From mboxrd@z Thu Jan 1 00:00:00 1970 From: Stephen Hemminger Subject: Re: Kernel oops on setting sky2 interfaces down Date: Tue, 4 Aug 2009 16:08:05 -0700 Message-ID: <20090804160805.2a63699d@nehalam> References: <4A65EC3F.4050400@gibraltar.at> <20090723102848.00a56ad1@nehalam> <4A6D8975.4050000@gibraltar.at> <20090727153548.7c0d9f85@nehalam> <4A76D036.6090705@gibraltar.at> <4A772A1D.1030904@mayrhofer.eu.org> <4A77E56B.9030804@gibraltar.at> <392fb48f0908040445pc21105bo3182773b76d49596@mail.gmail.com> <4A78BC48.4060200@gibraltar.at> <4A78BD5F.2030901@mayrhofer.eu.org> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Cc: Mike McCormack , netdev@vger.kernel.org, Richard Leitner To: Rene Mayrhofer Return-path: Received: from mail.vyatta.com ([76.74.103.46]:42840 "EHLO mail.vyatta.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753569AbZHDXIK (ORCPT ); Tue, 4 Aug 2009 19:08:10 -0400 In-Reply-To: <4A78BD5F.2030901@mayrhofer.eu.org> Sender: netdev-owner@vger.kernel.org List-ID: On Wed, 05 Aug 2009 00:59:43 +0200 Rene Mayrhofer wrote: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > Rene Mayrhofer wrote: > > Nonetheless, the current unmodified version from netdev git solves the > > oops in sky2. > Actually, it doesn't. I managed to run networking restart twice without > an oops (with the netdev git version of sky2.c), but after generating > some minor traffic and trying to restart again, I still get this oops: > > [~]# /etc/init.d/networking restart > Reconfiguring network interfaces...[ 844.000236] sky2 0000:01:00.0: > error interrupt status=0xffffffff > > [ 844.007309] sky2 0000:01:00.0: PCI hardware error (0xffff) > > [ 844.013657] sky2 0000:01:00.0: PCI Express error (0xffffffff) There is something about the hardware on your system that causes the Marvell chip to not be present on the bus after the steps taken in sky2_down. Is there something unique about how it is wired to the PCI express bus? The sky2 driver has to handle the rare case of dual port board, so in sky2_down in only shuts off part of the chip. Driver turns off the PHY and stops receiver/transmitter. It could be the power control bits on your hardware turn off more than just the PHY. Or perhaps, most systems have a low power input to keep chip alive for Wake On Lan and that isn't present on your system. Maybe an option to not power down phy would be the simplest fix. --