From owner-freebsd-questions Fri Oct 20 16:52:39 2000 Delivered-To: freebsd-questions@freebsd.org Received: from guru.mired.org (okc-27-149-77.mmcable.com [24.27.149.77]) by hub.freebsd.org (Postfix) with SMTP id A6C3637B4C5 for ; Fri, 20 Oct 2000 16:52:35 -0700 (PDT) Received: (qmail 66457 invoked by uid 100); 20 Oct 2000 23:52:34 -0000 From: Mike Meyer MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Message-ID: <14832.56002.369564.835637@guru.mired.org> Date: Fri, 20 Oct 2000 18:52:34 -0500 (CDT) To: "Otter" Cc: Subject: RE: FW: booting problems with SMP enabled In-Reply-To: References: <14831.46693.860465.246500@guru.mired.org> X-Mailer: VM 6.75 under 21.1 (patch 10) "Capitol Reef" XEmacs Lucid X-face: "5Mnwy%?j>IIV\)A=):rjWL~NB2aH[}Yq8Z=u~vJ`"(,&SiLvbbz2W`;h9L,Yg`+vb1>RG% *h+%X^n0EZd>TM8_IB;a8F?(Fb"lw'IgCoyM.[Lg#r\ Sender: owner-freebsd-questions@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG Otter writes: > }Otter writes: > }> I've got a new install of 4.1-RELEASE here. > }> Hardware consists of a dual processor board (unknown brand), > }> 2 Intel P2-300's, a single 128MB DIMM, onboard scsi > }> (disabled), on a 13GB IDE drive. The machine boots and runs > }> fine with a single processor. For testing purposes, I've > }> swapped them around to make sure that both CPU's are working > }> as they should. No problems there. When I boot up with the > }> SMP kernel (only the SMP and APIC_IO lines uncommented) it > }> hangs in the boot process at where it says "APIC_IO: testing > }> 8254 interrupt delivery". At this point, the machine locks > }> and requires me cycling power via the switch on the power > }> supply. I'm at a loss. I've got a bank of dipdswitches on > }> the board and have changed a few at random, but no luck so > }> far. Without knowing the brand of the board, I'm not sure > }> how I can take this any further. Any tips/ideas/suggestions? TIA. > }Check the board manufacturers web site for a manual! > I'd love to, but as you can see above, I don't know who the > manufacturer is/was. I've found no markings on the board that identify > it as any particular brand. Mea culpa. Without that, you may well be hosed. After all - how do you know this is an Intel MP spec motherboard? If it isn't, it won't work in that mode. Anyway, another route to try is checking the site for the company that wrote the BIOS. At least one of those had instructions on how to identify the motherboard manufacturer based on the BIOS. > }Also, check the stepping number of the P2's (on the chip, somewhere, > }one hopes, or possibly in dmesg when the system boots) and then check > }the intel web to make sure the two chips will work *together*. While > }each may work fine, if you've don't have compatible stepping numbers, > }they won't work together. > The stepping numbers DO match. I've searched Intel's site and found > some CPUID info charts, but nothing that states they work together... > then again, I didn't find anything saying the contrary. I seem to > remember that was the big hype of P2 over Celerons-- they had more > cache and would support SMP. I could only assume that they all do... > and neer had problems getting any p2 or p3's to work in SMP as long as > the stepping numbers matched. I've set this up on several machines in > the past, so I know (or at least I used to!) what to look for and the > steps to take to make it happen. I'm just at a loss on this one. Well, if the numbers are the same, then they should work together. I've only been through this for P2 Xeons. For those, incompatable stepping numbers are a problem, and there's a chart on the web site showing which stepping numbers work together. I assumed the standard P2 would be the same. As for the P2 vs. Celeron, I don't pay a lot of attention. I think the difference is packaging, and the size of the on-chip cache. The Celerons started with *no* cache, then showed up with some, and these days you can get them with full speed cache, though not as much as a P2. As for SMP, Intel claims that Celerons don't do it, but I understand that third parties sell socket adapters that let them work, or even Celeron SMP motherboards.