Hello All

I have been having many problems with NetBSD 1.4 on an Ebsa 285.

Trying to use various flavours of ethernet "Tulip" (i.e "de") cards or Intel
82557/8/9 (fxp) ethernet cards, I get a PCI status "Master ABort" error in
the PCI status registers of the respective CARDS. This occurs repeatedly,
rendering the cards unuseable under NetBSD. (Note an ne2000 pci bus card
works).

When trying the Tulip cards with the Angel debugger (No NetBSD involved
here), I can connect via the Arm SDT debugger and do stuff (e.g. examine
memory) and (surprise surprise), the 21285 chip has recorded a "Master
Abort" status in its PCI status register. However the ether cards "Master
Abort" are clear and appear to remain so thru program downloads to debugger
etc...If I use an Angel Image that does not have ethernet support, then the
21285 also doesn't reflect a "Master Abort". Nothe that this is different to
the NetBSD case where the ether devices reflected the Master Abort error.

Now I have tried two different EBSA boards, two different (digital) PCI
development backplanes (and all combinations thereof), more or less every
tulip flavour card I could get my hands on (5 different 21x4x's), Intel
82557/8/9, different RAM Dimms (including those supplied with the EBSA
boards) and the symptoms are the same. I have also tried making the EBSA
board handle the PCI Bus arbitration (normally that is left to the PCI
backplane).

I wont even mention my SCSI drive problems (also using a PCI Bus SCSI
controller....).

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?

I'm now at my wits end.....
Any help/suggestions very much appreciated.

Thanks

David McConnell