHADER_STORAGE_BUFFER, 0);
+        glBindBuffer(GL_UNIFORM_BUFFER, buffer_draw_state.ssbo);
+        glBufferData(GL_UNIFORM_BUFFER, data.size(), data.data(), GL_DYNAMIC_DRAW);
+        glBindBuffer(GL_UNIFORM_BUFFER, 0);
 
         // Now configure the bindpoint of the buffer inside the shader
         std::string buffer_name = used_buffer.GetName();
-        GLuint index =
-            glGetProgramResourceIndex(program, GL_SHADER_STORAGE_BLOCK, buffer_name.c_str());
+        GLuint index = glGetProgramResourceIndex(program, GL_UNIFORM_BLOCK, buffer_name.c_str());
         if (index != -1)
-            glShaderStorageBlockBinding(program, index, buffer_draw_state.bindpoint);
+            glUniformBlockBinding(program, index, buffer_draw_state.bindpoint);
     }
 
     state.Apply();
 
-    return current_bindpoint + entries.size();
+    return current_bindpoint + static_cast<u32>(entries.size());
 }
 
 u32 RasterizerOpenGL::SetupTextures(Maxwell::ShaderStage stage, GLuint program, u32 current_unit,
@@ -697,7 +691,7 @@ u32 RasterizerOpenGL::SetupTextures(Maxwell::ShaderStage stage, GLuint program,
 
     state.Apply();
 
-    return current_unit + entries.size();
+    return current_unit + static_cast<u32>(entries.size());
 }
 
 void RasterizerOpenGL::BindFramebufferSurfaces(const Surface& color_surface,
@@ -756,14 +750,21 @@ void RasterizerOpenGL::SyncDepthOffset() {
     UNREACHABLE();
 }
 
-void RasterizerOpenGL::SyncBlendEnabled() {
-    UNREACHABLE();
-}
+void RasterizerOpenGL::SyncBlendState() {
+    const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
 
-void RasterizerOpenGL::SyncBlendFuncs() {
-    UNREACHABLE();
-}
+    // TODO(Subv): Support more than just render target 0.
+    state.blend.enabled = regs.blend.enable[0] != 0;
 
-void RasterizerOpenGL::SyncBlendColor() {
-    UNREACHABLE();
+    if (!state.blend.enabled)
+        return;
+
+    ASSERT_MSG(regs.independent_blend_enable == 1, "Only independent blending is implemented");
+    ASSERT_MSG(!regs.independent_blend[0].separate_alpha, "Unimplemented");
+    state.blend.rgb_equation = MaxwellToGL::BlendEquation(regs.independent_blend[0].equation_rgb);
+    state.blend.src_rgb_func = MaxwellToGL::BlendFunc(regs.independent_blend[0].factor_source_rgb);
+    state.blend.dst_rgb_func = MaxwellToGL::BlendFunc(regs.independent_blend[0].factor_dest_rgb);
+    state.blend.a_equation = MaxwellToGL::BlendEquation(regs.independent_blend[0].equation_a);
+    state.blend.src_a_func = MaxwellToGL::BlendFunc(regs.independent_blend[0].factor_source_a);
+    state.blend.dst_a_func = MaxwellToGL::BlendFunc(regs.independent_blend[0].factor_dest_a);
 }

src/video_core/renderer_opengl/gl_rasterizer.h

@@ -54,6 +54,11 @@ public:
         OGLShader shader;
     };
 
+    /// Maximum supported size that a constbuffer can have in bytes.
+    static constexpr size_t MaxConstbufferSize = 0x10000;
+    static_assert(MaxConstbufferSize % sizeof(GLvec4) == 0,
+                  "The maximum size of a constbuffer must be a multiple of the size of GLvec4");
+
 private:
     class SamplerInfo {
     public:
@@ -121,14 +126,8 @@ private:
     /// Syncs the depth offset to match the guest state
     void SyncDepthOffset();
 
-    /// Syncs the blend enabled status to match the guest state
-    void SyncBlendEnabled();
-
-    /// Syncs the blend functions to match the guest state
-    void SyncBlendFuncs();
-
-    /// Syncs the blend color to match the guest state
-    void SyncBlendColor();
+    /// Syncs the blend state to match the guest state
+    void SyncBlendState();
 
     bool has_ARB_buffer_storage;
     bool has_ARB_direct_state_access;

src/video_core/renderer_opengl/gl_rasterizer_cache.cpp

@@ -28,6 +28,7 @@
 #include "video_core/engines/maxwell_3d.h"
 #include "video_core/renderer_opengl/gl_rasterizer_cache.h"
 #include "video_core/renderer_opengl/gl_state.h"
+#include "video_core/textures/astc.h"
 #include "video_core/textures/decoders.h"
 #include "video_core/utils.h"
 #include "video_core/video_core.h"
@@ -45,7 +46,7 @@ struct FormatTuple {
 
 static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_format_tuples = {{
     {GL_RGBA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8_REV, false},                    // ABGR8
-    {GL_RGB, GL_RGB, GL_UNSIGNED_SHORT_5_6_5, false},                           // B5G6R5
+    {GL_RGB, GL_RGB, GL_UNSIGNED_SHORT_5_6_5_REV, false},                       // B5G6R5
     {GL_RGB10_A2, GL_RGBA, GL_UNSIGNED_INT_2_10_10_10_REV, false},              // A2B10G10R10
     {GL_RGB5_A1, GL_RGBA, GL_UNSIGNED_SHORT_1_5_5_5_REV, false},                // A1B5G5R5
     {GL_R8, GL_RED, GL_UNSIGNED_BYTE, false},                                   // R8
@@ -55,6 +56,7 @@ static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_form
     {GL_COMPRESSED_RGBA_S3TC_DXT3_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, true}, // DXT23
     {GL_COMPRESSED_RGBA_S3TC_DXT5_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, true}, // DXT45
     {GL_COMPRESSED_RED_RGTC1, GL_RED, GL_UNSIGNED_INT_8_8_8_8, true},           // DXN1
+    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, false},                               // ASTC_2D_4X4
 }};
 
 static const FormatTuple& GetFormatTuple(PixelFormat pixel_format, ComponentType component_type) {
@@ -86,6 +88,23 @@ static u16 GetResolutionScaleFactor() {
                                 : Settings::values.resolution_factor);
 }
 
+static void ConvertASTCToRGBA8(std::vector<u8>& data, PixelFormat format, u32 width, u32 height) {
+    u32 block_width{};
+    u32 block_height{};
+
+    switch (format) {
+    case PixelFormat::ASTC_2D_4X4:
+        block_width = 4;
+        block_height = 4;
+        break;
+    default:
+        NGLOG_CRITICAL(HW_GPU, "Unhandled format: {}", static_cast<u32>(format));
+        UNREACHABLE();
+    }
+
+    data = Tegra::Texture::ASTC::Decompress(data, width, height, block_width, block_height);
+}
+
 template <bool morton_to_gl, PixelFormat format>
 void MortonCopy(u32 stride, u32 block_height, u32 height, u8* gl_buffer, Tegra::GPUVAddr base,
                 Tegra::GPUVAddr start, Tegra::GPUVAddr end) {
@@ -97,6 +116,12 @@ void MortonCopy(u32 stride, u32 block_height, u32 height, u8* gl_buffer, Tegra::
         auto data = Tegra::Texture::UnswizzleTexture(
             *gpu.memory_manager->GpuToCpuAddress(base),
             SurfaceParams::TextureFormatFromPixelFormat(format), stride, height, block_height);
+
+        if (SurfaceParams::IsFormatASTC(format)) {
+            // ASTC formats are converted to RGBA8 in software, as most PC GPUs do not support this
+            ConvertASTCToRGBA8(data, format, stride, height);
+        }
+
         std::memcpy(gl_buffer, data.data(), data.size());
     } else {
         // TODO(bunnei): Assumes the default rendering GOB size of 16 (128 lines). We should check
@@ -118,7 +143,7 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr, Tegra:
         MortonCopy<true, PixelFormat::R8>,           MortonCopy<true, PixelFormat::RGBA16F>,
         MortonCopy<true, PixelFormat::R11FG11FB10F>, MortonCopy<true, PixelFormat::DXT1>,
         MortonCopy<true, PixelFormat::DXT23>,        MortonCopy<true, PixelFormat::DXT45>,
-        MortonCopy<true, PixelFormat::DXN1>,
+        MortonCopy<true, PixelFormat::DXN1>,         MortonCopy<true, PixelFormat::ASTC_2D_4X4>,
 };
 
 static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr, Tegra::GPUVAddr,
@@ -137,6 +162,7 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr, Tegra:
         nullptr,
         nullptr,
         nullptr,
+        MortonCopy<false, PixelFormat::ABGR8>,
 };
 
 // Allocate an uninitialized texture of appropriate size and format for the surface
@@ -169,60 +195,10 @@ static void AllocateSurfaceTexture(GLuint texture, const FormatTuple& format_tup
 static bool BlitTextures(GLuint src_tex, const MathUtil::Rectangle<u32>& src_rect, GLuint dst_tex,
                          const MathUtil::Rectangle<u32>& dst_rect, SurfaceType type,
                          GLuint read_fb_handle, GLuint draw_fb_handle) {
-    OpenGLState state = OpenGLState::GetCurState();
-
-    OpenGLState prev_state = state;
-    SCOPE_EXIT({ prev_state.Apply(); });
-
-    // Make sure textures aren't bound to texture units, since going to bind them to framebuffer
-    // components
-    state.ResetTexture(src_tex);
-    state.ResetTexture(dst_tex);
-
-    state.draw.read_framebuffer = read_fb_handle;
-    state.draw.draw_framebuffer = draw_fb_handle;
-    state.Apply();
-
-    u32 buffers = 0;
-
-    if (type == SurfaceType::ColorTexture) {
-        glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, src_tex,
-                               0);
-        glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0,
-                               0);
-
-        glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, dst_tex,
-                               0);
-        glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0,
-                               0);
-
-        buffers = GL_COLOR_BUFFER_BIT;
-    } else if (type == SurfaceType::Depth) {
-        glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, 0, 0);
-        glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_DEPTH_ATTACHMENT, GL_TEXTURE_2D, src_tex, 0);
-        glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0, 0);
-
-        glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, 0, 0);
-        glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_ATTACHMENT, GL_TEXTURE_2D, dst_tex, 0);
-        glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0, 0);
-
-        buffers = GL_DEPTH_BUFFER_BIT;
-    } else if (type == SurfaceType::DepthStencil) {
-        glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, 0, 0);
-        glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D,
-                               src_tex, 0);
-
-        glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, 0, 0);
-        glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D,
-                               dst_tex, 0);
-
-        buffers = GL_DEPTH_BUFFER_BIT | GL_STENCIL_BUFFER_BIT;
-    }
-
-    glBlitFramebuffer(src_rect.left, src_rect.bottom, src_rect.right, src_rect.top, dst_rect.left,
-                      dst_rect.bottom, dst_rect.right, dst_rect.top, buffers,
-                      buffers == GL_COLOR_BUFFER_BIT ? GL_LINEAR : GL_NEAREST);
 
+    glCopyImageSubData(src_tex, GL_TEXTURE_2D, 0, src_rect.left, src_rect.bottom, 0, dst_tex,
+                       GL_TEXTURE_2D, 0, dst_rect.left, dst_rect.bottom, 0, src_rect.GetWidth(),
+                       src_rect.GetHeight(), 0);
     return true;
 }
 
@@ -599,7 +575,7 @@ void CachedSurface::UploadGLTexture(const MathUtil::Rectangle<u32>& rect, GLuint
         glCompressedTexImage2D(GL_TEXTURE_2D, 0, tuple.internal_format,
                                static_cast<GLsizei>(rect.GetWidth() * GetCompresssionFactor()),
                                static_cast<GLsizei>(rect.GetHeight() * GetCompresssionFactor()), 0,
-                               size, &gl_buffer[buffer_offset]);
+                               static_cast<GLsizei>(size), &gl_buffer[buffer_offset]);
     } else {
         glTexSubImage2D(GL_TEXTURE_2D, 0, x0, y0, static_cast<GLsizei>(rect.GetWidth()),
                         static_cast<GLsizei>(rect.GetHeight()), tuple.format, tuple.type,
@@ -669,7 +645,7 @@ void CachedSurface::DownloadGLTexture(const MathUtil::Rectangle<u32>& rect, GLui
         glActiveTexture(GL_TEXTURE0);
         glGetTexImage(GL_TEXTURE_2D, 0, tuple.format, tuple.type, &gl_buffer[buffer_offset]);
     } else {
-        state.ResetTexture(texture.handle);
+        state.UnbindTexture(texture.handle);
         state.draw.read_framebuffer = read_fb_handle;
         state.Apply();
 
@@ -938,9 +914,6 @@ Surface RasterizerCacheOpenGL::GetSurface(const SurfaceParams& params, ScaleMatc
     // Use GetSurfaceSubRect instead
     ASSERT(params.width == params.stride);
 
-    ASSERT(!params.is_tiled ||
-           (params.GetActualWidth() % 8 == 0 && params.GetActualHeight() % 8 == 0));
-
     // Check for an exact match in existing surfaces
     Surface surface =
         FindMatch<MatchFlags::Exact | MatchFlags::Invalid>(surface_cache, params, match_res_scale);
@@ -1083,8 +1056,11 @@ Surface RasterizerCacheOpenGL::GetTextureSurface(const Tegra::Texture::FullTextu
     params.addr = config.tic.Address();
     params.is_tiled = config.tic.IsTiled();
     params.pixel_format = SurfaceParams::PixelFormatFromTextureFormat(config.tic.format);
-    params.width = config.tic.Width() / params.GetCompresssionFactor();
-    params.height = config.tic.Height() / params.GetCompresssionFactor();
+
+    params.width = Common::AlignUp(config.tic.Width(), params.GetCompresssionFactor()) /
+                   params.GetCompresssionFactor();
+    params.height = Common::AlignUp(config.tic.Height(), params.GetCompresssionFactor()) /
+                    params.GetCompresssionFactor();
 
     // TODO(Subv): Different types per component are not supported.
     ASSERT(config.tic.r_type.Value() == config.tic.g_type.Value() &&
@@ -1095,6 +1071,13 @@ Surface RasterizerCacheOpenGL::GetTextureSurface(const Tegra::Texture::FullTextu
 
     if (config.tic.IsTiled()) {
         params.block_height = config.tic.BlockHeight();
+
+        // TODO(bunnei): The below align up is a hack. This is here because some compressed textures
+        // are not a multiple of their own compression factor, and so this accounts for that. This
+        // could potentially result in an extra row of 4px being decoded if a texture is not a
+        // multiple of 4.
+        params.width = Common::AlignUp(params.width, 4);
+        params.height = Common::AlignUp(params.height, 4);
     } else {
         // Use the texture-provided stride value if the texture isn't tiled.
         params.stride = static_cast<u32>(params.PixelsInBytes(config.tic.Pitch()));
@@ -1102,23 +1085,6 @@ Surface RasterizerCacheOpenGL::GetTextureSurface(const Tegra::Texture::FullTextu
 
     params.UpdateParams();
 
-    if (config.tic.Width() % 8 != 0 || config.tic.Height() % 8 != 0 ||
-        params.stride != params.width) {
-        Surface src_surface;
-        MathUtil::Rectangle<u32> rect;
-        std::tie(src_surface, rect) = GetSurfaceSubRect(params, ScaleMatch::Ignore, true);
-
-        params.res_scale = src_surface->res_scale;
-        Surface tmp_surface = CreateSurface(params);
-        BlitTextures(src_surface->texture.handle, rect, tmp_surface->texture.handle,
-                     tmp_surface->GetScaledRect(),
-                     SurfaceParams::GetFormatType(params.pixel_format), read_framebuffer.handle,
-                     draw_framebuffer.handle);
-
-        remove_surfaces.emplace(tmp_surface);
-        return tmp_surface;
-    }
-
     return GetSurface(params, ScaleMatch::Ignore, true);
 }
 
@@ -1293,7 +1259,7 @@ void RasterizerCacheOpenGL::ValidateSurface(const Surface& surface, Tegra::GPUVA
 
         const auto interval = *it & validate_interval;
         // Look for a valid surface to copy from
-        SurfaceParams params = surface->FromInterval(interval);
+        SurfaceParams params = *surface;
 
         Surface copy_surface =
             FindMatch<MatchFlags::Copy>(surface_cache, params, ScaleMatch::Ignore, interval);

src/video_core/renderer_opengl/gl_rasterizer_cache.h

@@ -65,6 +65,7 @@ struct SurfaceParams {
         DXT23 = 8,
         DXT45 = 9,
         DXN1 = 10, // This is also known as BC4
+        ASTC_2D_4X4 = 11,
 
         Max,
         Invalid = 255,
@@ -111,6 +112,7 @@ struct SurfaceParams {
             4, // DXT23
             4, // DXT45
             4, // DXN1
+            1, // ASTC_2D_4X4
         }};
 
         ASSERT(static_cast<size_t>(format) < compression_factor_table.size());
@@ -136,6 +138,7 @@ struct SurfaceParams {
             128, // DXT23
             128, // DXT45
             64,  // DXN1
+            32,  // ASTC_2D_4X4
         }};
 
         ASSERT(static_cast<size_t>(format) < bpp_table.size());
@@ -162,6 +165,15 @@ struct SurfaceParams {
         }
     }
 
+    static bool IsFormatASTC(PixelFormat format) {
+        switch (format) {
+        case PixelFormat::ASTC_2D_4X4:
+            return true;
+        default:
+            return false;
+        }
+    }
+
     static PixelFormat PixelFormatFromGPUPixelFormat(Tegra::FramebufferConfig::PixelFormat format) {
         switch (format) {
         case Tegra::FramebufferConfig::PixelFormat::ABGR8:
@@ -197,6 +209,8 @@ struct SurfaceParams {
             return PixelFormat::DXT45;
         case Tegra::Texture::TextureFormat::DXN1:
             return PixelFormat::DXN1;
+        case Tegra::Texture::TextureFormat::ASTC_2D_4X4:
+            return PixelFormat::ASTC_2D_4X4;
         default:
             NGLOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
             UNREACHABLE();
@@ -228,6 +242,8 @@ struct SurfaceParams {
             return Tegra::Texture::TextureFormat::DXT45;
         case PixelFormat::DXN1:
             return Tegra::Texture::TextureFormat::DXN1;
+        case PixelFormat::ASTC_2D_4X4:
+            return Tegra::Texture::TextureFormat::ASTC_2D_4X4;
         default:
             UNREACHABLE();
         }

src/video_core/renderer_opengl/gl_resource_manager.h

@@ -38,7 +38,7 @@ public:
         if (handle == 0)
             return;
         glDeleteTextures(1, &handle);
-        OpenGLState::GetCurState().ResetTexture(handle).Apply();
+        OpenGLState::GetCurState().UnbindTexture(handle).Apply();
         handle = 0;
     }
 

src/video_core/renderer_opengl/gl_shader_decompiler.cpp

@@ -9,6 +9,7 @@
 #include "common/assert.h"
 #include "common/common_types.h"
 #include "video_core/engines/shader_bytecode.h"
+#include "video_core/renderer_opengl/gl_rasterizer.h"
 #include "video_core/renderer_opengl/gl_shader_decompiler.h"
 
 namespace GLShader {
@@ -16,6 +17,7 @@ namespace Decompiler {
 
 using Tegra::Shader::Attribute;
 using Tegra::Shader::Instruction;
+using Tegra::Shader::LogicOperation;
 using Tegra::Shader::OpCode;
 using Tegra::Shader::Register;
 using Tegra::Shader::Sampler;
@@ -265,6 +267,27 @@ public:
         BuildRegisterList();
     }
 
+    /**
+     * Returns code that does an integer size conversion for the specified size.
+     * @param value Value to perform integer size conversion on.
+     * @param size Register size to use for conversion instructions.
+     * @returns GLSL string corresponding to the value converted to the specified size.
+     */
+    static std::string ConvertIntegerSize(const std::string& value, Register::Size size) {
+        switch (size) {
+        case Register::Size::Byte:
+            return "((" + value + " << 24) >> 24)";
+        case Register::Size::Short:
+            return "((" + value + " << 16) >> 16)";
+        case Register::Size::Word:
+            // Default - do nothing
+            return value;
+        default:
+            NGLOG_CRITICAL(HW_GPU, "Unimplemented conversion size {}", static_cast<u32>(size));
+            UNREACHABLE();
+        }
+    }
+
     /**
      * Gets a register as an float.
      * @param reg The register to get.
@@ -281,15 +304,18 @@ public:
      * @param reg The register to get.
      * @param elem The element to use for the operation.
      * @param is_signed Whether to get the register as a signed (or unsigned) integer.
+     * @param size Register size to use for conversion instructions.
      * @returns GLSL string corresponding to the register as an integer.
      */
-    std::string GetRegisterAsInteger(const Register& reg, unsigned elem = 0,
-                                     bool is_signed = true) {
+    std::string GetRegisterAsInteger(const Register& reg, unsigned elem = 0, bool is_signed = true,
+                                     Register::Size size = Register::Size::Word) {
         const std::string func = GetGLSLConversionFunc(
             GLSLRegister::Type::Float,
             is_signed ? GLSLRegister::Type::Integer : GLSLRegister::Type::UnsignedInteger);
 
-        return func + '(' + GetRegister(reg, elem) + ')';
+        std::string value = func + '(' + GetRegister(reg, elem) + ')';
+
+        return ConvertIntegerSize(value, size);
     }
 
     /**
@@ -299,13 +325,15 @@ public:
      * @param value The code representing the value to assign.
      * @param dest_num_components Number of components in the destination.
      * @param value_num_components Number of components in the value.
-     * @param is_abs Optional, when True, applies absolute value to output.
+     * @param is_saturated Optional, when True, saturates the provided value.
      * @param dest_elem Optional, the destination element to use for the operation.
      */
     void SetRegisterToFloat(const Register& reg, u64 elem, const std::string& value,
-                            u64 dest_num_components, u64 value_num_components, bool is_abs = false,
-                            u64 dest_elem = 0) {
-        SetRegister(reg, elem, value, dest_num_components, value_num_components, is_abs, dest_elem);
+                            u64 dest_num_components, u64 value_num_components,
+                            bool is_saturated = false, u64 dest_elem = 0) {
+
+        SetRegister(reg, elem, is_saturated ? "clamp(" + value + ", 0.0, 1.0)" : value,
+                    dest_num_components, value_num_components, dest_elem);
     }
 
     /**
@@ -315,18 +343,22 @@ public:
      * @param value The code representing the value to assign.
      * @param dest_num_components Number of components in the destination.
      * @param value_num_components Number of components in the value.
-     * @param is_abs Optional, when True, applies absolute value to output.
+     * @param is_saturated Optional, when True, saturates the provided value.
      * @param dest_elem Optional, the destination element to use for the operation.
+     * @param size Register size to use for conversion instructions.
      */
     void SetRegisterToInteger(const Register& reg, bool is_signed, u64 elem,
                               const std::string& value, u64 dest_num_components,
-                              u64 value_num_components, bool is_abs = false, u64 dest_elem = 0) {
+                              u64 value_num_components, bool is_saturated = false,
+                              u64 dest_elem = 0, Register::Size size = Register::Size::Word) {
+        ASSERT_MSG(!is_saturated, "Unimplemented");
+
         const std::string func = GetGLSLConversionFunc(
             is_signed ? GLSLRegister::Type::Integer : GLSLRegister::Type::UnsignedInteger,
             GLSLRegister::Type::Float);
 
-        SetRegister(reg, elem, func + '(' + value + ')', dest_num_components, value_num_components,
-                    is_abs, dest_elem);
+        SetRegister(reg, elem, func + '(' + ConvertIntegerSize(value, size) + ')',
+                    dest_num_components, value_num_components, dest_elem);
     }
 
     /**
@@ -366,7 +398,8 @@ public:
     /// Generates code representing a uniform (C buffer) register, interpreted as the input type.
     std::string GetUniform(u64 index, u64 offset, GLSLRegister::Type type) {
         declr_const_buffers[index].MarkAsUsed(index, offset, stage);
-        std::string value = 'c' + std::to_string(index) + '[' + std::to_string(offset) + ']';
+        std::string value = 'c' + std::to_string(index) + '[' + std::to_string(offset / 4) + "][" +
+                            std::to_string(offset % 4) + ']';
 
         if (type == GLSLRegister::Type::Float) {
             return value;
@@ -380,8 +413,12 @@ public:
     std::string GetUniformIndirect(u64 index, s64 offset, const Register& index_reg,
                                    GLSLRegister::Type type) {
         declr_const_buffers[index].MarkAsUsedIndirect(index, stage);
-        std::string value = 'c' + std::to_string(index) + "[(floatBitsToInt(" +
-                            GetRegister(index_reg, 0) + ") + " + std::to_string(offset) + ") / 4]";
+
+        std::string final_offset = "((floatBitsToInt(" + GetRegister(index_reg, 0) + ") + " +
+                                   std::to_string(offset) + ") / 4)";
+
+        std::string value =
+            'c' + std::to_string(index) + '[' + final_offset + " / 4][" + final_offset + " % 4]";
 
         if (type == GLSLRegister::Type::Float) {
             return value;
@@ -423,9 +460,10 @@ public:
 
         unsigned const_buffer_layout = 0;
         for (const auto& entry : GetConstBuffersDeclarations()) {
-            declarations.AddLine("layout(std430) buffer " + entry.GetName());
+            declarations.AddLine("layout(std140) uniform " + entry.GetName());
             declarations.AddLine('{');
-            declarations.AddLine("    float c" + std::to_string(entry.GetIndex()) + "[];");
+            declarations.AddLine("    vec4 c" + std::to_string(entry.GetIndex()) +
+                                 "[MAX_CONSTBUFFER_ELEMENTS];");
             declarations.AddLine("};");
             declarations.AddNewLine();
             ++const_buffer_layout;
@@ -500,13 +538,11 @@ private:
      * @param value The code representing the value to assign.
      * @param dest_num_components Number of components in the destination.
      * @param value_num_components Number of components in the value.
-     * @param is_abs Optional, when True, applies absolute value to output.
      * @param dest_elem Optional, the destination element to use for the operation.
      */
     void SetRegister(const Register& reg, u64 elem, const std::string& value,
-                     u64 dest_num_components, u64 value_num_components, bool is_abs,
-                     u64 dest_elem) {
-        std::string dest = GetRegister(reg, dest_elem);
+                     u64 dest_num_components, u64 value_num_components, u64 dest_elem) {
+        std::string dest = GetRegister(reg, static_cast<u32>(dest_elem));
         if (dest_num_components > 1) {
             dest += GetSwizzle(elem);
         }
@@ -516,8 +552,6 @@ private:
             src += GetSwizzle(elem);
         }
 
-        src = is_abs ? "abs(" + src + ')' : src;
-
         shader.AddLine(dest + " = " + src + ';');
     }
 
@@ -538,7 +572,7 @@ private:
             // vertex shader, and what's the value of the fourth element when inside a Tess Eval
             // shader.
             ASSERT(stage == Maxwell3D::Regs::ShaderStage::Vertex);
-            return "vec4(0, 0, gl_InstanceID, gl_VertexID)";
+            return "vec4(0, 0, uintBitsToFloat(gl_InstanceID), uintBitsToFloat(gl_VertexID))";
         default:
             const u32 index{static_cast<u32>(attribute) -
                             static_cast<u32>(Attribute::Index::Attribute_0)};
@@ -733,6 +767,31 @@ private:
         return (absolute_offset % SchedPeriod) == 0;
     }
 
+    void WriteLogicOperation(Register dest, LogicOperation logic_op, const std::string& op_a,
+                             const std::string& op_b) {
+        switch (logic_op) {
+        case LogicOperation::And: {
+            regs.SetRegisterToInteger(dest, true, 0, '(' + op_a + " & " + op_b + ')', 1, 1);
+            break;
+        }
+        case LogicOperation::Or: {
+            regs.SetRegisterToInteger(dest, true, 0, '(' + op_a + " | " + op_b + ')', 1, 1);
+            break;
+        }
+        case LogicOperation::Xor: {
+            regs.SetRegisterToInteger(dest, true, 0, '(' + op_a + " ^ " + op_b + ')', 1, 1);
+            break;
+        }
+        case LogicOperation::PassB: {
+            regs.SetRegisterToInteger(dest, true, 0, op_b, 1, 1);
+            break;
+        }
+        default:
+            NGLOG_CRITICAL(HW_GPU, "Unimplemented logic operation: {}", static_cast<u32>(logic_op));
+            UNREACHABLE();
+        }
+    }
+
     /**
      * Compiles a single instruction from Tegra to GLSL.
      * @param offset the offset of the Tegra shader instruction.
@@ -752,6 +811,7 @@ private:
         if (!opcode) {
             NGLOG_CRITICAL(HW_GPU, "Unhandled instruction: {0:x}", instr.value);
             UNREACHABLE();
+            return offset + 1;
         }
 
         shader.AddLine("// " + std::to_string(offset) + ": " + opcode->GetName());
@@ -770,22 +830,25 @@ private:
 
         switch (opcode->GetType()) {
         case OpCode::Type::Arithmetic: {
-            std::string op_a = instr.alu.negate_a ? "-" : "";
-            op_a += regs.GetRegisterAsFloat(instr.gpr8);
+            std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
             if (instr.alu.abs_a) {
                 op_a = "abs(" + op_a + ')';
             }
 
-            std::string op_b = instr.alu.negate_b ? "-" : "";
+            if (instr.alu.negate_a) {
+                op_a = "-(" + op_a + ')';
+            }
+
+            std::string op_b;
 
             if (instr.is_b_imm) {
-                op_b += GetImmediate19(instr);
+                op_b = GetImmediate19(instr);
             } else {
                 if (instr.is_b_gpr) {
-                    op_b += regs.GetRegisterAsFloat(instr.gpr20);
+                    op_b = regs.GetRegisterAsFloat(instr.gpr20);
                 } else {
-                    op_b += regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
-                                            GLSLRegister::Type::Float);
+                    op_b = regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
+                                           GLSLRegister::Type::Float);
                 }
             }
 
@@ -793,6 +856,10 @@ private:
                 op_b = "abs(" + op_b + ')';
             }
 
+            if (instr.alu.negate_b) {
+                op_b = "-(" + op_b + ')';
+            }
+
             switch (opcode->GetId()) {
             case OpCode::Id::MOV_C:
             case OpCode::Id::MOV_R: {
@@ -800,58 +867,49 @@ private:
                 break;
             }
 
-            case OpCode::Id::MOV32_IMM: {
-                // mov32i doesn't have abs or neg bits.
-                regs.SetRegisterToFloat(instr.gpr0, 0, GetImmediate32(instr), 1, 1);
-                break;
-            }
             case OpCode::Id::FMUL_C:
             case OpCode::Id::FMUL_R:
             case OpCode::Id::FMUL_IMM: {
-                regs.SetRegisterToFloat(instr.gpr0, 0, op_a + " * " + op_b, 1, 1, instr.alu.abs_d);
-                break;
-            }
-            case OpCode::Id::FMUL32_IMM: {
-                // fmul32i doesn't have abs or neg bits.
-                regs.SetRegisterToFloat(
-                    instr.gpr0, 0,
-                    regs.GetRegisterAsFloat(instr.gpr8) + " * " + GetImmediate32(instr), 1, 1);
+                regs.SetRegisterToFloat(instr.gpr0, 0, op_a + " * " + op_b, 1, 1,
+                                        instr.alu.saturate_d);
                 break;
             }
             case OpCode::Id::FADD_C:
             case OpCode::Id::FADD_R:
             case OpCode::Id::FADD_IMM: {
-                regs.SetRegisterToFloat(instr.gpr0, 0, op_a + " + " + op_b, 1, 1, instr.alu.abs_d);
+                regs.SetRegisterToFloat(instr.gpr0, 0, op_a + " + " + op_b, 1, 1,
+                                        instr.alu.saturate_d);
                 break;
             }
             case OpCode::Id::MUFU: {
                 switch (instr.sub_op) {
                 case SubOp::Cos:
                     regs.SetRegisterToFloat(instr.gpr0, 0, "cos(" + op_a + ')', 1, 1,
-                                            instr.alu.abs_d);
+                                            instr.alu.saturate_d);
                     break;
                 case SubOp::Sin:
                     regs.SetRegisterToFloat(instr.gpr0, 0, "sin(" + op_a + ')', 1, 1,
-                                            instr.alu.abs_d);
+                                            instr.alu.saturate_d);
                     break;
                 case SubOp::Ex2:
                     regs.SetRegisterToFloat(instr.gpr0, 0, "exp2(" + op_a + ')', 1, 1,
-                                            instr.alu.abs_d);
+                                            instr.alu.saturate_d);
                     break;
                 case SubOp::Lg2:
                     regs.SetRegisterToFloat(instr.gpr0, 0, "log2(" + op_a + ')', 1, 1,
-                                            instr.alu.abs_d);
+                                            instr.alu.saturate_d);
                     break;
                 case SubOp::Rcp:
-                    regs.SetRegisterToFloat(instr.gpr0, 0, "1.0 / " + op_a, 1, 1, instr.alu.abs_d);
+                    regs.SetRegisterToFloat(instr.gpr0, 0, "1.0 / " + op_a, 1, 1,
+                                            instr.alu.saturate_d);
                     break;
                 case SubOp::Rsq:
                     regs.SetRegisterToFloat(instr.gpr0, 0, "inversesqrt(" + op_a + ')', 1, 1,
-                                            instr.alu.abs_d);
+                                            instr.alu.saturate_d);
                     break;
                 case SubOp::Min:
                     regs.SetRegisterToFloat(instr.gpr0, 0, "min(" + op_a + "," + op_b + ')', 1, 1,
-                                            instr.alu.abs_d);
+                                            instr.alu.saturate_d);
                     break;
                 default:
                     NGLOG_CRITICAL(HW_GPU, "Unhandled MUFU sub op: {0:x}",
@@ -888,47 +946,44 @@ private:
             }
             break;
         }
-        case OpCode::Type::Logic: {
-            std::string op_a = regs.GetRegisterAsInteger(instr.gpr8, 0, false);
+        case OpCode::Type::ArithmeticImmediate: {
+            switch (opcode->GetId()) {
+            case OpCode::Id::MOV32_IMM: {
+                regs.SetRegisterToFloat(instr.gpr0, 0, GetImmediate32(instr), 1, 1);
+                break;
+            }
+            case OpCode::Id::FMUL32_IMM: {
+                regs.SetRegisterToFloat(
+                    instr.gpr0, 0,
+                    regs.GetRegisterAsFloat(instr.gpr8) + " * " + GetImmediate32(instr), 1, 1);
+                break;
+            }
+            }
+            break;
+        }
+        case OpCode::Type::Bfe: {
+            ASSERT_MSG(!instr.bfe.negate_b, "Unimplemented");
 
-            if (instr.alu.lop.invert_a)
-                op_a = "~(" + op_a + ')';
+            std::string op_a = instr.bfe.negate_a ? "-" : "";
+            op_a += regs.GetRegisterAsInteger(instr.gpr8);
 
             switch (opcode->GetId()) {
-            case OpCode::Id::LOP32I: {
-                u32 imm = static_cast<u32>(instr.alu.imm20_32.Value());
+            case OpCode::Id::BFE_IMM: {
+                std::string inner_shift =
+                    '(' + op_a + " << " + std::to_string(instr.bfe.GetLeftShiftValue()) + ')';
+                std::string outer_shift =
+                    '(' + inner_shift + " >> " +
+                    std::to_string(instr.bfe.GetLeftShiftValue() + instr.bfe.shift_position) + ')';
 
-                if (instr.alu.lop.invert_b)
-                    imm = ~imm;
-
-                switch (instr.alu.lop.operation) {
-                case Tegra::Shader::LogicOperation::And: {
-                    regs.SetRegisterToInteger(instr.gpr0, false, 0,
-                                              '(' + op_a + " & " + std::to_string(imm) + ')', 1, 1);
-                    break;
-                }
-                case Tegra::Shader::LogicOperation::Or: {
-                    regs.SetRegisterToInteger(instr.gpr0, false, 0,
-                                              '(' + op_a + " | " + std::to_string(imm) + ')', 1, 1);
-                    break;
-                }
-                case Tegra::Shader::LogicOperation::Xor: {
-                    regs.SetRegisterToInteger(instr.gpr0, false, 0,
-                                              '(' + op_a + " ^ " + std::to_string(imm) + ')', 1, 1);
-                    break;
-                }
-                default:
-                    NGLOG_CRITICAL(HW_GPU, "Unimplemented lop32i operation: {}",
-                                   static_cast<u32>(instr.alu.lop.operation.Value()));
-                    UNREACHABLE();
-                }
+                regs.SetRegisterToInteger(instr.gpr0, true, 0, outer_shift, 1, 1);
                 break;
             }
             default: {
-                NGLOG_CRITICAL(HW_GPU, "Unhandled logic instruction: {}", opcode->GetName());
+                NGLOG_CRITICAL(HW_GPU, "Unhandled BFE instruction: {}", opcode->GetName());
                 UNREACHABLE();
             }
             }
+
             break;
         }
 
@@ -948,6 +1003,19 @@ private:
             }
 
             switch (opcode->GetId()) {
+            case OpCode::Id::SHR_C:
+            case OpCode::Id::SHR_R:
+            case OpCode::Id::SHR_IMM: {
+                if (!instr.shift.is_signed) {
+                    // Logical shift right
+                    op_a = "uint(" + op_a + ')';
+                }
+
+                // Cast to int is superfluous for arithmetic shift, it's only for a logical shift
+                regs.SetRegisterToInteger(instr.gpr0, true, 0, "int(" + op_a + " >> " + op_b + ')',
+                                          1, 1);
+                break;
+            }
             case OpCode::Id::SHL_C:
             case OpCode::Id::SHL_R:
             case OpCode::Id::SHL_IMM:
@@ -961,14 +1029,39 @@ private:
             break;
         }
 
-        case OpCode::Type::ScaledAdd: {
+        case OpCode::Type::ArithmeticIntegerImmediate: {
             std::string op_a = regs.GetRegisterAsInteger(instr.gpr8);
+            std::string op_b = std::to_string(instr.alu.imm20_32.Value());
 
-            if (instr.iscadd.negate_a)
-                op_a = '-' + op_a;
+            switch (opcode->GetId()) {
+            case OpCode::Id::IADD32I:
+                if (instr.iadd32i.negate_a)
+                    op_a = "-(" + op_a + ')';
 
-            std::string op_b = instr.iscadd.negate_b ? "-" : "";
+                regs.SetRegisterToInteger(instr.gpr0, true, 0, op_a + " + " + op_b, 1, 1,
+                                          instr.iadd32i.saturate != 0);
+                break;
+            case OpCode::Id::LOP32I: {
+                if (instr.alu.lop32i.invert_a)
+                    op_a = "~(" + op_a + ')';
 
+                if (instr.alu.lop32i.invert_b)
+                    op_b = "~(" + op_b + ')';
+
+                WriteLogicOperation(instr.gpr0, instr.alu.lop32i.operation, op_a, op_b);
+                break;
+            }
+            default: {
+                NGLOG_CRITICAL(HW_GPU, "Unhandled ArithmeticIntegerImmediate instruction: {}",
+                               opcode->GetName());
+                UNREACHABLE();
+            }
+            }
+            break;
+        }
+        case OpCode::Type::ArithmeticInteger: {
+            std::string op_a = regs.GetRegisterAsInteger(instr.gpr8);
+            std::string op_b;
             if (instr.is_b_imm) {
                 op_b += '(' + std::to_string(instr.alu.GetSignedImm20_20()) + ')';
             } else {
@@ -980,10 +1073,57 @@ private:
                 }
             }
 
-            std::string shift = std::to_string(instr.iscadd.shift_amount.Value());
+            switch (opcode->GetId()) {
+            case OpCode::Id::IADD_C:
+            case OpCode::Id::IADD_R:
+            case OpCode::Id::IADD_IMM: {
+                if (instr.alu_integer.negate_a)
+                    op_a = "-(" + op_a + ')';
+
+                if (instr.alu_integer.negate_b)
+                    op_b = "-(" + op_b + ')';
+
+                regs.SetRegisterToInteger(instr.gpr0, true, 0, op_a + " + " + op_b, 1, 1,
+                                          instr.alu.saturate_d);
+                break;
+            }
+            case OpCode::Id::ISCADD_C:
+            case OpCode::Id::ISCADD_R:
+            case OpCode::Id::ISCADD_IMM: {
+                if (instr.alu_integer.negate_a)
+                    op_a = "-(" + op_a + ')';
+
+                if (instr.alu_integer.negate_b)
+                    op_b = "-(" + op_b + ')';
+
+                std::string shift = std::to_string(instr.alu_integer.shift_amount.Value());
+
+                regs.SetRegisterToInteger(instr.gpr0, true, 0,
+                                          "((" + op_a + " << " + shift + ") + " + op_b + ')', 1, 1);
+                break;
+            }
+            case OpCode::Id::LOP_C:
+            case OpCode::Id::LOP_R:
+            case OpCode::Id::LOP_IMM: {
+                ASSERT_MSG(!instr.alu.lop.unk44, "Unimplemented");
+                ASSERT_MSG(instr.alu.lop.pred48 == Pred::UnusedIndex, "Unimplemented");
+
+                if (instr.alu.lop.invert_a)
+                    op_a = "~(" + op_a + ')';
+
+                if (instr.alu.lop.invert_b)
+                    op_b = "~(" + op_b + ')';
+
+                WriteLogicOperation(instr.gpr0, instr.alu.lop.operation, op_a, op_b);
+                break;
+            }
+            default: {
+                NGLOG_CRITICAL(HW_GPU, "Unhandled ArithmeticInteger instruction: {}",
+                               opcode->GetName());
+                UNREACHABLE();
+            }
+            }
 
-            regs.SetRegisterToInteger(instr.gpr0, true, 0,
-                                      "((" + op_a + " << " + shift + ") + " + op_b + ')', 1, 1);
             break;
         }
         case OpCode::Type::Ffma: {
@@ -1020,33 +1160,33 @@ private:
             }
             }
 
-            regs.SetRegisterToFloat(instr.gpr0, 0, op_a + " * " + op_b + " + " + op_c, 1, 1);
+            regs.SetRegisterToFloat(instr.gpr0, 0, op_a + " * " + op_b + " + " + op_c, 1, 1,
+                                    instr.alu.saturate_d);
             break;
         }
         case OpCode::Type::Conversion: {
-            ASSERT_MSG(instr.conversion.size == Register::Size::Word, "Unimplemented");
             ASSERT_MSG(!instr.conversion.negate_a, "Unimplemented");
-            ASSERT_MSG(!instr.conversion.saturate_a, "Unimplemented");
 
             switch (opcode->GetId()) {
             case OpCode::Id::I2I_R: {
                 ASSERT_MSG(!instr.conversion.selector, "Unimplemented");
 
-                std::string op_a =
-                    regs.GetRegisterAsInteger(instr.gpr20, 0, instr.conversion.is_input_signed);
+                std::string op_a = regs.GetRegisterAsInteger(
+                    instr.gpr20, 0, instr.conversion.is_input_signed, instr.conversion.src_size);
 
                 if (instr.conversion.abs_a) {
                     op_a = "abs(" + op_a + ')';
                 }
 
                 regs.SetRegisterToInteger(instr.gpr0, instr.conversion.is_output_signed, 0, op_a, 1,
-                                          1);
+                                          1, instr.alu.saturate_d, 0, instr.conversion.dest_size);
                 break;
             }
             case OpCode::Id::I2F_R: {
+                ASSERT_MSG(instr.conversion.dest_size == Register::Size::Word, "Unimplemented");
                 ASSERT_MSG(!instr.conversion.selector, "Unimplemented");
-                std::string op_a =
-                    regs.GetRegisterAsInteger(instr.gpr20, 0, instr.conversion.is_input_signed);
+                std::string op_a = regs.GetRegisterAsInteger(
+                    instr.gpr20, 0, instr.conversion.is_input_signed, instr.conversion.src_size);
 
                 if (instr.conversion.abs_a) {
                     op_a = "abs(" + op_a + ')';
@@ -1056,35 +1196,54 @@ private:
                 break;
             }
             case OpCode::Id::F2F_R: {
-                // TODO(Subv): Implement rounding operations.
-                ASSERT_MSG(instr.conversion.f2f.rounding == 0, "Unimplemented rounding operation");
+                ASSERT_MSG(instr.conversion.dest_size == Register::Size::Word, "Unimplemented");
+                ASSERT_MSG(instr.conversion.src_size == Register::Size::Word, "Unimplemented");
                 std::string op_a = regs.GetRegisterAsFloat(instr.gpr20);
 
+                switch (instr.conversion.f2f.rounding) {
+                case Tegra::Shader::F2fRoundingOp::None:
+                    break;
+                case Tegra::Shader::F2fRoundingOp::Floor:
+                    op_a = "floor(" + op_a + ')';
+                    break;
+                case Tegra::Shader::F2fRoundingOp::Ceil:
+                    op_a = "ceil(" + op_a + ')';
+                    break;
+                case Tegra::Shader::F2fRoundingOp::Trunc:
+                    op_a = "trunc(" + op_a + ')';
+                    break;
+                default:
+                    NGLOG_CRITICAL(HW_GPU, "Unimplemented f2f rounding mode {}",
+                                   static_cast<u32>(instr.conversion.f2f.rounding.Value()));
+                    UNREACHABLE();
+                    break;
+                }
+
                 if (instr.conversion.abs_a) {
                     op_a = "abs(" + op_a + ')';
                 }
 
-                regs.SetRegisterToFloat(instr.gpr0, 0, op_a, 1, 1);
+                regs.SetRegisterToFloat(instr.gpr0, 0, op_a, 1, 1, instr.alu.saturate_d);
                 break;
             }
             case OpCode::Id::F2I_R: {
+                ASSERT_MSG(instr.conversion.src_size == Register::Size::Word, "Unimplemented");
                 std::string op_a = regs.GetRegisterAsFloat(instr.gpr20);
 
                 if (instr.conversion.abs_a) {
                     op_a = "abs(" + op_a + ')';
                 }
 
-                using Tegra::Shader::FloatRoundingOp;
                 switch (instr.conversion.f2i.rounding) {
-                case FloatRoundingOp::None:
+                case Tegra::Shader::F2iRoundingOp::None:
                     break;
-                case FloatRoundingOp::Floor:
+                case Tegra::Shader::F2iRoundingOp::Floor:
                     op_a = "floor(" + op_a + ')';
                     break;
-                case FloatRoundingOp::Ceil:
+                case Tegra::Shader::F2iRoundingOp::Ceil:
                     op_a = "ceil(" + op_a + ')';
                     break;
-                case FloatRoundingOp::Trunc:
+                case Tegra::Shader::F2iRoundingOp::Trunc:
                     op_a = "trunc(" + op_a + ')';
                     break;
                 default:
@@ -1101,7 +1260,7 @@ private:
                 }
 
                 regs.SetRegisterToInteger(instr.gpr0, instr.conversion.is_output_signed, 0, op_a, 1,
-                                          1);
+                                          1, false, 0, instr.conversion.dest_size);
                 break;
             }
             default: {
@@ -1112,13 +1271,11 @@ private:
             break;
         }
         case OpCode::Type::Memory: {
-            const Attribute::Index attribute = instr.attribute.fmt20.index;
-
             switch (opcode->GetId()) {
             case OpCode::Id::LD_A: {
                 ASSERT_MSG(instr.attribute.fmt20.size == 0, "untested");
                 regs.SetRegisterToInputAttibute(instr.gpr0, instr.attribute.fmt20.element,
-                                                attribute);
+                                                instr.attribute.fmt20.index);
                 break;
             }
             case OpCode::Id::LD_C: {
@@ -1150,25 +1307,24 @@ private:
             }
             case OpCode::Id::ST_A: {
                 ASSERT_MSG(instr.attribute.fmt20.size == 0, "untested");
-                regs.SetOutputAttributeToRegister(attribute, instr.attribute.fmt20.element,
-                                                  instr.gpr0);
+                regs.SetOutputAttributeToRegister(instr.attribute.fmt20.index,
+                                                  instr.attribute.fmt20.element, instr.gpr0);
                 break;
             }
             case OpCode::Id::TEX: {
-                ASSERT_MSG(instr.attribute.fmt20.size == 4, "untested");
                 const std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
                 const std::string op_b = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
                 const std::string sampler = GetSampler(instr.sampler);
                 const std::string coord = "vec2 coords = vec2(" + op_a + ", " + op_b + ");";
-                // Add an extra scope and declare the texture coords inside to prevent overwriting
-                // them in case they are used as outputs of the texs instruction.
+                // Add an extra scope and declare the texture coords inside to prevent
+                // overwriting them in case they are used as outputs of the texs instruction.
                 shader.AddLine("{");
                 ++shader.scope;
                 shader.AddLine(coord);
                 const std::string texture = "texture(" + sampler + ", coords)";
 
                 size_t dest_elem{};
-                for (size_t elem = 0; elem < instr.attribute.fmt20.size; ++elem) {
+                for (size_t elem = 0; elem < 4; ++elem) {
                     if (!instr.tex.IsComponentEnabled(elem)) {
                         // Skip disabled components
                         continue;
@@ -1181,7 +1337,6 @@ private:
                 break;
             }
             case OpCode::Id::TEXS: {
-                ASSERT_MSG(instr.attribute.fmt20.size == 4, "untested");
                 const std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
                 const std::string op_b = regs.GetRegisterAsFloat(instr.gpr20);
                 const std::string sampler = GetSampler(instr.sampler);
@@ -1193,8 +1348,8 @@ private:
                 shader.AddLine(coord);
                 const std::string texture = "texture(" + sampler + ", coords)";
 
-                // TEXS has two destination registers. RG goes into gpr0+0 and gpr0+1, and BA goes
-                // into gpr28+0 and gpr28+1
+                // TEXS has two destination registers. RG goes into gpr0+0 and gpr0+1, and BA
+                // goes into gpr28+0 and gpr28+1
                 size_t offset{};
 
                 for (const auto& dest : {instr.gpr0.Value(), instr.gpr28.Value()}) {
@@ -1278,16 +1433,17 @@ private:
         }
         case OpCode::Type::IntegerSetPredicate: {
             std::string op_a = regs.GetRegisterAsInteger(instr.gpr8, 0, instr.isetp.is_signed);
+            std::string op_b;
 
-            std::string op_b{};
-
-            ASSERT_MSG(!instr.is_b_imm, "ISETP_IMM not implemented");
-
-            if (instr.is_b_gpr) {
-                op_b += regs.GetRegisterAsInteger(instr.gpr20, 0, instr.isetp.is_signed);
+            if (instr.is_b_imm) {
+                op_b += '(' + std::to_string(instr.alu.GetSignedImm20_20()) + ')';
             } else {
-                op_b += regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
-                                        GLSLRegister::Type::Integer);
+                if (instr.is_b_gpr) {
+                    op_b += regs.GetRegisterAsInteger(instr.gpr20, 0, instr.isetp.is_signed);
+                } else {
+                    op_b += regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
+                                            GLSLRegister::Type::Integer);
+                }
             }
 
             using Tegra::Shader::Pred;
@@ -1342,8 +1498,8 @@ private:
                 op_b = "abs(" + op_b + ')';
             }
 
-            // The fset instruction sets a register to 1.0 if the condition is true, and to 0
-            // otherwise.
+            // The fset instruction sets a register to 1.0 or -1 (depending on the bf bit) if the
+            // condition is true, and to 0 otherwise.
             std::string second_pred =
                 GetPredicateCondition(instr.fset.pred39, instr.fset.neg_pred != 0);
 
@@ -1361,6 +1517,41 @@ private:
             }
             break;
         }
+        case OpCode::Type::IntegerSet: {
+            std::string op_a = regs.GetRegisterAsInteger(instr.gpr8, 0, instr.iset.is_signed);
+
+            std::string op_b;
+
+            if (instr.is_b_imm) {
+                op_b = std::to_string(instr.alu.GetSignedImm20_20());
+            } else {
+                if (instr.is_b_gpr) {
+                    op_b = regs.GetRegisterAsInteger(instr.gpr20, 0, instr.iset.is_signed);
+                } else {
+                    op_b = regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
+                                           GLSLRegister::Type::Integer);
+                }
+            }
+
+            // The iset instruction sets a register to 1.0 or -1 (depending on the bf bit) if the
+            // condition is true, and to 0 otherwise.
+            std::string second_pred =
+                GetPredicateCondition(instr.iset.pred39, instr.iset.neg_pred != 0);
+
+            std::string comparator = GetPredicateComparison(instr.iset.cond);
+            std::string combiner = GetPredicateCombiner(instr.iset.op);
+
+            std::string predicate = "(((" + op_a + ") " + comparator + " (" + op_b + ")) " +
+                                    combiner + " (" + second_pred + "))";
+
+            if (instr.iset.bf) {
+                regs.SetRegisterToFloat(instr.gpr0, 0, predicate + " ? 1.0 : 0.0", 1, 1);
+            } else {
+                regs.SetRegisterToInteger(instr.gpr0, false, 0, predicate + " ? 0xFFFFFFFF : 0", 1,
+                                          1);
+            }
+            break;
+        }
         default: {
             switch (opcode->GetId()) {
             case OpCode::Id::EXIT: {
@@ -1374,8 +1565,8 @@ private:
 
                 shader.AddLine("return true;");
                 if (instr.pred.pred_index == static_cast<u64>(Pred::UnusedIndex)) {
-                    // If this is an unconditional exit then just end processing here, otherwise we
-                    // have to account for the possibility of the condition not being met, so
+                    // If this is an unconditional exit then just end processing here, otherwise
+                    // we have to account for the possibility of the condition not being met, so
                     // continue processing the next instruction.
                     offset = PROGRAM_END - 1;
                 }
@@ -1397,6 +1588,11 @@ private:
                 regs.SetRegisterToInputAttibute(instr.gpr0, attribute.element, attribute.index);
                 break;
             }
+            case OpCode::Id::SSY: {
+                // The SSY opcode tells the GPU where to re-converge divergent execution paths, we
+                // can ignore this when generating GLSL code.
+                break;
+            }
             default: {
                 NGLOG_CRITICAL(HW_GPU, "Unhandled instruction: {}", opcode->GetName());
                 UNREACHABLE();
@@ -1525,7 +1721,10 @@ private:
 }; // namespace Decompiler
 
 std::string GetCommonDeclarations() {
-    return "bool exec_shader();";
+    std::string declarations = "bool exec_shader();\n";
+    declarations += "#define MAX_CONSTBUFFER_ELEMENTS " +
+                    std::to_string(RasterizerOpenGL::MaxConstbufferSize / (sizeof(GLvec4)));
+    return declarations;
 }
 
 boost::optional<ProgramResult> DecompileProgram(const ProgramCode& program_code, u32 main_offset,

src/video_core/renderer_opengl/gl_shader_gen.cpp

@@ -39,6 +39,10 @@ void main() {
     // Viewport can be flipped, which is unsupported by glViewport
     position.xy *= viewport_flip.xy;
     gl_Position = position;
+
+    // TODO(bunnei): This is likely a hack, position.w should be interpolated as 1.0
+    // For now, this is here to bring order in lieu of proper emulation
+    position.w = 1.0;
 }
 )";
     out += program.first;

src/video_core/renderer_opengl/gl_shader_manager.cpp

@@ -38,8 +38,8 @@ void MaxwellUniformData::SetFromRegs(const Maxwell3D::State::ShaderStageInfo& sh
     const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
 
     // TODO(bunnei): Support more than one viewport
-    viewport_flip[0] = regs.viewport_transform[0].scale_x < 0.0 ? -1.0 : 1.0;
-    viewport_flip[1] = regs.viewport_transform[0].scale_y < 0.0 ? -1.0 : 1.0;
+    viewport_flip[0] = regs.viewport_transform[0].scale_x < 0.0 ? -1.0f : 1.0f;
+    viewport_flip[1] = regs.viewport_transform[0].scale_y < 0.0 ? -1.0f : 1.0f;
 }
 
 } // namespace GLShader

src/video_core/renderer_opengl/gl_state.cpp

@@ -48,24 +48,9 @@ OpenGLState::OpenGLState() {
     logic_op = GL_COPY;
 
     for (auto& texture_unit : texture_units) {
-        texture_unit.texture_2d = 0;
-        texture_unit.sampler = 0;
-        texture_unit.swizzle.r = GL_RED;
-        texture_unit.swizzle.g = GL_GREEN;
-        texture_unit.swizzle.b = GL_BLUE;
-        texture_unit.swizzle.a = GL_ALPHA;
+        texture_unit.Reset();
     }
 
-    lighting_lut.texture_buffer = 0;
-
-    fog_lut.texture_buffer = 0;
-
-    proctex_lut.texture_buffer = 0;
-    proctex_diff_lut.texture_buffer = 0;
-    proctex_color_map.texture_buffer = 0;
-    proctex_alpha_map.texture_buffer = 0;
-    proctex_noise_lut.texture_buffer = 0;
-
     draw.read_framebuffer = 0;
     draw.draw_framebuffer = 0;
     draw.vertex_array = 0;
@@ -196,13 +181,13 @@ void OpenGLState::Apply() const {
     }
 
     // Textures
-    for (size_t i = 0; i < std::size(texture_units); ++i) {
+    for (int i = 0; i < std::size(texture_units); ++i) {
         if (texture_units[i].texture_2d != cur_state.texture_units[i].texture_2d) {
             glActiveTexture(TextureUnits::MaxwellTexture(i).Enum());
             glBindTexture(GL_TEXTURE_2D, texture_units[i].texture_2d);
         }
         if (texture_units[i].sampler != cur_state.texture_units[i].sampler) {
-            glBindSampler(i, texture_units[i].sampler);
+            glBindSampler(static_cast<GLuint>(i), texture_units[i].sampler);
         }
         // Update the texture swizzle
         if (texture_units[i].swizzle.r != cur_state.texture_units[i].swizzle.r ||
@@ -223,54 +208,12 @@ void OpenGLState::Apply() const {
             if (current.enabled != new_state.enabled || current.bindpoint != new_state.bindpoint ||
                 current.ssbo != new_state.ssbo) {
                 if (new_state.enabled) {
-                    glBindBufferBase(GL_SHADER_STORAGE_BUFFER, new_state.bindpoint, new_state.ssbo);
+                    glBindBufferBase(GL_UNIFORM_BUFFER, new_state.bindpoint, new_state.ssbo);
                 }
             }
         }
     }
 
-    // Lighting LUTs
-    if (lighting_lut.texture_buffer != cur_state.lighting_lut.texture_buffer) {
-        glActiveTexture(TextureUnits::LightingLUT.Enum());
-        glBindTexture(GL_TEXTURE_BUFFER, lighting_lut.texture_buffer);
-    }
-
-    // Fog LUT
-    if (fog_lut.texture_buffer != cur_state.fog_lut.texture_buffer) {
-        glActiveTexture(TextureUnits::FogLUT.Enum());
-        glBindTexture(GL_TEXTURE_BUFFER, fog_lut.texture_buffer);
-    }
-
-    // ProcTex Noise LUT
-    if (proctex_noise_lut.texture_buffer != cur_state.proctex_noise_lut.texture_buffer) {
-        glActiveTexture(TextureUnits::ProcTexNoiseLUT.Enum());
-        glBindTexture(GL_TEXTURE_BUFFER, proctex_noise_lut.texture_buffer);
-    }
-
-    // ProcTex Color Map
-    if (proctex_color_map.texture_buffer != cur_state.proctex_color_map.texture_buffer) {
-        glActiveTexture(TextureUnits::ProcTexColorMap.Enum());
-        glBindTexture(GL_TEXTURE_BUFFER, proctex_color_map.texture_buffer);
-    }
-
-    // ProcTex Alpha Map
-    if (proctex_alpha_map.texture_buffer != cur_state.proctex_alpha_map.texture_buffer) {
-        glActiveTexture(TextureUnits::ProcTexAlphaMap.Enum());
-        glBindTexture(GL_TEXTURE_BUFFER, proctex_alpha_map.texture_buffer);
-    }
-
-    // ProcTex LUT
-    if (proctex_lut.texture_buffer != cur_state.proctex_lut.texture_buffer) {
-        glActiveTexture(TextureUnits::ProcTexLUT.Enum());
-        glBindTexture(GL_TEXTURE_BUFFER, proctex_lut.texture_buffer);
-    }
-
-    // ProcTex Diff LUT
-    if (proctex_diff_lut.texture_buffer != cur_state.proctex_diff_lut.texture_buffer) {
-        glActiveTexture(TextureUnits::ProcTexDiffLUT.Enum());
-        glBindTexture(GL_TEXTURE_BUFFER, proctex_diff_lut.texture_buffer);
-    }
-
     // Framebuffer
     if (draw.read_framebuffer != cur_state.draw.read_framebuffer) {
         glBindFramebuffer(GL_READ_FRAMEBUFFER, draw.read_framebuffer);
@@ -338,26 +281,12 @@ void OpenGLState::Apply() const {
     cur_state = *this;
 }
 
-OpenGLState& OpenGLState::ResetTexture(GLuint handle) {
+OpenGLState& OpenGLState::UnbindTexture(GLuint handle) {
     for (auto& unit : texture_units) {
         if (unit.texture_2d == handle) {
-            unit.texture_2d = 0;
+            unit.Unbind();
         }
     }
-    if (lighting_lut.texture_buffer == handle)
-        lighting_lut.texture_buffer = 0;
-    if (fog_lut.texture_buffer == handle)
-        fog_lut.texture_buffer = 0;
-    if (proctex_noise_lut.texture_buffer == handle)
-        proctex_noise_lut.texture_buffer = 0;
-    if (proctex_color_map.texture_buffer == handle)
-        proctex_color_map.texture_buffer = 0;
-    if (proctex_alpha_map.texture_buffer == handle)
-        proctex_alpha_map.texture_buffer = 0;
-    if (proctex_lut.texture_buffer == handle)
-        proctex_lut.texture_buffer = 0;
-    if (proctex_diff_lut.texture_buffer == handle)
-        proctex_diff_lut.texture_buffer = 0;
     return *this;
 }
 

src/video_core/renderer_opengl/gl_state.h

@@ -91,36 +91,21 @@ public:
             GLint b; // GL_TEXTURE_SWIZZLE_B
             GLint a; // GL_TEXTURE_SWIZZLE_A
         } swizzle;
+
+        void Unbind() {
+            texture_2d = 0;
+            swizzle.r = GL_RED;
+            swizzle.g = GL_GREEN;
+            swizzle.b = GL_BLUE;
+            swizzle.a = GL_ALPHA;
+        }
+
+        void Reset() {
+            Unbind();
+            sampler = 0;
+        }
     } texture_units[32];
 
-    struct {
-        GLuint texture_buffer; // GL_TEXTURE_BINDING_BUFFER
-    } lighting_lut;
-
-    struct {
-        GLuint texture_buffer; // GL_TEXTURE_BINDING_BUFFER
-    } fog_lut;
-
-    struct {
-        GLuint texture_buffer; // GL_TEXTURE_BINDING_BUFFER
-    } proctex_noise_lut;
-
-    struct {
-        GLuint texture_buffer; // GL_TEXTURE_BINDING_BUFFER
-    } proctex_color_map;
-
-    struct {
-        GLuint texture_buffer; // GL_TEXTURE_BINDING_BUFFER
-    } proctex_alpha_map;
-
-    struct {
-        GLuint texture_buffer; // GL_TEXTURE_BINDING_BUFFER
-    } proctex_lut;
-
-    struct {
-        GLuint texture_buffer; // GL_TEXTURE_BINDING_BUFFER
-    } proctex_diff_lut;
-
     struct {
         GLuint read_framebuffer; // GL_READ_FRAMEBUFFER_BINDING
         GLuint draw_framebuffer; // GL_DRAW_FRAMEBUFFER_BINDING
@@ -165,7 +150,7 @@ public:
     void Apply() const;
 
     /// Resets any references to the given resource
-    OpenGLState& ResetTexture(GLuint handle);
+    OpenGLState& UnbindTexture(GLuint handle);
     OpenGLState& ResetSampler(GLuint handle);
     OpenGLState& ResetProgram(GLuint handle);
     OpenGLState& ResetPipeline(GLuint handle);

src/video_core/textures/astc.h

@@ -0,0 +1,15 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <cstdint>
+#include <vector>
+
+namespace Tegra::Texture::ASTC {
+
+std::vector<uint8_t> Decompress(std::vector<uint8_t>& data, uint32_t width, uint32_t height,
+                                uint32_t block_width, uint32_t block_height);
+
+} // namespace Tegra::Texture::ASTC

src/video_core/textures/decoders.cpp

@@ -53,6 +53,7 @@ u32 BytesPerPixel(TextureFormat format) {
     case TextureFormat::DXT45:
         // In this case a 'pixel' actually refers to a 4x4 tile.
         return 16;
+    case TextureFormat::ASTC_2D_4X4:
     case TextureFormat::A8R8G8B8:
     case TextureFormat::A2B10G10R10:
     case TextureFormat::BF10GF11RF11:
@@ -94,6 +95,7 @@ std::vector<u8> UnswizzleTexture(VAddr address, TextureFormat format, u32 width,
     case TextureFormat::R8:
     case TextureFormat::R16_G16_B16_A16:
     case TextureFormat::BF10GF11RF11:
+    case TextureFormat::ASTC_2D_4X4:
         CopySwizzledData(width, height, bytes_per_pixel, bytes_per_pixel, data,
                          unswizzled_data.data(), true, block_height);
         break;
@@ -115,6 +117,7 @@ std::vector<u8> DecodeTexture(const std::vector<u8>& texture_data, TextureFormat
     case TextureFormat::DXT23:
     case TextureFormat::DXT45:
     case TextureFormat::DXN1:
+    case TextureFormat::ASTC_2D_4X4:
     case TextureFormat::A8R8G8B8:
     case TextureFormat::A2B10G10R10:
     case TextureFormat::A1B5G5R5:

src/yuzu/CMakeLists.txt

@@ -32,8 +32,6 @@ add_executable(yuzu
     debugger/graphics/graphics_surface.h
     debugger/profiler.cpp
     debugger/profiler.h
-    debugger/registers.cpp
-    debugger/registers.h
     debugger/wait_tree.cpp
     debugger/wait_tree.h
     game_list.cpp
@@ -60,7 +58,6 @@ set(UIS
     configuration/configure_graphics.ui
     configuration/configure_input.ui
     configuration/configure_system.ui
-    debugger/registers.ui
     hotkeys.ui
     main.ui
 )

src/yuzu/debugger/registers.cpp

@@ -1,190 +0,0 @@
-// Copyright 2014 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include <QTreeWidgetItem>
-#include "core/arm/arm_interface.h"
-#include "core/core.h"
-#include "yuzu/debugger/registers.h"
-#include "yuzu/util/util.h"
-
-RegistersWidget::RegistersWidget(QWidget* parent) : QDockWidget(parent) {
-    cpu_regs_ui.setupUi(this);
-
-    tree = cpu_regs_ui.treeWidget;
-    tree->addTopLevelItem(core_registers = new QTreeWidgetItem(QStringList(tr("Registers"))));
-    tree->addTopLevelItem(vfp_registers = new QTreeWidgetItem(QStringList(tr("VFP Registers"))));
-    tree->addTopLevelItem(vfp_system_registers =
-                              new QTreeWidgetItem(QStringList(tr("VFP System Registers"))));
-    tree->addTopLevelItem(cpsr = new QTreeWidgetItem(QStringList("CPSR")));
-
-    for (int i = 0; i < 16; ++i) {
-        QTreeWidgetItem* child = new QTreeWidgetItem(QStringList(QString("R[%1]").arg(i)));
-        core_registers->addChild(child);
-    }
-
-    for (int i = 0; i < 32; ++i) {
-        QTreeWidgetItem* child = new QTreeWidgetItem(QStringList(QString("S[%1]").arg(i)));
-        vfp_registers->addChild(child);
-    }
-
-    QFont font = GetMonospaceFont();
-
-    CreateCPSRChildren();
-    CreateVFPSystemRegisterChildren();
-
-    // Set Registers to display in monospace font
-    for (int i = 0; i < core_registers->childCount(); ++i)
-        core_registers->child(i)->setFont(1, font);
-
-    for (int i = 0; i < vfp_registers->childCount(); ++i)
-        vfp_registers->child(i)->setFont(1, font);
-
-    for (int i = 0; i < vfp_system_registers->childCount(); ++i) {
-        vfp_system_registers->child(i)->setFont(1, font);
-        for (int x = 0; x < vfp_system_registers->child(i)->childCount(); ++x) {
-            vfp_system_registers->child(i)->child(x)->setFont(1, font);
-        }
-    }
-    // Set CSPR to display in monospace font
-    cpsr->setFont(1, font);
-    for (int i = 0; i < cpsr->childCount(); ++i) {
-        cpsr->child(i)->setFont(1, font);
-        for (int x = 0; x < cpsr->child(i)->childCount(); ++x) {
-            cpsr->child(i)->child(x)->setFont(1, font);
-        }
-    }
-    setEnabled(false);
-}
-
-void RegistersWidget::OnDebugModeEntered() {
-    if (!Core::System::GetInstance().IsPoweredOn())
-        return;
-
-    for (int i = 0; i < core_registers->childCount(); ++i)
-        core_registers->child(i)->setText(
-            1, QString("0x%1").arg(Core::CurrentArmInterface().GetReg(i), 8, 16, QLatin1Char('0')));
-
-    UpdateCPSRValues();
-}
-
-void RegistersWidget::OnDebugModeLeft() {}
-
-void RegistersWidget::OnEmulationStarting(EmuThread* emu_thread) {
-    setEnabled(true);
-}
-
-void RegistersWidget::OnEmulationStopping() {
-    // Reset widget text
-    for (int i = 0; i < core_registers->childCount(); ++i)
-        core_registers->child(i)->setText(1, QString(""));
-
-    for (int i = 0; i < vfp_registers->childCount(); ++i)
-        vfp_registers->child(i)->setText(1, QString(""));
-
-    for (int i = 0; i < cpsr->childCount(); ++i)
-        cpsr->child(i)->setText(1, QString(""));
-
-    cpsr->setText(1, QString(""));
-
-    // FPSCR
-    for (int i = 0; i < vfp_system_registers->child(0)->childCount(); ++i)
-        vfp_system_registers->child(0)->child(i)->setText(1, QString(""));
-
-    // FPEXC
-    for (int i = 0; i < vfp_system_registers->child(1)->childCount(); ++i)
-        vfp_system_registers->child(1)->child(i)->setText(1, QString(""));
-
-    vfp_system_registers->child(0)->setText(1, QString(""));
-    vfp_system_registers->child(1)->setText(1, QString(""));
-    vfp_system_registers->child(2)->setText(1, QString(""));
-    vfp_system_registers->child(3)->setText(1, QString(""));
-
-    setEnabled(false);
-}
-
-void RegistersWidget::CreateCPSRChildren() {
-    cpsr->addChild(new QTreeWidgetItem(QStringList("M")));
-    cpsr->addChild(new QTreeWidgetItem(QStringList("T")));
-    cpsr->addChild(new QTreeWidgetItem(QStringList("F")));
-    cpsr->addChild(new QTreeWidgetItem(QStringList("I")));
-    cpsr->addChild(new QTreeWidgetItem(QStringList("A")));
-    cpsr->addChild(new QTreeWidgetItem(QStringList("E")));
-    cpsr->addChild(new QTreeWidgetItem(QStringList("IT")));
-    cpsr->addChild(new QTreeWidgetItem(QStringList("GE")));
-    cpsr->addChild(new QTreeWidgetItem(QStringList("DNM")));
-    cpsr->addChild(new QTreeWidgetItem(QStringList("J")));
-    cpsr->addChild(new QTreeWidgetItem(QStringList("Q")));
-    cpsr->addChild(new QTreeWidgetItem(QStringList("V")));
-    cpsr->addChild(new QTreeWidgetItem(QStringList("C")));
-    cpsr->addChild(new QTreeWidgetItem(QStringList("Z")));
-    cpsr->addChild(new QTreeWidgetItem(QStringList("N")));
-}
-
-void RegistersWidget::UpdateCPSRValues() {
-    const u32 cpsr_val = Core::CurrentArmInterface().GetCPSR();
-
-    cpsr->setText(1, QString("0x%1").arg(cpsr_val, 8, 16, QLatin1Char('0')));
-    cpsr->child(0)->setText(
-        1, QString("b%1").arg(cpsr_val & 0x1F, 5, 2, QLatin1Char('0'))); // M - Mode
-    cpsr->child(1)->setText(1, QString::number((cpsr_val >> 5) & 1));    // T - State
-    cpsr->child(2)->setText(1, QString::number((cpsr_val >> 6) & 1));    // F - FIQ disable
-    cpsr->child(3)->setText(1, QString::number((cpsr_val >> 7) & 1));    // I - IRQ disable
-    cpsr->child(4)->setText(1, QString::number((cpsr_val >> 8) & 1));    // A - Imprecise abort
-    cpsr->child(5)->setText(1, QString::number((cpsr_val >> 9) & 1));    // E - Data endianness
-    cpsr->child(6)->setText(1,
-                            QString::number((cpsr_val >> 10) & 0x3F)); // IT - If-Then state (DNM)
-    cpsr->child(7)->setText(1,
-                            QString::number((cpsr_val >> 16) & 0xF)); // GE - Greater-than-or-Equal
-    cpsr->child(8)->setText(1, QString::number((cpsr_val >> 20) & 0xF)); // DNM - Do not modify
-    cpsr->child(9)->setText(1, QString::number((cpsr_val >> 24) & 1));   // J - Jazelle
-    cpsr->child(10)->setText(1, QString::number((cpsr_val >> 27) & 1));  // Q - Saturation
-    cpsr->child(11)->setText(1, QString::number((cpsr_val >> 28) & 1));  // V - Overflow
-    cpsr->child(12)->setText(1, QString::number((cpsr_val >> 29) & 1));  // C - Carry/Borrow/Extend
-    cpsr->child(13)->setText(1, QString::number((cpsr_val >> 30) & 1));  // Z - Zero
-    cpsr->child(14)->setText(1, QString::number((cpsr_val >> 31) & 1));  // N - Negative/Less than
-}
-
-void RegistersWidget::CreateVFPSystemRegisterChildren() {
-    QTreeWidgetItem* const fpscr = new QTreeWidgetItem(QStringList("FPSCR"));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("IOC")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("DZC")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("OFC")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("UFC")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("IXC")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("IDC")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("IOE")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("DZE")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("OFE")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("UFE")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("IXE")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("IDE")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList(tr("Vector Length"))));
-    fpscr->addChild(new QTreeWidgetItem(QStringList(tr("Vector Stride"))));
-    fpscr->addChild(new QTreeWidgetItem(QStringList(tr("Rounding Mode"))));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("FZ")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("DN")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("V")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("C")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("Z")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("N")));
-
-    QTreeWidgetItem* const fpexc = new QTreeWidgetItem(QStringList("FPEXC"));
-    fpexc->addChild(new QTreeWidgetItem(QStringList("IOC")));
-    fpexc->addChild(new QTreeWidgetItem(QStringList("OFC")));
-    fpexc->addChild(new QTreeWidgetItem(QStringList("UFC")));
-    fpexc->addChild(new QTreeWidgetItem(QStringList("INV")));
-    fpexc->addChild(new QTreeWidgetItem(QStringList(tr("Vector Iteration Count"))));
-    fpexc->addChild(new QTreeWidgetItem(QStringList("FP2V")));
-    fpexc->addChild(new QTreeWidgetItem(QStringList("EN")));
-    fpexc->addChild(new QTreeWidgetItem(QStringList("EX")));
-
-    vfp_system_registers->addChild(fpscr);
-    vfp_system_registers->addChild(fpexc);
-    vfp_system_registers->addChild(new QTreeWidgetItem(QStringList("FPINST")));
-    vfp_system_registers->addChild(new QTreeWidgetItem(QStringList("FPINST2")));
-}
-
-void RegistersWidget::UpdateVFPSystemRegisterValues() {
-    UNIMPLEMENTED();
-}

src/yuzu/debugger/registers.h

@@ -1,42 +0,0 @@
-// Copyright 2014 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <QDockWidget>
-#include "ui_registers.h"
-
-class QTreeWidget;
-class QTreeWidgetItem;
-class EmuThread;
-
-class RegistersWidget : public QDockWidget {
-    Q_OBJECT
-
-public:
-    explicit RegistersWidget(QWidget* parent = nullptr);
-
-public slots:
-    void OnDebugModeEntered();
-    void OnDebugModeLeft();
-
-    void OnEmulationStarting(EmuThread* emu_thread);
-    void OnEmulationStopping();
-
-private:
-    void CreateCPSRChildren();
-    void UpdateCPSRValues();
-
-    void CreateVFPSystemRegisterChildren();
-    void UpdateVFPSystemRegisterValues();
-
-    Ui::ARMRegisters cpu_regs_ui;
-
-    QTreeWidget* tree;
-
-    QTreeWidgetItem* core_registers;
-    QTreeWidgetItem* vfp_registers;
-    QTreeWidgetItem* vfp_system_registers;
-    QTreeWidgetItem* cpsr;
-};

src/yuzu/debugger/registers.ui

@@ -1,40 +0,0 @@
-<?xml version="1.0" encoding="UTF-8"?>
-<ui version="4.0">
- <class>ARMRegisters</class>
- <widget class="QDockWidget" name="ARMRegisters">
-  <property name="geometry">
-   <rect>
-    <x>0</x>
-    <y>0</y>
-    <width>400</width>
-    <height>300</height>
-   </rect>
-  </property>
-  <property name="windowTitle">
-   <string>ARM Registers</string>
-  </property>
-  <widget class="QWidget" name="dockWidgetContents">
-   <layout class="QVBoxLayout" name="verticalLayout">
-    <item>
-     <widget class="QTreeWidget" name="treeWidget">
-      <property name="alternatingRowColors">
-       <bool>true</bool>
-      </property>
-      <column>
-       <property name="text">
-        <string>Register</string>
-       </property>
-      </column>
-      <column>
-       <property name="text">
-        <string>Value</string>
-       </property>
-      </column>
-     </widget>
-    </item>
-   </layout>
-  </widget>
- </widget>
- <resources/>
- <connections/>
-</ui>

src/yuzu/debugger/wait_tree.cpp

@@ -213,6 +213,9 @@ QString WaitTreeThread::GetText() const {
     case THREADSTATUS_WAIT_MUTEX:
         status = tr("waiting for mutex");
         break;
+    case THREADSTATUS_WAIT_ARB:
+        status = tr("waiting for address arbiter");
+        break;
     case THREADSTATUS_DORMANT:
         status = tr("dormant");
         break;
@@ -240,6 +243,7 @@ QColor WaitTreeThread::GetColor() const {
     case THREADSTATUS_WAIT_SYNCH_ALL:
     case THREADSTATUS_WAIT_SYNCH_ANY:
     case THREADSTATUS_WAIT_MUTEX:
+    case THREADSTATUS_WAIT_ARB:
         return QColor(Qt::GlobalColor::red);
     case THREADSTATUS_DORMANT:
         return QColor(Qt::GlobalColor::darkCyan);

src/yuzu/game_list.cpp

@@ -3,6 +3,7 @@
 // Refer to the license.txt file included.
 
 #include <QApplication>
+#include <QDir>
 #include <QFileInfo>
 #include <QHeaderView>
 #include <QKeyEvent>
@@ -264,8 +265,17 @@ void GameList::ValidateEntry(const QModelIndex& item) {
     if (file_path.isEmpty())
         return;
     std::string std_file_path(file_path.toStdString());
-    if (!FileUtil::Exists(std_file_path) || FileUtil::IsDirectory(std_file_path))
+    if (!FileUtil::Exists(std_file_path))
         return;
+    if (FileUtil::IsDirectory(std_file_path)) {
+        QDir dir(std_file_path.c_str());
+        QStringList matching_main = dir.entryList(QStringList("main"), QDir::Files);
+        if (matching_main.size() == 1) {
+            emit GameChosen(dir.path() + DIR_SEP + matching_main[0]);
+        }
+        return;
+    }
+
     // Users usually want to run a diffrent game after closing one
     search_field->clear();
     emit GameChosen(file_path);
@@ -356,13 +366,26 @@ void GameList::LoadInterfaceLayout() {
     item_model->sort(header->sortIndicatorSection(), header->sortIndicatorOrder());
 }
 
-const QStringList GameList::supported_file_extensions = {"nso", "nro"};
+const QStringList GameList::supported_file_extensions = {"nso", "nro", "nca"};
 
 static bool HasSupportedFileExtension(const std::string& file_name) {
     QFileInfo file = QFileInfo(file_name.c_str());
     return GameList::supported_file_extensions.contains(file.suffix(), Qt::CaseInsensitive);
 }
 
+static bool IsExtractedNCAMain(const std::string& file_name) {
+    return QFileInfo(file_name.c_str()).fileName() == "main";
+}
+
+static QString FormatGameName(const std::string& physical_name) {
+    QFileInfo file_info(physical_name.c_str());
+    if (IsExtractedNCAMain(physical_name)) {
+        return file_info.dir().path();
+    } else {
+        return QString::fromStdString(physical_name);
+    }
+}
+
 void GameList::RefreshGameDirectory() {
     if (!UISettings::values.gamedir.isEmpty() && current_worker != nullptr) {
         NGLOG_INFO(Frontend, "Change detected in the games directory. Reloading game list.");
@@ -380,7 +403,8 @@ void GameListWorker::AddFstEntriesToGameList(const std::string& dir_path, unsign
             return false; // Breaks the callback loop.
 
         bool is_dir = FileUtil::IsDirectory(physical_name);
-        if (!is_dir && HasSupportedFileExtension(physical_name)) {
+        if (!is_dir &&
+            (HasSupportedFileExtension(physical_name) || IsExtractedNCAMain(physical_name))) {
             std::unique_ptr<Loader::AppLoader> loader = Loader::GetLoader(physical_name);
             if (!loader)
                 return true;
@@ -392,7 +416,7 @@ void GameListWorker::AddFstEntriesToGameList(const std::string& dir_path, unsign
             loader->ReadProgramId(program_id);
 
             emit EntryReady({
-                new GameListItemPath(QString::fromStdString(physical_name), smdh, program_id),
+                new GameListItemPath(FormatGameName(physical_name), smdh, program_id),
                 new GameListItem(
                     QString::fromStdString(Loader::GetFileTypeString(loader->GetFileType()))),
                 new GameListItemSize(FileUtil::GetSize(physical_name)),

src/yuzu/main.cpp

@@ -13,6 +13,7 @@
 #include <QMessageBox>
 #include <QtGui>
 #include <QtWidgets>
+#include "common/common_paths.h"
 #include "common/logging/backend.h"
 #include "common/logging/filter.h"
 #include "common/logging/log.h"
@@ -33,7 +34,6 @@
 #include "yuzu/debugger/graphics/graphics_breakpoints.h"
 #include "yuzu/debugger/graphics/graphics_surface.h"
 #include "yuzu/debugger/profiler.h"
-#include "yuzu/debugger/registers.h"
 #include "yuzu/debugger/wait_tree.h"
 #include "yuzu/game_list.h"
 #include "yuzu/hotkeys.h"
@@ -169,15 +169,6 @@ void GMainWindow::InitializeDebugWidgets() {
     debug_menu->addAction(microProfileDialog->toggleViewAction());
 #endif
 
-    registersWidget = new RegistersWidget(this);
-    addDockWidget(Qt::RightDockWidgetArea, registersWidget);
-    registersWidget->hide();
-    debug_menu->addAction(registersWidget->toggleViewAction());
-    connect(this, &GMainWindow::EmulationStarting, registersWidget,
-            &RegistersWidget::OnEmulationStarting);
-    connect(this, &GMainWindow::EmulationStopping, registersWidget,
-            &RegistersWidget::OnEmulationStopping);
-
     graphicsBreakpointsWidget = new GraphicsBreakPointsWidget(debug_context, this);
     addDockWidget(Qt::RightDockWidgetArea, graphicsBreakpointsWidget);
     graphicsBreakpointsWidget->hide();
@@ -288,6 +279,7 @@ void GMainWindow::ConnectWidgetEvents() {
 void GMainWindow::ConnectMenuEvents() {
     // File
     connect(ui.action_Load_File, &QAction::triggered, this, &GMainWindow::OnMenuLoadFile);
+    connect(ui.action_Load_Folder, &QAction::triggered, this, &GMainWindow::OnMenuLoadFolder);
     connect(ui.action_Select_Game_List_Root, &QAction::triggered, this,
             &GMainWindow::OnMenuSelectGameListRoot);
     connect(ui.action_Exit, &QAction::triggered, this, &QMainWindow::close);
@@ -342,8 +334,6 @@ bool GMainWindow::SupportsRequiredGLExtensions() {
         unsupported_ext.append("ARB_program_interface_query");
     if (!GLAD_GL_ARB_separate_shader_objects)
         unsupported_ext.append("ARB_separate_shader_objects");
-    if (!GLAD_GL_ARB_shader_storage_buffer_object)
-        unsupported_ext.append("ARB_shader_storage_buffer_object");
     if (!GLAD_GL_ARB_vertex_attrib_binding)
         unsupported_ext.append("ARB_vertex_attrib_binding");
 
@@ -460,17 +450,12 @@ void GMainWindow::BootGame(const QString& filename) {
     connect(render_window, &GRenderWindow::Closed, this, &GMainWindow::OnStopGame);
     // BlockingQueuedConnection is important here, it makes sure we've finished refreshing our views
     // before the CPU continues
-    connect(emu_thread.get(), &EmuThread::DebugModeEntered, registersWidget,
-            &RegistersWidget::OnDebugModeEntered, Qt::BlockingQueuedConnection);
     connect(emu_thread.get(), &EmuThread::DebugModeEntered, waitTreeWidget,
             &WaitTreeWidget::OnDebugModeEntered, Qt::BlockingQueuedConnection);
-    connect(emu_thread.get(), &EmuThread::DebugModeLeft, registersWidget,
-            &RegistersWidget::OnDebugModeLeft, Qt::BlockingQueuedConnection);
     connect(emu_thread.get(), &EmuThread::DebugModeLeft, waitTreeWidget,
             &WaitTreeWidget::OnDebugModeLeft, Qt::BlockingQueuedConnection);
 
     // Update the GUI
-    registersWidget->OnDebugModeEntered();
     if (ui.action_Single_Window_Mode->isChecked()) {
         game_list->hide();
     }
@@ -565,6 +550,8 @@ void GMainWindow::OnMenuLoadFile() {
     for (const auto& piece : game_list->supported_file_extensions)
         extensions += "*." + piece + " ";
 
+    extensions += "main ";
+
     QString file_filter = tr("Switch Executable") + " (" + extensions + ")";
     file_filter += ";;" + tr("All Files (*.*)");
 
@@ -577,6 +564,18 @@ void GMainWindow::OnMenuLoadFile() {
     }
 }
 
+void GMainWindow::OnMenuLoadFolder() {
+    QDir dir = QFileDialog::getExistingDirectory(this, tr("Open Extracted ROM Directory"));
+
+    QStringList matching_main = dir.entryList(QStringList("main"), QDir::Files);
+    if (matching_main.size() == 1) {
+        BootGame(dir.path() + DIR_SEP + matching_main[0]);
+    } else {
+        QMessageBox::warning(this, tr("Invalid Directory Selected"),
+                             tr("The directory you have selected does not contain a 'main' file."));
+    }
+}
+
 void GMainWindow::OnMenuSelectGameListRoot() {
     QString dir_path = QFileDialog::getExistingDirectory(this, tr("Select Directory"));
     if (!dir_path.isEmpty()) {

src/yuzu/main.h

@@ -19,7 +19,6 @@ class GraphicsSurfaceWidget;
 class GRenderWindow;
 class MicroProfileDialog;
 class ProfilerWidget;
-class RegistersWidget;
 class WaitTreeWidget;
 
 namespace Tegra {
@@ -124,6 +123,7 @@ private slots:
     void OnGameListLoadFile(QString game_path);
     void OnGameListOpenSaveFolder(u64 program_id);
     void OnMenuLoadFile();
+    void OnMenuLoadFolder();
     /// Called whenever a user selects the "File->Select Game List Root" menu item
     void OnMenuSelectGameListRoot();
     void OnMenuRecentFile();
@@ -163,7 +163,6 @@ private:
     // Debugger panes
     ProfilerWidget* profilerWidget;
     MicroProfileDialog* microProfileDialog;
-    RegistersWidget* registersWidget;
     GraphicsBreakPointsWidget* graphicsBreakpointsWidget;
     GraphicsSurfaceWidget* graphicsSurfaceWidget;
     WaitTreeWidget* waitTreeWidget;

src/yuzu/main.ui

@@ -58,6 +58,7 @@
      </property>
     </widget>
     <addaction name="action_Load_File"/>
+    <addaction name="action_Load_Folder"/>
     <addaction name="separator"/>
     <addaction name="action_Select_Game_List_Root"/>
     <addaction name="menu_recent_files"/>
@@ -106,6 +107,11 @@
     <string>Load File...</string>
    </property>
   </action>
+  <action name="action_Load_Folder">
+   <property name="text">
+    <string>Load Folder...</string>
+   </property>
+  </action>
   <action name="action_Load_Symbol_Map">
    <property name="text">
     <string>Load Symbol Map...</string>