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 S1753592AbbGKJhQ (ORCPT <rfc822;w@1wt.eu>);
	Sat, 11 Jul 2015 05:37:16 -0400
Received: from mail-wg0-f52.google.com ([74.125.82.52]:36666 "EHLO
	mail-wg0-f52.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
	with ESMTP id S1753374AbbGKJhO (ORCPT
	<rfc822;linux-kernel@vger.kernel.org>);
	Sat, 11 Jul 2015 05:37:14 -0400
Date: Sat, 11 Jul 2015 11:37:09 +0200
From: Ingo Molnar <mingo@kernel.org>
To: Thomas Gleixner <tglx@linutronix.de>
Cc: Dave Jones <davej@codemonkey.org.uk>, x86@kernel.org,
        Linus Torvalds <torvalds@linux-foundation.org>,
        Linux Kernel <linux-kernel@vger.kernel.org>
Subject: Re: CONFIG_DEBUG_VIRTUAL instant reboot.
Message-ID: <20150711093709.GA27975@gmail.com>
References: <20150710145047.GA16804@codemonkey.org.uk>
 <alpine.DEB.2.11.1507111009530.17353@nanos>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <alpine.DEB.2.11.1507111009530.17353@nanos>
User-Agent: Mutt/1.5.23 (2014-03-12)
Sender: linux-kernel-owner@vger.kernel.org
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org


* Thomas Gleixner <tglx@linutronix.de> wrote:

> On Fri, 10 Jul 2015, Dave Jones wrote:
> > I bought a Haswell Xeon E3-1225 v3 box, and found that
> > my self-built kernel would instantly reboot the machine once
> > grub had loaded it.
> > 
> > Through trial and error I found the following options matter..
> > 
> > if I _enable_ DEBUG_PAGEALLOC, it boots fine.
> > disabling it again, instant death.
> > 
> > further narrowing down revealed that if I have DEBUG_PAGEALLOC off,
> > I also _must_ also have CONFIG_DEBUG_VIRTUAL turned off.
> > With both disabled, it also boots fine.
> > 
> > I tried earlyprintk, but this happens so soon I don't think
> > we're even at the point of initializing that code.
> > 
> > Any ideas how I can further narrow this down ?
> 
> I assume that VIRTUAL_BUG_ON() triggers very early. Changing it to
> WARN_ON() should keep the machine alive and give you a hint which of
> those checks explodes.

I suspect an early console might be useful as well, in case the kernel doesn't 
manage to boot.

> --- a/include/linux/mmdebug.h
> +++ b/include/linux/mmdebug.h
> @@ -50,7 +50,7 @@ void dump_mm(const struct mm_struct *mm);
>  #endif
>  
>  #ifdef CONFIG_DEBUG_VIRTUAL
> -#define VIRTUAL_BUG_ON(cond) BUG_ON(cond)
> +#define VIRTUAL_BUG_ON(cond) WARN_ON(cond)

I'd make this WARN_ON_ONCE(), to see the first instance and to cross fingers 
afterwards and maybe boot up.

Thanks,

	Ingo