Date: Fri, 24 Sep 2004 16:26:31 +0200
From: "Oliver M. Bolzer" <oliver@fakeroot.net>
To: linux-kernel@vger.kernel.org
Subject: Re: qla2xxx: frequent total lockups (2.6.8, 2.6.9-rc{1-mm5,2})
Message-ID: <20040924142631.GA15004@magi.fakeroot.net>
Reply-To: linux-kernel@vger.kernel.org
References: <20040915231657.GA2005@magi.fakeroot.net>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20040915231657.GA2005@magi.fakeroot.net>
User-Agent: Mutt/1.5.4i
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1248
Lines: 35


On Thu, Sep 16, 2004 at 01:16:59AM +0200, "Oliver M. Bolzer" <oliver@fakeroot.net> wrote...
 
>As soon as there is I/O load on the HBA, I start seeing
> 
>qla2300 0000:01:03.0: qla2xxx_eh_abort: cmd already done sp=0000000000000000
> 
>messages every few seconds, eventually leading to complete system lockups after
>several minutes up to several hours, due to kernel NULL pointer dereferences
>in qla2x00_cmd_timeout().
 
> I've tested and reproduced the error on the following kernels, all
> compiled for x86_64.
> 2.6.8.1
> 2.6.9-rc1-mm5 (with dma_fixups patch posted by Andrew Vasquez on 13.9)
> 2.6.9-rc2 

2.6.9-rc2-mm2 seems to have fixed the problem.
It is rather puzzling since the difference in the qla2xxx driver between
2.6.9-rc1-mm5+qlogic-oops-fix and 2.6.9-rc2-mm2 is rather minimal. Maybe
the problem was caused by some other part of the kernel that has now
been fixed.


-- 
	Oliver M. Bolzer
	oliver@gol.com

GPG (PGP) Fingerprint = 621B 52F6 2AC1 36DB 8761  018F 8786 87AD EF50 D1FF
-
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/