From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <linux-kernel-owner@vger.kernel.org>
Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand
	id S1753486AbbGaRW4 (ORCPT <rfc822;w@1wt.eu>);
	Fri, 31 Jul 2015 13:22:56 -0400
Received: from bombadil.infradead.org ([198.137.202.9]:50978 "EHLO
	bombadil.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
	with ESMTP id S1752409AbbGaRWz (ORCPT
	<rfc822;linux-kernel@vger.kernel.org>);
	Fri, 31 Jul 2015 13:22:55 -0400
Date: Fri, 31 Jul 2015 19:22:46 +0200
From: Peter Zijlstra <peterz@infradead.org>
To: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: linux-kernel@vger.kernel.org, Rusty Russell <rusty@rustcorp.com.au>,
        Ananth N Mavinakayanahalli <ananth@in.ibm.com>,
        Anil S Keshavamurthy <anil.s.keshavamurthy@intel.com>,
        tglx@linutronix.de
Subject: Re: [RFC 0/4] use rcu_read_lock() during module list walk
Message-ID: <20150731172246.GB25159@twins.programming.kicks-ass.net>
References: <1438362488-29857-1-git-send-email-bigeasy@linutronix.de>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <1438362488-29857-1-git-send-email-bigeasy@linutronix.de>
User-Agent: Mutt/1.5.21 (2012-12-30)
Sender: linux-kernel-owner@vger.kernel.org
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org

On Fri, Jul 31, 2015 at 07:08:04PM +0200, Sebastian Andrzej Siewior wrote:
> Hi Peter,
> 
> this series was made before I noticed that you introduced a RB tree for
> lookup but the old way still remains under !CONFIG_MODULES_TREE_LOOKUP.

> In the old way the caller had preempt_disable() while invoking
> list_for_each_safe_rcu() which is (according to the RCU checklist) not a
> substitute for rcu_readlock().

This is true. preempt_disable() != rcu_read_lock().

> With your CONFIG_MODULES_TREE_LOOKUP I fail to understand what blocks
> free_module() until all mod_find() callers have dropped their refrence to
> the obtained struct mod. We had synchronize_sched() in RCU case.

We still have synchronize_sched(), but that is not RCU, that is
RCU-sched, and rcu_read_lock_sched() == preempt_disable() (+- some
debugging bits).

The code used to be broken in this regard, see 0be964be0d45 ("module:
Sanitize RCU usage and locking"), that fixed things to be consistent.