From mboxrd@z Thu Jan  1 00:00:00 1970
From: "Michael S. Tsirkin" <mst@redhat.com>
Subject: Re: [PATCH UPDATED 1/3] vhost: replace vhost_workqueue with
 per-vhost kthread
Date: Mon, 26 Jul 2010 22:57:14 +0300
Message-ID: <20100726195714.GD27644@redhat.com>
References: <4C48B664.9000109@kernel.org>
 <20100724191447.GA4972@redhat.com>
 <4C4BEAA2.6040301@kernel.org>
 <20100726152510.GA26223@redhat.com>
 <4C4DAB14.5050809@kernel.org>
 <20100726155014.GA26412@redhat.com>
 <4C4DB247.9060709@kernel.org>
 <4C4DB466.6000409@kernel.org>
 <20100726163108.GE26412@redhat.com>
 <4C4DD946.2080502@kernel.org>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Cc: Oleg Nesterov <oleg@redhat.com>,
	Sridhar Samudrala <sri@us.ibm.com>,
	netdev <netdev@vger.kernel.org>,
	lkml <linux-kernel@vger.kernel.org>,
	"kvm@vger.kernel.org" <kvm@vger.kernel.org>,
	Andrew Morton <akpm@linux-foundation.org>,
	Dmitri Vorobiev <dmitri.vorobiev@movial.com>,
	Jiri Kosina <jkosina@suse.cz>,
	Thomas Gleixner <tglx@linutronix.de>,
	Ingo Molnar <mingo@elte.hu>, Andi Kleen <ak@linux.intel.com>
To: Tejun Heo <tj@kernel.org>
Return-path: <linux-kernel-owner@vger.kernel.org>
Content-Disposition: inline
In-Reply-To: <4C4DD946.2080502@kernel.org>
Sender: linux-kernel-owner@vger.kernel.org
List-Id: netdev.vger.kernel.org

On Mon, Jul 26, 2010 at 08:51:50PM +0200, Tejun Heo wrote:
> Hello,
> 
> On 07/26/2010 06:31 PM, Michael S. Tsirkin wrote:
> >> On 07/26/2010 06:05 PM, Tejun Heo wrote:
> >>> * Placing try_to_freeze() could be a bit annoying.  It shouldn't be
> >>>   executed when there's a work to flush.
> > 
> > BTW why is this important?
> > We could always get another work and flush right after
> > try_to_freeze, and then flush would block for a long time.
> > 
> > BTW the vhost patch you sent does not do this at all.
> > I am guessing it is because our thread is not freezable?
> 
> Yeap, I think so.
> 
> >> * Similar issue exists for kthread_stop().  The kthread shouldn't exit
> >>   while there's a work to flush (please note that kthread_worker
> >>   interface allows detaching / attaching worker kthread during
> >>   operation, so it should remain in consistent state with regard to
> >>   flushing).
> > 
> > Not sure I agree here. Users must synchronise flush and stop calls.
> > Otherwise a work might get queued after stop is called, and
> > you won't be able to flush it.
> 
> For freeze, it probably is okay but for stop, I think it's better to
> keep the semantics straight forward.

What are the semantics then? What do we want stop followed
by queue and flush to do?

>  It may be okay to do otherwise
> but having such oddity in generic interface is nasty and may lead to
> surprises which can be pretty difficult to track down later on.  It's
> just a bit more of annoyance while writing the generic code, so...
> 
> Thanks.
> 
> -- 
> tejun