[Fwd: FIO bug]

-------- Forwarded Message --------
> From: Zhang, Yanmin <yanmin_zhang_at_linux.intel.com>
> To: Jens Axboe <jens.axboe_at_oracle.com>
> Subject: FIO bug
> Date: Sun, 03 Feb 2008 16:49:26 +0800
>
> Jens,
>
> I tried again to integrate FIO into my automation testing system.
> Last year, a coulpe of interns fixed a couple of bugs. But I found new
> bug now.
>
> With the attached job file, fio hangs. Basically the sem->cond isn't initiated
> correctly.
>
> Below patch fixes it.
>
> Yanmin
>
> ---
>
> --- fio-1.17.3/mutex.c 2007-03-09 19:39:14.000000000 +0800
> +++ fio-1.17.3_work/mutex.c 2008-02-03 09:49:51.000000000 +0800
> @@ -19,6 +19,8 @@ struct fio_sem *fio_sem_init(int value)
> char sem_name[] = "/tmp/.fio_sem.XXXXXX";
> struct fio_sem *sem = NULL;
> pthread_mutexattr_t attr;
> + pthread_condattr_t cond;
> +
> int fd;
>
> fd = mkstemp(sem_name);
> @@ -53,6 +55,11 @@ struct fio_sem *fio_sem_init(int value)
> perror("pthread_mutexattr_setpshared");
> goto err;
> }
> +
> + pthread_condattr_init (&cond);
> + pthread_condattr_setpshared (&cond, PTHREAD_PROCESS_SHARED);
> + pthread_cond_init (&sem->cond, &cond);
> +
> if (pthread_mutex_init(&sem->lock, &attr)) {
> perror("pthread_mutex_init");
> goto err;
>
>
Received on Mon Feb 04 2008 - 08:16:16 CET