1 ext4: allow inode expansion for nojournal file systems
3 From: Eric Whitney <enwlinux@gmail.com>
5 Runs of xfstest ext4/022 on nojournal file systems result in failures
6 because the inodes of some of its test files do not expand as expected.
7 The cause is a conditional in ext4_mark_inode_dirty() that prevents inode
8 expansion unless the test file system has a journal. Remove this
9 unnecessary restriction.
11 Signed-off-by: Eric Whitney <enwlinux@gmail.com>
12 Signed-off-by: Theodore Ts'o <tytso@mit.edu>
14 fs/ext4/inode.c | 12 +++++++-----
15 1 file changed, 7 insertions(+), 5 deletions(-)
17 diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
18 index 9c06472..260da4d 100644
21 @@ -5455,18 +5455,20 @@ int ext4_mark_inode_dirty(handle_t *handle, struct inode *inode)
22 err = ext4_reserve_inode_write(handle, inode, &iloc);
25 - if (ext4_handle_valid(handle) &&
26 - EXT4_I(inode)->i_extra_isize < sbi->s_want_extra_isize &&
27 + if (EXT4_I(inode)->i_extra_isize < sbi->s_want_extra_isize &&
28 !ext4_test_inode_state(inode, EXT4_STATE_NO_EXPAND)) {
30 - * We need extra buffer credits since we may write into EA block
31 + * In nojournal mode, we can immediately attempt to expand
32 + * the inode. When journaled, we first need to obtain extra
33 + * buffer credits since we may write into the EA block
34 * with this same handle. If journal_extend fails, then it will
35 * only result in a minor loss of functionality for that inode.
36 * If this is felt to be critical, then e2fsck should be run to
37 * force a large enough s_min_extra_isize.
39 - if ((jbd2_journal_extend(handle,
40 - EXT4_DATA_TRANS_BLOCKS(inode->i_sb))) == 0) {
41 + if (!ext4_handle_valid(handle) ||
42 + jbd2_journal_extend(handle,
43 + EXT4_DATA_TRANS_BLOCKS(inode->i_sb)) == 0) {
44 ret = ext4_expand_extra_isize(inode,
45 sbi->s_want_extra_isize,