From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S270624AbTHJSom (ORCPT ); Sun, 10 Aug 2003 14:44:42 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S270622AbTHJSom (ORCPT ); Sun, 10 Aug 2003 14:44:42 -0400 Received: from hades.mk.cvut.cz ([147.32.96.3]:37032 "EHLO hades.mk.cvut.cz") by vger.kernel.org with ESMTP id S270627AbTHJSoj (ORCPT ); Sun, 10 Aug 2003 14:44:39 -0400 Message-ID: <3F369296.10709@kmlinux.fjfi.cvut.cz> Date: Sun, 10 Aug 2003 20:44:38 +0200 From: Jindrich Makovicka User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.5b) Gecko/20030808 X-Accept-Language: cs, en-us, en MIME-Version: 1.0 To: linux-kernel@vger.kernel.org Subject: hpt366 kernel panic on EPoX 8K9A3+ fix (?) Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org [please cc:] Hello, with the 2.6.0-test3 kernel, hpt366 still does nothing else than kernel panic on my machine. However, I found that with my EPoX 8K9A3+, the 0x78 (frequency) register of HPT374 has values in the interval of 0x9B..0x9E, although the FSB is set to 133MHz and PCI should be at 33. Then, in init_hpt37x, the execution usually goes thru the "freq < 0xb0" branch, which doesn't do the pci_set_drvdata stuff. Later, the frequency detection loop fails because hpt374 internal PLL isn't supported. However, changing the respective branch to } else if (freq < 0xb0) { pll = F_LOW_PCI_40; if (hpt_minimum_revision(dev,8)) { pci_set_drvdata(dev, (void *) thirty_three_base_hpt374); printk("HPT37X: using 33MHz PCI clock\n"); } } ... made the driver work, although I didn't have time to test it (just mounted the disk and copied a few files). Regards, -- Jindrich Makovicka