From owner-aic7xxx Fri Apr 16 15:33:42 1999 Delivered-To: aic7xxx@freebsd.org Received: from news.internal.ids.net (news.internal.ids.net [155.212.1.8]) by hub.freebsd.org (Postfix) with ESMTP id BEBF215373 for ; Fri, 16 Apr 1999 15:33:25 -0700 (PDT) (envelope-from mikebw@bilow.bilow.uu.ids.net) Received: from bilow.bilow.uu.ids.net (uucp@localhost) by news.internal.ids.net (8.9.1a/8.9.1.1) with UUCP id SAA13392; Fri, 16 Apr 1999 18:31:02 -0400 (EDT) Received: by bilow.bilow.uu.ids.net (uugate-034-OS/2-230); Fri, 16 Apr 99 22:25:47 -0000 Date: Fri, 16 Apr 99 20:19:00 -0000 Message-ID: <71786813@bilow.bilow.uu.ids.net> From: mikebw@bilow.bilow.uu.ids.net (Mike Bilow) Subject: Problem w/ SMP and aic7xxx To: aic7xxx@freebsd.org Reply-To: mikebw@bilow.bilow.uu.ids.net X-Mailer: uugate 0.34 (OS/2 2.30) Sender: owner-aic7xxx@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org Joseph T. Trudeau wrote in a message to Mike Bilow: JTT> Hardware: JTT> HP Netserver LH Pro JTT> 128 Meg RAM (2 - 64 Meg DIMM) JTT> 2 - Pentium Pro 200's JTT> 2 - aic7880 on-board (PCI): They share interrupt 11 and JTT> cannot be changed to have unique interrupts for each (The EISA JTT> config utility promptly configures both adapters to the same JTT> interrupt when either is changed). The HP Netserver LH Pro does have this IRQ problem, but using a 2.2 kernel which supports IO-APIC should handle this, I would think. It does cause problems on the 2.0 kernels. JTT> NOTE: I noticed that the 1st CPU has 512K cache while the JTT> 2nd CPU only has 256K cache. That is ABSOLUTELY NOT a supported configuration in the HP Netserver. In fact, the CPUs must have exactly identical steppings to work stably; check /proc/cpuinfo to see what you have. It would also be worth pulling the CPUs and doing a physical inspection to make sure they are rated for the proper speed, since obviously a mismatched set cannot have been factory installed. If you run dual PPro CPUs with different cache sizes, the cache coherency is going to fall apart when you run DMA devices such as the SCSI controller. My suggestion is to test by building a 2.2 kernel for non-SMP and running with the second CPU physically removed: if that works stably, you have diagnosed it. -- Mike To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe aic7xxx" in the body of the message