From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S261420AbVGHUes (ORCPT ); Fri, 8 Jul 2005 16:34:48 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S262840AbVGHUdu (ORCPT ); Fri, 8 Jul 2005 16:33:50 -0400 Received: from ss1000.ms.mff.cuni.cz ([195.113.20.8]:64730 "EHLO ss1000.ms.mff.cuni.cz") by vger.kernel.org with ESMTP id S262821AbVGHU2D (ORCPT ); Fri, 8 Jul 2005 16:28:03 -0400 Date: Fri, 8 Jul 2005 22:27:53 +0200 From: Rudo Thomas To: Con Kolivas Cc: linux kernel mailing list , ck list Subject: Re: 2.6.12-ck3 Message-ID: <20050708202753.GA32317@ss1000.ms.mff.cuni.cz> Mail-Followup-To: Con Kolivas , linux kernel mailing list , ck list References: <200506301241.00593.kernel@kolivas.org> <20050704091648.GA14759@ss1000.ms.mff.cuni.cz> <20050707213034.GA9306@ss1000.ms.mff.cuni.cz> <200507081352.55660.kernel@kolivas.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <200507081352.55660.kernel@kolivas.org> User-Agent: Mutt/1.5.7i Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org > > > Time seems to pass very fast with this kernel. > > > > Am I the only one who gets this strange behaviour? Kernel's notion of > > time seems to be about 30 times faster than real time. > > Sorry I really have no idea. If you can retest with latest stable mainline > that this kernel is based on (2.6.12.2) and reproduce the problem The following one-liner (from 2.6.12.2) seems to be the problem: diff -urN linux-2.6.12-ck2-rudo/drivers/acpi/pci_irq.c linux-2.6.12-ck-rudo/drivers/acpi/pci_irq.c --- linux-2.6.12-ck2-rudo/drivers/acpi/pci_irq.c 2005-07-08 10:16:53.000000000 +0200 +++ linux-2.6.12-ck-rudo/drivers/acpi/pci_irq.c 2005-07-03 21:06:10.000000000 +0200 @@ -435,6 +435,7 @@ /* Interrupt Line values above 0xF are forbidden */ if (dev->irq >= 0 && (dev->irq <= 0xF)) { printk(" - using IRQ %d\n", dev->irq); + acpi_register_gsi(dev->irq, ACPI_LEVEL_SENSITIVE, ACPI_ACTIVE_LOW); return_VALUE(0); } else { I will report this to the lkml shortly. Rudo.