From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andi Kleen Subject: Re: cmwq and dm-crypt devices? Date: Wed, 3 Nov 2010 12:51:44 +0100 Message-ID: <20101103115144.GA18935@basil.fritz.box> References: <1271958538-11193-1-git-send-email-san@google.com> <4BD08F78.4000701@redhat.com> <4BD099DB.2020108@redhat.com> <20101102220207.GD23680@redhat.com> <4CD12F6E.8040501@kernel.org> Reply-To: device-mapper development Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: Content-Disposition: inline In-Reply-To: <4CD12F6E.8040501@kernel.org> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: dm-devel-bounces@redhat.com Errors-To: dm-devel-bounces@redhat.com To: Tejun Heo Cc: Christophe Saout , Mike Snitzer , Brian Swetland , San Mehat , device-mapper development , andi@firstfloor.org, Andrew Morton , Alasdair G Kergon , Milan Broz List-Id: dm-devel.ids > > > 2) scale up the number of workqueue threads used for a single dm-crypt > > device so that a device can realize per-cpu concurrency (to address > > Andi's scalability concerns: https://patchwork.kernel.org/patch/244031/) They are already addressed in my patchkit and the patches seem to be used by more and more users. It's just you guys who are behind. > > > > [the desired locality is currently missing due to dm-crypt's current > > use of WQ_UNBOUND; so it is clear the way the workqueues are created > > will be important] > > I don't know enough about dm-crypt workload to tell whether per-cpu > affinity would be better or not, but it's really a simple matter of CPU affinity and an own thread makes sense for the crypto helper because it uses up a lot of CPU time. For the IO helper you probably still want CPU affinity, but it can be concurrency managed. -andi