releases/4.10.7/ext4-mark-inode-dirty-after-converting-inline-directory.patch

   1 From b9cf625d6ecde0d372e23ae022feead72b4228a6 Mon Sep 17 00:00:00 2001
   2 From: Eric Biggers <ebiggers@google.com>
   3 Date: Wed, 15 Mar 2017 14:52:02 -0400
   4 Subject: ext4: mark inode dirty after converting inline directory
   5
   6 From: Eric Biggers <ebiggers@google.com>
   7
   8 commit b9cf625d6ecde0d372e23ae022feead72b4228a6 upstream.
   9
  10 If ext4_convert_inline_data() was called on a directory with inline
  11 data, the filesystem was left in an inconsistent state (as considered by
  12 e2fsck) because the file size was not increased to cover the new block.
  13 This happened because the inode was not marked dirty after i_disksize
  14 was updated.  Fix this by marking the inode dirty at the end of
  15 ext4_finish_convert_inline_dir().
  16
  17 This bug was probably not noticed before because most users mark the
  18 inode dirty afterwards for other reasons.  But if userspace executed
  19 FS_IOC_SET_ENCRYPTION_POLICY with invalid parameters, as exercised by
  20 'kvm-xfstests -c adv generic/396', then the inode was never marked dirty
  21 after updating i_disksize.
  22
  23 Fixes: 3c47d54170b6a678875566b1b8d6dcf57904e49b
  24 Signed-off-by: Eric Biggers <ebiggers@google.com>
  25 Signed-off-by: Theodore Ts'o <tytso@mit.edu>
  26 Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
  27
  28 ---
  29  fs/ext4/inline.c |    5 ++---
  30  1 file changed, 2 insertions(+), 3 deletions(-)
  31
  32 --- a/fs/ext4/inline.c
  33 +++ b/fs/ext4/inline.c
  34 @@ -1167,10 +1167,9 @@ static int ext4_finish_convert_inline_di
  35         set_buffer_uptodate(dir_block);
  36         err = ext4_handle_dirty_dirent_node(handle, inode, dir_block);
  37         if (err)
  38 -               goto out;
  39 +               return err;
  40         set_buffer_verified(dir_block);
  41 -out:
  42 -       return err;
  43 +       return ext4_mark_inode_dirty(handle, inode);
  44  }
  45
  46  static int ext4_convert_inline_data_nolock(handle_t *handle,