>rendering the cards unuseable under NetBSD. (Note an ne2000 pci bus card
>works).

NE2000s, even PCI versions, are not busmastering so they can never suffer 
master aborts.

>I hear others are using the EBSA 285 with these devices and all is OK? Well
>I'd really like to know what it is we are doing wrong. An "undocumented"
>jumper setting on the EBSA board (or backplane) perhaps?

It is almost certainly a driver bug.  A master abort happens when you try to 
access an area of PCI memory or I/O that doesn't exist.  If the ethernet chip 
is recording an abort, one of the descriptors it was fed contained duff 
information, or the Footbridge is not set up to respond as a PCI target.  It's 
remotely possible that a hardware fault could cause this but I would exhaust 
all software avenues first if I were you.

p.