MOVBU	internal∕cpu·ARM64+const_offsetARM64HasATOMICS(SB), R4
	CBZ 	R4, load_store_loop
#endif
	SWPALD	R1, (R0), R2
	MOVD	R2, ret+16(FP)
	RET
#ifndef GOARM64_LSE
load_store_loop:
	LDAXR	(R0), R2
	STLXR	R1, (R0), R3
	CBNZ	R3, load_store_loop
	MOVD	R2, ret+16(FP)
	RET
#endif

// bool Cas(uint32 *ptr, uint32 old, uint32 new)
// Atomically:
//	if(*val == old){
//		*val = new;
//		return 1;