From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751258AbVJSTrN (ORCPT ); Wed, 19 Oct 2005 15:47:13 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1751259AbVJSTrM (ORCPT ); Wed, 19 Oct 2005 15:47:12 -0400 Received: from quechua.inka.de ([193.197.184.2]:56301 "EHLO mail.inka.de") by vger.kernel.org with ESMTP id S1751258AbVJSTrL (ORCPT ); Wed, 19 Oct 2005 15:47:11 -0400 To: linux-kernel@vger.kernel.org Subject: Kernel hang under high network load (b44?) Organization: private Linux site, southern Germany From: Olaf Titz Date: Wed, 19 Oct 2005 21:44:58 +0200 Message-ID: Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org This happens under all of the following kernels: 2.6.10, 2.6.12.3, 2.6.13.2. I'm trying to process VDR recordings from another box. These are up to 2GB per file. I have mounted them via NFS, both machines have 100Mbps network interfaces connected via a 100Mbps switch. Sometimes when the reading process is halfway through the file, the system locks up completely. No trace in the logs or on the console. This happens regardless of mount options, it even happens when scp-ing the file for local processing (so I think this is not NFS-related at all). It does _not_ happen when I configure the VDR box's network interface to use only 10 Mbps speed. This makes me think it's a problem with high network load. The box in question has an on-board b44 interface: 0000:00:09.0 Ethernet controller: Broadcom Corporation BCM4401 100Base-T (rev 01) Subsystem: Asustek Computer, Inc. A7V8X motherboard Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- SERR-