I updated to 4.0_BETA2 on the machine I mention in PR 34892
(http://www.netbsd.org/cgi-bin/query-pr-single.pl?number=34892)

With the 4.0_BETA kernel I was running I had the problem in the
PR above, as well as a panic every few days due to needing to
change the KVA split (too much RAM is a problem).  I am using
a 4.0_BETA2 GENERIC kernel for this test.

I updated to 20070403 4.0_BETA2, and now the machine hangs
within about 30 minutes of coming up and starting the backups
the machine is responsible for.  Since it seems repeatable
(happened twice in a row before I reverted), I should be able
to test, although I'm not very fond of fsck'ing this machine.
There were a lot of changes between 4.0_BETA and 4.0_BETA2, of
course, so I'm not sure what to look at.

The debugger tells me this:
Stopped at      netbsd:cpu_Debugger+0x4:        popl    %ebp
db> t
cpu_Debugger(c4d95400,0,c4da96c0,6,c4d95460) at netbsd:cpu_Debugger+0x4
comintr(c4d95400,6,10,c4f40030,10) at netbsd:comintr+0x70a
Xintr_ioapic_edge4() at netbsd:Xintr_ioapic_edge4+0x97
--- interrupt ---
DDB lost frame for netbsd:Xdoreti, trying 0xc0ae4a04
Xdoreti() at netbsd:Xdoreti
--- interrupt ---
0x206:
db> c
Stopped at      netbsd:cpu_Debugger+0x4:        popl    %ebp
db> t
cpu_Debugger(c4d95400,c4ea9050,c4da96c0,6,c4d95460) at netbsd:cpu_Debugger+0x4
comintr(c4d95400,6,10,c4f40030,10) at netbsd:comintr+0x70a
Xintr_ioapic_edge4() at netbsd:Xintr_ioapic_edge4+0x97
--- interrupt ---
DDB lost frame for netbsd:Xdoreti, trying 0xc0ae4a04
Xdoreti() at netbsd:Xdoreti
--- interrupt ---
0x206:
db> 

(I am dropping it into the debugger, of course.)
What else can I do to help find out what is going on?

-Tracy