Date: Wed, 19 Apr 1995 00:37:18 -0700 (PDT) From: "Rodney W. Grimes" <rgrimes@gndrsh.aac.dev.com> To: fenner@parc.xerox.com (Bill Fenner) Cc: freebsd-hardware@FreeBSD.org, fenner@parc.xerox.com Subject: Re: PCI plug-n-play on Intel Premiere Baby II? Message-ID: <199504190737.AAA01354@gndrsh.aac.dev.com> In-Reply-To: <95Apr19.001812pdt.49864@crevenia.parc.xerox.com> from "Bill Fenner" at Apr 19, 95 00:18:05 am
next in thread | previous in thread | raw e-mail | index | archive | help
>
I pulled the subject line inside, as it is *important* to what I
am about to say:
Subject: PCI plug-n-play on Intel Premiere Baby II?
^^^^^^^^^^^^^^^^^^^^^^
This board is also, I believe, called the Intel Plato card, check you
user manual to see if it says Plato in it.
> I just bought joe random PCI ethernet card, with a DEC 21040 chip on it.
> I already had an NCR 53810 with a "PCI interrupt jumper" on it, which
> allows the selection of PCI interrupt A through D.
>
> The NCR was jumpered to PCI interrupt A. When I plugged the ethernet card
> in, they got probed as follows:
>
> de0 <Digital DC21040 Ethernet> int a irq 9 on pci0:6
> reg20: virtual=0xf2925000 physical=0xc0000000
> ncr0 <ncr 53c810 scsi> int a irq 9 on pci0:14
> reg20: virtual=0xf2926000 physical=0xc0001000
>
> The SCSI probe hung while trying to scan the bus. Not horribly surprising.
>
> I changed the jumper on the NCR card to its interrupt B position. When I
> rebooted with this configuration, the probes looked *the*same* but the
> SCSI probe no longer hung.
>
> Any ideas what might be going on, or what I can do to make sure that these
> cards aren't going to interfere with each other when I start using the
> machine more heavily?
What version of the BIOS does the motherboard have on it? Intel had
some very brain dead ideas on this board when they did the BIOS. It
will force any card using the same PCI interrupt signal (A, B, C or D)
to use the same interrupt no mater what you do. They also left you
with very little flexiability in what you could do to stop cards from
sharing an interrupt. Your luck in that your cards are of different
class (ie, one is storage, one is network). Setting them to use PCI A
and PCI B works becuase the NCR driver falls back to polled mode (or
at least I think it still does) if it does not see interrupts from the
card.
Basically this was one of the first PCI P54C-90 motherboards on the market
and I talked it down then, and still tell people to avoid it if they can.
I brought up the issue with the Intel BIOS author for this board, and
he basically told me that the PCI spec allowed them to do it that way
and if Unix or any other OS doesn't like things on the same interrupt,
then that OS is not following the SPEC. (Which is true, but for performance
reasons doing this and forcing the OS to do run time dispatch of interrupts
is just silly).
If you have set the jumper on the NCR card to interrupt B, this motherboard
will never see the interrupt from that card as far as I can tell.
> Thanks,
> Bill
--
Rod Grimes rgrimes@gndrsh.aac.dev.com
Accurate Automation Company Custom computers for FreeBSD
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199504190737.AAA01354>