[PATCH v2 for-5.8-rc 1/6] nvme: fix possible deadlock when I/O is blocked

[Christoph Hellwig]

On Tue, Jun 23, 2020 at 11:54:31PM -0700, Sagi Grimberg wrote:
>
>>> Revert fab7772bfbcf ("nvme-multipath: revalidate nvme_ns_head gendisk
>>> in nvme_validate_ns")
>>>
>>> When adding a new namespace to the head disk (via nvme_mpath_set_live)
>>> we will see partition scan which triggers I/O on the mpath device node.
>>> This process will usually be triggered from the scan_work which holds
>>> the scan_lock. If I/O blocks (if we got ana change currently have only
>>> available paths but none are accessible) this can deadlock on the head
>>> disk bd_mutex as both partition scan I/O takes it, and head disk revalidation
>>> takes it to check for resize (also triggered from scan_work on a different
>>> path). See trace [1].
>>>
>>> This is no longer needed since commit cb224c3af4df ("nvme: Convert to
>>> use set_capacity_revalidate_and_notify") which already updates resize
>>> info without unnecessarily revalidating the disk.
>>
>> Did you look if any other revalidation gets skipped this way? E.g.
>> LBA size change?
>>
>>  From purely the size POV this looks good, though.
>
> The only reason the revalidate was added was to update on a size change,
> which is redundant.
>
> And if you look at revalidate, it only calls fops->revalidate_disk
> (which is a noop in the mpath device) and checks the size change.

Please add this blurb to the changelog for the next version.