/*

Simple DirectMedia Layer

Copyright (C) 1997-2020 Sam Lantinga <slouken@libsdl.org>

This software is provided 'as-is', without any express or implied

warranty. In no event will the authors be held liable for any damages

arising from the use of this software.

Permission is granted to anyone to use this software for any purpose,

including commercial applications, and to alter it and redistribute it

freely, subject to the following restrictions:

1. The origin of this software must not be misrepresented; you must not

claim that you wrote the original software. If you use this software

in a product, an acknowledgment in the product documentation would be

appreciated but is not required.

2. Altered source versions must be plainly marked as such, and must not be

misrepresented as being the original software.

3. This notice may not be removed or altered from any source distribution.

*/

#include "../SDL_internal.h"

#if defined(__WIN32__) || defined(__WINRT__)

#include "../core/windows/SDL_windows.h"

#endif

#include "SDL_atomic.h"

#include "SDL_mutex.h"

#include "SDL_timer.h"

#if !defined(HAVE_GCC_ATOMICS) && defined(__SOLARIS__)

#include <atomic.h>

#endif

#if !defined(HAVE_GCC_ATOMICS) && defined(__RISCOS__)

#include <unixlib/local.h>

#endif

#if defined(_MSC_VER) && (defined(_M_IX86) || defined(_M_X64))

#include <xmmintrin.h>

#endif

#if defined(__WATCOMC__) && defined(__386__)

SDL_COMPILE_TIME_ASSERT(locksize, 4==sizeof(SDL_SpinLock));

extern _inline int _SDL_xchg_watcom(volatile int *a, int v);

#pragma aux _SDL_xchg_watcom = \

"lock xchg [ecx], eax" \

parm [ecx] [eax] \

value [eax] \

modify exact [eax];

#endif /* __WATCOMC__ && __386__ */

/* This function is where all the magic happens... */

SDL_bool

SDL_AtomicTryLock(SDL_SpinLock *lock)

{

#if SDL_ATOMIC_DISABLED

/* Terrible terrible damage */

static SDL_mutex *_spinlock_mutex;

if (!_spinlock_mutex) {

/* Race condition on first lock... */

_spinlock_mutex = SDL_CreateMutex();

}

SDL_LockMutex(_spinlock_mutex);

if (*lock == 0) {

*lock = 1;

SDL_UnlockMutex(_spinlock_mutex);

return SDL_TRUE;

} else {

SDL_UnlockMutex(_spinlock_mutex);

return SDL_FALSE;

}

#elif defined(_MSC_VER)

SDL_COMPILE_TIME_ASSERT(locksize, sizeof(*lock) == sizeof(long));

return (InterlockedExchange((long*)lock, 1) == 0);

#elif defined(__WATCOMC__) && defined(__386__)

return _SDL_xchg_watcom(lock, 1) == 0;

#elif HAVE_GCC_ATOMICS || HAVE_GCC_SYNC_LOCK_TEST_AND_SET

return (__sync_lock_test_and_set(lock, 1) == 0);

#elif defined(__GNUC__) && defined(__arm__) && \

(defined(__ARM_ARCH_3__) || defined(__ARM_ARCH_3M__) || \

defined(__ARM_ARCH_4__) || defined(__ARM_ARCH_4T__) || \

defined(__ARM_ARCH_5__) || defined(__ARM_ARCH_5TE__) || \

defined(__ARM_ARCH_5TEJ__))

int result;

#if defined(__RISCOS__)

if (__cpucap_have_rex()) {

__asm__ __volatile__ (

"ldrex %0, [%2]\nteq %0, #0\nstrexeq %0, %1, [%2]"

: "=&r" (result) : "r" (1), "r" (lock) : "cc", "memory");

return (result == 0);

}

#endif

__asm__ __volatile__ (

"swp %0, %1, [%2]\n"

: "=&r,&r" (result) : "r,0" (1), "r,r" (lock) : "memory");

return (result == 0);

#elif defined(__GNUC__) && defined(__arm__)

int result;

__asm__ __volatile__ (

"ldrex %0, [%2]\nteq %0, #0\nstrexeq %0, %1, [%2]"

: "=&r" (result) : "r" (1), "r" (lock) : "cc", "memory");

return (result == 0);

#elif defined(__GNUC__) && (defined(__i386__) || defined(__x86_64__))

int result;

__asm__ __volatile__(

"lock ; xchgl %0, (%1)\n"

: "=r" (result) : "r" (lock), "0" (1) : "cc", "memory");

return (result == 0);

#elif defined(__MACOSX__) || defined(__IPHONEOS__)

/* Maybe used for PowerPC, but the Intel asm or gcc atomics are favored. */

return OSAtomicCompareAndSwap32Barrier(0, 1, lock);

#elif defined(__SOLARIS__) && defined(_LP64)

/* Used for Solaris with non-gcc compilers. */

return (SDL_bool) ((int) atomic_cas_64((volatile uint64_t*)lock, 0, 1) == 0);

#elif defined(__SOLARIS__) && !defined(_LP64)

/* Used for Solaris with non-gcc compilers. */

return (SDL_bool) ((int) atomic_cas_32((volatile uint32_t*)lock, 0, 1) == 0);

#else

#error Please implement for your platform.

return SDL_FALSE;

#endif

}

/* "REP NOP" is PAUSE, coded for tools that don't know it by that name. */

#if (defined(__GNUC__) || defined(__clang__)) && (defined(__i386__) || defined(__x86_64__))

#define PAUSE_INSTRUCTION() __asm__ __volatile__("pause\n") /* Some assemblers can't do REP NOP, so go with PAUSE. */

#elif defined(_MSC_VER) && (defined(_M_IX86) || defined(_M_X64))

#define PAUSE_INSTRUCTION() _mm_pause() /* this is actually "rep nop" and not a SIMD instruction. No inline asm in MSVC x86-64! */

#elif defined(__WATCOMC__) && defined(__386__)

/* watcom assembler rejects PAUSE if CPU < i686, and it refuses REP NOP as an invalid combination. Hardcode the bytes. */

extern _inline void PAUSE_INSTRUCTION(void);

#pragma aux PAUSE_INSTRUCTION = "db 0f3h,90h"

#else

#define PAUSE_INSTRUCTION()

#endif

void

SDL_AtomicLock(SDL_SpinLock *lock)

{

int iterations = 0;

/* FIXME: Should we have an eventual timeout? */

while (!SDL_AtomicTryLock(lock)) {

if (iterations < 32) {

iterations++;

PAUSE_INSTRUCTION();

} else {

/* !!! FIXME: this doesn't definitely give up the current timeslice, it does different things on various platforms. */

SDL_Delay(0);

}

}

}

void

SDL_AtomicUnlock(SDL_SpinLock *lock)

{

#if defined(_MSC_VER)

_ReadWriteBarrier();

*lock = 0;

#elif defined(__WATCOMC__) && defined(__386__)

SDL_CompilerBarrier ();

*lock = 0;

#elif HAVE_GCC_ATOMICS || HAVE_GCC_SYNC_LOCK_TEST_AND_SET

__sync_lock_release(lock);

#elif defined(__SOLARIS__)

/* Used for Solaris when not using gcc. */

*lock = 0;

membar_producer();

#else

*lock = 0;

#endif

}

/* vi: set ts=4 sw=4 expandtab: */