>> >The existance of an EISA config file does *not* indicate that it's an
>> >EISA device.  You can find EISA config files for quite a variety of
>> >ISA devices; they mainly exist as a repository of info about jumpers
>> >and whatnot.
>> 
>> The 3c509 actually shows up in slot space when in eisa configuration mode.
>> The only way you know that its not an EISA device from eisaconf is by
>> having the driver tell you when it matches the ID.  Your statement
>> above is generally correct, but the 3c509 is an exception.
>
>Again, "show up as an EISA device != an EISA device."
>
>One of the hallmarks of an EISA device is that, because of them there
>extra pins, it's not going to work in an ISA motherboard.  As far as i
>know, that's not true of the '509.  It may configure as an EISA
>device, but that doesn't make it an EISA device.

Okay, it configures like an EISA device, it shows up in slot space as an
EISA device, the ECU thinks its an EISA device, NetBSD-current even thinks
its an EISA device...  The only thing it doesn't do is 32bit PIO, but there
are some broken EISA cards (with "them there extra pins") that break when
you do this too, so I'd call it a "short" EISA device when installed in an
EISA system.  From a software point of view, the only way you can tell
that its "short" is by matching its ID in the driver and having it tell you
to report it as an ISA device and only do 16bit PIO to the device.

>That is, unless you're going to take the tack that "all ISA devices
>are EISA devices," but that's quite a waste...

All ISA devices can't be fully configured through an ECU, nor sport EISA ID
registers, etc.  I didn't even come close to saying, "all ISA devices are
EISA devices."

In the end, when I was doing this in FreeBSD, I decided that forcing things
like the 3c509 (in EISA configuration mode) and the 2842 to be
registerd/probed as ISA devices yet still automatically find all cards
(implied re-scanning of the EISA slot address space) was pure code
duplication (with all of its associated headaches) with no difference in
performance or robustness to the system.  Again YMMV.

>cgd

--
Justin T. Gibbs
===========================================
  FreeBSD: Turning PCs into workstations
===========================================