From: "Luck, Tony" <tony.luck@intel.com>
To: Chen Gong <gong.chen@linux.intel.com>
CC: "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
        Ingo Molnar <mingo@elte.hu>, Borislav Petkov <bp@amd64.org>,
        "Huang, Ying" <ying.huang@intel.com>,
        Hidetoshi Seto <seto.hidetoshi@jp.fujitsu.com>
Subject: RE: [PATCH 2/2] x86/mce: Add instruction recovery signatures to
 mce-severity table
Thread-Topic: [PATCH 2/2] x86/mce: Add instruction recovery signatures to
 mce-severity table
Thread-Index: AQHNL0lNX6PGPdBkbUOeJPxXhK/15JbE0YwA
Date: Fri, 11 May 2012 17:42:09 +0000
Message-ID: <3908561D78D1C84285E8C5FCA982C28F192ED617@ORSMSX104.amr.corp.intel.com>
References: <cover.1336674796.git.tony.luck@intel.com>
 <0e89440bf6444635bde0b876eedd26e5c51f9ca3.1336674796.git.tony.luck@intel.com>
 <4FACC25B.80305@linux.intel.com>
In-Reply-To: <4FACC25B.80305@linux.intel.com>
Accept-Language: en-US
Content-Language: en-US
Content-Type: text/plain; charset="utf-8"
MIME-Version: 1.0
Sender: linux-kernel-owner@vger.kernel.org
Content-Transfer-Encoding: 8bit
Content-Length: 929
Lines: 21

> For IFU, on affected logical processors, RIPV and EIPV both are 0,
> since now new IFU entries are added into severity table, the old
> entry as below should be removed:
>
> /* Neither return not error IP -- no chance to recover -> PANIC */
>        MCESEV(
>                PANIC, "Neither restart nor error IP",
>                MCGMASK(MCG_STATUS_RIPV|MCG_STATUS_EIPV, 0)
>                ),

We need to keep this. If EIPV is not set, then CS and IP on the
stack are not guaranteed ... so we can't tell whether the error
happened in user or kernel mode. This makes recovery "challenging".

I'm trying to figure out a quirk for processors that do generate
EIPV=RIPV=0 signature for IFU errors. There are some case where
we can work around the lack of EIPV.

-Tony

????{.n?+???????+%?????ݶ??w??{.n?+????{??G?????{ay?ʇڙ?,j??f???h?????????z_??(?階?ݢj"???m??????G????????????&???~???iO???z??v?^?m????????????I?