From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S264292AbTIIXj0 (ORCPT ); Tue, 9 Sep 2003 19:39:26 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S265080AbTIIXj0 (ORCPT ); Tue, 9 Sep 2003 19:39:26 -0400 Received: from hstntx01.bsius.com ([64.246.32.38]:5814 "EHLO mx1.bsius.com") by vger.kernel.org with ESMTP id S264292AbTIIXjQ (ORCPT ); Tue, 9 Sep 2003 19:39:16 -0400 From: "Bill Church" To: Subject: 2.4.2x kernel + ICH4 DMA errors Date: Tue, 9 Sep 2003 19:39:12 -0400 Organization: Bayside Solution, Inc. Message-ID: <001a01c3772b$92fba410$6900000a@bsi000> MIME-Version: 1.0 Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: 7bit X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook, Build 10.0.4024 Importance: Normal X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org I have a GigaByte GA-8IGX with an Intel 845G chipset. Two new Hitachi 180GXP 80G DeskStar drives. Both set as master on two ATA/100 channels (/dev/hda and /dev/hdc). P4 2.53 Processor at 533FSB 2 Sticks of 512MB Corsair 333MHz DDR Ram My issue: Started out with Gentoo gs-sources Kernel (2.4.22_pre2-gss). When I would boot with both drives connected, /dev/hdc would timeout and hang when enumerating the partitions. If I disabled DMA, the system would boot with out error. I disabled ACPI also, but that seemed to have no effect. I reverted to the vanilla-sources Kernel (2.4.22) with the same config file and experienced the same results. However, /dev/hdc would time out several times but no hang. Upon investigation I found that /dev/hda had DMA enabled but /dev/hdc had DMA disabled. So with hdparm I enabled dma on /dev/hdc and tried some operations on that drive. Just an fdisk /dev/hdc produced timeouts. Checking hdparm again on /dev/hdc revealed that it was again disabled. I then tried setting both drives on the same channel, manually setting master on one drive and slave on another. Same issues as before. I also tried swapping cables and drive positions. Moving /dev/hdc drive to /dev/hda and vice-versa. So I decided to revert to 2.4.20, I experienced the same results as did originally (time out on /dev/hdc to lockup). I then decided to try 2.6.0-test4, same results... I am using 80 conductor 40 pin ATA-100 cables. And I'm using the PIIXn option under ATA/IDE/MFM/RLL support. Thanks in advance... -Bill bill at bsius dot com