fix-an-endianness-bug-in-ext4_encrypted_follow_link

   1 ext4: fix an endianness bug in ext4_encrypted_follow_link()
   2
   3 From: Al Viro <viro@ZenIV.linux.org.uk>
   4
   5 applying le32_to_cpu() to 16bit value is a bad idea...
   6
   7 Cc: stable@vger.kernel.org # v4.1+
   8 Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
   9 Signed-off-by: Theodore Ts'o <tytso@mit.edu>
  10 ---
  11
  12 diff --git a/fs/ext4/symlink.c b/fs/ext4/symlink.c
  13 index abe2401..e8e7af6 100644
  14 --- a/fs/ext4/symlink.c
  15 +++ b/fs/ext4/symlink.c
  16 @@ -52,7 +52,7 @@ static const char *ext4_encrypted_follow_link(struct dentry *dentry, void **cook
  17         /* Symlink is encrypted */
  18         sd = (struct ext4_encrypted_symlink_data *)caddr;
  19         cstr.name = sd->encrypted_path;
  20 -       cstr.len  = le32_to_cpu(sd->len);
  21 +       cstr.len  = le16_to_cpu(sd->len);
  22         if ((cstr.len +
  23              sizeof(struct ext4_encrypted_symlink_data) - 1) >
  24             max_size) {
  25