[thirdparty/linux.git] / arch / arm64 / include / asm / cmpxchg.h

/* SPDX-License-Identifier: GPL-2.0-only */
/*
 * Based on arch/arm/include/asm/cmpxchg.h
 *
 * Copyright (C) 2012 ARM Ltd.
 */
#ifndef __ASM_CMPXCHG_H
#define __ASM_CMPXCHG_H

#include <linux/build_bug.h>
#include <linux/compiler.h>

#include <asm/atomic.h>
#include <asm/barrier.h>
#include <asm/lse.h>

/*
 * We need separate acquire parameters for ll/sc and lse, since the full
 * barrier case is generated as release+dmb for the former and
 * acquire+release for the latter.
 */
#define __XCHG_CASE(w, sfx, name, sz, mb, nop_lse, acq, acq_lse, rel, cl)	\
static inline u##sz __xchg_case_##name##sz(u##sz x, volatile void *ptr)		\
{										\
	u##sz ret;								\
	unsigned long tmp;							\
										\
	asm volatile(ARM64_LSE_ATOMIC_INSN(					\
	/* LL/SC */								\
	"	prfm	pstl1strm, %2\n"					\
	"1:	ld" #acq "xr" #sfx "\t%" #w "0, %2\n"				\
	"	st" #rel "xr" #sfx "\t%w1, %" #w "3, %2\n"			\
	"	cbnz	%w1, 1b\n"						\
	"	" #mb,								\
	/* LSE atomics */							\
	"	swp" #acq_lse #rel #sfx "\t%" #w "3, %" #w "0, %2\n"		\
		__nops(3)							\
	"	" #nop_lse)							\
	: "=&r" (ret), "=&r" (tmp), "+Q" (*(u##sz *)ptr)			\
	: "r" (x)								\
	: cl);									\
										\
	return ret;								\
}

__XCHG_CASE(w, b,     ,  8,        ,    ,  ,  ,  ,         )
__XCHG_CASE(w, h,     , 16,        ,    ,  ,  ,  ,         )
__XCHG_CASE(w,  ,     , 32,        ,    ,  ,  ,  ,         )
__XCHG_CASE( ,  ,     , 64,        ,    ,  ,  ,  ,         )
__XCHG_CASE(w, b, acq_,  8,        ,    , a, a,  , "memory")
__XCHG_CASE(w, h, acq_, 16,        ,    , a, a,  , "memory")
__XCHG_CASE(w,  , acq_, 32,        ,    , a, a,  , "memory")
__XCHG_CASE( ,  , acq_, 64,        ,    , a, a,  , "memory")
__XCHG_CASE(w, b, rel_,  8,        ,    ,  ,  , l, "memory")
__XCHG_CASE(w, h, rel_, 16,        ,    ,  ,  , l, "memory")
__XCHG_CASE(w,  , rel_, 32,        ,    ,  ,  , l, "memory")
__XCHG_CASE( ,  , rel_, 64,        ,    ,  ,  , l, "memory")
__XCHG_CASE(w, b,  mb_,  8, dmb ish, nop,  , a, l, "memory")
__XCHG_CASE(w, h,  mb_, 16, dmb ish, nop,  , a, l, "memory")
__XCHG_CASE(w,  ,  mb_, 32, dmb ish, nop,  , a, l, "memory")
__XCHG_CASE( ,  ,  mb_, 64, dmb ish, nop,  , a, l, "memory")

#undef __XCHG_CASE

#define __XCHG_GEN(sfx)							\
static inline unsigned long __xchg##sfx(unsigned long x,		\
					volatile void *ptr,		\
					int size)			\
{									\
	switch (size) {							\
	case 1:								\
		return __xchg_case##sfx##_8(x, ptr);			\
	case 2:								\
		return __xchg_case##sfx##_16(x, ptr);			\
	case 4:								\
		return __xchg_case##sfx##_32(x, ptr);			\
	case 8:								\
		return __xchg_case##sfx##_64(x, ptr);			\
	default:							\
		BUILD_BUG();						\
	}								\
									\
	unreachable();							\
}

__XCHG_GEN()
__XCHG_GEN(_acq)
__XCHG_GEN(_rel)
__XCHG_GEN(_mb)

#undef __XCHG_GEN

#define __xchg_wrapper(sfx, ptr, x)					\
({									\
	__typeof__(*(ptr)) __ret;					\
	__ret = (__typeof__(*(ptr)))					\
		__xchg##sfx((unsigned long)(x), (ptr), sizeof(*(ptr))); \
	__ret;								\
})

/* xchg */
#define arch_xchg_relaxed(...)	__xchg_wrapper(    , __VA_ARGS__)
#define arch_xchg_acquire(...)	__xchg_wrapper(_acq, __VA_ARGS__)
#define arch_xchg_release(...)	__xchg_wrapper(_rel, __VA_ARGS__)
#define arch_xchg(...)		__xchg_wrapper( _mb, __VA_ARGS__)

#define __CMPXCHG_GEN(sfx)						\
static inline unsigned long __cmpxchg##sfx(volatile void *ptr,		\
					   unsigned long old,		\
					   unsigned long new,		\
					   int size)			\
{									\
	switch (size) {							\
	case 1:								\
		return __cmpxchg_case##sfx##_8(ptr, old, new);		\
	case 2:								\
		return __cmpxchg_case##sfx##_16(ptr, old, new);		\
	case 4:								\
		return __cmpxchg_case##sfx##_32(ptr, old, new);		\
	case 8:								\
		return __cmpxchg_case##sfx##_64(ptr, old, new);		\
	default:							\
		BUILD_BUG();						\
	}								\
									\
	unreachable();							\
}

__CMPXCHG_GEN()
__CMPXCHG_GEN(_acq)
__CMPXCHG_GEN(_rel)
__CMPXCHG_GEN(_mb)

#undef __CMPXCHG_GEN

#define __cmpxchg_wrapper(sfx, ptr, o, n)				\
({									\
	__typeof__(*(ptr)) __ret;					\
	__ret = (__typeof__(*(ptr)))					\
		__cmpxchg##sfx((ptr), (unsigned long)(o),		\
				(unsigned long)(n), sizeof(*(ptr)));	\
	__ret;								\
})

/* cmpxchg */
#define arch_cmpxchg_relaxed(...)	__cmpxchg_wrapper(    , __VA_ARGS__)
#define arch_cmpxchg_acquire(...)	__cmpxchg_wrapper(_acq, __VA_ARGS__)
#define arch_cmpxchg_release(...)	__cmpxchg_wrapper(_rel, __VA_ARGS__)
#define arch_cmpxchg(...)		__cmpxchg_wrapper( _mb, __VA_ARGS__)
#define arch_cmpxchg_local		arch_cmpxchg_relaxed

/* cmpxchg64 */
#define arch_cmpxchg64_relaxed		arch_cmpxchg_relaxed
#define arch_cmpxchg64_acquire		arch_cmpxchg_acquire
#define arch_cmpxchg64_release		arch_cmpxchg_release
#define arch_cmpxchg64			arch_cmpxchg
#define arch_cmpxchg64_local		arch_cmpxchg_local

/* cmpxchg_double */
#define system_has_cmpxchg_double()     1

#define __cmpxchg_double_check(ptr1, ptr2)					\
({										\
	if (sizeof(*(ptr1)) != 8)						\
		BUILD_BUG();							\
	VM_BUG_ON((unsigned long *)(ptr2) - (unsigned long *)(ptr1) != 1);	\
})

#define arch_cmpxchg_double(ptr1, ptr2, o1, o2, n1, n2)				\
({										\
	int __ret;								\
	__cmpxchg_double_check(ptr1, ptr2);					\
	__ret = !__cmpxchg_double_mb((unsigned long)(o1), (unsigned long)(o2),	\
				     (unsigned long)(n1), (unsigned long)(n2),	\
				     ptr1);					\
	__ret;									\
})

#define arch_cmpxchg_double_local(ptr1, ptr2, o1, o2, n1, n2)			\
({										\
	int __ret;								\
	__cmpxchg_double_check(ptr1, ptr2);					\
	__ret = !__cmpxchg_double((unsigned long)(o1), (unsigned long)(o2),	\
				  (unsigned long)(n1), (unsigned long)(n2),	\
				  ptr1);					\
	__ret;									\
})

#define __CMPWAIT_CASE(w, sfx, sz)					\
static inline void __cmpwait_case_##sz(volatile void *ptr,		\
				       unsigned long val)		\
{									\
	unsigned long tmp;						\
									\
	asm volatile(							\
	"	sevl\n"							\
	"	wfe\n"							\
	"	ldxr" #sfx "\t%" #w "[tmp], %[v]\n"			\
	"	eor	%" #w "[tmp], %" #w "[tmp], %" #w "[val]\n"	\
	"	cbnz	%" #w "[tmp], 1f\n"				\
	"	wfe\n"							\
	"1:"								\
	: [tmp] "=&r" (tmp), [v] "+Q" (*(unsigned long *)ptr)		\
	: [val] "r" (val));						\
}

__CMPWAIT_CASE(w, b, 8);
__CMPWAIT_CASE(w, h, 16);
__CMPWAIT_CASE(w,  , 32);
__CMPWAIT_CASE( ,  , 64);

#undef __CMPWAIT_CASE

#define __CMPWAIT_GEN(sfx)						\
static inline void __cmpwait##sfx(volatile void *ptr,			\
				  unsigned long val,			\
				  int size)				\
{									\
	switch (size) {							\
	case 1:								\
		return __cmpwait_case##sfx##_8(ptr, (u8)val);		\
	case 2:								\
		return __cmpwait_case##sfx##_16(ptr, (u16)val);		\
	case 4:								\
		return __cmpwait_case##sfx##_32(ptr, val);		\
	case 8:								\
		return __cmpwait_case##sfx##_64(ptr, val);		\
	default:							\
		BUILD_BUG();						\
	}								\
									\
	unreachable();							\
}

__CMPWAIT_GEN()

#undef __CMPWAIT_GEN

#define __cmpwait_relaxed(ptr, val) \
	__cmpwait((ptr), (unsigned long)(val), sizeof(*(ptr)))

#endif	/* __ASM_CMPXCHG_H */
Commit	Line	Data
caab277b	1	/* SPDX-License-Identifier: GPL-2.0-only */
10b663ae CM	2	/*
	3	* Based on arch/arm/include/asm/cmpxchg.h
	4	*
	5	* Copyright (C) 2012 ARM Ltd.
10b663ae CM	6	*/
	7	#ifndef __ASM_CMPXCHG_H
	8	#define __ASM_CMPXCHG_H
	9
e8a2d040	10	#include <linux/build_bug.h>
2a58fca9	11	#include <linux/compiler.h>
10b663ae	12
c342f782	13	#include <asm/atomic.h>
10b663ae	14	#include <asm/barrier.h>
c8366ba0	15	#include <asm/lse.h>
10b663ae	16
305d454a WD	17	/*
	18	* We need separate acquire parameters for ll/sc and lse, since the full
	19	* barrier case is generated as release+dmb for the former and
	20	* acquire+release for the latter.
	21	*/
5ef3fe4c WD	22	#define __XCHG_CASE(w, sfx, name, sz, mb, nop_lse, acq, acq_lse, rel, cl) \
	23	static inline u##sz __xchg_case_##name##sz(u##sz x, volatile void *ptr) \
	24	{ \
	25	u##sz ret; \
	26	unsigned long tmp; \
	27	\
	28	asm volatile(ARM64_LSE_ATOMIC_INSN( \
	29	/* LL/SC */ \
	30	" prfm pstl1strm, %2\n" \
	31	"1: ld" #acq "xr" #sfx "\t%" #w "0, %2\n" \
	32	" st" #rel "xr" #sfx "\t%w1, %" #w "3, %2\n" \
	33	" cbnz %w1, 1b\n" \
	34	" " #mb, \
	35	/* LSE atomics */ \
	36	" swp" #acq_lse #rel #sfx "\t%" #w "3, %" #w "0, %2\n" \
	37	__nops(3) \
	38	" " #nop_lse) \
	39	: "=&r" (ret), "=&r" (tmp), "+Q" ((u##sz )ptr) \
	40	: "r" (x) \
	41	: cl); \
	42	\
	43	return ret; \
10b663ae CM	44	}
10b663ae CM	45
5ef3fe4c WD	46	__XCHG_CASE(w, b, , 8, , , , , , )
	47	__XCHG_CASE(w, h, , 16, , , , , , )
	48	__XCHG_CASE(w, , , 32, , , , , , )
	49	__XCHG_CASE( , , , 64, , , , , , )
	50	__XCHG_CASE(w, b, acq_, 8, , , a, a, , "memory")
	51	__XCHG_CASE(w, h, acq_, 16, , , a, a, , "memory")
	52	__XCHG_CASE(w, , acq_, 32, , , a, a, , "memory")
	53	__XCHG_CASE( , , acq_, 64, , , a, a, , "memory")
	54	__XCHG_CASE(w, b, rel_, 8, , , , , l, "memory")
	55	__XCHG_CASE(w, h, rel_, 16, , , , , l, "memory")
	56	__XCHG_CASE(w, , rel_, 32, , , , , l, "memory")
	57	__XCHG_CASE( , , rel_, 64, , , , , l, "memory")
	58	__XCHG_CASE(w, b, mb_, 8, dmb ish, nop, , a, l, "memory")
	59	__XCHG_CASE(w, h, mb_, 16, dmb ish, nop, , a, l, "memory")
	60	__XCHG_CASE(w, , mb_, 32, dmb ish, nop, , a, l, "memory")
	61	__XCHG_CASE( , , mb_, 64, dmb ish, nop, , a, l, "memory")
305d454a WD	62
	63	#undef __XCHG_CASE
	64
	65	#define __XCHG_GEN(sfx) \
	66	static inline unsigned long __xchg##sfx(unsigned long x, \
	67	volatile void *ptr, \
	68	int size) \
	69	{ \
	70	switch (size) { \
	71	case 1: \
5ef3fe4c	72	return __xchg_case##sfx##_8(x, ptr); \
305d454a	73	case 2: \
5ef3fe4c	74	return __xchg_case##sfx##_16(x, ptr); \
305d454a	75	case 4: \
5ef3fe4c	76	return __xchg_case##sfx##_32(x, ptr); \
305d454a	77	case 8: \
5ef3fe4c	78	return __xchg_case##sfx##_64(x, ptr); \
305d454a WD	79	default: \
	80	BUILD_BUG(); \
	81	} \
	82	\
	83	unreachable(); \
	84	}
	85
	86	__XCHG_GEN()
	87	__XCHG_GEN(_acq)
	88	__XCHG_GEN(_rel)
	89	__XCHG_GEN(_mb)
	90
	91	#undef __XCHG_GEN
	92
	93	#define __xchg_wrapper(sfx, ptr, x) \
	94	({ \
	95	__typeof__(*(ptr)) __ret; \
	96	__ret = (__typeof__(*(ptr))) \
	97	__xchg##sfx((unsigned long)(x), (ptr), sizeof(*(ptr))); \
	98	__ret; \
e1dfda9c	99	})
10b663ae	100
305d454a	101	/* xchg */
c0df1081 MR	102	#define arch_xchg_relaxed(...) __xchg_wrapper( , __VA_ARGS__)
	103	#define arch_xchg_acquire(...) __xchg_wrapper(_acq, __VA_ARGS__)
	104	#define arch_xchg_release(...) __xchg_wrapper(_rel, __VA_ARGS__)
	105	#define arch_xchg(...) __xchg_wrapper( _mb, __VA_ARGS__)
305d454a WD	106
	107	#define __CMPXCHG_GEN(sfx) \
	108	static inline unsigned long __cmpxchg##sfx(volatile void *ptr, \
	109	unsigned long old, \
	110	unsigned long new, \
	111	int size) \
	112	{ \
	113	switch (size) { \
	114	case 1: \
b4f9209b	115	return __cmpxchg_case##sfx##_8(ptr, old, new); \
305d454a	116	case 2: \
b4f9209b	117	return __cmpxchg_case##sfx##_16(ptr, old, new); \
305d454a	118	case 4: \
5ef3fe4c	119	return __cmpxchg_case##sfx##_32(ptr, old, new); \
305d454a	120	case 8: \
5ef3fe4c	121	return __cmpxchg_case##sfx##_64(ptr, old, new); \
305d454a WD	122	default: \
	123	BUILD_BUG(); \
	124	} \
	125	\
	126	unreachable(); \
10b663ae CM	127	}
10b663ae CM	128
305d454a WD	129	__CMPXCHG_GEN()
	130	__CMPXCHG_GEN(_acq)
	131	__CMPXCHG_GEN(_rel)
	132	__CMPXCHG_GEN(_mb)
10b663ae	133
305d454a	134	#undef __CMPXCHG_GEN
60010e50	135
305d454a WD	136	#define __cmpxchg_wrapper(sfx, ptr, o, n) \
	137	({ \
	138	__typeof__(*(ptr)) __ret; \
	139	__ret = (__typeof__(*(ptr))) \
	140	__cmpxchg##sfx((ptr), (unsigned long)(o), \
	141	(unsigned long)(n), sizeof(*(ptr))); \
	142	__ret; \
60010e50	143	})
10b663ae	144
305d454a	145	/* cmpxchg */
c0df1081 MR	146	#define arch_cmpxchg_relaxed(...) __cmpxchg_wrapper( , __VA_ARGS__)
	147	#define arch_cmpxchg_acquire(...) __cmpxchg_wrapper(_acq, __VA_ARGS__)
	148	#define arch_cmpxchg_release(...) __cmpxchg_wrapper(_rel, __VA_ARGS__)
	149	#define arch_cmpxchg(...) __cmpxchg_wrapper( _mb, __VA_ARGS__)
	150	#define arch_cmpxchg_local arch_cmpxchg_relaxed
305d454a WD	151
305d454a WD	152	/* cmpxchg64 */
c0df1081 MR	153	#define arch_cmpxchg64_relaxed arch_cmpxchg_relaxed
	154	#define arch_cmpxchg64_acquire arch_cmpxchg_acquire
	155	#define arch_cmpxchg64_release arch_cmpxchg_release
	156	#define arch_cmpxchg64 arch_cmpxchg
	157	#define arch_cmpxchg64_local arch_cmpxchg_local
305d454a WD	158
305d454a WD	159	/* cmpxchg_double */
e9a4b795 WD	160	#define system_has_cmpxchg_double() 1
	161
	162	#define __cmpxchg_double_check(ptr1, ptr2) \
	163	({ \
	164	if (sizeof(*(ptr1)) != 8) \
	165	BUILD_BUG(); \
	166	VM_BUG_ON((unsigned long )(ptr2) - (unsigned long )(ptr1) != 1); \
	167	})
	168
c0df1081 MR	169	#define arch_cmpxchg_double(ptr1, ptr2, o1, o2, n1, n2) \
	170	({ \
	171	int __ret; \
	172	__cmpxchg_double_check(ptr1, ptr2); \
	173	__ret = !__cmpxchg_double_mb((unsigned long)(o1), (unsigned long)(o2), \
	174	(unsigned long)(n1), (unsigned long)(n2), \
	175	ptr1); \
	176	__ret; \
5284e1b4 SC	177	})
5284e1b4 SC	178
c0df1081 MR	179	#define arch_cmpxchg_double_local(ptr1, ptr2, o1, o2, n1, n2) \
	180	({ \
	181	int __ret; \
	182	__cmpxchg_double_check(ptr1, ptr2); \
	183	__ret = !__cmpxchg_double((unsigned long)(o1), (unsigned long)(o2), \
	184	(unsigned long)(n1), (unsigned long)(n2), \
	185	ptr1); \
	186	__ret; \
5284e1b4 SC	187	})
5284e1b4 SC	188
5ef3fe4c WD	189	#define __CMPWAIT_CASE(w, sfx, sz) \
	190	static inline void __cmpwait_case_##sz(volatile void *ptr, \
	191	unsigned long val) \
03e3c2b7 WD	192	{ \
	193	unsigned long tmp; \
	194	\
	195	asm volatile( \
1cfc63b5 WD	196	" sevl\n" \
1cfc63b5 WD	197	" wfe\n" \
5ef3fe4c	198	" ldxr" #sfx "\t%" #w "[tmp], %[v]\n" \
03e3c2b7 WD	199	" eor %" #w "[tmp], %" #w "[tmp], %" #w "[val]\n" \
	200	" cbnz %" #w "[tmp], 1f\n" \
	201	" wfe\n" \
	202	"1:" \
	203	: [tmp] "=&r" (tmp), [v] "+Q" ((unsigned long )ptr) \
	204	: [val] "r" (val)); \
	205	}
	206
5ef3fe4c WD	207	__CMPWAIT_CASE(w, b, 8);
	208	__CMPWAIT_CASE(w, h, 16);
	209	__CMPWAIT_CASE(w, , 32);
	210	__CMPWAIT_CASE( , , 64);
03e3c2b7 WD	211
	212	#undef __CMPWAIT_CASE
	213
	214	#define __CMPWAIT_GEN(sfx) \
	215	static inline void __cmpwait##sfx(volatile void *ptr, \
	216	unsigned long val, \
	217	int size) \
	218	{ \
	219	switch (size) { \
	220	case 1: \
5ef3fe4c	221	return __cmpwait_case##sfx##_8(ptr, (u8)val); \
03e3c2b7	222	case 2: \
5ef3fe4c	223	return __cmpwait_case##sfx##_16(ptr, (u16)val); \
03e3c2b7	224	case 4: \
5ef3fe4c	225	return __cmpwait_case##sfx##_32(ptr, val); \
03e3c2b7	226	case 8: \
5ef3fe4c	227	return __cmpwait_case##sfx##_64(ptr, val); \
03e3c2b7 WD	228	default: \
	229	BUILD_BUG(); \
	230	} \
	231	\
	232	unreachable(); \
	233	}
	234
	235	__CMPWAIT_GEN()
	236
	237	#undef __CMPWAIT_GEN
	238
	239	#define __cmpwait_relaxed(ptr, val) \
	240	__cmpwait((ptr), (unsigned long)(val), sizeof(*(ptr)))
	241
10b663ae	242	#endif /* __ASM_CMPXCHG_H */