From mboxrd@z Thu Jan  1 00:00:00 1970
From: Oleg Nesterov <oleg@tv-sign.ru>
Subject: Re: [PATCH] RT: Add priority-queuing and priority-inheritance to workqueue infrastructure
Date: Mon, 6 Aug 2007 19:38:16 +0400
Message-ID: <20070806153816.GA265@tv-sign.ru>
References: <1186005598.9513.261.camel@ghaskins-t60p.haskins.net> <20070801222201.GA316@tv-sign.ru> <1186012439.9513.321.camel@ghaskins-t60p.haskins.net> <20070802195049.GA361@tv-sign.ru> <20070806114954.GC1903@elte.hu> <20070806131814.GC91@tv-sign.ru> <1186406963.7182.13.camel@twins> <1186407173.7182.16.camel@twins> <20070806144536.GA232@tv-sign.ru> <1186412689.21381.91.camel@ghaskins-t60p.haskins.net>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Cc: Peter Zijlstra <peterz@infradead.org>, Ingo Molnar <mingo@elte.hu>,
	Daniel Walker <dwalker@mvista.com>,
	linux-rt-users@vger.kernel.org, linux-kernel@vger.kernel.org
To: Gregory Haskins <ghaskins@novell.com>
Return-path: <linux-rt-users-owner@vger.kernel.org>
Received: from mail.screens.ru ([213.234.233.54]:48402 "EHLO mail.screens.ru"
	rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP
	id S1754187AbXHFPfy (ORCPT <rfc822;linux-rt-users@vger.kernel.org>);
	Mon, 6 Aug 2007 11:35:54 -0400
Content-Disposition: inline
In-Reply-To: <1186412689.21381.91.camel@ghaskins-t60p.haskins.net>
Sender: linux-rt-users-owner@vger.kernel.org
List-Id: linux-rt-users.vger.kernel.org

On 08/06, Gregory Haskins wrote:
>
> On Mon, 2007-08-06 at 18:45 +0400, Oleg Nesterov wrote:
> > On 08/06, Peter Zijlstra wrote:
> 
> > > 
> > > still this does not change the fundamental issue of a high prio piece of
> > > work waiting on a lower prio task.
> >        ^^^^^^^
> > waiting. This is a "key" word, and this was my (perhaps wrong) point.
> 
> Actually, I think Peter is making a really important point here.

Yes. Please see another email I just sent.

> "Waiting" can be defined in more ways than the REQUEST/RESPONSE pattern
> that I have been rambling about.
> 
> Using Peters NIC vs USB example:  What if a NIC driver is using a
> workqueue as a bottom-half mechanism for its RX packet queuing.  In a
> nice RT environment it would be highly ideal if we allow the deferred
> work to complete with respect to the priority that was assigned to the
> subsystem.
> 
> So while the submitter isn't technically blocking on the work, the
> application that is receiving packets is now subject to the arbitrary
> priority of the keventd as opposed to the NIC irq.  Thus there is still
> "waiting" being subject to inversion, its just not in a REQUEST/RESPONSE
> pattern.

Oleg.