ext4-fix-mmap-data-corruption-when-blocksize-lt-pagesize

   1 ext4: fix mmap data corruption when blocksize < pagesize
   2
   3 From: Jan Kara <jack@suse.cz>
   4
   5 Use truncate_isize_extended() when hole is being created in a file so that
   6 ->page_mkwrite() will get called for the partial tail page if it is
   7 mmaped (see the first patch in the series for details).
   8
   9 Signed-off-by: Jan Kara <jack@suse.cz>
  10 Signed-off-by: Theodore Ts'o <tytso@mit.edu>
  11 ---
  12  fs/ext4/inode.c | 6 +++++-
  13  1 file changed, 5 insertions(+), 1 deletion(-)
  14
  15 diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
  16 index 3aa26e9117c4..9e269489e094 100644
  17 --- a/fs/ext4/inode.c
  18 +++ b/fs/ext4/inode.c
  19 @@ -4536,8 +4536,12 @@ int ext4_setattr(struct dentry *dentry, struct iattr *attr)
  20                                 ext4_orphan_del(NULL, inode);
  21                                 goto err_out;
  22                         }
  23 -               } else
  24 +               } else {
  25 +                       loff_t oldsize = inode->i_size;
  26 +
  27                         i_size_write(inode, attr->ia_size);
  28 +                       pagecache_isize_extended(inode, oldsize, inode->i_size);
  29 +               }
  30
  31                 /*
  32                  * Blocks are going to be removed from the inode. Wait
  33 --
  34 1.8.1.4
  35
  36 --
  37 To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
  38 the body of a message to majordomo@vger.kernel.org
  39 More majordomo info at  http://vger.kernel.org/majordomo-info.html
  40