On Wed, 2017-08-09 at 11:32 -0700, Bart Van Assche wrote:
> Detected by sparse.
> 
> Fixes: commit 4e4cbee93d56 ("block: switch bios to blk_status_t")
> Signed-off-by: Bart Van Assche <bart.vanass...@wdc.com>
> Cc: Christoph Hellwig <h...@lst.de>
> Cc: Laurence Oberman <lober...@redhat.com>
> ---
>  drivers/md/dm.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/md/dm.c b/drivers/md/dm.c
> index 2edbcc2d7d3f..8298670757e9 100644
> --- a/drivers/md/dm.c
> +++ b/drivers/md/dm.c
> @@ -1523,7 +1523,7 @@ static void __split_and_process_bio(struct
> mapped_device *md,
>       }
>  
>       /* drop the extra reference count */
> -     dec_pending(ci.io, error);
> +     dec_pending(ci.io, errno_to_blk_status(error));
>  }
>  /*-----------------------------------------------------------------
>   * CRUD END

Hello Bart

I applied and tested all patches except patch 2/7 (as discussed)
I also still have the prior patch in place you sent me.

I built a new kernel on 4.13-rc3

The soft lockups are gone for me with 10 parallel writes.
I also have not seen a hard lockup so far.

What is important to note is that if I use the [none] scheduler I am
still stable but I will sporadically get the      ******
"[38071.773471] device-mapper: multipath: blk_get_request() returned
-11 - requeuing
"

However that is not unexpected I guess.

The ( requeuing messages) do NOT show up with [mq-deadline] which is
all I was testing with last night and yesterday. So when I said they
are gone, that was specifically for [mq-deadline]

Patches look good and are helping. I will circle back and patch and
test on Ming's kernel.

For the series except PATCH2/7 and including

[PATCH] block: Make blk_mq_delay_kick_requeue_list() rerun the queue at
a quiet time


Tested-by: Laurence Oberman <lober...@redhat.com>

Thanks
Laurence




--
dm-devel mailing list
dm-devel@redhat.com
https://www.redhat.com/mailman/listinfo/dm-devel

Reply via email to