Re: [PATCH UPDATED] block: In blk_execute_rq_nowait, init rq->end_iobefore checking for dead queue.

From: Muthu Kumar
Date: Wed Jun 20 2012 - 20:53:33 EST


On Wed, Jun 20, 2012 at 11:53 AM, Bart Van Assche <bvanassche@xxxxxxx> wrote:
> On 06/18/12 22:42, Tejun Heo wrote:
>
>> On Mon, Jun 11, 2012 at 02:23:23PM -0700, Muthu Kumar wrote:
>>> On Mon, Jun 11, 2012 at 10:33 AM, Muthu Kumar <muthu.lkml@xxxxxxxxx> wrote:
>>>> On Sun, Jun 10, 2012 at 10:40 AM, Bart Van Assche <bvanassche@xxxxxxx> wrote:
>>>>> On 06/09/12 23:57, Muthu Kumar wrote:
>>>>>
>>>>>> Locking change is the one you posted already (the link above). Anyway,
>>>>>> I have the attached patch *including* the locking change. Original
>>>>>> mail has attachment without locking change. Please use whatever you
>>>>>> need.
>>>>>
>>>>> While we are at it ... the rq->rq_disk and rq->end_io assignments can be
>>>>> performed safely before the spinlock is taken, isn't it ?
>>>>
>>>> Yes.. that spinlock is to protect the q.
>>>
>>> Attached patch with assignment performed before taking the spinlock.
>>
>> This looks correct to me.  Bart, can you please include this patch in
>> your series and repost the series?
>
>
> I'll start testing the patch below:
>
> [PATCH] block: Fix blk_execute_rq_nowait() dead queue handling
>
> Make sure that blk_execute_rq_nowait() invokes the proper end_io
> function if the queue is dead and also that this call is performed
> with the queue lock held. Found this through source code review.
>
> Notes:
> - Since blk_execute_rq_nowait() must not be invoked from interrupt
>  context it is safe to invoke end_io directly from this function.
> - blk_finish_request() already invokes request.end_io with the
>  queue lock held.
>
> Signed-off-by: Bart Van Assche <bvanassche@xxxxxxx>
> Reported-by: Muthukumar Ratty <muthur@xxxxxxxxx>

Hmm... Shouldn't this be:

Signed-off-by: Muthukumar Ratty <muthur@xxxxxxxxx>
Tested-by: Bart Van Assche <bvanassche@xxxxxxx>

???

> Cc: Jens Axboe <axboe@xxxxxxxxx>
> Cc: Tejun Heo <tj@xxxxxxxxxx>
> ---
>  block/blk-exec.c |   11 +++++++----
>  1 files changed, 7 insertions(+), 4 deletions(-)
>
> diff --git a/block/blk-exec.c b/block/blk-exec.c
> index fb2cbd5..c0fd83a 100644
> --- a/block/blk-exec.c
> +++ b/block/blk-exec.c
> @@ -42,7 +42,8 @@ static void blk_end_sync_rq(struct request *rq, int error)
>  *
>  * Description:
>  *    Insert a fully prepared request at the back of the I/O scheduler queue
> - *    for execution.  Don't wait for completion.
> + *    for execution.  Don't wait for completion.  This function may invoke
> + *    rq->end_io directly.
>  */
>  void blk_execute_rq_nowait(struct request_queue *q, struct gendisk *bd_disk,
>                           struct request *rq, int at_head,
> @@ -51,18 +52,20 @@ void blk_execute_rq_nowait(struct request_queue *q, struct gendisk *bd_disk,
>        int where = at_head ? ELEVATOR_INSERT_FRONT : ELEVATOR_INSERT_BACK;
>
>        WARN_ON(irqs_disabled());
> +
> +       rq->rq_disk = bd_disk;
> +       rq->end_io = done;
> +
>        spin_lock_irq(q->queue_lock);
>
>        if (unlikely(blk_queue_dead(q))) {
> -               spin_unlock_irq(q->queue_lock);
>                rq->errors = -ENXIO;
>                if (rq->end_io)
>                        rq->end_io(rq, rq->errors);
> +               spin_unlock_irq(q->queue_lock);

This part already went in, its here only for testing - So in the final
patch, before applying this should be removed.

>                return;
>        }
>
> -       rq->rq_disk = bd_disk;
> -       rq->end_io = done;
>        __elv_add_request(q, rq, where);
>        __blk_run_queue(q);
>        /* the queue is stopped so it won't be run */
> --
> 1.7.7
>
>
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/