On 5/21/21 11:06 AM, Darrick J. Wong wrote: > On Fri, May 21, 2021 at 11:03:11AM +0200, lukas@xxxxxxxxxxx wrote: >>> Are you going to wire up fs uuid logging for the other filesystems that >>> support them? >> Well, I wasn't planning to but I can take a look on other FS as well >> Ext4 and Btrfs for start. >> >>> What happens w.r.t. uuid disambiguation if someone uses a nouuid mount >>> to mount a filesystem with the same uuid as an already-mounted xfs? >> >> I a not sure I understand the "nouuid mount". I don't think there can >> be XFS with empty uuid value in SB. And printing the message is independent >> on the mount method (mount UUID="" ...; mount /dev/sdX ...;). > > I meant specifically: > > mount /dev/mapper/fubar /mnt > <snapshot fubar to fubar.bak> > > Oh no, I deleted something in fubar, let's retrieve it from fubar.bak! > > mount /dev/mapper/fubar.bak /opt # fails because same uuid as fubar > mount /dev/mapper/fubar.bak /opt -o nouuid While it is possible to end up with 2 mounted devices with the same UUID, that's still useful information I think; to me, the -o nouuid case doesn't really argue against this patch. In fact, it could be /very/ useful information if someone force-mounted 2 paths to the same underlying device by using -o nouuid. -Eric
