Merge branch 'master' into TemporaryHackishBranch

# Conflicts: # .gitignore # CMakeLists.txt # Core/HLE/sceNet.cpp # UI/GameSettingsScreen.cpp # UI/MainScreen.cpp # UI/MiscScreens.cpp # UI/ui_atlas.cpp # UI/ui_atlas.h # Windows/resource.h # android/.gitignore # android/assets/ui_atlas.zim # assets/ui_atlas.zim # assets/ui_atlas_lowmem.zim # atlasscript_lowmem.txt # ext/native/tools/atlastool/atlastool.vcxproj # ext/native/tools/tools.sln # ext/native/ui/ui_context.cpp # gradle/wrapper/gradle-wrapper.properties also fix atlas generation.
2026-11-15 02:29:51 +01:00 · 2020-04-05 19:34:24 +02:00
commit 6e202e5bdd
@@ -28,8 +28,6 @@ Windows/ARM64
 Windows/ipch
 Windows/bin-release
 Windows/win-version.h
-ui_atlas.zim.png
-ui_atlas_high.zim.png
 android/assets
 ext/_Output
 android/lint.xml
@@ -87,10 +85,10 @@ __testfinish.txt
 __testfailure.bmp
 GameLogNotes.txt
 screenshots
-android/ui_atlas_luna.zim
 android/assets/lang
 android/assets/flash0
-ppge_atlas.zim.png
+ui_atlas_luna.zim.png
+ppge_atlas_luna.zim.png
 local.properties
 r.sh

@@ -28,3 +28,6 @@
 [submodule "ext/rapidjson"]
 	path = ext/rapidjson
 	url = https://github.com/Tencent/rapidjson.git
+[submodule "ext/native/tools/prebuilt"]
+	path = ext/native/tools/prebuilt
+	url = https://github.com/hrydgard/ppsspp-freetype.git
@@ -86,11 +86,11 @@ matrix:
      env: PPSSPP_BUILD_TYPE=Linux
           LIBRETRO=TRUE
    - os: osx
-      osx_image: xcode8.3
+      osx_image: xcode9
      compiler: "clang"
      env: PPSSPP_BUILD_TYPE=macOS
    - os: osx
-      osx_image: xcode8.3
+      osx_image: xcode9
      compiler: "clang"
      env: PPSSPP_BUILD_TYPE=iOS
    - os: windows
@@ -166,6 +166,10 @@ endif()

 include_directories(ext/native)

+# Work around for some misfeature of the current glslang build system 
+include_directories(ext/glslang)
+
+
 if(NOT OPENGL_LIBRARIES AND USING_GLES2)
 	set(OPENGL_LIBRARIES GLESv2 EGL)
 endif()
@@ -268,6 +272,7 @@ if(NOT MSVC)
 	# Disable some warnings
 	add_definitions(-Wno-multichar)
 	add_definitions(-Wno-deprecated-register)
+	add_definitions(-Wno-tautological-pointer-compare)

 	# Don't compile with strict aliasing, we're not 100% aliasing-safe
 	add_compile_options(-fno-strict-aliasing)
@@ -485,12 +490,6 @@ add_library(gason STATIC
 	ext/native/ext/gason/gason.h
 )

-add_library(rg_etc1 STATIC
-	ext/native/ext/rg_etc1/rg_etc1.cpp
-	ext/native/ext/rg_etc1/rg_etc1.h
-)
-include_directories(ext/native/ext/rg_etc1)
-
 if(USE_FFMPEG)
 	if(NOT FFMPEG_DIR)
 		if(NOT USE_SYSTEM_FFMPEG)
@@ -661,13 +660,16 @@ else()
 	set(LIBZIP_LIBRARY libzip)
 endif()

-# FindPNG does a few things we don't want. So do it ourselves. Fixed to libpng17
-find_path(PNG_PNG_INCLUDE_DIR NAMES "libpng17/png.h")
-find_library(PNG_LIBRARY NAMES png17 libpng17)
-find_package(PackageHandleStandardArgs)
-find_package_handle_standard_args(PNG REQUIRED_VARS PNG_LIBRARY PNG_PNG_INCLUDE_DIR)
+# Arm platforms require at least libpng17.
+if(ANDROID OR ARMV7 OR ARM64 OR ARM OR IOS)
+	set(PNG_REQUIRED_VERSION 1.7)
+else()
+	set(PNG_REQUIRED_VERSION 1.6)
+endif()
+
+find_package(PNG ${PNG_REQUIRED_VERSION})
 if(PNG_FOUND)
-	include_directories(${PNG_PNG_INCLUDE_DIR})
+	include_directories(${PNG_INCLUDE_DIRS})
 else()
 	if(ARM)
 		set(PNG_ARM_INCLUDES
@@ -701,8 +703,8 @@ else()
 		ext/native/ext/libpng17/pngwutil.c
 		${PNG_ARM_INCLUDES}
 	)
-	set(PNG_LIBRARY png17)
-	include_directories(ext/native/ext)
+	set(PNG_LIBRARIES png17)
+	include_directories(ext/native/ext/libpng17)
 endif()

 set(nativeExtra)
@@ -844,6 +846,7 @@ list(APPEND NativeAppSource
 	UI/DiscordIntegration.cpp
 	UI/NativeApp.cpp
 	UI/BackgroundAudio.cpp
+	UI/ChatScreen.cpp
 	UI/DevScreens.cpp
 	UI/DisplayLayoutEditor.cpp
 	UI/DisplayLayoutScreen.cpp
@@ -869,7 +872,6 @@ list(APPEND NativeAppSource
 	UI/CwCheatScreen.cpp
 	UI/InstallZipScreen.cpp
 	UI/ProfilerDraw.cpp
-	UI/ui_atlas.cpp
 	UI/TextureUtil.cpp
 	UI/ComboKeyMappingScreen.cpp
 )
@@ -1042,6 +1044,8 @@ add_library(native STATIC
 	ext/native/thread/threadutil.h
 	ext/native/thread/threadpool.cpp
 	ext/native/thread/threadpool.h
+	ext/native/ui/root.cpp
+	ext/native/ui/root.h
 	ext/native/ui/screen.cpp
 	ext/native/ui/screen.h
 	ext/native/ui/ui.cpp
@@ -1082,7 +1086,7 @@ if(ANDROID)
 	set(ATOMIC_LIB atomic)
 endif()

-target_link_libraries(native ${LIBZIP_LIBRARY} ${PNG_LIBRARY} ${ZLIB_LIBRARY} rg_etc1 gason udis86 ${RT_LIB} ${nativeExtraLibs} ${ATOMIC_LIB})
+target_link_libraries(native ${LIBZIP_LIBRARY} ${PNG_LIBRARIES} ${ZLIB_LIBRARY} gason udis86 ${RT_LIB} ${nativeExtraLibs} ${ATOMIC_LIB})
 if(TARGET Ext::GLEW)
 	target_link_libraries(native Ext::GLEW)
 endif()
@@ -1760,8 +1764,6 @@ add_library(${CoreLibName} ${CoreLinkType}
 	Core/Util/BlockAllocator.h
 	Core/Util/PPGeDraw.cpp
 	Core/Util/PPGeDraw.h
-	Core/Util/ppge_atlas.cpp
-	Core/Util/ppge_atlas.h
 	${CORE_NEON}
 	${GPU_SOURCES}
 	ext/disarm.cpp
@@ -1805,7 +1807,7 @@ if(FFmpeg_FOUND)
 endif()

 # Discord integration
-if(USE_DISCORD AND NOT IOS)
+if(USE_DISCORD AND NOT IOS AND NOT LIBRETRO)
 	add_definitions(-DUSE_DISCORD=1)
 	target_link_libraries(${CoreLibName} discord-rpc)
 endif()
@@ -1933,6 +1935,7 @@ endif()

 set(NativeAssets
 	android/assets/ui_atlas_luna.zim
+	android/assets/ui_atlas_luna.meta
 	assets/lang
 	assets/shaders
 	assets/scalers
@@ -1941,7 +1944,8 @@ set(NativeAssets
 	assets/compat.ini
 	assets/gamecontrollerdb.txt
 	assets/langregion.ini
-	assets/ppge_atlas.zim
+	assets/ppge_atlas_luna.zim
+	assets/ppge_atlas_luna.meta
 	assets/rargray.png
 	assets/unknown.png
 	assets/zip.png
@@ -117,6 +117,19 @@ void PointerWrap::Do(std::wstring &x) {
 	(*ptr) += stringLen;
 }

+void PointerWrap::Do(std::u16string &x) {
+	int stringLen = sizeof(char16_t) * ((int)x.length() + 1);
+	Do(stringLen);
+
+	switch (mode) {
+	case MODE_READ: x = (char16_t*)*ptr; break;
+	case MODE_WRITE: memcpy(*ptr, x.c_str(), stringLen); break;
+	case MODE_MEASURE: break;
+	case MODE_VERIFY: _dbg_assert_msg_(COMMON, x == (char16_t*)*ptr, "Savestate verification failure: (at %p).\n", x.c_str()); break;
+	}
+	(*ptr) += stringLen;
+}
+
 struct standard_tm {
 	int tm_sec;
 	int tm_min;
@@ -459,7 +459,8 @@ public:

 	// Store strings.
 	void Do(std::string &x);
-	void Do(std::wstring &x);
+	void Do(std::wstring &x);  // DEPRECATED, do not save wstrings
+	void Do(std::u16string &x);

 	void Do(tm &t);

@@ -137,7 +137,7 @@
      <PreprocessorDefinitions>_CRTDBG_MAP_ALLOC;USING_WIN_UI;_CRT_SECURE_NO_WARNINGS;WIN32;_ARCH_32=1;_M_IX86=1;_DEBUG;_LIB;_UNICODE;UNICODE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
      <ForcedIncludeFiles>stdafx.h;Common/DbgNew.h</ForcedIncludeFiles>
      <RuntimeLibrary>MultiThreadedDebug</RuntimeLibrary>
-      <AdditionalIncludeDirectories>../ext/native;../ext/snappy;..</AdditionalIncludeDirectories>
+      <AdditionalIncludeDirectories>../ext/native;../ext/snappy;../ext/glslang;..</AdditionalIncludeDirectories>
      <EnableEnhancedInstructionSet>StreamingSIMDExtensions2</EnableEnhancedInstructionSet>
      <FloatingPointModel>Precise</FloatingPointModel>
      <MultiProcessorCompilation>true</MultiProcessorCompilation>
@@ -164,7 +164,7 @@
      <ForcedIncludeFiles>stdafx.h;Common/DbgNew.h</ForcedIncludeFiles>
      <RuntimeLibrary>MultiThreadedDebug</RuntimeLibrary>
      <PreprocessorDefinitions>_CRTDBG_MAP_ALLOC;USING_WIN_UI;_CRT_SECURE_NO_WARNINGS;WIN32;_ARCH_64=1;_M_X64=1;_DEBUG;_LIB;_UNICODE;UNICODE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
-      <AdditionalIncludeDirectories>../ext/native;../ext/snappy;..</AdditionalIncludeDirectories>
+      <AdditionalIncludeDirectories>../ext/native;../ext/snappy;../ext/glslang;..</AdditionalIncludeDirectories>
      <EnableEnhancedInstructionSet>NotSet</EnableEnhancedInstructionSet>
      <FloatingPointModel>Precise</FloatingPointModel>
      <OmitFramePointers>false</OmitFramePointers>
@@ -192,7 +192,7 @@
      <ForcedIncludeFiles>stdafx.h;Common/DbgNew.h</ForcedIncludeFiles>
      <RuntimeLibrary>MultiThreadedDebug</RuntimeLibrary>
      <PreprocessorDefinitions>_CRTDBG_MAP_ALLOC;USING_WIN_UI;_CRT_SECURE_NO_WARNINGS;WIN32;_ARCH_64=1;_DEBUG;_LIB;_UNICODE;UNICODE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
-      <AdditionalIncludeDirectories>../ext/native;../ext/snappy;..</AdditionalIncludeDirectories>
+      <AdditionalIncludeDirectories>../ext/native;../ext/snappy;../ext/glslang;..</AdditionalIncludeDirectories>
      <EnableEnhancedInstructionSet>NotSet</EnableEnhancedInstructionSet>
      <FloatingPointModel>Precise</FloatingPointModel>
      <OmitFramePointers>false</OmitFramePointers>
@@ -220,7 +220,7 @@
      <ForcedIncludeFiles>stdafx.h</ForcedIncludeFiles>
      <RuntimeLibrary>MultiThreadedDebug</RuntimeLibrary>
      <PreprocessorDefinitions>_CRTDBG_MAP_ALLOC;USING_WIN_UI;_CRT_SECURE_NO_WARNINGS;WIN32;_ARCH_32=1;_DEBUG;_LIB;_UNICODE;UNICODE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
-      <AdditionalIncludeDirectories>../ext/native;../ext/snappy;..</AdditionalIncludeDirectories>
+      <AdditionalIncludeDirectories>../ext/native;../ext/snappy;../ext/glslang;..</AdditionalIncludeDirectories>
      <EnableEnhancedInstructionSet>NotSet</EnableEnhancedInstructionSet>
      <FloatingPointModel>Precise</FloatingPointModel>
      <OmitFramePointers>false</OmitFramePointers>
@@ -253,7 +253,7 @@
      <BufferSecurityCheck>false</BufferSecurityCheck>
      <EnableEnhancedInstructionSet>StreamingSIMDExtensions2</EnableEnhancedInstructionSet>
      <FloatingPointModel>Precise</FloatingPointModel>
-      <AdditionalIncludeDirectories>../ext/native;../ext/snappy;..</AdditionalIncludeDirectories>
+      <AdditionalIncludeDirectories>../ext/native;../ext/snappy;../ext/glslang;..</AdditionalIncludeDirectories>
      <FavorSizeOrSpeed>Speed</FavorSizeOrSpeed>
      <MultiProcessorCompilation>true</MultiProcessorCompilation>
      <RuntimeTypeInfo>false</RuntimeTypeInfo>
@@ -283,7 +283,7 @@
      <IntrinsicFunctions>true</IntrinsicFunctions>
      <PreprocessorDefinitions>USING_WIN_UI;_CRT_SECURE_NO_WARNINGS;WIN32;_ARCH_64=1;_M_X64=1;NDEBUG;_LIB;_UNICODE;UNICODE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
      <ForcedIncludeFiles>stdafx.h</ForcedIncludeFiles>
-      <AdditionalIncludeDirectories>../ext/native;../ext/snappy;..</AdditionalIncludeDirectories>
+      <AdditionalIncludeDirectories>../ext/native;../ext/snappy;../ext/glslang;..</AdditionalIncludeDirectories>
      <BufferSecurityCheck>false</BufferSecurityCheck>
      <EnableEnhancedInstructionSet>NotSet</EnableEnhancedInstructionSet>
      <FloatingPointModel>Precise</FloatingPointModel>
@@ -318,7 +318,7 @@
      <IntrinsicFunctions>true</IntrinsicFunctions>
      <PreprocessorDefinitions>USING_WIN_UI;_CRT_SECURE_NO_WARNINGS;WIN32;_ARCH_64=1;NDEBUG;_LIB;_UNICODE;UNICODE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
      <ForcedIncludeFiles>stdafx.h</ForcedIncludeFiles>
-      <AdditionalIncludeDirectories>../ext/native;../ext/snappy;..</AdditionalIncludeDirectories>
+      <AdditionalIncludeDirectories>../ext/native;../ext/snappy;../ext/glslang;..</AdditionalIncludeDirectories>
      <BufferSecurityCheck>false</BufferSecurityCheck>
      <EnableEnhancedInstructionSet>NotSet</EnableEnhancedInstructionSet>
      <FloatingPointModel>Precise</FloatingPointModel>
@@ -353,7 +353,7 @@
      <IntrinsicFunctions>true</IntrinsicFunctions>
      <PreprocessorDefinitions>USING_WIN_UI;_CRT_SECURE_NO_WARNINGS;WIN32;_ARCH_32=1;NDEBUG;_LIB;_UNICODE;UNICODE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
      <ForcedIncludeFiles>stdafx.h</ForcedIncludeFiles>
-      <AdditionalIncludeDirectories>../ext/native;../ext/snappy;..</AdditionalIncludeDirectories>
+      <AdditionalIncludeDirectories>../ext/native;../ext/snappy;../ext/glslang;..</AdditionalIncludeDirectories>
      <BufferSecurityCheck>false</BufferSecurityCheck>
      <EnableEnhancedInstructionSet>NotSet</EnableEnhancedInstructionSet>
      <FloatingPointModel>Precise</FloatingPointModel>
@@ -74,6 +74,7 @@
    <ClInclude Include="Hashmaps.h" />
    <ClInclude Include="Vulkan\VulkanDebug.h" />
    <ClInclude Include="BitScan.h" />
+    <ClInclude Include="MakeUnique.h" />
  </ItemGroup>
  <ItemGroup>
    <ClCompile Include="stdafx.cpp" />
@@ -64,11 +64,7 @@ ConsoleListener::ConsoleListener() : bHidden(true)
 		logPending = new char[LOG_PENDING_MAX];
 	}
 	++refCount;
-#elif defined(ANDROID)
-	bUseColor = false;
-#elif defined(IOS)
-	bUseColor = false;
-#elif PPSSPP_PLATFORM(UWP) || PPSSPP_PLATFORM(SWITCH)
+#elif PPSSPP_PLATFORM(ANDROID) || PPSSPP_PLATFORM(IOS) || PPSSPP_PLATFORM(UWP) || PPSSPP_PLATFORM(SWITCH)
 	bUseColor = false;
 #elif defined(_MSC_VER)
 	bUseColor = false;
@@ -0,0 +1,21 @@
+// Copyright 2014 Dolphin Emulator Project
+// Licensed under GPLv2+
+// Refer to the license.txt file included.
+
+#include "ppsspp_config.h"
+#if PPSSPP_PLATFORM(SWITCH)
+#include <switch.h>
+#include "Common/Log.h"
+#include "Common/GL/GLInterface/EGLSwitch.h"
+
+EGLDisplay cInterfaceEGLSwitch::OpenDisplay() {
+	return eglGetDisplay(EGL_DEFAULT_DISPLAY);
+}
+
+EGLNativeWindowType cInterfaceEGLSwitch::InitializePlatform(EGLNativeWindowType host_window, EGLConfig config) {
+	return nwindowGetDefault();
+}
+
+void cInterfaceEGLSwitch::ShutdownPlatform() {
+}
+#endif
@@ -0,0 +1,24 @@
+// Copyright 2014 Dolphin Emulator Project
+// Licensed under GPLv2+
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "Common/GL/GLInterface/EGL.h"
+
+class cInterfaceEGLSwitch : public cInterfaceEGL {
+public:
+	cInterfaceEGLSwitch() {}
+protected:
+	EGLDisplay OpenDisplay() override;
+	EGLNativeWindowType InitializePlatform(EGLNativeWindowType host_window, EGLConfig config) override;
+	void ShutdownPlatform() override;
+	void OverrideBackbufferDimensions(int internalWidth, int internalHeight) override {
+		internalWidth_ = internalWidth;
+		internalHeight_ = internalHeight;
+	}
+
+private:
+	int internalWidth_ = 0;
+	int internalHeight_ = 0;
+};
@@ -2,10 +2,13 @@
 // Licensed under GPLv2+
 // Refer to the license.txt file included.

+#include "ppsspp_config.h"
 #include "Common/GL/GLInterfaceBase.h"

 #ifdef __ANDROID__
 #include "Common/GL/GLInterface/EGLAndroid.h"
+#elif PPSSPP_PLATFORM(SWITCH)
+#include "Common/GL/GLInterface/EGLSwitch.h"
 #elif defined(__APPLE__)
 #include "Common/GL/GLInterface/AGL.h"
 #elif defined(_WIN32)
@@ -23,6 +26,8 @@
 cInterfaceBase* HostGL_CreateGLInterface(){
 	#ifdef __ANDROID__
 		return new cInterfaceEGLAndroid;
+	#elif if PPSSPP_PLATFORM(SWITCH)
+		return new cInterfaceEGLSwitch;
 	#elif defined(__APPLE__)
 		return new cInterfaceAGL;
 	#elif defined(_WIN32)
@@ -343,6 +343,7 @@ void UpdateNativeMenuKeys() {
 		KeyDef(DEVICE_ID_KEYBOARD, NKCODE_ESCAPE),
 		KeyDef(DEVICE_ID_ANY, NKCODE_BACK),
 		KeyDef(DEVICE_ID_ANY, NKCODE_BUTTON_B),
+		KeyDef(DEVICE_ID_MOUSE, NKCODE_EXT_MOUSEBUTTON_4),
 	};

 	for (size_t i = 0; i < ARRAY_SIZE(hardcodedCancelKeys); i++) {
@@ -696,12 +697,16 @@ const KeyMap_IntStrPair psp_button_names[] = {
 	{VIRTKEY_AXIS_RIGHT_Y_MIN, "RightAn.Down"},
 	{VIRTKEY_AXIS_RIGHT_X_MIN, "RightAn.Left"},
 	{VIRTKEY_AXIS_RIGHT_X_MAX, "RightAn.Right"},
+	{VIRTKEY_OPENCHAT, "OpenChat" },

 	{VIRTKEY_AXIS_SWAP, "AxisSwap"},
 	{VIRTKEY_DEVMENU, "DevMenu"},
 	{VIRTKEY_TEXTURE_DUMP, "Texture Dumping"},
 	{VIRTKEY_TEXTURE_REPLACE, "Texture Replacement"},
 	{VIRTKEY_SCREENSHOT, "Screenshot"},
+	{VIRTKEY_MUTE_TOGGLE, "Mute toggle"},
+	{VIRTKEY_ANALOG_ROTATE_CW, "Rotate Analog (CW)"},
+	{VIRTKEY_ANALOG_ROTATE_CCW, "Rotate Analog (CCW)"},

 	{CTRL_HOME, "Home"},
 	{CTRL_HOLD, "Hold"},
@@ -57,6 +57,10 @@ enum {
 	VIRTKEY_TEXTURE_DUMP = 0x40000019,
 	VIRTKEY_TEXTURE_REPLACE = 0x4000001A,
 	VIRTKEY_SCREENSHOT = 0x4000001B,
+	VIRTKEY_MUTE_TOGGLE = 0x4000001C,
+	VIRTKEY_OPENCHAT = 0x4000001D,
+	VIRTKEY_ANALOG_ROTATE_CW = 0x4000001E,
+	VIRTKEY_ANALOG_ROTATE_CCW = 0x4000001F,
 	VIRTKEY_LAST,
 	VIRTKEY_COUNT = VIRTKEY_LAST - VIRTKEY_FIRST
 };
@@ -122,7 +122,7 @@ void AndroidAssertLog(const char *func, const char *file, int line, const char *
 #if MAX_LOGLEVEL >= DEBUG_LEVEL
 #define _dbg_assert_(_t_, _a_) \
 	if (!(_a_)) {\
-		ERROR_LOG(_t_, "Error...\n\n  Line: %d\n  File: %s\n\nIgnore and continue?", \
+		ERROR_LOG(_t_, #_a_ "\n\nError...\n\n  Line: %d\n  File: %s\n\nIgnore and continue?", \
 					   __LINE__, __FILE__); \
 		if (!PanicYesNo("*** Assertion ***\n")) { Crash(); } \
 	}
@@ -184,4 +184,4 @@ void AndroidAssertLog(const char *func, const char *file, int line, const char *
    Crash(); \
 	}

-#endif  // __ANDROID__
+#endif  // __ANDROID__
@@ -50,7 +50,7 @@ static SYSTEM_INFO sys_info;
 #endif

 #define MEM_PAGE_MASK ((MEM_PAGE_SIZE)-1)
-#define round_page(x) ((((uintptr_t)(x)) + MEM_PAGE_MASK) & ~(MEM_PAGE_MASK))
+#define ppsspp_round_page(x) ((((uintptr_t)(x)) + MEM_PAGE_MASK) & ~(MEM_PAGE_MASK))

 #ifdef _WIN32
 // Win32 flags are odd...
@@ -127,7 +127,7 @@ void *AllocateExecutableMemory(size_t size) {
 		GetSystemInfo(&sys_info);
 #if defined(_M_X64)
 	if ((uintptr_t)&hint_location > 0xFFFFFFFFULL) {
-		size_t aligned_size = round_page(size);
+		size_t aligned_size = ppsspp_round_page(size);
 #if 1   // Turn off to hunt for RIP bugs on x86-64.
 		ptr = SearchForFreeMem(aligned_size);
 		if (!ptr) {
@@ -161,11 +161,11 @@ void *AllocateExecutableMemory(size_t size) {
 	// We use a dummy global variable to give us a good location to start from.
 	if (!map_hint) {
 		if ((uintptr_t) &hint_location > 0xFFFFFFFFULL)
-			map_hint = (char*)round_page(&hint_location) - 0x20000000; // 0.5gb lower than our approximate location
+			map_hint = (char*)ppsspp_round_page(&hint_location) - 0x20000000; // 0.5gb lower than our approximate location
 		else
 			map_hint = (char*)0x20000000; // 0.5GB mark in memory
 	} else if ((uintptr_t) map_hint > 0xFFFFFFFFULL) {
-		map_hint -= round_page(size); /* round down to the next page if we're in high memory */
+		map_hint -= ppsspp_round_page(size); /* round down to the next page if we're in high memory */
 	}
 #endif

@@ -191,7 +191,7 @@ void *AllocateExecutableMemory(size_t size) {
 #if defined(_M_X64) && !defined(_WIN32)
 	else if ((uintptr_t)map_hint <= 0xFFFFFFFF) {
 		// Round up if we're below 32-bit mark, probably allocating sequentially.
-		map_hint += round_page(size);
+		map_hint += ppsspp_round_page(size);

 		// If we moved ahead too far, skip backwards and recalculate.
 		// When we free, we keep moving forward and eventually move too far.
@@ -204,7 +204,7 @@ void *AllocateExecutableMemory(size_t size) {
 }

 void *AllocateMemoryPages(size_t size, uint32_t memProtFlags) {
-	size = round_page(size);
+	size = ppsspp_round_page(size);
 #ifdef _WIN32
 	if (sys_info.dwPageSize == 0)
 		GetSystemInfo(&sys_info);
@@ -285,6 +285,14 @@ void VulkanContext::EndFrame() {
 	}
 }

+void VulkanContext::UpdateInflightFrames(int n) {
+	assert(n >= 1 && n <= MAX_INFLIGHT_FRAMES);
+	inflightFrames_ = n;
+	if (curFrame_ >= inflightFrames_) {
+		curFrame_ = 0;
+	}
+}
+
 void VulkanContext::WaitUntilQueueIdle() {
 	// Should almost never be used
 	vkQueueWaitIdle(gfx_queue_);
@@ -949,20 +957,17 @@ bool VulkanContext::InitSwapchain() {
 		ILOG("Supported present mode: %d (%s)", presentModes[i], PresentModeString(presentModes[i]));
 	}
 	for (size_t i = 0; i < presentModeCount; i++) {
-		if (swapchainPresentMode == VK_PRESENT_MODE_MAX_ENUM_KHR) {
-			// Default to the first present mode from the list.
+		bool match = false;
+		match = match || ((flags_ & VULKAN_FLAG_PRESENT_MAILBOX) && presentModes[i] == VK_PRESENT_MODE_MAILBOX_KHR);
+		match = match || ((flags_ & VULKAN_FLAG_PRESENT_FIFO_RELAXED) && presentModes[i] == VK_PRESENT_MODE_FIFO_RELAXED_KHR);
+		match = match || ((flags_ & VULKAN_FLAG_PRESENT_FIFO) && presentModes[i] == VK_PRESENT_MODE_FIFO_KHR);
+		match = match || ((flags_ & VULKAN_FLAG_PRESENT_IMMEDIATE) && presentModes[i] == VK_PRESENT_MODE_IMMEDIATE_KHR);
+
+		// Default to the first present mode from the list.
+		if (match || swapchainPresentMode == VK_PRESENT_MODE_MAX_ENUM_KHR) {
 			swapchainPresentMode = presentModes[i];
 		}
-		if ((flags_ & VULKAN_FLAG_PRESENT_MAILBOX) && presentModes[i] == VK_PRESENT_MODE_MAILBOX_KHR) {
-			swapchainPresentMode = VK_PRESENT_MODE_MAILBOX_KHR;
-			break;
-		}
-		if ((flags_ & VULKAN_FLAG_PRESENT_FIFO_RELAXED) && presentModes[i] == VK_PRESENT_MODE_FIFO_RELAXED_KHR) {
-			swapchainPresentMode = VK_PRESENT_MODE_FIFO_RELAXED_KHR;
-			break;
-		}
-		if ((flags_ & VULKAN_FLAG_PRESENT_IMMEDIATE) && presentModes[i] == VK_PRESENT_MODE_IMMEDIATE_KHR) {
-			swapchainPresentMode = VK_PRESENT_MODE_IMMEDIATE_KHR;
+		if (match) {
 			break;
 		}
 	}
@@ -13,6 +13,7 @@ enum {
 	VULKAN_FLAG_PRESENT_MAILBOX = 2,
 	VULKAN_FLAG_PRESENT_IMMEDIATE = 4,
 	VULKAN_FLAG_PRESENT_FIFO_RELAXED = 8,
+	VULKAN_FLAG_PRESENT_FIFO = 16,
 };

 enum {
@@ -136,6 +137,7 @@ public:
 	VkDevice GetDevice() const { return device_; }
 	VkInstance GetInstance() const { return instance_; }
 	uint32_t GetFlags() const { return flags_; }
+	void UpdateFlags(uint32_t flags) { flags_ = flags; }

 	VulkanDeleteList &Delete() { return globalDeleteList_; }

@@ -174,10 +176,13 @@ public:
 	VkResult InitDebugMsgCallback(PFN_vkDebugReportCallbackEXT dbgFunc, int bits, void *userdata);
 	void DestroyDebugMsgCallback();

-	VkPhysicalDevice GetPhysicalDevice(int n = 0) const {
+	VkPhysicalDevice GetPhysicalDevice(int n) const {
 		return physical_devices_[n];
 	}
-	int GetCurrentPhysicalDevice() const {
+	VkPhysicalDevice GetCurrentPhysicalDevice() const {
+		return physical_devices_[physical_device_];
+	}
+	int GetCurrentPhysicalDeviceIndex() const {
 		return physical_device_;
 	}
 	int GetNumPhysicalDevices() const {
@@ -200,7 +205,7 @@ public:

 	const PhysicalDeviceProps &GetPhysicalDeviceProperties(int i = -1) const {
 		if (i < 0)
-			i = GetCurrentPhysicalDevice();
+			i = GetCurrentPhysicalDeviceIndex();
 		return physicalDeviceProperties_[i];
 	}

@@ -249,6 +254,8 @@ public:
 	int GetInflightFrames() const {
 		return inflightFrames_;
 	}
+	// Don't call while a frame is in progress.
+	void UpdateInflightFrames(int n);

 	int GetCurFrame() const {
 		return curFrame_;
@@ -236,6 +236,7 @@ bool g_vulkanMayBeAvailable = false;

 static const char *device_name_blacklist[] = {
 	"NVIDIA:SHIELD Tablet K1",
+	"SDL:Horizon",
 };

 static const char *so_names[] = {
@@ -441,6 +441,7 @@ static ConfigSetting generalSettings[] = {
 	ConfigSetting("AutoLoadSaveState", &g_Config.iAutoLoadSaveState, 0, true, true),
 	ReportedConfigSetting("EnableCheats", &g_Config.bEnableCheats, true, true, true),
 	ConfigSetting("CwCheatRefreshRate", &g_Config.iCwCheatRefreshRate, 77, true, true),
+	ConfigSetting("CwCheatScrollPosition", &g_Config.fCwCheatScrollPosition, 0.0f, true, true),

 	ConfigSetting("ScreenshotsAsPNG", &g_Config.bScreenshotsAsPNG, false, true, true),
 	ConfigSetting("ScreenshotsAtRenderRes", &g_Config.bScreenshotsAtRenderRes, false, true, true),
@@ -454,6 +455,9 @@ static ConfigSetting generalSettings[] = {
 	ConfigSetting("EnableStateUndo", &g_Config.bEnableStateUndo, false/*&DefaultEnableStateUndo*/, true, true),
 	ConfigSetting("RewindFlipFrequency", &g_Config.iRewindFlipFrequency, 0, true, true),

+	ConfigSetting("ShowRegionOnGameIcon", &g_Config.bShowRegionOnGameIcon, false, true, true),
+	ConfigSetting("ShowIDOnGameIcon", &g_Config.bShowIDOnGameIcon, false, true, true),
+	ConfigSetting("GameGridScale", &g_Config.fGameGridScale, 1.0, true, true),
 	ConfigSetting("GridView1", &g_Config.bGridView1, true),
 	ConfigSetting("GridView2", &g_Config.bGridView2, true),
 	ConfigSetting("GridView3", &g_Config.bGridView3, false),
@@ -786,6 +790,9 @@ static ConfigSetting graphicsSettings[] = {
 	ConfigSetting("GfxDebugSplitSubmit", &g_Config.bGfxDebugSplitSubmit, false, false, false),
 	ConfigSetting("LogFrameDrops", &g_Config.bLogFrameDrops, false, true, false),

+	ConfigSetting("InflightFrames", &g_Config.iInflightFrames, 3, true, false),
+	ConfigSetting("RenderDuplicateFrames", &g_Config.bRenderDuplicateFrames, false, true, true),
+
 	ConfigSetting(false),
 };

@@ -840,6 +847,12 @@ static ConfigSetting controlSettings[] = {
 	ConfigSetting("ComboKey3Mapping", &g_Config.iCombokey3, 0, true, true),
 	ConfigSetting("ComboKey4Mapping", &g_Config.iCombokey4, 0, true, true),

+	ConfigSetting("ComboKey0Toggle", &g_Config.bComboToggle0, false, true, true),
+	ConfigSetting("ComboKey1Toggle", &g_Config.bComboToggle1, false, true, true),
+	ConfigSetting("ComboKey2Toggle", &g_Config.bComboToggle2, false, true, true),
+	ConfigSetting("ComboKey3Toggle", &g_Config.bComboToggle3, false, true, true),
+	ConfigSetting("ComboKey4Toggle", &g_Config.bComboToggle4, false, true, true),
+
 #if defined(_WIN32)
 	// A win32 user seeing touch controls is likely using PPSSPP on a tablet. There it makes
 	// sense to default this to on.
@@ -861,6 +874,7 @@ static ConfigSetting controlSettings[] = {
 	ConfigSetting("TiltSensitivityX", &g_Config.iTiltSensitivityX, 100, true, true),
 	ConfigSetting("TiltSensitivityY", &g_Config.iTiltSensitivityY, 100, true, true),
 	ConfigSetting("DeadzoneRadius", &g_Config.fDeadzoneRadius, 0.2f, true, true),
+	ConfigSetting("TiltDeadzoneSkip", &g_Config.fTiltDeadzoneSkip, 0.0f, true, true),
 	ConfigSetting("TiltInputType", &g_Config.iTiltInputType, 0, true, true),
 #endif

@@ -870,6 +884,7 @@ static ConfigSetting controlSettings[] = {
 	ConfigSetting("TouchButtonOpacity", &g_Config.iTouchButtonOpacity, 65, true, true),
 	ConfigSetting("TouchButtonHideSeconds", &g_Config.iTouchButtonHideSeconds, 20, true, true),
 	ConfigSetting("AutoCenterTouchAnalog", &g_Config.bAutoCenterTouchAnalog, false, true, true),
+	ConfigSetting("AnalogAutoRotSpeed", &g_Config.fAnalogAutoRotSpeed, 15.0f, true, true),

 	// Snap touch control position
 	ConfigSetting("TouchSnapToGrid", &g_Config.bTouchSnapToGrid, false, true, true),
@@ -898,6 +913,8 @@ static ConfigSetting controlSettings[] = {
 	ConfigSetting("Speed1KeyX", "Speed1KeyY", "Speed1KeyScale", "ShowSpeed1Key", &g_Config.touchSpeed1Key, defaultTouchPosHide, true, true),
 	ConfigSetting("Speed2KeyX", "Speed2KeyY", "Speed2KeyScale", "ShowSpeed2Key", &g_Config.touchSpeed2Key, defaultTouchPosHide, true, true),
 	ConfigSetting("RapidFireKeyX", "RapidFireKeyY", "RapidFireKeyScale", "ShowRapidFireKey", &g_Config.touchRapidFireKey, defaultTouchPosHide, true, true),
+	ConfigSetting("AnalogRotationCWKeyX", "AnalogRotationKeyCWY", "AnalogRotationKeyCWScale", "ShowAnalogRotationCWKey", &g_Config.touchAnalogRotationCWKey, defaultTouchPosHide, true, true),
+	ConfigSetting("AnalogRotationCCWKeyX", "AnalogRotationKeyCCWY", "AnalogRotationKeyCCWScale", "ShowAnalogRotationCCWKey", &g_Config.touchAnalogRotationCCWKey, defaultTouchPosHide, true, true),

 #ifdef _WIN32
 	ConfigSetting("DInputAnalogDeadzone", &g_Config.fDInputAnalogDeadzone, 0.1f, true, true),
@@ -927,7 +944,15 @@ static ConfigSetting controlSettings[] = {
 static ConfigSetting networkSettings[] = {
 	ConfigSetting("EnableWlan", &g_Config.bEnableWlan, false, true, true),
 	ConfigSetting("EnableAdhocServer", &g_Config.bEnableAdhocServer, false, true, true),
-
+	ConfigSetting("EnableNetworkChat", &g_Config.bEnableNetworkChat, false, true, true),
+	ConfigSetting("ChatButtonPosition",&g_Config.iChatButtonPosition,BOTTOM_LEFT,true,true),
+	ConfigSetting("ChatScreenPosition",&g_Config.iChatScreenPosition,BOTTOM_LEFT,true,true),
+	ConfigSetting("EnableQuickChat", &g_Config.bEnableQuickChat, true, true, true),
+	ConfigSetting("QuickChat1", &g_Config.sQuickChat0, "Quick Chat 1", true, true),
+	ConfigSetting("QuickChat2", &g_Config.sQuickChat1, "Quick Chat 2", true, true),
+	ConfigSetting("QuickChat3", &g_Config.sQuickChat2, "Quick Chat 3", true, true),
+	ConfigSetting("QuickChat4", &g_Config.sQuickChat3, "Quick Chat 4", true, true),
+	ConfigSetting("QuickChat5", &g_Config.sQuickChat4, "Quick Chat 5", true, true),
 	ConfigSetting(false),
 };

@@ -968,7 +993,7 @@ static ConfigSetting systemParamSettings[] = {
 	ReportedConfigSetting("ButtonPreference", &g_Config.iButtonPreference, PSP_SYSTEMPARAM_BUTTON_CROSS, true, true),
 	ConfigSetting("LockParentalLevel", &g_Config.iLockParentalLevel, 0, true, true),
 	ConfigSetting("WlanAdhocChannel", &g_Config.iWlanAdhocChannel, PSP_SYSTEMPARAM_ADHOC_CHANNEL_AUTOMATIC, true, true),
-#if defined(USING_WIN_UI)
+#if defined(USING_WIN_UI) || defined(USING_QT_UI) || PPSSPP_PLATFORM(ANDROID)
 	ConfigSetting("BypassOSKWithKeyboard", &g_Config.bBypassOSKWithKeyboard, false, true, true),
 	ConfigSetting("DisableWinMenu", &g_Config.bDisableWinMenu, false, true, true),
 	ConfigSetting("DisableWinBorders", &g_Config.bDisableWinBorders, false, true, true),
@@ -1116,6 +1141,12 @@ std::map<std::string, std::pair<std::string, int>> GetLangValuesMapping() {
 	return langValuesMapping;
 }

+void Config::Reload() {
+	reload_ = true;
+	Load();
+	reload_ = false;
+}
+
 void Config::Load(const char *iniFileName, const char *controllerIniFilename) {
 	const bool useIniFilename = iniFileName != nullptr && strlen(iniFileName) > 0;
 	iniFilename_ = FindConfigFile(useIniFilename ? iniFileName : "ppsspp.ini");
@@ -1492,7 +1523,8 @@ bool Config::hasGameConfig(const std::string &pGameId) {
 }

 void Config::changeGameSpecific(const std::string &pGameId, const std::string &title) {
-	Save("changeGameSpecific");
+	if (!reload_)
+		Save("changeGameSpecific");
 	gameId_ = pGameId;
 	gameIdTitle_ = title;
 	bGameSpecific = !pGameId.empty();
@@ -1571,7 +1603,7 @@ bool Config::loadGameConfig(const std::string &pGameId, const std::string &title
 }

 void Config::unloadGameConfig() {
-	if (bGameSpecific){
+	if (bGameSpecific) {
 		changeGameSpecific();

 		IniFile iniFile;
@@ -1624,6 +1656,8 @@ void Config::ResetControlLayout() {
 	reset(g_Config.touchSpeed1Key);
 	reset(g_Config.touchSpeed2Key);
 	reset(g_Config.touchRapidFireKey);
+	reset(g_Config.touchAnalogRotationCWKey);
+	reset(g_Config.touchAnalogRotationCCWKey);
 }

 void Config::GetReportingInfo(UrlEncoder &data) {
@@ -28,6 +28,17 @@
 extern const char *PPSSPP_GIT_VERSION;
 const int MAX_CONFIG_VOLUME = 8;

+enum ChatPositions {
+	BOTTOM_LEFT = 0,
+	BOTTOM_CENTER = 1,
+	BOTOM_RIGHT = 2,
+	TOP_LEFT = 3,
+	TOP_CENTER = 4,
+	TOP_RIGHT = 5,
+	CENTER_LEFT = 6,
+	CENTER_RIGHT = 7,
+};
+
 namespace http {
 	class Download;
 	class Downloader;
@@ -196,6 +207,7 @@ public:
 	bool bEnableCheats;
 	bool bReloadCheats;
 	int iCwCheatRefreshRate;
+	float fCwCheatScrollPosition;
 	int iBloomHack; //0 = off, 1 = safe, 2 = balanced, 3 = aggressive
 	bool bBlockTransferGPU;
 	bool bDisableSlowFramebufEffects;
@@ -205,6 +217,8 @@ public:
 	std::string sPostShaderName;  // Off for off.
 	bool bGfxDebugOutput;
 	bool bGfxDebugSplitSubmit;
+	int iInflightFrames;
+	bool bRenderDuplicateFrames;

 	// Sound
 	bool bEnableSound;
@@ -219,6 +233,9 @@ public:
 	// UI
 	bool bShowDebuggerOnLoad;
 	int iShowFPSCounter;
+	bool bShowRegionOnGameIcon;
+	bool bShowIDOnGameIcon;
+	float fGameGridScale;

 	// TODO: Maybe move to a separate theme system.
 	uint32_t uItemStyleFg;
@@ -268,6 +285,8 @@ public:
 	int iTiltSensitivityY;
 	//the deadzone radius of the tilt
 	float fDeadzoneRadius;
+	// deadzone skip
+	float fTiltDeadzoneSkip;
 	//type of tilt input currently selected: Defined in TiltEventProcessor.h
 	//0 - no tilt, 1 - analog stick, 2 - D-Pad, 3 - Action Buttons (Tri, Cross, Square, Circle)
 	int iTiltInputType;
@@ -286,6 +305,8 @@ public:
 	int iTouchButtonStyle;
 	int iTouchButtonOpacity;
 	int iTouchButtonHideSeconds;
+	// Auto rotation speed
+	float fAnalogAutoRotSpeed;

 	// Snap touch control position
 	bool bTouchSnapToGrid;
@@ -319,6 +340,8 @@ public:
 	ConfigTouchPos touchSpeed1Key;
 	ConfigTouchPos touchSpeed2Key;
 	ConfigTouchPos touchRapidFireKey;
+	ConfigTouchPos touchAnalogRotationCWKey;
+	ConfigTouchPos touchAnalogRotationCCWKey;

 	// Controls Visibility
 	bool bShowTouchControls;
@@ -335,6 +358,12 @@ public:
 	int iCombokey3;
 	int iCombokey4;

+	bool bComboToggle0;
+	bool bComboToggle1;
+	bool bComboToggle2;
+	bool bComboToggle3;
+	bool bComboToggle4;
+
 	// Ignored on iOS and other platforms that lack pause.
 	bool bShowTouchPause;

@@ -388,10 +417,20 @@ public:
 	bool bEnableAdhocServer;
 	int iWlanAdhocChannel;
 	bool bWlanPowerSave;
+	bool bEnableNetworkChat;
+	//for chat position , moveable buttons is better than this 
+	int iChatButtonPosition;
+	int iChatScreenPosition;
+
+	bool bEnableQuickChat;
+	std::string sQuickChat0;
+	std::string sQuickChat1;
+	std::string sQuickChat2;
+	std::string sQuickChat3;
+	std::string sQuickChat4;

 	int iPSPModel;
 	int iFirmwareVersion;
-	// TODO: Make this work with your platform, too!
 	bool bBypassOSKWithKeyboard;
 #if defined(USING_WIN_UI)
 	bool bDisableWinMenu;
@@ -438,6 +477,7 @@ public:

 	void Load(const char *iniFileName = nullptr, const char *controllerIniFilename = nullptr);
 	void Save(const char *saveReason);
+	void Reload();
 	void RestoreDefaults();

 	//per game config managment, should maybe be in it's own class
@@ -476,6 +516,7 @@ protected:
 	void LoadStandardControllerIni();

 private:
+	bool reload_ = false;
 	std::string gameId_;
 	std::string gameIdTitle_;
 	std::string iniFilename_;
@@ -71,7 +71,6 @@
    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
    <Import Project="..\Windows\fix_2017.props" />
  </ImportGroup>
-
  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.Default.props" />
  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'" Label="Configuration">
    <ConfigurationType>StaticLibrary</ConfigurationType>
@@ -134,7 +133,7 @@
  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
    <ClCompile>
      <WarningLevel>Level3</WarningLevel>
-      <AdditionalIncludeDirectories>..\ffmpeg\WindowsInclude;..\ffmpeg\Windows\x86\include;../common;..;../ext/native;../ext/glew;../ext/snappy;../ext/zlib;../ext/native/ext</AdditionalIncludeDirectories>
+      <AdditionalIncludeDirectories>..\ffmpeg\WindowsInclude;..\ffmpeg\Windows\x86\include;../common;..;../ext/native;../ext/glew;../ext/snappy;../ext/native/ext/libpng17;../ext/zlib;../ext/native/ext</AdditionalIncludeDirectories>
      <PreprocessorDefinitions>_CRTDBG_MAP_ALLOC;USING_WIN_UI;_CRT_SECURE_NO_WARNINGS;USE_FFMPEG;WIN32;_ARCH_32=1;_M_IX86=1;_DEBUG;_LIB;_UNICODE;UNICODE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
      <EnableEnhancedInstructionSet>StreamingSIMDExtensions2</EnableEnhancedInstructionSet>
      <FloatingPointModel>Precise</FloatingPointModel>
@@ -156,7 +155,7 @@
  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
    <ClCompile>
      <WarningLevel>Level3</WarningLevel>
-      <AdditionalIncludeDirectories>..\ffmpeg\WindowsInclude;..\ffmpeg\Windows\x86_64\include;../common;..;../ext/native;../ext/glew;../ext/snappy;../ext/zlib;../ext/native/ext</AdditionalIncludeDirectories>
+      <AdditionalIncludeDirectories>..\ffmpeg\WindowsInclude;..\ffmpeg\Windows\x86_64\include;../common;..;../ext/native;../ext/glew;../ext/snappy;../ext/native/ext/libpng17;../ext/zlib;../ext/native/ext</AdditionalIncludeDirectories>
      <PreprocessorDefinitions>_CRTDBG_MAP_ALLOC;USING_WIN_UI;_CRT_SECURE_NO_WARNINGS;USE_FFMPEG;WIN32;_ARCH_64=1;_M_X64=1;_DEBUG;_LIB;_UNICODE;UNICODE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
      <EnableEnhancedInstructionSet>NotSet</EnableEnhancedInstructionSet>
      <FloatingPointModel>Precise</FloatingPointModel>
@@ -179,7 +178,7 @@
  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|ARM64'">
    <ClCompile>
      <WarningLevel>Level3</WarningLevel>
-      <AdditionalIncludeDirectories>..\ffmpeg\WindowsInclude;..\ffmpeg\Windows\aarch64\include;../common;..;../ext/native;../ext/glew;../ext/snappy;../ext/zlib;../ext/native/ext</AdditionalIncludeDirectories>
+      <AdditionalIncludeDirectories>..\ffmpeg\WindowsInclude;..\ffmpeg\Windows\aarch64\include;../common;..;../ext/native;../ext/glew;../ext/snappy;../ext/native/ext/libpng17;../ext/zlib;../ext/native/ext</AdditionalIncludeDirectories>
      <PreprocessorDefinitions>_CRTDBG_MAP_ALLOC;USING_WIN_UI;_CRT_SECURE_NO_WARNINGS;USE_FFMPEG;WIN32;_ARCH_64=1;_DEBUG;_LIB;_UNICODE;UNICODE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
      <EnableEnhancedInstructionSet>NotSet</EnableEnhancedInstructionSet>
      <FloatingPointModel>Precise</FloatingPointModel>
@@ -202,7 +201,7 @@
  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|ARM'">
    <ClCompile>
      <WarningLevel>Level3</WarningLevel>
-      <AdditionalIncludeDirectories>..\ffmpeg\WindowsInclude;..\ffmpeg\Windows\arm\include;../common;..;../ext/native;../ext/glew;../ext/snappy;../ext/zlib;../ext/native/ext</AdditionalIncludeDirectories>
+      <AdditionalIncludeDirectories>..\ffmpeg\WindowsInclude;..\ffmpeg\Windows\arm\include;../common;..;../ext/native;../ext/glew;../ext/snappy;../ext/native/ext/libpng17;../ext/zlib;../ext/native/ext</AdditionalIncludeDirectories>
      <PreprocessorDefinitions>_CRTDBG_MAP_ALLOC;USING_WIN_UI;_CRT_SECURE_NO_WARNINGS;USE_FFMPEG;WIN32;_ARCH_32=1;_DEBUG;_LIB;_UNICODE;UNICODE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
      <EnableEnhancedInstructionSet>NotSet</EnableEnhancedInstructionSet>
      <FloatingPointModel>Precise</FloatingPointModel>
@@ -229,7 +228,7 @@
      <Optimization>MaxSpeed</Optimization>
      <FunctionLevelLinking>true</FunctionLevelLinking>
      <IntrinsicFunctions>true</IntrinsicFunctions>
-      <AdditionalIncludeDirectories>..\ffmpeg\WindowsInclude;..\ffmpeg\Windows\x86\include;../common;..;../ext/native;../ext/glew;../ext/snappy;../ext/zlib;../ext/native/ext</AdditionalIncludeDirectories>
+      <AdditionalIncludeDirectories>..\ffmpeg\WindowsInclude;..\ffmpeg\Windows\x86\include;../common;..;../ext/native;../ext/glew;../ext/snappy;../ext/native/ext/libpng17;../ext/zlib;../ext/native/ext</AdditionalIncludeDirectories>
      <BufferSecurityCheck>false</BufferSecurityCheck>
      <EnableEnhancedInstructionSet>StreamingSIMDExtensions2</EnableEnhancedInstructionSet>
      <FloatingPointModel>Precise</FloatingPointModel>
@@ -260,7 +259,7 @@
      <Optimization>MaxSpeed</Optimization>
      <FunctionLevelLinking>true</FunctionLevelLinking>
      <IntrinsicFunctions>true</IntrinsicFunctions>
-      <AdditionalIncludeDirectories>..\ffmpeg\WindowsInclude;..\ffmpeg\Windows\x86_64\include;../common;..;../ext/native;../ext/glew;../ext/snappy;../ext/zlib;../ext/native/ext</AdditionalIncludeDirectories>
+      <AdditionalIncludeDirectories>..\ffmpeg\WindowsInclude;..\ffmpeg\Windows\x86_64\include;../common;..;../ext/native;../ext/glew;../ext/snappy;../ext/native/ext/libpng17;../ext/zlib;../ext/native/ext</AdditionalIncludeDirectories>
      <EnableEnhancedInstructionSet>NotSet</EnableEnhancedInstructionSet>
      <FloatingPointModel>Precise</FloatingPointModel>
      <BufferSecurityCheck>false</BufferSecurityCheck>
@@ -290,7 +289,7 @@
      <Optimization>MaxSpeed</Optimization>
      <FunctionLevelLinking>true</FunctionLevelLinking>
      <IntrinsicFunctions>true</IntrinsicFunctions>
-      <AdditionalIncludeDirectories>..\ffmpeg\WindowsInclude;..\ffmpeg\Windows\aarch64\include;../common;..;../ext/native;../ext/glew;../ext/snappy;../ext/zlib;../ext/native/ext</AdditionalIncludeDirectories>
+      <AdditionalIncludeDirectories>..\ffmpeg\WindowsInclude;..\ffmpeg\Windows\aarch64\include;../common;..;../ext/native;../ext/glew;../ext/snappy;../ext/native/ext/libpng17;../ext/zlib;../ext/native/ext</AdditionalIncludeDirectories>
      <EnableEnhancedInstructionSet>NotSet</EnableEnhancedInstructionSet>
      <FloatingPointModel>Precise</FloatingPointModel>
      <BufferSecurityCheck>false</BufferSecurityCheck>
@@ -320,7 +319,7 @@
      <Optimization>MaxSpeed</Optimization>
      <FunctionLevelLinking>true</FunctionLevelLinking>
      <IntrinsicFunctions>true</IntrinsicFunctions>
-      <AdditionalIncludeDirectories>..\ffmpeg\WindowsInclude;..\ffmpeg\Windows\arm\include;../common;..;../ext/native;../ext/glew;../ext/snappy;../ext/zlib;../ext/native/ext</AdditionalIncludeDirectories>
+      <AdditionalIncludeDirectories>..\ffmpeg\WindowsInclude;..\ffmpeg\Windows\arm\include;../common;..;../ext/native;../ext/glew;../ext/snappy;../ext/native/ext/libpng17;../ext/zlib;../ext/native/ext</AdditionalIncludeDirectories>
      <EnableEnhancedInstructionSet>NotSet</EnableEnhancedInstructionSet>
      <FloatingPointModel>Precise</FloatingPointModel>
      <BufferSecurityCheck>false</BufferSecurityCheck>
@@ -834,7 +833,6 @@
    <ClCompile Include="Util\DisArm64.cpp" />
    <ClCompile Include="Util\GameManager.cpp" />
    <ClCompile Include="Util\PPGeDraw.cpp" />
-    <ClCompile Include="Util\ppge_atlas.cpp" />
    <ClCompile Include="..\ext\xxhash.c">
      <Optimization Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">MaxSpeed</Optimization>
      <IntrinsicFunctions Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">true</IntrinsicFunctions>
@@ -1166,7 +1164,6 @@
    <ClInclude Include="Util\DisArm64.h" />
    <ClInclude Include="Util\GameManager.h" />
    <ClInclude Include="Util\PPGeDraw.h" />
-    <ClInclude Include="Util\ppge_atlas.h" />
    <ClInclude Include="..\ext\xxhash.h" />
    <ClInclude Include="WaveFile.h" />
    <ClInclude Include="WebServer.h" />
@@ -336,9 +336,6 @@
    <ClCompile Include="Util\PPGeDraw.cpp">
      <Filter>Util</Filter>
    </ClCompile>
-    <ClCompile Include="Util\ppge_atlas.cpp">
-      <Filter>Util</Filter>
-    </ClCompile>
    <ClCompile Include="HLE\sceFont.cpp">
      <Filter>HLE\Libraries</Filter>
    </ClCompile>
@@ -1000,9 +997,6 @@
    <ClInclude Include="Util\PPGeDraw.h">
      <Filter>Util</Filter>
    </ClInclude>
-    <ClInclude Include="Util\ppge_atlas.h">
-      <Filter>Util</Filter>
-    </ClInclude>
    <ClInclude Include="HLE\sceFont.h">
      <Filter>HLE\Libraries</Filter>
    </ClInclude>
@@ -17,7 +17,7 @@

 #include <algorithm>
 #ifndef USING_QT_UI
-#include <libpng17/png.h>
+#include <png.h>
 #include <zlib.h>
 #endif
 #include "data/base64.h"
@@ -144,8 +144,13 @@ void PSPDialog::DoState(PointerWrap &p)
 	p.Do(isFading);
 	p.Do(fadeIn);
 	p.Do(fadeValue);
+
+	// I don't think we should save these two... Let's just ignore them for now for compat.
+	int okButtonImg = 0;
 	p.Do(okButtonImg);
+	int cancelButtonImg = 0;
 	p.Do(cancelButtonImg);
+
 	p.Do(okButtonFlag);
 	p.Do(cancelButtonFlag);

@@ -17,6 +17,8 @@

 #pragma once

+#include "gfx/texture_atlas.h"
+
 #include "Common/Common.h"
 #include "Common/CommonTypes.h"
 #include "Common/Swap.h"
@@ -112,8 +114,8 @@ protected:
 	bool fadeIn;
 	u32 fadeValue;

-	int okButtonImg;
-	int cancelButtonImg;
+	ImageID okButtonImg;
+	ImageID cancelButtonImg;
 	int okButtonFlag;
 	int cancelButtonFlag;
 };
@@ -211,7 +211,7 @@ void PSPGamedataInstallDialog::WriteSfoFile() {
 	size_t sfoSize;
 	sfoFile.WriteSFO(&sfoData,&sfoSize);

-	u32 handle = pspFileSystem.OpenFile(sfopath, (FileAccess)(FILEACCESS_WRITE | FILEACCESS_CREATE | FILEACCESS_TRUNCATE));
+	int handle = pspFileSystem.OpenFile(sfopath, (FileAccess)(FILEACCESS_WRITE | FILEACCESS_CREATE | FILEACCESS_TRUNCATE));
 	if (handle >= 0) {
 		pspFileSystem.WriteFile(handle, sfoData, sfoSize);
 		pspFileSystem.CloseFile(handle);
@@ -62,7 +62,7 @@ private:
 	u64 allReadSize;   // use this to calculate progress value.
 	int progressValue;

-	u32 currentInputFile;
+	int currentInputFile;
 	u32 currentInputBytesLeft;
-	u32 currentOutputFile;
+	int currentOutputFile;
 };
@@ -45,7 +45,7 @@ int PSPMsgDialog::Init(unsigned int paramAddr) {
 	// Ignore if already running
 	if (GetStatus() != SCE_UTILITY_STATUS_NONE) {
 		ERROR_LOG_REPORT(SCEUTILITY, "sceUtilityMsgDialogInitStart: invalid status");
-		return 0;
+		return SCE_ERROR_UTILITY_INVALID_STATUS;
 	}

 	messageDialogAddr = paramAddr;
@@ -148,10 +148,9 @@ void PSPMsgDialog::DisplayMessage(std::string text, bool hasYesNo, bool hasOK)
 		WRAP_WIDTH = 372.0f;

 	float y = 140.0f;
-	float h, sy ,ey;
-	int n;
-	PPGeMeasureText(0, &h, &n, text.c_str(), FONT_SCALE, PPGE_LINE_WRAP_WORD, WRAP_WIDTH);
-	float h2 = h * n / 2.0f;
+	float h, sy, ey;
+	PPGeMeasureText(nullptr, &h, text.c_str(), FONT_SCALE, PPGE_LINE_WRAP_WORD, WRAP_WIDTH);
+	float h2 = h / 2.0f;
 	ey = y + h2 + 20.0f;

 	if (hasYesNo)
@@ -172,7 +171,7 @@ void PSPMsgDialog::DisplayMessage(std::string text, bool hasYesNo, bool hasOK)
 			yesColor = 0xFFFFFFFF;
 			noColor  = 0xFFFFFFFF;
 		}
-		PPGeMeasureText(&w, &h, 0, choiceText, FONT_SCALE);
+		PPGeMeasureText(&w, &h, choiceText, FONT_SCALE);
 		w = 15.0f;
 		h = 8.0f;
 		float y2 = y + h2 + 8.0f;
@@ -225,14 +224,14 @@ int PSPMsgDialog::Update(int animSpeed) {
 		UpdateButtons();
 		UpdateFade(animSpeed);

-		okButtonImg = I_CIRCLE;
-		cancelButtonImg = I_CROSS;
+		okButtonImg = ImageID("I_CIRCLE");
+		cancelButtonImg = ImageID("I_CROSS");
 		okButtonFlag = CTRL_CIRCLE;
 		cancelButtonFlag = CTRL_CROSS;
 		if (messageDialog.common.buttonSwap == 1)
 		{
-			okButtonImg = I_CROSS;
-			cancelButtonImg = I_CIRCLE;
+			okButtonImg = ImageID("I_CROSS");
+			cancelButtonImg = ImageID("I_CIRCLE");
 			okButtonFlag = CTRL_CROSS;
 			cancelButtonFlag = CTRL_CIRCLE;
 		}
@@ -28,9 +28,14 @@
 #define NETCONF_CONNECT_APNET 0
 #define NETCONF_STATUS_APNET 1
 #define NETCONF_CONNECT_ADHOC 2
+#define NETCONF_CONNECT_APNET_LAST 3
 #define NETCONF_CREATE_ADHOC 4
 #define NETCONF_JOIN_ADHOC 5

+// Needs testing.
+const static int NET_INIT_DELAY_US = 300000;
+const static int NET_SHUTDOWN_DELAY_US = 26000;
+
 PSPNetconfDialog::PSPNetconfDialog() {
 }

@@ -39,7 +44,7 @@ PSPNetconfDialog::~PSPNetconfDialog() {

 int PSPNetconfDialog::Init(u32 paramAddr) {
 	// Already running
-	if (status != SCE_UTILITY_STATUS_NONE && status != SCE_UTILITY_STATUS_SHUTDOWN)
+	if (status != SCE_UTILITY_STATUS_NONE)
 		return SCE_ERROR_UTILITY_INVALID_STATUS;

 	int size = Memory::Read_U32(paramAddr);
@@ -47,7 +52,7 @@ int PSPNetconfDialog::Init(u32 paramAddr) {
 	// Only copy the right size to support different request format
 	Memory::Memcpy(&request, paramAddr, size);

-	status = SCE_UTILITY_STATUS_INITIALIZE;
+	ChangeStatusInit(NET_INIT_DELAY_US);

 	// Eat any keys pressed before the dialog inited.
 	UpdateButtons();
@@ -67,16 +72,18 @@ void PSPNetconfDialog::DrawBanner() {
 }

 int PSPNetconfDialog::Update(int animSpeed) {
+	if (GetStatus() != SCE_UTILITY_STATUS_RUNNING) {
+		return SCE_ERROR_UTILITY_INVALID_STATUS;
+	}
+
 	UpdateButtons();
 	auto di = GetI18NCategory("Dialog");
 	auto err = GetI18NCategory("Error");
 	const float WRAP_WIDTH = 254.0f;
-	const int confirmBtnImage = g_Config.iButtonPreference == PSP_SYSTEMPARAM_BUTTON_CROSS ? I_CROSS : I_CIRCLE;
+	const ImageID confirmBtnImage = g_Config.iButtonPreference == PSP_SYSTEMPARAM_BUTTON_CROSS ? ImageID("I_CROSS") : ImageID("I_CIRCLE");
 	const int confirmBtn = g_Config.iButtonPreference == PSP_SYSTEMPARAM_BUTTON_CROSS ? CTRL_CROSS : CTRL_CIRCLE;

-	if (status == SCE_UTILITY_STATUS_INITIALIZE) {
-		status = SCE_UTILITY_STATUS_RUNNING;
-	} else if (status == SCE_UTILITY_STATUS_RUNNING && (request.netAction == NETCONF_CONNECT_APNET || request.netAction == NETCONF_STATUS_APNET)) {
+	if (request.netAction == NETCONF_CONNECT_APNET || request.netAction == NETCONF_STATUS_APNET || request.netAction == NETCONF_CONNECT_APNET_LAST) {
 		UpdateFade(animSpeed);
 		StartDraw();
 		DrawBanner();
@@ -87,26 +94,24 @@ int PSPNetconfDialog::Update(int animSpeed) {

 		if (IsButtonPressed(confirmBtn)) {
 			StartFade(false);
-			status = SCE_UTILITY_STATUS_FINISHED;
+			ChangeStatus(SCE_UTILITY_STATUS_FINISHED, 0);
 			// TODO: When the dialog is aborted, does it really set the result to this?
 			// It seems to make Phantasy Star Portable 2 happy, so it should be okay for now.
 			request.common.result = SCE_UTILITY_DIALOG_RESULT_ABORT;
 		}
-		
-	} else if (status == SCE_UTILITY_STATUS_RUNNING && (request.netAction == NETCONF_CONNECT_ADHOC || request.netAction == NETCONF_CREATE_ADHOC || request.netAction == NETCONF_JOIN_ADHOC)) {
-		if (request.NetconfData != NULL) {
+
+		EndDraw();
+	} else if (request.netAction == NETCONF_CONNECT_ADHOC || request.netAction == NETCONF_CREATE_ADHOC || request.netAction == NETCONF_JOIN_ADHOC) {
+		if (request.NetconfData.IsValid()) {
 			Shutdown(true);
 			if (sceNetAdhocctlCreate(request.NetconfData->groupName) == 0) {
-				status = SCE_UTILITY_STATUS_FINISHED;
+				ChangeStatus(SCE_UTILITY_STATUS_FINISHED, 0);
 				return 0;
 			}
 			return -1;
 		}
-	} else if (status == SCE_UTILITY_STATUS_FINISHED) {
-		status = SCE_UTILITY_STATUS_SHUTDOWN;
 	}

-	EndDraw();
 	return 0;
 }

@@ -114,7 +119,12 @@ int PSPNetconfDialog::Shutdown(bool force) {
 	if (status != SCE_UTILITY_STATUS_FINISHED && !force)
 		return SCE_ERROR_UTILITY_INVALID_STATUS;

-	return PSPDialog::Shutdown(force);
+	PSPDialog::Shutdown(force);
+	if (!force) {
+		ChangeStatusShutdown(NET_SHUTDOWN_DELAY_US);
+	}
+
+	return 0;
 }

 void PSPNetconfDialog::DoState(PointerWrap &p) {	
@@ -44,6 +44,11 @@ public:
 	virtual int Shutdown(bool force = false) override;
 	virtual void DoState(PointerWrap &p) override;

+protected:
+	bool UseAutoStatus() override {
+		return false;
+	}
+
 private:
 	void DrawBanner();
 	SceUtilityNetconfParam request;
@@ -15,9 +15,11 @@
 // Official git repository and contact information can be found at
 // https://github.com/hrydgard/ppsspp and http://www.ppsspp.org/.

+#include <algorithm>
+#include "base/NativeApp.h"
 #include "i18n/i18n.h"
 #include "math/math_util.h"
-#include <algorithm>
+#include "util/text/utf8.h"

 #include "Core/Dialog/PSPOskDialog.h"
 #include "Core/Util/PPGeDraw.h"
@@ -29,10 +31,6 @@
 #include "Common/ChunkFile.h"
 #include "GPU/GPUState.h"

-#if defined(USING_WIN_UI)
-#include "base/NativeApp.h"
-#endif
-
 #ifndef _WIN32
 #include <ctype.h>
 #include <math.h>
@@ -73,73 +71,73 @@ const int kor_lconsCom[] = {18,0,2,21,3,4,26,3,5,0,7,8,15,7,9,16,7,10,18,7,11,24
 // Korean (Hangul) last consonant Separation key
 const int kor_lconsSpr[] = {2,1,9,4,4,12,5,4,18,8,8,0,9,8,6,10,8,7,11,8,9,12,8,16,13,8,17,14,8,18,17,17,9};

-static const wchar_t oskKeys[OSK_KEYBOARD_COUNT][5][14] =
+static const char16_t oskKeys[OSK_KEYBOARD_COUNT][5][14] =
 {
 	{
 		// Latin Lowercase
-		{L"1234567890-+"},
-		{L"qwertyuiop[]"},
-		{L"asdfghjkl;@~"},
-		{L"zxcvbnm,./?\\"},
+		{u"1234567890-+"},
+		{u"qwertyuiop[]"},
+		{u"asdfghjkl;@~"},
+		{u"zxcvbnm,./?\\"},
 	},
 	{
 		// Latin Uppercase
-		{L"!@#$%^&*()_+"},
-		{L"QWERTYUIOP{}"},
-		{L"ASDFGHJKL:\"`"},
-		{L"ZXCVBNM<>/?|"},
+		{u"!@#$%^&*()_+"},
+		{u"QWERTYUIOP{}"},
+		{u"ASDFGHJKL:\"`"},
+		{u"ZXCVBNM<>/?|"},
 	},
 	{
 		// Hiragana
-		{L"あかさたなはまやらわぁゃっ"},
-		{L"いきしちにひみ　り　ぃ　　"},
-		{L"うくすつぬふむゆるをぅゅ゛"},
-		{L"えけせてねへめ　れ　ぇ　゜"},
-		{L"おこそとのほもよろんぉょー"},
+		{u"あかさたなはまやらわぁゃっ"},
+		{u"いきしちにひみ　り　ぃ　　"},
+		{u"うくすつぬふむゆるをぅゅ゛"},
+		{u"えけせてねへめ　れ　ぇ　゜"},
+		{u"おこそとのほもよろんぉょー"},
 	},
 	{
 		// Katakana
-		{L"アカサタナハマヤラワァャッ"},
-		{L"イキシチニヒミ　リ　ィ　　"},
-		{L"ウクスツヌフムユルヲゥュ゛"},
-		{L"エケセテネヘメ　レ　ェ　゜"},
-		{L"オコソトノホモヨロンォョー"},
+		{u"アカサタナハマヤラワァャッ"},
+		{u"イキシチニヒミ　リ　ィ　　"},
+		{u"ウクスツヌフムユルヲゥュ゛"},
+		{u"エケセテネヘメ　レ　ェ　゜"},
+		{u"オコソトノホモヨロンォョー"},
 	},
 	{
 		// Korean(Hangul)
-		{L"1234567890-+"},
-		{L"ㅃㅉㄸㄲㅆ!@#$%^&"},
-		{L"ㅂㅈㄷㄱㅅㅛㅕㅑㅐㅔ[]"},
-		{L"ㅁㄴㅇㄹㅎㅗㅓㅏㅣ;@~"},
-		{L"ㅋㅌㅊㅍㅠㅜㅡ<>/?|"},
+		{u"1234567890-+"},
+		{u"ㅃㅉㄸㄲㅆ!@#$%^&"},
+		{u"ㅂㅈㄷㄱㅅㅛㅕㅑㅐㅔ[]"},
+		{u"ㅁㄴㅇㄹㅎㅗㅓㅏㅣ;@~"},
+		{u"ㅋㅌㅊㅍㅠㅜㅡ<>/?|"},
 	},
 	{
 		// Russian Lowercase
-		{L"1234567890-+"},
-		{L"йцукенгшщзхъ"},
-		{L"фывапролджэё"},
-		{L"ячсмитьбю/?|"},
+		{u"1234567890-+"},
+		{u"йцукенгшщзхъ"},
+		{u"фывапролджэё"},
+		{u"ячсмитьбю/?|"},
 	},
 	{
 		// Russian Uppercase
-		{L"!@#$%^&*()_+"},
-		{L"ЙЦУКЕНГШЩЗХЪ"},
-		{L"ФЫВАПРОЛДЖЭЁ"},
-		{L"ЯЧСМИТЬБЮ/?|"},
+		{u"!@#$%^&*()_+"},
+		{u"ЙЦУКЕНГШЩЗХЪ"},
+		{u"ФЫВАПРОЛДЖЭЁ"},
+		{u"ЯЧСМИТЬБЮ/?|"},
 	},
 	{
 		// Latin Full-width Lowercase
-		{ L"１２３４５６７８９０－＋" },
-		{ L"ｑｗｅｒｔｙｕｉｏｐ［］" },
-		{ L"ａｓｄｆｇｈｊｋｌ；＠～" },
-		{ L"ｚｘｃｖｂｎｍ，．／？￥￥" },
+		{ u"１２３４５６７８９０－＋" },
+		{ u"ｑｗｅｒｔｙｕｉｏｐ［］" },
+		{ u"ａｓｄｆｇｈｊｋｌ；＠～" },
+		{ u"ｚｘｃｖｂｎｍ，．／？￥￥" },
 	},
 	{
 		// Latin Full-width Uppercase
-		{ L"！＠＃＄％＾＆＊（）＿＋" },
-		{ L"ＱＷＥＲＴＹＵＩＯＰ｛｝" },
-		{ L"ＡＳＤＦＧＨＪＫＬ：￥”‘" },
-		{ L"ＺＸＣＶＢＮＭ＜＞／？｜" },
+		{ u"！＠＃＄％＾＆＊（）＿＋" },
+		{ u"ＱＷＥＲＴＹＵＩＯＰ｛｝" },
+		{ u"ＡＳＤＦＧＨＪＫＬ：￥”‘" },
+		{ u"ＺＸＣＶＢＮＭ＜＞／？｜" },
 	},
 };

@@ -169,7 +167,7 @@ void PSPOskDialog::ConvertUCS2ToUTF8(std::string& _string, const PSPPointer<u16_
 {
 	if (!em_address.IsValid())
 	{
-		_string = "";
+		_string.clear();
 		return;
 	}

@@ -196,17 +194,17 @@ void PSPOskDialog::ConvertUCS2ToUTF8(std::string& _string, const PSPPointer<u16_
 	_string = stringBuffer;
 }

-void GetWideStringFromPSPPointer(std::wstring& _string, const PSPPointer<u16_le>& em_address)
+void GetWideStringFromPSPPointer(std::u16string& _string, const PSPPointer<u16_le>& em_address)
 {
 	if (!em_address.IsValid())
 	{
-		_string = L"";
+		_string.clear();
 		return;
 	}

 	const size_t maxLength = 2047;
-	wchar_t stringBuffer[maxLength + 1];
-	wchar_t *string = stringBuffer;
+	char16_t stringBuffer[maxLength + 1];
+	char16_t *string = stringBuffer;

 	u16_le *input = &em_address[0];
 	int c;
@@ -216,7 +214,7 @@ void GetWideStringFromPSPPointer(std::wstring& _string, const PSPPointer<u16_le>
 	_string = stringBuffer;
 }

-void PSPOskDialog::ConvertUCS2ToUTF8(std::string& _string, const wchar_t *input)
+void PSPOskDialog::ConvertUCS2ToUTF8(std::string& _string, const char16_t *input)
 {
 	char stringBuffer[2048];
 	char *string = stringBuffer;
@@ -317,7 +315,7 @@ int PSPOskDialog::Init(u32 oskPtr) {

 	i_level = 0;

-	inputChars = L"";
+	inputChars.clear();

 	if (oskParams->fields[0].intext.IsValid()) {
 		auto src = oskParams->fields[0].intext;
@@ -331,13 +329,16 @@ int PSPOskDialog::Init(u32 oskPtr) {
 	// Eat any keys pressed before the dialog inited.
 	UpdateButtons();

+	std::lock_guard<std::mutex> guard(nativeMutex_);
+	nativeStatus_ = PSPOskNativeStatus::IDLE;
+
 	StartFade(true);
 	return 0;
 }

-std::wstring PSPOskDialog::CombinationKorean(bool isInput)
+std::u16string PSPOskDialog::CombinationKorean(bool isInput)
 {
-	std::wstring string;
+	std::u16string string;

 	isCombinated = true;

@@ -577,9 +578,9 @@ std::wstring PSPOskDialog::CombinationKorean(bool isInput)
 	return string;
 }

-std::wstring PSPOskDialog::CombinationString(bool isInput)
+std::u16string PSPOskDialog::CombinationString(bool isInput)
 {
-	std::wstring string;
+	std::u16string string;

 	isCombinated = false;

@@ -658,7 +659,7 @@ std::wstring PSPOskDialog::CombinationString(bool isInput)
 				string += inputChars[i];
 			}

-			if (string.size() <= FieldMaxLength())
+			if (string.size() < FieldMaxLength())
 			{
 				string += oskKeys[currentKeyboard][selectedRow][selectedCol];
 			}
@@ -751,7 +752,7 @@ void PSPOskDialog::RenderKeyboard()
 	int selectedRow = selectedChar / numKeyCols[currentKeyboard];
 	int selectedCol = selectedChar % numKeyCols[currentKeyboard];

-	wchar_t temp[2];
+	char16_t temp[2];
 	temp[1] = '\0';

 	std::string buffer;
@@ -768,7 +769,7 @@ void PSPOskDialog::RenderKeyboard()

 	PPGeDrawText(oskDesc.c_str(), title , 20, PPGE_ALIGN_CENTER, 0.5f, CalcFadedColor(0xFFFFFFFF));

-	std::wstring result;
+	std::u16string result;

 	result = CombinationString(false);

@@ -836,7 +837,6 @@ void PSPOskDialog::RenderKeyboard()
 	}
 }

-#if defined(USING_WIN_UI)
 // TODO: Why does this have a 2 button press lag/delay when
 // re-opening the dialog box? I don't get it.
 int PSPOskDialog::NativeKeyboard() {
@@ -844,23 +844,52 @@ int PSPOskDialog::NativeKeyboard() {
 		return SCE_ERROR_UTILITY_INVALID_STATUS;
 	}

-	std::wstring titleText;
-	GetWideStringFromPSPPointer(titleText, oskParams->fields[0].desc);
+#if defined(USING_WIN_UI) || defined(USING_QT_UI) || PPSSPP_PLATFORM(ANDROID)
+	bool beginInputBox = false;
+	if (nativeStatus_ == PSPOskNativeStatus::IDLE) {
+		std::lock_guard<std::mutex> guard(nativeMutex_);
+		if (nativeStatus_ == PSPOskNativeStatus::IDLE) {
+			nativeStatus_ = PSPOskNativeStatus::WAITING;
+			beginInputBox = true;
+		}
+	}

-	std::wstring defaultText;
-	GetWideStringFromPSPPointer(defaultText, oskParams->fields[0].intext);
+	if (beginInputBox) {
+		std::u16string titleText;
+		GetWideStringFromPSPPointer(titleText, oskParams->fields[0].desc);

-	if (defaultText.empty())
-		defaultText.assign(L"VALUE");
+		std::u16string defaultText;
+		GetWideStringFromPSPPointer(defaultText, oskParams->fields[0].intext);
+
+		if (defaultText.empty())
+			defaultText.assign(u"VALUE");
+
+		// There's already ConvertUCS2ToUTF8 in this file. Should we use that instead of the global ones?
+		System_InputBoxGetString(::ConvertUCS2ToUTF8(titleText), ::ConvertUCS2ToUTF8(defaultText), [&](bool result, const std::string &value) {
+			std::lock_guard<std::mutex> guard(nativeMutex_);
+			if (nativeStatus_ != PSPOskNativeStatus::WAITING) {
+				return;
+			}
+
+			nativeValue_ = value;
+			nativeStatus_ = result ? PSPOskNativeStatus::SUCCESS : PSPOskNativeStatus::FAILURE;
+		});
+	} else if (nativeStatus_ == PSPOskNativeStatus::SUCCESS) {
+		inputChars = ConvertUTF8ToUCS2(nativeValue_);
+		nativeValue_.clear();

-	if (System_InputBoxGetWString(titleText.c_str(), defaultText, inputChars)) {
 		u32 maxLength = FieldMaxLength();
 		if (inputChars.length() > maxLength) {
 			ERROR_LOG(SCEUTILITY, "NativeKeyboard: input text too long(%d characters/glyphs max), truncating to game-requested length.", maxLength);
 			inputChars.erase(maxLength, std::string::npos);
 		}
+		ChangeStatus(SCE_UTILITY_STATUS_FINISHED, 0);
+		nativeStatus_ = PSPOskNativeStatus::DONE;
+	} else if (nativeStatus_ == PSPOskNativeStatus::FAILURE) {
+		ChangeStatus(SCE_UTILITY_STATUS_FINISHED, 0);
+		nativeStatus_ = PSPOskNativeStatus::DONE;
 	}
-	ChangeStatus(SCE_UTILITY_STATUS_FINISHED, 0);
+#endif
 	
 	u16_le *outText = oskParams->fields[0].outtext;

@@ -870,7 +899,7 @@ int PSPOskDialog::NativeKeyboard() {
 	// Only write the bytes of the output and the null terminator, don't write the rest.
 	for (size_t i = 0; i < end; ++i) {
 		u16 value = 0;
-		if (i < FieldMaxLength())
+		if (i < FieldMaxLength() && i < inputChars.size())
 			value = inputChars[i];
 		outText[i] = value;
 	}
@@ -880,7 +909,6 @@ int PSPOskDialog::NativeKeyboard() {

 	return 0;
 }
-#endif

 int PSPOskDialog::Update(int animSpeed) {
 	if (GetStatus() != SCE_UTILITY_STATUS_RUNNING) {
@@ -902,12 +930,10 @@ int PSPOskDialog::Update(int animSpeed) {
 	int selectedRow = selectedChar / numKeyCols[currentKeyboard];
 	int selectedExtra = selectedChar % numKeyCols[currentKeyboard];

-	// TODO: Add your platforms here when you have a NativeKeyboard func.
-
-#if defined(USING_WIN_UI)
+#if defined(USING_WIN_UI) || defined(USING_QT_UI) || PPSSPP_PLATFORM(ANDROID)
 	// Windows: Fall back to the OSK/continue normally if we're in fullscreen.
 	// The dialog box doesn't work right if in fullscreen.
-	if(g_Config.bBypassOSKWithKeyboard && !g_Config.bFullScreen)
+	if (g_Config.bBypassOSKWithKeyboard && !g_Config.bFullScreen)
 		return NativeKeyboard();
 #endif

@@ -919,15 +945,15 @@ int PSPOskDialog::Update(int animSpeed) {

 	auto di = GetI18NCategory("Dialog");

-	PPGeDrawImage(I_SQUARE, 365, 222, 16, 16, 0, CalcFadedColor(0xFFFFFFFF));
+	PPGeDrawImage(ImageID("I_SQUARE"), 365, 222, 16, 16, 0, CalcFadedColor(0xFFFFFFFF));
 	PPGeDrawText(di->T("Space"), 390, 222, PPGE_ALIGN_LEFT, 0.5f, CalcFadedColor(0xFFFFFFFF));

 	if (g_Config.iButtonPreference != PSP_SYSTEMPARAM_BUTTON_CIRCLE) {
-		PPGeDrawImage(I_CROSS, 45, 222, 16, 16, 0, CalcFadedColor(0xFFFFFFFF));
-		PPGeDrawImage(I_CIRCLE, 45, 247, 16, 16, 0, CalcFadedColor(0xFFFFFFFF));
+		PPGeDrawImage(ImageID("I_CROSS"), 45, 222, 16, 16, 0, CalcFadedColor(0xFFFFFFFF));
+		PPGeDrawImage(ImageID("I_CIRCLE"), 45, 247, 16, 16, 0, CalcFadedColor(0xFFFFFFFF));
 	} else {
-		PPGeDrawImage(I_CIRCLE, 45, 222, 16, 16, 0, CalcFadedColor(0xFFFFFFFF));
-		PPGeDrawImage(I_CROSS, 45, 247, 16, 16, 0, CalcFadedColor(0xFFFFFFFF));
+		PPGeDrawImage(ImageID("I_CIRCLE"), 45, 222, 16, 16, 0, CalcFadedColor(0xFFFFFFFF));
+		PPGeDrawImage(ImageID("I_CROSS"), 45, 247, 16, 16, 0, CalcFadedColor(0xFFFFFFFF));
 	}

 	PPGeDrawText(di->T("Select"), 75, 222, PPGE_ALIGN_LEFT, 0.5f, CalcFadedColor(0xFFFFFFFF));
@@ -1055,9 +1081,9 @@ int PSPOskDialog::Update(int animSpeed) {
 	} else if (IsButtonPressed(CTRL_SQUARE) && inputChars.size() < FieldMaxLength()) {
 		// Use a regular space if the current keyboard isn't Japanese nor full-width English
 		if (currentKeyboardLanguage != OSK_LANGUAGE_JAPANESE && currentKeyboardLanguage != OSK_LANGUAGE_ENGLISH_FW)
-			inputChars += L" ";
+			inputChars += u" ";
 		else
-			inputChars += L"　";
+			inputChars += u"　";
 	}

 	EndDraw();
@@ -1070,7 +1096,7 @@ int PSPOskDialog::Update(int animSpeed) {
 	for (size_t i = 0; i < end; ++i)
 	{
 		u16 value = 0;
-		if (i < inputChars.size())
+		if (i < FieldMaxLength() && i < inputChars.size())
 			value = inputChars[i];
 		outText[i] = value;
 	}
@@ -1089,6 +1115,7 @@ int PSPOskDialog::Shutdown(bool force)
 	if (!force) {
 		ChangeStatusShutdown(OSK_SHUTDOWN_DELAY_US);
 	}
+	nativeStatus_ = PSPOskNativeStatus::IDLE;

 	return 0;
 }
@@ -1097,7 +1124,7 @@ void PSPOskDialog::DoState(PointerWrap &p)
 {
 	PSPDialog::DoState(p);

-	auto s = p.Section("PSPOskDialog", 1);
+	auto s = p.Section("PSPOskDialog", 1, 2);
 	if (!s)
 		return;

@@ -1106,7 +1133,14 @@ void PSPOskDialog::DoState(PointerWrap &p)
 	p.Do(oskIntext);
 	p.Do(oskOuttext);
 	p.Do(selectedChar);
-	p.Do(inputChars);
+	if (s >= 2) {
+		p.Do(inputChars);
+	} else {
+		// Discard the wstring.
+		std::wstring wstr;
+		p.Do(wstr);
+	}
+	// Don't need to save state native status or value.
 }

 pspUtilityDialogCommon *PSPOskDialog::GetCommonParam()
@@ -17,6 +17,7 @@

 #pragma once

+#include <mutex>
 #include <string>

 #include "Core/Dialog/PSPDialog.h"
@@ -202,6 +203,14 @@ static const std::string OskKeyboardNames[] =
 	"English Full-width",
 };

+enum class PSPOskNativeStatus {
+	IDLE,
+	DONE,
+	WAITING,
+	SUCCESS,
+	FAILURE,
+};
+
 class PSPOskDialog: public PSPDialog {
 public:
 	PSPOskDialog();
@@ -220,14 +229,12 @@ protected:

 private:
 	void ConvertUCS2ToUTF8(std::string& _string, const PSPPointer<u16_le>& em_address);
-	void ConvertUCS2ToUTF8(std::string& _string, const wchar_t *input);
+	void ConvertUCS2ToUTF8(std::string& _string, const char16_t *input);
 	void RenderKeyboard();
-#if defined(USING_WIN_UI)
 	int NativeKeyboard();
-#endif

-	std::wstring CombinationString(bool isInput); // for Japanese, Korean
-	std::wstring CombinationKorean(bool isInput); // for Korea
+	std::u16string CombinationString(bool isInput); // for Japanese, Korean
+	std::u16string CombinationKorean(bool isInput); // for Korea
 	void RemoveKorean(); // for Korean character removal

 	u32 FieldMaxLength();
@@ -239,11 +246,15 @@ private:
 	std::string oskOuttext;

 	int selectedChar;
-	std::wstring inputChars;
+	std::u16string inputChars;
 	OskKeyboardDisplay currentKeyboard;
 	OskKeyboardLanguage currentKeyboardLanguage;
 	bool isCombinated;

+	std::mutex nativeMutex_;
+	PSPOskNativeStatus nativeStatus_ = PSPOskNativeStatus::IDLE;
+	std::string nativeValue_;
+
 	int i_level; // for Korean Keyboard support
 	int i_value[3]; // for Korean Keyboard support
 };
@@ -534,9 +534,8 @@ void PSPSaveDialog::DisplayMessage(std::string text, bool hasYesNo)
 {
 	const float WRAP_WIDTH = 254.0f;
 	float y = 136.0f, h;
-	int n;
-	PPGeMeasureText(0, &h, &n, text.c_str(), FONT_SCALE, PPGE_LINE_WRAP_WORD, WRAP_WIDTH);
-	float h2 = h * (float)n / 2.0f;
+	PPGeMeasureText(nullptr, &h, text.c_str(), FONT_SCALE, PPGE_LINE_WRAP_WORD, WRAP_WIDTH);
+	float h2 = h / 2.0f;
 	if (hasYesNo)
 	{
 		auto di = GetI18NCategory("Dialog");
@@ -555,7 +554,7 @@ void PSPSaveDialog::DisplayMessage(std::string text, bool hasYesNo)
 			yesColor = 0xFFFFFFFF;
 			noColor  = 0xFFFFFFFF;
 		}
-		PPGeMeasureText(&w, &h, 0, choiceText, FONT_SCALE);
+		PPGeMeasureText(&w, &h, choiceText, FONT_SCALE);
 		w = w / 2.0f + 5.5f;
 		h /= 2.0f;
 		float y2 = y + h2 + 4.0f;
@@ -610,13 +609,13 @@ int PSPSaveDialog::Update(int animSpeed)
 	UpdateButtons();
 	UpdateFade(animSpeed);

-	okButtonImg = I_CIRCLE;
-	cancelButtonImg = I_CROSS;
+	okButtonImg = ImageID("I_CIRCLE");
+	cancelButtonImg = ImageID("I_CROSS");
 	okButtonFlag = CTRL_CIRCLE;
 	cancelButtonFlag = CTRL_CROSS;
 	if (param.GetPspParam()->common.buttonSwap == 1) {
-		okButtonImg = I_CROSS;
-		cancelButtonImg = I_CIRCLE;
+		okButtonImg = ImageID("I_CROSS");
+		cancelButtonImg = ImageID("I_CIRCLE");
 		okButtonFlag = CTRL_CROSS;
 		cancelButtonFlag = CTRL_CIRCLE;
 	}
@@ -61,7 +61,7 @@ namespace

 	bool ReadPSPFile(std::string filename, u8 **data, s64 dataSize, s64 *readSize)
 	{
-		u32 handle = pspFileSystem.OpenFile(filename, FILEACCESS_READ);
+		int handle = pspFileSystem.OpenFile(filename, FILEACCESS_READ);
 		if (handle < 0)
 			return false;

@@ -81,7 +81,7 @@ namespace

 	bool WritePSPFile(std::string filename, u8 *data, SceSize dataSize)
 	{
-		u32 handle = pspFileSystem.OpenFile(filename, (FileAccess)(FILEACCESS_WRITE | FILEACCESS_CREATE | FILEACCESS_TRUNCATE));
+		int handle = pspFileSystem.OpenFile(filename, (FileAccess)(FILEACCESS_WRITE | FILEACCESS_CREATE | FILEACCESS_TRUNCATE));
 		if (handle < 0)
 			return false;

@@ -21,6 +21,7 @@
 #include <mutex>
 #include <cstring>

+#include "ppsspp_config.h"
 #include "file/file_util.h"
 #include "file/free.h"
 #include "util/text/utf8.h"
@@ -29,6 +30,11 @@
 #include "Core/FileLoaders/DiskCachingFileLoader.h"
 #include "Core/System.h"

+#if PPSSPP_PLATFORM(SWITCH)
+// Far from optimal, but I guess it works...
+#define fseeko fseek
+#endif
+
 static const char *CACHEFILE_MAGIC = "ppssppDC";
 static const s64 SAFETY_FREE_DISK_SPACE = 768 * 1024 * 1024; // 768 MB
 // Aim to allow this many files cached at once.
@@ -117,7 +117,12 @@ std::string LocalFileLoader::Path() const {
 }

 size_t LocalFileLoader::ReadAt(s64 absolutePos, size_t bytes, size_t count, void *data, Flags flags) {
-#if PPSSPP_PLATFORM(ANDROID)
+#if PPSSPP_PLATFORM(SWITCH)
+	// Toolchain has no fancy IO API.  We must lock.
+	std::lock_guard<std::mutex> guard(readLock_);
+	lseek(fd_, absolutePos, SEEK_SET);
+	return read(fd_, data, bytes * count) / bytes;
+#elif PPSSPP_PLATFORM(ANDROID)
 	// pread64 doesn't appear to actually be 64-bit safe, though such ISOs are uncommon.  See #10862.
 	if (absolutePos <= 0x7FFFFFFF) {
 #if defined(_FILE_OFFSET_BITS) && _FILE_OFFSET_BITS < 64
@@ -78,7 +78,7 @@ size_t RamCachingFileLoader::ReadAt(s64 absolutePos, size_t bytes, void *data, F
 			size_t bytesFromCache = ReadFromCache(absolutePos + readSize, bytes - readSize, (u8 *)data + readSize);
 			readSize += bytesFromCache;
 			if (bytesFromCache == 0) {
-			// We can't read any more.
+				// We can't read any more.
 				break;
 			}
 		}
@@ -197,7 +197,8 @@ CISOFileBlockDevice::CISOFileBlockDevice(FileLoader *fileLoader)
 	u64 lastIndexPos = index[indexSize - 1] & 0x7FFFFFFF;
 	u64 expectedFileSize = lastIndexPos << indexShift;
 	if (expectedFileSize > fileSize) {
-		ERROR_LOG(LOADER, "Expected CSO to at least be %lld bytes, but file is %lld bytes", expectedFileSize, fileSize);
+		ERROR_LOG(LOADER, "Expected CSO to at least be %lld bytes, but file is %lld bytes. File: '%s'",
+			expectedFileSize, fileSize, fileLoader->Path().c_str());
 		NotifyReadError();
 	}
 }
@@ -442,7 +442,8 @@ void DirectoryFileHandle::Close()
 		if (SetEndOfFile(hFile) == 0) {
 			ERROR_LOG_REPORT(FILESYS, "Failed to truncate file.");
 		}
-#else
+#elif !PPSSPP_PLATFORM(SWITCH)
+		// Note: it's not great that Switch cannot truncate appropriately...
 		if (ftruncate(hFile, (off_t)needsTrunc_) != 0) {
 			ERROR_LOG_REPORT(FILESYS, "Failed to truncate file.");
 		}
@@ -794,10 +795,17 @@ static std::string SimulateVFATBug(std::string filename) {
 	// These are the characters allowed in DOS filenames.
 	static const char *FAT_UPPER_CHARS = "ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789!@#$%^&'(){}-_`~";
 	static const char *FAT_LOWER_CHARS = "abcdefghijklmnopqrstuvwxyz0123456789!@#$%^&'(){}-_`~";
+	static const char *LOWER_CHARS = "abcdefghijklmnopqrstuvwxyz";
+
+	// To avoid logging/comparing, skip all this if it has no lowercase chars to begin with.
+	size_t lowerchar = filename.find_first_of(LOWER_CHARS);
+	if (lowerchar == filename.npos) {
+		return filename;
+	}

 	bool apply_hack = false;
 	size_t dot_pos = filename.find('.');
-	if (dot_pos == filename.npos) {
+	if (dot_pos == filename.npos && filename.length() <= 8) {
 		size_t badchar = filename.find_first_not_of(FAT_LOWER_CHARS);
 		if (badchar == filename.npos) {
 			// It's all lowercase.  Convert to upper.
@@ -824,6 +832,7 @@ static std::string SimulateVFATBug(std::string filename) {
 	}

 	if (apply_hack) {
+		VERBOSE_LOG(FILESYS, "Applying VFAT hack to filename: %s", filename.c_str());
 		// In this situation, NT would write UPPERCASE, and just set a flag to say "actually lowercase".
 		// That VFAT flag isn't read by the PSP firmware, so let's pretend to "not read it."
 		std::transform(filename.begin(), filename.end(), filename.begin(), toupper);
@@ -59,6 +59,8 @@ enum
 	HLE_AFTER_DEBUG_BREAK       = 0x20,
 	// Don't fill temp regs with 0xDEADBEEF.
 	HLE_AFTER_SKIP_DEADBEEF     = 0x40,
+	// Execute pending mips calls.
+	HLE_AFTER_QUEUED_CALLS      = 0x80,
 };

 static std::vector<HLEModule> moduleDB;
@@ -68,6 +70,33 @@ static const char *hleAfterSyscallReschedReason;
 static const HLEFunction *latestSyscall = nullptr;
 static int idleOp;

+struct HLEMipsCallInfo {
+	u32 func;
+	PSPAction *action;
+	std::vector<u32> args;
+};
+
+struct HLEMipsCallStack {
+	u32_le nextOff;
+	union {
+		struct {
+			u32_le func;
+			u32_le actionIndex;
+			u32_le argc;
+		};
+		struct {
+			u32_le ra;
+			u32_le v0;
+			u32_le v1;
+		};
+	};
+};
+
+// No need to save state, always flushed at a syscall end.
+static std::vector<HLEMipsCallInfo> enqueuedMipsCalls;
+// Does need to be saved, referenced by the stack and owned.
+static std::vector<PSPAction *> mipsCallActions;
+
 void hleDelayResultFinish(u64 userdata, int cycleslate)
 {
 	u32 error;
@@ -86,16 +115,14 @@ void hleDelayResultFinish(u64 userdata, int cycleslate)
 		WARN_LOG(HLE, "Someone else woke up HLE-blocked thread?");
 }

-void HLEInit()
-{
+void HLEInit() {
 	RegisterAllModules();
 	delayedResultEvent = CoreTiming::RegisterEvent("HLEDelayedResult", hleDelayResultFinish);
 	idleOp = GetSyscallOp("FakeSysCalls", NID_IDLE);
 }

-void HLEDoState(PointerWrap &p)
-{
-	auto s = p.Section("HLE", 1);
+void HLEDoState(PointerWrap &p) {
+	auto s = p.Section("HLE", 1, 2);
 	if (!s)
 		return;

@@ -103,13 +130,35 @@ void HLEDoState(PointerWrap &p)
 	latestSyscall = nullptr;
 	p.Do(delayedResultEvent);
 	CoreTiming::RestoreRegisterEvent(delayedResultEvent, "HLEDelayedResult", hleDelayResultFinish);
+
+	if (s >= 2) {
+		int actions = (int)mipsCallActions.size();
+		p.Do(actions);
+		if (actions != (int)mipsCallActions.size()) {
+			mipsCallActions.resize(actions);
+		}
+
+		for (auto &action : mipsCallActions) {
+			int actionTypeID = action != nullptr ? action->actionTypeID : -1;
+			p.Do(actionTypeID);
+			if (actionTypeID != -1) {
+				if (p.mode == p.MODE_READ)
+					action = __KernelCreateAction(actionTypeID);
+				action->DoState(p);
+			}
+		}
+	}
 }

-void HLEShutdown()
-{
+void HLEShutdown() {
 	hleAfterSyscall = HLE_AFTER_NOTHING;
 	latestSyscall = nullptr;
 	moduleDB.clear();
+	enqueuedMipsCalls.clear();
+	for (auto p : mipsCallActions) {
+		delete p;
+	}
+	mipsCallActions.clear();
 }

 void RegisterModule(const char *name, int numFunctions, const HLEFunction *funcTable)
@@ -354,6 +403,137 @@ bool hleIsKernelMode() {
 	return latestSyscall && (latestSyscall->flags & HLE_KERNEL_SYSCALL) != 0;
 }

+void hleEnqueueCall(u32 func, int argc, const u32 *argv, PSPAction *afterAction) {
+	std::vector<u32> args;
+	args.resize(argc);
+	memcpy(args.data(), argv, argc * sizeof(u32));
+
+	enqueuedMipsCalls.push_back({ func, afterAction, args });
+
+	hleAfterSyscall |= HLE_AFTER_QUEUED_CALLS;
+}
+
+void hleFlushCalls() {
+	u32 &sp = currentMIPS->r[MIPS_REG_SP];
+	PSPPointer<HLEMipsCallStack> stackData;
+	VERBOSE_LOG(HLE, "Flushing %d HLE mips calls from %s, sp=%08x", (int)enqueuedMipsCalls.size(), latestSyscall ? latestSyscall->name : "?", sp);
+
+	// First, we'll add a marker for the final return.
+	sp -= sizeof(HLEMipsCallStack);
+	stackData.ptr = sp;
+	stackData->nextOff = 0xFFFFFFFF;
+	stackData->ra = currentMIPS->pc;
+	stackData->v0 = currentMIPS->r[MIPS_REG_V0];
+	stackData->v1 = currentMIPS->r[MIPS_REG_V1];
+
+	// Now we'll set up the first in the chain.
+	currentMIPS->pc = enqueuedMipsCalls[0].func;
+	currentMIPS->r[MIPS_REG_RA] = HLEMipsCallReturnAddress();
+	for (int i = 0; i < (int)enqueuedMipsCalls[0].args.size(); i++) {
+		currentMIPS->r[MIPS_REG_A0 + i] = enqueuedMipsCalls[0].args[i];
+	}
+
+	// For stack info, process the first enqueued call last, so we run it first.
+	// We don't actually need to store 0's args, but keep it consistent.
+	for (int i = (int)enqueuedMipsCalls.size() - 1; i >= 0; --i) {
+		auto &info = enqueuedMipsCalls[i];
+		u32 stackRequired = (int)info.args.size() * sizeof(u32) + sizeof(HLEMipsCallStack);
+		u32 stackAligned = (stackRequired + 0xF) & ~0xF;
+
+		sp -= stackAligned;
+		stackData.ptr = sp;
+		stackData->nextOff = stackAligned;
+		stackData->func = info.func;
+		if (info.action) {
+			stackData->actionIndex = (int)mipsCallActions.size();
+			mipsCallActions.push_back(info.action);
+		} else {
+			stackData->actionIndex = 0xFFFFFFFF;
+		}
+		stackData->argc = (int)info.args.size();
+		for (int j = 0; j < (int)info.args.size(); ++j) {
+			Memory::Write_U32(info.args[j], sp + sizeof(HLEMipsCallStack) + j * sizeof(u32));
+		}
+	}
+	enqueuedMipsCalls.clear();
+
+	DEBUG_LOG(HLE, "Executing HLE mips call at %08x, sp=%08x", currentMIPS->pc, sp);
+}
+
+void HLEReturnFromMipsCall() {
+	u32 &sp = currentMIPS->r[MIPS_REG_SP];
+	PSPPointer<HLEMipsCallStack> stackData;
+
+	// At this point, we may have another mips call to run, or be at the end...
+	stackData.ptr = sp;
+
+	if ((stackData->nextOff & 0x0000000F) != 0 || !Memory::IsValidAddress(sp + stackData->nextOff)) {
+		ERROR_LOG(HLE, "Corrupt stack on HLE mips call return: %08x", stackData->nextOff);
+		Core_UpdateState(CORE_ERROR);
+		return;
+	}
+
+	if (stackData->actionIndex != 0xFFFFFFFF && stackData->actionIndex < (u32)mipsCallActions.size()) {
+		PSPAction *&action = mipsCallActions[stackData->actionIndex];
+		VERBOSE_LOG(HLE, "Executing action for HLE mips call at %08x, sp=%08x", stackData->func, sp);
+
+		// Search for the saved v0/v1 values, to preserve the PSPAction API...
+		PSPPointer<HLEMipsCallStack> finalMarker = stackData;
+		while ((finalMarker->nextOff & 0x0000000F) == 0 && Memory::IsValidAddress(finalMarker.ptr + finalMarker->nextOff)) {
+			finalMarker.ptr += finalMarker->nextOff;
+		}
+		if (finalMarker->nextOff != 0xFFFFFFFF) {
+			ERROR_LOG(HLE, "Corrupt stack on HLE mips call return action: %08x", finalMarker->nextOff);
+			Core_UpdateState(CORE_ERROR);
+			return;
+		}
+
+		MipsCall mc;
+		mc.savedV0 = finalMarker->v0;
+		mc.savedV1 = finalMarker->v1;
+		action->run(mc);
+		finalMarker->v0 = mc.savedV0;
+		finalMarker->v1 = mc.savedV1;
+
+		delete action;
+		action = nullptr;
+
+		// Note: the action could actually enqueue more, adding another layer on stack after this.
+	}
+
+	sp += stackData->nextOff;
+	stackData.ptr = sp;
+
+	if (stackData->nextOff == 0xFFFFFFFF) {
+		// We're done.  Grab the HLE result's v0/v1 and return from the syscall.
+		currentMIPS->pc = stackData->ra;
+		currentMIPS->r[MIPS_REG_V0] = stackData->v0;
+		currentMIPS->r[MIPS_REG_V1] = stackData->v1;
+
+		sp += sizeof(HLEMipsCallStack);
+
+		bool canClear = true;
+		for (auto p : mipsCallActions) {
+			canClear = canClear && p == nullptr;
+		}
+		if (canClear) {
+			mipsCallActions.clear();
+		}
+
+		VERBOSE_LOG(HLE, "Finished HLE mips calls, v0=%08x, sp=%08x", currentMIPS->r[MIPS_REG_V0], sp);
+		return;
+	}
+
+	// Alright, we have another to call.
+	hleSkipDeadbeef();
+	currentMIPS->pc = stackData->func;
+	currentMIPS->r[MIPS_REG_RA] = HLEMipsCallReturnAddress();
+	for (int i = 0; i < (int)stackData->argc; i++) {
+		currentMIPS->r[MIPS_REG_A0 + i] = Memory::Read_U32(sp + sizeof(HLEMipsCallStack) + i * sizeof(u32));
+	}
+	DEBUG_LOG(HLE, "Executing next HLE mips call at %08x, sp=%08x", currentMIPS->pc, sp);
+}
+
 const static u32 deadbeefRegs[12] = {0xDEADBEEF, 0xDEADBEEF, 0xDEADBEEF, 0xDEADBEEF, 0xDEADBEEF, 0xDEADBEEF, 0xDEADBEEF, 0xDEADBEEF, 0xDEADBEEF, 0xDEADBEEF, 0xDEADBEEF, 0xDEADBEEF};
 inline static void SetDeadbeefRegs()
 {
@@ -375,6 +555,8 @@ inline void hleFinishSyscall(const HLEFunction &info)
 	if ((hleAfterSyscall & HLE_AFTER_SKIP_DEADBEEF) == 0)
 		SetDeadbeefRegs();

+	if ((hleAfterSyscall & HLE_AFTER_QUEUED_CALLS) != 0)
+		hleFlushCalls();
 	if ((hleAfterSyscall & HLE_AFTER_CURRENT_CALLBACKS) != 0 && (hleAfterSyscall & HLE_AFTER_RESCHED_CALLBACKS) == 0)
 		__KernelForceCallbacks();

@@ -661,6 +843,7 @@ void hleDoLogInternal(LogTypes::LOG_TYPE t, LogTypes::LOG_LEVELS level, u64 res,
 	const char *funcName = "?";
 	u32 funcFlags = 0;
 	if (latestSyscall) {
+		_dbg_assert_(HLE, latestSyscall->argmask != nullptr);
 		hleFormatLogArgs(formatted_args, sizeof(formatted_args), latestSyscall->argmask);

 		// This acts as an override (for error returns which are usually hex.)
@@ -24,6 +24,7 @@
 #include "Core/MIPS/MIPS.h"

 class PointerWrap;
+class PSPAction;
 typedef void (* HLEFunc)();

 enum {
@@ -116,6 +117,8 @@ void hleSkipDeadbeef();
 void hleSetSteppingTime(double t);
 // Check if the current syscall context is kernel.
 bool hleIsKernelMode();
+// Enqueue a MIPS function to be called after this HLE call finishes.
+void hleEnqueueCall(u32 func, int argc, const u32 *argv, PSPAction *afterAction = nullptr);

 // Delays the result for usec microseconds, allowing other threads to run during this time.
 u32 hleDelayResult(u32 result, const char *reason, int usec);
@@ -144,6 +147,8 @@ void CallSyscall(MIPSOpcode op);
 void WriteFuncStub(u32 stubAddr, u32 symAddr);
 void WriteFuncMissingStub(u32 stubAddr, u32 nid);

+void HLEReturnFromMipsCall();
+
 const HLEFunction *GetSyscallFuncPointer(MIPSOpcode op);
 // For jit, takes arg: const HLEFunction *
 void *GetQuickSyscallFunc(MIPSOpcode op);
@@ -89,13 +89,14 @@
 //sound
 //zlibdec
 const HLEFunction FakeSysCalls[] = {
-	{NID_THREADRETURN, __KernelReturnFromThread, "__KernelReturnFromThread"},
-	{NID_CALLBACKRETURN, __KernelReturnFromMipsCall, "__KernelReturnFromMipsCall"},
-	{NID_INTERRUPTRETURN, __KernelReturnFromInterrupt, "__KernelReturnFromInterrupt"},
-	{NID_EXTENDRETURN, __KernelReturnFromExtendStack, "__KernelReturnFromExtendStack"},
-	{NID_MODULERETURN, __KernelReturnFromModuleFunc, "__KernelReturnFromModuleFunc"},
-	{NID_IDLE, __KernelIdle, "_sceKernelIdle"},
-	{NID_GPUREPLAY, __KernelGPUReplay, "__KernelGPUReplay"},
+	{NID_THREADRETURN, __KernelReturnFromThread, "__KernelReturnFromThread", 'x', ""},
+	{NID_CALLBACKRETURN, __KernelReturnFromMipsCall, "__KernelReturnFromMipsCall", 'x', ""},
+	{NID_INTERRUPTRETURN, __KernelReturnFromInterrupt, "__KernelReturnFromInterrupt", 'x', ""},
+	{NID_EXTENDRETURN, __KernelReturnFromExtendStack, "__KernelReturnFromExtendStack", 'x', ""},
+	{NID_MODULERETURN, __KernelReturnFromModuleFunc, "__KernelReturnFromModuleFunc", 'x', ""},
+	{NID_IDLE, __KernelIdle, "_sceKernelIdle", 'x', ""},
+	{NID_GPUREPLAY, __KernelGPUReplay, "__KernelGPUReplay", 'x', ""},
+	{NID_HLECALLRETURN, HLEReturnFromMipsCall, "HLEReturnFromMipsCall", 'x', ""},
 };

 const HLEFunction UtilsForUser[] = 
@@ -22,6 +22,7 @@
 #define NID_INTERRUPTRETURN 0xbadd00d5
 #define NID_EXTENDRETURN 0xbad0b0c9
 #define NID_MODULERETURN 0xbad0d318
+#define NID_HLECALLRETURN 0xbad0259b
 #define NID_IDLE 0x1d7e1d7e
 #define NID_GPUREPLAY 0x9e45bd95

@@ -21,7 +21,7 @@

 class KernelThreadDebugInterface : public MIPSDebugInterface {
 public:
-	KernelThreadDebugInterface(MIPSState *c, ThreadContext &t) : MIPSDebugInterface(c), ctx(t) {
+	KernelThreadDebugInterface(MIPSState *c, PSPThreadContext &t) : MIPSDebugInterface(c), ctx(t) {
 	}

 	unsigned int getPC() override { return ctx.pc; }
@@ -86,5 +86,5 @@ public:
 	}

 protected:
-	ThreadContext &ctx;
+	PSPThreadContext &ctx;
 };
@@ -63,6 +63,13 @@ std::recursive_mutex peerlock;
 SceNetAdhocPdpStat * pdp[255];
 SceNetAdhocPtpStat * ptp[255];
 uint32_t localip;
+std::vector<std::string> chatLog;
+std::string name = "";
+std::string incoming = "";
+std::string message = "";
+bool chatScreenVisible = false;
+bool updateChatScreen = false;
+int newChat = 0;

 int isLocalMAC(const SceNetEtherAddr * addr) {
 	SceNetEtherAddr saddr;
@@ -1011,6 +1018,45 @@ void freeFriendsRecursive(SceNetAdhocctlPeerInfo * node) {
 	free(node);
 }

+void sendChat(std::string chatString) {
+	SceNetAdhocctlChatPacketC2S chat;
+	auto n = GetI18NCategory("Networking");
+	chat.base.opcode = OPCODE_CHAT;
+	//TODO check network inited, check send success or not, chatlog.pushback error on failed send, pushback error on not connected
+	if (friendFinderRunning)
+	{
+		// Send Chat to Server 
+		if (!chatString.empty()) {
+			//maximum char allowed is 64 character for compability with original server (pro.coldbird.net)
+			message = chatString.substr(0, 60); // 64 return chat variable corrupted is it out of memory?
+			strcpy(chat.message, message.c_str());
+			//Send Chat Messages
+			int chatResult = send(metasocket, (const char *)&chat, sizeof(chat), 0);
+			NOTICE_LOG(SCENET, "Send Chat %s to Adhoc Server", chat.message);
+			name = g_Config.sNickName.c_str();
+			chatLog.push_back(name.substr(0, 8) + ": " + chat.message);
+			if (chatScreenVisible) {
+				updateChatScreen = true;
+			}
+		}
+	}
+	else {
+		chatLog.push_back(n->T("You're in Offline Mode, go to lobby or online hall"));
+		if (chatScreenVisible) {
+			updateChatScreen = true;
+		}
+	}
+}
+
+std::vector<std::string> getChatLog() {
+	// this log used by chat screen
+	if (chatLog.size() > 50) {
+		//erase the first 40 element limit the chatlog size
+		chatLog.erase(chatLog.begin(), chatLog.begin() + 40);
+	}
+	return chatLog;
+}
+
 int friendFinder(){
 	// Receive Buffer
 	int rxpos = 0;
@@ -1054,13 +1100,6 @@ int friendFinder(){
 			}*/
 		}

-		// Send Chat Messages
-		//while(popFromOutbox(chat.message))
-		//{
-		//  // Send Chat to Server
-		//  sceNetInetSend(metasocket, (const char *)&chat, sizeof(chat), 0);
-		//}
-
 		// Wait for Incoming Data
 		int received = recv(metasocket, (char *)(rx + rxpos), sizeof(rx) - rxpos, 0);

@@ -1110,14 +1149,23 @@ int friendFinder(){
 				if (rxpos >= (int)sizeof(SceNetAdhocctlChatPacketS2C)) {
 					// Cast Packet
 					SceNetAdhocctlChatPacketS2C * packet = (SceNetAdhocctlChatPacketS2C *)rx;
-
-					// Fix for Idiots that try to troll the "ME" Nametag
-					if (strcasecmp((char *)packet->name.data, "ME") == 0) strcpy((char *)packet->name.data, "NOT ME");
-
 					// Add Incoming Chat to HUD
-					//printf("Receive chat message %s", packet->base.message);
-					DEBUG_LOG(SCENET, "Received chat message %s", packet->base.message);
-
+					NOTICE_LOG(SCENET, "Received chat message %s", packet->base.message);
+					incoming = "";
+					name = (char *)packet->name.data;
+					incoming.append(name.substr(0, 8));
+					incoming.append(": ");
+					incoming.append((char *)packet->base.message);
+					chatLog.push_back(incoming);
+					//im new to pointer btw :( doesn't know its safe or not this should update the chat screen when data coming
+					if (chatScreenVisible) {
+						updateChatScreen = true;
+					}
+					else {
+						if (newChat < 50) {
+							newChat += 1;
+						}
+					}
 					// Move RX Buffer
 					memmove(rx, rx + sizeof(SceNetAdhocctlChatPacketS2C), sizeof(rx) - sizeof(SceNetAdhocctlChatPacketS2C));

@@ -1139,7 +1187,16 @@ int friendFinder(){

 					// Add User
 					addFriend(packet);
-
+					incoming = "";
+					incoming.append((char *)packet->name.data);
+					incoming.append(" Joined ");
+					//do we need ip?
+					//joined.append((char *)packet->ip);
+					chatLog.push_back(incoming);
+					//im new to pointer btw :( doesn't know its safe or not this should update the chat screen when data coming
+					if (chatScreenVisible) {
+						updateChatScreen = true;
+					}
 					// Update HUD User Count
 #ifdef LOCALHOST_AS_PEER
 					setUserCount(getActivePeerCount());
@@ -764,10 +764,10 @@ typedef struct {
 #pragma pack(pop)
 #endif

-class AfterMatchingMipsCall : public Action {
+class AfterMatchingMipsCall : public PSPAction {
 public:
 	AfterMatchingMipsCall() {}
-	static Action *Create() { return new AfterMatchingMipsCall(); }
+	static PSPAction *Create() { return new AfterMatchingMipsCall(); }
 	void DoState(PointerWrap &p) override {
 		auto s = p.Section("AfterMatchingMipsCall", 1, 2);
 		if (!s)
@@ -851,6 +851,16 @@ SceNetAdhocMatchingMemberInternal* addMember(SceNetAdhocMatchingContext * contex
 */
 void addFriend(SceNetAdhocctlConnectPacketS2C * packet);

+/**
+* Send chat or get that
+* @param std::string ChatString 
+*/
+void sendChat(std::string chatString);
+std::vector<std::string> getChatLog();
+extern bool chatScreenVisible;
+extern bool updateChatScreen;
+extern int newChat;
+
 /*
 * Find a Peer/Friend by MAC address
 */
@@ -28,9 +28,9 @@
 #include "Core/Reporting.h"

 typedef PSPPointer<char> PSPCharPointer;
-typedef PSPPointer<u16> PSPWCharPointer;
+typedef PSPPointer<char16_t> PSPWCharPointer;
 typedef PSPPointer<const char> PSPConstCharPointer;
-typedef PSPPointer<const u16> PSPConstWCharPointer;
+typedef PSPPointer<const char16_t> PSPConstWCharPointer;

 static u16 errorUTF8;
 static u16 errorUTF16;
@@ -49,6 +49,7 @@
 #include "Core/HLE/sceKernel.h"
 #include "Core/HLE/sceKernelThread.h"
 #include "Core/HLE/sceKernelInterrupt.h"
+#include "Core/Util/PPGeDraw.h"

 #include "GPU/GPU.h"
 #include "GPU/GPUState.h"
@@ -549,6 +550,9 @@ static int CalculateFrameSkip() {
 static void DoFrameTiming(bool &throttle, bool &skipFrame, float timestep) {
 	PROFILE_THIS_SCOPE("timing");
 	FPSLimit fpsLimiter = PSP_CoreParameter().fpsLimit;
+	int fpsLimit = 60;
+	if (fpsLimiter != FPSLimit::NORMAL)
+		fpsLimit = fpsLimiter == FPSLimit::CUSTOM1 ? g_Config.iFpsLimit1 : g_Config.iFpsLimit2;
 	throttle = FrameTimingThrottled();
 	skipFrame = false;

@@ -556,7 +560,16 @@ static void DoFrameTiming(bool &throttle, bool &skipFrame, float timestep) {
 	// we have nothing to do here.
 	bool doFrameSkip = g_Config.iFrameSkip != 0;

-	if (!throttle && g_Config.bFrameSkipUnthrottle) {
+	bool unthrottleNeedsSkip = g_Config.bFrameSkipUnthrottle;
+	if (g_Config.bVSync && GetGPUBackend() == GPUBackend::VULKAN) {
+		// Vulkan doesn't support the interval setting, so we force frameskip.
+		unthrottleNeedsSkip = true;
+		// If it's not a clean multiple of 60, we may need frameskip to achieve it.
+		if (fpsLimit == 0 || (fpsLimit >= 0 && fpsLimit != 15 && fpsLimit != 30 && fpsLimit != 60)) {
+			doFrameSkip = true;
+		}
+	}
+	if (!throttle && unthrottleNeedsSkip) {
 		doFrameSkip = true;
 		skipFrame = true;
 		if (numSkippedFrames >= 7) {
@@ -571,10 +584,8 @@ static void DoFrameTiming(bool &throttle, bool &skipFrame, float timestep) {
 	time_update();

 	float scaledTimestep = timestep;
-	if (fpsLimiter == FPSLimit::CUSTOM1 && g_Config.iFpsLimit1 > 0) {
-		scaledTimestep *= 60.0f / g_Config.iFpsLimit1;
-	} else if (fpsLimiter == FPSLimit::CUSTOM2 && g_Config.iFpsLimit2 > 0) {
-		scaledTimestep *= 60.0f / g_Config.iFpsLimit2;
+	if (fpsLimit > 0 && fpsLimit != 60) {
+		scaledTimestep *= 60.0f / fpsLimit;
 	}

 	if (lastFrameTime == 0.0 || wasPaused) {
@@ -594,13 +605,16 @@ static void DoFrameTiming(bool &throttle, bool &skipFrame, float timestep) {

 	// Auto-frameskip automatically if speed limit is set differently than the default.
 	bool useAutoFrameskip = g_Config.bAutoFrameSkip && g_Config.iRenderingMode != FB_NON_BUFFERED_MODE;
-	bool forceFrameskip = (fpsLimiter == FPSLimit::CUSTOM1 && g_Config.iFpsLimit1 > 60) || (fpsLimiter == FPSLimit::CUSTOM2 && g_Config.iFpsLimit2 > 60);
+	bool forceFrameskip = fpsLimit > 60 && unthrottleNeedsSkip;
 	int frameSkipNum = CalculateFrameSkip();
 	if (g_Config.bAutoFrameSkip || forceFrameskip) {
 		// autoframeskip
 		// Argh, we are falling behind! Let's skip a frame and see if we catch up.
 		if (curFrameTime > nextFrameTime && doFrameSkip) {
 			skipFrame = true;
+			if (forceFrameskip) {
+				throttle = false;
+			}
 		}
 	} else if (frameSkipNum >= 1) {
 		// fixed frameskip
@@ -740,9 +754,17 @@ void __DisplayFlip(int cyclesLate) {
 	// Also let's always flip for animated shaders.
 	const ShaderInfo *shaderInfo = g_Config.sPostShaderName == "Off" ? nullptr : GetPostShaderInfo(g_Config.sPostShaderName);
 	bool postEffectRequiresFlip = false;
-	if (shaderInfo && g_Config.iRenderingMode != FB_NON_BUFFERED_MODE)
-		postEffectRequiresFlip = shaderInfo->requires60fps;
+	// postEffectRequiresFlip is not compatible with frameskip unthrottling, see #12325.
+	if (g_Config.iRenderingMode != FB_NON_BUFFERED_MODE && !(g_Config.bFrameSkipUnthrottle && !FrameTimingThrottled())) {
+		if (shaderInfo) {
+			postEffectRequiresFlip = (shaderInfo->requires60fps || g_Config.bRenderDuplicateFrames);
+		} else {
+			postEffectRequiresFlip = g_Config.bRenderDuplicateFrames;
+		}
+	}
+
 	const bool fbDirty = gpu->FramebufferDirty();
+
 	if (fbDirty || noRecentFlip || postEffectRequiresFlip) {
 		int frameSleepPos = frameTimeHistoryPos;
 		CalculateFPS();
@@ -768,7 +790,7 @@ void __DisplayFlip(int cyclesLate) {
 			// Check first though, might've just quit / been paused.
 			if (coreState == CORE_RUNNING) {
 				coreState = CORE_NEXTFRAME;
-				gpu->CopyDisplayToOutput();
+				gpu->CopyDisplayToOutput(fbReallyDirty);
 				if (fbReallyDirty) {
 					actualFlips++;
 				}
@@ -819,6 +841,7 @@ void __DisplayFlip(int cyclesLate) {

 void hleAfterFlip(u64 userdata, int cyclesLate) {
 	gpu->BeginFrame();  // doesn't really matter if begin or end of frame.
+	PPGeNotifyFrame();

 	// This seems like as good a time as any to check if the config changed.
 	if (lagSyncScheduled != g_Config.bForceLagSync) {
@@ -929,6 +952,8 @@ void __DisplaySetFramebuf(u32 topaddr, int linesize, int pixelFormat, int sync)
 	if (sync == PSP_DISPLAY_SETBUF_IMMEDIATE) {
 		// Write immediately to the current framebuffer parameters.
 		framebuf = fbstate;
+		// Also update latchedFramebuf for any sceDisplayGetFramebuf() after this.
+		latchedFramebuf = fbstate;
 		gpu->SetDisplayFramebuffer(framebuf.topaddr, framebuf.stride, framebuf.fmt);
 		// IMMEDIATE means that the buffer is fine. We can just flip immediately.
 		// Doing it in non-buffered though creates problems (black screen) on occasion though
@@ -1033,9 +1058,7 @@ bool __DisplayGetFramebuf(PSPPointer<u8> *topaddr, u32 *linesize, u32 *pixelForm
 }

 static u32 sceDisplayGetFramebuf(u32 topaddrPtr, u32 linesizePtr, u32 pixelFormatPtr, int latchedMode) {
-	// NOTE: This is wrong and partially reverts #8753. Presumably there's something else involved here as well.
-	// See #8816. Could also be a firmware version difference, there are a few of those...
-	const FrameBufferState &fbState = (latchedMode == PSP_DISPLAY_SETBUF_NEXTFRAME && framebufIsLatched) ? latchedFramebuf : framebuf;
+	const FrameBufferState &fbState = latchedMode == PSP_DISPLAY_SETBUF_NEXTFRAME ? latchedFramebuf : framebuf;

 	if (Memory::IsValidAddress(topaddrPtr))
 		Memory::Write_U32(fbState.topaddr, topaddrPtr);
@@ -35,9 +35,15 @@ enum {
 	FONT_IS_OPEN   = 1,
 };

+// For the save states.
+static bool useAllocCallbacks = true;
+
 // Actions
 static int actionPostAllocCallback;
 static int actionPostOpenCallback;
+static int actionPostOpenAllocCallback;
+static int actionPostCharInfoAllocCallback;
+static int actionPostCharInfoFreeCallback;

 // Monster Hunter sequence:
 // 36:46:998 c:\dev\ppsspp\core\hle\scefont.cpp:469 E[HLE]: sceFontNewLib 89ad4a0, 9fff5cc
@@ -82,30 +88,31 @@ struct FontRegistryEntry {
 	const char *fontName;
 	int expireDate;
 	int shadow_option;
+	u32 fontFileSize;
 	bool ignoreIfMissing;
 };

 static const FontRegistryEntry fontRegistry[] = {
 	// This was added for Chinese translations and is not normally loaded on a PSP.
-	{ 0x288, 0x288, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SANS_SERIF, FONT_STYLE_DB, 0, FONT_LANGUAGE_CHINESE, 0, 1, "zh_gb.pgf", "FTT-NewRodin Pro DB", 0, 0, true },
-	{ 0x288, 0x288, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SANS_SERIF, FONT_STYLE_DB, 0, FONT_LANGUAGE_JAPANESE, 0, 1, "jpn0.pgf", "FTT-NewRodin Pro DB", 0, 0 },
-	{0x288, 0x288, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SANS_SERIF, FONT_STYLE_REGULAR, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn0.pgf", "FTT-NewRodin Pro Latin", 0, 0},
-	{0x288, 0x288, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SERIF, FONT_STYLE_REGULAR, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn1.pgf", "FTT-Matisse Pro Latin", 0, 0},
-	{0x288, 0x288, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SANS_SERIF, FONT_STYLE_ITALIC, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn2.pgf", "FTT-NewRodin Pro Latin", 0, 0},
-	{0x288, 0x288, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SERIF, FONT_STYLE_ITALIC, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn3.pgf", "FTT-Matisse Pro Latin", 0, 0},
-	{0x288, 0x288, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SANS_SERIF, FONT_STYLE_BOLD, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn4.pgf", "FTT-NewRodin Pro Latin", 0, 0},
-	{0x288, 0x288, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SERIF, FONT_STYLE_BOLD, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn5.pgf", "FTT-Matisse Pro Latin", 0, 0},
-	{0x288, 0x288, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SANS_SERIF, FONT_STYLE_BOLD_ITALIC, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn6.pgf", "FTT-NewRodin Pro Latin", 0, 0},
-	{0x288, 0x288, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SERIF, FONT_STYLE_BOLD_ITALIC, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn7.pgf", "FTT-Matisse Pro Latin", 0, 0},
-	{0x1c0, 0x1c0, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SANS_SERIF, FONT_STYLE_REGULAR, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn8.pgf", "FTT-NewRodin Pro Latin", 0, 0},
-	{0x1c0, 0x1c0, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SERIF, FONT_STYLE_REGULAR, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn9.pgf", "FTT-Matisse Pro Latin", 0, 0},
-	{0x1c0, 0x1c0, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SANS_SERIF, FONT_STYLE_ITALIC, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn10.pgf", "FTT-NewRodin Pro Latin", 0, 0},
-	{0x1c0, 0x1c0, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SERIF, FONT_STYLE_ITALIC, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn11.pgf", "FTT-Matisse Pro Latin", 0, 0},
-	{0x1c0, 0x1c0, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SANS_SERIF, FONT_STYLE_BOLD, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn12.pgf", "FTT-NewRodin Pro Latin", 0, 0},
-	{0x1c0, 0x1c0, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SERIF, FONT_STYLE_BOLD, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn13.pgf", "FTT-Matisse Pro Latin", 0, 0},
-	{0x1c0, 0x1c0, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SANS_SERIF, FONT_STYLE_BOLD_ITALIC, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn14.pgf", "FTT-NewRodin Pro Latin", 0, 0},
-	{0x1c0, 0x1c0, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SERIF, FONT_STYLE_BOLD_ITALIC, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn15.pgf", "FTT-Matisse Pro Latin", 0, 0},
-	{0x288, 0x288, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SANS_SERIF, FONT_STYLE_REGULAR, 0, FONT_LANGUAGE_KOREAN, 0, 3, "kr0.pgf", "AsiaNHH(512Johab)", 0, 0},
+	{ 0x288, 0x288, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SANS_SERIF, FONT_STYLE_DB, 0, FONT_LANGUAGE_CHINESE, 0, 1, "zh_gb.pgf", "FTT-NewRodin Pro DB", 0, 0, 1581700, true },
+	{ 0x288, 0x288, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SANS_SERIF, FONT_STYLE_DB, 0, FONT_LANGUAGE_JAPANESE, 0, 1, "jpn0.pgf", "FTT-NewRodin Pro DB", 0, 0, 1581700 },
+	{ 0x288, 0x288, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SANS_SERIF, FONT_STYLE_REGULAR, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn0.pgf", "FTT-NewRodin Pro Latin", 0, 0, 69108 },
+	{ 0x288, 0x288, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SERIF, FONT_STYLE_REGULAR, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn1.pgf", "FTT-Matisse Pro Latin", 0, 0, 65124 },
+	{ 0x288, 0x288, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SANS_SERIF, FONT_STYLE_ITALIC, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn2.pgf", "FTT-NewRodin Pro Latin", 0, 0, 72948 },
+	{ 0x288, 0x288, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SERIF, FONT_STYLE_ITALIC, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn3.pgf", "FTT-Matisse Pro Latin", 0, 0, 67700},
+	{ 0x288, 0x288, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SANS_SERIF, FONT_STYLE_BOLD, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn4.pgf", "FTT-NewRodin Pro Latin", 0, 0, 72828 },
+	{ 0x288, 0x288, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SERIF, FONT_STYLE_BOLD, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn5.pgf", "FTT-Matisse Pro Latin", 0, 0, 68220 },
+	{ 0x288, 0x288, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SANS_SERIF, FONT_STYLE_BOLD_ITALIC, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn6.pgf", "FTT-NewRodin Pro Latin", 0, 0, 77032 },
+	{ 0x288, 0x288, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SERIF, FONT_STYLE_BOLD_ITALIC, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn7.pgf", "FTT-Matisse Pro Latin", 0, 0, 71144 },
+	{ 0x1c0, 0x1c0, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SANS_SERIF, FONT_STYLE_REGULAR, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn8.pgf", "FTT-NewRodin Pro Latin", 0, 0, 41000 },
+	{ 0x1c0, 0x1c0, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SERIF, FONT_STYLE_REGULAR, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn9.pgf", "FTT-Matisse Pro Latin", 0, 0, 40164 },
+	{ 0x1c0, 0x1c0, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SANS_SERIF, FONT_STYLE_ITALIC, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn10.pgf", "FTT-NewRodin Pro Latin", 0, 0, 42692 },
+	{ 0x1c0, 0x1c0, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SERIF, FONT_STYLE_ITALIC, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn11.pgf", "FTT-Matisse Pro Latin", 0, 0, 41488 },
+	{ 0x1c0, 0x1c0, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SANS_SERIF, FONT_STYLE_BOLD, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn12.pgf", "FTT-NewRodin Pro Latin", 0, 0, 43136 },
+	{ 0x1c0, 0x1c0, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SERIF, FONT_STYLE_BOLD, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn13.pgf", "FTT-Matisse Pro Latin", 0, 0, 41772 },
+	{ 0x1c0, 0x1c0, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SANS_SERIF, FONT_STYLE_BOLD_ITALIC, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn14.pgf", "FTT-NewRodin Pro Latin", 0, 0, 45184 },
+	{ 0x1c0, 0x1c0, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SERIF, FONT_STYLE_BOLD_ITALIC, 0, FONT_LANGUAGE_LATIN, 0, 1, "ltn15.pgf", "FTT-Matisse Pro Latin", 0, 0, 43044 },
+	{ 0x288, 0x288, 0x2000, 0x2000, 0, 0, FONT_FAMILY_SANS_SERIF, FONT_STYLE_REGULAR, 0, FONT_LANGUAGE_KOREAN, 0, 3, "kr0.pgf", "AsiaNHH(512Johab)", 0, 0, 394192 },
 };

 static const float pointDPI = 72.f;
@@ -204,6 +211,7 @@ public:

 	PGF *GetPGF() { return &pgf_; }
 	const PGF *GetPGF() const { return &pgf_; }
+	u32 getSize() const { return dataSize_; }
 	bool IsValid() const { return valid_; }

 	void DoState(PointerWrap &p) {
@@ -228,6 +236,7 @@ private:
 		style_.fontV = (float)pgf_.header.vSize / 64.0f;
 		style_.fontHRes = (float)pgf_.header.hResolution / 64.0f;
 		style_.fontVRes = (float)pgf_.header.vResolution / 64.0f;
+		this->dataSize_ = (u32)dataSize;
 	}

 	void Init(const u8 *data, size_t dataSize, const FontRegistryEntry &entry) {
@@ -247,11 +256,13 @@ private:
 		strncpy(style_.fontFileName, entry.fileName, sizeof(style_.fontFileName));
 		style_.fontAttributes = entry.extraAttributes;
 		style_.fontExpire = entry.expireDate;
+		this->dataSize_ = entry.fontFileSize;
 	}

 	PGF pgf_;
 	PGFFontStyle style_;
 	bool valid_;
+	u32 dataSize_;
 	DISALLOW_COPY_AND_ASSIGN(Font);
 };

@@ -338,10 +349,10 @@ private:
 	DISALLOW_COPY_AND_ASSIGN(LoadedFont);
 };

-class PostAllocCallback : public Action {
+class PostAllocCallback : public PSPAction {
 public:
 	PostAllocCallback() {}
-	static Action *Create() { return new PostAllocCallback(); }
+	static PSPAction *Create() { return new PostAllocCallback(); }
 	void DoState(PointerWrap &p) override {
 		auto s = p.Section("PostAllocCallback", 1, 2);
 		if (!s)
@@ -360,10 +371,10 @@ private:
 	u32 errorCodePtr_;
 };

-class PostOpenCallback : public Action {
+class PostOpenCallback : public PSPAction {
 public:
 	PostOpenCallback() {}
-	static Action *Create() { return new PostOpenCallback(); }
+	static PSPAction *Create() { return new PostOpenCallback(); }
 	void DoState(PointerWrap &p) override {
 		auto s = p.Section("PostOpenCallback", 1);
 		if (!s)
@@ -378,6 +389,69 @@ private:
 	u32 fontLibID_;
 };

+class PostOpenAllocCallback : public PSPAction {
+public:
+	PostOpenAllocCallback() {}
+	static PSPAction *Create() { return new PostOpenAllocCallback(); }
+	void DoState(PointerWrap &p) override {
+		auto s = p.Section("PostOpenAllocCallback", 1);
+		if (!s)
+			return;
+
+		p.Do(fontLibID_);
+		p.Do(fontHandle_);
+		p.Do(fontIndex_);
+	}
+	void run(MipsCall &call) override;
+	void SetFontLib(u32 fontLibID) { fontLibID_ = fontLibID; }
+	void SetFont(u32 handle, int index) { fontHandle_ = handle; fontIndex_ = index; }
+
+private:
+	u32 fontLibID_;
+	u32 fontHandle_;
+	int fontIndex_;
+};
+
+class PostCharInfoAllocCallback : public PSPAction {
+public:
+	PostCharInfoAllocCallback() {}
+	static PSPAction *Create() { return new PostCharInfoAllocCallback(); }
+	void DoState(PointerWrap &p) override {
+		auto s = p.Section("PostCharInfoAllocCallback", 1);
+		if (!s)
+			return;
+
+		p.Do(fontLibID_);
+	}
+	void run(MipsCall &call) override;
+	void SetFontLib(u32 fontLibID) { fontLibID_ = fontLibID; }
+
+private:
+	u32 fontLibID_;
+};
+
+class PostCharInfoFreeCallback : public PSPAction {
+public:
+	PostCharInfoFreeCallback() {}
+	static PSPAction *Create() { return new PostCharInfoFreeCallback(); }
+	void DoState(PointerWrap &p) override {
+		auto s = p.Section("PostCharInfoFreeCallback", 1);
+		if (!s)
+			return;
+
+		p.Do(fontLibID_);
+		p.Do(charInfo_);
+	}
+	void run(MipsCall &call) override;
+	void SetFontLib(u32 fontLibID) { fontLibID_ = fontLibID; }
+	void SetCharInfo(PSPPointer<PGFCharInfo> charInfo) { charInfo_ = charInfo; }
+
+private:
+	u32 fontLibID_;
+	PSPPointer<PGFCharInfo> charInfo_;
+};
+
+
 struct NativeFontLib {
 	FontNewLibParams params;
 	// TODO
@@ -406,7 +480,7 @@ public:
 		// For save states only.
 	}

-	FontLib(u32 paramPtr, u32 errorCodePtr) : fontHRes_(128.0f), fontVRes_(128.0f), altCharCode_(0x5F) {
+	FontLib(u32 paramPtr, u32 errorCodePtr) : fontHRes_(128.0f), fontVRes_(128.0f), altCharCode_(0x5F), charInfoBitmapAddress_(0) {
 		nfl_ = 0;
 		Memory::ReadStruct(paramPtr, &params_);
 		if (params_.numFonts > 9) {
@@ -418,8 +492,8 @@ public:
 		PostAllocCallback *action = (PostAllocCallback *) __KernelCreateAction(actionPostAllocCallback);
 		action->SetFontLib(GetListID(), errorCodePtr);

-		u32 args[2] = { params_.userDataAddr, allocSize };
-		__KernelDirectMipsCall(params_.allocFuncAddr, action, args, 2, true);
+		u32 args[2] = { userDataAddr(), allocSize };
+		hleEnqueueCall(allocFuncAddr(), 2, args, action);
 	}

 	u32 GetListID() {
@@ -429,25 +503,29 @@ public:
 	void Done() {
 		for (size_t i = 0; i < fonts_.size(); i++) {
 			if (isfontopen_[i] == FONT_IS_OPEN) {
-				fontMap[fonts_[i]]->Close();
+				CloseFont(fontMap[fonts_[i]]);
 				delete fontMap[fonts_[i]];
 				fontMap.erase(fonts_[i]);
 			}
 		}
-		u32 args[2] = { params_.userDataAddr, (u32)handle_ };
+		u32 args[2] = { userDataAddr(), (u32)handle_ };
 		// TODO: The return value of this is leaking.
 		if (handle_) {  // Avoid calling free-callback on double-free
-			__KernelDirectMipsCall(params_.freeFuncAddr, 0, args, 2, false);
+			if (coreState != CORE_POWERDOWN) {
+				hleEnqueueCall(freeFuncAddr(), 2, args);
+			}
 		}
 		handle_ = 0;
 		fonts_.clear();
 		isfontopen_.clear();
+		openAllocatedAddresses_.clear();
 	}

 	void AllocDone(u32 allocatedAddr) {
 		handle_ = allocatedAddr;
 		fonts_.resize(params_.numFonts);
 		isfontopen_.resize(params_.numFonts);
+		openAllocatedAddresses_.resize(params_.numFonts);
 		for (size_t i = 0; i < fonts_.size(); i++) {
 			u32 addr = allocatedAddr + 0x4C + (u32)i * 0x4C;
 			isfontopen_[i] = 0;
@@ -471,6 +549,9 @@ public:

 	u32 handle() const { return handle_; }
 	int numFonts() const { return params_.numFonts; }
+	u32_le userDataAddr() const{ return params_.userDataAddr; }
+	u32_le allocFuncAddr() const { return params_.allocFuncAddr; }
+	u32_le freeFuncAddr() const { return params_.freeFuncAddr; }

 	void SetResolution(float hres, float vres) {
 		fontHRes_ = hres;
@@ -524,6 +605,24 @@ public:
 			delete prevFont->second;
 		}
 		fontMap[loadedFont->Handle()] = loadedFont;
+
+		if (!useAllocCallbacks)
+			return loadedFont;
+
+		u32 allocSize = 12;
+		if (mode == FONT_OPEN_INTERNAL_STINGY) {
+			allocSize = 0x239B4;
+		} else if (mode == FONT_OPEN_INTERNAL_FULL) {
+			allocSize += loadedFont->GetFont()->getSize();
+		}
+
+		PostOpenAllocCallback *action = (PostOpenAllocCallback *)__KernelCreateAction(actionPostOpenAllocCallback);
+		action->SetFontLib(GetListID());
+		action->SetFont(loadedFont->Handle(), freeFontIndex);
+
+		u32 args[2] = { userDataAddr(), allocSize };
+		hleEnqueueCall(allocFuncAddr(), 2, args, action);
+
 		return loadedFont;
 	}

@@ -531,14 +630,28 @@ public:
 		for (size_t i = 0; i < fonts_.size(); i++) {
 			if (fonts_[i] == font->Handle()) {
 				isfontopen_[i] = 0;
-
+				if (openAllocatedAddresses_[i] != 0 && coreState != CORE_POWERDOWN) {
+					u32 args[2] = { userDataAddr(), openAllocatedAddresses_[i] };
+					hleEnqueueCall(freeFuncAddr(), 2, args);
+					openAllocatedAddresses_[i] = 0;
+				}
+				break;
 			}
 		}
+		flushFont();
 		font->Close();
 	}

+	void flushFont() {
+		if (charInfoBitmapAddress_ != 0 && coreState != CORE_POWERDOWN) {
+			u32 args[2] = { userDataAddr(), charInfoBitmapAddress_ };
+			hleEnqueueCall(freeFuncAddr(), 2, args);
+			charInfoBitmapAddress_ = 0;
+		}
+	}
+
 	void DoState(PointerWrap &p) {
-		auto s = p.Section("FontLib", 1, 2);
+		auto s = p.Section("FontLib", 1, 3);
 		if (!s)
 			return;

@@ -555,6 +668,14 @@ public:
 		} else {
 			nfl_ = 0;
 		}
+
+		if (s >= 3) {
+			p.Do(openAllocatedAddresses_);
+			p.Do(charInfoBitmapAddress_);
+		} else {
+			openAllocatedAddresses_.resize(params_.numFonts);
+			charInfoBitmapAddress_ = 0;
+		}
 	}

 	void SetFileFontHandle(u32 handle) {
@@ -563,6 +684,20 @@ public:

 	u32 GetAltCharCode() const { return altCharCode_; }

+	u32 GetOpenAllocatedAddress(int index) const { 
+		if(index < numFonts())
+			return openAllocatedAddresses_[index];
+		return 0;
+	}
+
+	void SetOpenAllocatedAddress(int index, u32 addr) {
+		if (index < numFonts())
+			openAllocatedAddresses_[index] = addr;
+	}
+
+	u32 GetCharInfoBitmapAddress() const { return charInfoBitmapAddress_; }
+	void SetCharInfoBitmapAddress(u32 addr) { charInfoBitmapAddress_ = addr; }
+
 private:
 	std::vector<u32> fonts_;
 	std::vector<u32> isfontopen_;
@@ -573,6 +708,8 @@ private:
 	int fileFontHandle_;
 	int handle_;
 	int altCharCode_;
+	std::vector<u32> openAllocatedAddresses_;
+	u32 charInfoBitmapAddress_;
 	PSPPointer<NativeFontLib> nfl_;

 	DISALLOW_COPY_AND_ASSIGN(FontLib);
@@ -590,6 +727,7 @@ void PostAllocCallback::run(MipsCall &call) {
 		FontLib *fontLib = fontLibList[fontLibID_];
 		fontLib->AllocDone(v0);
 		fontLibMap[fontLib->handle()] = fontLibID_;
+		// This is the same as v0 above.
 		call.setReturnValue(fontLib->handle());
 	}
 	INFO_LOG(SCEFONT, "Leaving PostAllocCallback::run");
@@ -601,6 +739,34 @@ void PostOpenCallback::run(MipsCall &call) {
 	fontLib->SetFileFontHandle(v0);
 }

+void PostOpenAllocCallback::run(MipsCall &call) {
+	FontLib *fontLib = fontLibList[fontLibID_];
+	u32 v0 = currentMIPS->r[MIPS_REG_V0];
+	fontLib->SetOpenAllocatedAddress(fontIndex_, v0);
+}
+
+void PostCharInfoAllocCallback::run(MipsCall &call) {
+	FontLib *fontLib = fontLibList[fontLibID_];
+	u32 v0 = currentMIPS->r[MIPS_REG_V0];
+	if (v0 == 0) {
+		call.setReturnValue(ERROR_FONT_OUT_OF_MEMORY); // From JPCSP, if alloc size is 0, still this error value?
+	} else {
+		fontLib->SetCharInfoBitmapAddress(v0);
+	}
+}
+
+void PostCharInfoFreeCallback::run(MipsCall &call) {
+	FontLib *fontLib = fontLibList[fontLibID_];
+	fontLib->SetCharInfoBitmapAddress(0);
+
+	u32 allocSize = charInfo_->bitmapWidth * charInfo_->bitmapHeight;
+	PostCharInfoAllocCallback *action = (PostCharInfoAllocCallback *)__KernelCreateAction(actionPostCharInfoAllocCallback);
+	action->SetFontLib(fontLibID_);
+
+	u32 args[2] = { fontLib->userDataAddr(), allocSize };
+	hleEnqueueCall(fontLib->allocFuncAddr(), 2, args, action);
+}
+
 inline bool LoadedFont::GetCharInfo(int charCode, PGFCharInfo *charInfo, int glyphType) const {
 	auto fontLib = GetFontLib();
 	int altCharCode = fontLib == NULL ? -1 : fontLib->GetAltCharCode();
@@ -709,8 +875,12 @@ int GetInternalFontIndex(Font *font) {
 }

 void __FontInit() {
+	useAllocCallbacks = true;
 	actionPostAllocCallback = __KernelRegisterActionType(PostAllocCallback::Create);
 	actionPostOpenCallback = __KernelRegisterActionType(PostOpenCallback::Create);
+	actionPostOpenAllocCallback = __KernelRegisterActionType(PostOpenAllocCallback::Create);
+	actionPostCharInfoAllocCallback = __KernelRegisterActionType(PostCharInfoAllocCallback::Create);
+	actionPostCharInfoFreeCallback = __KernelRegisterActionType(PostCharInfoFreeCallback::Create);
 }

 void __FontShutdown() {
@@ -733,7 +903,7 @@ void __FontShutdown() {
 }

 void __FontDoState(PointerWrap &p) {
-	auto s = p.Section("sceFont", 1);
+	auto s = p.Section("sceFont", 1, 2);
 	if (!s)
 		return;

@@ -747,6 +917,16 @@ void __FontDoState(PointerWrap &p) {
 	__KernelRestoreActionType(actionPostAllocCallback, PostAllocCallback::Create);
 	p.Do(actionPostOpenCallback);
 	__KernelRestoreActionType(actionPostOpenCallback, PostOpenCallback::Create);
+	if (s >= 2) {
+		p.Do(actionPostOpenAllocCallback);
+		__KernelRestoreActionType(actionPostOpenAllocCallback, PostOpenAllocCallback::Create);
+		p.Do(actionPostCharInfoAllocCallback);
+		__KernelRestoreActionType(actionPostCharInfoAllocCallback, PostCharInfoAllocCallback::Create);
+		p.Do(actionPostCharInfoFreeCallback);
+		__KernelRestoreActionType(actionPostCharInfoFreeCallback, PostCharInfoFreeCallback::Create);
+	} else {
+		useAllocCallbacks = false;
+	}
 }

 static u32 sceFontNewLib(u32 paramPtr, u32 errorCodePtr) {
@@ -912,14 +1092,13 @@ static u32 sceFontOpenUserFile(u32 libHandle, const char *fileName, u32 mode, u3

 static int sceFontClose(u32 fontHandle) {
 	LoadedFont *font = GetLoadedFont(fontHandle, false);
-	if (font)
-	{
+	if (font) {
 		DEBUG_LOG(SCEFONT, "sceFontClose(%x)", fontHandle);
 		FontLib *fontLib = font->GetFontLib();
-		if (fontLib)
+		if (fontLib) {
 			fontLib->CloseFont(font);
-	}
-	else
+		}
+	} else
 		ERROR_LOG(SCEFONT, "sceFontClose(%x) - font not open?", fontHandle);
 	return 0;
 }
@@ -1100,6 +1279,27 @@ static int sceFontGetCharInfo(u32 fontHandle, u32 charCode, u32 charInfoPtr) {
 	auto charInfo = PSPPointer<PGFCharInfo>::Create(charInfoPtr);
 	font->GetCharInfo(charCode, charInfo);

+	if (!useAllocCallbacks)
+		return 0;
+
+	u32 allocSize = charInfo->bitmapWidth * charInfo->bitmapHeight;
+	if (charInfo->sfp26AdvanceH != 0 || charInfo->sfp26AdvanceV != 0) {
+		if (font->GetFontLib()->GetCharInfoBitmapAddress() != 0) {
+			PostCharInfoFreeCallback *action = (PostCharInfoFreeCallback *)__KernelCreateAction(actionPostCharInfoFreeCallback);
+			action->SetFontLib(font->GetFontLib()->GetListID());
+			action->SetCharInfo(charInfo);
+
+			u32 args[2] = { font->GetFontLib()->userDataAddr(), font->GetFontLib()->GetCharInfoBitmapAddress() };
+			hleEnqueueCall(font->GetFontLib()->freeFuncAddr(), 2, args, action);
+		} else {
+			PostCharInfoAllocCallback *action = (PostCharInfoAllocCallback *)__KernelCreateAction(actionPostCharInfoAllocCallback);
+			action->SetFontLib(font->GetFontLib()->GetListID());
+
+			u32 args[2] = { font->GetFontLib()->userDataAddr(), allocSize };
+			hleEnqueueCall(font->GetFontLib()->allocFuncAddr(), 2, args, action);
+		}
+	}
+
 	return 0;
 }

@@ -1215,7 +1415,15 @@ static int sceFontSetAltCharacterCode(u32 fontLibHandle, u32 charCode) {

 static int sceFontFlush(u32 fontHandle) {
 	INFO_LOG(SCEFONT, "sceFontFlush(%i)", fontHandle);
-	// Probably don't need to do anything here.
+	
+	LoadedFont *font = GetLoadedFont(fontHandle, true);
+	if (!font) {
+		ERROR_LOG_REPORT(SCEFONT, "sceFontFlush(%08x): bad font", fontHandle);
+		return ERROR_FONT_INVALID_PARAMETER;
+	}
+
+	font->GetFontLib()->flushFont();
+
 	return 0;
 }

@@ -930,7 +930,7 @@ static u32 sceIoGetstat(const char *filename, u32 addr) {
 }

 static u32 sceIoChstat(const char *filename, u32 iostatptr, u32 changebits) {
-	ERROR_LOG(SCEIO, "UNIMPL sceIoChstat(%s, %08x, %08x)", filename, iostatptr, changebits);
+	ERROR_LOG_REPORT(SCEIO, "UNIMPL sceIoChstat(%s, %08x, %08x)", filename, iostatptr, changebits);
 	if (changebits & SCE_CST_MODE)
 		ERROR_LOG(SCEIO, "sceIoChstat: change mode requested");
 	if (changebits & SCE_CST_ATTR)
@@ -2731,7 +2731,8 @@ static int IoAsyncFinish(int id) {
 			break;

 		default:
-			ERROR_LOG_REPORT(SCEIO, "Unknown async op %d", params.op);
+			ERROR_LOG_REPORT(SCEIO, "Unknown async op %d", (int)params.op);
+			us = 0;
 			break;
 		}

@@ -37,16 +37,14 @@ struct NativeAlarm
 	u32_le commonPtr;
 };

-struct Alarm : public KernelObject
-{
+struct PSPAlarm : public KernelObject {
 	const char *GetName() override {return "[Alarm]";}
 	const char *GetTypeName() override {return "Alarm";}
 	static u32 GetMissingErrorCode() { return SCE_KERNEL_ERROR_UNKNOWN_ALMID; }
 	static int GetStaticIDType() { return SCE_KERNEL_TMID_Alarm; }
 	int GetIDType() const override { return SCE_KERNEL_TMID_Alarm; }

-	void DoState(PointerWrap &p) override
-	{
+	void DoState(PointerWrap &p) override {
 		auto s = p.Section("Alarm", 1);
 		if (!s)
 			return;
@@ -57,7 +55,7 @@ struct Alarm : public KernelObject
 	NativeAlarm alm;
 };

-void __KernelScheduleAlarm(Alarm *alarm, u64 micro);
+void __KernelScheduleAlarm(PSPAlarm *alarm, u64 micro);

 class AlarmIntrHandler : public IntrHandler
 {
@@ -69,7 +67,7 @@ public:
 		u32 error;
 		int alarmID = triggeredAlarm.front();

-		Alarm *alarm = kernelObjects.Get<Alarm>(alarmID, error);
+		PSPAlarm *alarm = kernelObjects.Get<PSPAlarm>(alarmID, error);
 		if (error)
 		{
 			WARN_LOG(SCEKERNEL, "Ignoring deleted alarm %08x", alarmID);
@@ -95,7 +93,7 @@ public:
 		{
 			DEBUG_LOG(SCEKERNEL, "Rescheduling alarm %08x for +%dms", alarmID, result);
 			u32 error;
-			Alarm *alarm = kernelObjects.Get<Alarm>(alarmID, error);
+			PSPAlarm *alarm = kernelObjects.Get<PSPAlarm>(alarmID, error);
 			__KernelScheduleAlarm(alarm, result);
 		}
 		else
@@ -106,21 +104,19 @@ public:
 			DEBUG_LOG(SCEKERNEL, "Finished alarm %08x", alarmID);

 			// Delete the alarm if it's not rescheduled.
-			kernelObjects.Destroy<Alarm>(alarmID);
+			kernelObjects.Destroy<PSPAlarm>(alarmID);
 		}
 	}
 };

 static int alarmTimer = -1;

-static void __KernelTriggerAlarm(u64 userdata, int cyclesLate)
-{
+static void __KernelTriggerAlarm(u64 userdata, int cyclesLate) {
 	int uid = (int) userdata;

 	u32 error;
-	Alarm *alarm = kernelObjects.Get<Alarm>(uid, error);
-	if (alarm)
-	{
+	PSPAlarm *alarm = kernelObjects.Get<PSPAlarm>(uid, error);
+	if (alarm) {
 		triggeredAlarm.push_back(uid);
 		__TriggerInterrupt(PSP_INTR_IMMEDIATE, PSP_SYSTIMER0_INTR);
 	}
@@ -144,14 +140,12 @@ void __KernelAlarmDoState(PointerWrap &p)
 	CoreTiming::RestoreRegisterEvent(alarmTimer, "Alarm", __KernelTriggerAlarm);
 }

-KernelObject *__KernelAlarmObject()
-{
+KernelObject *__KernelAlarmObject() {
 	// Default object to load from state.
-	return new Alarm;
+	return new PSPAlarm;
 }

-void __KernelScheduleAlarm(Alarm *alarm, u64 micro)
-{
+void __KernelScheduleAlarm(PSPAlarm *alarm, u64 micro) {
 	alarm->alm.schedule = CoreTiming::GetGlobalTimeUs() + micro;
 	CoreTiming::ScheduleEvent(usToCycles(micro), alarmTimer, alarm->GetUID());
 }
@@ -161,7 +155,7 @@ static SceUID __KernelSetAlarm(u64 micro, u32 handlerPtr, u32 commonPtr)
 	if (!Memory::IsValidAddress(handlerPtr))
 		return SCE_KERNEL_ERROR_ILLEGAL_ADDR;

-	Alarm *alarm = new Alarm;
+	PSPAlarm *alarm = new PSPAlarm();
 	SceUID uid = kernelObjects.Create(alarm);

 	alarm->alm.size = NATIVEALARM_SIZE;
@@ -197,13 +191,13 @@ int sceKernelCancelAlarm(SceUID uid)

 	CoreTiming::UnscheduleEvent(alarmTimer, uid);

-	return kernelObjects.Destroy<Alarm>(uid);
+	return kernelObjects.Destroy<PSPAlarm>(uid);
 }

 int sceKernelReferAlarmStatus(SceUID uid, u32 infoPtr)
 {
 	u32 error;
-	Alarm *alarm = kernelObjects.Get<Alarm>(uid, error);
+	PSPAlarm *alarm = kernelObjects.Get<PSPAlarm>(uid, error);
 	if (!alarm)
 	{
 		ERROR_LOG(SCEKERNEL, "sceKernelReferAlarmStatus(%08x, %08x): invalid alarm", uid, infoPtr);
@@ -228,4 +222,4 @@ int sceKernelReferAlarmStatus(SceUID uid, u32 infoPtr)
 		Memory::Write_U32(alarm->alm.commonPtr, infoPtr + 16);

 	return 0;
-}
+}
@@ -48,15 +48,13 @@ static const u32 PSP_NUMBER_SUBINTERRUPTS = 32;
 // INTERRUPT MANAGEMENT
 //////////////////////////////////////////////////////////////////////////

-class InterruptState
-{
+class InterruptState {
 public:
 	void save();
 	void restore();
 	void clear();

-	void DoState(PointerWrap &p)
-	{
+	void DoState(PointerWrap &p) {
 		auto s = p.Section("InterruptState", 1);
 		if (!s)
 			return;
@@ -64,7 +62,7 @@ public:
 		p.Do(savedCpu);
 	}

-	ThreadContext savedCpu;
+	PSPThreadContext savedCpu;
 };

 // STATE
@@ -213,10 +213,6 @@ struct SceKernelVplHeader {
 		do {
 			auto b = prev->next;
 			if (b->sizeInBlocks > allocBlocks) {
-				if (nextFreeBlock_ == b) {
-					nextFreeBlock_ = prev;
-				}
-				prev = b;
 				b = SplitBlock(b, allocBlocks);
 			}

@@ -289,16 +285,13 @@ struct SceKernelVplHeader {
 	void UnlinkFreeBlock(PSPPointer<SceKernelVplBlock> b, PSPPointer<SceKernelVplBlock> prev) {
 		allocatedInBlocks_ += b->sizeInBlocks;
 		prev->next = b->next;
-		if (nextFreeBlock_ == b) {
-			nextFreeBlock_ = prev;
-		}
+		nextFreeBlock_ = prev;
 		b->next = SentinelPtr();
 	}

 	PSPPointer<SceKernelVplBlock> SplitBlock(PSPPointer<SceKernelVplBlock> b, u32 allocBlocks) {
-		u32 prev = b->next.ptr;
+		u32 prev = b.ptr;
 		b->sizeInBlocks -= allocBlocks;
-		b->next = b + b->sizeInBlocks;

 		b += b->sizeInBlocks;
 		b->sizeInBlocks = allocBlocks;
@@ -155,8 +155,8 @@ void ImportFuncSymbol(const FuncSymbolImport &func, bool reimporting);
 void ExportFuncSymbol(const FuncSymbolExport &func);
 void UnexportFuncSymbol(const FuncSymbolExport &func);

-class Module;
-static bool KernelImportModuleFuncs(Module *module, u32 *firstImportStubAddr, bool reimporting = false);
+class PSPModule;
+static bool KernelImportModuleFuncs(PSPModule *module, u32 *firstImportStubAddr, bool reimporting = false);

 struct NativeModule {
 	u32_le next;
@@ -228,10 +228,10 @@ enum NativeModuleStatus {
 	MODULE_STATUS_UNLOADING = 8,
 };

-class Module : public KernelObject {
+class PSPModule : public KernelObject {
 public:
-	Module() : textStart(0), textEnd(0), libstub(0), libstubend(0), memoryBlockAddr(0), isFake(false) {}
-	~Module() {
+	PSPModule() : textStart(0), textEnd(0), libstub(0), libstubend(0), memoryBlockAddr(0), isFake(false) {}
+	~PSPModule() {
 		if (memoryBlockAddr) {
 			// If it's either below user memory, or using a high kernel bit, it's in kernel.
 			if (memoryBlockAddr < PSP_GetUserMemoryBase() || memoryBlockAddr > PSP_GetUserMemoryEnd()) {
@@ -421,10 +421,10 @@ public:

 KernelObject *__KernelModuleObject()
 {
-	return new Module;
+	return new PSPModule;
 }

-class AfterModuleEntryCall : public Action {
+class AfterModuleEntryCall : public PSPAction {
 public:
 	AfterModuleEntryCall() {}
 	SceUID moduleID_;
@@ -438,7 +438,7 @@ public:
 		p.Do(moduleID_);
 		p.Do(retValAddr);
 	}
-	static Action *Create() {
+	static PSPAction *Create() {
 		return new AfterModuleEntryCall;
 	}
 };
@@ -507,7 +507,7 @@ void __KernelModuleDoState(PointerWrap &p)
 		u32 error;
 		// We process these late, since they depend on loadedModules for interlinking.
 		for (SceUID moduleId : loadedModules) {
-			Module *module = kernelObjects.Get<Module>(moduleId, error);
+			PSPModule *module = kernelObjects.Get<PSPModule>(moduleId, error);
 			if (module && module->libstub != 0) {
 				if (!KernelImportModuleFuncs(module, nullptr, true)) {
 					ERROR_LOG(LOADER, "Something went wrong loading imports on load state");
@@ -655,7 +655,7 @@ void ImportVarSymbol(const VarSymbolImport &var) {

 	u32 error;
 	for (SceUID moduleId : loadedModules) {
-		Module *module = kernelObjects.Get<Module>(moduleId, error);
+		PSPModule *module = kernelObjects.Get<PSPModule>(moduleId, error);
 		if (!module || !module->ImportsOrExportsModuleName(var.moduleName)) {
 			continue;
 		}
@@ -676,7 +676,7 @@ void ImportVarSymbol(const VarSymbolImport &var) {
 void ExportVarSymbol(const VarSymbolExport &var) {
 	u32 error;
 	for (SceUID moduleId : loadedModules) {
-		Module *module = kernelObjects.Get<Module>(moduleId, error);
+		PSPModule *module = kernelObjects.Get<PSPModule>(moduleId, error);
 		if (!module || !module->ImportsOrExportsModuleName(var.moduleName)) {
 			continue;
 		}
@@ -694,7 +694,7 @@ void ExportVarSymbol(const VarSymbolExport &var) {
 void UnexportVarSymbol(const VarSymbolExport &var) {
 	u32 error;
 	for (SceUID moduleId : loadedModules) {
-		Module *module = kernelObjects.Get<Module>(moduleId, error);
+		PSPModule *module = kernelObjects.Get<PSPModule>(moduleId, error);
 		if (!module || !module->ImportsOrExportsModuleName(var.moduleName)) {
 			continue;
 		}
@@ -724,7 +724,7 @@ void ImportFuncSymbol(const FuncSymbolImport &func, bool reimporting) {

 	u32 error;
 	for (SceUID moduleId : loadedModules) {
-		Module *module = kernelObjects.Get<Module>(moduleId, error);
+		PSPModule *module = kernelObjects.Get<PSPModule>(moduleId, error);
 		if (!module || !module->ImportsOrExportsModuleName(func.moduleName)) {
 			continue;
 		}
@@ -765,7 +765,7 @@ void ExportFuncSymbol(const FuncSymbolExport &func) {

 	u32 error;
 	for (SceUID moduleId : loadedModules) {
-		Module *module = kernelObjects.Get<Module>(moduleId, error);
+		PSPModule *module = kernelObjects.Get<PSPModule>(moduleId, error);
 		if (!module || !module->ImportsOrExportsModuleName(func.moduleName)) {
 			continue;
 		}
@@ -790,7 +790,7 @@ void UnexportFuncSymbol(const FuncSymbolExport &func) {

 	u32 error;
 	for (SceUID moduleId : loadedModules) {
-		Module *module = kernelObjects.Get<Module>(moduleId, error);
+		PSPModule *module = kernelObjects.Get<PSPModule>(moduleId, error);
 		if (!module || !module->ImportsOrExportsModuleName(func.moduleName)) {
 			continue;
 		}
@@ -806,7 +806,7 @@ void UnexportFuncSymbol(const FuncSymbolExport &func) {
 	}
 }

-void Module::Cleanup() {
+void PSPModule::Cleanup() {
 	MIPSAnalyst::ForgetFunctions(textStart, textEnd);

 	loadedModules.erase(GetUID());
@@ -909,7 +909,7 @@ static bool IsHLEVersionedModule(const char *name) {
 	return false;
 }

-static bool KernelImportModuleFuncs(Module *module, u32 *firstImportStubAddr, bool reimporting) {
+static bool KernelImportModuleFuncs(PSPModule *module, u32 *firstImportStubAddr, bool reimporting) {
 	struct PspLibStubEntry {
 		u32_le name;
 		u16_le version;
@@ -1090,8 +1090,8 @@ static int gzipDecompress(u8 *OutBuffer, int OutBufferLength, u8 *InBuffer) {
 	return stream.total_out;
 }

-static Module *__KernelLoadELFFromPtr(const u8 *ptr, size_t elfSize, u32 loadAddress, bool fromTop, std::string *error_string, u32 *magic, u32 &error) {
-	Module *module = new Module;
+static PSPModule *__KernelLoadELFFromPtr(const u8 *ptr, size_t elfSize, u32 loadAddress, bool fromTop, std::string *error_string, u32 *magic, u32 &error) {
+	PSPModule *module = new PSPModule();
 	kernelObjects.Create(module);
 	loadedModules.insert(module->GetUID());
 	memset(&module->nm, 0, sizeof(module->nm));
@@ -1133,7 +1133,7 @@ static Module *__KernelLoadELFFromPtr(const u8 *ptr, size_t elfSize, u32 loadAdd
 		if (size > elfSize) {
 			*error_string = StringFromFormat("ELF/PRX truncated: %d > %d", (int)size, (int)elfSize);
 			module->Cleanup();
-			kernelObjects.Destroy<Module>(module->GetUID());
+			kernelObjects.Destroy<PSPModule>(module->GetUID());
 			return nullptr;
 		}
 		const auto maxElfSize = std::max(head->elf_size, head->psp_size);
@@ -1166,7 +1166,7 @@ static Module *__KernelLoadELFFromPtr(const u8 *ptr, size_t elfSize, u32 loadAdd
 			if (addr == (u32)-1) {
 				error = SCE_KERNEL_ERROR_MEMBLOCK_ALLOC_FAILED;
 				module->Cleanup();
-				kernelObjects.Destroy<Module>(module->GetUID());
+				kernelObjects.Destroy<PSPModule>(module->GetUID());
 			} else {
 				error = 0;
 				module->memoryBlockAddr = addr;
@@ -1207,7 +1207,7 @@ static Module *__KernelLoadELFFromPtr(const u8 *ptr, size_t elfSize, u32 loadAdd
 		if (newptr)
 			delete [] newptr;
 		module->Cleanup();
-		kernelObjects.Destroy<Module>(module->GetUID());
+		kernelObjects.Destroy<PSPModule>(module->GetUID());
 		error = SCE_KERNEL_ERROR_UNSUPPORTED_PRX_TYPE;
 		return nullptr;
 	}
@@ -1221,7 +1221,7 @@ static Module *__KernelLoadELFFromPtr(const u8 *ptr, size_t elfSize, u32 loadAdd
 		if (newptr)
 			delete [] newptr;
 		module->Cleanup();
-		kernelObjects.Destroy<Module>(module->GetUID());
+		kernelObjects.Destroy<PSPModule>(module->GetUID());
 		error = result;
 		return nullptr;
 	}
@@ -1536,8 +1536,8 @@ static Module *__KernelLoadELFFromPtr(const u8 *ptr, size_t elfSize, u32 loadAdd
 	return module;
 }

-static Module *__KernelLoadModule(u8 *fileptr, size_t fileSize, SceKernelLMOption *options, std::string *error_string) {
-	Module *module = 0;
+static PSPModule *__KernelLoadModule(u8 *fileptr, size_t fileSize, SceKernelLMOption *options, std::string *error_string) {
+	PSPModule *module = nullptr;
 	// Check for PBP
 	if (memcmp(fileptr, "\0PBP", 4) == 0) {
 		// PBP!
@@ -1583,8 +1583,7 @@ static Module *__KernelLoadModule(u8 *fileptr, size_t fileSize, SceKernelLMOptio
 	return module;
 }

-static void __KernelStartModule(Module *m, int args, const char *argp, SceKernelSMOption *options)
-{
+static void __KernelStartModule(PSPModule *m, int args, const char *argp, SceKernelSMOption *options) {
 	m->nm.status = MODULE_STATUS_STARTED;
 	if (m->nm.module_start_func != 0 && m->nm.module_start_func != (u32)-1)
 	{
@@ -1599,16 +1598,12 @@ static void __KernelStartModule(Module *m, int args, const char *argp, SceKernel
 }


-u32 __KernelGetModuleGP(SceUID uid)
-{
+u32 __KernelGetModuleGP(SceUID uid) {
 	u32 error;
-	Module *module = kernelObjects.Get<Module>(uid, error);
-	if (module)
-	{
+	PSPModule *module = kernelObjects.Get<PSPModule>(uid, error);
+	if (module) {
 		return module->nm.gp_value;
-	}
-	else
-	{
+	} else {
 		return 0;
 	}
 }
@@ -1619,7 +1614,7 @@ void __KernelLoadReset() {
 		u32 error;
 		while (!loadedModules.empty()) {
 			SceUID moduleID = *loadedModules.begin();
-			Module *module = kernelObjects.Get<Module>(moduleID, error);
+			PSPModule *module = kernelObjects.Get<PSPModule>(moduleID, error);
 			if (module) {
 				module->Cleanup();
 			} else {
@@ -1635,8 +1630,6 @@ void __KernelLoadReset() {
 		HLEShutdown();
 		Replacement_Init();
 		HLEInit();
-		assert(gpu);
-		gpu->Reinitialize();
 	}

 	__KernelModuleInit();
@@ -1688,12 +1681,12 @@ bool __KernelLoadExec(const char *filename, u32 paramPtr, std::string *error_str
 	pspFileSystem.ReadFile(handle, temp, (size_t)info.size);

 	PSP_SetLoading("Loading modules...");
-	Module *module = __KernelLoadModule(temp, (size_t)info.size, 0, error_string);
+	PSPModule *module = __KernelLoadModule(temp, (size_t)info.size, 0, error_string);

 	if (!module || module->isFake) {
 		if (module) {
 			module->Cleanup();
-			kernelObjects.Destroy<Module>(module->GetUID());
+			kernelObjects.Destroy<PSPModule>(module->GetUID());
 		}
 		ERROR_LOG(LOADER, "Failed to load module %s", filename);
 		*error_string = "Failed to load executable: " + *error_string;
@@ -1772,7 +1765,7 @@ bool __KernelLoadGEDump(const std::string &base_filename, std::string *error_str
 		Memory::WriteUnchecked_U32(runDumpCode[i], mipsr4k.pc + (int)i * sizeof(u32_le));
 	}

-	Module *module = new Module;
+	PSPModule *module = new PSPModule();
 	kernelObjects.Create(module);
 	loadedModules.insert(module->GetUID());
 	memset(&module->nm, 0, sizeof(module->nm));
@@ -1836,6 +1829,9 @@ int sceKernelLoadExec(const char *filename, u32 paramPtr)
 		Core_UpdateState(CORE_ERROR);
 		return -1;
 	}
+	if (gpu) {
+		gpu->Reinitialize();
+	}
 	return 0;
 }

@@ -1846,7 +1842,7 @@ u32 sceKernelLoadModule(const char *name, u32 flags, u32 optionAddr) {

 	for (size_t i = 0; i < ARRAY_SIZE(lieAboutSuccessModules); i++) {
 		if (!strcmp(name, lieAboutSuccessModules[i])) {
-			Module *module = new Module;
+			PSPModule *module = new PSPModule();
 			kernelObjects.Create(module);
 			loadedModules.insert(module->GetUID());
 			memset(&module->nm, 0, sizeof(module->nm));
@@ -1897,7 +1893,7 @@ u32 sceKernelLoadModule(const char *name, u32 flags, u32 optionAddr) {
 		WARN_LOG_REPORT(LOADER, "sceKernelLoadModule: unsupported options size=%08x, flags=%08x, pos=%d, access=%d, data=%d, text=%d", lmoption->size, lmoption->flags, lmoption->position, lmoption->access, lmoption->mpiddata, lmoption->mpidtext);
 	}

-	Module *module = 0;
+	PSPModule *module = nullptr;
 	u8 *temp = new u8[(int)size];
 	u32 handle = pspFileSystem.OpenFile(name, FILEACCESS_READ);
 	pspFileSystem.ReadFile(handle, temp, (size_t)size);
@@ -1920,6 +1916,9 @@ u32 sceKernelLoadModule(const char *name, u32 flags, u32 optionAddr) {
 			NOTICE_LOG_REPORT(LOADER, "Module %s is blacklisted or undecryptable - we try __KernelLoadExec", name);
 			// Name might get deleted.
 			const std::string safeName = name;
+			if (gpu) {
+				gpu->Reinitialize();
+			}
 			return __KernelLoadExec(safeName.c_str(), 0, &error_string);
 		} else {
 			hleLogError(LOADER, error, "failed to load");
@@ -1957,7 +1956,7 @@ static void sceKernelStartModule(u32 moduleId, u32 argsize, u32 argAddr, u32 ret
 		Memory::ReadStruct(optionAddr, &smoption);
 	}
 	u32 error;
-	Module *module = kernelObjects.Get<Module>(moduleId, error);
+	PSPModule *module = kernelObjects.Get<PSPModule>(moduleId, error);
 	if (!module) {
 		INFO_LOG(SCEMODULE, "sceKernelStartModule(%d,asize=%08x,aptr=%08x,retptr=%08x,%08x): error %08x", moduleId, argsize, argAddr, returnValueAddr, optionAddr, error);
 		RETURN(error);
@@ -2056,7 +2055,7 @@ static u32 sceKernelStopModule(u32 moduleId, u32 argSize, u32 argAddr, u32 retur
 	// TODO: In a lot of cases (even for errors), this should resched.  Needs testing.

 	u32 error;
-	Module *module = kernelObjects.Get<Module>(moduleId, error);
+	PSPModule *module = kernelObjects.Get<PSPModule>(moduleId, error);
 	if (!module)
 	{
 		ERROR_LOG(SCEMODULE, "sceKernelStopModule(%08x, %08x, %08x, %08x, %08x): invalid module id", moduleId, argSize, argAddr, returnValueAddr, optionAddr);
@@ -2129,12 +2128,12 @@ static u32 sceKernelUnloadModule(u32 moduleId)
 {
 	INFO_LOG(SCEMODULE,"sceKernelUnloadModule(%i)", moduleId);
 	u32 error;
-	Module *module = kernelObjects.Get<Module>(moduleId, error);
+	PSPModule *module = kernelObjects.Get<PSPModule>(moduleId, error);
 	if (!module)
 		return hleDelayResult(error, "module unloaded", 150);

 	module->Cleanup();
-	kernelObjects.Destroy<Module>(moduleId);
+	kernelObjects.Destroy<PSPModule>(moduleId);
 	return hleDelayResult(moduleId, "module unloaded", 500);
 }

@@ -2151,7 +2150,7 @@ u32 hleKernelStopUnloadSelfModuleWithOrWithoutStatus(u32 exitCode, u32 argSize,
 		// TODO: In a lot of cases (even for errors), this should resched.  Needs testing.

 		u32 error;
-		Module *module = kernelObjects.Get<Module>(moduleID, error);
+		PSPModule *module = kernelObjects.Get<PSPModule>(moduleID, error);
 		if (!module) {
 			if (WithStatus)
 				ERROR_LOG(SCEMODULE, "sceKernelStopUnloadSelfModuleWithStatus(%08x, %08x, %08x, %08x, %08x): invalid module id", exitCode, argSize, argp, statusAddr, optionAddr);
@@ -2200,7 +2199,7 @@ u32 hleKernelStopUnloadSelfModuleWithOrWithoutStatus(u32 exitCode, u32 argSize,
 				INFO_LOG(SCEMODULE, "sceKernelSelfStopUnloadModule(%08x, %08x, %08x): no stop func", exitCode, argSize, argp);
 			sceKernelExitDeleteThread(exitCode);
 			module->Cleanup();
-			kernelObjects.Destroy<Module>(moduleID);
+			kernelObjects.Destroy<PSPModule>(moduleID);
 		} else {
 			if (WithStatus)
 				ERROR_LOG_REPORT(SCEMODULE, "sceKernelStopUnloadSelfModuleWithStatus(%08x, %08x, %08x, %08x, %08x): bad stop func address", exitCode, argSize, argp, statusAddr, optionAddr);
@@ -2208,7 +2207,7 @@ u32 hleKernelStopUnloadSelfModuleWithOrWithoutStatus(u32 exitCode, u32 argSize,
 				ERROR_LOG_REPORT(SCEMODULE, "sceKernelSelfStopUnloadModule(%08x, %08x, %08x): bad stop func address", exitCode, argSize, argp);
 			sceKernelExitDeleteThread(exitCode);
 			module->Cleanup();
-			kernelObjects.Destroy<Module>(moduleID);
+			kernelObjects.Destroy<PSPModule>(moduleID);
 		}
 	} else {
 		if (WithStatus)
@@ -2244,7 +2243,7 @@ void __KernelReturnFromModuleFunc()
 	sceKernelDeleteThread(leftThreadID);

 	u32 error;
-	Module *module = kernelObjects.Get<Module>(leftModuleID, error);
+	PSPModule *module = kernelObjects.Get<PSPModule>(leftModuleID, error);
 	if (!module) {
 		ERROR_LOG_REPORT(SCEMODULE, "Returned from deleted module start/stop func");
 		return;
@@ -2274,7 +2273,7 @@ void __KernelReturnFromModuleFunc()
 	if (module->nm.status == MODULE_STATUS_UNLOADING) {
 		// TODO: Delete the waiting thread?
 		module->Cleanup();
-		kernelObjects.Destroy<Module>(leftModuleID);
+		kernelObjects.Destroy<PSPModule>(leftModuleID);
 	}
 }

@@ -2284,11 +2283,9 @@ struct GetModuleIdByAddressArg
 	SceUID result;
 };

-static bool __GetModuleIdByAddressIterator(Module *module, GetModuleIdByAddressArg *state)
-{
+static bool __GetModuleIdByAddressIterator(PSPModule *module, GetModuleIdByAddressArg *state) {
 	const u32 start = module->memoryBlockAddr, size = module->memoryBlockSize;
-	if (start != 0 && start <= state->addr && start + size > state->addr)
-	{
+	if (start != 0 && start <= state->addr && start + size > state->addr) {
 		state->result = module->GetUID();
 		return false;
 	}
@@ -2347,7 +2344,7 @@ static u32 sceKernelLoadModuleByID(u32 id, u32 flags, u32 lmoptionPtr)
 	size_t size = pspFileSystem.SeekFile(handle, 0, FILEMOVE_END);
 	std::string error_string;
 	pspFileSystem.SeekFile(handle, pos, FILEMOVE_BEGIN);
-	Module *module = 0;
+	PSPModule *module = nullptr;
 	u8 *temp = new u8[size - pos];
 	pspFileSystem.ReadFile(handle, temp, size - pos);
 	u32 magic;
@@ -2405,7 +2402,7 @@ static SceUID sceKernelLoadModuleBufferUsbWlan(u32 size, u32 bufPtr, u32 flags,
 		WARN_LOG_REPORT(LOADER, "sceKernelLoadModuleBufferUsbWlan: unsupported options size=%08x, flags=%08x, pos=%d, access=%d, data=%d, text=%d", lmoption->size, lmoption->flags, lmoption->position, lmoption->access, lmoption->mpiddata, lmoption->mpidtext);
 	}
 	std::string error_string;
-	Module *module = 0;
+	PSPModule *module = nullptr;
 	u32 magic;
 	u32 error;
 	module = __KernelLoadELFFromPtr(Memory::GetPointer(bufPtr), size, 0, lmoption ? lmoption->position == PSP_SMEM_High : false, &error_string, &magic, error);
@@ -2447,7 +2444,7 @@ static u32 sceKernelQueryModuleInfo(u32 uid, u32 infoAddr)
 {
 	INFO_LOG(SCEMODULE, "sceKernelQueryModuleInfo(%i, %08x)", uid, infoAddr);
 	u32 error;
-	Module *module = kernelObjects.Get<Module>(uid, error);
+	PSPModule *module = kernelObjects.Get<PSPModule>(uid, error);
 	if (!module)
 		return error;
 	if (!Memory::IsValidAddress(infoAddr)) {
@@ -2488,7 +2485,7 @@ static u32 sceKernelGetModuleIdList(u32 resultBuffer, u32 resultBufferSize, u32

 	u32 error;
 	for (SceUID moduleId : loadedModules) {
-		Module *module = kernelObjects.Get<Module>(moduleId, error);
+		PSPModule *module = kernelObjects.Get<PSPModule>(moduleId, error);
 		if (!module->isFake) {
 			if (resultBufferOffset < resultBufferSize) {
 				Memory::Write_U32(module->GetUID(), resultBuffer + resultBufferOffset);
@@ -62,7 +62,7 @@ struct NativeMutex
 	s32_le numWaitThreads;
 };

-struct Mutex : public KernelObject
+struct PSPMutex : public KernelObject
 {
 	const char *GetName() override { return nm.name; }
 	const char *GetTypeName() override { return "Mutex"; }
@@ -189,7 +189,7 @@ void __KernelMutexDoState(PointerWrap &p)

 KernelObject *__KernelMutexObject()
 {
-	return new Mutex;
+	return new PSPMutex;
 }

 KernelObject *__KernelLwMutexObject()
@@ -202,8 +202,7 @@ void __KernelMutexShutdown()
 	mutexHeldLocks.clear();
 }

-static void __KernelMutexAcquireLock(Mutex *mutex, int count, SceUID thread)
-{
+static void __KernelMutexAcquireLock(PSPMutex *mutex, int count, SceUID thread) {
 #if defined(_DEBUG)
 	auto locked = mutexHeldLocks.equal_range(thread);
 	for (MutexMap::iterator iter = locked.first; iter != locked.second; ++iter)
@@ -216,13 +215,11 @@ static void __KernelMutexAcquireLock(Mutex *mutex, int count, SceUID thread)
 	mutex->nm.lockThread = thread;
 }

-static void __KernelMutexAcquireLock(Mutex *mutex, int count)
-{
+static void __KernelMutexAcquireLock(PSPMutex *mutex, int count) {
 	__KernelMutexAcquireLock(mutex, count, __KernelGetCurThread());
 }

-static void __KernelMutexEraseLock(Mutex *mutex)
-{
+static void __KernelMutexEraseLock(PSPMutex *mutex) {
 	if (mutex->nm.lockThread != -1)
 	{
 		SceUID id = mutex->GetUID();
@@ -259,8 +256,7 @@ static std::vector<SceUID>::iterator __KernelMutexFindPriority(std::vector<SceUI
 	return best;
 }

-static bool __KernelUnlockMutexForThread(Mutex *mutex, SceUID threadID, u32 &error, int result)
-{
+static bool __KernelUnlockMutexForThread(PSPMutex *mutex, SceUID threadID, u32 &error, int result) {
 	if (!HLEKernel::VerifyWait(threadID, WAITTYPE_MUTEX, mutex->GetUID()))
 		return false;

@@ -283,8 +279,7 @@ static bool __KernelUnlockMutexForThread(Mutex *mutex, SceUID threadID, u32 &err
 	return true;
 }

-static bool __KernelUnlockMutexForThreadCheck(Mutex *mutex, SceUID threadID, u32 &error, int result, bool &wokeThreads)
-{
+static bool __KernelUnlockMutexForThreadCheck(PSPMutex *mutex, SceUID threadID, u32 &error, int result, bool &wokeThreads) {
 	if (mutex->nm.lockThread == -1 && __KernelUnlockMutexForThread(mutex, threadID, error, 0))
 		return true;
 	return false;
@@ -292,7 +287,7 @@ static bool __KernelUnlockMutexForThreadCheck(Mutex *mutex, SceUID threadID, u32

 void __KernelMutexBeginCallback(SceUID threadID, SceUID prevCallbackId)
 {
-	auto result = HLEKernel::WaitBeginCallback<Mutex, WAITTYPE_MUTEX, SceUID>(threadID, prevCallbackId, mutexWaitTimer);
+	auto result = HLEKernel::WaitBeginCallback<PSPMutex, WAITTYPE_MUTEX, SceUID>(threadID, prevCallbackId, mutexWaitTimer);
 	if (result == HLEKernel::WAIT_CB_SUCCESS)
 		DEBUG_LOG(SCEKERNEL, "sceKernelLockMutexCB: Suspending lock wait for callback");
 	else
@@ -301,7 +296,7 @@ void __KernelMutexBeginCallback(SceUID threadID, SceUID prevCallbackId)

 void __KernelMutexEndCallback(SceUID threadID, SceUID prevCallbackId)
 {
-	auto result = HLEKernel::WaitEndCallback<Mutex, WAITTYPE_MUTEX, SceUID>(threadID, prevCallbackId, mutexWaitTimer, __KernelUnlockMutexForThreadCheck);
+	auto result = HLEKernel::WaitEndCallback<PSPMutex, WAITTYPE_MUTEX, SceUID>(threadID, prevCallbackId, mutexWaitTimer, __KernelUnlockMutexForThreadCheck);
 	if (result == HLEKernel::WAIT_CB_RESUMED_WAIT)
 		DEBUG_LOG(SCEKERNEL, "sceKernelLockMutexCB: Resuming lock wait for callback");
 }
@@ -324,7 +319,7 @@ int sceKernelCreateMutex(const char *name, u32 attr, int initialCount, u32 optio
 	if ((attr & PSP_MUTEX_ATTR_ALLOW_RECURSIVE) == 0 && initialCount > 1)
 		return SCE_KERNEL_ERROR_ILLEGAL_COUNT;

-	Mutex *mutex = new Mutex();
+	PSPMutex *mutex = new PSPMutex();
 	SceUID id = kernelObjects.Create(mutex);

 	mutex->nm.size = sizeof(mutex->nm);
@@ -357,7 +352,7 @@ int sceKernelCreateMutex(const char *name, u32 attr, int initialCount, u32 optio
 int sceKernelDeleteMutex(SceUID id)
 {
 	u32 error;
-	Mutex *mutex = kernelObjects.Get<Mutex>(id, error);
+	PSPMutex *mutex = kernelObjects.Get<PSPMutex>(id, error);
 	if (mutex)
 	{
 		DEBUG_LOG(SCEKERNEL, "sceKernelDeleteMutex(%i)", id);
@@ -373,7 +368,7 @@ int sceKernelDeleteMutex(SceUID id)
 		if (wokeThreads)
 			hleReSchedule("mutex deleted");

-		return kernelObjects.Destroy<Mutex>(id);
+		return kernelObjects.Destroy<PSPMutex>(id);
 	}
 	else
 	{
@@ -382,8 +377,7 @@ int sceKernelDeleteMutex(SceUID id)
 	}
 }

-static bool __KernelLockMutexCheck(Mutex *mutex, int count, u32 &error)
-{
+static bool __KernelLockMutexCheck(PSPMutex *mutex, int count, u32 &error) {
 	if (error)
 		return false;

@@ -411,8 +405,7 @@ static bool __KernelLockMutexCheck(Mutex *mutex, int count, u32 &error)
 	return false;
 }

-static bool __KernelLockMutex(Mutex *mutex, int count, u32 &error)
-{
+static bool __KernelLockMutex(PSPMutex *mutex, int count, u32 &error) {
 	if (!__KernelLockMutexCheck(mutex, count, error))
 		return false;

@@ -433,8 +426,7 @@ static bool __KernelLockMutex(Mutex *mutex, int count, u32 &error)
 	return false;
 }

-static bool __KernelUnlockMutex(Mutex *mutex, u32 &error)
-{
+static bool __KernelUnlockMutex(PSPMutex *mutex, u32 &error) {
 	__KernelMutexEraseLock(mutex);

 	bool wokeThreads = false;
@@ -459,7 +451,7 @@ static bool __KernelUnlockMutex(Mutex *mutex, u32 &error)
 void __KernelMutexTimeout(u64 userdata, int cyclesLate)
 {
 	SceUID threadID = (SceUID)userdata;
-	HLEKernel::WaitExecTimeout<Mutex, WAITTYPE_MUTEX>(threadID);
+	HLEKernel::WaitExecTimeout<PSPMutex, WAITTYPE_MUTEX>(threadID);
 }

 void __KernelMutexThreadEnd(SceUID threadID)
@@ -470,7 +462,7 @@ void __KernelMutexThreadEnd(SceUID threadID)
 	SceUID waitingMutexID = __KernelGetWaitID(threadID, WAITTYPE_MUTEX, error);
 	if (waitingMutexID)
 	{
-		Mutex *mutex = kernelObjects.Get<Mutex>(waitingMutexID, error);
+		PSPMutex *mutex = kernelObjects.Get<PSPMutex>(waitingMutexID, error);
 		if (mutex)
 			HLEKernel::RemoveWaitingThread(mutex->waitingThreads, threadID);
 	}
@@ -481,7 +473,7 @@ void __KernelMutexThreadEnd(SceUID threadID)
 	{
 		// Need to increment early so erase() doesn't invalidate.
 		SceUID mutexID = (*iter++).second;
-		Mutex *mutex = kernelObjects.Get<Mutex>(mutexID, error);
+		PSPMutex *mutex = kernelObjects.Get<PSPMutex>(mutexID, error);

 		if (mutex)
 		{
@@ -491,8 +483,7 @@ void __KernelMutexThreadEnd(SceUID threadID)
 	}
 }

-static void __KernelWaitMutex(Mutex *mutex, u32 timeoutPtr)
-{
+static void __KernelWaitMutex(PSPMutex *mutex, u32 timeoutPtr) {
 	if (timeoutPtr == 0 || mutexWaitTimer == -1)
 		return;

@@ -511,7 +502,7 @@ static void __KernelWaitMutex(Mutex *mutex, u32 timeoutPtr)
 int sceKernelCancelMutex(SceUID uid, int count, u32 numWaitThreadsPtr)
 {
 	u32 error;
-	Mutex *mutex = kernelObjects.Get<Mutex>(uid, error);
+	PSPMutex *mutex = kernelObjects.Get<PSPMutex>(uid, error);
 	if (mutex)
 	{
 		bool lockable = count <= 0 || __KernelLockMutexCheck(mutex, count, error);
@@ -566,7 +557,7 @@ int sceKernelLockMutex(SceUID id, int count, u32 timeoutPtr)
 {
 	DEBUG_LOG(SCEKERNEL, "sceKernelLockMutex(%i, %i, %08x)", id, count, timeoutPtr);
 	u32 error;
-	Mutex *mutex = kernelObjects.Get<Mutex>(id, error);
+	PSPMutex *mutex = kernelObjects.Get<PSPMutex>(id, error);

 	if (__KernelLockMutex(mutex, count, error))
 		return 0;
@@ -591,7 +582,7 @@ int sceKernelLockMutexCB(SceUID id, int count, u32 timeoutPtr)
 {
 	DEBUG_LOG(SCEKERNEL, "sceKernelLockMutexCB(%i, %i, %08x)", id, count, timeoutPtr);
 	u32 error;
-	Mutex *mutex = kernelObjects.Get<Mutex>(id, error);
+	PSPMutex *mutex = kernelObjects.Get<PSPMutex>(id, error);

 	if (!__KernelLockMutexCheck(mutex, count, error))
 	{
@@ -630,7 +621,7 @@ int sceKernelTryLockMutex(SceUID id, int count)
 {
 	DEBUG_LOG(SCEKERNEL, "sceKernelTryLockMutex(%i, %i)", id, count);
 	u32 error;
-	Mutex *mutex = kernelObjects.Get<Mutex>(id, error);
+	PSPMutex *mutex = kernelObjects.Get<PSPMutex>(id, error);

 	if (__KernelLockMutex(mutex, count, error))
 		return 0;
@@ -645,7 +636,7 @@ int sceKernelUnlockMutex(SceUID id, int count)
 {
 	DEBUG_LOG(SCEKERNEL, "sceKernelUnlockMutex(%i, %i)", id, count);
 	u32 error;
-	Mutex *mutex = kernelObjects.Get<Mutex>(id, error);
+	PSPMutex *mutex = kernelObjects.Get<PSPMutex>(id, error);

 	if (error)
 		return error;
@@ -672,7 +663,7 @@ int sceKernelUnlockMutex(SceUID id, int count)
 int sceKernelReferMutexStatus(SceUID id, u32 infoAddr)
 {
 	u32 error;
-	Mutex *m = kernelObjects.Get<Mutex>(id, error);
+	PSPMutex *m = kernelObjects.Get<PSPMutex>(id, error);
 	if (!m)
 	{
 		ERROR_LOG(SCEKERNEL, "sceKernelReferMutexStatus(%i, %08x): invalid mutex id", id, infoAddr);
@@ -1136,4 +1127,4 @@ int sceKernelReferLwMutexStatus(u32 workareaPtr, u32 infoPtr)
 		ERROR_LOG(SCEKERNEL, "%08x=sceKernelReferLwMutexStatus(%08x, %08x)", error, workareaPtr, infoPtr);
 		return error;
 	}
-}
+}
@@ -53,8 +53,7 @@ struct NativeSemaphore
 };


-struct Semaphore : public KernelObject 
-{
+struct PSPSemaphore : public KernelObject {
 	const char *GetName() override { return ns.name; }
 	const char *GetTypeName() override { return "Semaphore"; }

@@ -103,12 +102,11 @@ void __KernelSemaDoState(PointerWrap &p)

 KernelObject *__KernelSemaphoreObject()
 {
-	return new Semaphore;
+	return new PSPSemaphore;
 }

 // Returns whether the thread should be removed.
-static bool __KernelUnlockSemaForThread(Semaphore *s, SceUID threadID, u32 &error, int result, bool &wokeThreads)
-{
+static bool __KernelUnlockSemaForThread(PSPSemaphore *s, SceUID threadID, u32 &error, int result, bool &wokeThreads) {
 	if (!HLEKernel::VerifyWait(threadID, WAITTYPE_SEMA, s->GetUID()))
 		return true;

@@ -139,7 +137,7 @@ static bool __KernelUnlockSemaForThread(Semaphore *s, SceUID threadID, u32 &erro

 void __KernelSemaBeginCallback(SceUID threadID, SceUID prevCallbackId)
 {
-	auto result = HLEKernel::WaitBeginCallback<Semaphore, WAITTYPE_SEMA, SceUID>(threadID, prevCallbackId, semaWaitTimer);
+	auto result = HLEKernel::WaitBeginCallback<PSPSemaphore, WAITTYPE_SEMA, SceUID>(threadID, prevCallbackId, semaWaitTimer);
 	if (result == HLEKernel::WAIT_CB_SUCCESS)
 		DEBUG_LOG(SCEKERNEL, "sceKernelWaitSemaCB: Suspending sema wait for callback");
 	else
@@ -148,15 +146,14 @@ void __KernelSemaBeginCallback(SceUID threadID, SceUID prevCallbackId)

 void __KernelSemaEndCallback(SceUID threadID, SceUID prevCallbackId)
 {
-	auto result = HLEKernel::WaitEndCallback<Semaphore, WAITTYPE_SEMA, SceUID>(threadID, prevCallbackId, semaWaitTimer, __KernelUnlockSemaForThread);
+	auto result = HLEKernel::WaitEndCallback<PSPSemaphore, WAITTYPE_SEMA, SceUID>(threadID, prevCallbackId, semaWaitTimer, __KernelUnlockSemaForThread);
 	if (result == HLEKernel::WAIT_CB_RESUMED_WAIT)
 		DEBUG_LOG(SCEKERNEL, "sceKernelWaitSemaCB: Resuming sema wait for callback");
 }

 // Resume all waiting threads (for delete / cancel.)
 // Returns true if it woke any threads.
-static bool __KernelClearSemaThreads(Semaphore *s, int reason)
-{
+static bool __KernelClearSemaThreads(PSPSemaphore *s, int reason) {
 	u32 error;
 	bool wokeThreads = false;
 	std::vector<SceUID>::iterator iter, end;
@@ -170,7 +167,7 @@ static bool __KernelClearSemaThreads(Semaphore *s, int reason)
 int sceKernelCancelSema(SceUID id, int newCount, u32 numWaitThreadsPtr)
 {
 	u32 error;
-	Semaphore *s = kernelObjects.Get<Semaphore>(id, error);
+	PSPSemaphore *s = kernelObjects.Get<PSPSemaphore>(id, error);
 	if (s)
 	{
 		if (newCount > s->ns.maxCount)
@@ -215,7 +212,7 @@ int sceKernelCreateSema(const char* name, u32 attr, int initVal, int maxVal, u32
 		return SCE_KERNEL_ERROR_ILLEGAL_ATTR;
 	}

-	Semaphore *s = new Semaphore;
+	PSPSemaphore *s = new PSPSemaphore();
 	SceUID id = kernelObjects.Create(s);

 	s->ns.size = sizeof(NativeSemaphore);
@@ -244,7 +241,7 @@ int sceKernelCreateSema(const char* name, u32 attr, int initVal, int maxVal, u32
 int sceKernelDeleteSema(SceUID id)
 {
 	u32 error;
-	Semaphore *s = kernelObjects.Get<Semaphore>(id, error);
+	PSPSemaphore *s = kernelObjects.Get<PSPSemaphore>(id, error);
 	if (s)
 	{
 		DEBUG_LOG(SCEKERNEL, "sceKernelDeleteSema(%i)", id);
@@ -253,7 +250,7 @@ int sceKernelDeleteSema(SceUID id)
 		if (wokeThreads)
 			hleReSchedule("semaphore deleted");

-		return kernelObjects.Destroy<Semaphore>(id);
+		return kernelObjects.Destroy<PSPSemaphore>(id);
 	}
 	else
 	{
@@ -265,7 +262,7 @@ int sceKernelDeleteSema(SceUID id)
 int sceKernelReferSemaStatus(SceUID id, u32 infoPtr)
 {
 	u32 error;
-	Semaphore *s = kernelObjects.Get<Semaphore>(id, error);
+	PSPSemaphore *s = kernelObjects.Get<PSPSemaphore>(id, error);
 	if (s)
 	{
 		DEBUG_LOG(SCEKERNEL, "sceKernelReferSemaStatus(%i, %08x)", id, infoPtr);
@@ -290,7 +287,7 @@ int sceKernelReferSemaStatus(SceUID id, u32 infoPtr)
 int sceKernelSignalSema(SceUID id, int signal)
 {
 	u32 error;
-	Semaphore *s = kernelObjects.Get<Semaphore>(id, error);
+	PSPSemaphore *s = kernelObjects.Get<PSPSemaphore>(id, error);
 	if (s)
 	{
 		if (s->ns.currentCount + signal - (int) s->waitingThreads.size() > s->ns.maxCount)
@@ -336,10 +333,10 @@ void __KernelSemaTimeout(u64 userdata, int cycleslate)
 	u32 error;
 	SceUID uid = __KernelGetWaitID(threadID, WAITTYPE_SEMA, error);

-	HLEKernel::WaitExecTimeout<Semaphore, WAITTYPE_SEMA>(threadID);
+	HLEKernel::WaitExecTimeout<PSPSemaphore, WAITTYPE_SEMA>(threadID);

 	// If in FIFO mode, that may have cleared another thread to wake up.
-	Semaphore *s = kernelObjects.Get<Semaphore>(uid, error);
+	PSPSemaphore *s = kernelObjects.Get<PSPSemaphore>(uid, error);
 	if (s && (s->ns.attr & PSP_SEMA_ATTR_PRIORITY) == PSP_SEMA_ATTR_FIFO) {
 		bool wokeThreads;
 		std::vector<SceUID>::iterator iter = s->waitingThreads.begin();
@@ -351,8 +348,7 @@ void __KernelSemaTimeout(u64 userdata, int cycleslate)
 	}
 }

-static void __KernelSetSemaTimeout(Semaphore *s, u32 timeoutPtr)
-{
+static void __KernelSetSemaTimeout(PSPSemaphore *s, u32 timeoutPtr) {
 	if (timeoutPtr == 0 || semaWaitTimer == -1)
 		return;

@@ -378,7 +374,7 @@ static int __KernelWaitSema(SceUID id, int wantedCount, u32 timeoutPtr, bool pro
 	hleEatCycles(500);

 	u32 error;
-	Semaphore *s = kernelObjects.Get<Semaphore>(id, error);
+	PSPSemaphore *s = kernelObjects.Get<PSPSemaphore>(id, error);
 	if (s)
 	{
 		if (wantedCount > s->ns.maxCount)
@@ -438,7 +434,7 @@ int sceKernelPollSema(SceUID id, int wantedCount)
 	}

 	u32 error;
-	Semaphore *s = kernelObjects.Get<Semaphore>(id, error);
+	PSPSemaphore *s = kernelObjects.Get<PSPSemaphore>(id, error);
 	if (s)
 	{
 		if (s->ns.currentCount >= wantedCount && s->waitingThreads.size() == 0)
@@ -26,7 +26,7 @@
 // There's a good description of the thread scheduling rules in:
 // http://code.google.com/p/jpcsp/source/browse/trunk/src/jpcsp/HLE/modules150/ThreadManForUser.java

-class Thread;
+class PSPThread;
 class DebugInterface;

 int sceKernelChangeThreadPriority(SceUID threadID, int priority);
@@ -119,8 +119,7 @@ typedef void (* WaitEndCallbackFunc)(SceUID threadID, SceUID prevCallbackId);

 void __KernelRegisterWaitTypeFuncs(WaitType type, WaitBeginCallbackFunc beginFunc, WaitEndCallbackFunc endFunc);

-struct ThreadContext
-{
+struct PSPThreadContext {
 	void reset();

 	// r must be followed by f.
@@ -168,8 +167,8 @@ u32 __KernelGetCurThreadStackStart();
 const char *__KernelGetThreadName(SceUID threadID);
 bool KernelIsThreadDormant(SceUID threadID);

-void __KernelSaveContext(ThreadContext *ctx, bool vfpuEnabled);
-void __KernelLoadContext(ThreadContext *ctx, bool vfpuEnabled);
+void __KernelSaveContext(PSPThreadContext *ctx, bool vfpuEnabled);
+void __KernelLoadContext(PSPThreadContext *ctx, bool vfpuEnabled);

 u32 __KernelResumeThreadFromWait(SceUID threadID, u32 retval); // can return an error value
 u32 __KernelResumeThreadFromWait(SceUID threadID, u64 retval);
@@ -205,7 +204,8 @@ void __KernelReturnFromExtendStack();

 void __KernelIdle();

-u32 __KernelMipsCallReturnAddress();
+u32 HLEMipsCallReturnAddress();
+u32 __KernelCallbackReturnAddress();
 u32 __KernelInterruptReturnAddress();  // TODO: remove

 SceUID sceKernelCreateCallback(const char *name, u32 entrypoint, u32 signalArg);
@@ -216,10 +216,10 @@ int sceKernelGetCallbackCount(SceUID cbId);
 void sceKernelCheckCallback();
 int sceKernelReferCallbackStatus(SceUID cbId, u32 statusAddr);

-class Action;
+class PSPAction;

 // Not an official Callback object, just calls a mips function on the current thread.
-void __KernelDirectMipsCall(u32 entryPoint, Action *afterAction, u32 args[], int numargs, bool reschedAfter);
+void __KernelDirectMipsCall(u32 entryPoint, PSPAction *afterAction, u32 args[], int numargs, bool reschedAfter);

 void __KernelReturnFromMipsCall();  // Called as HLE function
 bool __KernelInCallback();
@@ -228,8 +228,8 @@ bool __KernelInCallback();
 bool __KernelCheckCallbacks();
 bool __KernelForceCallbacks();
 bool __KernelCurHasReadyCallbacks();
-void __KernelSwitchContext(Thread *target, const char *reason);
-bool __KernelExecutePendingMipsCalls(Thread *currentThread, bool reschedAfter);
+void __KernelSwitchContext(PSPThread *target, const char *reason);
+bool __KernelExecutePendingMipsCalls(PSPThread *currentThread, bool reschedAfter);
 void __KernelNotifyCallback(SceUID cbId, int notifyArg);

 // Switch to an idle / non-user thread, if not already on one.
@@ -243,8 +243,8 @@ u32 __KernelSetThreadRA(SceUID threadID, u32 nid);

 // A call into game code. These can be pending on a thread.
 // Similar to Callback-s (NOT CallbackInfos) in JPCSP.
-typedef Action *(*ActionCreator)();
-Action *__KernelCreateAction(int actionType);
+typedef PSPAction *(*ActionCreator)();
+PSPAction *__KernelCreateAction(int actionType);
 int __KernelRegisterActionType(ActionCreator creator);
 void __KernelRestoreActionType(int actionType, ActionCreator creator);

@@ -258,7 +258,7 @@ struct MipsCall {
 	u32 cbId;
 	u32 args[6];
 	int numArgs;
-	Action *doAfter;
+	PSPAction *doAfter;
 	u32 savedPc;
 	u32 savedV0;
 	u32 savedV1;
@@ -279,10 +279,10 @@ struct MipsCall {
 	}
 };

-class Action
+class PSPAction
 {
 public:
-	virtual ~Action() {}
+	virtual ~PSPAction() {}
 	virtual void run(MipsCall &call) = 0;
 	virtual void DoState(PointerWrap &p) = 0;
 	int actionTypeID;
@@ -300,7 +300,7 @@ enum ThreadStatus
 	THREADSTATUS_WAITSUSPEND = THREADSTATUS_WAIT | THREADSTATUS_SUSPEND
 };

-void __KernelChangeThreadState(Thread *thread, ThreadStatus newStatus);
+void __KernelChangeThreadState(PSPThread *thread, ThreadStatus newStatus);

 typedef void (*ThreadCallback)(SceUID threadID);
 void __KernelListenThreadEnd(ThreadCallback callback);
@@ -355,11 +355,11 @@ static void AnalyzeMpeg(u8 *buffer, u32 validSize, MpegContext *ctx) {
 	INFO_LOG(ME, "First timestamp: %lld, Last timestamp: %lld", ctx->mpegFirstTimestamp, ctx->mpegLastTimestamp);
 }

-class PostPutAction : public Action {
+class PostPutAction : public PSPAction {
 public:
 	PostPutAction() {}
 	void setRingAddr(u32 ringAddr) { ringAddr_ = ringAddr; }
-	static Action *Create() { return new PostPutAction; }
+	static PSPAction *Create() { return new PostPutAction; }
 	void DoState(PointerWrap &p) override {
 		auto s = p.Section("PostPutAction", 1);
 		if (!s)
@@ -1524,7 +1524,7 @@ static u32 sceMpegRingbufferPut(u32 ringbufferAddr, int numPackets, int availabl
 		int writeOffset = ringbuffer->packetsWritePos % (s32)ringbuffer->packets;
 		u32 packetsThisRound = std::min(numPackets, (s32)ringbuffer->packets - writeOffset);
 		u32 args[3] = {(u32)ringbuffer->data + (u32)writeOffset * 2048, packetsThisRound, (u32)ringbuffer->callback_args};
-		__KernelDirectMipsCall(ringbuffer->callback_addr, action, args, 3, false);
+		hleEnqueueCall(ringbuffer->callback_addr, 3, args, action);
 	} else {
 		ERROR_LOG_REPORT(ME, "sceMpegRingbufferPut: callback_addr zero");
 	}
@@ -1939,6 +1939,14 @@ static u32 sceMpegAvcCsc(u32 mpeg, u32 sourceAddr, u32 rangeAddr, int frameWidth

 	DEBUG_LOG(ME, "sceMpegAvcCsc(%08x, %08x, %08x, %i, %08x)", mpeg, sourceAddr, rangeAddr, frameWidth, destAddr);

+	if (frameWidth == 0) {
+		if (!ctx->defaultFrameWidth) {
+			frameWidth = ctx->avc.avcDetailFrameWidth;
+		} else {
+			frameWidth = ctx->defaultFrameWidth;
+		}
+	}
+
 	int x = Memory::Read_U32(rangeAddr);
 	int y = Memory::Read_U32(rangeAddr + 4);
 	int width = Memory::Read_U32(rangeAddr + 8);
@@ -28,6 +28,7 @@
 #include "Common/ChunkFile.h"
 #include "Core/HLE/HLE.h"
 #include "Core/HLE/FunctionWrappers.h"
+#include "Core/HLE/sceKernelMemory.h"
 #include "Core/MIPS/MIPS.h"
 #include "Core/Config.h"
 #include "Core/MemMapHelpers.h"
@@ -47,6 +48,9 @@ static bool netInetInited;
 static bool netApctlInited;
 u32 netDropRate = 0;
 u32 netDropDuration = 0;
+u32 netPoolAddr = 0;
+u32 netThread1Addr = 0;
+u32 netThread2Addr = 0;

 static struct SceNetMallocStat netMallocStat;

@@ -239,14 +243,13 @@ static void __UpdateApctlHandlers(int oldState, int newState, int flag, int erro

 	for(std::map<int, ApctlHandler>::iterator it = apctlHandlers.begin(); it != apctlHandlers.end(); ++it) {
 		args[4] = it->second.argument;
-
-		__KernelDirectMipsCall(it->second.entryPoint, NULL, args, 5, true);
+		hleEnqueueCall(it->second.entryPoint, 5, args);
 	}
 }

 // This feels like a dubious proposition, mostly...
 void __NetDoState(PointerWrap &p) {
-	auto s = p.Section("sceNet", 1, 2);
+	auto s = p.Section("sceNet", 1, 3);
 	if (!s)
 		return;

@@ -262,6 +265,28 @@ void __NetDoState(PointerWrap &p) {
 		p.Do(netDropRate);
 		p.Do(netDropDuration);
 	}
+	if (s < 3) {
+		netPoolAddr = 0;
+		netThread1Addr = 0;
+		netThread2Addr = 0;
+	} else {
+		p.Do(netPoolAddr);
+		p.Do(netThread1Addr);
+		p.Do(netThread2Addr);
+	}
+}
+
+static inline u32 AllocUser(u32 size, bool fromTop, const char *name) {
+	u32 addr = userMemory.Alloc(size, fromTop, name);
+	if (addr == -1)
+		return 0;
+	return addr;
+}
+
+static inline void FreeUser(u32 &addr) {
+	if (addr != 0)
+		userMemory.Free(addr);
+	addr = 0;
 }

 static u32 sceNetTerm() {
@@ -271,37 +296,76 @@ static u32 sceNetTerm() {

 	WARN_LOG(SCENET, "sceNetTerm()");
 	netInited = false;
+	FreeUser(netPoolAddr);
+	FreeUser(netThread1Addr);
+	FreeUser(netThread2Addr);

 	return 0;
 }

 // TODO: should that struct actually be initialized here?
-static u32 sceNetInit(u32 poolSize, u32 calloutPri, u32 calloutStack, u32 netinitPri, u32 netinitStack)  {
-	// May need to Terminate old one first since the game (ie. GTA:VCS) might not called sceNetTerm before the next sceNetInit and behave strangely
-	if (netInited) 
+static int sceNetInit(u32 poolSize, u32 calloutPri, u32 calloutStack, u32 netinitPri, u32 netinitStack)  {
+	// TODO: The correct behavior is actually to allocate more and leak the other threads/pool.
+	// But we reset here for historic reasons (GTA:VCS potentially triggers this.)
+	if (netInited)
 		sceNetTerm();

+	if (poolSize == 0) {
+		return hleLogError(SCENET, SCE_KERNEL_ERROR_ILLEGAL_MEMSIZE, "invalid pool size");
+	} else if (calloutPri < 0x08 || calloutPri > 0x77) {
+		return hleLogError(SCENET, SCE_KERNEL_ERROR_ILLEGAL_PRIORITY, "invalid callout thread priority");
+	} else if (netinitPri < 0x08 || netinitPri > 0x77) {
+		return hleLogError(SCENET, SCE_KERNEL_ERROR_ILLEGAL_PRIORITY, "invalid init thread priority");
+	}
+
+	// TODO: Should also start the threads, probably?  For now, let's just allocate.
+	// TODO: Respect the stack size if firmware set to 1.50?
+	u32 stackSize = 4096;
+	netThread1Addr = AllocUser(stackSize, true, "netstack1");
+	if (netThread1Addr == 0) {
+		return hleLogError(SCENET, SCE_KERNEL_ERROR_NO_MEMORY, "unable to allocate thread");
+	}
+	netThread2Addr = AllocUser(stackSize, true, "netstack2");
+	if (netThread2Addr == 0) {
+		FreeUser(netThread1Addr);
+		return hleLogError(SCENET, SCE_KERNEL_ERROR_NO_MEMORY, "unable to allocate thread");
+	}
+
+	netPoolAddr = AllocUser(poolSize, false, "netpool");
+	if (netPoolAddr == 0) {
+		FreeUser(netThread1Addr);
+		FreeUser(netThread2Addr);
+		return hleLogError(SCENET, SCE_KERNEL_ERROR_NO_MEMORY, "unable to allocate pool");
+	}
+
 	WARN_LOG(SCENET, "sceNetInit(poolsize=%d, calloutpri=%i, calloutstack=%d, netintrpri=%i, netintrstack=%d) at %08x", poolSize, calloutPri, calloutStack, netinitPri, netinitStack, currentMIPS->pc);
 	netInited = true;
 	netMallocStat.maximum = poolSize;
 	netMallocStat.free = poolSize;
 	netMallocStat.pool = 0;
 	
-	return 0;
+	return hleLogSuccessI(SCENET, 0);
 }

 static u32 sceWlanGetEtherAddr(u32 addrAddr) {
+	if (!Memory::IsValidRange(addrAddr, 6)) {
+		// More correctly, it should crash.
+		return hleLogError(SCENET, SCE_KERNEL_ERROR_ILLEGAL_ADDR, "illegal address");
+	}
+
+	u8 *addr = Memory::GetPointer(addrAddr);
 	// Read MAC Address from config
 	uint8_t mac[6] = {0};
 	if (PPSSPP_ID > 1) {
 		memset(&mac, PPSSPP_ID, sizeof(mac));
-	} else	if (!ParseMacAddress(g_Config.sMACAddress.c_str(), mac)) {
+	} else	if (!ParseMacAddress(g_Config.sMACAddress.c_str(), addr)) {
 		ERROR_LOG(SCENET, "Error parsing mac address %s", g_Config.sMACAddress.c_str());
+		Memory::Memset(addrAddr, 0, 6);
+	} else {
+		CBreakPoints::ExecMemCheck(addrAddr, true, 6, currentMIPS->pc);
 	}
-	DEBUG_LOG(SCENET, "sceWlanGetEtherAddr(%08x)", addrAddr);
-	for (int i = 0; i < 6; i++)
-		Memory::Write_U8(mac[i], addrAddr + i);
-	return 0;
+
+	return hleLogSuccessI(SCENET, hleDelayResult(0, "get ether mac", 200));
 }

 static u32 sceNetGetLocalEtherAddr(u32 addrAddr) {
@@ -309,13 +373,11 @@ static u32 sceNetGetLocalEtherAddr(u32 addrAddr) {
 }

 static u32 sceWlanDevIsPowerOn() {
-	DEBUG_LOG(SCENET, "UNTESTED sceWlanDevIsPowerOn()");
-	return g_Config.bEnableWlan ? 1 : 0;
+	return hleLogSuccessVerboseI(SCENET, g_Config.bEnableWlan ? 1 : 0);
 }

 static u32 sceWlanGetSwitchState() {
-	VERBOSE_LOG(SCENET, "sceWlanGetSwitchState()");
-	return g_Config.bEnableWlan ? 1 : 0;
+	return hleLogSuccessVerboseI(SCENET, g_Config.bEnableWlan ? 1 : 0);
 }

 // Probably a void function, but often returns a useful value.
@@ -638,7 +700,7 @@ static int sceNetSetDropRate(u32 dropRate, u32 dropDuration)
 }

 const HLEFunction sceNet[] = {
-	{0X39AF39A6, &WrapU_UUUUU<sceNetInit>,           "sceNetInit",                      'x', "xxxxx"},
+	{0X39AF39A6, &WrapI_UUUUU<sceNetInit>,           "sceNetInit",                      'i', "xxxxx"},
 	{0X281928A9, &WrapU_V<sceNetTerm>,               "sceNetTerm",                      'x', ""     },
 	{0X89360950, &WrapI_UU<sceNetEtherNtostr>,       "sceNetEtherNtostr",               'i', "xx"   },
 	{0XD27961C9, &WrapI_UU<sceNetEtherStrton>,       "sceNetEtherStrton",               'i', "xx"   },
@@ -160,8 +160,6 @@ void __NetAdhocInit() {
 }

 u32 sceNetAdhocInit() {
-	// Library uninitialized
-	INFO_LOG(SCENET, "sceNetAdhocInit() at %08x", currentMIPS->pc);
 	if (!netAdhocInited) {
 		// Clear Translator Memory
 		memset(&pdp, 0, sizeof(pdp));
@@ -173,17 +171,17 @@ u32 sceNetAdhocInit() {
 		// Create fake PSP Thread for callback
 		// TODO: Should use a separated threads for friendFinder, matchingEvent, and matchingInput and created on AdhocctlInit & AdhocMatchingStart instead of here
 		#define PSP_THREAD_ATTR_KERNEL 0x00001000 // PSP_THREAD_ATTR_KERNEL is located in sceKernelThread.cpp instead of sceKernelThread.h :(
-		//threadAdhocID = __KernelCreateThreadInternal("AdhocThread", __KernelGetCurThreadModuleId(), dummyThreadHackAddr, 0x30, 4096, PSP_THREAD_ATTR_KERNEL);
-		threadAdhocID = __KernelCreateThread("AdhocThread", __KernelGetCurThreadModuleId(), dummyThreadHackAddr, 0x10, 0x1000, 0, 0, false);
+		// TODO: This should probably be a user thread, but maybe from sceNetAdhocctlInit?
+		threadAdhocID = __KernelCreateThread("AdhocThread", __KernelGetCurThreadModuleId(), dummyThreadHackAddr, 0x10, 0x1000, PSP_THREAD_ATTR_KERNEL, 0, true);
 		if (threadAdhocID > 0) {
 			__KernelStartThread(threadAdhocID, 0, 0);
 		}

 		// Return Success
-		return 0;
+		return hleLogSuccessInfoI(SCENET, 0, "at %08x", currentMIPS->pc);
 	}
 	// Already initialized
-	return ERROR_NET_ADHOC_ALREADY_INITIALIZED;
+	return hleLogWarning(SCENET, ERROR_NET_ADHOC_ALREADY_INITIALIZED, "already initialized");
 }

 static u32 sceNetAdhocctlInit(int stackSize, int prio, u32 productAddr) {
@@ -1424,7 +1422,6 @@ int sceNetAdhocctlCreateEnterGameModeMin(const char *group_name, int game_type,
 }

 int sceNetAdhocTerm() {
-	INFO_LOG(SCENET, "sceNetAdhocTerm()");
 	// WLAN might be disabled in the middle of successfull multiplayer, but we still need to cleanup all the sockets right?

 	if (netAdhocctlInited) sceNetAdhocctlTerm();
@@ -1453,10 +1450,11 @@ int sceNetAdhocTerm() {
 		// if (_manage_modules != 0) sceUtilityUnloadModule(PSP_MODULE_NET_INET);
 		// Library shutdown
 		netAdhocInited = false;
-		return 0;
+		return hleLogSuccessInfoI(SCENET, 0);
 	} else {
-		// Seems to return this when called a second time after being terminated without another initialisation
-		return SCE_KERNEL_ERROR_LWMUTEX_NOT_FOUND;
+		// TODO: Reportedly returns SCE_KERNEL_ERROR_LWMUTEX_NOT_FOUND in some cases?
+		// Only seen returning 0 in tests.
+		return hleLogWarning(SCENET, 0, "already uninitialized");
 	}
 }

@@ -234,10 +234,10 @@ public:

 	bool HasReachedEnd() {
 		// The pts are ignored - the end is when we're out of data.
-		return mediaengine->IsVideoEnd() && mediaengine->IsNoAudioData();
+		return mediaengine->IsVideoEnd() && (mediaengine->IsNoAudioData() || !mediaengine->IsActuallyPlayingAudio());
 	}

-	u32 filehandle;
+	int filehandle;
 	u32 fileoffset;
 	int readSize;
 	int streamSize;
@@ -1617,7 +1617,7 @@ static int scePsmfPlayerGetVideoData(u32 psmfPlayer, u32 videoDataAddr)
 	bool doVideoStep = true;
 	if (psmfplayer->playMode == PSMF_PLAYER_MODE_PAUSE) {
 		doVideoStep = false;
-	} else if (!psmfplayer->mediaengine->IsNoAudioData()) {
+	} else if (!psmfplayer->mediaengine->IsNoAudioData() && psmfplayer->mediaengine->IsActuallyPlayingAudio()) {
 		s64 deltapts = psmfplayer->mediaengine->getVideoTimeStamp() - psmfplayer->mediaengine->getAudioTimeStamp();
 		// Don't skip the very first frame, sometimes audio starts with an early timestamp.
 		if (deltapts > 0 && psmfplayer->mediaengine->getVideoTimeStamp() > 0) {
@@ -37,6 +37,13 @@
 #include "Core/HLE/sceKernel.h"
 #include "Core/HLE/sceRtc.h"

+#ifdef HAVE_LIBNX
+// I guess that works...
+#define setenv(x, y, z) (void*)0
+#define tzset() (void*)0
+#define unsetenv(x) (void*)0
+#endif // HAVE_LIBNX
+
 // This is a base time that everything is relative to.
 // This way, time doesn't move strangely with savestates, turbo speed, etc.
 static PSPTimeval rtcBaseTime;
@@ -494,7 +501,7 @@ static int sceRtcConvertLocalTimeToUTC(u32 tickLocalPtr,u32 tickUTCPtr)
 		long timezone_val;
 		_get_timezone(&timezone_val);
 		srcTick -= -timezone_val * 1000000ULL;
-#elif !defined(_AIX) && !defined(__sgi) && !defined(__hpux)
+#elif !defined(_AIX) && !defined(__sgi) && !defined(__hpux) && !defined(HAVE_LIBNX)
 		time_t timezone = 0;
 		tm *time = localtime(&timezone);
 		srcTick -= time->tm_gmtoff*1000000ULL;
@@ -519,7 +526,7 @@ static int sceRtcConvertUtcToLocalTime(u32 tickUTCPtr,u32 tickLocalPtr)
 		long timezone_val;
 		_get_timezone(&timezone_val);
 		srcTick += -timezone_val * 1000000ULL;
-#elif !defined(_AIX) && !defined(__sgi) && !defined(__hpux)
+#elif !defined(_AIX) && !defined(__sgi) && !defined(__hpux) && !defined(HAVE_LIBNX)
 		time_t timezone = 0;
 		tm *time = localtime(&timezone);
 		srcTick += time->tm_gmtoff*1000000ULL;
@@ -1054,7 +1061,7 @@ static int sceRtcFormatRFC2822LocalTime(u32 outPtr, u32 srcTickPtr)
 		long timezone_val;
 		_get_timezone(&timezone_val);
 		tz_seconds = -timezone_val;
-#elif !defined(_AIX) && !defined(__sgi) && !defined(__hpux)
+#elif !defined(_AIX) && !defined(__sgi) && !defined(__hpux) && !defined(HAVE_LIBNX)
 		time_t timezone = 0;
 		tm *time = localtime(&timezone);
 		tz_seconds = time->tm_gmtoff;
@@ -1091,7 +1098,7 @@ static int sceRtcFormatRFC3339LocalTime(u32 outPtr, u32 srcTickPtr)
 		long timezone_val;
 		_get_timezone(&timezone_val);
 		tz_seconds = -timezone_val;
-#elif !defined(_AIX) && !defined(__sgi) && !defined(__hpux)
+#elif !defined(_AIX) && !defined(__sgi) && !defined(__hpux) && !defined(HAVE_LIBNX)
 		time_t timezone = 0;
 		tm *time = localtime(&timezone);
 		tz_seconds = time->tm_gmtoff;
@@ -269,7 +269,7 @@ static u32 sceSasGetEndFlag(u32 core) {
 			endFlag |= (1 << i);
 	}

-	DEBUG_LOG(SCESAS, "%08x=sceSasGetEndFlag(%08x)", endFlag, core);
+	VERBOSE_LOG(SCESAS, "%08x=sceSasGetEndFlag(%08x)", endFlag, core);
 	return endFlag;
 }

@@ -470,55 +470,46 @@ static int sceUtilityOskGetStatus()
 }


-static int sceUtilityNetconfInitStart(u32 paramsAddr)
-{
+static int sceUtilityNetconfInitStart(u32 paramsAddr) {
 	if (currentDialogActive && currentDialogType != UTILITY_DIALOG_NET) {
-		WARN_LOG(SCEUTILITY, "sceUtilityNetconfInitStart(%08x): wrong dialog type", paramsAddr);
-		return SCE_ERROR_UTILITY_WRONG_TYPE;
+		return hleLogWarning(SCEUTILITY, SCE_ERROR_UTILITY_WRONG_TYPE, "wrong dialog type");
 	}
 	
 	oldStatus = 100;
 	currentDialogType = UTILITY_DIALOG_NET;
-	currentDialogActive = true;	
-	int ret = netDialog.Init(paramsAddr);
-	INFO_LOG(SCEUTILITY, "%08x=sceUtilityNetconfInitStart(%08x)", ret, paramsAddr);
-	return ret;
+	currentDialogActive = true;
+	return hleLogSuccessInfoI(SCEUTILITY, netDialog.Init(paramsAddr));
 }

-static int sceUtilityNetconfShutdownStart()
-{
+static int sceUtilityNetconfShutdownStart() {
 	if (currentDialogType != UTILITY_DIALOG_NET) {
-		WARN_LOG(SCEUTILITY, "sceUtilityNetconfShutdownStart(): wrong dialog type");
-		return SCE_ERROR_UTILITY_WRONG_TYPE;
+		return hleLogWarning(SCEUTILITY, SCE_ERROR_UTILITY_WRONG_TYPE, "wrong dialog type");
 	}
 	
 	currentDialogActive = false;
-	int ret = netDialog.Shutdown();
-	DEBUG_LOG(SCEUTILITY, "%08x=sceUtilityNetconfShutdownStart()",ret);
-	return ret;
+	return hleLogSuccessI(SCEUTILITY, netDialog.Shutdown());
 }

-static int sceUtilityNetconfUpdate(int animSpeed)
-{
-	int ret = netDialog.Update(animSpeed);
-	ERROR_LOG(SCEUTILITY, "UNIMPL %08x=sceUtilityNetconfUpdate(%i)", ret, animSpeed);
-	return ret;
-}
-
-static int sceUtilityNetconfGetStatus()
-{
-	// Spam in Danball Senki BOOST
+static int sceUtilityNetconfUpdate(int animSpeed) {
 	if (currentDialogType != UTILITY_DIALOG_NET) {
-		DEBUG_LOG(SCEUTILITY, "sceUtilityNetconfGetStatus(): wrong dialog type");
-		return SCE_ERROR_UTILITY_WRONG_TYPE;
+		return hleLogWarning(SCEUTILITY, SCE_ERROR_UTILITY_WRONG_TYPE, "wrong dialog type");
+	}
+
+	return hleLogSuccessI(SCEUTILITY, netDialog.Update(animSpeed));
+}
+
+static int sceUtilityNetconfGetStatus() {
+	if (currentDialogType != UTILITY_DIALOG_NET) {
+		// Spam in Danball Senki BOOST.
+		return hleLogDebug(SCEUTILITY, SCE_ERROR_UTILITY_WRONG_TYPE, "wrong dialog type");
 	}

 	int status = netDialog.GetStatus();
 	if (oldStatus != status) {
 		oldStatus = status;
-		DEBUG_LOG(SCEUTILITY, "%08x=sceUtilityNetconfGetStatus()", status);
+		return hleLogSuccessI(SCEUTILITY, status);
 	}
-	return status;
+	return hleLogSuccessVerboseI(SCEUTILITY, status);
 }

 static int sceUtilityCheckNetParam(int id)
@@ -769,6 +760,18 @@ static u32 sceUtilityUnloadNetModule(u32 module)
 	return 0;
 }

+static int sceUtilityNpSigninInitStart(u32 paramsPtr) {
+	return hleLogError(SCEUTILITY, 0, "not implemented");
+}
+
+static int sceUtilityNpSigninUpdate(int animSpeed) {
+	return hleLogError(SCEUTILITY, 0, "not implemented");
+}
+
+static int sceUtilityNpSigninGetStatus() {
+	return hleLogError(SCEUTILITY, 0, "not implemented");
+}
+
 static void sceUtilityInstallInitStart(u32 unknown)
 {
 	WARN_LOG_REPORT(SCEUTILITY, "UNIMPL sceUtilityInstallInitStart()");
@@ -934,7 +937,7 @@ const HLEFunction sceUtility[] =

 	{0X0251B134, &WrapI_U<sceUtilityScreenshotInitStart>,          "sceUtilityScreenshotInitStart",          'i', "x"  },
 	{0XF9E0008C, &WrapI_V<sceUtilityScreenshotShutdownStart>,      "sceUtilityScreenshotShutdownStart",      'i', ""   },
-	{0XAB083EA9, &WrapI_U<sceUtilityScreenshotUpdate>,             "sceUtilityScreenshotUpdate",             'i', "x"  },
+	{0XAB083EA9, &WrapI_U<sceUtilityScreenshotUpdate>,             "sceUtilityScreenshotUpdate",             'i', "i"  },
 	{0XD81957B7, &WrapI_V<sceUtilityScreenshotGetStatus>,          "sceUtilityScreenshotGetStatus",          'i', ""   },
 	{0X86A03A27, &WrapI_U<sceUtilityScreenshotContStart>,          "sceUtilityScreenshotContStart",          'i', "x"  },

@@ -947,10 +950,10 @@ const HLEFunction sceUtility[] =
 	{0XB57E95D9, &WrapI_V<sceUtilityGamedataInstallGetStatus>,     "sceUtilityGamedataInstallGetStatus",     'i', ""   },
 	{0X180F7B62, &WrapI_V<sceUtilityGamedataInstallAbort>,         "sceUtilityGamedataInstallAbort",         'i', ""   },

-	{0X16D02AF0, nullptr,                                          "sceUtilityNpSigninInitStart",            '?', ""   },
-	{0XE19C97D6, nullptr,                                          "sceUtilityNpSigninShutdownStart",        '?', ""   },
-	{0XF3FBC572, nullptr,                                          "sceUtilityNpSigninUpdate",               '?', ""   },
-	{0X86ABDB1B, nullptr,                                          "sceUtilityNpSigninGetStatus",            '?', ""   },
+	{0X16D02AF0, &WrapI_U<sceUtilityNpSigninInitStart>,            "sceUtilityNpSigninInitStart",            'i', "x"  },
+	{0XE19C97D6, nullptr,                                          "sceUtilityNpSigninShutdownStart",        'i', ""   },
+	{0XF3FBC572, &WrapI_I<sceUtilityNpSigninUpdate>,               "sceUtilityNpSigninUpdate",               'i', "i"  },
+	{0X86ABDB1B, &WrapI_V<sceUtilityNpSigninGetStatus>,            "sceUtilityNpSigninGetStatus",            'i', ""   },

 	{0X1281DA8E, &WrapV_U<sceUtilityInstallInitStart>,             "sceUtilityInstallInitStart",             'v', "x"  },
 	{0X5EF1C24A, nullptr,                                          "sceUtilityInstallShutdownStart",         '?', ""   },
@@ -231,10 +231,10 @@ void *v4l_loop(void *data) {
 		if (ioctl(v4l_fd, VIDIOC_DQBUF, &buf) == -1) {
 			ERROR_LOG(HLE, "VIDIOC_DQBUF; errno=%d(%s)", errno, strerror(errno));
 			switch (errno) {
-				case EAGAIN:
-					continue;
-				default:
-					return NULL;
+			case EAGAIN:
+				continue;
+			default:
+				return nullptr;
 			}
 		}

@@ -265,9 +265,10 @@ void *v4l_loop(void *data) {
 		buf.memory = V4L2_MEMORY_MMAP;
 		if (ioctl(v4l_fd, VIDIOC_QBUF, &buf) == -1) {
 			ERROR_LOG(HLE, "VIDIOC_QBUF");
-			return NULL;
+			return nullptr;
 		}
 	}
+	return nullptr;
 }

 int __v4l_startCapture(int ideal_width, int ideal_height) {
@@ -338,10 +339,10 @@ int __v4l_startCapture(int ideal_width, int ideal_height) {
 		frmsize.index++;
 		if (frmsize.type == V4L2_FRMSIZE_TYPE_DISCRETE) {
 			INFO_LOG(HLE, "V4L2: frame size supported: %dx%d", frmsize.discrete.width, frmsize.discrete.height);
-			if (frmsize.discrete.width >= ideal_width && frmsize.discrete.height >= ideal_height
-					&& fmt.fmt.pix.width == 0 && fmt.fmt.pix.height == 0
-			 || frmsize.discrete.width >= ideal_width && frmsize.discrete.height >= ideal_height
-					&& frmsize.discrete.width < fmt.fmt.pix.width && frmsize.discrete.height < fmt.fmt.pix.height) {
+			bool matchesIdeal = frmsize.discrete.width >= ideal_width && frmsize.discrete.height >= ideal_height;
+			bool zeroPix = fmt.fmt.pix.width == 0 && fmt.fmt.pix.height == 0;
+			bool pixLarger = frmsize.discrete.width < fmt.fmt.pix.width && frmsize.discrete.height < fmt.fmt.pix.height;
+			if (matchesIdeal && (zeroPix || pixLarger)) {
 				fmt.fmt.pix.width  = frmsize.discrete.width;
 				fmt.fmt.pix.height = frmsize.discrete.height;
 			}
@@ -992,6 +992,10 @@ bool MediaEngine::IsNoAudioData() {
 	return !m_demux->hasNextAudioFrame(NULL, NULL, NULL, NULL);
 }

+bool MediaEngine::IsActuallyPlayingAudio() {
+	return getAudioTimeStamp() >= 0;
+}
+
 s64 MediaEngine::getVideoTimeStamp() {
 	return m_videopts;
 }
@@ -89,6 +89,7 @@ public:

 	bool IsVideoEnd() { return m_isVideoEnd; }
 	bool IsNoAudioData();
+	bool IsActuallyPlayingAudio();
 	int VideoWidth() { return m_desWidth; }
 	int VideoHeight() { return m_desHeight; }

@@ -1022,6 +1022,8 @@ skip:
 	bool ScanForFunctions(u32 startAddr, u32 endAddr, bool insertSymbols) {
 		std::lock_guard<std::recursive_mutex> guard(functions_lock);

+		FunctionsVector new_functions;
+
 		AnalyzedFunction currentFunction = {startAddr};

 		u32 furthestBranch = 0;
@@ -1142,7 +1144,7 @@ skip:
 					}
 				}

-				functions.push_back(currentFunction);
+				new_functions.push_back(currentFunction);

 				furthestBranch = 0;
 				addr += 4;
@@ -1157,10 +1159,10 @@ skip:

 		if (addr <= endAddr) {
 			currentFunction.end = addr + 4;
-			functions.push_back(currentFunction);
+			new_functions.push_back(currentFunction);
 		}

-		for (auto iter = functions.begin(); iter != functions.end(); iter++) {
+		for (auto iter = new_functions.begin(); iter != new_functions.end(); iter++) {
 			iter->size = iter->end - iter->start + 4;
 			if (insertSymbols && !iter->foundInSymbolMap) {
 				char temp[256];
@@ -1168,6 +1170,8 @@ skip:
 			}
 		}

+		// Concatenate the new functions to the end of the old ones.
+		functions.insert(functions.end(), new_functions.begin(), new_functions.end());
 		return insertSymbols;
 	}

@@ -249,7 +249,7 @@ bool Load_PSP_ISO(FileLoader *fileLoader, std::string *error_string) {
 	}

 	bool hasEncrypted = false;
-	u32 fd;
+	int fd;
 	if ((fd = pspFileSystem.OpenFile(bootpath, FILEACCESS_READ)) >= 0)
 	{
 		u8 head[4];
@@ -21,7 +21,7 @@
 #ifdef USING_QT_UI
 #include <QtGui/QImage>
 #else
-#include <libpng17/png.h>
+#include <png.h>
 #include "ext/jpge/jpge.h"
 #endif

@@ -220,6 +220,12 @@ void CPU_Init() {
 		// ERROR_LOG(LOADER, "PBP directory resolution failed.");
 		InitMemoryForGamePBP(loadedFile);
 		break;
+	case IdentifiedFileType::PSP_ELF:
+		if (Memory::g_PSPModel != PSP_MODEL_FAT) {
+			INFO_LOG(LOADER, "ELF, using full PSP-2000 memory access");
+			Memory::g_MemorySize = Memory::RAM_DOUBLE_SIZE;
+		}
+		break;
 	default:
 		break;
 	}
@@ -609,7 +615,7 @@ void InitSysDirectories() {
 		INFO_LOG(COMMON, "Memstick directory not present, creating at '%s'", g_Config.memStickDirectory.c_str());
 	}

-	const std::string testFile = g_Config.memStickDirectory + "/_writable_test.$$$";
+	const std::string testFile = g_Config.memStickDirectory + "_writable_test.$$$";

 	// If any directory is read-only, fall back to the Documents directory.
 	// We're screwed anyway if we can't write to Documents, or can't detect it.
@@ -18,7 +18,7 @@
 #ifdef USING_QT_UI
 #include <QtGui/QImage>
 #else
-#include <libpng17/png.h>
+#include <png.h>
 #endif

 #include <algorithm>
@@ -55,7 +55,7 @@ std::string GameManager::GetTempFilename() const {
 	GetTempFileName(tempPath, L"PSP", 1, buffer);
 	return ConvertWStringToUTF8(buffer);
 #else
-	return g_Config.memStickDirectory + "/ppsspp.dl";
+	return g_Config.memStickDirectory + "ppsspp.dl";
 #endif
 }

@@ -18,6 +18,9 @@
 #include <algorithm>

 #include "base/stringutil.h"
+#include "file/vfs.h"
+#include "gfx/texture_atlas.h"
+#include "gfx_es2/draw_text.h"
 #include "image/zim_load.h"
 #include "image/png_load.h"
 #include "util/text/utf8.h"
@@ -36,6 +39,8 @@
 #include "Core/MemMapHelpers.h"
 #include "Core/System.h"

+Atlas g_ppge_atlas;
+
 static u32 atlasPtr;
 static int atlasWidth;
 static int atlasHeight;
@@ -99,6 +104,38 @@ static AtlasCharLine char_one_line;
 static AtlasLineArray char_lines;
 static AtlasTextMetrics char_lines_metrics;

+static bool textDrawerInited = false;
+static TextDrawer *textDrawer = nullptr;
+struct PPGeTextDrawerCacheKey {
+	bool operator < (const PPGeTextDrawerCacheKey &other) const {
+		if (align != other.align)
+			return align < other.align;
+		if (wrapWidth != other.wrapWidth)
+			return wrapWidth < other.wrapWidth;
+		return text < other.text;
+	}
+	std::string text;
+	int align;
+	float wrapWidth;
+};
+struct PPGeTextDrawerImage {
+	TextStringEntry entry;
+	u32 ptr;
+};
+std::map<PPGeTextDrawerCacheKey, PPGeTextDrawerImage> textDrawerImages;
+
+// Overwrite the current text lines buffer so it can be drawn later.
+void PPGePrepareText(const char *text, float x, float y, int align, float scale, float lineHeightScale,
+	int WrapType = PPGE_LINE_NONE, int wrapWidth = 0);
+
+// These functions must be called between PPGeBegin and PPGeEnd.
+
+// Draw currently buffered text using the state from PPGeGetTextBoundingBox() call.
+// Clears the buffer and state when done.
+void PPGeDrawCurrentText(u32 color = 0xFFFFFFFF);
+
+void PPGeSetTexture(u32 dataAddr, int width, int height);
+
 //only 0xFFFFFF of data is used
 static void WriteCmd(u8 cmd, u32 data) {
 	Memory::Write_U32((cmd << 24) | (data & 0xFFFFFF), dlWritePtr);
@@ -125,18 +162,18 @@ static void BeginVertexData() {
 	vertexStart = dataWritePtr;
 }

-static void Vertex(float x, float y, float u, float v, int tw, int th, u32 color = 0xFFFFFFFF) {
+static void Vertex(float x, float y, float u, float v, int tw, int th, u32 color = 0xFFFFFFFF, float off = -0.5f) {
 	if (g_RemasterMode) {
 		PPGeRemasterVertex vtx;
-		vtx.x = x - 0.5f; vtx.y = y - 0.5f; vtx.z = 0;
-		vtx.u = u * tw - 0.5f; vtx.v = v * th - 0.5f;
+		vtx.x = x + off; vtx.y = y + off; vtx.z = 0;
+		vtx.u = u * tw + off; vtx.v = v * th + off;
 		vtx.color = color;
 		Memory::WriteStruct(dataWritePtr, &vtx);
 		dataWritePtr += sizeof(vtx);
 	} else {
 		PPGeVertex vtx;
-		vtx.x = x - 0.5f; vtx.y = y - 0.5f; vtx.z = 0;
-		vtx.u = u * tw - 0.5f; vtx.v = v * th - 0.5f;
+		vtx.x = x + off; vtx.y = y + off; vtx.z = 0;
+		vtx.u = u * tw + off; vtx.v = v * th + off;
 		vtx.color = color;
 		Memory::WriteStruct(dataWritePtr, &vtx);
 		dataWritePtr += sizeof(vtx);
@@ -171,23 +208,30 @@ void __PPGeSetupListArgs()
 	}
 }

-void __PPGeInit()
-{
+void __PPGeInit() {
 	// PPGe isn't really important for headless, and LoadZIM takes a long time.
-	if (PSP_CoreParameter().gpuCore == GPUCORE_NULL || host->ShouldSkipUI()) {
-		// Let's just not bother.
-		dlPtr = 0;
-		NOTICE_LOG(SCEGE, "Not initializing PPGe - GPU is NullGpu");
-		return;
-	}
-	u8 *imageData[12];
-	int width[12];
-	int height[12];
-	int flags;
-	if (!LoadZIM("ppge_atlas.zim", width, height, &flags, imageData)) {
+	bool skipZIM = PSP_CoreParameter().gpuCore == GPUCORE_NULL || host->ShouldSkipUI();
+
+	u8 *imageData[12]{};
+	int width[12]{};
+	int height[12]{};
+	int flags = 0;
+
+	bool loadedZIM = !skipZIM && LoadZIM("ppge_atlas.zim", width, height, &flags, imageData);
+	if (!skipZIM && !loadedZIM) {
 		PanicAlert("Failed to load ppge_atlas.zim.\n\nPlace it in the directory \"assets\" under your PPSSPP directory.");
 		ERROR_LOG(SCEGE, "PPGe init failed - no atlas texture. PPGe stuff will not be drawn.");
-		return;
+	}
+
+	if (loadedZIM) {
+		size_t atlas_data_size;
+		uint8_t *atlas_data;
+		if (!g_ppge_atlas.IsMetadataLoaded()) {
+			atlas_data = VFSReadFile("ppge_atlas.meta", &atlas_data_size);
+			if (atlas_data)
+				g_ppge_atlas.Load(atlas_data, atlas_data_size);
+			delete[] atlas_data;
+		}
 	}

 	u32 atlasSize = height[0] * width[0] / 2;  // it's a 4-bit paletted texture in ram
@@ -196,7 +240,7 @@ void __PPGeInit()
 	dlPtr = __PPGeDoAlloc(dlSize, false, "PPGe Display List");
 	dataPtr = __PPGeDoAlloc(dataSize, false, "PPGe Vertex Data");
 	__PPGeSetupListArgs();
-	atlasPtr = __PPGeDoAlloc(atlasSize, false, "PPGe Atlas Texture");
+	atlasPtr = atlasSize == 0 ? 0 : __PPGeDoAlloc(atlasSize, false, "PPGe Atlas Texture");
 	palette = __PPGeDoAlloc(paletteSize, false, "PPGe Texture Palette");

 	// Generate 16-greyscale palette. All PPGe graphics are greyscale so we can use a tiny paletted texture.
@@ -206,7 +250,7 @@ void __PPGeInit()
 	}

 	const u32_le *imagePtr = (u32_le *)imageData[0];
-	u8 *ramPtr = (u8 *)Memory::GetPointer(atlasPtr);
+	u8 *ramPtr = atlasPtr == 0 ? nullptr : (u8 *)Memory::GetPointer(atlasPtr);

 	// Palettize to 4-bit, the easy way.
 	for (int i = 0; i < width[0] * height[0] / 2; i++) {
@@ -221,13 +265,18 @@ void __PPGeInit()
 	
 	free(imageData[0]);

+	// We can't create it here, because Android needs it on the right thread.
+	textDrawerInited = false;
+	textDrawer = nullptr;
+	textDrawerImages.clear();
+
 	DEBUG_LOG(SCEGE, "PPGe drawing library initialized. DL: %08x Data: %08x Atlas: %08x (%i) Args: %08x",
 		dlPtr, dataPtr, atlasPtr, atlasSize, listArgs.ptr);
 }

 void __PPGeDoState(PointerWrap &p)
 {
-	auto s = p.Section("PPGeDraw", 1, 2);
+	auto s = p.Section("PPGeDraw", 1, 3);
 	if (!s)
 		return;

@@ -245,6 +294,30 @@ void __PPGeDoState(PointerWrap &p)
 		p.Do(listArgs);
 	}

+	if (s >= 3) {
+		uint32_t sz = (uint32_t)textDrawerImages.size();
+		p.Do(sz);
+
+		switch (p.mode) {
+		case PointerWrap::MODE_READ:
+			textDrawerImages.clear();
+			for (uint32_t i = 0; i < sz; ++i) {
+				// We only care about the pointers, so we can free them.  We'll decimate right away.
+				PPGeTextDrawerCacheKey key{ StringFromFormat("__savestate__%d", i), -1, -1 };
+				textDrawerImages[key] = PPGeTextDrawerImage{};
+				p.Do(textDrawerImages[key].ptr);
+			}
+			break;
+		default:
+			for (const auto &im : textDrawerImages) {
+				p.Do(im.second.ptr);
+			}
+			break;
+		}
+	} else {
+		textDrawerImages.clear();
+	}
+
 	p.Do(dlPtr);
 	p.Do(dlWritePtr);
 	p.Do(dlSize);
@@ -280,6 +353,9 @@ void __PPGeShutdown()
 	dlPtr = 0;
 	savedContextPtr = 0;
 	listArgs = 0;
+
+	delete textDrawer;
+	textDrawer = nullptr;
 }

 void PPGeBegin()
@@ -614,31 +690,63 @@ static AtlasTextMetrics BreakLines(const char *text, const AtlasFont &atlasfont,
 	return metrics;
 }

-void PPGeMeasureText(float *w, float *h, int *n, 
-					const char *text, float scale, int WrapType, int wrapWidth)
-{
-	const AtlasFont &atlasfont = *ppge_atlas.fonts[0];
+static bool HasTextDrawer() {
+	// We create this on first use so it's on the correct thread.
+	if (textDrawerInited) {
+		return textDrawer != nullptr;
+	}
+
+	// TODO: Should we pass a draw_?
+	textDrawer = TextDrawer::Create(nullptr);
+	if (textDrawer) {
+		textDrawer->SetFontScale(1.0f, 1.0f);
+		textDrawer->SetForcedDPIScale(1.0f);
+		textDrawer->SetFont(g_Config.sFont.c_str(), 18, 0);
+	}
+	textDrawerInited = true;
+	return textDrawer != nullptr;
+}
+
+void PPGeMeasureText(float *w, float *h, const char *text, float scale, int WrapType, int wrapWidth) {
+	if (HasTextDrawer()) {
+		float mw, mh;
+		textDrawer->SetFontScale(scale, scale);
+		int dtalign = (WrapType & PPGE_LINE_WRAP_WORD) ? FLAG_WRAP_TEXT : 0;
+		if (WrapType & PPGE_LINE_USE_ELLIPSIS)
+			dtalign |= FLAG_ELLIPSIZE_TEXT;
+		Bounds b(0, 0, wrapWidth <= 0 ? 480.0f : wrapWidth, 272.0f);
+		textDrawer->MeasureStringRect(text, strlen(text), b, &mw, &mh, dtalign);
+
+		if (w)
+			*w = mw;
+		if (h)
+			*h = mh;
+		return;
+	}
+
+	if (!g_ppge_atlas.IsMetadataLoaded() || g_ppge_atlas.num_fonts < 1) {
+		if (w)
+			*w = 0;
+		if (h)
+			*h = 0;
+		return;
+	}
+
+	const AtlasFont &atlasfont = g_ppge_atlas.fonts[0];
 	AtlasTextMetrics metrics = BreakLines(text, atlasfont, 0, 0, 0, scale, scale, WrapType, wrapWidth, true);
 	if (w) *w = metrics.maxWidth;
-	if (h) *h = metrics.lineHeight;
-	if (n) *n = metrics.numLines;
+	if (h) *h = metrics.lineHeight * metrics.numLines;
 }

 void PPGePrepareText(const char *text, float x, float y, int align, float scale, float lineHeightScale, int WrapType, int wrapWidth)
 {
-	const AtlasFont &atlasfont = *ppge_atlas.fonts[0];
+	const AtlasFont &atlasfont = g_ppge_atlas.fonts[0];
+	if (!g_ppge_atlas.IsMetadataLoaded() || g_ppge_atlas.num_fonts < 1) {
+		return;
+	}
 	char_lines_metrics = BreakLines(text, atlasfont, x, y, align, scale, lineHeightScale, WrapType, wrapWidth, false);
 }

-void PPGeMeasureCurrentText(float *x, float *y, float *w, float *h, int *n)
-{
-	if (x) *x = char_lines_metrics.x;
-	if (y) *y = char_lines_metrics.y;
-	if (w) *w = char_lines_metrics.maxWidth;
-	if (h) *h = char_lines_metrics.lineHeight;
-	if (n) *n = char_lines_metrics.numLines;
-}
-
 static void PPGeResetCurrentText() {
 	char_one_line.clear();
 	char_lines.clear();
@@ -672,8 +780,110 @@ void PPGeDrawCurrentText(u32 color)
 	PPGeResetCurrentText();
 }

-void PPGeDrawText(const char *text, float x, float y, int align, float scale, u32 color)
-{
+// Return a value such that (1 << value) >= x
+int GetPow2(int x) {
+#ifdef __GNUC__
+	int ret = 31 - __builtin_clz(x | 1);
+	if ((1 << ret) < x)
+#else
+	int ret = 0;
+	while ((1 << ret) < x)
+#endif
+		ret++;
+	return ret;
+}
+
+static PPGeTextDrawerImage PPGeGetTextImage(const char *text, int align, float scale, float maxWidth, bool wrap) {
+	int tdalign = (align & PPGE_ALIGN_HCENTER) ? ALIGN_HCENTER : 0;
+	tdalign |= FLAG_ELLIPSIZE_TEXT;
+	if (wrap) {
+		tdalign |= FLAG_WRAP_TEXT;
+	}
+
+	PPGeTextDrawerCacheKey key{ text, tdalign, maxWidth / scale };
+	PPGeTextDrawerImage im;
+
+	auto cacheItem = textDrawerImages.find(key);
+	if (cacheItem != textDrawerImages.end()) {
+		im = cacheItem->second;
+		cacheItem->second.entry.lastUsedFrame = gpuStats.numFlips;
+	} else {
+		std::vector<uint8_t> bitmapData;
+		textDrawer->SetFontScale(scale, scale);
+		Bounds b(0, 0, maxWidth, 272.0f);
+		std::string cleaned = ReplaceAll(text, "\r", "");
+		textDrawer->DrawStringBitmapRect(bitmapData, im.entry, Draw::DataFormat::R8_UNORM, cleaned.c_str(), b, tdalign);
+
+		int bufwBytes = ((im.entry.bmWidth + 31) / 32) * 16;
+		u32 sz = bufwBytes * (im.entry.bmHeight + 1);
+		u32 origSz = sz;
+		im.ptr = __PPGeDoAlloc(sz, true, "PPGeText");
+
+		if (bitmapData.size() & 1)
+			bitmapData.resize(bitmapData.size() + 1);
+
+		if (im.ptr) {
+			int wBytes = (im.entry.bmWidth + 1) / 2;
+			u8 *ramPtr = (u8 *)Memory::GetPointer(im.ptr);
+			for (int y = 0; y < im.entry.bmHeight; ++y) {
+				for (int x = 0; x < wBytes; ++x) {
+					uint8_t c1 = bitmapData[y * im.entry.bmWidth + x * 2];
+					uint8_t c2 = bitmapData[y * im.entry.bmWidth + x * 2 + 1];
+					// Convert this to 4-bit palette values.
+					ramPtr[y * bufwBytes + x] = (c2 & 0xF0) | (c1 >> 4);
+				}
+				if (bufwBytes != wBytes) {
+					memset(ramPtr + y * bufwBytes + wBytes, 0, bufwBytes - wBytes);
+				}
+			}
+			memset(ramPtr + im.entry.bmHeight * bufwBytes, 0, bufwBytes + sz - origSz);
+		}
+
+		im.entry.lastUsedFrame = gpuStats.numFlips;
+		textDrawerImages[key] = im;
+	}
+
+	return im;
+}
+
+static void PPGeDrawTextImage(PPGeTextDrawerImage im, float x, float y, int align, float scale, u32 color) {
+	int bufw = ((im.entry.bmWidth + 31) / 32) * 32;
+	int wp2 = GetPow2(im.entry.bmWidth);
+	int hp2 = GetPow2(im.entry.bmHeight);
+	WriteCmd(GE_CMD_TEXADDR0, im.ptr & 0xFFFFF0);
+	WriteCmd(GE_CMD_TEXBUFWIDTH0, bufw | ((im.ptr & 0xFF000000) >> 8));
+	WriteCmd(GE_CMD_TEXSIZE0, wp2 | (hp2 << 8));
+	WriteCmd(GE_CMD_TEXFLUSH, 0);
+
+	float w = im.entry.width * scale;
+	float h = im.entry.height * scale;
+
+	if (align & PPGE_ALIGN_HCENTER)
+		x -= w / 2.0f;
+	else if (align & PPGE_ALIGN_RIGHT)
+		x -= w;
+	if (align & PPGE_ALIGN_VCENTER)
+		y -= h / 2.0f;
+	else if (align & PPGE_ALIGN_BOTTOM)
+		y -= h;
+
+	BeginVertexData();
+	float u1 = (float)im.entry.width / (1 << wp2);
+	float v1 = (float)im.entry.height / (1 << hp2);
+	Vertex(x, y, 0, 0, 1 << wp2, 1 << hp2, color, 0.0f);
+	Vertex(x + w, y + h, u1, v1, 1 << wp2, 1 << hp2, color, 0.0f);
+	EndVertexDataAndDraw(GE_PRIM_RECTANGLES);
+
+	PPGeSetDefaultTexture();
+}
+
+void PPGeDrawText(const char *text, float x, float y, int align, float scale, u32 color) {
+	if (HasTextDrawer()) {
+		PPGeTextDrawerImage im = PPGeGetTextImage(text, align, scale, 480.0f - x, false);
+		PPGeDrawTextImage(im, x, y, align, scale, color);
+		return;
+	}
+
 	PPGePrepareText(text, x, y, align, scale, scale, PPGE_LINE_USE_ELLIPSIS);
 	PPGeDrawCurrentText(color);
 }
@@ -707,10 +917,40 @@ void PPGeDrawTextWrapped(const char *text, float x, float y, float wrapWidth, fl
 		s = StripTrailingWhite(s);
 	}

-	PPGePrepareText(s.c_str(), x, y, align, scale, scale, PPGE_LINE_USE_ELLIPSIS | PPGE_LINE_WRAP_WORD, wrapWidth);
-
 	int zoom = (PSP_CoreParameter().pixelHeight + 479) / 480;
 	float maxScaleDown = zoom == 1 ? 1.3f : 2.0f;
+
+	if (HasTextDrawer()) {
+		float actualWidth, actualHeight;
+		Bounds b(0, 0, wrapWidth <= 0 ? 480.0f - x : wrapWidth, wrapHeight);
+		int tdalign = (align & PPGE_ALIGN_HCENTER) ? ALIGN_HCENTER : 0;
+		textDrawer->SetFontScale(scale, scale);
+		textDrawer->MeasureStringRect(s.c_str(), s.size(), b, &actualWidth, &actualHeight, tdalign | FLAG_WRAP_TEXT);
+		if (wrapHeight != 0.0f && actualHeight > wrapHeight) {
+			// Cheap way to get the line height.
+			float oneLine, twoLines;
+			textDrawer->MeasureString("|", 1, &actualWidth, &oneLine);
+			textDrawer->MeasureStringRect("|\n|", 3, Bounds(0, 0, 480, 272), &actualWidth, &twoLines);
+
+			float lineHeight = twoLines - oneLine;
+			if (actualHeight > wrapHeight * maxScaleDown) {
+				float maxLines = floor(wrapHeight * maxScaleDown / lineHeight);
+				actualHeight = (maxLines + 1) * lineHeight;
+				// Add an ellipsis if it's just too long to be readable.
+				// On a PSP, it does this without scaling it down.
+				s = StripTrailingWhite(CropLinesToCount(s, (int)maxLines)) + "\n...";
+			}
+
+			scale *= wrapHeight / actualHeight;
+		}
+
+		PPGeTextDrawerImage im = PPGeGetTextImage(s.c_str(), align, scale, wrapWidth <= 0 ? 480.0f - x : wrapWidth, true);
+		PPGeDrawTextImage(im, x, y, align, scale, color);
+		return;
+	}
+
+	PPGePrepareText(s.c_str(), x, y, align, scale, scale, PPGE_LINE_USE_ELLIPSIS | PPGE_LINE_WRAP_WORD, wrapWidth);
+
 	float actualHeight = char_lines_metrics.lineHeight * char_lines_metrics.numLines;
 	if (wrapHeight != 0.0f && actualHeight > wrapHeight) {
 		if (actualHeight > wrapHeight * maxScaleDown) {
@@ -731,15 +971,16 @@ void PPGeDrawTextWrapped(const char *text, float x, float y, float wrapWidth, fl
 }

 // Draws a "4-patch" for button-like things that can be resized
-void PPGeDraw4Patch(int atlasImage, float x, float y, float w, float h, u32 color)
-{
+void PPGeDraw4Patch(ImageID atlasImage, float x, float y, float w, float h, u32 color) {
 	if (!dlPtr)
 		return;
-	const AtlasImage &img = ppge_images[atlasImage];
-	float borderx = img.w / 20;
-	float bordery = img.h / 20;
-	float u1 = img.u1, uhalf = (img.u1 + img.u2) / 2, u2 = img.u2;
-	float v1 = img.v1, vhalf = (img.v1 + img.v2) / 2, v2 = img.v2;
+	const AtlasImage *img = g_ppge_atlas.getImage(atlasImage);
+	if (!img)
+		return;
+	float borderx = img->w / 20;
+	float bordery = img->h / 20;
+	float u1 = img->u1, uhalf = (img->u1 + img->u2) / 2, u2 = img->u2;
+	float v1 = img->v1, vhalf = (img->v1 + img->v2) / 2, v2 = img->v2;
 	float xmid1 = x + borderx;
 	float xmid2 = x + w - borderx;
 	float ymid1 = y + bordery;
@@ -771,8 +1012,7 @@ void PPGeDraw4Patch(int atlasImage, float x, float y, float w, float h, u32 colo
 	EndVertexDataAndDraw(GE_PRIM_RECTANGLES);
 }

-void PPGeDrawRect(float x1, float y1, float x2, float y2, u32 color)
-{
+void PPGeDrawRect(float x1, float y1, float x2, float y2, u32 color) {
 	if (!dlPtr)
 		return;

@@ -787,29 +1027,35 @@ void PPGeDrawRect(float x1, float y1, float x2, float y2, u32 color)
 }

 // Just blits an image to the screen, multiplied with the color.
-void PPGeDrawImage(int atlasImage, float x, float y, int align, u32 color)
+void PPGeDrawImage(ImageID atlasImage, float x, float y, int align, u32 color)
 {
 	if (!dlPtr)
 		return;

-	const AtlasImage &img = ppge_atlas.images[atlasImage];
-	float w = img.w;
-	float h = img.h;
+	const AtlasImage *img = g_ppge_atlas.getImage(atlasImage);
+	if (!img) {
+		return;
+	}
+	float w = img->w;
+	float h = img->h;
 	BeginVertexData();
-	Vertex(x, y, img.u1, img.v1, atlasWidth, atlasHeight, color);
-	Vertex(x + w, y + h, img.u2, img.v2, atlasWidth, atlasHeight, color);
+	Vertex(x, y, img->u1, img->v1, atlasWidth, atlasHeight, color);
+	Vertex(x + w, y + h, img->u2, img->v2, atlasWidth, atlasHeight, color);
 	EndVertexDataAndDraw(GE_PRIM_RECTANGLES);
 }

-void PPGeDrawImage(int atlasImage, float x, float y, float w, float h, int align, u32 color)
+void PPGeDrawImage(ImageID atlasImage, float x, float y, float w, float h, int align, u32 color)
 {
 	if (!dlPtr)
 		return;

-	const AtlasImage &img = ppge_atlas.images[atlasImage];
+	const AtlasImage *img = g_ppge_atlas.getImage(atlasImage);
+	if (!img) {
+		return;
+	}
 	BeginVertexData();
-	Vertex(x, y, img.u1, img.v1, atlasWidth, atlasHeight, color);
-	Vertex(x + w, y + h, img.u2, img.v2, atlasWidth, atlasHeight, color);
+	Vertex(x, y, img->u1, img->v1, atlasWidth, atlasHeight, color);
+	Vertex(x + w, y + h, img->u2, img->v2, atlasWidth, atlasHeight, color);
 	EndVertexDataAndDraw(GE_PRIM_RECTANGLES);
 }

@@ -823,20 +1069,6 @@ void PPGeDrawImage(float x, float y, float w, float h, float u1, float v1, float
 	EndVertexDataAndDraw(GE_PRIM_RECTANGLES);
 }

-// Return a value such that (1 << value) >= x
-int GetPow2(int x)
-{
-#ifdef __GNUC__
-	int ret = 31 - __builtin_clz(x|1);
-	if ((1 << ret) < x)
-#else
-	int ret = 0;
-	while ((1 << ret) < x)
-#endif
-		ret++;
-	return ret;
-}
-
 void PPGeSetDefaultTexture()
 {
 	WriteCmd(GE_CMD_TEXTUREMAPENABLE, 1);
@@ -990,3 +1222,20 @@ void PPGeImage::SetTexture() {
 	}
 }

+void PPGeNotifyFrame() {
+	if (textDrawer) {
+		textDrawer->OncePerFrame();
+	}
+
+	// Do this always, in case the platform has no TextDrawer but save state did.
+	for (auto it = textDrawerImages.begin(); it != textDrawerImages.end(); ) {
+		if (it->second.entry.lastUsedFrame - gpuStats.numFlips >= 97) {
+			kernelMemory.Free(it->second.ptr);
+			it = textDrawerImages.erase(it);
+		} else {
+			++it;
+		}
+	}
+
+	PPGeImage::Decimate();
+}
@@ -19,7 +19,9 @@

 #include <vector>
 #include <string>
-#include "ppge_atlas.h"
+
+#include "gfx/texture_atlas.h"
+
 #include "Common/CommonTypes.h"

 class PointerWrap;
@@ -43,10 +45,6 @@ void __PPGeShutdown();
 void PPGeBegin();
 void PPGeEnd();

-// If you want to draw using this texture but not go through the PSP GE emulation,
-// jsut call this. Will bind the texture to unit 0.
-void PPGeBindTexture();
-
 enum {
 	PPGE_ALIGN_LEFT        = 0,
 	PPGE_ALIGN_RIGHT       = 16,
@@ -63,11 +61,6 @@ enum {
 	PPGE_ALIGN_BOTTOMRIGHT = PPGE_ALIGN_BOTTOM | PPGE_ALIGN_RIGHT,
 };

-enum {
-	PPGE_ESCAPE_NONE,
-	PPGE_ESCAPE_BACKSLASHED,
-};
-
 enum {
 	PPGE_LINE_NONE         = 0,
 	PPGE_LINE_USE_ELLIPSIS = 1, // use ellipses in too long words
@@ -76,21 +69,7 @@ enum {
 };

 // Get the metrics of the bounding box of the text without changing the buffer or state.
-void PPGeMeasureText(float *w, float *h, int *n, 
-					const char *text, float scale, int WrapType = PPGE_LINE_NONE, int wrapWidth = 0);
-
-// Overwrite the current text lines buffer so it can be drawn later.
-void PPGePrepareText(const char *text, float x, float y, int align, float scale, float lineHeightScale,
-					int WrapType = PPGE_LINE_NONE, int wrapWidth = 0);
-
-// Get the metrics of the bounding box of the currently stated text.
-void PPGeMeasureCurrentText(float *x, float *y, float *w, float *h, int *n);
-
-// These functions must be called between PPGeBegin and PPGeEnd.
-
-// Draw currently buffered text using the state from PPGeGetTextBoundingBox() call.
-// Clears the buffer and state when done.
-void PPGeDrawCurrentText(u32 color = 0xFFFFFFFF);
+void PPGeMeasureText(float *w, float *h, const char *text, float scale, int WrapType = PPGE_LINE_NONE, int wrapWidth = 0);

 // Draws some text using the one font we have.
 // Clears the text buffer when done.
@@ -98,13 +77,15 @@ void PPGeDrawText(const char *text, float x, float y, int align, float scale = 1
 void PPGeDrawTextWrapped(const char *text, float x, float y, float wrapWidth, float wrapHeight, int align, float scale = 1.0f, u32 color = 0xFFFFFFFF);

 // Draws a "4-patch" for button-like things that can be resized.
-void PPGeDraw4Patch(int atlasImage, float x, float y, float w, float h, u32 color = 0xFFFFFFFF);
+void PPGeDraw4Patch(ImageID atlasImage, float x, float y, float w, float h, u32 color = 0xFFFFFFFF);

 // Just blits an image to the screen, multiplied with the color.
-void PPGeDrawImage(int atlasImage, float x, float y, int align, u32 color = 0xFFFFFFFF);
-void PPGeDrawImage(int atlasImage, float x, float y, float w, float h, int align, u32 color = 0xFFFFFFFF);
+void PPGeDrawImage(ImageID atlasImage, float x, float y, int align, u32 color = 0xFFFFFFFF);
+void PPGeDrawImage(ImageID atlasImage, float x, float y, float w, float h, int align, u32 color = 0xFFFFFFFF);
 void PPGeDrawImage(float x, float y, float w, float h, float u1, float v1, float u2, float v2, int tw, int th, u32 color);

+void PPGeNotifyFrame();
+
 class PPGeImage {
 public:
 	PPGeImage(const std::string &pspFilename);
@@ -130,8 +111,9 @@ public:
 		return height_;
 	}

-private:
 	static void Decimate();
+
+private:
 	static std::vector<PPGeImage *> loadedTextures_;

 	std::string filename_;
@@ -150,6 +132,5 @@ private:
 void PPGeDrawRect(float x1, float y1, float x2, float y2, u32 color);

 void PPGeSetDefaultTexture();
-void PPGeSetTexture(u32 dataAddr, int width, int height);
 void PPGeDisableTexture();

@@ -1,18 +0,0 @@
-// Header generated by atlastool from ppge_atlasscript.txt (hrydgard@gmail.com)
-
-#pragma once
-#include "gfx/texture_atlas.h"
-
-// FONTS_ppge
-#define UBUNTU24 0
-
-
-// IMAGES_ppge
-#define I_CROSS 0
-#define I_CIRCLE 1
-#define I_SQUARE 2
-#define I_TRIANGLE 3
-
-
-extern const Atlas ppge_atlas;
-extern const AtlasImage ppge_images[4];
@@ -37,6 +37,8 @@ DrawEngineCommon::DrawEngineCommon() : decoderMap_(16) {
 	decJitCache_ = new VertexDecoderJitCache();
 	transformed = (TransformedVertex *)AllocateMemoryPages(TRANSFORMED_VERTEX_BUFFER_SIZE, MEM_PROT_READ | MEM_PROT_WRITE);
 	transformedExpanded = (TransformedVertex *)AllocateMemoryPages(3 * TRANSFORMED_VERTEX_BUFFER_SIZE, MEM_PROT_READ | MEM_PROT_WRITE);
+	useHWTransform_ = g_Config.bHardwareTransform;
+	useHWTessellation_ = UpdateUseHWTessellation(g_Config.bHardwareTessellation);
 }

 DrawEngineCommon::~DrawEngineCommon() {
@@ -171,6 +173,9 @@ void DrawEngineCommon::Resized() {
 	});
 	decoderMap_.Clear();
 	ClearTrackedVertexArrays();
+
+	useHWTransform_ = g_Config.bHardwareTransform;
+	useHWTessellation_ = UpdateUseHWTessellation(g_Config.bHardwareTessellation);
 }

 u32 DrawEngineCommon::NormalizeVertices(u8 *outPtr, u8 *bufPtr, const u8 *inPtr, int lowerBound, int upperBound, u32 vertType, int *vertexSize) {
@@ -737,6 +742,19 @@ void DrawEngineCommon::SubmitPrim(void *verts, void *inds, GEPrimitiveType prim,
 	}
 }

+bool DrawEngineCommon::CanUseHardwareTransform(int prim) {
+	if (!useHWTransform_)
+		return false;
+	return !gstate.isModeThrough() && prim != GE_PRIM_RECTANGLES;
+}
+
+bool DrawEngineCommon::CanUseHardwareTessellation(GEPatchPrimType prim) {
+	if (useHWTessellation_) {
+		return CanUseHardwareTransform(PatchPrimToPrim(prim));
+	}
+	return false;
+}
+
 void TessellationDataTransfer::CopyControlPoints(float *pos, float *tex, float *col, int posStride, int texStride, int colStride, const SimpleVertex *const *points, int size, u32 vertType) {
 	bool hasColor = (vertType & GE_VTYPE_COL_MASK) != 0;
 	bool hasTexCoord = (vertType & GE_VTYPE_TC_MASK) != 0;
@@ -86,6 +86,9 @@ public:
 	void SubmitCurve(const void *control_points, const void *indices, Surface &surface, u32 vertType, int *bytesRead, const char *scope);
 	void ClearSplineBezierWeights();

+	bool CanUseHardwareTransform(int prim);
+	bool CanUseHardwareTessellation(GEPatchPrimType prim);
+
 	std::vector<std::string> DebugGetVertexLoaderIDs();
 	std::string DebugGetVertexLoaderString(std::string id, DebugShaderStringType stringType);

@@ -101,6 +104,7 @@ public:
 	VertexDecoder *GetVertexDecoder(u32 vtype);

 protected:
+	virtual bool UpdateUseHWTessellation(bool enabled) { return enabled; }
 	virtual void ClearTrackedVertexArrays() {}

 	int ComputeNumVertsToDecode() const;
@@ -128,6 +132,9 @@ protected:
 		return 1;
 	}

+	bool useHWTransform_ = false;
+	bool useHWTessellation_ = false;
+
 	// Vertex collector buffers
 	u8 *decoded = nullptr;
 	u16 *decIndex = nullptr;
@@ -236,6 +236,10 @@ void FramebufferManagerCommon::EstimateDrawingSize(u32 fb_address, GEBufferForma
 			drawing_width = scissor_width;
 			drawing_height = std::max(drawing_height, scissor_height);
 		}
+		if (scissor_width == 481 && region_width == 480 && scissor_height == 273 && region_height == 272) {
+			drawing_width = 480;
+			drawing_height = 272;
+		}
 	} else {
 		// If viewport wasn't valid, let's just take the greatest anything regardless of stride.
 		drawing_width = std::min(std::max(region_width, scissor_width), fb_stride);
@@ -848,7 +852,7 @@ void FramebufferManagerCommon::SetViewport2D(int x, int y, int w, int h) {
 	draw_->SetViewports(1, &vp);
 }

-void FramebufferManagerCommon::CopyDisplayToOutput() {
+void FramebufferManagerCommon::CopyDisplayToOutput(bool reallyDirty) {
 	DownloadFramebufferOnSwitch(currentRenderVfb_);
 	shaderManager_->DirtyLastShader();

@@ -871,12 +875,16 @@ void FramebufferManagerCommon::CopyDisplayToOutput() {
 	CardboardSettings cardboardSettings;
 	GetCardboardSettings(&cardboardSettings);

-	VirtualFramebuffer *vfb = GetVFBAt(displayFramebufPtr_);
+	// If it's not really dirty, we're probably frameskipping.  Use the last working one.
+	u32 fbaddr = reallyDirty ? displayFramebufPtr_ : prevDisplayFramebufPtr_;
+	prevDisplayFramebufPtr_ = fbaddr;
+
+	VirtualFramebuffer *vfb = GetVFBAt(fbaddr);
 	if (!vfb) {
 		// Let's search for a framebuf within this range. Note that we also look for
 		// "framebuffers" sitting in RAM (created from block transfer or similar) so we only take off the kernel
 		// and uncached bits of the address when comparing.
-		const u32 addr = displayFramebufPtr_ & 0x3FFFFFFF;
+		const u32 addr = fbaddr & 0x3FFFFFFF;
 		for (size_t i = 0; i < vfbs_.size(); ++i) {
 			VirtualFramebuffer *v = vfbs_[i];
 			const u32 v_addr = v->fb_address & 0x3FFFFFFF;
@@ -914,7 +922,7 @@ void FramebufferManagerCommon::CopyDisplayToOutput() {
 	}

 	if (!vfb) {
-		if (Memory::IsValidAddress(displayFramebufPtr_)) {
+		if (Memory::IsValidAddress(fbaddr)) {
 			// The game is displaying something directly from RAM. In GTA, it's decoded video.
 			if (!vfb) {
 				shaderManager_->DirtyLastShader();
@@ -925,12 +933,12 @@ void FramebufferManagerCommon::CopyDisplayToOutput() {
 				// Just a pointer to plain memory to draw. We should create a framebuffer, then draw to it.
 				SetViewport2D(0, 0, pixelWidth_, pixelHeight_);
 				draw_->SetScissorRect(0, 0, pixelWidth_, pixelHeight_);
-				DrawFramebufferToOutput(Memory::GetPointer(displayFramebufPtr_), displayFormat_, displayStride_, true);
+				DrawFramebufferToOutput(Memory::GetPointer(fbaddr), displayFormat_, displayStride_, true);
 				gstate_c.Dirty(DIRTY_BLEND_STATE);
 				return;
 			}
 		} else {
-			DEBUG_LOG(FRAMEBUF, "Found no FBO to display! displayFBPtr = %08x", displayFramebufPtr_);
+			DEBUG_LOG(FRAMEBUF, "Found no FBO to display! displayFBPtr = %08x", fbaddr);
 			// No framebuffer to display! Clear to black.
 			if (useBufferedRendering_) {
 				shaderManager_->DirtyLastShader();
@@ -228,7 +228,7 @@ public:
 	void RebindFramebuffer();
 	std::vector<FramebufferInfo> GetFramebufferList();

-	void CopyDisplayToOutput();
+	void CopyDisplayToOutput(bool reallyDirty);

 	bool NotifyFramebufferCopy(u32 src, u32 dest, int size, bool isMemset, u32 skipDrawReason);
 	void NotifyVideoUpload(u32 addr, int size, int width, GEBufferFormat fmt);
@@ -264,6 +264,10 @@ public:
 		return displayFramebuf_ ? displayFormat_ : GE_FORMAT_INVALID;
 	}

+	bool UseBufferedRendering() {
+		return useBufferedRendering_;
+	}
+
 	bool MayIntersectFramebuffer(u32 start) {
 		// Clear the cache/kernel bits.
 		start = start & 0x3FFFFFFF;
@@ -383,6 +387,7 @@ protected:
 	u32 displayFramebufPtr_ = 0;
 	u32 displayStride_ = 0;
 	GEBufferFormat displayFormat_;
+	u32 prevDisplayFramebufPtr_ = 0;

 	VirtualFramebuffer *displayFramebuf_ = nullptr;
 	VirtualFramebuffer *prevDisplayFramebuf_ = nullptr;
@@ -34,12 +34,6 @@

 #include "GPU/Common/GPUStateUtils.h"

-bool CanUseHardwareTransform(int prim) {
-	if (!g_Config.bHardwareTransform)
-		return false;
-	return !gstate.isModeThrough() && prim != GE_PRIM_RECTANGLES;
-}
-
 bool IsStencilTestOutputDisabled() {
 	// The mask applies on all stencil ops.
 	if (gstate.isStencilTestEnabled() && (gstate.pmska & 0xFF) != 0xFF) {
@@ -51,7 +51,6 @@ StencilValueType ReplaceAlphaWithStencilType();
 ReplaceAlphaType ReplaceAlphaWithStencil(ReplaceBlendType replaceBlend);
 ReplaceBlendType ReplaceBlendWithShader(bool allowShaderBlend, GEBufferFormat bufferFormat);

-bool CanUseHardwareTransform(int prim);
 LogicOpReplaceType ReplaceLogicOpType();


@@ -60,7 +60,7 @@ std::string VertexShaderDesc(const VShaderID &id) {
 	return desc.str();
 }

-void ComputeVertexShaderID(VShaderID *id_out, u32 vertType, bool useHWTransform) {
+void ComputeVertexShaderID(VShaderID *id_out, u32 vertType, bool useHWTransform, bool useHWTessellation) {
 	bool isModeThrough = (vertType & GE_VTYPE_THROUGH) != 0;
 	bool doTexture = gstate.isTextureMapEnabled() && !gstate.isModeClear();
 	bool doTextureTransform = gstate.getUVGenMode() == GE_TEXMAP_TEXTURE_MATRIX;
@@ -131,7 +131,7 @@ void ComputeVertexShaderID(VShaderID *id_out, u32 vertType, bool useHWTransform)
 		id.SetBit(VS_BIT_NORM_REVERSE, gstate.areNormalsReversed());
 		id.SetBit(VS_BIT_HAS_TEXCOORD, hasTexcoord);

-		if (g_Config.bHardwareTessellation) {
+		if (useHWTessellation) {
 			id.SetBit(VS_BIT_BEZIER, doBezier);
 			id.SetBit(VS_BIT_SPLINE, doSpline);
 			if (doBezier || doSpline) {
@@ -223,7 +223,7 @@ class Bugs;
 }


-void ComputeVertexShaderID(VShaderID *id, uint32_t vertexType, bool useHWTransform);
+void ComputeVertexShaderID(VShaderID *id, uint32_t vertexType, bool useHWTransform, bool useHWTessellation);
 // Generates a compact string that describes the shader. Useful in a list to get an overview
 // of the current flora of shaders.
 std::string VertexShaderDesc(const VShaderID &id);
@@ -11,8 +11,6 @@
 #include "GPU/Common/FramebufferCommon.h"
 #include "GPU/Common/GPUStateUtils.h"
 #include "GPU/Math3D.h"
-#include "Core/Reporting.h"
-#include "Core/Config.h"

 using namespace Lin;

@@ -91,7 +89,7 @@ void CalcCullRange(float minValues[4], float maxValues[4], bool flipViewport, bo
 	maxValues[3] = NAN;
 }

-void BaseUpdateUniforms(UB_VS_FS_Base *ub, uint64_t dirtyUniforms, bool flipViewport) {
+void BaseUpdateUniforms(UB_VS_FS_Base *ub, uint64_t dirtyUniforms, bool flipViewport, bool useBufferedRendering) {
 	if (dirtyUniforms & DIRTY_TEXENV) {
 		Uint8x3ToFloat4(ub->texEnvColor, gstate.texenvcolor);
 	}
@@ -149,7 +147,7 @@ void BaseUpdateUniforms(UB_VS_FS_Base *ub, uint64_t dirtyUniforms, bool flipView
 			ConvertProjMatrixToVulkan(flippedMatrix);
 		}

-		if (g_Config.iRenderingMode == FB_NON_BUFFERED_MODE && g_display_rotation != DisplayRotation::ROTATE_0) {
+		if (!useBufferedRendering && g_display_rotation != DisplayRotation::ROTATE_0) {
 			flippedMatrix = flippedMatrix * g_display_rot_matrix;
 		}
 		CopyMatrix4x4(ub->proj, flippedMatrix.getReadPtr());
@@ -162,7 +160,7 @@ void BaseUpdateUniforms(UB_VS_FS_Base *ub, uint64_t dirtyUniforms, bool flipView
 		} else {
 			proj_through.setOrthoVulkan(0.0f, gstate_c.curRTWidth, 0, gstate_c.curRTHeight, 0, 1);
 		}
-		if (g_Config.iRenderingMode == FB_NON_BUFFERED_MODE && g_display_rotation != DisplayRotation::ROTATE_0) {
+		if (!useBufferedRendering && g_display_rotation != DisplayRotation::ROTATE_0) {
 			proj_through = proj_through * g_display_rot_matrix;
 		}
 		// proj_through.translateAndScale(Vec3(0, 0, 0), Vec3(1.0f / debugscale, 1.0f / debugscale, 0));
@@ -186,7 +186,7 @@ R"(	float4x3 u_bone[8];

 void CalcCullRange(float minValues[4], float maxValues[4], bool flipViewport, bool hasNegZ);

-void BaseUpdateUniforms(UB_VS_FS_Base *ub, uint64_t dirtyUniforms, bool flipViewport);
+void BaseUpdateUniforms(UB_VS_FS_Base *ub, uint64_t dirtyUniforms, bool flipViewport, bool useBufferedRendering);
 void LightUpdateUniforms(UB_VS_Lights *ub, uint64_t dirtyUniforms);
 void BoneUpdateUniforms(UB_VS_Bones *ub, uint64_t dirtyUniforms);

@@ -517,7 +517,7 @@ void SoftwareTransform(
 	numTrans = 0;
 	drawIndexed = false;

-	bool useBufferedRendering = g_Config.iRenderingMode != FB_NON_BUFFERED_MODE;
+	bool useBufferedRendering = fbman->UseBufferedRendering();

 	bool flippedY = g_Config.iGPUBackend == (int)GPUBackend::OPENGL && !useBufferedRendering;

@@ -28,13 +28,6 @@
 #include "GPU/ge_constants.h"
 #include "GPU/GPUState.h"  // only needed for UVScale stuff

-bool CanUseHardwareTessellation(GEPatchPrimType prim) {
-	if (g_Config.bHardwareTessellation && !g_Config.bSoftwareRendering) {
-		return CanUseHardwareTransform(PatchPrimToPrim(prim));
-	}
-	return false;
-}
-
 class SimpleBufferManager {
 private:
 	u8 *buf_;
@@ -207,8 +207,6 @@ void SoftwareTessellation(OutputBuffers &output, const Surface &surface, u32 ori

 } // namespace Spline

-bool CanUseHardwareTessellation(GEPatchPrimType prim);
-
 // Define function object for TemplateParameterDispatcher
 #define TEMPLATE_PARAMETER_DISPATCHER_FUNCTION(NAME, FUNCNAME, FUNCTYPE) \
 struct NAME { \
@@ -795,7 +795,7 @@ bool TextureCacheCommon::AttachFramebuffer(TexCacheEntry *entry, u32 address, Vi
 		}
 	} else {
 		// Apply to buffered mode only.
-		if (!(g_Config.iRenderingMode == FB_BUFFERED_MODE))
+		if (!framebufferManager_->UseBufferedRendering())
 			return false;

 		const bool clutFormat =
@@ -875,8 +875,7 @@ void TextureCacheCommon::SetTextureFramebuffer(TexCacheEntry *entry, VirtualFram
 	_dbg_assert_msg_(G3D, framebuffer != nullptr, "Framebuffer must not be null.");

 	framebuffer->usageFlags |= FB_USAGE_TEXTURE;
-	bool useBufferedRendering = g_Config.iRenderingMode != FB_NON_BUFFERED_MODE;
-	if (useBufferedRendering) {
+	if (framebufferManager_->UseBufferedRendering()) {
 		const u64 cachekey = entry->CacheKey();
 		const auto &fbInfo = fbTexInfo_[cachekey];

@@ -918,7 +917,7 @@ void TextureCacheCommon::SetTextureFramebuffer(TexCacheEntry *entry, VirtualFram
 }

 bool TextureCacheCommon::SetOffsetTexture(u32 offset) {
-	if (g_Config.iRenderingMode != FB_BUFFERED_MODE) {
+	if (!framebufferManager_->UseBufferedRendering()) {
 		return false;
 	}
 	u32 texaddr = gstate.getTextureAddress(0);
@@ -1307,7 +1306,7 @@ void TextureCacheCommon::DecodeTextureLevel(u8 *out, int outPitch, GETextureForm
 					memcpy(out + outPitch * y, texptr + bufw * sizeof(u16) * y, w * sizeof(u16));
 				}
 			}
-		} else if (h >= 8 && !expandTo32bit) {
+		} else if (h >= 8 && bufw <= w && !expandTo32bit) {
 			// Note: this is always safe since h must be a power of 2, so a multiple of 8.
 			UnswizzleFromMem((u32 *)out, outPitch, texptr, bufw, h, 2);
 			if (reverseColors) {
@@ -1346,7 +1345,7 @@ void TextureCacheCommon::DecodeTextureLevel(u8 *out, int outPitch, GETextureForm
 					memcpy(out + outPitch * y, texptr + bufw * sizeof(u32) * y, w * sizeof(u32));
 				}
 			}
-		} else if (h >= 8) {
+		} else if (h >= 8 && bufw <= w) {
 			UnswizzleFromMem((u32 *)out, outPitch, texptr, bufw, h, 4);
 			if (reverseColors) {
 				ReverseColors(out, out, format, h * outPitch / 4, useBGRA);
@@ -93,7 +93,7 @@ u32 GetTextureBufw(int level, u32 texaddr, GETextureFormat format) {
 		return gstate.texbufwidth[level] & 0x1FFF;

 	u32 bufw = gstate.texbufwidth[level] & textureAlignMask16[format];
-	if (bufw == 0) {
+	if (bufw == 0 && format <= GE_TFMT_DXT5) {
 		// If it's less than 16 bytes, use 16 bytes.
 		bufw = (8 * 16) / textureBitsPerPixel[format];
 	}
@@ -514,11 +514,11 @@ rotateVBO:

 		D3D11VertexShader *vshader;
 		D3D11FragmentShader *fshader;
-		shaderManager_->GetShaders(prim, lastVType_, &vshader, &fshader, useHWTransform);
+		shaderManager_->GetShaders(prim, lastVType_, &vshader, &fshader, useHWTransform, useHWTessellation_);
 		ID3D11InputLayout *inputLayout = SetupDecFmtForDraw(vshader, dec_->GetDecVtxFmt(), dec_->VertexType());
 		context_->PSSetShader(fshader->GetShader(), nullptr, 0);
 		context_->VSSetShader(vshader->GetShader(), nullptr, 0);
-		shaderManager_->UpdateUniforms();
+		shaderManager_->UpdateUniforms(framebufferManager_->UseBufferedRendering());
 		shaderManager_->BindUniforms();

 		context_->IASetInputLayout(inputLayout);
@@ -598,10 +598,10 @@ rotateVBO:

 			D3D11VertexShader *vshader;
 			D3D11FragmentShader *fshader;
-			shaderManager_->GetShaders(prim, lastVType_, &vshader, &fshader, false);
+			shaderManager_->GetShaders(prim, lastVType_, &vshader, &fshader, false, false);
 			context_->PSSetShader(fshader->GetShader(), nullptr, 0);
 			context_->VSSetShader(vshader->GetShader(), nullptr, 0);
-			shaderManager_->UpdateUniforms();
+			shaderManager_->UpdateUniforms(framebufferManager_->UseBufferedRendering());
 			shaderManager_->BindUniforms();

 			// We really do need a vertex layout for each vertex shader (or at least check its ID bits for what inputs it uses)!
@@ -73,7 +73,6 @@ GPU_D3D11::GPU_D3D11(GraphicsContext *gfxCtx, Draw::DrawContext *draw)
 	device_ = (ID3D11Device *)draw->GetNativeObject(Draw::NativeObject::DEVICE);
 	context_ = (ID3D11DeviceContext *)draw->GetNativeObject(Draw::NativeObject::CONTEXT);
 	D3D_FEATURE_LEVEL featureLevel = (D3D_FEATURE_LEVEL)draw->GetNativeObject(Draw::NativeObject::FEATURE_LEVEL);
-	lastVsync_ = g_Config.bVSync ? 1 : 0;

 	stockD3D11.Create(device_);

@@ -212,8 +211,7 @@ void GPU_D3D11::DeviceRestore() {
 }

 void GPU_D3D11::InitClear() {
-	bool useNonBufferedRendering = g_Config.iRenderingMode == FB_NON_BUFFERED_MODE;
-	if (useNonBufferedRendering) {
+	if (!framebufferManager_->UseBufferedRendering()) {
 		// device_->Clear(0, NULL, D3DCLEAR_STENCIL | D3DCLEAR_TARGET | D3DCLEAR_ZBUFFER, D3DCOLOR_XRGB(0, 0, 0), 1.f, 0);
 	}
 }
@@ -263,13 +261,13 @@ void GPU_D3D11::SetDisplayFramebuffer(u32 framebuf, u32 stride, GEBufferFormat f
 	framebufferManagerD3D11_->SetDisplayFramebuffer(framebuf, stride, format);
 }

-void GPU_D3D11::CopyDisplayToOutput() {
+void GPU_D3D11::CopyDisplayToOutput(bool reallyDirty) {
 	float blendColor[4]{};
 	context_->OMSetBlendState(stockD3D11.blendStateDisabledWithColorMask[0xF], blendColor, 0xFFFFFFFF);

 	drawEngine_.Flush();

-	framebufferManagerD3D11_->CopyDisplayToOutput();
+	framebufferManagerD3D11_->CopyDisplayToOutput(reallyDirty);
 	framebufferManagerD3D11_->EndFrame();

 	// shaderManager_->EndFrame();
@@ -71,7 +71,7 @@ private:

 	void InitClear() override;
 	void BeginFrame() override;
-	void CopyDisplayToOutput() override;
+	void CopyDisplayToOutput(bool reallyDirty) override;

 	ID3D11Device *device_;
 	ID3D11DeviceContext *context_;
@@ -81,6 +81,4 @@ private:
 	DepalShaderCacheD3D11 *depalShaderCache_;
 	DrawEngineD3D11 drawEngine_;
 	ShaderManagerD3D11 *shaderManagerD3D11_;
-
-	int lastVsync_;
 };
@@ -149,12 +149,12 @@ void ShaderManagerD3D11::DirtyLastShader() {
 	gstate_c.Dirty(DIRTY_VERTEXSHADER_STATE | DIRTY_FRAGMENTSHADER_STATE);
 }

-uint64_t ShaderManagerD3D11::UpdateUniforms() {
+uint64_t ShaderManagerD3D11::UpdateUniforms(bool useBufferedRendering) {
 	uint64_t dirty = gstate_c.GetDirtyUniforms();
 	if (dirty != 0) {
 		D3D11_MAPPED_SUBRESOURCE map;
 		if (dirty & DIRTY_BASE_UNIFORMS) {
-			BaseUpdateUniforms(&ub_base, dirty, true);
+			BaseUpdateUniforms(&ub_base, dirty, true, useBufferedRendering);
 			context_->Map(push_base, 0, D3D11_MAP_WRITE_DISCARD, 0, &map);
 			memcpy(map.pData, &ub_base, sizeof(ub_base));
 			context_->Unmap(push_base, 0);
@@ -183,13 +183,13 @@ void ShaderManagerD3D11::BindUniforms() {
 	context_->PSSetConstantBuffers(0, 1, ps_cbs);
 }

-void ShaderManagerD3D11::GetShaders(int prim, u32 vertType, D3D11VertexShader **vshader, D3D11FragmentShader **fshader, bool useHWTransform) {
+void ShaderManagerD3D11::GetShaders(int prim, u32 vertType, D3D11VertexShader **vshader, D3D11FragmentShader **fshader, bool useHWTransform, bool useHWTessellation) {
 	VShaderID VSID;
 	FShaderID FSID;

 	if (gstate_c.IsDirty(DIRTY_VERTEXSHADER_STATE)) {
 		gstate_c.Clean(DIRTY_VERTEXSHADER_STATE);
-		ComputeVertexShaderID(&VSID, vertType, useHWTransform);
+		ComputeVertexShaderID(&VSID, vertType, useHWTransform, useHWTessellation);
 	} else {
 		VSID = lastVSID_;
 	}
@@ -84,7 +84,7 @@ public:
 	ShaderManagerD3D11(Draw::DrawContext *draw, ID3D11Device *device, ID3D11DeviceContext *context, D3D_FEATURE_LEVEL featureLevel);
 	~ShaderManagerD3D11();

-	void GetShaders(int prim, u32 vertType, D3D11VertexShader **vshader, D3D11FragmentShader **fshader, bool useHWTransform);
+	void GetShaders(int prim, u32 vertType, D3D11VertexShader **vshader, D3D11FragmentShader **fshader, bool useHWTransform, bool useHWTessellation);
 	void ClearShaders();
 	void DirtyLastShader() override;

@@ -94,7 +94,7 @@ public:
 	std::vector<std::string> DebugGetShaderIDs(DebugShaderType type);
 	std::string DebugGetShaderString(std::string id, DebugShaderType type, DebugShaderStringType stringType);

-	uint64_t UpdateUniforms();
+	uint64_t UpdateUniforms(bool useBufferedRendering);
 	void BindUniforms();

 	// TODO: Avoid copying these buffers if same as last draw, can still point to it assuming we're still in the same pushbuffer.
@@ -139,7 +139,7 @@ void DrawEngineD3D11::ApplyDrawState(int prim) {
 		return;
 	}

-	bool useBufferedRendering = g_Config.iRenderingMode != FB_NON_BUFFERED_MODE;
+	bool useBufferedRendering = framebufferManager_->UseBufferedRendering();
 	// Blend
 	if (gstate_c.IsDirty(DIRTY_BLEND_STATE)) {
 		gstate_c.SetAllowShaderBlend(!g_Config.bDisableSlowFramebufEffects);