Merge pull request #1335 from turtleli/freebsd

Support FreeBSD

cmake/BuildParameters.cmake

@@ -201,6 +201,11 @@ if(${PCSX2_TARGET_ARCHITECTURES} MATCHES "i386")
         endif()
     endif()
 
+    # Don't bother porting SuperVU
+    if (NOT Linux)
+        set(DISABLE_SVU TRUE)
+    endif()
+
     add_definitions(-D_ARCH_32=1 -D_M_X86=1 -D_M_X86_32=1)
     set(_ARCH_32 1)
     set(_M_X86 1)
@@ -293,7 +298,7 @@ set(AGGRESSIVE_WARNING "-Wstrict-aliasing -Wstrict-overflow=2 ")
 if (USE_CLANG)
     # -Wno-deprecated-register: glib issue...
     set(DEFAULT_WARNINGS "${DEFAULT_WARNINGS}  -Wno-deprecated-register -Wno-c++14-extensions")
-    if (NOT APPLE)
+    if (Linux)
         set(COMMON_FLAG "${COMMON_FLAG} -no-integrated-as")
     endif()
     set(DBG "-g -fno-omit-frame-pointer")

cmake/FindLibc.cmake

@@ -7,15 +7,20 @@ if(LIBC_LIBRARIES)
     set(LIBC_FIND_QUIETLY TRUE)
 endif(LIBC_LIBRARIES)
 
-find_library(libdl NAMES dl)
 find_library(libm NAMES m)
 
 # OSX doesn't have rt. On Linux timer and aio dependency.
 if(APPLE)
+	find_library(libdl NAMES dl)
 	set(LIBC_LIBRARIES ${librt} ${libdl} ${libm})    
-else()
+elseif(Linux)
+	find_library(libdl NAMES dl)
 	find_library(librt NAMES rt)
 	set(LIBC_LIBRARIES ${librt} ${libdl} ${libm})
+else()
+	# FreeBSD doesn't have libdl
+	find_library(librt NAMES rt)
+	set(LIBC_LIBRARIES ${librt} ${libm})
 endif()
 
 # handle the QUIETLY and REQUIRED arguments and set LIBC_FOUND to TRUE if 

cmake/SearchForStuff.cmake

@@ -2,7 +2,9 @@
 #						Search all libraries on the system
 #-------------------------------------------------------------------------------
 ## Use cmake package to find module
-find_package(ALSA)
+if (Linux)
+    find_package(ALSA)
+endif()
 find_package(Gettext) # translation tool
 if(EXISTS ${PROJECT_SOURCE_DIR}/.git)
     find_package(Git)
@@ -50,6 +52,9 @@ if(CMAKE_CROSSCOMPILING)
         endif()
     endif()
 else()
+    if (${CMAKE_SYSTEM_NAME} MATCHES "FreeBSD")
+        set(wxWidgets_CONFIG_EXECUTABLE "/usr/local/bin/wxgtk2u-3.0-config")
+    endif()
     if(EXISTS "/usr/bin/wx-config-3.0")
         set(wxWidgets_CONFIG_EXECUTABLE "/usr/bin/wx-config-3.0")
     endif()

cmake/SelectPcsx2Plugins.cmake

@@ -218,7 +218,7 @@ endif()
 #---------------------------------------
 # Not ready to be packaged
 if(EXTRA_PLUGINS OR NOT PACKAGE_MODE)
-    if(GTKn_FOUND AND X11_FOUND)
+    if(Linux AND GTKn_FOUND AND X11_FOUND)
         set(LilyPad TRUE)
     endif()
 endif()
@@ -257,8 +257,8 @@ endif()
 #           -SDL
 #           -common_libs
 #---------------------------------------
-if((APPLE AND PORTAUDIO_FOUND AND SOUNDTOUCH_FOUND AND SDLn_FOUND AND common_libs)
+if((PORTAUDIO_FOUND AND SOUNDTOUCH_FOUND AND SDLn_FOUND AND common_libs)
-    OR (Linux AND ALSA_FOUND AND PORTAUDIO_FOUND AND SOUNDTOUCH_FOUND AND SDLn_FOUND AND common_libs))
+	AND ((Linux AND ALSA_FOUND) OR (UNIX AND NOT Linux)))
 	set(spu2-x TRUE)
 elseif(NOT EXISTS "${CMAKE_SOURCE_DIR}/plugins/spu2-x")
 	set(spu2-x FALSE)

common/include/PS2Eext.h

@@ -30,7 +30,7 @@
 
 #define EXPORT_C_(type) extern "C" type CALLBACK
 
-#elif defined(__linux__)
+#elif defined(__unix__)
 
 #include <gtk/gtk.h>
 #include <cstring>
@@ -193,7 +193,7 @@ struct PluginConf
     }
 };
 
-#if defined(__linux__)
+#if defined(__unix__)
 
 static void SysMessage(const char *fmt, ...)
 {

common/include/Pcsx2Defs.h

@@ -213,9 +213,12 @@ static const int __pagesize	= PCSX2_PAGESIZE;
 //  GCC / Intel Compilers Section
 // --------------------------------------------------------------------------------------
 
+#ifndef __packed
 #	define __packed			__attribute__((packed))
-
+#endif
+#ifndef __aligned
 #	define __aligned(alig)	__attribute__((aligned(alig)))
+#endif
 #	define __aligned16		__attribute__((aligned(16)))
 #	define __aligned32		__attribute__((aligned(32)))
 #	define __pagealigned	__attribute__((aligned(PCSX2_PAGESIZE)))
@@ -232,14 +235,18 @@ static const int __pagesize	= PCSX2_PAGESIZE;
 // warnings when a static inlined function isn't used in the scope of a single file (which
 // happens *by design* like all the friggen time >_<)
 
+#ifndef __fastcall
 #	define __fastcall		__attribute__((fastcall))
+#endif
 #	define _inline			__inline__ __attribute__((unused))
 #	ifdef NDEBUG
 #		define __forceinline	__attribute__((always_inline,unused))
 #	else
 #		define __forceinline	__attribute__((unused))
 #	endif
+#ifndef __noinline
 #	define __noinline		__attribute__((noinline))
+#endif
 #	define __threadlocal	__thread
 #	define likely(x)		__builtin_expect(!!(x), 1)
 #	define unlikely(x)		__builtin_expect(!!(x), 0)

common/include/Utilities/Console.h

@@ -234,7 +234,7 @@ public:
 
 extern IConsoleWriter	Console;
 
-#ifdef __linux__
+#if defined(__unix__)
 extern void Console_SetStdout(FILE *fp);
 #endif
 extern void Console_SetActiveHandler( const IConsoleWriter& writer, FILE* flushfp=NULL );

common/src/Utilities/Console.cpp

@@ -113,7 +113,7 @@ const IConsoleWriter ConsoleWriter_Null =
 //  Console_Stdout
 // --------------------------------------------------------------------------------------
 
-#ifdef __linux__
+#if defined(__unix__)
 static __fi const char* GetLinuxConsoleColor(ConsoleColors color)
 {
     switch(color)
@@ -174,7 +174,7 @@ static void __concall ConsoleStdout_Newline()
 
 static void __concall ConsoleStdout_DoSetColor( ConsoleColors color )
 {
-#ifdef __linux__
+#if defined(__unix__)
 	fprintf(stdout_fp, "\033[0m%s", GetLinuxConsoleColor(color));
 	fflush(stdout_fp);
 #endif
@@ -182,7 +182,7 @@ static void __concall ConsoleStdout_DoSetColor( ConsoleColors color )
 
 static void __concall ConsoleStdout_SetTitle( const wxString& title )
 {
-#ifdef __linux__
+#if defined(__unix__)
 	fputs("\033]0;", stdout_fp);
 	fputs(title.utf8_str(), stdout_fp);
 	fputs("\007", stdout_fp);

common/src/Utilities/Linux/LnxThreads.cpp

@@ -16,16 +16,20 @@
 
 #include "../PrecompiledHeader.h"
 #include "PersistentThread.h"
-#include <sys/prctl.h>
 #include <unistd.h>
+#if defined(__linux__)
+#include <sys/prctl.h>
+#elif defined(__unix__)
+#include <pthread_np.h>
+#endif
 
 // We wont need this until we actually have this more then just stubbed out, so I'm commenting this out
 // to remove an unneeded dependency.
 //#include "x86emitter/tools.h"
 
-#if !defined(__linux__) && !defined(__WXMAC__)
+#if !defined(__unix__)
 
-#	pragma message( "LnxThreads.cpp should only be compiled by projects or makefiles targeted at Linux/Mac distros.")
+#	pragma message( "LnxThreads.cpp should only be compiled by projects or makefiles targeted at Linux/BSD distros.")
 
 #else
 
@@ -68,7 +72,7 @@ static u64 get_thread_time(uptr id = 0)
 {
 	clockid_t cid;
 	if (id) {
-		int err = pthread_getcpuclockid(id, &cid);
+		int err = pthread_getcpuclockid((pthread_t)id, &cid);
 		if (err) return 0;
 	} else {
 		cid = CLOCK_THREAD_CPUTIME_ID;
@@ -113,9 +117,13 @@ void Threading::pxThread::_platform_specific_OnCleanupInThread()
 
 void Threading::pxThread::_DoSetThreadName( const char* name )
 {
+#if defined(__linux__)
 	// Extract of manpage: "The name can be up to 16 bytes long, and should be
 	//						null-terminated if it contains fewer bytes."
 	prctl(PR_SET_NAME, name, 0, 0, 0);
+#elif defined(__unix__)
+	pthread_set_name_np(pthread_self(), name);
+#endif
 }
 
 #endif

pcsx2/AsyncFileReader.h

@@ -20,7 +20,7 @@
 #	undef Yield
 #elif defined(__linux__)
 #	include <libaio.h>
-#elif defined(__APPLE__)
+#elif defined(__POSIX__)
 #	include <aio.h>
 #endif
 #include <memory>
@@ -78,7 +78,8 @@ class FlatFileReader : public AsyncFileReader
 	io_context_t m_aio_context;
 #elif defined(__POSIX__)
 	int m_fd; // TODO OSX don't know if overlap as an equivalent on OSX
-	struct aiocb m_aio_context;
+	struct aiocb m_aiocb;
+	bool m_read_in_progress;
 #endif
 
 	bool shareWrite;

pcsx2/CMakeLists.txt

@@ -397,6 +397,12 @@ set(pcsx2OSXSources
 	Darwin/DarwinFlatFileReader.cpp
 	)
 
+set(pcsx2FreeBSDSources
+	Linux/LnxConsolePipe.cpp
+	Linux/LnxKeyCodes.cpp
+	Darwin/DarwinFlatFileReader.cpp
+	)
+
 # Linux headers
 set(pcsx2LinuxHeaders
 	)
@@ -615,6 +621,12 @@ if(APPLE)
 		${pcsx2LinuxHeaders})
 endif()
 
+if(${CMAKE_SYSTEM_NAME} MATCHES "FreeBSD")
+	set(Platform
+		${pcsx2FreeBSDSources}
+		${pcsx2LinuxHeaders})
+endif()
+
 set(pcsx2FinalSources
 	${Common}
 	${Platform}

pcsx2/Darwin/DarwinFlatFileReader.cpp

@@ -16,15 +16,16 @@
 #include "PrecompiledHeader.h"
 #include "AsyncFileReader.h"
 
-#warning This reader is not yet implemented. Crash boom bang if used
+#if defined(__APPLE__)
+#warning Tested on FreeBSD, not OS X. Be very afraid.
+#endif
 
 //FlatFileReader::FlatFileReader(void)
 FlatFileReader::FlatFileReader(bool shareWrite) : shareWrite(shareWrite)
 {
-	printf("FLATC\n");
 	m_blocksize = 2048;
-	m_fd = 0;
+	m_fd = -1;
-	//m_aio_context = 0;
+	m_read_in_progress = false;
 }
 
 FlatFileReader::~FlatFileReader(void)
@@ -34,79 +35,74 @@ FlatFileReader::~FlatFileReader(void)
 
 bool FlatFileReader::Open(const wxString& fileName)
 {
-    printf("OB\n");
     m_filename = fileName;
 
-    int err = 0; //io_setup(64, &m_aio_context);
-	if (err) return false;
-
     m_fd = wxOpen(fileName, O_RDONLY, 0);
 
-	return (m_fd != 0);
+	return (m_fd != -1);
 }
 
 int FlatFileReader::ReadSync(void* pBuffer, uint sector, uint count)
 {
-    printf("RAD\n");
 	BeginRead(pBuffer, sector, count);
 	return FinishRead();
 }
 
 void FlatFileReader::BeginRead(void* pBuffer, uint sector, uint count)
 {
-    printf("RWEADB\n");
+	u64 offset = sector * (u64)m_blocksize + m_dataoffset;
-	u64 offset;
-	offset = sector * (u64)m_blocksize + m_dataoffset;
 
 	u32 bytesToRead = count * m_blocksize;
 
-    struct aiocb iocb;
+	m_aiocb = {0};
-    struct aiocb* iocbs = &iocb;
+	m_aiocb.aio_fildes = m_fd;
+	m_aiocb.aio_offset = offset;
+	m_aiocb.aio_nbytes = bytesToRead;
+	m_aiocb.aio_buf = pBuffer;
 
-    //io_prep_pread(&iocb, m_fd, pBuffer, bytesToRead, offset);
+	if (aio_read(&m_aiocb) != 0) {
-    //io_submit(m_aio_context, 1, &iocbs);
+#if defined(__FreeBSD__)
+		if (errno == ENOSYS)
+			Console.Error("AIO read failed: Check the aio kernel module is loaded");
+		else
+			Console.Error("AIO read failed: error code %d", errno);
+#else
+		Console.Error("AIO read failed: error code %d\n", errno);
+#endif
+		return;
+	}
+	m_read_in_progress = true;
 }
 
 int FlatFileReader::FinishRead(void)
 {
-    printf("FINISH\n");
+	struct aiocb *aiocb_list[] = {&m_aiocb};
-	u32 bytes;
 
-	int min_nr = 1;
+	while (aio_suspend(aiocb_list, 1, nullptr) == -1)
-	int max_nr = 1;
+		if (errno != EINTR)
-   /* struct io_event* events = new io_event[max_nr];
+			break;
 
-	int event = io_getevents(m_aio_context, min_nr, max_nr, events, NULL);
+	m_read_in_progress = false;
-	if (event < 1) {
+	return aio_return(&m_aiocb) == -1? -1: 1;
-		return -1;
-    }*/
-
-	return 1;
 }
 
 void FlatFileReader::CancelRead(void)
 {
-    printf("CANCEL\n");
+	aio_cancel(m_fd, &m_aiocb);
-	// Will be done when m_aio_context context is destroyed
+	m_read_in_progress = false;
-	// Note: io_cancel exists but need the iocb structure as parameter
-	// int io_cancel(aio_context_t ctx_id, struct iocb *iocb,
-	//                struct io_event *result);
 }
 
 void FlatFileReader::Close(void)
 {
-    printf("CLOSE\n");
+	if (m_read_in_progress)
-	if (m_fd) close(m_fd);
+		CancelRead();
+	if (m_fd != -1)
+		close(m_fd);
 
-    //io_destroy(m_aio_context);
+	m_fd = -1;
-    aio_cancel(m_fd, &m_aio_context);
-
-	m_fd = 0;
-    //m_aio_context = 0;
 }
 
 uint FlatFileReader::GetBlockCount(void) const
 {
-    printf("BLOCKS\n");
 	return (int)(Path::GetFileSize(m_filename) / m_blocksize);
 }

pcsx2/IPU/yuv2rgb.cpp

@@ -1,5 +1,5 @@
 /*  PCSX2 - PS2 Emulator for PCs
- *  Copyright (C) 2002-2010  PCSX2 Dev Team
+ *  Copyright (C) 2002-2016  PCSX2 Dev Team
  *
  *  PCSX2 is free software: you can redistribute it and/or modify it under the terms
  *  of the GNU Lesser General Public License as published by the Free Software Found-
@@ -58,356 +58,94 @@ void yuv2rgb_reference(void)
 		}
 }
 
-// TODO OSX optimize me
+// Suikoden Tactics FMV speed results: Reference - ~72fps, SSE2 - ~120fps
-#ifdef __APPLE__
+// An AVX2 version is only slightly faster than an SSE2 version (+2-3fps)
-void yuv2rgb_sse2() {
+// (or I'm a poor optimiser), though it might be worth attempting again
-	yuv2rgb_reference();
+// once we've ported to 64 bits (the extra registers should help).
-}
+__ri void yuv2rgb_sse2()
-
-#elif defined(_M_X86_32)
-// Everything below is bit accurate to the IPU specification (except maybe rounding).
-// Know the specification before you touch it.
-#define SSE_BYTES(x) {x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x}
-#define SSE_WORDS(x) {x, x, x, x, x, x, x, x}
-#define SSE_COEFFICIENTS(x) SSE_WORDS((s16)((x)<<2))
-
-struct SSE2_Tables
 {
-	u16	C_bias[8];			// offset -64
+	const __m128i c_bias = _mm_set1_epi8(s8(IPU_C_BIAS));
-	u8	Y_bias[16];			// offset -48
+	const __m128i y_bias = _mm_set1_epi8(IPU_Y_BIAS);
-	u16 Y_mask[8];			// offset -32
+	const __m128i y_mask = _mm_set1_epi16(s16(0xFF00));
-	u16 round_1bit[8];		// offset -16
-
-	s16 Y_coefficients[8];	// offset 0
-	s16 GCr_coefficients[8];// offset 16
-	s16 GCb_coefficients[8];// offset 32
-	s16 RCr_coefficients[8];// offset 48
-	s16 BCb_coefficients[8];// offset 64
-};
-
-enum
-{
-	C_BIAS     = -0x40,
-	Y_BIAS     = -0x30,
-	Y_MASK     = -0x20,
-	ROUND_1BIT = -0x10,
-
-	Y_COEFF     = 0x00,
-	GCr_COEFF   = 0x10,
-	GCb_COEFF   = 0x20,
-	RCr_COEFF   = 0x30,
-	BCb_COEFF   = 0x40
-};
-
-static const __aligned16 SSE2_Tables sse2_tables =
-{
-	SSE_WORDS(0x8000),		// c_bias
-	SSE_BYTES(IPU_Y_BIAS),	// y_bias
-	SSE_WORDS(0xff00),		// y_mask
-
 	// Specifying round off instead of round down as everywhere else
 	// implies that this is right
-	SSE_WORDS(1),		// round_1bit
+	const __m128i round_1bit = _mm_set1_epi16(0x0001);;
 
-	SSE_COEFFICIENTS(IPU_Y_COEFF),
+	const __m128i y_coefficient = _mm_set1_epi16(s16(IPU_Y_COEFF << 2));
-	SSE_COEFFICIENTS(IPU_GCR_COEFF),
+	const __m128i gcr_coefficient = _mm_set1_epi16(s16(u16(IPU_GCR_COEFF) << 2));
-	SSE_COEFFICIENTS(IPU_GCB_COEFF),
+	const __m128i gcb_coefficient = _mm_set1_epi16(s16(u16(IPU_GCB_COEFF) << 2));
-	SSE_COEFFICIENTS(IPU_RCR_COEFF),
+	const __m128i rcr_coefficient = _mm_set1_epi16(s16(IPU_RCR_COEFF << 2));
-	SSE_COEFFICIENTS(IPU_BCB_COEFF),
+	const __m128i bcb_coefficient = _mm_set1_epi16(s16(IPU_BCB_COEFF << 2));
-};
 
-static __aligned16 u16 yuv2rgb_temp[3][8];
+	// Alpha set to 0x80 here. The threshold stuff is done later.
+	const __m128i& alpha = c_bias;
 
-// This could potentially be improved for SSE4
+	for (int n = 0; n < 8; ++n) {
-__ri void yuv2rgb_sse2(void)
+		// could skip the loadl_epi64 but most SSE instructions require 128-bit
-{
+		// alignment so two versions would be needed.
-#if defined(_MSC_VER) || defined(__INTEL_COMPILER)
+		__m128i cb = _mm_loadl_epi64(reinterpret_cast<__m128i*>(&decoder.mb8.Cb[n][0]));
-	__asm {
+		__m128i cr = _mm_loadl_epi64(reinterpret_cast<__m128i*>(&decoder.mb8.Cr[n][0]));
-		mov eax, 1
-		xor esi, esi
-		xor edi, edi
 
-		// Use ecx and edx as base pointers, to allow for Mod/RM form on memOps.
+		// (Cb - 128) << 8, (Cr - 128) << 8
-		// This saves 2-3 bytes per instruction where these are used. :)
+		cb = _mm_xor_si128(cb, c_bias);
-		mov ecx, offset yuv2rgb_temp
+		cr = _mm_xor_si128(cr, c_bias);
-		mov edx, offset sse2_tables+64;
+		cb = _mm_unpacklo_epi8(_mm_setzero_si128(), cb);
+		cr = _mm_unpacklo_epi8(_mm_setzero_si128(), cr);
 
-		align 16
+		__m128i rc = _mm_mulhi_epi16(cr, rcr_coefficient);
-tworows:
+		__m128i gc = _mm_adds_epi16(_mm_mulhi_epi16(cr, gcr_coefficient), _mm_mulhi_epi16(cb, gcb_coefficient));
-		movq xmm3, qword ptr [decoder.mb8+256+esi]
+		__m128i bc = _mm_mulhi_epi16(cb, bcb_coefficient);
-		movq xmm1, qword ptr [decoder.mb8+320+esi]
-		pxor xmm2, xmm2
-		pxor xmm0, xmm0
-		// could skip the movq but punpck requires 128-bit alignment
-		// for some reason, so two versions would be needed,
-		// bloating the function (further)
-		punpcklbw xmm2, xmm3
-		punpcklbw xmm0, xmm1
-		// unfortunately I don't think this will matter despite being
-		// technically potentially a little faster, but this is
-		// equivalent to an add or sub
-		pxor xmm2, xmmword ptr [edx+C_BIAS] // xmm2 <-- 8 x (Cb - 128) << 8
-		pxor xmm0, xmmword ptr [edx+C_BIAS] // xmm0 <-- 8 x (Cr - 128) << 8
 
-		movaps xmm1, xmm0
+		for (int m = 0; m < 2; ++m) {
-		movaps xmm3, xmm2
+			__m128i y = _mm_load_si128(reinterpret_cast<__m128i*>(&decoder.mb8.Y[n * 2 + m][0]));
-		pmulhw xmm1, xmmword ptr [edx+GCr_COEFF]
+			y = _mm_subs_epu8(y, y_bias);
-		pmulhw xmm3, xmmword ptr [edx+GCb_COEFF]
+			// Y << 8 for pixels 0, 2, 4, 6, 8, 10, 12, 14
-		pmulhw xmm0, xmmword ptr [edx+RCr_COEFF]
+			__m128i y_even = _mm_slli_epi16(y, 8);
-		pmulhw xmm2, xmmword ptr [edx+BCb_COEFF]
+			// Y << 8 for pixels 1, 3, 5, 7 ,9, 11, 13, 15
-		paddsw xmm1, xmm3
+			__m128i y_odd = _mm_and_si128(y, y_mask);
-		// store for the next line; looking at the code above
-		// compared to the code below, I have to wonder whether
-		// this was worth the hassle
-		movaps xmmword ptr [ecx], xmm0
-		movaps xmmword ptr [ecx+16], xmm1
-		movaps xmmword ptr [ecx+32], xmm2
-		jmp ihatemsvc
 
-		align 16
+			y_even = _mm_mulhi_epu16(y_even, y_coefficient);
-onerow:
+			y_odd  = _mm_mulhi_epu16(y_odd,  y_coefficient);
-		movaps xmm0, xmmword ptr [ecx]
-		movaps xmm1, xmmword ptr [ecx+16]
-		movaps xmm2, xmmword ptr [ecx+32]
 
-// If masm directives worked properly in inline asm, I'd be using them,
+			__m128i r_even = _mm_adds_epi16(rc, y_even);
-// but I'm not inclined to write ~70 line #defines to simulate them.
+			__m128i r_odd  = _mm_adds_epi16(rc, y_odd);
-// Maybe the function's faster like this anyway because it's smaller?
+			__m128i g_even = _mm_adds_epi16(gc, y_even);
-// I'd have to write a 70 line #define to benchmark it.
+			__m128i g_odd  = _mm_adds_epi16(gc, y_odd);
+			__m128i b_even = _mm_adds_epi16(bc, y_even);
+			__m128i b_odd  = _mm_adds_epi16(bc, y_odd);
 
-ihatemsvc:
+			// round
-		movaps xmm3, xmm0
+			r_even = _mm_srai_epi16(_mm_add_epi16(r_even, round_1bit), 1);
-		movaps xmm4, xmm1
+			r_odd  = _mm_srai_epi16(_mm_add_epi16(r_odd,  round_1bit), 1);
-		movaps xmm5, xmm2
+			g_even = _mm_srai_epi16(_mm_add_epi16(g_even, round_1bit), 1);
+			g_odd  = _mm_srai_epi16(_mm_add_epi16(g_odd,  round_1bit), 1);
+			b_even = _mm_srai_epi16(_mm_add_epi16(b_even, round_1bit), 1);
+			b_odd  = _mm_srai_epi16(_mm_add_epi16(b_odd,  round_1bit), 1);
 
-		movaps xmm6, xmmword ptr [decoder.mb8+edi]
+			// combine even and odd bytes in original order
-		psubusb xmm6, xmmword ptr [edx+Y_BIAS]
+			__m128i r = _mm_packus_epi16(r_even, r_odd);
-		movaps xmm7, xmm6
+			__m128i g = _mm_packus_epi16(g_even, g_odd);
-		psllw xmm6, 8                    // xmm6 <- Y << 8 for pixels 0,2,4,6,8,10,12,14
+			__m128i b = _mm_packus_epi16(b_even, b_odd);
-		pand xmm7, xmmword ptr [edx+Y_MASK]  // xmm7 <- Y << 8 for pixels 1,3,5,7,9,11,13,15
 
-		pmulhuw xmm6, xmmword ptr [edx+Y_COEFF]
+			r = _mm_unpacklo_epi8(r, _mm_shuffle_epi32(r, _MM_SHUFFLE(3, 2, 3, 2)));
-		pmulhuw xmm7, xmmword ptr [edx+Y_COEFF]
+			g = _mm_unpacklo_epi8(g, _mm_shuffle_epi32(g, _MM_SHUFFLE(3, 2, 3, 2)));
+			b = _mm_unpacklo_epi8(b, _mm_shuffle_epi32(b, _MM_SHUFFLE(3, 2, 3, 2)));
 
-		paddsw xmm0, xmm6
+			// Create RGBA (we could generate A here, but we don't) quads
-		paddsw xmm3, xmm7
+			__m128i rg_l = _mm_unpacklo_epi8(r, g);
-		paddsw xmm1, xmm6
+			__m128i ba_l = _mm_unpacklo_epi8(b, alpha);
-		paddsw xmm4, xmm7
+			__m128i rgba_ll = _mm_unpacklo_epi16(rg_l, ba_l);
-		paddsw xmm2, xmm6
+			__m128i rgba_lh = _mm_unpackhi_epi16(rg_l, ba_l);
-		paddsw xmm5, xmm7
 
-		// 0x80; a constant is probably so much better
+			__m128i rg_h = _mm_unpackhi_epi8(r, g);
-		pcmpeqb xmm7, xmm7
+			__m128i ba_h = _mm_unpackhi_epi8(b, alpha);
-		psllw xmm7, 15
+			__m128i rgba_hl = _mm_unpacklo_epi16(rg_h, ba_h);
-		psrlw xmm7, 8
+			__m128i rgba_hh = _mm_unpackhi_epi16(rg_h, ba_h);
-		packuswb xmm7, xmm7
 
-		// round
+			_mm_store_si128(reinterpret_cast<__m128i*>(&decoder.rgb32.c[n * 2 + m][0]), rgba_ll);
-		movaps xmm6, xmmword ptr [edx+ROUND_1BIT]
+			_mm_store_si128(reinterpret_cast<__m128i*>(&decoder.rgb32.c[n * 2 + m][4]), rgba_lh);
-		paddw xmm0, xmm6
+			_mm_store_si128(reinterpret_cast<__m128i*>(&decoder.rgb32.c[n * 2 + m][8]), rgba_hl);
-		paddw xmm1, xmm6
+			_mm_store_si128(reinterpret_cast<__m128i*>(&decoder.rgb32.c[n * 2 + m][12]), rgba_hh);
-		paddw xmm2, xmm6
+		}
-		paddw xmm3, xmm6
-		paddw xmm4, xmm6
-		paddw xmm5, xmm6
-		psraw xmm0, 1
-		psraw xmm1, 1
-		psraw xmm2, 1
-		psraw xmm3, 1
-		psraw xmm4, 1
-		psraw xmm5, 1
-
-		// combine even and odd bytes
-		packuswb xmm0, xmm3
-		packuswb xmm1, xmm4
-		packuswb xmm2, xmm5
-		movhlps xmm3, xmm0
-		movhlps xmm4, xmm1
-		movhlps xmm5, xmm2
-		punpcklbw xmm0, xmm3 // Red bytes, back in order
-		punpcklbw xmm1, xmm4 // Green ""
-		punpcklbw xmm2, xmm5 // Blue ""
-		movaps xmm3, xmm0
-		movaps xmm4, xmm1
-		movaps xmm5, xmm2
-
-		// Create RGBA (we could generate A here, but we don't) quads
-		punpcklbw xmm0, xmm1
-		punpcklbw xmm2, xmm7
-		movaps xmm1, xmm0
-		punpcklwd xmm0, xmm2
-		punpckhwd xmm1, xmm2
-
-		punpckhbw xmm3, xmm4
-		punpckhbw xmm5, xmm7
-		movaps xmm4, xmm3
-		punpcklwd xmm3, xmm5
-		punpckhwd xmm4, xmm5
-
-		// at last
-		movaps xmmword ptr [decoder.rgb32+edi*4+0], xmm0
-		movaps xmmword ptr [decoder.rgb32+edi*4+16], xmm1
-		movaps xmmword ptr [decoder.rgb32+edi*4+32], xmm3
-		movaps xmmword ptr [decoder.rgb32+edi*4+48], xmm4
-
-		add edi, 16
-
-		neg eax
-		jl onerow // run twice
-
-		add esi, 8
-		cmp esi, 64
-		jne tworows
 	}
-
-#elif defined(__GNUC__)
-
-	// offset to the middle of the sse2 table, so that we can use 1-byte address displacement
-	// to access all fields:
-	static const u8* sse2_tableoffset = ((u8*)&sse2_tables) + 64;
-	static const u8* mb8 = (u8*)&decoder.mb8;
-	static u8* rgb32 = (u8*)&decoder.rgb32;
-
-	__asm__ __volatile__ (
-		".intel_syntax noprefix\n"
-		"xor esi, esi\n"
-		"xor edi, edi\n"
-
-		".align 16\n"
-"tworows_%=:\n"
-		"movq xmm3, qword ptr [%[mb8]+256+esi]\n"
-		"movq xmm1, qword ptr [%[mb8]+320+esi]\n"
-		"pxor xmm2, xmm2\n"
-		"pxor xmm0, xmm0\n"
-		// could skip the movq but punpck requires 128-bit alignment
-		// for some reason, so two versions would be needed,
-		// bloating the function (further)
-		"punpcklbw xmm2, xmm3\n"
-		"punpcklbw xmm0, xmm1\n"
-		// unfortunately I don't think this will matter despite being
-		// technically potentially a little faster, but this is
-		// equivalent to an add or sub
-		"pxor xmm2, xmmword ptr [%[sse2_tables]+%c[C_BIAS]]\n" // xmm2 <-- 8 x (Cb - 128) << 8
-		"pxor xmm0, xmmword ptr [%[sse2_tables]+%c[C_BIAS]]\n" // xmm0 <-- 8 x (Cr - 128) << 8
-
-		"movaps xmm1, xmm0\n"
-		"movaps xmm3, xmm2\n"
-		"pmulhw xmm1, xmmword ptr [%[sse2_tables]+%c[GCr_COEFF]]\n"
-		"pmulhw xmm3, xmmword ptr [%[sse2_tables]+%c[GCb_COEFF]]\n"
-		"pmulhw xmm0, xmmword ptr [%[sse2_tables]+%c[RCr_COEFF]]\n"
-		"pmulhw xmm2, xmmword ptr [%[sse2_tables]+%c[BCb_COEFF]]\n"
-		"paddsw xmm1, xmm3\n"
-		// store for the next line; looking at the code above
-		// compared to the code below, I have to wonder whether
-		// this was worth the hassle
-		"movaps xmmword ptr [%[yuv2rgb_temp]], xmm0\n"
-		"movaps xmmword ptr [%[yuv2rgb_temp]+16], xmm1\n"
-		"movaps xmmword ptr [%[yuv2rgb_temp]+32], xmm2\n"
-		"jmp ihategcctoo_%=\n"
-
-		".align 16\n"
-"onerow_%=:\n"
-		"movaps xmm0, xmmword ptr [%[yuv2rgb_temp]]\n"
-		"movaps xmm1, xmmword ptr [%[yuv2rgb_temp]+16]\n"
-		"movaps xmm2, xmmword ptr [%[yuv2rgb_temp]+32]\n"
-
-"ihategcctoo_%=:\n"
-		"movaps xmm3, xmm0\n"
-		"movaps xmm4, xmm1\n"
-		"movaps xmm5, xmm2\n"
-
-		"movaps xmm6, xmmword ptr [%[mb8]+edi]\n"
-		"psubusb xmm6, xmmword ptr [%[sse2_tables]+%c[Y_BIAS]]\n"
-		"movaps xmm7, xmm6\n"
-		"psllw xmm6, 8\n"                   // xmm6 <- Y << 8 for pixels 0,2,4,6,8,10,12,14
-		"pand xmm7, xmmword ptr [%[sse2_tables]+%c[Y_MASK]]\n" // xmm7 <- Y << 8 for pixels 1,3,5,7,9,11,13,15
-
-		"pmulhuw xmm6, xmmword ptr [%[sse2_tables]+%c[Y_COEFF]]\n"
-		"pmulhuw xmm7, xmmword ptr [%[sse2_tables]+%c[Y_COEFF]]\n"
-
-		"paddsw xmm0, xmm6\n"
-		"paddsw xmm3, xmm7\n"
-		"paddsw xmm1, xmm6\n"
-		"paddsw xmm4, xmm7\n"
-		"paddsw xmm2, xmm6\n"
-		"paddsw xmm5, xmm7\n"
-
-		// 0x80; a constant is probably so much better
-		"pcmpeqb xmm7, xmm7\n"
-		"psllw xmm7, 15\n"
-		"psrlw xmm7, 8\n"
-		"packuswb xmm7, xmm7\n"
-
-		// round
-		"movaps xmm6, xmmword ptr [%[sse2_tables]+%c[ROUND_1BIT]]\n"
-		"paddw xmm0, xmm6\n"
-		"paddw xmm1, xmm6\n"
-		"paddw xmm2, xmm6\n"
-		"paddw xmm3, xmm6\n"
-		"paddw xmm4, xmm6\n"
-		"paddw xmm5, xmm6\n"
-		"psraw xmm0, 1\n"
-		"psraw xmm1, 1\n"
-		"psraw xmm2, 1\n"
-		"psraw xmm3, 1\n"
-		"psraw xmm4, 1\n"
-		"psraw xmm5, 1\n"
-
-		// combine even and odd bytes
-		"packuswb xmm0, xmm3\n"
-		"packuswb xmm1, xmm4\n"
-		"packuswb xmm2, xmm5\n"
-		"movhlps xmm3, xmm0\n"
-		"movhlps xmm4, xmm1\n"
-		"movhlps xmm5, xmm2\n"
-		"punpcklbw xmm0, xmm3\n" // Red bytes, back in order
-		"punpcklbw xmm1, xmm4\n" // Green ""
-		"punpcklbw xmm2, xmm5\n" // Blue ""
-		"movaps xmm3, xmm0\n"
-		"movaps xmm4, xmm1\n"
-		"movaps xmm5, xmm2\n"
-
-		// Create RGBA (we could generate A here, but we don't) quads
-		"punpcklbw xmm0, xmm1\n"
-		"punpcklbw xmm2, xmm7\n"
-		"movaps xmm1, xmm0\n"
-		"punpcklwd xmm0, xmm2\n"
-		"punpckhwd xmm1, xmm2\n"
-
-		"punpckhbw xmm3, xmm4\n"
-		"punpckhbw xmm5, xmm7\n"
-		"movaps xmm4, xmm3\n"
-		"punpcklwd xmm3, xmm5\n"
-		"punpckhwd xmm4, xmm5\n"
-
-		// at last
-		"movaps xmmword ptr [%[rgb32]+edi*4+0], xmm0\n"
-		"movaps xmmword ptr [%[rgb32]+edi*4+16], xmm1\n"
-		"movaps xmmword ptr [%[rgb32]+edi*4+32], xmm3\n"
-		"movaps xmmword ptr [%[rgb32]+edi*4+48], xmm4\n"
-
-		"add edi, 16\n"
-
-        // run twice the onerow <=> edi = 16 or 48 or 80 etc... <=> check bit 5
-		"test edi, 16\n"
-		"jnz onerow_%=\n"
-
-		"add esi, 8\n"
-		"cmp esi, 64\n"
-		"jne tworows_%=\n"
-		".att_syntax\n"
-		:
-		:[C_BIAS]"i"(C_BIAS), [Y_BIAS]"i"(Y_BIAS), [Y_MASK]"i"(Y_MASK),
-			[ROUND_1BIT]"i"(ROUND_1BIT), [Y_COEFF]"i"(Y_COEFF), [GCr_COEFF]"i"(GCr_COEFF),
-			[GCb_COEFF]"i"(GCb_COEFF), [RCr_COEFF]"i"(RCr_COEFF), [BCb_COEFF]"i"(BCb_COEFF),
-            // Use ecx and edx as base pointers, to allow for Mod/RM form on memOps.
-            // This saves 2-3 bytes per instruction where these are used. :)
-			[yuv2rgb_temp]"c"(yuv2rgb_temp), [sse2_tables]"d"(sse2_tableoffset),
-			[mb8]"r"(mb8), [rgb32]"r"(rgb32)
-		: "esi", "edi", "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7", "memory"
-	);
-#else
-#	error Unsupported compiler
-#endif
 }
-#endif

pcsx2/IPU/yuv2rgb.h

@@ -1,5 +1,5 @@
 /*  PCSX2 - PS2 Emulator for PCs
- *  Copyright (C) 2002-2010  PCSX2 Dev Team
+ *  Copyright (C) 2002-2016  PCSX2 Dev Team
  *
  *  PCSX2 is free software: you can redistribute it and/or modify it under the terms
  *  of the GNU Lesser General Public License as published by the Free Software Found-
@@ -16,9 +16,6 @@
 #pragma once
 
 extern void yuv2rgb_reference();
-#ifdef _M_X86_32
+
 #define yuv2rgb yuv2rgb_sse2
 extern void yuv2rgb_sse2();
-#else
-#define yuv2rgb yuv2rgb_reference
-#endif

pcsx2/IopGte.cpp

@@ -20,6 +20,7 @@
 #include "IopGte.h"
 //#include "R3000A.h"
 #include "IopCommon.h"
+#include "Utilities/MathUtils.h"
 #ifdef GTE_DUMP
 #define G_OP(name,delay) fprintf(gteLog, "* : %08X : %02d : %s\n", psxRegs.code, delay, name);
 #define G_SD(reg)  fprintf(gteLog, "+D%02d : %08X\n", reg, psxRegs.CP2D.r[reg]);
@@ -175,8 +176,6 @@ __inline unsigned long MFC2(int reg) {
 }
 
 __inline void MTC2(unsigned long value, int reg) {
-	int a;
-
 	switch (reg) {
 	case 8: case 9: case 10: case 11:
 		psxRegs.CP2D.r[reg] = (short)value;
@@ -208,58 +207,7 @@ __inline void MTC2(unsigned long value, int reg) {
 
 	case 30:
 		psxRegs.CP2D.r[30] = value;
-
+		psxRegs.CP2D.r[31] = count_leading_sign_bits(value);
-		a = psxRegs.CP2D.r[30];
-#if defined(_MSC_VER_)
-		if (a > 0) {
-			__asm {
-				mov eax, a;
-				bsr eax, eax;
-				mov a, eax;
-			}
-			psxRegs.CP2D.r[31] = 31 - a;
-		}
-		else if (a < 0) {
-			__asm {
-				mov eax, a;
-				xor eax, 0xffffffff;
-				bsr eax, eax;
-				mov a, eax;
-			}
-			psxRegs.CP2D.r[31] = 31 - a;
-		}
-		else {
-			psxRegs.CP2D.r[31] = 32;
-		}
-#elif defined(__linux__) || defined(__MINGW32__)
-		if (a > 0) {
-			__asm__("bsrl %1, %0\n" : "=r"(a) : "r"(a));
-			psxRegs.CP2D.r[31] = 31 - a;
-		}
-		else if (a < 0) {
-			a ^= 0xffffffff;
-			__asm__("bsrl %1, %0\n" : "=r"(a) : "r"(a));
-			psxRegs.CP2D.r[31] = 31 - a;
-		}
-		else {
-			psxRegs.CP2D.r[31] = 32;
-		}
-#else
-		if (a > 0) {
-			int i;
-			for (i = 31; (a & (1 << i)) == 0 && i >= 0; i--);
-			psxRegs.CP2D.r[31] = 31 - i;
-		}
-		else if (a < 0) {
-			int i;
-			a ^= 0xffffffff;
-			for (i = 31; (a & (1 << i)) == 0 && i >= 0; i--);
-			psxRegs.CP2D.r[31] = 31 - i;
-		}
-		else {
-			psxRegs.CP2D.r[31] = 32;
-		}
-#endif
 		break;
 
 	default:

pcsx2/PrecompiledHeader.h

@@ -96,13 +96,7 @@ typedef FnType_Void* Fnptr_Void;
 
 #else	// must be GCC or Clang
 
-#	include <sys/types.h>
+#include <sys/types.h>
-#	include <sys/timeb.h>
-
-// Definitions added Feb 16, 2006 by efp
-#	ifndef __declspec
-#		define __declspec(x)
-#	endif
 
 #endif
 

pcsx2/gui/ConsoleLogger.cpp

@@ -1017,11 +1017,11 @@ void Pcsx2App::ProgramLog_PostEvent( wxEvent& evt )
 
 static void __concall ConsoleToFile_Newline()
 {
-#ifdef __linux__
+#if defined(__unix__)
 	if ((g_Conf) && (g_Conf->EmuOptions.ConsoleToStdio)) ConsoleWriter_Stdout.Newline();
 #endif
 
-#ifdef __linux__
+#if defined(__unix__)
 	fputc( '\n', emuLog );
 #else
 	fputs( "\r\n", emuLog );
@@ -1030,7 +1030,7 @@ static void __concall ConsoleToFile_Newline()
 
 static void __concall ConsoleToFile_DoWrite( const wxString& fmt )
 {
-#ifdef __linux__
+#if defined(__unix__)
 	if ((g_Conf) && (g_Conf->EmuOptions.ConsoleToStdio)) ConsoleWriter_Stdout.WriteRaw(fmt);
 #endif
 

pcsx2/gui/FrameForGS.cpp

@@ -222,7 +222,7 @@ void GSPanel::OnMouseEvent( wxMouseEvent& evt )
 		DoShowMouse();
 	}
 
-#ifdef __linux__
+#if defined(__unix__)
 	// HACK2: In gsopen2 there is one event buffer read by both wx/gui and pad plugin. Wx deletes
 	// the event before the pad see it. So you send key event directly to the pad.
 	if( (PADWriteEvent != NULL) && (GSopen2 != NULL) ) {
@@ -285,7 +285,7 @@ void GSPanel::OnKeyDownOrUp( wxKeyEvent& evt )
 	// to the APP level message handler, which in turn routes them right back here -- yes it's
 	// silly, but oh well).
 
-#ifdef __linux__
+#if defined(__unix__)
 	// HACK2: In gsopen2 there is one event buffer read by both wx/gui and pad plugin. Wx deletes
 	// the event before the pad see it. So you send key event directly to the pad.
 	if( (PADWriteEvent != NULL) && (GSopen2 != NULL) ) {
@@ -367,7 +367,7 @@ void GSPanel::OnFocus( wxFocusEvent& evt )
 	else
 		DoShowMouse();
 
-#ifdef __linux__
+#if defined(__unix__)
 	// HACK2: In gsopen2 there is one event buffer read by both wx/gui and pad plugin. Wx deletes
 	// the event before the pad see it. So you send key event directly to the pad.
 	if( (PADWriteEvent != NULL) && (GSopen2 != NULL) ) {
@@ -385,7 +385,7 @@ void GSPanel::OnFocusLost( wxFocusEvent& evt )
 	evt.Skip();
 	m_HasFocus = false;
 	DoShowMouse();
-#ifdef __linux__
+#if defined(__unix__)
 	// HACK2: In gsopen2 there is one event buffer read by both wx/gui and pad plugin. Wx deletes
 	// the event before the pad see it. So you send key event directly to the pad.
 	if( (PADWriteEvent != NULL) && (GSopen2 != NULL) ) {

pcsx2/gui/MainFrame.cpp

@@ -503,7 +503,7 @@ MainEmuFrame::MainEmuFrame(wxWindow* parent, const wxString& title)
 	// ------------------------------------------------------------------------
 
 	m_menuMisc.Append( &m_MenuItem_Console );
-#ifdef __linux__
+#if defined(__unix__)
 	m_menuMisc.Append( &m_MenuItem_Console_Stdio );
 #endif
 	//Todo: Though not many people need this one :p
@@ -695,7 +695,7 @@ void MainEmuFrame::ApplyConfigToGui(AppConfig& configToApply, int flags)
 		menubar.Check( MenuId_EnableCheats,  configToApply.EmuOptions.EnableCheats );
 		menubar.Check( MenuId_EnableWideScreenPatches,  configToApply.EmuOptions.EnableWideScreenPatches );
 		menubar.Check( MenuId_EnableHostFs,  configToApply.EmuOptions.HostFs );
-#ifdef __linux__
+#if defined(__unix__)
 		menubar.Check( MenuId_Console_Stdio, configToApply.EmuOptions.ConsoleToStdio );
 #endif
 

pcsx2/gui/i18n.cpp

@@ -362,7 +362,7 @@ void i18n_SetLanguagePath()
 	// default location for windows
 	wxLocale::AddCatalogLookupPathPrefix( wxGetCwd() );
 	// additional location for linux
-#ifdef __linux__
+#ifdef __unix__
 	wxLocale::AddCatalogLookupPathPrefix( PathDefs::GetLangs().ToString() );
 #endif
 

plugins/GSdx/GPUDrawScanlineCodeGenerator.cpp

@@ -1022,7 +1022,7 @@ const GSVector4i GPUDrawScanlineCodeGenerator::m_test[8] =
 	GSVector4i::zero(),
 };
 
-__aligned(const uint16, 32) GPUDrawScanlineCodeGenerator::m_dither[4][16] =
+alignas(32) const uint16_t GPUDrawScanlineCodeGenerator::m_dither[4][16] =
 {
 	{7, 0, 6, 1, 7, 0, 6, 1, 7, 0, 6, 1, 7, 0, 6, 1},
 	{2, 5, 3, 4, 2, 5, 3, 4, 2, 5, 3, 4, 2, 5, 3, 4},

plugins/GSdx/GPUDrawScanlineCodeGenerator.h

@@ -56,5 +56,5 @@ public:
 	GPUDrawScanlineCodeGenerator(void* param, uint32 key, void* code, size_t maxsize);
 
 	static const GSVector4i m_test[8];
-	static __aligned(const uint16, 32) m_dither[4][16];
+	alignas(32) static const uint16 m_dither[4][16];
 };

plugins/GSdx/GPUDrawingEnvironment.h

@@ -23,7 +23,7 @@
 
 #include "GPU.h"
 
-__aligned(class, 32) GPUDrawingEnvironment
+class alignas(32) GPUDrawingEnvironment
 {
 public:
 	GPURegSTATUS STATUS;

plugins/GSdx/GPUScanlineEnvironment.h

@@ -56,7 +56,7 @@ union GPUScanlineSelector
 	operator uint32() const {return key;}
 };
 
-__aligned(struct, 32) GPUScanlineGlobalData
+struct alignas(32) GPUScanlineGlobalData
 {
 	GPUScanlineSelector sel;
 
@@ -66,7 +66,7 @@ __aligned(struct, 32) GPUScanlineGlobalData
 	GSVector4i twin; // TWW, TWH, TWX, TWY
 };
 
-__aligned(struct, 32) GPUScanlineLocalData
+struct alignas(32) GPUScanlineLocalData
 {
 	const GPUScanlineGlobalData* gd;
 

plugins/GSdx/GPUVertex.h

@@ -26,7 +26,7 @@
 
 #pragma pack(push, 1)
 
-__aligned(struct, 32) GPUVertex
+struct alignas(32) GPUVertex
 {
 	union
 	{

plugins/GSdx/GS.cpp

@@ -365,7 +365,7 @@ static int _GSopen(void** dsp, const char* title, GSRendererType renderer, int t
 		int w = theApp.GetConfig("ModeWidth", 0);
 		int h = theApp.GetConfig("ModeHeight", 0);
 
-#ifdef __linux__
+#if defined(__unix__)
 		for(uint32 i = 0; i < 2; i++) {
 			try
 			{
@@ -408,7 +408,7 @@ static int _GSopen(void** dsp, const char* title, GSRendererType renderer, int t
 	{
 		s_gs->SetMultithreaded(true);
 
-#ifdef __linux__
+#if defined(__unix__)
 		if (s_gs->m_wnd) {
 			// A window was already attached to s_gs so we also
 			// need to restore the window state (Attach)
@@ -510,7 +510,7 @@ EXPORT_C_(int) GSopen2(void** dsp, uint32 flags)
 		}
 
 #endif
-#ifdef __linux__
+#if defined(__unix__)
 		switch(renderer) {
 			// Use alternative renderer (SW if currently using HW renderer, and vice versa)
 		case GSRendererType::OGL_SW: renderer = GSRendererType::OGL_HW; break;
@@ -882,7 +882,7 @@ EXPORT_C_(int) GSsetupRecording(int start, void* data)
 		printf("GSdx: no s_gs for recording\n");
 		return 0;
 	}
-#ifdef __linux__
+#if defined(__unix__)
 	if (!theApp.GetConfig("capture_enabled", 0)) {
 		printf("GSdx: Recording is disabled\n");
 		return 0;
@@ -1497,18 +1497,15 @@ EXPORT_C GSBenchmark(HWND hwnd, HINSTANCE hinst, LPSTR lpszCmdLine, int nCmdShow
 
 #endif
 
-#ifdef __linux__
+#if defined(__unix__)
 
-#include <sys/time.h>
-#include <sys/timeb.h>	// ftime(), struct timeb
 #include "GSLzma.h"
 
 inline unsigned long timeGetTime()
 {
-	timeb t;
+	struct timespec t;
-	ftime(&t);
+	clock_gettime(CLOCK_REALTIME, &t);
-
+	return (unsigned long)(t.tv_sec*1000 + t.tv_nsec/1000000);
-	return (unsigned long)(t.time*1000 + t.millitm);
 }
 
 // Note

plugins/GSdx/GS.h

@@ -1101,7 +1101,7 @@ REG128_SET(GIFPackedReg)
 	GIFPackedNOP	NOP;
 REG_SET_END
 
-__aligned(struct, 32) GIFPath
+struct alignas(32) GIFPath
 {
 	GIFTag tag;
 	uint32 nloop;

plugins/GSdx/GSBlock.h

@@ -1917,7 +1917,7 @@ public:
 
 		#else
 		
-		__aligned(uint16, 32) block[16 * 8];
+		alignas(32) uint16 block[16 * 8];
 	
 		ReadBlock16(src, (uint8*)block, sizeof(block) / 8);
 
@@ -1976,7 +1976,7 @@ public:
 
 		#else
 
-		__aligned(uint8, 32) block[16 * 16];
+		alignas(32) uint8 block[16 * 16];
 
 		ReadBlock8(src, (uint8*)block, sizeof(block) / 16);
 
@@ -2053,7 +2053,7 @@ public:
 
 		#else
 
-		__aligned(uint8, 32) block[(32 / 2) * 16];
+		alignas(32) uint8 block[(32 / 2) * 16];
 
 		ReadBlock4(src, (uint8*)block, sizeof(block) / 16);
 
@@ -2096,7 +2096,7 @@ public:
 
 		#else
 
-		__aligned(uint32, 32) block[8 * 8];
+		alignas(32) uint32 block[8 * 8];
 
 		ReadBlock32(src, (uint8*)block, sizeof(block) / 8);
 
@@ -2139,7 +2139,7 @@ public:
 
 		#else
 
-		__aligned(uint32, 32) block[8 * 8];
+		alignas(32) uint32 block[8 * 8];
 
 		ReadBlock32(src, (uint8*)block, sizeof(block) / 8);
 
@@ -2182,7 +2182,7 @@ public:
 
 		#else
 
-		__aligned(uint32, 32) block[8 * 8];
+		alignas(32) uint32 block[8 * 8];
 
 		ReadBlock32(src, (uint8*)block, sizeof(block) / 8);
 

plugins/GSdx/GSCapture.cpp

@@ -382,7 +382,7 @@ GSCapture::GSCapture()
 {
 	m_out_dir = theApp.GetConfig("capture_out_dir", "/tmp/GSdx_Capture");
 	m_threads = theApp.GetConfig("capture_threads", 4);
-#ifdef __linux__
+#if defined(__unix__)
 	m_compression_level = theApp.GetConfig("png_compression_level", Z_BEST_SPEED);
 #endif
 }
@@ -481,7 +481,7 @@ bool GSCapture::BeginCapture(float fps, GSVector2i recomendedResolution, float a
 
 	CComQIPtr<IGSSource>(m_src)->DeliverNewSegment();
 
-#elif __linux__
+#elif defined(__unix__)
 	// Note I think it doesn't support multiple depth creation
 	GSmkdir(m_out_dir.c_str());
 
@@ -521,7 +521,7 @@ bool GSCapture::DeliverFrame(const void* bits, int pitch, bool rgba)
 		return true;
 	}
 
-#elif __linux__
+#elif defined(__unix__)
 
 	std::string out_file = m_out_dir + format("/frame.%010d.png", m_frame);
 	//GSPng::Save(GSPng::RGB_PNG, out_file, (uint8*)bits, m_size.x, m_size.y, pitch, m_compression_level);
@@ -554,7 +554,7 @@ bool GSCapture::EndCapture()
 		m_graph = NULL;
 	}
 
-#elif __linux__
+#elif defined(__unix__)
 	for(size_t i = 0; i < m_workers.size(); i++) {
 		m_workers[i]->Wait();
 	}

plugins/GSdx/GSCapture.h

@@ -42,7 +42,7 @@ class GSCapture
 	CComPtr<IGraphBuilder> m_graph;
 	CComPtr<IBaseFilter> m_src;
 
-	#elif __linux__
+	#elif defined(__unix__)
 
 	vector<GSPng::Worker*> m_workers;
 	int m_compression_level;

plugins/GSdx/GSClut.h

@@ -28,7 +28,7 @@
 
 class GSLocalMemory;
 
-__aligned(class, 32) GSClut : public GSAlignedClass<32>
+class alignas(32) GSClut : public GSAlignedClass<32>
 {
 	GSLocalMemory* m_mem;
 
@@ -37,7 +37,7 @@ __aligned(class, 32) GSClut : public GSAlignedClass<32>
 	uint32* m_buff32;
 	uint64* m_buff64;
 
-	__aligned(struct, 32) WriteState
+	struct alignas(32) WriteState
 	{
 		GIFRegTEX0 TEX0;
 		GIFRegTEXCLUT TEXCLUT;
@@ -45,7 +45,7 @@ __aligned(class, 32) GSClut : public GSAlignedClass<32>
 		bool IsDirty(const GIFRegTEX0& TEX0, const GIFRegTEXCLUT& TEXCLUT);
 	} m_write;
 
-	__aligned(struct, 32) ReadState
+	struct alignas(32) ReadState
 	{
 		GIFRegTEX0 TEX0;
 		GIFRegTEXA TEXA;

plugins/GSdx/GSDeviceDX.h

@@ -30,7 +30,7 @@ class GSDeviceDX : public GSDevice
 public:
 	#pragma pack(push, 1)
 
-	__aligned(struct, 32) VSConstantBuffer
+	struct alignas(32) VSConstantBuffer
 	{
 		GSVector4 VertexScale;
 		GSVector4 VertexOffset;
@@ -86,7 +86,7 @@ public:
 		VSSelector() : key(0) {}
 	};
 
-	__aligned(struct, 32) PSConstantBuffer
+	struct alignas(32) PSConstantBuffer
 	{
 		GSVector4 FogColor_AREF;
 		GSVector4 HalfTexel;

plugins/GSdx/GSDeviceOGL.h

@@ -116,8 +116,8 @@ public:
 
 class GSDeviceOGL final : public GSDevice
 {
-	public:
+public:
-	__aligned(struct, 32) VSConstantBuffer
+	struct alignas(32) VSConstantBuffer
 	{
 		GSVector4 Vertex_Scale_Offset;
 		GSVector4 TextureScale;
@@ -187,7 +187,7 @@ class GSDeviceOGL final : public GSDevice
 		GSSelector(uint32 k) : key(k) {}
 	};
 
-	__aligned(struct, 32) PSConstantBuffer
+	struct alignas(32) PSConstantBuffer
 	{
 		GSVector4 FogColor_AREF;
 		GSVector4 WH;

plugins/GSdx/GSDeviceSW.cpp

@@ -186,7 +186,7 @@ public:
 	}
 };
 
-__aligned(class, 16) ShaderFactorBlend : public ShaderBase
+class alignas(16) ShaderFactorBlend : public ShaderBase
 {
 	GSVector4i m_f;
 

plugins/GSdx/GSDrawScanlineCodeGenerator.cpp

@@ -24,7 +24,7 @@
 
 #if _M_SSE >= 0x501
 
-__aligned(const uint8, 8) GSDrawScanlineCodeGenerator::m_test[16][8] =
+alignas(8) const uint8 GSDrawScanlineCodeGenerator::m_test[16][8] =
 {
 	{0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00},
 	{0xff, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00},

plugins/GSdx/GSDrawScanlineCodeGenerator.h

@@ -135,7 +135,7 @@ public:
 	GSDrawScanlineCodeGenerator(void* param, uint64 key, void* code, size_t maxsize);
 
 	#if _M_SSE >= 0x501
-	static __aligned(const uint8, 8) m_test[16][8];
+	alignas(8) static const uint8 m_test[16][8];
 	static const GSVector8 m_log2_coef[4];
 	#else
 	static const GSVector4i m_test[8];

plugins/GSdx/GSDrawingContext.h

@@ -24,7 +24,7 @@
 #include "GS.h"
 #include "GSLocalMemory.h"
 
-__aligned(class, 32) GSDrawingContext
+class alignas(32) GSDrawingContext
 {
 public:
 	GIFRegXYOFFSET	XYOFFSET;

plugins/GSdx/GSDrawingEnvironment.h

@@ -23,7 +23,7 @@
 
 #include "GS.h"
 
-__aligned(class, 32) GSDrawingEnvironment
+class alignas(32) GSDrawingEnvironment
 {
 public:
 	GIFRegPRIM			PRIM;

plugins/GSdx/GSLocalMemory.cpp

@@ -770,7 +770,7 @@ void GSLocalMemory::WriteImageLeftRight(int l, int r, int y, int h, const uint8*
 template<int psm, int bsx, int bsy, int trbpp>
 void GSLocalMemory::WriteImageTopBottom(int l, int r, int y, int h, const uint8* src, int srcpitch, const GIFRegBITBLTBUF& BITBLTBUF)
 {
-	__aligned(uint8, 32) buff[64]; // merge buffer for one column
+	alignas(32) uint8 buff[64]; // merge buffer for one column
 
 	uint32 bp = BITBLTBUF.DBP;
 	uint32 bw = BITBLTBUF.DBW;
@@ -2055,7 +2055,7 @@ uint32* GSOffset::GetPages(const GSVector4i& rect, uint32* pages, GSVector4i* bb
 		pages = new uint32[limit];
 	}
 
-	__aligned(uint32, 16) tmp[16];
+	alignas(16) uint32 tmp[16];
 
 	((GSVector4i*)tmp)[0] = GSVector4i::zero();
 	((GSVector4i*)tmp)[1] = GSVector4i::zero();

plugins/GSdx/GSLocalMemory.h

@@ -30,13 +30,13 @@
 class GSOffset : public GSAlignedClass<32>
 {
 public:
-	__aligned(struct, 32) Block
+	struct alignas(32) Block
 	{
 		short row[256]; // yn (n = 0 8 16 ...)
 		short* col; // blockOffset*
 	};
 	
-	__aligned(struct, 32) Pixel
+	struct alignas(32) Pixel
 	{
 		int row[4096]; // yn (n = 0 1 2 ...) NOTE: this wraps around above 2048, only transfers should address the upper half (dark cloud 2 inventing)
 		int* col[8]; // rowOffset*
@@ -95,7 +95,7 @@ public:
 	typedef void (GSLocalMemory::*readTexture)(const GSOffset* RESTRICT off, const GSVector4i& r, uint8* dst, int dstpitch, const GIFRegTEXA& TEXA);
 	typedef void (GSLocalMemory::*readTextureBlock)(uint32 bp, uint8* dst, int dstpitch, const GIFRegTEXA& TEXA) const;
 
-	__aligned(struct, 128) psm_t
+	struct alignas(128) psm_t
 	{
 		pixelAddress pa, bn;
 		readPixel rp;

plugins/GSdx/GSLzma.cpp

@@ -21,7 +21,7 @@
 #include "stdafx.h"
 #include "GSLzma.h"
 
-#ifdef __linux__
+#if defined(__unix__)
 
 GSDumpFile::GSDumpFile(char* filename) {
 	m_fp = fopen(filename, "rb");

plugins/GSdx/GSLzma.h

@@ -18,7 +18,7 @@
  *
  */
 
-#ifdef __linux__
+#if defined(__unix__)
 
 #ifdef LZMA_SUPPORTED
 #include <lzma.h>

plugins/GSdx/GSPerfMon.cpp

@@ -38,7 +38,7 @@ void GSPerfMon::Put(counter_t c, double val)
 #ifndef DISABLE_PERF_MON
 	if(c == Frame)
 	{
-#ifdef __linux__
+#if defined(__unix__)
 		// clock on linux will return CLOCK_PROCESS_CPUTIME_ID.
 		// CLOCK_THREAD_CPUTIME_ID is much more useful to measure the fps
 		struct timespec ts;

plugins/GSdx/GSRasterizer.h

@@ -28,7 +28,7 @@
 #include "GSPerfMon.h"
 #include "GSThread_CXX11.h"
 
-__aligned(class, 32) GSRasterizerData : public GSAlignedClass<32>
+class alignas(32) GSRasterizerData : public GSAlignedClass<32>
 {
 	static int s_counter;
 
@@ -122,7 +122,7 @@ public:
 	virtual void PrintStats() = 0;
 };
 
-__aligned(class, 32) GSRasterizer : public IRasterizer
+class alignas(32) GSRasterizer : public IRasterizer
 {
 protected:
 	GSPerfMon* m_perfmon;

plugins/GSdx/GSRenderer.cpp

@@ -21,7 +21,7 @@
 
 #include "stdafx.h"
 #include "GSRenderer.h"
-#ifdef __linux__
+#if defined(__unix__)
 #include <X11/keysym.h>
 #endif
 
@@ -585,7 +585,7 @@ void GSRenderer::KeyEvent(GSKeyEventData* e)
 		}
 
 	}
-#elif defined(__linux__)
+#elif defined(__unix__)
 	if(e->type == KEYPRESS)
 	{
 		int step = m_shift_key ? -1 : 1;

plugins/GSdx/GSRendererCL.h

@@ -25,7 +25,7 @@
 
 #ifdef ENABLE_OPENCL
 
-__aligned(struct, 32) GSVertexCL
+struct alignas(32) GSVertexCL
 {
 	GSVector4 p, t;
 };
@@ -145,7 +145,7 @@ class GSRendererCL : public GSRenderer
 		}
 	};
 
-	__aligned(struct, 32) TFXParameter
+	struct alignas(32) TFXParameter
 	{
 		GSVector4i scissor;
 		GSVector4i dimx; // 4x4 signed char

plugins/GSdx/GSRendererCS.h

@@ -44,7 +44,7 @@ class GSRendererCS : public GSRenderer
 		VSSelector() : key(0) {}
 	};
 
-	__aligned(struct, 32) VSConstantBuffer
+	struct alignas(32) VSConstantBuffer
 	{
 		GSVector4 VertexScale;
 		GSVector4 VertexOffset;
@@ -86,7 +86,7 @@ class GSRendererCS : public GSRenderer
 		PSSelector() : key(0) {}
 	};
 
-	__aligned(struct, 32) PSConstantBuffer
+	struct alignas(32) PSConstantBuffer
 	{
 		uint32 fm;
 		uint32 zm;

plugins/GSdx/GSRendererSW.h

@@ -29,7 +29,7 @@ class GSRendererSW : public GSRenderer
 {
 	class SharedData : public GSDrawScanline::SharedData
 	{
-		__aligned(struct, 16) TextureLevel 
+		struct alignas(16) TextureLevel
 		{
 			GSVector4i r; 
 			GSTextureCacheSW::Texture* t;

plugins/GSdx/GSScanlineEnvironment.h

@@ -105,7 +105,7 @@ union GSScanlineSelector
 	}
 };
 
-__aligned(struct, 32) GSScanlineGlobalData // per batch variables, this is like a pixel shader constant buffer
+struct alignas(32) GSScanlineGlobalData // per batch variables, this is like a pixel shader constant buffer
 {
 	GSScanlineSelector sel;
 
@@ -150,7 +150,7 @@ __aligned(struct, 32) GSScanlineGlobalData // per batch variables, this is like
 	#endif
 };
 
-__aligned(struct, 32) GSScanlineLocalData // per prim variables, each thread has its own
+struct alignas(32) GSScanlineLocalData // per prim variables, each thread has its own
 {
 	#if _M_SSE >= 0x501
 

plugins/GSdx/GSSetting.cpp

@@ -21,7 +21,7 @@
 
 #include "stdafx.h"
 #include "GSSetting.h"
-#ifndef __linux__
+#ifdef _WIN32
 #include "resource.h"
 #endif
 

plugins/GSdx/GSSetting.h

@@ -40,7 +40,7 @@ struct GSSetting
 
 const char* dialog_message(int ID, bool* updateText = NULL);
 
-#ifdef __linux__
+#ifndef _WIN32
 enum {
 	IDC_FILTER,
 	IDC_SKIPDRAWHACK,

plugins/GSdx/GSState.cpp

@@ -56,7 +56,7 @@ GSState::GSState()
 	s_savef = !!theApp.GetConfig("savef", 0);
 	s_saven = theApp.GetConfig("saven", 0);
 	s_savel = theApp.GetConfig("savel", 5000);
-#ifdef __linux__
+#if defined(__unix__)
 	if (s_dump) {
 		GSmkdir("/tmp/GS_HW_dump");
 		GSmkdir("/tmp/GS_SW_dump");

plugins/GSdx/GSUtil.h

@@ -65,6 +65,6 @@ public:
 #endif
 };
 
-#ifdef __linux__
+#if defined(__unix__)
 void GSmkdir(const char* dir);
 #endif

plugins/GSdx/GSVector.h

@@ -90,7 +90,7 @@ class GSVector8i;
 
 #endif
 
-__aligned(class, 16) GSVector4i
+class alignas(16) GSVector4i
 {
 	static const GSVector4i m_xff[17];
 	static const GSVector4i m_x0f[17];
@@ -2421,7 +2421,7 @@ public:
 	__forceinline static GSVector4i x0f(int n) {return m_x0f[n];}
 };
 
-__aligned(class, 16) GSVector4
+class alignas(16) GSVector4
 {
 public:
 	union
@@ -3341,7 +3341,7 @@ GSVector.h:2973:15: error:  shadows template parm 'int i'
 
 #if _M_SSE >= 0x501
 
-__aligned(class, 32) GSVector8i
+class alignas(32) GSVector8i
 {
 	static const GSVector8i m_xff[33];
 	static const GSVector8i m_x0f[33];
@@ -5133,7 +5133,7 @@ public:
 
 #if _M_SSE >= 0x500
 
-__aligned(class, 32) GSVector8
+class alignas(32) GSVector8
 {
 public:
 	union

plugins/GSdx/GSVertex.h

@@ -28,7 +28,7 @@
 
 #pragma pack(push, 1)
 
-__aligned(struct, 32) GSVertex
+struct alignas(32) GSVertex
 {
 	union
 	{
@@ -52,7 +52,7 @@ struct GSVertexP
 	GSVector4 p;
 };
 
-__aligned(struct, 32) GSVertexPT1
+struct alignas(32) GSVertexPT1
 {
 	GSVector4 p;
 	GSVector2 t;

plugins/GSdx/GSVertexHW.h

@@ -26,7 +26,7 @@
 
 #pragma pack(push, 1)
 
-__aligned(struct, 32) GSVertexHW9
+struct alignas(32) GSVertexHW9
 {
 	GSVector4 t; 
 	GSVector4 p;

plugins/GSdx/GSVertexSW.h

@@ -23,7 +23,7 @@
 
 #include "GSVector.h"
 
-__aligned(struct, 32) GSVertexSW
+struct alignas(32) GSVertexSW
 {
 	GSVector4 p, _pad, t, c;
 
@@ -237,7 +237,7 @@ __aligned(struct, 32) GSVertexSW
 
 #if _M_SSE >= 0x501
 
-__aligned(struct, 32) GSVertexSW2
+struct alignas(32) GSVertexSW2
 {
 	GSVector4 p, _pad;
 	GSVector8 tc;

plugins/GSdx/GSVertexTrace.h

@@ -29,7 +29,7 @@
 
 class GSState;
 
-__aligned(class, 32) GSVertexTrace : public GSAlignedClass<32>
+class alignas(32) GSVertexTrace : public GSAlignedClass<32>
 {
 public:
 	struct Vertex {GSVector4i c; GSVector4 p, t;};

plugins/GSdx/GSWndEGL.cpp

@@ -22,7 +22,7 @@
 #include "stdafx.h"
 #include "GSWndEGL.h"
 
-#if defined(__linux__) && defined(EGL_SUPPORTED)
+#if defined(__unix__) && defined(EGL_SUPPORTED)
 
 GSWndEGL::GSWndEGL()
 	: m_NativeWindow(0), m_NativeDisplay(NULL)

plugins/GSdx/GSWndEGL.h

@@ -21,7 +21,7 @@
 
 #include "GSWnd.h"
 
-#if defined(__linux__) && defined(EGL_SUPPORTED)
+#if defined(__unix__) && defined(EGL_SUPPORTED)
 #include <X11/Xlib.h>
 #include <EGL/egl.h>
 #include <EGL/eglext.h>

plugins/GSdx/GSWndOGL.cpp

@@ -22,7 +22,7 @@
 #include "stdafx.h"
 #include "GSWndOGL.h"
 
-#if defined(__linux__)
+#if defined(__unix__)
 GSWndOGL::GSWndOGL()
 	: m_NativeWindow(0), m_NativeDisplay(NULL), m_context(0), m_swapinterval(NULL)
 {

plugins/GSdx/GSWndOGL.h

@@ -21,7 +21,7 @@
 
 #include "GSWnd.h"
 
-#if defined(__linux__)
+#if defined(__unix__)
 #include <X11/Xlib.h>
 #include <GL/glx.h>
 

plugins/GSdx/GSdx.cpp

@@ -172,7 +172,7 @@ GSdxApp::GSdxApp()
 	m_gs_upscale_multiplier.push_back(GSSetting(5, "5x Native", ""));
 	m_gs_upscale_multiplier.push_back(GSSetting(6, "6x Native", ""));
 	m_gs_upscale_multiplier.push_back(GSSetting(8, "8x Native", ""));
-#ifndef __linux__
+#ifndef __unix__
 	m_gs_upscale_multiplier.push_back(GSSetting(0, "Custom", ""));
 #endif
 
@@ -237,7 +237,7 @@ GSdxApp::GSdxApp()
 	m_gpu_scale.push_back(GSSetting(2 | (2 << 2), "H x 4 - V x 4", ""));
 }
 
-#ifdef __linux__
+#if defined(__unix__)
 void GSdxApp::ReloadConfig()
 {
 	if (m_configuration_map.empty()) return;

plugins/GSdx/GSdx.h

@@ -27,7 +27,7 @@ class GSdxApp
 {
 	std::string m_ini;
 	std::string m_section;
-#ifdef __linux__
+#if defined(__unix__)
 	std::map< std::string, std::string > m_configuration_map;
 #endif
 
@@ -40,7 +40,7 @@ public:
  	HMODULE GetModuleHandle() {return (HMODULE)GetModuleHandlePtr();}
 #endif
 
-#ifdef __linux__
+#if defined(__unix__)
 	void BuildConfigurationMap(const char* lpFileName);
 	void ReloadConfig();
 

plugins/GSdx/config.h

@@ -45,7 +45,7 @@
 //#define ENABLE_OGL_DEBUG_MEM_BW // compute the quantity of data transfered (debug purpose)
 #endif
 
-#if defined(__linux__) && !(defined(_DEBUG) || defined(_DEVEL))
+#if defined(__unix__) && !(defined(_DEBUG) || defined(_DEVEL))
 #define DISABLE_PERF_MON // Burn cycle for nothing in release mode
 #endif
 

plugins/GSdx/stdafx.h

@@ -197,18 +197,20 @@ using namespace stdext;
 #endif
 
 #ifdef _MSC_VER
-
+#if _MSC_VER < 1900
-    #define __aligned(t, n) __declspec(align(n)) t
+    #define alignas(n) __declspec(align(n))
+#endif
 
     #define EXPORT_C_(type) extern "C" __declspec(dllexport) type __stdcall
     #define EXPORT_C EXPORT_C_(void)
 
-    #define ALIGN_STACK(n) __aligned(int, n) __dummy;
+    #define ALIGN_STACK(n) alignas(n) int dummy__;
 
 #else
 
-    #define __aligned(t, n) t __attribute__((aligned(n)))
+    #ifndef __fastcall
-    #define __fastcall __attribute__((fastcall))
+        #define __fastcall __attribute__((fastcall))
+    #endif
 
     #define EXPORT_C_(type) extern "C" __attribute__((stdcall,externally_visible,visibility("default"))) type
     #define EXPORT_C EXPORT_C_(void)
@@ -227,7 +229,7 @@ using namespace stdext;
     #else
 
         // TODO Check clang behavior
-        #define ALIGN_STACK(n) __aligned(int, n) __dummy;
+        #define ALIGN_STACK(n) alignas(n) int dummy__;
 
     #endif
 

plugins/onepad/controller.h

@@ -21,11 +21,7 @@
 
 #pragma once
 #include <string.h> // for memset
-#ifdef __linux__
 #define MAX_KEYS 24
-#else
-#define MAX_KEYS 20
-#endif
 
 enum KeyType
 {

plugins/onepad/keyboard.cpp

@@ -28,7 +28,7 @@
 #include <gdk/gdkkeysyms.h>
 #include "keyboard.h"
 
-#ifndef __linux__
+#ifdef _WIN32
 char* KeysymToChar(int keysym)
 {
 	LPWORD temp;
@@ -40,7 +40,7 @@ char* KeysymToChar(int keysym)
 
 void SetAutoRepeat(bool autorep)
 {
- #ifdef __linux__
+#if defined(__unix__)
     if (toggleAutoRepeat)
     {
         if (autorep)
@@ -51,7 +51,7 @@ void SetAutoRepeat(bool autorep)
 #endif
 }
 
-#ifdef __linux__
+#if defined(__unix__)
 static bool s_grab_input = false;
 static bool s_Shift = false;
 static unsigned int  s_previous_mouse_x = 0;

plugins/onepad/keyboard.h

@@ -24,7 +24,7 @@
 
 #include "onepad.h"
 
-#ifdef __linux__
+#if defined(__unix__)
 
 #include "Linux/linux.h"
 

plugins/onepad/onepad.cpp

@@ -263,7 +263,7 @@ EXPORT_C_(s32) PADopen(void *pDsp)
 	pthread_spin_init(&mutex_KeyEvent, PTHREAD_PROCESS_PRIVATE);
 	mutex_WasInit = true;
 
-#ifdef __linux__
+#if defined(__unix__)
 	GamePad::EnumerateGamePads(s_vgamePad);
 #endif
 	return _PADopen(pDsp);
@@ -595,7 +595,7 @@ EXPORT_C_(keyEvent*) PADkeyEvent()
 	return &s_event;
 }
 
-#ifdef __linux__
+#if defined(__unix__)
 EXPORT_C_(void) PADWriteEvent(keyEvent &evt)
 {
 	// This function call be called before PADopen. Therefore we cann't

plugins/onepad/onepad.h

@@ -49,7 +49,7 @@ using namespace std;
 #define PADdefs
 #include "PS2Edefs.h"
 
-#ifdef __linux__
+#if defined(__unix__)
 #include "GamePad.h"
 #endif
 #include "bitwise.h"

plugins/spu2-x/src/Linux/Config.cpp

@@ -19,8 +19,10 @@
 #include "Dialogs.h"
 #include "Config.h"
 
+#ifdef __linux__
 #include <SDL.h>
 #include <SDL_audio.h>
+#endif
 
 #ifdef PCSX2_DEVBUILD
 static const int LATENCY_MAX = 3000;
@@ -121,12 +123,20 @@ void ReadSettings()
 	OutputModule = FindOutputModuleById( temp.c_str() );// find the driver index of this module
 
 	// find current API
+#ifdef __linux__
 	CfgReadStr( L"PORTAUDIO", L"HostApi", temp, L"ALSA" );
 	OutputAPI = -1;
 	if (temp == L"ALSA") OutputAPI = 0;
 	if (temp == L"OSS")  OutputAPI = 1;
 	if (temp == L"JACK") OutputAPI = 2;
+#else
+	CfgReadStr( L"PORTAUDIO", L"HostApi", temp, L"OSS" );
+	OutputAPI = -1;
 
+	if (temp == L"OSS")  OutputAPI = 0;
+#endif
+
+#ifdef __linux__
 	CfgReadStr( L"SDL", L"HostApi", temp, L"pulseaudio" );
 	SdlOutputAPI = -1;
 #if SDL_MAJOR_VERSION >= 2
@@ -135,13 +145,14 @@ void ReadSettings()
 		if (!temp.Cmp(wxString(SDL_GetAudioDriver(i), wxConvUTF8)))
 			SdlOutputAPI = i;
 	}
+#endif
 #endif
 
 	SndOutLatencyMS = CfgReadInt(L"OUTPUT",L"Latency", 300);
 	SynchMode = CfgReadInt( L"OUTPUT", L"Synch_Mode", 0);
 
 	PortaudioOut->ReadSettings();
-#ifndef __APPLE__
+#ifdef __linux__
 	SDLOut->ReadSettings();
 #endif
 	SoundtouchCfg::ReadSettings();
@@ -187,7 +198,7 @@ void WriteSettings()
 	CfgWriteInt(L"DEBUG", L"DelayCycles", delayCycles);
 
 	PortaudioOut->WriteSettings();
-#ifndef __APPLE__
+#ifdef __linux__
 	SDLOut->WriteSettings();
 #endif
 	SoundtouchCfg::WriteSettings();
@@ -204,7 +215,7 @@ void debug_dialog()
 	DebugConfig::DisplayDialog();
 }
 
-#ifdef __linux__
+#if defined(__unix__)
 void DisplayDialog()
 {
     int return_value;
@@ -257,16 +268,22 @@ void DisplayDialog()
     mod_box = gtk_combo_box_text_new ();
     gtk_combo_box_text_append_text(GTK_COMBO_BOX_TEXT(mod_box), "0 - No Sound (emulate SPU2 only)");
     gtk_combo_box_text_append_text(GTK_COMBO_BOX_TEXT(mod_box), "1 - PortAudio (cross-platform)");
+#ifdef __linux__
     gtk_combo_box_text_append_text(GTK_COMBO_BOX_TEXT(mod_box), "2 - SDL Audio (recommended for PulseAudio)");
+#endif
     //gtk_combo_box_text_append_text(GTK_COMBO_BOX_TEXT(mod_box), "3 - Alsa (probably doesn't work)");
     gtk_combo_box_set_active(GTK_COMBO_BOX(mod_box), OutputModule);
 
     api_label = gtk_label_new ("PortAudio API:");
     api_box = gtk_combo_box_text_new ();
+#ifdef __linux__
 	// In order to keep it the menu light, I only put linux major api
     gtk_combo_box_text_append_text(GTK_COMBO_BOX_TEXT(api_box), "0 - ALSA (recommended)");
     gtk_combo_box_text_append_text(GTK_COMBO_BOX_TEXT(api_box), "1 - OSS (legacy)");
     gtk_combo_box_text_append_text(GTK_COMBO_BOX_TEXT(api_box), "2 - JACK");
+#else
+    gtk_combo_box_text_append_text(GTK_COMBO_BOX_TEXT(api_box), "OSS");
+#endif
     gtk_combo_box_set_active(GTK_COMBO_BOX(api_box), OutputAPI);
 
 #if SDL_MAJOR_VERSION >= 2
@@ -360,12 +377,19 @@ void DisplayDialog()
 
     	if (gtk_combo_box_get_active(GTK_COMBO_BOX(api_box)) != -1) {
 			OutputAPI = gtk_combo_box_get_active(GTK_COMBO_BOX(api_box));
+#ifdef __linux__
 			switch(OutputAPI) {
 				case 0: PortaudioOut->SetApiSettings(L"ALSA"); break;
 				case 1: PortaudioOut->SetApiSettings(L"OSS"); break;
 				case 2: PortaudioOut->SetApiSettings(L"JACK"); break;
 				default: PortaudioOut->SetApiSettings(L"Unknown");
 			}
+#else
+			switch(OutputAPI) {
+				case 0: PortaudioOut->SetApiSettings(L"OSS"); break;
+				default: PortaudioOut->SetApiSettings(L"Unknown");
+			}
+#endif
 		}
 
 #if SDL_MAJOR_VERSION >= 2

plugins/spu2-x/src/Linux/Config.h

@@ -18,7 +18,7 @@
 #ifndef CONFIG_H_INCLUDED
 #define CONFIG_H_INCLUDED
 
-#ifdef __linux__
+#if defined(__unix__)
 #include <gtk/gtk.h>
 #endif
 

plugins/spu2-x/src/Linux/ConfigDebug.cpp

@@ -154,7 +154,7 @@ void WriteSettings()
 
 }
 
-#ifdef __linux__
+#ifdef __unix__
 void DisplayDialog()
 {
     GtkWidget *dialog;

plugins/spu2-x/src/Linux/ConfigSoundTouch.cpp

@@ -71,7 +71,7 @@ namespace SoundtouchCfg
 		CfgWriteInt( L"SOUNDTOUCH", L"OverlapMS", OverlapMS );
 	}
 
-#ifdef __linux__
+#ifdef __unix__
 	static GtkWidget *seq_label, *seek_label, *over_label;
 	static GtkWidget *seq_slide, *seek_slide, *over_slide;
 

plugins/spu2-x/src/Linux/Dialogs.cpp

@@ -20,7 +20,7 @@
 #include "Dialogs.h"
 #include <cstring>
 
-#ifdef __linux__
+#if defined(__unix__)
 #include <gtk/gtk.h>
 
 void SysMessage(const char *fmt, ...)