Subject: Re: pchb.c
To: Carl Shapiro <>
From: Frank van der Linden <>
List: port-i386
Date: 01/10/1998 16:54:03
On Fri, Jan 09, 1998 at 08:33:39PM -0500, Carl Shapiro wrote:
> Is there any reason why a host bridge has to be matched against a list
> of product and vendor ID's rather than just trusting the devices bridge
> subclass code (which in this case would be PCI_SUBCLASS_BRIDGE_ISA)?
> For that matter, is there any reason why any of the pci bridge code
> has to match an IC's pid and vid against a list of known ID's?  Doing
> things in the existing fashion seems like a losing battle as it will
> be almost impossible to keep the match functions in up to date with
> the ever growing number of PCI-to-whatever bridges.

Yes, it is essentially useless to do this. Since the idea behind
pchb.c is to do some special case frobbing for PCI chipsets that
need it, the original thought must have been something like
"if we can't even match the ID, let's not bother with attach".

I think I'll remove all those matches the next time pchb.c needs
a change.

- Frank