Date: Wed, 23 Jul 2008 13:55:30 -0700 (PDT)
Message-Id: <20080723.135530.63039342.davem@davemloft.net> (sfid-20080723_225534_324707_1666A25D)
To: jarkao2@gmail.com
Cc: peterz@infradead.org, Larry.Finger@lwfinger.net, kaber@trash.net,
	torvalds@linux-foundation.org, akpm@linux-foundation.org,
	netdev@vger.kernel.org, linux-kernel@vger.kernel.org,
	linux-wireless@vger.kernel.org, mingo@redhat.com
Subject: Re: Kernel WARNING: at net/core/dev.c:1330
 __netif_schedule+0x2c/0x98()
From: David Miller <davem@davemloft.net>
In-Reply-To: <20080723204335.GA14788@ami.dom.local>
References: <20080723114914.GF4561@ff.dom.local>
	<20080723.131607.79681752.davem@davemloft.net>
	<20080723204335.GA14788@ami.dom.local>
Mime-Version: 1.0
Content-Type: Text/Plain; charset=us-ascii
Sender: linux-wireless-owner@vger.kernel.org

From: Jarek Poplawski <jarkao2@gmail.com>
Date: Wed, 23 Jul 2008 22:43:35 +0200

> On Wed, Jul 23, 2008 at 01:16:07PM -0700, David Miller wrote:
> ...
> > There will always be a need for a "stop all the TX queues" operation.
> 
> The question is if the current way is "all correct". As a matter of
> fact I think Peter's doubts could be justified: taking "USHORT_MAX"
> locks looks really dubious (so maybe it's not so strange lockdep
> didn't get used to this).

There are, of course, potentially other ways to achieve the objective.

And for non-multiqueue aware devices (which is the vast majority of
the 400 or so networking drivers we have) there is only one queue and
thus one lock taken.