Return-Path: linux-nfs-owner@vger.kernel.org Received: from bombadil.infradead.org ([198.137.202.9]:40295 "EHLO bombadil.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753412Ab3KDUNX (ORCPT ); Mon, 4 Nov 2013 15:13:23 -0500 Date: Mon, 4 Nov 2013 12:13:20 -0800 From: Christoph Hellwig To: "Haynes, Tom" Cc: Chuck Lever , Christoph Hellwig , "Schumaker, Bryan" , "J. Bruce Fields" , Mailing List Linux NFS Subject: Re: [PATCH 3/4] NFSD: Add WRITE_PLUS support for hole punches Message-ID: <20131104201320.GA21884@infradead.org> References: <1382972247-1108-4-git-send-email-bjschuma@netapp.com> <20131102135238.GB18961@infradead.org> <5277CE23.6010207@netapp.com> <20131104170345.GA31499@infradead.org> <5277D80F.8050106@netapp.com> <20131104185309.GA14378@infradead.org> <3AC4ECC6-0496-4593-8791-92BCAD2567CC@oracle.com> <20131104191907.GA9337@infradead.org> <8EB0049A-9225-4C4A-8099-774B2EB48323@oracle.com> <1A74FF84-00E2-4C5F-A209-350792D2E0AE@netapp.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii In-Reply-To: <1A74FF84-00E2-4C5F-A209-350792D2E0AE@netapp.com> Sender: linux-nfs-owner@vger.kernel.org List-ID: On Mon, Nov 04, 2013 at 08:03:58PM +0000, Haynes, Tom wrote: > Do you have an issue with READ_PLUS? > > I.e., are you okay with that operation because the answer is not known beforehand? The hole vs data part of it looks reasonable. The whole ADH complex looks a little fishy to me, but I will have to dig deeper into before writing down a detailed critism. But one very obvious question already comes up: what data is supposed to be returned when a client that doesn't suport 4.2 at all are doesn't support ADHs reads a file that contains them. From a quick glance it seems a mapping from ADHs to flat blocks is possible, but such a mapping should be specified in the standard. > There was a strong consensus in the biweekly calls to merge all of the operations together into WRITE_PLUS. Do you remember any arguments for it? Also any rational for adding the plain old data write to it?