Date: Thu, 31 Jul 2008 05:44:01 -0700 (PDT)
Message-Id: <20080731.054401.06991829.davem@davemloft.net> (sfid-20080731_144407_498379_33DFEC62)
To: nickpiggin@yahoo.com.au
Cc: jarkao2@gmail.com, johannes@sipsolutions.net, netdev@axxeo.de,
	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: <200807312238.20264.nickpiggin@yahoo.com.au>
References: <20080727203757.GA2527@ami.dom.local>
	<20080731.052932.110299354.davem@davemloft.net>
	<200807312238.20264.nickpiggin@yahoo.com.au>
Mime-Version: 1.0
Content-Type: Text/Plain; charset=us-ascii
Sender: linux-wireless-owner@vger.kernel.org

From: Nick Piggin <nickpiggin@yahoo.com.au>
Date: Thu, 31 Jul 2008 22:38:19 +1000

> Except for the braindead volatile that gets stuck on the bitops pointer.
> 
> Last time I complained about this, a lot of noise was made and I think
> Linus wanted it to stay around so we could pass volatile pointers to
> bitops & co without warnings. I say we should just remove the volatile
> and kill any callers that might warn...

Ho hum... :)

Another way to approach that, and keep the volatile, is to have
a "test_flags()" interface that takes the bit mask of values
you want to test for cases where you know it is a single word
flags value.

The downside is that this kind of interface is easy to use
incorrectly especially when accesses to the same flags use
bot test_bit() and test_flags().