Comment # 6 on bug 942925 from Filipe Manana

(In reply to Jeff Mahoney from comment #3)
> Actually, SLE12 passes the xfstests that test these problems. It looks like
> it's just openSUSE releases that are affected.  I've applied the fixes to
> master and will test previous openSUSE releases.

That's weird Jeff. Can you double check you were testing the correct kernel
(unpatched sle12)?

These issues have been present for several years, essentially since the log
trees were added to btrfs, and for me the tests fail on SLE12 as expected:

root 14:05:12 /home/fdmanana/git/xfstests-dev (master)> ./check generic/104
FSTYP         -- btrfs
PLATFORM      -- Linux/x86_64 suse1 3.12.45-debug-5+
MKFS_OPTIONS  -- /dev/sdc
MOUNT_OPTIONS -- /dev/sdc /home/fdmanana/btrfs-tests/scratch_1

generic/104     - output mismatch (see
/home/fdmanana/git/xfstests-dev/results//generic/104.out.bad)
    --- tests/generic/104.out    2015-08-06 11:00:44.302727292 +0100
    +++ /home/fdmanana/git/xfstests-dev/results//generic/104.out.bad   
2015-08-28 14:05:20.277346562 +0100
    @@ -1,3 +1,5 @@
     QA output created by 104
    -Link count for file foo: 2
    +Link count for file foo: 1
     Link count for file bar: 2
    +rm: cannot remove '/home/fdmanana/btrfs-tests/scratch_1/testdir/foo_link':
Stale file handle
    +rmdir: failed to remove '/home/fdmanana/btrfs-tests/scratch_1/testdir':
Directory not empty
    ...
    (Run 'diff -u tests/generic/104.out
/home/fdmanana/git/xfstests-dev/results//generic/104.out.bad'  to see the
entire diff)
_check_btrfs_filesystem: filesystem on /dev/sdc is inconsistent (see
/home/fdmanana/git/xfstests-dev/results//generic/104.full)
Ran: generic/104
Failures: generic/104
Failed 1 of 1 tests

root 14:05:20 /home/fdmanana/git/xfstests-dev (master)> ./check generic/106
FSTYP         -- btrfs
PLATFORM      -- Linux/x86_64 suse1 3.12.45-debug-5+
MKFS_OPTIONS  -- /dev/sdc
MOUNT_OPTIONS -- /dev/sdc /home/fdmanana/btrfs-tests/scratch_1

generic/106     - output mismatch (see
/home/fdmanana/git/xfstests-dev/results//generic/106.out.bad)
    --- tests/generic/106.out    2015-08-06 11:00:44.302727292 +0100
    +++ /home/fdmanana/git/xfstests-dev/results//generic/106.out.bad   
2015-08-28 14:05:27.409333855 +0100
    @@ -1,3 +1,4 @@
     QA output created by 106
     Entries in testdir:
    +bar
     foo
    ...
    (Run 'diff -u tests/generic/106.out
/home/fdmanana/git/xfstests-dev/results//generic/106.out.bad'  to see the
entire diff)
Ran: generic/106
Failures: generic/106
Failed 1 of 1 tests

root 14:05:27 /home/fdmanana/git/xfstests-dev (master)> uname -a
Linux suse1 3.12.45-debug-5+ #1 SMP Fri Jul 31 16:25:23 WEST 2015 x86_64 x86_64
x86_64 GNU/Linux


Plus, there's at least a couple more fixes for this type of problem (wrong link
counts, stale dentries):

1) recent one in linux-next/for 4.3: 
https://git.kernel.org/cgit/linux/kernel/git/next/linux-next.git/commit/?id=18aa09229741364280d0a1670597b5207fc05b8d

2) This one landed in 4.0:
https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=2f2ff0ee5e4303e727cfd7abd4133d1a8ee68394,
and its test is:
https://git.kernel.org/cgit/fs/xfs/xfstests-dev.git/commit/?id=036a163772697950bad537e4b7267cb9d3754147

I was planning on adding all these to sle12 once I get back from vacations, and
they've been part of my regular local sle12 testing.