From: Trond Myklebust Subject: Re: [PATCH] NFS: Ensure that 'noac' and/or 'actimeo=0' turn off attribute caching Date: Mon, 05 May 2008 11:59:43 -0400 Message-ID: <1210003183.7409.5.camel@localhost> References: <20080502202502.29449.81285.stgit@c-69-242-210-120.hsd1.mi.comcast.net> <6A5A94A9-6F7A-469C-926D-A39B18381FFE@oracle.com> <1210001033.7409.1.camel@localhost> <95989F05-2BCE-4923-94C6-46D46270D120@oracle.com> Mime-Version: 1.0 Content-Type: text/plain Cc: linux-nfs@vger.kernel.org To: Chuck Lever Return-path: Received: from mx2.netapp.com ([216.240.18.37]:53988 "EHLO mx2.netapp.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754576AbYEEP7r (ORCPT ); Mon, 5 May 2008 11:59:47 -0400 In-Reply-To: <95989F05-2BCE-4923-94C6-46D46270D120@oracle.com> Sender: linux-nfs-owner@vger.kernel.org List-ID: On Mon, 2008-05-05 at 11:55 -0400, Chuck Lever wrote: > nfs_update_inode() is the only other place that actually uses the > value of nfsi->attrtimeo (and it uses it with time_in_range). I'm > merely asking if we have verified that the behavior is correct if > nfsi- > >attrtimeo == 0. > > The logic there is abstruse, so it's difficult to tell if the (nfsi- > >attrtimeo == 0) case is behaving as expected. I don't see any > obvious problem with it right now. That part of the code is irrelevant as far as actimeo=0/noac is concerned. It is the mechanism that decides whether or not to double the existing value of nfsi->attrtimeo for the case where the file/directory hasn't changed in a while. Cheers Trond -- Trond Myklebust Linux NFS client maintainer NetApp Trond.Myklebust@netapp.com www.netapp.com