From: Andreas Dilger Subject: Re: [PATCH][RFC] resize2fs and uninit_bg questions Date: Fri, 18 Sep 2009 07:56:53 -0600 Message-ID: <20090918135653.GR2537@webber.adilger.int> References: <20090916162457.GA84213@freezingfog.local> <20090916190831.GH2537@webber.adilger.int> <20090916204225.GB84213@freezingfog.local> <20090916212250.GL2537@webber.adilger.int> <20090916231131.GC84213@freezingfog.local> <4AB28DD5.30607@redhat.com> <4AB37248.4050309@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; CHARSET=US-ASCII Content-Transfer-Encoding: 7BIT Cc: "Martin K. Petersen" , Will Drewry , linux-ext4@vger.kernel.org To: Ric Wheeler Return-path: Received: from sca-es-mail-1.Sun.COM ([192.18.43.132]:47508 "EHLO sca-es-mail-1.sun.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751200AbZIRN4u (ORCPT ); Fri, 18 Sep 2009 09:56:50 -0400 Received: from fe-sfbay-09.sun.com ([192.18.43.129]) by sca-es-mail-1.sun.com (8.13.7+Sun/8.12.9) with ESMTP id n8IDurwr000559 for ; Fri, 18 Sep 2009 06:56:53 -0700 (PDT) Content-disposition: inline Received: from conversion-daemon.fe-sfbay-09.sun.com by fe-sfbay-09.sun.com (Sun Java(tm) System Messaging Server 7u2-7.04 64bit (built Jul 2 2009)) id <0KQ6000006EV0X00@fe-sfbay-09.sun.com> for linux-ext4@vger.kernel.org; Fri, 18 Sep 2009 06:56:53 -0700 (PDT) In-reply-to: <4AB37248.4050309@redhat.com> Sender: linux-ext4-owner@vger.kernel.org List-ID: On Sep 18, 2009 07:43 -0400, Ric Wheeler wrote: > On 09/18/2009 02:09 AM, Martin K. Petersen wrote: >>>>>>> "Ric" == Ric Wheeler writes: >>>>>>> >> Ric> One note - the WRITE_SAME command in SCSI has long been used by >> Ric> array vendors to do relatively high performance zero fills. >> >> I've been working on WRITE SAME and UNMAP support recently (layering on >> top of Christoph's patch kit). It might be worthwhile to expose the >> WRITE SAME functionality further up the stack. It could also be helpful >> for initializing RAID disks, for instance... > > Using it to initialize RAID disks is exactly what it is there for - that > sounds quite useful... In ext4 we already have a routine ext4_ext_zeroout() that is used to zero out chunks of disk in order to wipe uninitialized extents when they need to be put into use. It currently maps ZERO_PAGE() multiple times to avoid causing cache pressure, but if we could wire it up to some block-level routine that optimized this it would be even better. Cheers, Andreas -- Andreas Dilger Sr. Staff Engineer, Lustre Group Sun Microsystems of Canada, Inc.