From: Ingo Molnar Subject: Re: Revert "gro: Fix legacy path napi_complete crash", (was: Re: Linux 2.6.29) Date: Tue, 24 Mar 2009 14:12:08 +0100 Message-ID: <20090324131208.GA10157@elte.hu> References: <20090324130202.GA32469@elte.hu> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Linux Kernel Mailing List , netdev@vger.kernel.org To: Linus Torvalds , Herbert Xu , Frank Blaschka , "David S. Miller" , Thoma Return-path: Received: from mx2.mail.elte.hu ([157.181.151.9]:36367 "EHLO mx2.mail.elte.hu" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1758095AbZCXNMw (ORCPT ); Tue, 24 Mar 2009 09:12:52 -0400 Content-Disposition: inline In-Reply-To: <20090324130202.GA32469@elte.hu> Sender: netdev-owner@vger.kernel.org List-ID: (netdev Cc:-ed) * Ingo Molnar wrote: > Yesterday about half of my testboxes (3 out of 7) started getting > weird networking failures: their network interface just got stuck > completely - no rx and no tx at all. Restarting the interface did > not help. > I've attached the reproducer (non-SMP) .config. The system has: Note, the .config is randconfig derived. There was a stage of the tests when about every ~5-10th randconfig was failing, so i dont think it's a rare config combo that triggers this. (but there were other stages where 30 randconfig in a row went fine so it's hard to tell.) In the worst case the hang needed 2 million packets to trigger - that's why i set the limit in the tests to 6 million packets. Ingo