From mboxrd@z Thu Jan 1 00:00:00 1970 From: Larry Finger Subject: Re: [PATCH 1/2] kaweth: Fix locking to be SMP-safe Date: Mon, 30 Mar 2009 10:58:28 -0500 Message-ID: <49D0EC24.4080206@lwfinger.net> References: <49d00512.XAF19LdpY1dlK6+U%Larry.Finger@lwfinger.net> <200903300828.45337.oliver@neukum.org> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: jgarzik@pobox.com, linux-kernel@vger.kernel.org, netdev@vger.kernel.org To: Oliver Neukum Return-path: Received: from fmailhost04.isp.att.net ([207.115.11.54]:61156 "EHLO fmailhost04.isp.att.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752567AbZC3P7Q (ORCPT ); Mon, 30 Mar 2009 11:59:16 -0400 In-Reply-To: <200903300828.45337.oliver@neukum.org> Sender: netdev-owner@vger.kernel.org List-ID: Oliver Neukum wrote: > Am Montag 30 M=E4rz 2009 01:32:34 schrieb Larry Finger: >> On an SMP system, the following message is printed. The patch below = gets >> fixes the problem. >=20 > Thanks for this report and the patch. I think, however that it introd= uces > unneeded locking. It seems to me that we should be fine if we fix > kaweth_start_xmit(). That code assumes that it is called with interru= pts > off and under a spinlock. Is that incorrect? You are correct in that only the locking in kaweth_start_xmit() needs t= o be changed to lock out the other CPU's. In my testing under extreme condit= ions (X server over the network), the interface stalled with no logged messages= =2E My other changes in the locking were an unsuccessful attempt to fix that a= nd have been removed. Version 2 of the patches will be sent after I finish test= ing. Thanks, Larry