Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S262967AbTIJMxO (ORCPT ); Wed, 10 Sep 2003 08:53:14 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S262983AbTIJMxN (ORCPT ); Wed, 10 Sep 2003 08:53:13 -0400 Received: from tench.street-vision.com ([212.18.235.100]:49044 "EHLO tench.street-vision.com") by vger.kernel.org with ESMTP id S262967AbTIJMxD (ORCPT ); Wed, 10 Sep 2003 08:53:03 -0400 Subject: Re: [2.4] siimage locks hard on high load From: Justin Cormack To: Witold Krecicki Cc: Kernel mailing list In-Reply-To: <200309121344.43573.adasi@kernel.pl> References: <200309121344.43573.adasi@kernel.pl> Content-Type: text/plain Content-Transfer-Encoding: 7bit X-Mailer: Ximian Evolution 1.0.8 (1.0.8-11) Date: 10 Sep 2003 13:52:38 +0100 Message-Id: <1063198360.313.54.camel@lotte.street-vision.com> Mime-Version: 1.0 Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 1589 Lines: 40 On Fri, 2003-09-12 at 12:44, Witold Krecicki wrote: > I've got Asus a7n8x deluxe with on-board Silicon Image SATA and 2xBarracuda > 120GB connected to it in software RAID array. > On high disk load (e.g. cp -Rf /usr/src/linux somewhere_else) kernel locks > hard after about 10 seconds (magic sysrq is not working). > It happens only when DMA is enabled. > /dev/hda: > multcount = 16 (on) > IO_support = 1 (32-bit) > unmaskirq = 1 (on) > using_dma = 1 (on) > keepsettings = 0 (off) > readonly = 0 (off) > readahead = 16 (on) > geometry = 14593/255/63, sectors = 234441648, start = 0 > > Also, I'm not really satisfied with speed of this array: > /dev/md1: > Timing buffered disk reads: 64 MB in 1.77 seconds = 36.16 MB/sec > Any solutions/fixes? Cant help you on the lockups (I havent seen this myself), but the performance is due to the workaround that sets max_kb_per_request to 15 for siimage. On some drives (Western digital, maxtor) this only cuts the performance by 10-15MB/s or so but on the Seagates it completely trashes performance (you can easily test this but you may get data corruption...). A proper bug fix rather than a workaround would be nice (unless this isnt possible). Are the docs for this chipset actually available and is a bug fix possible? (Alan/Andre should know). Justin - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/