Return-Path: linux-nfs-owner@vger.kernel.org Received: from mail-vc0-f174.google.com ([209.85.220.174]:37570 "EHLO mail-vc0-f174.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754130AbbBZXGo (ORCPT ); Thu, 26 Feb 2015 18:06:44 -0500 Received: by mail-vc0-f174.google.com with SMTP id id10so5211333vcb.5 for ; Thu, 26 Feb 2015 15:06:43 -0800 (PST) MIME-Version: 1.0 In-Reply-To: <1DD59CAE-F2C8-45D3-9613-908064D8E476@oracle.com> References: <1424914057.41161.8.camel@primarydata.com> <1DA1252F-8D1D-4A77-BF8B-51B8AD30975A@oracle.com> <1DD59CAE-F2C8-45D3-9613-908064D8E476@oracle.com> Date: Thu, 26 Feb 2015 18:06:43 -0500 Message-ID: Subject: Re: fsx size error (was: File Read Returns Non-existent Null Bytes) From: Trond Myklebust To: Chuck Lever Cc: Linux NFS Mailing List Content-Type: text/plain; charset=UTF-8 Sender: linux-nfs-owner@vger.kernel.org List-ID: On Thu, Feb 26, 2015 at 2:00 PM, Chuck Lever wrote: >> The alternative would be to add a "barrier" operation to allow the >> setattr code to wait for all RPC calls before ours to complete (it >> would have to wait not just for READ, but also GETATTR and all other >> operations to the file). We could do that too. > > Yes. A barrier is usually heavyweight, though. And I thought there > was some exclusion between GETATTR and SETATTR, so forcing an > inode revalidation should work too? Or would that result in an > unneeded data cache invalidation? > I found a better solution, I think. Please see the 5 patches I just sent out. -- Trond Myklebust Linux NFS client maintainer, PrimaryData trond.myklebust@primarydata.com