> Yesterday my 1.5 install welcomed me home  with 19 inches of "panic:"

> I'm unable to break to debug, it's completely frozen.

> I previously had problems with locks like this, [...]

> It doesn't seem to be related to temperature, but I seem to be able
> to force a crash within a day or two by generating alot of CPU
> activity, while a idle machine usually can stay up for about a week.
> If I leave it in the boot prompt it'll stay up forever.

> Anyone experienced similar problems ?

This sounds reminiscent of my experiences.

I diverted a 2100 on its way to the dumpster at a local university.  I
set it up with a binary install off the 1.4.2 CD and tried to build the
same 1.4T source tree I run everywhere else.

And it would lock up hard.  Just freeze at an apparently arbitrary
place.  I observed a similar envelope to what you sketch above; it
would stay up "forever" (I didn't wait more than about a day) at the
single-user prompt, but trying to compile anything would take it out
fairly quickly.  And it's not just the console going out to lunch the
way my macppc does; I brought up its ethernet, and when it locks up it
stops answering pings.

I opened up the case and propped up a small fan blowing directly on the
CPU, thinking the problem might be thermal.  No help.

Then someone was kind enough to send me a 3100.  (Thank you - you know
who you are!)  I left it in its case, opening it only long enough to
connect up SCSI and disk drive power to cables that would reach outside
the case.  I moved the disks from the 2100 to it.  And it hangs in
exactly the same way.

It's a hard hang; a serial-line break doesn't do anything.  And as I
said, it stops responding to pings.

I suppose it's *possible* that I've got two machines that are broken in
exactly the same way.  It seems pretty unlikely, though.  But then, it
also seems pretty unlikely that this is just the way 1.4T works - was
there a time a year and a bit ago during which the pmax kernel was
broken?  I may have frozen at an unfortunate spot.

And in any case, does anyone have any idea how I might go about
debugging it?  I hate debugging hard hangs, because so little
post-mortem information is available....

/~\ The ASCII				der Mouse
\ / Ribbon Campaign
 X  Against HTML	       mouse@rodents.montreal.qc.ca
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B