Return-Path: Received: from fieldses.org ([173.255.197.46]:40072 "EHLO fieldses.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753044AbcLCDxj (ORCPT ); Fri, 2 Dec 2016 22:53:39 -0500 Date: Fri, 2 Dec 2016 22:53:02 -0500 From: "J. Bruce Fields" To: Trond Myklebust , Anna Schumaker Cc: Andreas Gruenbacher , Linux NFS Mailing List Subject: NFSv4.2 mode_umask support Message-ID: <20161203035302.GC21785@fieldses.org> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Sender: linux-nfs-owner@vger.kernel.org List-ID: Since the last version, just two minor changes: - client uses the stored supported attribute results instead of a new capability flag. - if a nutty client attempts to set both mode and new attribute, the server returns INVAL instead of ignoring the mode. Description, as before: The following patches allow the umask to be ignored in the presence of inheritable NFSv4 ACLs. Otherwise inheritable ACLs can be rendered mostly useless whenever the umask masks out group bits. This solves a problem we've seen complaints about for some time, both upstream and from RHEL users. The new protocol has been discussed in the IETF working group and is documented at: https://tools.ietf.org/html/draft-ietf-nfsv4-umask-02 It's unlikely that we'll discover problems requiring an incompatible change, so I think we should consider this for 4.10. --b.