From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail.lixom.net (lixom.net [66.141.50.11]) by ozlabs.org (Postfix) with ESMTP id 8F0BBDDE31 for ; Mon, 5 Feb 2007 15:11:54 +1100 (EST) Date: Sun, 4 Feb 2007 22:19:19 -0600 To: Benjamin Herrenschmidt Subject: Re: [PATCH] [3/9] pasemi: Machine check handler Message-ID: <20070205041919.GA4719@lixom.net> References: <20070202045448.145851000@lixom.net> <20070202045508.119282000@lixom.net>> <1170632998.2620.50.camel@localhost.localdomain> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii In-Reply-To: <1170632998.2620.50.camel@localhost.localdomain> From: olof@lixom.net (Olof Johansson) Cc: linuxppc-dev@ozlabs.org, Paul Mackerras List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , On Mon, Feb 05, 2007 at 10:49:57AM +1100, Benjamin Herrenschmidt wrote: > On Thu, 2007-02-01 at 22:54 -0600, Olof Johansson wrote: > > plain text document attachment (pasemi-machine-check) > > Print out decoded machine check information on PA6T. > > > > > > Signed-off-by: Olof Johansson > > Not specific to your patch, but in general, Machine Check might happen > while MSR:EE is off... which means the kernel might dislike printk being > called from the MC handler under some circumstances (for example, > re-entering printk itself in places it doesn't expect to be). > > Might be worth, when the MC is recoverable, to try to "defer" the > printout to some real irq or timer irq Right. There are some other fixups to be done too that I will look at later, for example all machine checks are still passed through a 32-bit decoder after the ppc_md one. It can be quite confusing and should probably be moved out to ppc_machdep handlers on all platforms instead. I'll try to have some cleanup of the generic code for 2.6.22. Meanwhile I think this can go in. -Olof