Return-Path: linux-nfs-owner@vger.kernel.org Received: from verein.lst.de ([213.95.11.211]:52946 "EHLO newverein.lst.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752218AbaIKPsh (ORCPT ); Thu, 11 Sep 2014 11:48:37 -0400 Date: Thu, 11 Sep 2014 17:48:34 +0200 From: Christoph Hellwig To: Trond Myklebust Cc: Christoph Hellwig , Peng Tao , Boaz Harrosh , Linux NFS Mailing List Subject: Re: [PATCH 8/9] pnfs/blocklayout: return layouts on setattr Message-ID: <20140911154834.GA9280@lst.de> References: <1410362617-28018-1-git-send-email-hch@lst.de> <1410362617-28018-9-git-send-email-hch@lst.de> <20140911152553.GD6690@lst.de> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii In-Reply-To: Sender: linux-nfs-owner@vger.kernel.org List-ID: On Thu, Sep 11, 2014 at 11:38:24AM -0400, Trond Myklebust wrote: > Why would it be needed? The layout isn't expected to change. If the > chown affects permissions then it is up to the DS to enforce that > (although POSIX does not require it to do that). I was wondering about the truncate case. Even if the DS needs to be able to enforce the new size it seems pointless to keep a layout beyond the size around. I don't really see a need to drop on a chown for the blocklayout or objlayout drivers either, given that these semantics are enforced at a higher level.