#include "libc.h"

#if __ARM_ARCH_4__ || __ARM_ARCH_4T__ || __ARM_ARCH == 4
#define BLX "mov lr,pc\n\tbx"
#else
#define BLX "blx"
#endif

extern hidden uintptr __atomic·casptr, __atomic·barrierptr;

#if((__ARM_ARCH_6__ || __ARM_ARCH_6K__ || __ARM_ARCH_6KZ__ || __ARM_ARCH_6ZK__) && !__thumb__) \
 || __ARM_ARCH_6T2__ || __ARM_ARCH_7A__ || __ARM_ARCH_7R__ || __ARM_ARCH >= 7

static inline int
atomic·ll(volatile int *p)
{
	int v;
	__asm__ __volatile__ ("ldrex %0, %1" : "=r"(v) : "Q"(*p));
	return v;
}

static inline int
atomic·sc(volatile int *p, int v)
{
	int r;
	__asm__ __volatile__ ("strex %0,%2,%1" : "=&r"(r), "=Q"(*p) : "r"(v) : "memory");
	return !r;
}

#if __ARM_ARCH_7A__ || __ARM_ARCH_7R__ ||  __ARM_ARCH >= 7

static inline void
atomic·barrier()
{
	__asm__ __volatile__ ("dmb ish" : : : "memory");
}

#endif

#define atomic·prellsc atomic·barrier
#define atomic·postllsc atomic·barrier

#else

static inline int
atomic·cas(volatile int *p, int t, int s)
{
	for(;;){
		register int r0 __asm__("r0") = t;
		register int r1 __asm__("r1") = s;
		register volatile int *r2 __asm__("r2") = p;
		register uintptr r3 __asm__("r3") = __atomic·casptr;
		int old;
		__asm__ __volatile__ (
			BLX " r3"
			: "+r"(r0), "+r"(r3) : "r"(r1), "r"(r2)
			: "memory", "lr", "ip", "cc" );
		if(!r0) return t;
		if((old=*p)!=t) return old;
	}
}

#endif

#ifndef atomic·barrier
static inline void
atomic·barrier()
{
	register uintptr ip __asm__("ip") = __atomic·barrierptr;
	__asm__ __volatile__( BLX " ip" : "+r"(ip) : : "memory", "cc", "lr" );
}
#endif

static inline void
atomic·crash()
{
	__asm__ __volatile__(
#ifndef __thumb__
		".word 0xe7f000f0"
#else
		".short 0xdeff"
#endif
		: : : "memory");
}

#if __ARM_ARCH >= 5 && (!__thumb__ || __thumb2__)

static inline int
atomic·clz32(uint32 x)
{
	__asm__ ("clz %0, %1" : "=r"(x) : "r"(x));
	return x;
}

#if __ARM_ARCH_6T2__ || __ARM_ARCH_7A__ || __ARM_ARCH_7R__ || __ARM_ARCH >= 7

static inline int
atomic·ctz32(uint32 x)
{
	uint32 xr;
	__asm__ ("rbit %0, %1" : "=r"(xr) : "r"(x));
	return atomic·clz32(xr);
}

#endif

#endif