> On Aug 9, 2023, at 12:11 PM, Christoph Hellwig <hch@xxxxxx> wrote:
>
> On Wed, Aug 09, 2023 at 09:51:18AM -0600, Jens Axboe wrote:
>>> Bisected to commit 615939a2ae73 ("blk-mq: defer to the normal
>>> submission path for post-flush requests"). I've tested reverting
>>> that commit on several kernels from 0aa69d53ac7c ("Merge tag
>>> 'for-6.5/io_uring-2023-06-23' of git://git.kernel.dk/linux") to
>>> v6.5-rc5, and it seems to address the problem.
>>>
>>> I also confirmed that commit 9f87fc4d72f5 ("block: queue data
>>> commands from the flush state machine at the head") is still
>>> necessary.
>>
>> Adding the author - Christoph?
>
> Hmm, weird. I though we had all this sorted out a few weeks
> ago when that 9f87fc4d72f5 fixed it for you? What is different
> in this round of testing?
I don't know the answer to that... I suspect that I originally
tested the first revert on a kernel that also did not have 615939
applied. At the time we thought only one of those commits was
broken.
I've bisected this every way I can think of, and both fixes are
needed to fully stop the hangs.
--
Chuck Lever
