From mboxrd@z Thu Jan 1 00:00:00 1970 From: Hannes Reinecke Subject: blk-mq and interrupt affinity Date: Fri, 29 Jan 2016 10:40:27 +0100 Message-ID: <56AB338B.4020804@suse.de> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Return-path: Received: from mx2.suse.de ([195.135.220.15]:59059 "EHLO mx2.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751475AbcA2Jk3 (ORCPT ); Fri, 29 Jan 2016 04:40:29 -0500 Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: Jens Axboe Cc: "linux-block@vger.kernel.org" , "linux-scsi@vger.kernel.org" , Christoph Hellwig Hi all, Looking through the blk-mq code I tried to figure out if and how the driver is supposed to set any interrupt affinity when block-mq is enabled. Thing is, for your typical driver you would set the interrupt affinity during init time (eg for SCSI drivers during host_alloc). But to set the interrupt affinity you'd need the cpumask, which is only populated way later during queue setup. So now I got two bad choices: either I have to delay setting the interrupt affinity until after the queue is created, making things awkward with several queues, or I have to out-guess the blk-mq mapping algorithm and set the interrupt affinity prior to queue setup. Tests with the lpfc driver found that one absolutely _does_ want to do irq affinity; not doing so limits the performance to single-queue results. So what is the expected strategy here? Or am I just too stupid to find it? Cheers, Hannes --=20 Dr. Hannes Reinecke Teamlead Storage & Networking hare@suse.de +49 911 74053 688 SUSE LINUX GmbH, Maxfeldstr. 5, 90409 N=C3=BCrnberg GF: F. Imend=C3=B6rffer, J. Smithard, J. Guild, D. Upmanyu, G. Norton HRB 21284 (AG N=C3=BCrnberg) -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" i= n the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html