From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Nicholas A. Bellinger" Subject: Re: [PATCH 13/17] scsi: push host_lock down into scsi_{host,target}_queue_ready Date: Fri, 07 Feb 2014 11:30:55 -0800 Message-ID: <1391801455.1155.26.camel@haakon3.risingtidesystems.com> References: <20140205123930.150608699@bombadil.infradead.org> <20140205124021.286457268@bombadil.infradead.org> <1391705819.22335.8.camel@dabdike> <52F3C21F.70409@acm.org> <1391723900.14985.3.camel@haakon3.risingtidesystems.com> <52F4B622.1070900@acm.org> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 7bit Return-path: Received: from mail.linux-iscsi.org ([67.23.28.174]:50613 "EHLO linux-iscsi.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752469AbaBGT2R (ORCPT ); Fri, 7 Feb 2014 14:28:17 -0500 In-Reply-To: <52F4B622.1070900@acm.org> Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: Bart Van Assche Cc: James Bottomley , Christoph Hellwig , Jens Axboe , linux-scsi@vger.kernel.org On Fri, 2014-02-07 at 11:32 +0100, Bart Van Assche wrote: > On 02/06/14 22:58, Nicholas A. Bellinger wrote: > > Starting with a baseline using scsi_debug that NOPs REQ_TYPE_FS commands > > to measure improvements would be a better baseline vs. scsi_request_fn() > > existing code that what your doing above. > > > > That way at least it's easier to measure specific scsi-core overhead > > down to the LLD's queuecommand without everything else in the way. > > Running initiator and target workload on the same system makes initiator > and target code share CPUs and hence makes it hard to analyze which > performance change is caused by initiator optimizations and which > performance change is caused by interactions between initiator and > target workloads. > Huh..? There is no 'target workload' involved. All that scsi_debug with NOP'ed REQ_TYPE_FS commands is doing is calling scsi_cmd->done() as soon as the descriptor has been dispatched into LLD ->queuecommand() code. It's useful for determining an absolute performance ceiling between scsi_mq vs. scsi_request_fn() based codepaths without involving any other LLD specific overheads. --nab