Return-Path: Received: from mail-vk0-f48.google.com ([209.85.213.48]:37603 "EHLO mail-vk0-f48.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S2388036AbeGTUPT (ORCPT ); Fri, 20 Jul 2018 16:15:19 -0400 Received: by mail-vk0-f48.google.com with SMTP id v72-v6so925107vkd.4 for ; Fri, 20 Jul 2018 12:25:39 -0700 (PDT) MIME-Version: 1.0 In-Reply-To: <839932d583bf0962c636f4494cf9f387d4aba6fd.camel@hammerspace.com> References: <839932d583bf0962c636f4494cf9f387d4aba6fd.camel@hammerspace.com> From: Olga Kornievskaia Date: Fri, 20 Jul 2018 15:25:38 -0400 Message-ID: Subject: Re: Another question about v3 attribute handling To: Trond Myklebust Cc: "linux-nfs@vger.kernel.org" Content-Type: text/plain; charset="UTF-8" Sender: linux-nfs-owner@vger.kernel.org List-ID: On Fri, Jul 20, 2018 at 2:50 PM, Trond Myklebust wrote: > On Fri, 2018-07-20 at 14:44 -0400, Olga Kornievskaia wrote: >> Hi Trond, >> >> This is a different question but is related to the attribute >> handling. >> >> What I see in v4, is that say the client received a reply to the OPEN >> with before and after change_attr values for the directory. We check >> if the before value is the same as what we have in the cache and if >> it's not then we invalidate the directory attributes. >> >> It seems that we don't have similar check for the v3. If in v3, the >> server returned a before that was different from what could be seen >> on >> the network trace as the before for the directory, it does not >> trigger >> invalidation. Is this expected because it's v3 or is this a bug? > > Doesn't the call to nfs_post_op_update_inode(dir, data->res.dir_attr) > in nfs3_do_create() do the right thing? Dah, I knew I was missing something. Thanks.