From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id ; Tue, 13 Feb 2001 19:25:28 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id ; Tue, 13 Feb 2001 19:25:18 -0500 Received: from gnu.in-berlin.de ([192.109.42.4]:10502 "EHLO gnu.in-berlin.de") by vger.kernel.org with ESMTP id ; Tue, 13 Feb 2001 19:25:04 -0500 X-Envelope-From: michael.karcher@dpk.berlin.fido.de Date: Tue, 13 Feb 2001 20:17:04 +0100 From: Michael Karcher Subject: Crash in request_region while handling kernel parameters Message-ID: <20010213201704.A319@p9.dpk.berlin.fido.de> To: linux-kernel@vger.kernel.org Organization: Welt am Draht - Berlin X-Gateway: FIDO WAD.in-berlin.de [FIDOGATE 4.4.0] X-FTN-From: Michael Karcher @ 242:7000/710.9 X-FTN-To: UUCP @ 242:7000/2.0 X-FTN-Tearline: FIDOGATE 4.2.3 X-FTN-Via: FIDOGATE/rfc2ftn 242:7000/710.9, Tue Feb 13 2001 at 20:17:08 CET X-FTN-Via: 242:7000/710 @20010213.214005 FastEcho 1.46.1 1078 X-FTN-Via: 242:7000/704@fidode @20010213.220454 FastEcho 1.46.1 2212 X-FTN-Via: FIDOGATE/ftntoss 242:7000/2.0, Wed Feb 14 2001 at 01:23:41 CET X-FTN-Domain: Z242@FidoDE MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Hello kernel-hackers, I found a problem with kernel 2.4, that makes the kernel crash at bootup, for example when using the UMC8672 VLB IDE controller driver. The problem is in kernel/resource.c. In line 229 some memory for handling new io-regions is kmalloc()ed. This crashes the computer before mem_init(), as it seems. But some drivers, for example the above mentioned one in drivers/ide/umc8672.c, do already claim i/o ports in their kernel parameter driven initialization procedures, so they crash the system. The point to discuss is whether one needs to fix the drivers, to not request i/o space while handling kernel parameters or to fix the kernel to allow this behaviour. As I do not read this mailing list, a currently don't have web access ready, I don't know whether this topic has already been discussed. Currently I helped myself by just commenting out the calls for check_region and request_region in the umc8672.c file, but I know it smells. Please send me a cc if you think you have a decision how to fix the crash. Michael Karcher