From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mailhub.sw.ru ([195.214.233.200]:17546 "EHLO relay.sw.ru" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752109AbXGRHWc (ORCPT ); Wed, 18 Jul 2007 03:22:32 -0400 Message-ID: <469DBFA2.7020608@openvz.org> Date: Wed, 18 Jul 2007 11:22:10 +0400 From: Pavel Emelianov MIME-Version: 1.0 Subject: Re: [patch 041/234] Report that kernel is tainted if there was an OOPS References: <200707171103.l6HB3gDf013411@imap1.linux-foundation.org> <20070717181003.GA16147@flint.arm.linux.org.uk> <469DA543.2030601@openvz.org> <20070718071658.GA30270@flint.arm.linux.org.uk> In-Reply-To: <20070718071658.GA30270@flint.arm.linux.org.uk> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Sender: linux-arch-owner@vger.kernel.org To: Russell King Cc: akpm@linux-foundation.org, torvalds@linux-foundation.org, linux-arch@vger.kernel.org, randy.dunlap@oracle.com List-ID: Russell King wrote: > On Wed, Jul 18, 2007 at 09:29:39AM +0400, Pavel Emelianov wrote: >> Russell King wrote: >>> On Tue, Jul 17, 2007 at 04:03:42AM -0700, akpm@linux-foundation.org wrote: >>>> From: Pavel Emelianov >>>> >>>> If the kernel OOPSed or BUGed then it probably should be considered as >>>> tainted. Thus, all subsequent OOPSes and SysRq dumps will report the >>>> tainted kernel. This saves a lot of time explaining oddities in the >>>> calltraces. >>> A bug causes an oops. Oops are counted. So, why do we need this >>> additional complexity when we already have the '#' counter in oops >>> dumps? >>> >>> For instance, on ARM: >>> >>> Unable to handle kernel NULL pointer dereference at virtual address >>> 00000090 >>> pgd = c0004000 >>> [00000090] *pgd=00000000 >>> Internal error: Oops: 817 [#1] >>> ^^^^ >>> This is the oops counter. Anything oops report from anyone other than the >>> first should always be questioned. Also note that this counter is not >>> re-settable at run time, unlike the taint flags. >>> >> Press SysRq-P and you won't see any oops-counters, but just the info that >> the kernel is tainted. This is helpful to know that kernel oopsed when >> observing the SysRq-p output. This is just one of the reasons. > > Maybe it'd make more sense to print the oops counter in places where > the tainted status is printed? > Maybe. But the "tainted" mask is a common way to tell people that the kernel is not clean. Besides, the "D" taint flag and "non-zero" oops counter both mean the same thing, while "D" looks better. Pavel