From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754876Ab0IVOad (ORCPT ); Wed, 22 Sep 2010 10:30:33 -0400 Received: from mx1.redhat.com ([209.132.183.28]:48694 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754643Ab0IVOac (ORCPT ); Wed, 22 Sep 2010 10:30:32 -0400 Date: Wed, 22 Sep 2010 10:29:31 -0400 From: Jason Baron To: Mathieu Desnoyers Cc: Andi Kleen , rostedt@goodmis.org, linux-kernel@vger.kernel.org, mingo@elte.hu, hpa@zytor.com, tglx@linutronix.de, roland@redhat.com, rth@redhat.com, mhiramat@redhat.com, fweisbec@gmail.com, avi@redhat.com, davem@davemloft.net, vgoyal@redhat.com, sam@ravnborg.org, tony@bakeyournoodle.com, Andi Kleen Subject: Re: [PATCH 1/2] Add for_each_module iterator function v2 Message-ID: <20100922142931.GB2816@redhat.com> References: <1285162001-9556-1-git-send-email-andi@firstfloor.org> <20100922141432.GB30730@Krystal> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20100922141432.GB30730@Krystal> User-Agent: Mutt/1.5.18 (2008-05-17) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Wed, Sep 22, 2010 at 10:14:32AM -0400, Mathieu Desnoyers wrote: > * Andi Kleen (andi@firstfloor.org) wrote: > > From: Andi Kleen > > > > This is a generic function to iterate over all modules. > > To be used in the next patch. > > > > v2: Use module_mutex instead of preempt disabling > > > > Signed-off-by: Andi Kleen > > --- > > include/linux/module.h | 1 + > > kernel/module.c | 10 ++++++++++ > > 2 files changed, 11 insertions(+), 0 deletions(-) > > > > diff --git a/include/linux/module.h b/include/linux/module.h > > index 403ac26..809b6db 100644 > > --- a/include/linux/module.h > > +++ b/include/linux/module.h > > @@ -410,6 +410,7 @@ struct module *__module_address(unsigned long addr); > > bool is_module_address(unsigned long addr); > > bool is_module_percpu_address(unsigned long addr); > > bool is_module_text_address(unsigned long addr); > > +void for_each_module(void (*op)(struct module *, void *arg), void *arg); > > > > static inline int within_module_core(unsigned long addr, struct module *mod) > > { > > diff --git a/kernel/module.c b/kernel/module.c > > index eba1341..9e2a561 100644 > > --- a/kernel/module.c > > +++ b/kernel/module.c > > @@ -384,6 +384,16 @@ struct module *find_module(const char *name) > > } > > EXPORT_SYMBOL_GPL(find_module); > > > > +void for_each_module(void (*op)(struct module *, void *arg), void *arg) > > +{ > > + struct module *mod; > > + > > + mutex_lock(&module_mutex); > > I'm concerned about the following mutex dependency chain: > > module_mutex -> jump_label_mutex (on module load) with the new code, this path doesn't exist. No state needs to be updated at this point. > jump_label_mutex -> module_mutex (on jump_label_update) > > Actually, I think your module notifier is missing in your version. This > is why you don't see this dependency problem. > the jump label design in stateless. When a new module is loaded the module notifier in the tracepoint call will call jump_label_update via enable_jump_label(). Since the module is on the the module_list before the notifier, I think this is ok. thanks, -Jason