freebsd-dev/lib/libthr/thread/thr_umtx.h
David Xu 17ce606321 umtx operation UMTX_OP_MUTEX_WAKE has a side-effect that it accesses
a mutex after a thread has unlocked it, it event writes data to the mutex
memory to clear contention bit, there is a race that other threads
can lock it and unlock it, then destroy it, so it should not write
data to the mutex memory if there isn't any waiter.
The new operation UMTX_OP_MUTEX_WAKE2 try to fix the problem. It
requires thread library to clear the lock word entirely, then
call the WAKE2 operation to check if there is any waiter in kernel,
and try to wake up a thread, if necessary, the contention bit is set again
by the operation. This also mitgates the chance that other threads find
the contention bit and try to enter kernel to compete with each other
to wake up sleeping thread, this is unnecessary. With this change, the
mutex owner is no longer holding the mutex until it reaches a point
where kernel umtx queue is locked, it releases the mutex as soon as
possible.
Performance is improved when the mutex is contensted heavily.  On Intel
i3-2310M, the runtime of a benchmark program is reduced from 26.87 seconds
to 2.39 seconds, it even is better than UMTX_OP_MUTEX_WAKE which is
deprecated now. http://people.freebsd.org/~davidxu/bench/mutex_perf.c
2012-04-05 02:24:08 +00:00

225 lines
7.0 KiB
C

/*-
* Copyright (c) 2005 David Xu <davidxu@freebsd.org>
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* $FreeBSD$
*/
#ifndef _THR_FBSD_UMTX_H_
#define _THR_FBSD_UMTX_H_
#include <strings.h>
#include <sys/umtx.h>
#define DEFAULT_UMUTEX {0,0,{0,0},{0,0,0,0}}
#define DEFAULT_URWLOCK {0,0,0,0,{0,0,0,0}}
int _umtx_op_err(void *, int op, u_long, void *, void *) __hidden;
int __thr_umutex_lock(struct umutex *mtx, uint32_t id) __hidden;
int __thr_umutex_lock_spin(struct umutex *mtx, uint32_t id) __hidden;
int __thr_umutex_timedlock(struct umutex *mtx, uint32_t id,
const struct timespec *timeout) __hidden;
int __thr_umutex_unlock(struct umutex *mtx, uint32_t id) __hidden;
int __thr_umutex_trylock(struct umutex *mtx) __hidden;
int __thr_umutex_set_ceiling(struct umutex *mtx, uint32_t ceiling,
uint32_t *oldceiling) __hidden;
void _thr_umutex_init(struct umutex *mtx) __hidden;
void _thr_urwlock_init(struct urwlock *rwl) __hidden;
int _thr_umtx_wait(volatile long *mtx, long exp,
const struct timespec *timeout) __hidden;
int _thr_umtx_wait_uint(volatile u_int *mtx, u_int exp,
const struct timespec *timeout, int shared) __hidden;
int _thr_umtx_timedwait_uint(volatile u_int *mtx, u_int exp, int clockid,
const struct timespec *timeout, int shared) __hidden;
int _thr_umtx_wake(volatile void *mtx, int count, int shared) __hidden;
int _thr_ucond_wait(struct ucond *cv, struct umutex *m,
const struct timespec *timeout, int check_unpaking) __hidden;
void _thr_ucond_init(struct ucond *cv) __hidden;
int _thr_ucond_signal(struct ucond *cv) __hidden;
int _thr_ucond_broadcast(struct ucond *cv) __hidden;
int __thr_rwlock_rdlock(struct urwlock *rwlock, int flags,
const struct timespec *tsp) __hidden;
int __thr_rwlock_wrlock(struct urwlock *rwlock,
const struct timespec *tsp) __hidden;
int __thr_rwlock_unlock(struct urwlock *rwlock) __hidden;
/* Internal used only */
void _thr_rwl_rdlock(struct urwlock *rwlock) __hidden;
void _thr_rwl_wrlock(struct urwlock *rwlock) __hidden;
void _thr_rwl_unlock(struct urwlock *rwlock) __hidden;
static inline int
_thr_umutex_trylock(struct umutex *mtx, uint32_t id)
{
if (atomic_cmpset_acq_32(&mtx->m_owner, UMUTEX_UNOWNED, id))
return (0);
if ((mtx->m_flags & UMUTEX_PRIO_PROTECT) == 0)
return (EBUSY);
return (__thr_umutex_trylock(mtx));
}
static inline int
_thr_umutex_trylock2(struct umutex *mtx, uint32_t id)
{
if (atomic_cmpset_acq_32(&mtx->m_owner, UMUTEX_UNOWNED, id) != 0)
return (0);
if ((uint32_t)mtx->m_owner == UMUTEX_CONTESTED &&
__predict_true((mtx->m_flags & (UMUTEX_PRIO_PROTECT | UMUTEX_PRIO_INHERIT)) == 0))
if (atomic_cmpset_acq_32(&mtx->m_owner, UMUTEX_CONTESTED, id | UMUTEX_CONTESTED))
return (0);
return (EBUSY);
}
static inline int
_thr_umutex_lock(struct umutex *mtx, uint32_t id)
{
if (_thr_umutex_trylock2(mtx, id) == 0)
return (0);
return (__thr_umutex_lock(mtx, id));
}
static inline int
_thr_umutex_lock_spin(struct umutex *mtx, uint32_t id)
{
if (_thr_umutex_trylock2(mtx, id) == 0)
return (0);
return (__thr_umutex_lock_spin(mtx, id));
}
static inline int
_thr_umutex_timedlock(struct umutex *mtx, uint32_t id,
const struct timespec *timeout)
{
if (_thr_umutex_trylock2(mtx, id) == 0)
return (0);
return (__thr_umutex_timedlock(mtx, id, timeout));
}
static inline int
_thr_umutex_unlock(struct umutex *mtx, uint32_t id)
{
uint32_t flags = mtx->m_flags;
if ((flags & (UMUTEX_PRIO_PROTECT | UMUTEX_PRIO_INHERIT)) == 0) {
uint32_t owner;
do {
owner = mtx->m_owner;
if (__predict_false((owner & ~UMUTEX_CONTESTED) != id))
return (EPERM);
} while (__predict_false(!atomic_cmpset_rel_32(&mtx->m_owner,
owner, UMUTEX_UNOWNED)));
if ((owner & UMUTEX_CONTESTED))
(void)_umtx_op_err(mtx, UMTX_OP_MUTEX_WAKE2, flags, 0, 0);
return (0);
}
if (atomic_cmpset_rel_32(&mtx->m_owner, id, UMUTEX_UNOWNED))
return (0);
return (__thr_umutex_unlock(mtx, id));
}
static inline int
_thr_rwlock_tryrdlock(struct urwlock *rwlock, int flags)
{
int32_t state;
int32_t wrflags;
if (flags & URWLOCK_PREFER_READER || rwlock->rw_flags & URWLOCK_PREFER_READER)
wrflags = URWLOCK_WRITE_OWNER;
else
wrflags = URWLOCK_WRITE_OWNER | URWLOCK_WRITE_WAITERS;
state = rwlock->rw_state;
while (!(state & wrflags)) {
if (__predict_false(URWLOCK_READER_COUNT(state) == URWLOCK_MAX_READERS))
return (EAGAIN);
if (atomic_cmpset_acq_32(&rwlock->rw_state, state, state + 1))
return (0);
state = rwlock->rw_state;
}
return (EBUSY);
}
static inline int
_thr_rwlock_trywrlock(struct urwlock *rwlock)
{
int32_t state;
state = rwlock->rw_state;
while (!(state & URWLOCK_WRITE_OWNER) && URWLOCK_READER_COUNT(state) == 0) {
if (atomic_cmpset_acq_32(&rwlock->rw_state, state, state | URWLOCK_WRITE_OWNER))
return (0);
state = rwlock->rw_state;
}
return (EBUSY);
}
static inline int
_thr_rwlock_rdlock(struct urwlock *rwlock, int flags, struct timespec *tsp)
{
if (_thr_rwlock_tryrdlock(rwlock, flags) == 0)
return (0);
return (__thr_rwlock_rdlock(rwlock, flags, tsp));
}
static inline int
_thr_rwlock_wrlock(struct urwlock *rwlock, struct timespec *tsp)
{
if (_thr_rwlock_trywrlock(rwlock) == 0)
return (0);
return (__thr_rwlock_wrlock(rwlock, tsp));
}
static inline int
_thr_rwlock_unlock(struct urwlock *rwlock)
{
int32_t state;
state = rwlock->rw_state;
if (state & URWLOCK_WRITE_OWNER) {
if (atomic_cmpset_rel_32(&rwlock->rw_state, URWLOCK_WRITE_OWNER, 0))
return (0);
} else {
for (;;) {
if (__predict_false(URWLOCK_READER_COUNT(state) == 0))
return (EPERM);
if (!((state & (URWLOCK_WRITE_WAITERS |
URWLOCK_READ_WAITERS)) &&
URWLOCK_READER_COUNT(state) == 1)) {
if (atomic_cmpset_rel_32(&rwlock->rw_state,
state, state-1))
return (0);
state = rwlock->rw_state;
} else {
break;
}
}
}
return (__thr_rwlock_unlock(rwlock));
}
#endif