From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753329AbZH3C4S (ORCPT ); Sat, 29 Aug 2009 22:56:18 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1752777AbZH3C4S (ORCPT ); Sat, 29 Aug 2009 22:56:18 -0400 Received: from bombadil.infradead.org ([18.85.46.34]:51602 "EHLO bombadil.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752666AbZH3C4R (ORCPT ); Sat, 29 Aug 2009 22:56:17 -0400 Date: Sat, 29 Aug 2009 22:56:16 -0400 From: Christoph Hellwig To: Mark Lord Cc: Matthew Wilcox , Christoph Hellwig , linux-scsi@vger.kernel.org, linux-ide@vger.kernel.org, linux-kernel@vger.kernel.org, jens.axboe@oracle.com, dwmw2@infradead.org Subject: Re: [PATCH 6/7] block: allow large discard requests Message-ID: <20090830025616.GA26428@infradead.org> References: <20090829230332.017137693@bombadil.infradead.org> <20090829231122.302813585@bombadil.infradead.org> <4A99E89E.9020204@rtr.ca> <20090830025022.GE22870@parisc-linux.org> <4A99E98B.403@rtr.ca> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <4A99E98B.403@rtr.ca> User-Agent: Mutt/1.5.19 (2009-01-05) X-SRS-Rewrite: SMTP reverse-path rewritten from by bombadil.infradead.org See http://www.infradead.org/rpr.html Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Sat, Aug 29, 2009 at 10:52:59PM -0400, Mark Lord wrote: >> That depends on the firmware. I do have plans to group the extents, >> but let's get something working before making it quick. > .. > > In this situation, it's more a case of making it *not sucky slow* > than getting to the point where we think about making it "quick". > > One extent at a time is glacial. Really depends on the use case. With XFS you will only have very few free space extents per AG, so it should work reasonably well for the batched "offline" discard in my last patch. And for SCSI arrays that can actually process other I/O during that time the WRITE SAME is in progress it should not be that bad. But so far this is all a lot of talk of course until I can actually work against real hardware.