Re: [PATCH -mm 3/3][AIO] - AIO completion signal notification

[Sébastien Dugué <sebastien.dugue@bull.net>]

On Thu, 2006-11-09 at 10:46 -0800, Badari Pulavarty wrote:
> On Thu, 2006-11-09 at 16:59 +0100, Sébastien Dugué wrote:
> 
> > +
> > +static long aio_setup_sigevent(struct kiocb *iocb,
> > +			       struct sigevent __user *user_event)
> > +{
> > +	int error = 0;
> > +	sigevent_t event;
> > +	struct task_struct *target;
> > +	unsigned long flags;
> > +
> > +	if (!access_ok(VERIFY_READ, user_event, sizeof(struct sigevent)))
> > +		return -EFAULT;
> > +
> > +	if (copy_from_user(&event, user_event, sizeof (event)))
> > +		return -EFAULT;
> > +
> > +	/* Check for the SIGEV_NONE case */
> > +	if (event.sigev_notify == SIGEV_NONE)
> > +		return 0;
> > +
> > +	/* Setup the request completion notification parameters */
> > +	iocb->ki_notify.notify = event.sigev_notify;
> > +	iocb->ki_notify.signo = event.sigev_signo;
> > +	iocb->ki_notify.value = event.sigev_value;
> > +
> > +	/* Now get the notification target */
> > +	read_lock(&tasklist_lock);
> > +
> > +	if ((target = good_sigevent(&event))) {
> > +
> > +		spin_lock_irqsave(&target->sighand->siglock, flags);
> > +
> > +		if (!(target->flags & PF_EXITING)) {
> > +			iocb->ki_notify.target = target;
> > +
> > +			spin_unlock_irqrestore(&target->sighand->siglock, flags);
> > +
> > +			/*
> > +			 * Get a ref on the task. It is dropped in really_put_req()
> > +			 * when we're done with the iocb, be it from the normal
> > +			 * completion path, the cancellation path or an error path.
> > +			 */
> > +			if (iocb->ki_notify.notify == (SIGEV_SIGNAL|SIGEV_THREAD_ID))
> > +				get_task_struct(target);
> > +		} else {
> > +			spin_unlock_irqrestore(&target->sighand->siglock, flags);
> > +			error = -EINVAL;
> > +		}
> > +	} else
> > +		error = -EINVAL;
> > +
> > +	read_unlock(&tasklist_lock);
> > +
> > +	if (!error) {
> > +		/*
> > +		 * Alloc a sigqueue for this request
> > +		 *
> > +		 * NOTE: we cannot free the sigqueue in the completion path as
> > +		 * the signal may not have been delivered to the target task.
> > +		 * Therefore it has to be freed in __sigqueue_free() when the
> > +		 * signal is collected if si_code is SI_ASYNCIO.
> > +		 */
> > +		if (unlikely(!(iocb->ki_sigq = sigqueue_alloc())))
> > +			error =  -EAGAIN;
> 
> Don't you have to do put_task_struct() here ?

  Well, put_task_struct() is called in really_put_req() when we're 
freeing the request. This should be OK unless I missed something.

  Sébastien.

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