Re: INFO: task blocked for more than 120 seconds

["Michael Madore" <michael.madore@gmail.com>]

>>> Committing a transaction would means writing rest of the meta-data in
>>> the transaction. And that would imply forcing most of the buffer_heads
>>> to disk in ordered mode. This can result a lot of seeks and make take
>>> more thatn 120 seconds.
>>
>>
>> Both Randy and Greg reported getting this for 2.6.27-rc but not
>> for 2.6.26.
>>
>> Why are people getting such messages for 2.6.27-rc but not for 2.6.26?
>
> Hi,
>
> I have reported getting these messages on 2.6.26:
>
> http://marc.info/?l=linux-kernel&m=121796211813099&w=2
>
> In addition to the system mentioned in that posting, I have just
> reproduced it by stress testing a system with 2 Opteron processors,
> 2GB of RAM and 2 SATA disks.
>
> Backing out this patch seems to help:
>
> commit cc19747977824ece6aa1c56a29e974fef5ec2b32
> Author: Jens Axboe <jens.axboe@oracle.com>
> Date:   Fri Apr 20 20:45:39 2007 +0200
>
>    cfq-iosched: tighten queue request overlap condition
>
>    For tagged devices, allow overlap of requests if the idle window
>    isn't enabled on the current active queue.
>
>    Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
>
> diff --git a/block/cfq-iosched.c b/block/cfq-iosched.c
> index a8237be..e859b49 100644
> --- a/block/cfq-iosched.c
> +++ b/block/cfq-iosched.c
> @@ -989,7 +989,8 @@ static struct cfq_queue *cfq_select_queue(struct
> cfq_data *cfqd)
> * flight or is idling for a new request, allow either of these
> * conditions to happen (or time out) before selecting a new queue.
> */
> - if (cfqq->dispatched || timer_pending(&cfqd->idle_slice_timer)) {
> + if (timer_pending(&cfqd->idle_slice_timer) ||
> +     (cfqq->dispatched && cfq_cfqq_idle_window(cfqq))) {
> cfqq = NULL;
> goto keep_queue;
> }

Actually, I tested again with this commit reversed, and I'm still
seeing the problem.  Sorry for any confusion.

Mike