[BACK]Return to atomic.h CVS log [TXT][DIR] Up to [local] / src / sys / sys

File: [local] / src / sys / sys / atomic.h (download)

Revision 1.9, Mon Mar 21 05:45:52 2022 UTC (2 years, 2 months ago) by visa
Branch: MAIN
CVS Tags: OPENBSD_7_5_BASE, OPENBSD_7_5, OPENBSD_7_4_BASE, OPENBSD_7_4, OPENBSD_7_3_BASE, OPENBSD_7_3, OPENBSD_7_2_BASE, OPENBSD_7_2, OPENBSD_7_1_BASE, OPENBSD_7_1, HEAD
Changes since 1.8: +3 -13 lines

Remove data dependency barrier from atomic_load_* functions

This makes the atomic_load_* functions relaxed in terms of memory
ordering. Now it should be acceptable to use these functions in
assertions.

The need of the data dependency barrier is conditioned to usage.
The barrier is unnecessary for the control decisions that cond_wait()
and refcnt_finalize() make. READ_ONCE() and SMR_PTR_GET() use the
barrier so that loaded pointers would work as excepted in lock-free
contexts (some Alpha CPUs have a data cache design that can cause
unusual load-load reordering if not synchronized properly).

OK bluhm@

/*	$OpenBSD: atomic.h,v 1.9 2022/03/21 05:45:52 visa Exp $ */
/*
 * Copyright (c) 2014 David Gwynne <dlg@openbsd.org>
 * Copyright (c) 2022 Alexander Bluhm <bluhm@openbsd.org>
 *
 * Permission to use, copy, modify, and distribute this software for any
 * purpose with or without fee is hereby granted, provided that the above
 * copyright notice and this permission notice appear in all copies.
 *
 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
 */

#ifndef _SYS_ATOMIC_H_
#define _SYS_ATOMIC_H_

#include <machine/atomic.h>

/*
 * an arch wanting to provide its own implementations does so by defining
 * macros.
 */

/*
 * atomic_cas_*
 */

#ifndef atomic_cas_uint
static inline unsigned int
atomic_cas_uint(volatile unsigned int *p, unsigned int o, unsigned int n)
{
	return __sync_val_compare_and_swap(p, o, n);
}
#endif

#ifndef atomic_cas_ulong
static inline unsigned long
atomic_cas_ulong(volatile unsigned long *p, unsigned long o, unsigned long n)
{
	return __sync_val_compare_and_swap(p, o, n);
}
#endif

#ifndef atomic_cas_ptr
static inline void *
atomic_cas_ptr(volatile void *pp, void *o, void *n)
{
	void * volatile *p = (void * volatile *)pp;
	return __sync_val_compare_and_swap(p, o, n);
}
#endif

/*
 * atomic_swap_*
 */

#ifndef atomic_swap_uint
static inline unsigned int
atomic_swap_uint(volatile unsigned int *p, unsigned int v)
{
	return __sync_lock_test_and_set(p, v);
}
#endif

#ifndef atomic_swap_ulong
static inline unsigned long
atomic_swap_ulong(volatile unsigned long *p, unsigned long v)
{
	return __sync_lock_test_and_set(p, v);
}
#endif

#ifndef atomic_swap_ptr
static inline void *
atomic_swap_ptr(volatile void *pp, void *v)
{
	void * volatile *p = (void * volatile *)pp;
	return __sync_lock_test_and_set(p, v);
}
#endif

/*
 * atomic_add_*_nv - add and fetch
 */

#ifndef atomic_add_int_nv
static inline unsigned int
atomic_add_int_nv(volatile unsigned int *p, unsigned int v)
{
	return __sync_add_and_fetch(p, v);
}
#endif

#ifndef atomic_add_long_nv
static inline unsigned long
atomic_add_long_nv(volatile unsigned long *p, unsigned long v)
{
	return __sync_add_and_fetch(p, v);
}
#endif

/*
 * atomic_add - add
 */

#ifndef atomic_add_int
#define atomic_add_int(_p, _v) ((void)atomic_add_int_nv((_p), (_v)))
#endif

#ifndef atomic_add_long
#define atomic_add_long(_p, _v) ((void)atomic_add_long_nv((_p), (_v)))
#endif

/*
 * atomic_inc_*_nv - increment and fetch
 */

#ifndef atomic_inc_int_nv
#define atomic_inc_int_nv(_p) atomic_add_int_nv((_p), 1)
#endif

#ifndef atomic_inc_long_nv
#define atomic_inc_long_nv(_p) atomic_add_long_nv((_p), 1)
#endif

/*
 * atomic_inc_* - increment
 */

#ifndef atomic_inc_int
#define atomic_inc_int(_p) ((void)atomic_inc_int_nv(_p))
#endif

#ifndef atomic_inc_long
#define atomic_inc_long(_p) ((void)atomic_inc_long_nv(_p))
#endif

/*
 * atomic_sub_*_nv - sub and fetch
 */

#ifndef atomic_sub_int_nv
static inline unsigned int
atomic_sub_int_nv(volatile unsigned int *p, unsigned int v)
{
	return __sync_sub_and_fetch(p, v);
}
#endif

#ifndef atomic_sub_long_nv
static inline unsigned long
atomic_sub_long_nv(volatile unsigned long *p, unsigned long v)
{
	return __sync_sub_and_fetch(p, v);
}
#endif

/*
 * atomic_sub_* - sub
 */

#ifndef atomic_sub_int
#define atomic_sub_int(_p, _v) ((void)atomic_sub_int_nv((_p), (_v)))
#endif

#ifndef atomic_sub_long
#define atomic_sub_long(_p, _v) ((void)atomic_sub_long_nv((_p), (_v)))
#endif

/*
 * atomic_dec_*_nv - decrement and fetch
 */

#ifndef atomic_dec_int_nv
#define atomic_dec_int_nv(_p) atomic_sub_int_nv((_p), 1)
#endif

#ifndef atomic_dec_long_nv
#define atomic_dec_long_nv(_p) atomic_sub_long_nv((_p), 1)
#endif

/*
 * atomic_dec_* - decrement
 */

#ifndef atomic_dec_int
#define atomic_dec_int(_p) ((void)atomic_dec_int_nv(_p))
#endif

#ifndef atomic_dec_long
#define atomic_dec_long(_p) ((void)atomic_dec_long_nv(_p))
#endif

#ifdef _KERNEL
/*
 * atomic_load_* - read from memory
 */

static inline unsigned int
atomic_load_int(volatile unsigned int *p)
{
	return *p;
}

static inline unsigned long
atomic_load_long(volatile unsigned long *p)
{
	return *p;
}

/*
 * atomic_store_* - write to memory
 */

static inline void
atomic_store_int(volatile unsigned int *p, unsigned int v)
{
	*p = v;
}

static inline void
atomic_store_long(volatile unsigned long *p, unsigned long v)
{
	*p = v;
}
#endif /* _KERNEL */

/*
 * memory barriers
 */

#ifndef membar_enter
#define membar_enter() __sync_synchronize()
#endif

#ifndef membar_exit
#define membar_exit() __sync_synchronize()
#endif

#ifndef membar_producer
#define membar_producer() __sync_synchronize()
#endif

#ifndef membar_consumer
#define membar_consumer() __sync_synchronize()
#endif

#ifndef membar_sync
#define membar_sync() __sync_synchronize()
#endif

#ifndef membar_enter_after_atomic
#define membar_enter_after_atomic() membar_enter()
#endif

#ifndef membar_exit_before_atomic
#define membar_exit_before_atomic() membar_exit()
#endif

#ifdef _KERNEL

/*
 * Force any preceding reads to happen before any subsequent reads that
 * depend on the value returned by the preceding reads.
 */
static inline void
membar_datadep_consumer(void)
{
#ifdef __alpha__
	membar_consumer();
#endif
}

#define READ_ONCE(x) ({							\
	typeof(x) __tmp = *(volatile typeof(x) *)&(x);			\
	membar_datadep_consumer();					\
	__tmp;								\
})

#define WRITE_ONCE(x, val) ({						\
	typeof(x) __tmp = (val);					\
	*(volatile typeof(x) *)&(x) = __tmp;				\
	__tmp;								\
})

#endif /* _KERNEL */

#endif /* _SYS_ATOMIC_H_ */