From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: From: Jeff Moyer To: Jens Axboe Cc: Jens Axboe , linux-block@vger.kernel.org, Christoph Hellwig Subject: Re: [PATCH] cfq: priority boost on meta/prio marked IO References: <20160608204347.GA30146@kernel.dk> <57599734.40607@kernel.dk> <5759CE16.905@kernel.dk> Date: Thu, 09 Jun 2016 17:08:02 -0400 In-Reply-To: <5759CE16.905@kernel.dk> (Jens Axboe's message of "Thu, 9 Jun 2016 14:14:14 -0600") Message-ID: MIME-Version: 1.0 Content-Type: text/plain List-ID: Jens Axboe writes: >> I was curious about writes. ;-) Anyway, it's good to validate that the >> read case is also relevant. > > You mean O_DIRECT writes? Most of the buffered writes will come out of > the associated threads, so I don't think it's a big of an issue > there. We've only seen it for reads. Well, you had me confused with your initial report: "... because eg meta data updates..." So I assumed that meant REQ_META WRITES. My bad. [snip] >> Interesting. I would have thought that the cfqd->active_queue would >> have been preempted by a request marked with REQ_PRIO. But you're >> suggesting that did not happen? [snip] > We seem to handily mostly ignore prio_pending for the idle class. If Right, I forgot we were talking about idle class. Sorry. > the new queue is idle, then we don't look at prio pending. I'd rather > make this more explicit, the patch is pretty similar to what we had in > the past. It's somewhat of a regression caused by commit 4aede84b33d, > except I like using the request flags for this a lot more than the old > current->fs_excl. REQ_PRIO should always be set for cases where we > hold fs (or even directory) specific resources. Ah, thanks for the reference! Now I'll go back and finish reviewing the actual patch. -Jeff