Date: Tue, 27 Feb 1996 13:59:01 -0700 (MST) From: Terry Lambert <terry@lambert.org> To: imb@scgt.oz.au (michael butler) Cc: terry@lambert.org, current@FreeBSD.ORG Subject: Re: Opti 82C895 cache coherency ? Message-ID: <199602272059.NAA05791@phaeton.artisoft.com> In-Reply-To: <199602271505.CAA00676@asstdc.scgt.oz.au> from "michael butler" at Feb 28, 96 02:05:55 am
next in thread | previous in thread | raw e-mail | index | archive | help
> > Does it work with cache disabled? > > Nope. L1 an L2 both disabled? Peculiar... > > If so, I suspect the VLB slot the controller is in is not a master slot. > > Fortunately, there are three documented (!) in the manual but, of course, I > only want one. If I use "master slot 1" it won't boot. If I use either slot > labelled "master slot 0", it will but then it won't talk to the ethernet > card (WD8013EPC, 0x280, 5, 0xd8000) :-( It simply hangs in (NFS) mounting > some shared directories. Clearly, you only have one master slot. The ethernet card needs to be relocated. Try jumpering additional wait states and see if it is a DRAM refresh problem. > If, however, I replace the 2842 with a Buslogic 542B (ISA) and reboot with > an older version of -stable, it will fly quite happily even if strangled on > disk I/O. I am assuming that a cache problem would exhibit some symptoms > with this card as well. No. The problem is specific to the cache notification pin of the VLB. ISA will work correctly. Even on the old Gateway/Dell 60MHz Pentium (with math bug) and the old Saturn I chipset worked correctly with ISA (the Saturn I differs from the Saturn II in that the PCI DMA notification was left disconnected in the I -- that's why you have to turn of the cache in those machines when using bus mastering DMA PCI cards). Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199602272059.NAA05791>