fix-ZERO_RANGE-test-failure-in-data-journalling

   1 ext4: fix ZERO_RANGE test failure in data journalling
   2
   3 From: Namjae Jeon <namjae.jeon@samsung.com>
   4
   5 xfstests generic/091 is failing when mounting ext4 with data=journal.
   6 I think that this regression is same problem that occurred prior to collapse
   7 range issue. So ZERO RANGE also need to call ext4_force_commit as
   8 collapse range.
   9
  10 Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
  11 Signed-off-by: Ashish Sangwan <a.sangwan@samsung.com>
  12 Signed-off-by: Theodore Ts'o <tytso@mit.edu>
  13 ---
  14  fs/ext4/extents.c | 9 +++++++++
  15  1 file changed, 9 insertions(+)
  16
  17 diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c
  18 index 1c5e877..02143d6 100644
  19 --- a/fs/ext4/extents.c
  20 +++ b/fs/ext4/extents.c
  21 @@ -4743,6 +4743,15 @@ static long ext4_zero_range(struct file *file, loff_t offset,
  22
  23         mutex_lock(&EXT4_I(inode)->i_write_mutex);
  24
  25 +       /* Call ext4_force_commit to flush all data in case of data=journal. */
  26 +       if (ext4_should_journal_data(inode)) {
  27 +               ret = ext4_force_commit(inode->i_sb);
  28 +               if (ret) {
  29 +                       mutex_unlock(&EXT4_I(inode)->i_write_mutex);
  30 +                       return ret;
  31 +               }
  32 +       }
  33 +
  34         /*
  35          * Write out all dirty pages to avoid race conditions
  36          * Then release them.
  37 --
  38 1.7.11-rc0
  39
  40 --
  41 To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
  42 the body of a message to majordomo@vger.kernel.org
  43 More majordomo info at  http://vger.kernel.org/majordomo-info.html
  44