Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S262564AbTKRN0K (ORCPT ); Tue, 18 Nov 2003 08:26:10 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S262603AbTKRN0K (ORCPT ); Tue, 18 Nov 2003 08:26:10 -0500 Received: from gprs147-139.eurotel.cz ([160.218.147.139]:16257 "EHLO amd.ucw.cz") by vger.kernel.org with ESMTP id S262564AbTKRN0G (ORCPT ); Tue, 18 Nov 2003 08:26:06 -0500 Date: Tue, 18 Nov 2003 14:26:34 +0100 From: Pavel Machek To: Jens Axboe Cc: Guillaume Chazarain , Linux Kernel Subject: Re: [PATCH] cfq + io priorities Message-ID: <20031118132634.GB470@elf.ucw.cz> References: <20031109113928.GN2831@suse.de> <20031113125427.GB643@openzaurus.ucw.cz> <20031117081407.GI888@suse.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20031117081407.GI888@suse.de> X-Warning: Reading this can be dangerous to your mental health. User-Agent: Mutt/1.5.4i Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 1470 Lines: 40 Hi! > > OTOH it might make sense to make "nice" command set > > both by default. > > Yes, I can probably be talked into that. Good. > > > > > these end values are "special" - 0 means the process is only allowed to > > > > > do io if the disk is idle, and 20 means the process io is considered > > > > > > > > So a process with ioprio == 0 can be forever starved. As it's not > > > > > > Yes > > > > If semaphore is held over disk io somewhere (quota code? journaling?) > > you have ugly possibility of priority inversion there. > > Indeed yes. That's a general problem with all the io priorities though, > RT io might end up waiting for nice 10 io etc. Dunno what to do about > this yet... Well, traditional (== scheduler) solution is not to have idle classes and not guarantee anything about realtime classes. At least idle class can not be used to hold important semaphore forever (even low-priority prosses receive enough time not to hold important semaphores too long)... I believe you should do the same (== get rid of idle class for now, and clearly state that realtime ones are not _guaranteed_ anything). Pavel -- When do you have a heart between your knees? [Johanka's followup: and *two* hearts?] - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/