Re: PCI configuration problem in 2.3.99 prex

From: Jay Estabrook (
Date: Thu May 11 2000 - 14:59:36 EST

On Thu, May 11, 2000 at 10:47:09AM -0700, Richard Henderson wrote:
> The purpose of that minimum is to avoid the random PeeCee gunk
> that exists on the ISA bus that we don't know about. Secondary
> PCI busses are not connected to the ISA bus, and so do not have
> to worry about such things.

Until now... ;-}

Whether there's "PeeCee gunk" on a PCI bus is a function of whether
there's a PCI-ISA bridge on there or not (duh! :-), and it's pretty
much been the rule that we (COMPAQ Alpha) have only had one PCI-ISA
bridge in a system, and it was on PCI 0.

Wildfire (aka AlphaServer GS 40/80/160) changes all that...

Conceivably, *every* PCI bus in such a system could have one (though
why one might want that is not clear to me, nor is it clear how we'd
teach the legacy Linux device drivers how to access them :-).

It is a very common configuration option to have one set of the
standard "legacy" ISA devices (floppy/CDROM/serial/parallel/kbd/mouse)
for every 4 PCI buses actually present (we make a peripheral box for
Wildfire that has slots for 4 PCI buses plus an optional set of those
ISA devices).

So, perhaps it's not such a bad thing to allocate *all* PCI buses
with suitable "adjustment" to avoid the "gunk"? Doesn't seem to
be too hard, and it *will* prevent problems down the road ("maybe
not today, maybe not tomorrow, but soon and for the rest of our
lives..."<sic>, er... :-).


Jay A Estabrook Alpha Engineering - LINUX Project
Compaq Computer Corp. - MRO1-2/K20 (508) 467-2080
200 Forest Street, Marlboro MA 01752

This archive was generated by hypermail 2b29 : Mon May 15 2000 - 21:00:29 EST