Re: [Patch] shm bug introduced with pagecache in 2.3.11

Manfred Spraul (manfreds@colorfullife.com)
Sat, 20 Nov 1999 11:14:54 +0100

Alexander Viro wrote:
>
> IIRC, there was an implementation and analysis in Vahalia. Streamlining it
> shouldn't be too hard (I have something like that in AFFS patches, but
> there I didn't need {up,down}_write and did need upgrade/downgrade. IOW,
> writrs always started as readers).

I never figured out how to upgrade a reader to a writer:
what if 2 threads acquire the lock 'read', and and both decide to
promote to "write"?
Could you send me the patch?

I have attached below a pseudo-code version of a rw-mutex, what do you
think?
acquire_{shared,exclusive} should be possible with 2 asm instructions
inline
release_exclusive with 3 asm instructions inline

--
	Manfred
>>>>>>>>>>>>>>


struct rw_mutex
{
        rwlock_t rwl;
        wait_queue_head_t wait;
        int wake_me_up;
};


/* TODO: add asm, add TASK_EXCLUSIVE for acquire_exclusive,
 *       check if a waiting writer should stop new readers.
 *	 optimize release_shared
 */


void rw_mutex_init(struct rw_mutex* rwm)
{
        init_wait_queue_head(&rwm->wait);
        rwm->wake_me_up = 0;
        rwm->rwl = RW_LOCK_UNLOCKED;
}


/* 2 asm instructions inline */
void acquire_exclusive(struct rw_mutex* rwm)
{
        if(!write_lock_trylock(&rwm->rwl)) {
                DECLARE_WAIT_QUEUE(wait,current);
                add_wait_queue(wait,&rwm->wait);
                for(;;) {
                        current->state = TASK_UNINTERRUPTIBLE;
                        rwm->wake_me_up = 1;
                        if(write_lock_trylock(&rwm->rwl)) {
                                break;
                        }
                        schedule();
                }
                current->state = TASK_RUNNING;
                remove_wait_queue(wait,&rwm->wait);
        }
}


/* 2 asm instruction inline */
void acquire_shared(struct rw_mutex* rwm)
{
        if(!read_lock_trylock(&rwm->rwl)) {
                DECLARE_WAIT_QUEUE(wait,current);
                add_wait_queue(wait,&rwm->wait);
                for(;;) {
                        current->state = TASK_UNINTERRUPTIBLE;
                        rwm->wake_me_up = 1;
                        if(read_lock_trylock(&rwm->rwl)) {
                                break;
                        }
                        schedule();
                }
                current->state = TASK_RUNNING;
                remove_wait_queue(wait,&rwm->wait);
        }
}


/* 3 asm instructions */
void release_exclusive(struct rw_mutex* rwm)
{
        read_unlock(&rwm->rwl);
        if(rwm->wake_me_up) {
                rwm->wake_me_up = 0;
                wake_up(&rwm->wait);
        }
}
void release_shared(struct rw_mutex* rwm)
{
        read_unlock(&rwm->rwl);
        if(rwm->wake_me_up) {
 		/* FIXME: is this a race? */
          	if(rwm->rwl != 0)
			return;
	        rwm->wake_me_up = 0;
                wake_up(&rwm->wait);
        }
}
>>>>>>>>>>>>>>>>>




-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/