Subject: Re: Recent VME spurrious interupts...
To: None <abrown@eecs.harvard.edu, dgilbert@jaywon.pci.on.ca>
From: Chris Torek <torek@bsdi.com>
List: port-sparc
Date: 03/20/1996 20:47:22
Is this really `ie1', or `le1'?

The le's should not fail this way, but the VME `ie's will.  Chuck
Hedrick explained why once -- the problem is that ie0 is on the
CPU card and ie1 is on the VME bus, and ie0 actually interrupts
at a higher priority than ie1.  (I believe this was the case for
the sun3, and am not at all sure about the VME sun4.)

The Sbus-based le driver should simply service the first interrupting
le, and then return; the level-triggered interrupts will make the
code try the other le device at that point.

Chris