From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Date: Tue, 17 Jan 2017 12:53:13 -0500 From: Keith Busch To: Bart Van Assche Cc: "hch@lst.de" , "tglx@linutronix.de" , "linux-nvme@lists.infradead.org" , "linux-block@vger.kernel.org" , "axboe@fb.com" , "axboe@kernel.dk" , "marc@merlins.org" Subject: Re: [PATCH 5/6] blk-mq: Fix queue freeze deadlock Message-ID: <20170117175312.GA8989@localhost.localdomain> References: <1483569671-1462-1-git-send-email-keith.busch@intel.com> <1483569671-1462-6-git-send-email-keith.busch@intel.com> <1483601585.2811.5.camel@sandisk.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii In-Reply-To: <1483601585.2811.5.camel@sandisk.com> List-ID: On Thu, Jan 05, 2017 at 07:33:22AM +0000, Bart Van Assche wrote: > This patch adds a second code path to the blk-mq core for running queues and > hence will make the blk-mq core harder to maintain. Have you considered to > implement this functionality by introducing a new "fail all requests" flag > for hctx queues such that blk_mq_abandon_stopped_requests() can reuse the > existing mechanism for running a queue? Thanks for the suggestion. I'll look into that. I wanted to avoid more flags to test for in the fast-path, but I see that a special queue run method is problematic for maintenance.