From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1758042AbXFMLWS (ORCPT ); Wed, 13 Jun 2007 07:22:18 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1756187AbXFMLWE (ORCPT ); Wed, 13 Jun 2007 07:22:04 -0400 Received: from smtp-out001.kontent.com ([81.88.40.215]:48244 "EHLO smtp-out.kontent.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755543AbXFMLWD (ORCPT ); Wed, 13 Jun 2007 07:22:03 -0400 From: Oliver Neukum To: Patrick McHardy Subject: Re: cannot set IP for ethernet Date: Wed, 13 Jun 2007 13:21:29 +0200 User-Agent: KMail/1.9.7 Cc: linux-kernel@vger.kernel.org, Linux Netdev List , Herbert Xu References: <200706121441.13561.oliver@neukum.org> <200706131039.32011.oliver@neukum.org> <466FD17A.8040900@trash.net> In-Reply-To: <466FD17A.8040900@trash.net> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-15" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200706131321.30451.oliver@neukum.org> Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Am Mittwoch, 13. Juni 2007 schrieb Patrick McHardy: > Oliver Neukum wrote: > > Am Dienstag, 12. Juni 2007 schrieb Patrick McHardy: > > > >>Oliver Neukum wrote: > >> > >>>with 2.6.22-rc4-git2 I am getting errors when setting IP for ethernet > >>>interfaces: > >>> > >>>ioctl(4, SIOCSIFADDR, 0x7fff94931600) = -1 ENOBUFS (No buffer space available) > >>> > >>>The error is independant of the interface. It happens to all interfaces. > >>>There's nothing in the syslog. > >>> > >>>valisk:/home/oliver # uname -a > >>>Linux valisk 2.6.22-rc4-git2-default #3 SMP Tue Jun 12 13:27:54 CEST 2007 x86_64 x86_64 x86_64 GNU/Linux > >> > >> > >>This can happen if the initial inetdev allocation when the netdevice is > >>registered fails. I think it would make sense to try to allocate again > >>when adding addresses in that case, otherwise there is no way of > >>recovery other than unregistering and registering the device again. > > > > > > With your patch the problem has gone away. Is there a way to especially > > stress the system in that regard? > > > Its a failed allocation at NETDEV_REGISTER time. I guess you could > keep unloading + reloading the driver module to try to reproduce it. If that is the cause, why doesn't it strike from the very beginning? How come it fails after some time? Regards Oliver