From mboxrd@z Thu Jan 1 00:00:00 1970 From: Tejun Heo Subject: Re: [patch 3/3] raid5: only wakeup necessary threads Date: Tue, 30 Jul 2013 10:01:06 -0400 Message-ID: <20130730140106.GE12016@htj.dyndns.org> References: <20130730055207.698660010@kernel.org> <20130730055425.056034691@kernel.org> <20130730124655.GB2599@htj.dyndns.org> <20130730132414.GB30352@kernel.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: <20130730132414.GB30352@kernel.org> Sender: linux-kernel-owner@vger.kernel.org To: Shaohua Li Cc: linux-raid@vger.kernel.org, linux-kernel@vger.kernel.org, neilb@suse.de, djbw@fb.com List-Id: linux-raid.ids Hello, Shaohua. On Tue, Jul 30, 2013 at 09:24:14PM +0800, Shaohua Li wrote: > In the raid5 case, it's very hard to keep the order the bios passed in, because > we need read some disks, calculate parity, and write some disks, the timing > could break any kind of order. Besides the workqueue handles 8 stripes one > time, so I suppose this keeps some order if there is. Of course, it can't be absolute but still keeping the relative ordering and temporal locality (ie. issue IOs in order once the XOR calculation of the whole source bio is done rather than issuing as each block completes calculation) would help quite a bit. Thanks. -- tejun