From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andi Kleen Subject: Re: [ACPI] Re: [PATCH] 2.6.0-test4 ACPI fixes series (4/4) Date: Sat, 6 Sep 2003 15:45:50 +0200 Sender: linux-kernel-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org Message-ID: <20030906154550.0334a2d5.ak@suse.de> References: <200309051958.02818.adq_dvb@lidskialf.net> <200309060157.47121.adq_dvb@lidskialf.net> <3F5936D2.3060502@pobox.com> <200309061327.16347.adq_dvb@lidskialf.net> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <200309061327.16347.adq_dvb-fmPXVN3awWJAJAzL26g0SA@public.gmane.org> To: Andrew de Quincey Cc: jgarzik-e+AXbWqSrlAAvxtiuMwx3w@public.gmane.org, torvalds-3NddpPZAyC0@public.gmane.org, linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, acpi-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org, linux-acpi-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org, mikpe-gJWeRRR3xd2zQB+pC5nmwQ@public.gmane.org List-Id: linux-acpi@vger.kernel.org On Sat, 6 Sep 2003 13:27:16 +0100 Andrew de Quincey wrote: > On Saturday 06 September 2003 02:22, Jeff Garzik wrote: > > Andrew de Quincey wrote: > > > This patch removes some erroneous code from mpparse which breaks IO-APIC > > > programming > > > > > > > > > --- linux-2.6.0-test4.null_crs/arch/i386/kernel/mpparse.c 2003-09-06 > > > 00:23:10.000000000 +0100 +++ > > > linux-2.6.0-test4.duffmpparse/arch/i386/kernel/mpparse.c 2003-09-06 > > > 00:28:23.788124872 +0100 @@ -1129,9 +1129,6 @@ > > > continue; > > > ioapic_pin = irq - mp_ioapic_routing[ioapic].irq_start; > > > > > > - if (!ioapic && (irq < 16)) > > > - irq += 16; > > > - > > > > Even though I've been digging through stuff off and on, I consider > > myself pretty darn IOAPIC-clueless. Mikael, does this look sane to you? > > Really breaks on TX150 servers... All IRQs < 16 get +16 added onto them, which > breaks all IRQ routing. It's also already been removed from 2.4.23-pre3 It is needed at least for the Unisys ES7000. But that box needs further changes anyways which are not in tree yet and is even an own subarchitecture that can be tested for. -Andi