Return-Path: Received: from mail-ob0-f177.google.com ([209.85.214.177]:36089 "EHLO mail-ob0-f177.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750816AbbEFKMO (ORCPT ); Wed, 6 May 2015 06:12:14 -0400 Received: by obbkp3 with SMTP id kp3so3307404obb.3 for ; Wed, 06 May 2015 03:12:13 -0700 (PDT) MIME-Version: 1.0 In-Reply-To: <20150506065651.GA22454@lst.de> References: <1430895024-1403-1-git-send-email-andreas.gruenbacher@gmail.com> <20150506065651.GA22454@lst.de> Date: Wed, 6 May 2015 12:12:13 +0200 Message-ID: Subject: Re: [PATCH] nfsd: Disable NFSv2 timestamp workaround for NFSv3+ From: =?UTF-8?Q?Andreas_Gr=C3=BCnbacher?= To: Christoph Hellwig Cc: "J. Bruce Fields" , linux-nfs@vger.kernel.org Content-Type: text/plain; charset=UTF-8 Sender: linux-nfs-owner@vger.kernel.org List-ID: 2015-05-06 8:56 GMT+02:00 Christoph Hellwig : > How about moving the workaround into the NFSv2 specific code? Not trivially, we would have to fh_verify() the file handle in nfsd_proc_setattr() first. Is that preferable? > Looks like the call from nfsd_create() should never be used for this > workaround, Right, the workaround won't trigger there because setting the times arbitrarily will already succeed because we are the owner (nfsd_sanitize_attrs tries that first with inode_change_ok). Thanks, Andreas