On Wed, Jul 26, 2023 at 09:13:30PM -0400, Theodore Ts'o wrote: > On Wed, Jul 26, 2023 at 09:28:23AM -0700, Luis Chamberlain wrote: > > > I'm a little confused. Where are these "sanity checks" enforced? > > > I've been using > > > > > > SCRATCH_DEV=/dev/mapper/xt-vdc > > > > > > where /dev/mapper/xt-vdc is a symlink to /dev/dm-4 (or some such) > > > without any problems. So I don't quite understand why we need to > > > canonicalize devices? > > > > That might work, but try using /dev/disk/by-id/ stuff, that'll bust. So > > to keep existing expecations by fstests, it's needed. > > What goes wrong, and why? /dev/disk/by-id/<disk-id> is a symlink, > just like /dev/mapper/<vg>-<lv> is a symlink. > > What am I missing? # mkfs.xfs -f /dev/sda # mount /dev/sda /mnt # TEST_DIR=/mnt TEST_DEV=/dev/sda FSTYP=xfs ./check generic/110 FSTYP -- xfs (debug) PLATFORM -- Linux/x86_64 flax-mtr01 6.5.0-rc3-djwx #rc3 SMP PREEMPT_DYNAMIC Wed Jul 26 14:26:48 PDT 2023 generic/110 2s Ran: generic/110 Passed all 1 tests versus: # TEST_DIR=/mnt TEST_DEV=/dev/disk/by-id/scsi-0QEMU_RAMDISK_drive-scsi0-0-0-0 FSTYP=xfs ./check generic/110 mount: /mnt: /dev/sda already mounted on /mnt. common/rc: retrying test device mount with external set mount: /mnt: /dev/sda already mounted on /mnt. common/rc: could not mount /dev/disk/by-id/scsi-0QEMU_RAMDISK_drive-scsi0-0-0-0 on /mnt # umount /mnt # TEST_DIR=/mnt TEST_DEV=/dev/disk/by-id/scsi-0QEMU_RAMDISK_drive-scsi0-0-0-0 FSTYP=xfs ./check generic/110 TEST_DEV=/dev/disk/by-id/scsi-0QEMU_RAMDISK_drive-scsi0-0-0-0 is mounted but not on TEST_DIR=/mnt - aborting Already mounted result: /dev/sda /mnt (This is not really how I run fstests, it's just the minimum example.) --D > Thanks, > > - Ted >
