Return-Path: linux-nfs-owner@vger.kernel.org Received: from mail-vb0-f46.google.com ([209.85.212.46]:33471 "EHLO mail-vb0-f46.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1756264Ab2D3SO1 (ORCPT ); Mon, 30 Apr 2012 14:14:27 -0400 Received: by vbbff1 with SMTP id ff1so2251022vbb.19 for ; Mon, 30 Apr 2012 11:14:27 -0700 (PDT) MIME-Version: 1.0 In-Reply-To: <1335809108.11828.1.camel@lade.trondhjem.org> References: <1335808841-11623-1-git-send-email-Trond.Myklebust@netapp.com> <1335808841-11623-2-git-send-email-Trond.Myklebust@netapp.com> <1335809108.11828.1.camel@lade.trondhjem.org> Date: Mon, 30 Apr 2012 14:14:26 -0400 Message-ID: Subject: Re: [PATCH 2/3] NFS: O_DIRECT pgio_completion_ops error_cleanup must unlock the request From: Fred Isaman To: "Myklebust, Trond" Cc: "linux-nfs@vger.kernel.org" , "Isaman, Fred" Content-Type: text/plain; charset=ISO-8859-1 Sender: linux-nfs-owner@vger.kernel.org List-ID: On Mon, Apr 30, 2012 at 2:05 PM, Myklebust, Trond wrote: > Alternatively, we could probably just send the write requests in an > unlocked state. They aren't supposed to be visible to anything via any > global lists... > The existing lock/commit code fairly often assumes the reqs are locked. I found it simpler to just lock the req than to audit all the code paths looking for where the assumptions made a difference. Fred