NMI not working as expected on Dell 2850

[Tom Ivar Helbekkmo <tih%hamartun.priv.no@localhost>]

I've been using the NMI button on the front of my old Dell PowerEdge
2850 while debugging problems involving the amr and uhci drivers on it,
and there's some problem with the handling of that button that I hope
is something that can be fixed in the NetBSD kernel.

I'm running NetBSD/amd64-current.

In single CPU mode (boot -1), the NMI button will drop the machine into
the debugger, but only once.  If I "continue" from the debugger, the NMI
button will no longer work afterwards.

In SMP mode, the NMI button will, again, drop the machine into the
debugger, but two of the CPUs will do so, the text announcing this being
printed by both, and intermingled on the serial console.  Then, it seems
that only one of them manages to grab the console for interactive use,
because the debugger works just fine.  Backtraces for the various
processors will show that two of them have gone through a trap(),
kdb_trap(), printf() call sequence.  Which two will vary from time to
time.  If I "continue" from the debugger, the machine will hang, and a
forced power cycle is then the only way out.

Any hints as to how to figure out why this happens?

-tih
-- 
Popularity is the hallmark of mediocrity.  --Niles Crane, "Frasier"