4 # This test is motivated by an fsync issue discovered in btrfs.
5 # The issue was that after punching a hole for a small range, which affected
6 # only a partial page, an fsync operation would have no effect at all. This was
7 # because for this particular case the btrfs hole punching implementation did
8 # not update some btrfs specific inode metadata that is required to determine
9 # if an fsync operation needs to update the fsync log. For this to happen, it
10 # was also necessary that in the transaction where the hole punching was
11 # performed, and before the fsync operation, no other operation that modified
12 # the file (or its metadata) was performed.
14 # The btrfs issue was fixed by the following linux kernel patch:
16 # Btrfs: add missing inode update when punching hole
18 #-----------------------------------------------------------------------
19 # Copyright (C) 2015 SUSE Linux Products GmbH. All Rights Reserved.
20 # Author: Filipe Manana <fdmanana@suse.com>
22 # This program is free software; you can redistribute it and/or
23 # modify it under the terms of the GNU General Public License as
24 # published by the Free Software Foundation.
26 # This program is distributed in the hope that it would be useful,
27 # but WITHOUT ANY WARRANTY; without even the implied warranty of
28 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
29 # GNU General Public License for more details.
31 # You should have received a copy of the GNU General Public License
32 # along with this program; if not, write the Free Software Foundation,
33 # Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
34 #-----------------------------------------------------------------------
38 seqres=$RESULT_DIR/$seq
39 echo "QA output created by $seq"
43 status=1 # failure is the default!
50 trap "_cleanup; exit \$status" 0 1 2 3 15
52 # get standard environment, filters and checks
57 # real QA test starts here
63 _require_metadata_journaling $SCRATCH_DEV
64 _require_xfs_io_command "fpunch"
68 _scratch_mkfs >> $seqres.full 2>&1
72 # Create our test file.
73 $XFS_IO_PROG -f -c "pwrite -S 0x22 -b 16K 0 16K" \
74 $SCRATCH_MNT/foo | _filter_xfs_io
76 # Fsync the file, this makes btrfs update some btrfs inode specific fields
77 # that are used to track if the inode needs to be written/updated to the fsync
78 # log or not. After this fsync, the new values for those fields indicate that
79 # a subsequent fsync does not need to touch the fsync log.
80 $XFS_IO_PROG -c "fsync" $SCRATCH_MNT/foo
82 # Force a commit of the current transaction. After this point, any operation
83 # that modifies the data or metadata of our file, should update those fields in
84 # the btrfs inode with values that make the next fsync operation write to the
88 # Punch a hole in our file. This small range affects only 1 page.
89 # This made the btrfs hole punching implementation write only some zeroes in
90 # one page, but it did not update the btrfs inode fields used to determine if
91 # the next fsync needs to write to the fsync log.
92 $XFS_IO_PROG -c "fpunch 8000 4K" $SCRATCH_MNT/foo
94 # Another variation of the previously mentioned case.
95 $XFS_IO_PROG -c "fpunch 15000 100" $SCRATCH_MNT/foo
97 # Now fsync the file. This was a no-operation because the previous hole punch
98 # operation didn't update the inode's fields mentioned before, so they remained
99 # with the values they had after the first fsync - that is, they indicate that
100 # it is not needed to write to fsync log.
101 $XFS_IO_PROG -c "fsync" $SCRATCH_MNT/foo
103 echo "File content before:"
104 od -t x1 $SCRATCH_MNT/foo
106 # Simulate a crash/power loss.
107 _load_flakey_table $FLAKEY_DROP_WRITES
110 # Enable writes and mount the fs. This makes the fsync log replay code run.
111 _load_flakey_table $FLAKEY_ALLOW_WRITES
114 # Because the last fsync didn't do anything, here the file content matched what
115 # it was after the first fsync, before the holes were punched, and not what it
116 # was after the holes were punched.
117 echo "File content after:"
118 od -t x1 $SCRATCH_MNT/foo