Re: [PATCH 5.4 103/142] Revert "block: end bio with BLK_STS_AGAIN in case of non-mq devs and REQ_NOWAIT"

[Jens Axboe]

On 2/3/21 4:59 PM, Andres Freund wrote:
> Hi,
> 
> On 2021-02-03 15:58:33 -0700, Jens Axboe wrote:
>> On 2/3/21 3:06 PM, Greg Kroah-Hartman wrote:
>>> On Wed, Feb 03, 2021 at 01:28:26PM -0800, Andres Freund wrote:
>>>> On 2021-02-03 14:03:09 +0100, Greg Kroah-Hartman wrote:
>>>>>> On v5.4.43-101-gbba91cdba612 this fails with
>>>>>> fio: io_u error on file /mnt/t2/test.0.0: Input/output error: write offset=0, buflen=4096
>>>>>> fio: pid=734, err=5/file:io_u.c:1834, func=io_u error, error=Input/output error
>>>>>>
>>>>>> whereas previously it worked. libaio still works...
>>>>>>
>>>>>> I haven't checked which major kernel version fixed this again, but I did
>>>>>> verify that it's still broken in 5.4.94 and that 5.10.9 works.
>>>>>>
>>>>>> I would suspect it's
>>>>>>
>>>>>> commit 4503b7676a2e0abe69c2f2c0d8b03aec53f2f048
>>>>>> Author: Jens Axboe <axboe@xxxxxxxxx>
>>>>>> Date:   2020-06-01 10:00:27 -0600
>>>>>>
>>>>>>     io_uring: catch -EIO from buffered issue request failure
>>>>>>
>>>>>>     -EIO bubbles up like -EAGAIN if we fail to allocate a request at the
>>>>>>     lower level. Play it safe and treat it like -EAGAIN in terms of sync
>>>>>>     retry, to avoid passing back an errant -EIO.
>>>>>>
>>>>>>     Catch some of these early for block based file, as non-mq devices
>>>>>>     generally do not support NOWAIT. That saves us some overhead by
>>>>>>     not first trying, then retrying from async context. We can go straight
>>>>>>     to async punt instead.
>>>>>>
>>>>>>     Signed-off-by: Jens Axboe <axboe@xxxxxxxxx>
>>>>>>
>>>>>>
>>>>>> which isn't in stable/linux-5.4.y
>>>>>
>>>>> Can you test that if the above commit is added, all works well again?
>>>>
>>>> It doesn't apply cleanly, I'll try to resolve the conflict. However, I
>>>> assume that the revert was for a concrete reason - but I can't quite
>>>> figure out what b0beb28097fa04177b3769f4bb7a0d0d9c4ae76e was concretely
>>>> solving, and whether reverting the revert in 5.4 would re-introduce a
>>>> different problem.
>>>>
>>>> commit b0beb28097fa04177b3769f4bb7a0d0d9c4ae76e (tag: block-5.7-2020-05-29, linux-block/block-5.7)
>>>> Author: Jens Axboe <axboe@xxxxxxxxx>
>>>> Date:   2020-05-28 13:19:29 -0600
>>>>
>>>>     Revert "block: end bio with BLK_STS_AGAIN in case of non-mq devs and REQ_NOWAIT"
>>>>
>>>>     This reverts commit c58c1f83436b501d45d4050fd1296d71a9760bcb.
>>>>
>>>>     io_uring does do the right thing for this case, and we're still returning
>>>>     -EAGAIN to userspace for the cases we don't support. Revert this change
>>>>     to avoid doing endless spins of resubmits.
>>>>
>>>>     Cc: stable@xxxxxxxxxxxxxxx # v5.6
>>>>     Reported-by: Bijan Mottahedeh <bijan.mottahedeh@xxxxxxxxxx>
>>>>     Signed-off-by: Jens Axboe <axboe@xxxxxxxxx>
>>>>
>>>> I suspect it just wasn't aimed at 5.4, and that's that, but I'm not
>>>> sure. In which case presumably reverting
>>>> bba91cdba612fbce4f8575c5d94d2b146fb83ea3 would be the right fix, not
>>>> backporting 4503b7676a2e0abe69c2f2c0d8b03aec53f2f048 et al.
> 
> Having looked a bit more through the history, I suspect that the reason
> 5.6 doesn't need c58c1f83436b501d45d4050fd1296d71a9760bcb - which I have
> confirmed - is that ext4 was converted to the iomap infrastructure in
> 5.5, but not in 5.4.
> 
> I've confirmed that the repro I shared upthread triggers in
> 378f32bab3714f04c4e0c3aee4129f6703805550^ but not in
> 378f32bab3714f04c4e0c3aee4129f6703805550.

I checked up on this, and I do see the issue as well. As far as
io_uring is concerned, we don't need that revert in 5.4. So I think
the right solution here would be to... revert the revert :-)

-- 
Jens Axboe