Re: ublk: consequences of a dead server are only failing on the last /dev/ublkcN release

[Hannes Reinecke <hare@suse.de>]

On 9/20/26 3:39 PM, Ming Lei wrote:
> Hello,
> 
> On Fri, Sep 18, 2026 at 12:29:22PM +0200, Christian Brauner wrote:
>> Hi,
>>
[ .. ]
>> The commit 7fc4da6a304b ("ublk: scan partition in async way") made
>> partition scans run from a workqueue that holds disk->open_mutex while
>> the server serves its reads and START_DEV returns before it's finished.
>>
>> Any server that treats START_DEV's return as a signal that the disk is
>> ready and doesn't serve requests hangs every open() of /dev/ublkbN on
>> uninterruptible on open_mutex. You can combine that with (1) and (2).
> 
> START_DEV doesn't promise partitions scanning done when it is returned.
> 
> nvme multipath has similar usage too.
> 
Indeed, we hit the same issue with nvme. We got it fixed by commit
1f021341eef41e77a633186e9be5223de2ce5d48 ('nvme-multipath: defer 
partition scanning'), essentially suppress partition scan during
probing, and move the partition scan to a work thread. That way
probing can continue (even if partition scan is blocked), and
a failed probe can correctly abort partition scanning.

Cheers,

Hannes
-- 
Dr. Hannes Reinecke                  Kernel Storage Architect
hare@suse.de                                +49 911 74053 688
SUSE Software Solutions GmbH, Frankenstr. 146, 90461 Nürnberg
HRB 36809 (AG Nürnberg), GF: I. Totev, A. McDonald, W. Knoblich