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 S1755004AbcK3LbN (ORCPT <rfc822;w@1wt.eu>);
        Wed, 30 Nov 2016 06:31:13 -0500
Received: from merlin.infradead.org ([205.233.59.134]:44776 "EHLO
        merlin.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
        with ESMTP id S1751050AbcK3LbE (ORCPT
        <rfc822;linux-kernel@vger.kernel.org>);
        Wed, 30 Nov 2016 06:31:04 -0500
Date: Wed, 30 Nov 2016 12:30:49 +0100
From: Peter Zijlstra <peterz@infradead.org>
To: Petr Mladek <pmladek@suse.com>
Cc: linyongting@huawei.com, kejinling@huawei.com,
        akpm@linux-foundation.org, sergey.senozhatsky@gmail.com, bp@suse.de,
        tj@kernel.org, treding@nvidia.com, linux-kernel@vger.kernel.org,
        leisure.wang@huawei.com
Subject: Re: [PATCH] printk: Fix spinlock deadlock in printk reenty
Message-ID: <20161130113049.GN3045@worktop.programming.kicks-ass.net>
References: <1480490119-63559-1-git-send-email-linyongting@huawei.com>
 <20161130105653.GF24060@pathway.suse.cz>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20161130105653.GF24060@pathway.suse.cz>
User-Agent: Mutt/1.5.22.1 (2013-10-16)
Sender: linux-kernel-owner@vger.kernel.org
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org

On Wed, Nov 30, 2016 at 11:56:53AM +0100, Petr Mladek wrote:
> On Wed 2016-11-30 15:15:19, linyongting@huawei.com wrote:
> > In arm, arm64, x86 or other architecture, spinlock variable

x86 no longer uses ticket locks.

> > The solution is that In function zap_locks(), replace
> > raw_spin_lock_init(&logbuf_lock) with raw_spin_unlock(&logbuf_lock),

That's broken too. Imagine the CPU that actually holds the lock then
_also_ doing an unlock. At that point the tail is ahead of the head and
you're also up some creek without no paddle.

Note that I ran into all these scenarios many years ago..

> Another solution would be to make printk() to ignore locks
> when Oops is in progress. It was somewhere suggested by Peter
> Zijlstra. Well, it might cause some problems as well when
> there are more CPUs still running and printing.

Ignoring is the only option. There is no way to fudge the lock state and
live to tell about it.