1 Revert "ext4: use ext4_write_inode() when fsyncing w/o a journal"
3 This reverts commit ad211f3e94b314a910d4af03178a0b52a7d1ee0a.
5 As Jan Kara pointed out, this change was unsafe since it means we lose
6 the call to sync_mapping_buffers() in the nojournal case. The
7 original point of the commit was avoid taking the inode mutex (since
8 it causes a lockdep warning in generic/113); but we need the mutex in
9 order to call sync_mapping_buffers().
11 The real fix to this problem was discussed here:
13 https://lore.kernel.org/lkml/20181025150540.259281-4-bvanassche@acm.org
15 The proposed patch was to fix a syzbot complaint, but the problem can
16 also demonstrated via "kvm-xfstests -c nojournal generic/113".
17 Multiple solutions were discused in the e-mail thread, but none have
18 landed in the kernel as of this writing. Anyway, commit
19 ad211f3e94b314 is absolutely the wrong way to suppress the lockdep, so
22 Fixes: ad211f3e94b314a910d4af03178a0b52a7d1ee0a ("ext4: use ext4_write_inode() when fsyncing w/o a journal")
23 Signed-off-by: Theodore Ts'o <tytso@mit.edu>
24 Reported: Jan Kara <jack@suse.cz>
26 fs/ext4/fsync.c | 13 ++++---------
27 1 file changed, 4 insertions(+), 9 deletions(-)
29 diff --git a/fs/ext4/fsync.c b/fs/ext4/fsync.c
30 index 712f00995390..5508baa11bb6 100644
33 @@ -116,16 +116,8 @@ int ext4_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
37 - ret = file_write_and_wait_range(file, start, end);
42 - struct writeback_control wbc = {
43 - .sync_mode = WB_SYNC_ALL
46 - ret = ext4_write_inode(inode, &wbc);
47 + ret = __generic_file_fsync(file, start, end, datasync);
49 ret = ext4_sync_parent(inode);
50 if (test_opt(inode->i_sb, BARRIER))
51 @@ -133,6 +125,9 @@ int ext4_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
55 + ret = file_write_and_wait_range(file, start, end);
59 * data=writeback,ordered:
60 * The caller's filemap_fdatawrite()/wait will sync the data.