Re: [PATCH] loop: clear wb_err in bd_inode when detaching backing file

[Jens Axboe <axboe@fb.com>]

On 5/21/18 12:26 PM, Jeff Layton wrote:
> On Mon, 2018-05-21 at 08:37 -0400, Jeff Layton wrote:
>> From: Jeff Layton <jlayton@redhat.com>
>>
>> When a loop block device encounters a writeback error, that error will
>> get propagated to the bd_inode's wb_err field. If we then detach the
>> backing file from it, attach another and fsync it, we'll get back the
>> writeback error that we had from the previous backing file.
>>
>> This is a bit of a grey area as POSIX doesn't cover loop devices, but it
>> is somewhat counterintuitive.
>>
>> If we detach a backing file from the loopdev while there are still
>> unreported errors, take it as a sign that we're no longer interested in
>> the previous file, and clear out the wb_err in the loop blockdev.
>>
>> Signed-off-by: Jeff Layton <jlayton@redhat.com>
>> ---
>>  drivers/block/loop.c | 1 +
>>  1 file changed, 1 insertion(+)
>>
>> diff --git a/drivers/block/loop.c b/drivers/block/loop.c
>> index 5d4e31655d96..55cf554bc914 100644
>> --- a/drivers/block/loop.c
>> +++ b/drivers/block/loop.c
>> @@ -1068,6 +1068,7 @@ static int loop_clr_fd(struct loop_device *lo)
>>  	if (bdev) {
>>  		bdput(bdev);
>>  		invalidate_bdev(bdev);
>> +		bdev->bd_inode->i_mapping->wb_err = 0;
>>  	}
>>  	set_capacity(lo->lo_disk, 0);
>>  	loop_sysfs_exit(lo);
> 
> Jens,
> 
> I sent this to the mailing lists earlier, but it occurs to me that
> loop.c changes should probably go through the block tree. Would you
> mind picking this up and feeding it to next and on to Linus?
> 
> I think we probably want this in v4.17 if possible, as it is a
> regression from v4.16 (due to commit b4678df184b3) and was causing some
> failures in xfstests runs.
> 
> Hmm, now that I think about it, we should probably also give Ted
> "Reported-and-Tested-by" credit here too. Let me know if you'd rather I
> resend with that added.

Yeah, I can funnel it for 4.17 - a quick resend with the right
attributions would be appreciated, thanks.

-- 
Jens Axboe