From owner-freebsd-questions Fri Feb 24 09:36:15 1995 Return-Path: questions-owner Received: (from majordom@localhost) by freefall.cdrom.com (8.6.9/8.6.6) id JAA20762 for questions-outgoing; Fri, 24 Feb 1995 09:36:15 -0800 Received: from cs.weber.edu (cs.weber.edu [137.190.16.16]) by freefall.cdrom.com (8.6.9/8.6.6) with SMTP id JAA20756 for ; Fri, 24 Feb 1995 09:36:13 -0800 Received: by cs.weber.edu (4.1/SMI-4.1.1) id AA27584; Fri, 24 Feb 95 10:28:54 MST From: terry@cs.weber.edu (Terry Lambert) Message-Id: <9502241728.AA27584@cs.weber.edu> Subject: Re: Pentium w/ PCI and EISA problems To: steve@khoros.unm.edu (Steven Jorgensen) Date: Fri, 24 Feb 95 10:28:54 MST Cc: freebsd-questions@FreeBSD.org In-Reply-To: <9502240351.AA00545@borris.khoros.unm.edu> from "Steven Jorgensen" at Feb 23, 95 08:51:16 pm X-Mailer: ELM [version 2.4dev PL52] Sender: questions-owner@FreeBSD.org Precedence: bulk [ ... weird Pentium 66 PCI/EISA woes deleted (disk controller) ... ] How old is this box? You were aware that older Saturn chipsets did not correctly do writeback and thus could not be used with bus mastering DMA disk or other controllers, right? And that this applied to Neptune and Mercury chips too for a while, because the problem was in the PCI bridge masks, and those were shared between all of the chips? This is one of the reasons I've suggested auto-detect and a BINVD or other workaround for cache problems, with big messages at boot: THIS MACHINE DOES NOT SUPPORT CACHE WRITEBACK OR INVALIDATION THIS MACHINE HAS THE PENTIUM FDIV BUG THIS MACHINE DOES NOT SUPPORT DMA TRANSFERS ABOVE 16M etc. Then work around it in software. Bletcherous, but anything that works is better than anything that doesn't work. Can you check the date of manufacture and type of PCI bridge chip in your machine? If it isn't from March 1994 or later, you will need replacement chips to run most protected mode operating systems. Try turning off caching (external only, then internal only, then external and internal both off) as a workaround. Also, can you check your main bus controller chipset? Is it OPTi or HiNT and you have more than 16M of memory? If so, one workaround might be to yank memory out until you have 16M or less. You may want to try this anyway, and if it works, report the chipset type to warn others off. Terry Lambert terry@cs.weber.edu --- Any opinions in this posting are my own and not those of my present or previous employers.