Date: Fri, 4 Dec 2015 17:51:12 +0100
From: Borislav Petkov <bp@alien8.de>
To: "Raj, Ashok" <ashok.raj@intel.com>
Cc: linux-kernel@vger.kernel.org, linux-edac@vger.kernel.org,
        Tony Luck <tony.luck@intel.com>
Subject: Re: [Patch V0] x86, mce: Ensure offline CPU's don't participate in
 mce rendezvous process.
Message-ID: <20151204165112.GI21177@pd.tnic>
References: <1449188170-3909-1-git-send-email-ashok.raj@intel.com>
 <20151204143404.GF21177@pd.tnic>
 <20151204171419.GA4870@otc-brkl-03.jf.intel.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <20151204171419.GA4870@otc-brkl-03.jf.intel.com>
User-Agent: Mutt/1.5.23 (2014-03-12)
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1866
Lines: 46

On Fri, Dec 04, 2015 at 12:14:20PM -0500, Raj, Ashok wrote:
> Yes, thats possible to not do ist_enter() and the exception count. 
> 
> I tried to keep most of the part as is and leveraging code already
> doing the reading of MCG_STATUS. Architecturally we need to also check RIPV
> and if clear we should initiate shutdown.

So add that check too.

> When we add the logging from offline cpus as next step it would be safe to 
> use interrupt stack, and the offline

Franky, I'm not sure at all and very very wary of adding *any* code
which runs on an offlined CPU. Because *no one* does that and it hasn't
been tested at all. So who knows what happens.

What we should be doing is execute the *minimal* amount of code possible
and get out. No counting, no per-cpu variables. No nothing.

> I liked the observability part keeping the exception count. if and
> when we online the cpu again, it might look as it noticed nothing. Now
> we can check /proc/interrupts and see the offline cpu also observed
> the MCE.

And? Tell us what? That SMM fondled the hardware under our feet. TBH,
I'd tend to be much more drastic here and even taint the kernel. I mean,
seriously, what kind of MCEs which happen as a result of OS execution
are you expecting to get reported on an offlined CPU?

I can't think of very any.

Because we have been considering offlining a core as one possible RAS
action. So what happens is a user or a RAS agent offlines a core and
yet, that offlined core still reports MCEs. Something's terribly wrong
with that picture, IMO.

-- 
Regards/Gruss,
    Boris.

ECO tip #101: Trim your mails when you reply.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/