releases/4.11.9/nfsv4.2-don-t-send-mode-again-in-post-exclusive4_1-setattr-with-umask.patch

   1 From 501e7a4689378f8b1690089bfdd4f1e12ec22903 Mon Sep 17 00:00:00 2001
   2 From: Benjamin Coddington <bcodding@redhat.com>
   3 Date: Fri, 2 Jun 2017 11:21:34 -0400
   4 Subject: NFSv4.2: Don't send mode again in post-EXCLUSIVE4_1 SETATTR with umask
   5
   6 From: Benjamin Coddington <bcodding@redhat.com>
   7
   8 commit 501e7a4689378f8b1690089bfdd4f1e12ec22903 upstream.
   9
  10 Now that we have umask support, we shouldn't re-send the mode in a SETATTR
  11 following an exclusive CREATE, or we risk having the same problem fixed in
  12 commit 5334c5bdac92 ("NFS: Send attributes in OPEN request for
  13 NFS4_CREATE_EXCLUSIVE4_1"), which is that files with S_ISGID will have that
  14 bit stripped away.
  15
  16 Signed-off-by: Benjamin Coddington <bcodding@redhat.com>
  17 Fixes: dff25ddb4808 ("nfs: add support for the umask attribute")
  18 Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
  19 Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
  20
  21 ---
  22  fs/nfs/nfs4proc.c |    3 ++-
  23  1 file changed, 2 insertions(+), 1 deletion(-)
  24
  25 --- a/fs/nfs/nfs4proc.c
  26 +++ b/fs/nfs/nfs4proc.c
  27 @@ -2588,7 +2588,8 @@ static inline void nfs4_exclusive_attrse
  28
  29         /* Except MODE, it seems harmless of setting twice. */
  30         if (opendata->o_arg.createmode != NFS4_CREATE_EXCLUSIVE &&
  31 -               attrset[1] & FATTR4_WORD1_MODE)
  32 +               (attrset[1] & FATTR4_WORD1_MODE ||
  33 +                attrset[2] & FATTR4_WORD2_MODE_UMASK))
  34                 sattr->ia_valid &= ~ATTR_MODE;
  35
  36         if (attrset[2] & FATTR4_WORD2_SECURITY_LABEL)