From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id ; Wed, 20 Jun 2001 06:00:03 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id ; Wed, 20 Jun 2001 05:59:53 -0400 Received: from malvek.artecdesign.ee ([194.204.53.4]:61530 "EHLO malvek.artecdesign.ee") by vger.kernel.org with ESMTP id ; Wed, 20 Jun 2001 05:59:48 -0400 Message-ID: <3B307415.C80C0C59@artecdesign.ee> Date: Wed, 20 Jun 2001 11:59:50 +0200 From: Heiki Kask X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.18custom i686) X-Accept-Language: en MIME-Version: 1.0 To: linux-kernel@vger.kernel.org Subject: cs5530 and UDMA implementation questions 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, I have following problems/questions I could not find answer from the kernel source code: - cs5530.c does not initialize hwif->speedproc hook. This means that hdparm -X does not call routine for chipset registers reprogramming. Did I miss something? - how chipset/disk initialization is supposed to work? When kernel is compiled without CONFIG_IDEDMA_PCI_AUTO, then there are no calls performed to initialize chipset for PIO or UDMA mode and therefore there is no possibility to recalculate IDE timings for different bus speeds (if it is implemented) before root filesystem gets mounted. This is basically OK when BIOS or BIOS replacement is initializing IDE controller to match the bus speed and the kernel does not make attempt to reprogram IDE controller, but then there is no sense to implement the feature at all because nobody will change bus speed at runtime, I guess. Thanks for listening. heiki