[dm-devel] [PATCH RFC] dm: Fix a bio leak in dec_pending()

Bart Van Assche bart.vanassche at sandisk.com
Fri Aug 26 15:34:49 UTC 2016


On 08/26/2016 07:57 AM, Mike Snitzer wrote:
> On Thu, Aug 25 2016 at  4:57pm -0400,
> Bart Van Assche <bart.vanassche at sandisk.com> wrote:
>
>> Ensure that bio_endio() is called if io->error == DM_ENDIO_REQUEUE and
>> __noflush_suspending(md) returns false. Posting this as an RFC since I'm
>> not really familiar with the dm code.
>>
>> Fixes: commit 2e93ccc1933d ("dm: suspend: add noflush pushback")
>> Signed-off-by: Bart Van Assche <bart.vanassche at sandisk.com>
>> Cc: Kiyoshi Ueda <k-ueda at ct.jp.nec.com>
>
> I did reinstate bio-based multipath recently.  But I just want to make
> sure: you realize that dec_pending() is only ever used by bio-based DM
> right? (dm-mq and .request_fn multipath are request-based)

Hello Mike,

Yes, I am aware that this code is not triggered by my tests. This is 
something I noticed while reading the dm source code and something I 
wanted to report.

Bart.




More information about the dm-devel mailing list