From mboxrd@z Thu Jan 1 00:00:00 1970 From: Ingo Molnar Subject: Re: [patch 10/10] forcedeth: work around NULL skb dereference crash Date: Tue, 3 Apr 2007 08:26:29 +0200 Message-ID: <20070403062629.GA13210@elte.hu> References: <20070402115325.GA945@elte.hu> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: akpm@linux-foundation.org, jeff@garzik.org, netdev@vger.kernel.org, linux-kernel@vger.kernel.org To: Ayaz Abdulla Return-path: Content-Disposition: inline In-Reply-To: Sender: linux-kernel-owner@vger.kernel.org List-Id: netdev.vger.kernel.org * Ayaz Abdulla wrote: > I had responded eariler to the thread asking you to try out the patch > found in bug 8058: > http://bugzilla.kernel.org/show_bug.cgi?id=8058 > > I believe that is the caush of the NULL skb dereference issue. there's a different type of regression now: under high load i dont get a crash, i get a hung interface instead. No error packets or other weird interface state - just a hung interface. The condition happens after a stream of: eth1: too many iterations (6) in nv_nic_irq. eth1: too many iterations (6) in nv_nic_irq. eth1: too many iterations (6) in nv_nic_irq. eth1: too many iterations (6) in nv_nic_irq. eth1: too many iterations (6) in nv_nic_irq. If i increase max_interrupt_work from 5 to 50 to paper over this condition then the hangs go away. So i suspect there might still be a queue-wakeup problem in one of these paths - or that a filled up tx ring somehow gets stuck. (and i've got commits 3ba4d093fe8a26f and fcc5f2665c81e08, so all the latest stuff that is upstream at the moment) Ingo