From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S265996AbUHSGim (ORCPT ); Thu, 19 Aug 2004 02:38:42 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S265999AbUHSGil (ORCPT ); Thu, 19 Aug 2004 02:38:41 -0400 Received: from fmr01.intel.com ([192.55.52.18]:5274 "EHLO hermes.fm.intel.com") by vger.kernel.org with ESMTP id S265996AbUHSGii (ORCPT ); Thu, 19 Aug 2004 02:38:38 -0400 Subject: Re: 2.6.8.1-mm1 broke USB driver with ACPI pci irq routing... info follows From: Len Brown To: Bjorn Helgaas Cc: Shawn Starr , linux-kernel@vger.kernel.org, Cyrille Ch?p?lov In-Reply-To: <566B962EB122634D86E6EE29E83DD808182C37A2@hdsmsx403.hd.intel.com> References: <566B962EB122634D86E6EE29E83DD808182C37A2@hdsmsx403.hd.intel.com> Content-Type: text/plain Organization: Message-Id: <1092897504.25902.220.camel@dhcppc4> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.2.3 Date: 19 Aug 2004 02:38:25 -0400 Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Wed, 2004-08-18 at 19:40, Bjorn Helgaas wrote: > Make acpi_irq_penalty non-initdata, since it's used by the > non_init acpi_pci_link_allocate(). hmmm, now that we need to keep this around, we might consider squeezing it down to, say 64 bytes. This stuff will probably never run for IRQs > 64 and probably a byte/level is enough. > And make acpi_irq_penalty_init() > __init, since it is used only by the __init pci_acpi_init(). > > Signed-off-by: Bjorn Helgaas I'll apply this Bjorn. I have to complement you on your paranoia to 1. keep acpi_pci_irq_enable(dev) for all devices in 2.6.8 2. add pci=routeirq to diagnose the breakage when you took it out thanks, -Len