From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751208AbdJCIRQ (ORCPT ); Tue, 3 Oct 2017 04:17:16 -0400 Received: from mx1.redhat.com ([209.132.183.28]:11581 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751131AbdJCIRN (ORCPT ); Tue, 3 Oct 2017 04:17:13 -0400 DMARC-Filter: OpenDMARC Filter v1.3.2 mx1.redhat.com A51CB356F5 Authentication-Results: ext-mx06.extmail.prod.ext.phx2.redhat.com; dmarc=none (p=none dis=none) header.from=redhat.com Authentication-Results: ext-mx06.extmail.prod.ext.phx2.redhat.com; spf=fail smtp.mailfrom=ming.lei@redhat.com Date: Tue, 3 Oct 2017 16:17:03 +0800 From: Ming Lei To: Christoph Hellwig Cc: Jens Axboe , linux-block@vger.kernel.org, linux-scsi@vger.kernel.org, "Martin K . Petersen" , "James E . J . Bottomley" , Bart Van Assche , Oleksandr Natalenko , Johannes Thumshirn , Cathy Avery , Martin Steigerwald , linux-kernel@vger.kernel.org, Hannes Reinecke , stable@vger.kernel.org, Bart Van Assche Subject: Re: [PATCH V7 6/6] SCSI: set block queue at preempt only when SCSI device is put into quiesce Message-ID: <20171003081702.GB23564@ming.t460p> References: <20170930061214.10622-1-ming.lei@redhat.com> <20170930061214.10622-7-ming.lei@redhat.com> <20171002135225.GF19119@infradead.org> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20171002135225.GF19119@infradead.org> User-Agent: Mutt/1.8.3 (2017-05-23) X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.5.16 (mx1.redhat.com [10.5.110.30]); Tue, 03 Oct 2017 08:17:13 +0000 (UTC) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Mon, Oct 02, 2017 at 06:52:25AM -0700, Christoph Hellwig wrote: > > + /* > > + * Simply quiesing SCSI device isn't safe, it is easy > > + * to use up requests because all these allocated requests > > + * can't be dispatched when device is put in QIUESCE. > > + * Then no request can be allocated and we may hang > > + * somewhere, such as system suspend/resume. > > + * > > + * So we set block queue in preempt only first, no new > > + * normal request can enter queue any more, and all pending > > + * requests are drained once blk_set_preempt_only() > > + * returns. Only RQF_PREEMPT is allowed in preempt only mode. > > + */ > > + blk_set_preempt_only(sdev->request_queue, true); > > + > > mutex_lock(&sdev->state_mutex); > > err = scsi_device_set_state(sdev, SDEV_QUIESCE); > > Why is this not under state_mutex so that we guarantee it's atomic > vs sdev state changes? It isn't necessary since what we want is to make sure only RQF_PREEMPT is allowed once blk_set_preempt_only() returns, and putting a lock before freezing queue might risk to deadlock. -- Ming