From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1761135AbXISPiT (ORCPT ); Wed, 19 Sep 2007 11:38:19 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1760790AbXISPhz (ORCPT ); Wed, 19 Sep 2007 11:37:55 -0400 Received: from e5.ny.us.ibm.com ([32.97.182.145]:43126 "EHLO e5.ny.us.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1760729AbXISPhx (ORCPT ); Wed, 19 Sep 2007 11:37:53 -0400 Date: Wed, 19 Sep 2007 08:37:50 -0700 From: "Paul E. McKenney" To: Dmitry Torokhov Cc: Peter Zijlstra , linux-kernel@vger.kernel.org, Ingo Molnar , Andrew Morton , Nick Piggin Subject: Re: [RFC][PATCH 2/6] lockdep: validate rcu_dereference() vs rcu_read_lock() Message-ID: <20070919153750.GB8666@linux.vnet.ibm.com> Reply-To: paulmck@linux.vnet.ibm.com References: <20070919104125.286538000@chello.nl> <20070919105054.332308000@chello.nl> <20070919163111.05c18021@twins> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.13 (2006-08-11) Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Wed, Sep 19, 2007 at 11:16:21AM -0400, Dmitry Torokhov wrote: > On 9/19/07, Peter Zijlstra wrote: > > On Wed, 19 Sep 2007 10:17:25 -0400 "Dmitry Torokhov" > > wrote: > > > > > Hi Peter, > > > > > > On 9/19/07, Peter Zijlstra wrote: > > > > Warn when rcu_dereference() is not used in combination with rcu_read_lock() > > > > > > > > > > According to Paul it is fine to use RCU primitives (when accompanied > > > with proper comments) when the read-size critical section is guarded > > > by spin_lock_irqsave()/spin_lock_irqsrestore() instead of > > > rcu_read_lock()/rcu_read_unlock() and writers synchronize with > > > synchronize_sched(), not synchronize_rcu(). Your patch will trigger > > > warnign on such valid usages. > > > > > > > Sounds fragile to begin with. But you're right in that that is valid > > for Linux as you know it. However in -rt most/all spinlocks are > > converted to sleeping locks. In that case sync_sched() is not enough. > > OK, then it goes beyond RCU... We need to come up with something that > can be used to synchronize with IRQ handlers (quite often in driver > code one needs to be sure that current invocation of IRQ handler > completed before doing something). And once we have it splinlock + RCU > users can just use that method. But Peter's approach would not cause a problem here -- you wouldn't be doing an rcu_dereference from within the IRQ handler in this case, right? That said, we will need something to handle threaded interrupts, since synchronize_sched() only waits for hardirq, NMI, SMI, etc., and not threaded IRQs. Thanx, Paul