Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752044AbXAXRYu (ORCPT ); Wed, 24 Jan 2007 12:24:50 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1752049AbXAXRYt (ORCPT ); Wed, 24 Jan 2007 12:24:49 -0500 Received: from tomts10-srv.bellnexxia.net ([209.226.175.54]:40257 "EHLO tomts10-srv.bellnexxia.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752044AbXAXRYs (ORCPT ); Wed, 24 Jan 2007 12:24:48 -0500 Date: Wed, 24 Jan 2007 12:24:45 -0500 From: Mathieu Desnoyers To: Andrew Morton Cc: Greg Kroah-Hartman , Richard J Moore , linux-kernel@vger.kernel.org, "Martin J. Bligh" , Christoph Hellwig , Douglas Niehaus , Ingo Molnar , ltt-dev@shafik.org, systemtap@sources.redhat.com, Thomas Gleixner Subject: [PATCH] order of lockdep off/on in vprintk() should be changed Message-ID: <20070124172445.GA1145@Krystal> References: <20061220235216.GA28643@Krystal> <20070116175624.GA16022@Krystal> <20070123202637.970e467b.akpm@osdl.org> <20070124165150.GC4979@Krystal> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Content-Disposition: inline In-Reply-To: <20070124165150.GC4979@Krystal> X-Editor: vi X-Info: http://krystal.dyndns.org:8080 X-Operating-System: Linux/2.4.32-grsec (i686) X-Uptime: 12:16:56 up 154 days, 14:24, 4 users, load average: 0.70, 0.74, 0.74 User-Agent: Mutt/1.5.13 (2006-08-11) Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 2473 Lines: 84 The order of locking between lockdep_off/on() and local_irq_save/restore() in vprintk() should be changed. * In kernel/printk.c : vprintk() does : preempt_disable() local_irq_save() lockdep_off() spin_lock(&logbuf_lock) spin_unlock(&logbuf_lock) if(!down_trylock(&console_sem)) up(&console_sem) lockdep_on() local_irq_restore() preempt_enable() The goals here is to make sure we do not call printk() recursively from kernel/lockdep.c:__lock_acquire() (called from spin_* and down/up) nor from kernel/lockdep.c:trace_hardirqs_on/off() (called from local_irq_restore/save). It can then potentially call printk() through mark_held_locks/mark_lock. It correctly protects against the spin_lock call and the up/down call, but it does not protect against local_irq_restore. It could cause infinite recursive printk/trace_hardirqs_on() calls when printk() is called from the mark_lock() error handing path. We should change the locking so it becomes correct : preempt_disable() lockdep_off() local_irq_save() spin_lock(&logbuf_lock) spin_unlock(&logbuf_lock) if(!down_trylock(&console_sem)) up(&console_sem) local_irq_restore() lockdep_on() preempt_enable() Signed-off-by: Mathieu Desnoyers --- a/kernel/printk.c +++ b/kernel/printk.c @@ -530,8 +530,8 @@ asmlinkage int vprintk(const char *fmt, va_list args) zap_locks(); /* This stops the holder of console_sem just where we want him */ - local_irq_save(flags); lockdep_off(); + local_irq_save(flags); spin_lock(&logbuf_lock); printk_cpu = smp_processor_id(); @@ -640,8 +640,8 @@ asmlinkage int vprintk(const char *fmt, va_list args) console_locked = 0; up(&console_sem); } - lockdep_on(); local_irq_restore(flags); + lockdep_on(); } else { /* * Someone else owns the drivers. We drop the spinlock, which @@ -650,8 +650,8 @@ asmlinkage int vprintk(const char *fmt, va_list args) */ printk_cpu = UINT_MAX; spin_unlock(&logbuf_lock); - lockdep_on(); local_irq_restore(flags); + lockdep_on(); } preempt_enable(); -- OpenPGP public key: http://krystal.dyndns.org:8080/key/compudj.gpg Key fingerprint: 8CD5 52C3 8E3C 4140 715F BA06 3F25 A8FE 3BAE 9A68 - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/