Re: [PATCH] pkt_sched: Fix lockdep warning on est_tree_lock in gen_estimator

[Eric Dumazet <eric.dumazet@gmail.com>]

Le jeudi 02 septembre 2010 à 22:55 +0200, Jarek Poplawski a écrit :
> On Thu, Sep 02, 2010 at 10:32:01PM +0200, Eric Dumazet wrote:
> > Le jeudi 02 septembre 2010 ?? 13:22 -0700, David Miller a écrit :
> > > From: Jarek Poplawski <jarkao2@gmail.com>
> > > Date: Thu, 2 Sep 2010 22:12:58 +0200
> > > 
> > > > This patch fixes a lockdep warning:
> > > > 
> > > > [  516.287584] =========================================================
> > > > [  516.288386] [ INFO: possible irq lock inversion dependency detected ]
> > > > [  516.288386] 2.6.35b #7
> > > > [  516.288386] ---------------------------------------------------------
> > > > [  516.288386] swapper/0 just changed the state of lock:
> > > > [  516.288386]  (&qdisc_tx_lock){+.-...}, at: [<c12eacda>] est_timer+0x62/0x1b4
> > > > [  516.288386] but this lock took another, SOFTIRQ-unsafe lock in the past:
> > > > [  516.288386]  (est_tree_lock){+.+...}
> > > > [  516.288386] 
> > > > [  516.288386] and interrupts could create inverse lock ordering between them.
> > > > ...
> > > > 
> > > > So, est_tree_lock needs BH protection because it's taken by
> > > > qdisc_tx_lock, which is used both in BH and process contexts.
> > > > (Full warning with this patch at netdev, 02 Sep 2010.)
> > > > 
> > > > Fixes commit: ae638c47dc040b8def16d05dc6acdd527628f231
> > > > 
> > > > Signed-off-by: Jarek Poplawski <jarkao2@gmail.com>
> > > 
> > > Applied, thanks Jarek.
> > > 
> > > Eric, please make an effort to run with lockdep enabled when making
> > > locking changes, just for fun :-)
> > 
> > Hmm... I still trying to understand the problem.
> > 
> > I dont understand Jarek analysis yet.
> > 
> 
> cpu1:					cpu2:
> (process)				(process)
> qdisc_tx_lock				est_tree_lock
> (waiting for est_tree_lock)		(bh)
> 					(waiting for qdisc_tx_lock)
> 

Cool. Reading commit ae638c47dc040b, I thought only RTNL was possibly
hold in these paths.

Thanks Jarek !