From mboxrd@z Thu Jan 1 00:00:00 1970 From: Mike Fowler Subject: Re: I/O APIC: AMD Errata #22 may be present. Date: Tue, 25 Oct 2005 10:48:07 +0000 Message-ID: <435E0D67.9030505@mlfowler.com> References: <435D5670.1070803@ianbrandt.com> Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <435D5670.1070803@ianbrandt.com> Sender: linux-smp-owner@vger.kernel.org List-Id: Content-Type: text/plain; charset="us-ascii"; format="flowed" To: Ian Brandt Cc: Linux SMP Ian, I hava two Athlon MP 1.4 processors with the same errata present. I've used the noapic option without any "noticeable" performance loss. I say "noticeable" as I have not done any performance tests to qualify that statement, but it certainly hasn't hindered me in any way. Hope this helps! > Hi, > > I have an Appro 1124s server with a Tyan Thunder K7 S2462 motherboard, > two AMD Athlon MP 1600 processors, the 760 MP chipset, and a single > PCI device that is an Adaptec 2100S SCSI RAID controller (on a PCI > riser/adapter of unknown origin). I've just upgraded to 2.6.13, and > occasionally my server fails to boot. It seems to hang at this line: > > I/O APIC: AMD Errata #22 may be present. In the event of instability > try booting with the "noapic" option. > > That line is however present in my kernel messages even on a > successful boot. My entire dmesg from a successful boot, > /proc/interrupts, and proc/cpuinfo are below. > > I've read Documentation/i386/IO-APIC.txt and have a vague > understanding of what it means. I will certainly try the noapic > option to see if things settle down, but is there any real world > performance loss in doing so? Should I be looking into using "...the > pirq= boot parameter to 'hand-construct' IRQ entries"? Is there > anything else I could do to help squash this bug? > > Thanks! > > Ian > [snip] -- Mike Fowler Registered Linux user: 379787 "I could be a genius if I just put my mind to it, and I, I could do anything, if only I could get 'round to it" -PULP 'Glory Days'