Re: [PATCH 2/2] dm-thinp demo test

[Eric Sandeen <sandeen@redhat.com>]

On 10/22/15 9:14 AM, Eric Sandeen wrote:
>>> +_workout
>>> >> +_dmthin_check_fs
>> > 
>> > btrfs fails the fs check, because btrfs changes lv device name in df
>> > output, from /dev/mapper/xxx to /dev/dm-x, so _check_scratch_fs couldn't
>> > umount it and btrfsck runs on mounted btrfs.
> Huh, weird.  I'll admit that I did not test btrfs.  Other filesystems
> worked for you?  I'm not sure how btrfs can change df output...
> 
> mapper symlinks have been a constant source of pain for fstests.  :(
> 
>> > Adding "$UMOUNT_PROG $SCRATCH_MNT" before _dmthin_check_fs is a
>> > workaround, though I think it's a btrfs issue.
> Hm, ok, thanks; easy enough to do that I guess.

ALthough, it seems to be working for me here; I do fail because btrfs
emits a warning in dmesg:

[215098.204612] BTRFS warning (device dm-6): Skipping commit of aborted transaction.
[215098.212074] ------------[ cut here ]------------
[215098.216794] WARNING: CPU: 3 PID: 19408 at fs/btrfs/transaction.c:1708 cleanup_transaction+0x86/0x2c0 [btrfs]()

which leads to:

_check_dmesg: something found in dmesg (see /mnt/test2/git/xfstests/results//generic/115.dmesg)
Ran: generic/115
Failures: generic/115

It's "just" a warning, though, so I'm not really sure this should be failing the test.

-Eric