Re: 2.1.119 on alpha

[Richard Henderson <rth cygnus com>]

> When doing embedded debugging, I'd be getting out the scope and the schematics
> and looking for bus activity to see where things are locked up ...

Unfortunately, I don't really have anything in the way of hardware
for debugging.  What I wouldn't give for a connected JTAG...

> Have you narrowed the problem at all?

The problem _appears_ to be complete vapour lock with interrupts
enabled. 

I now know that it isn't the kernel corrupting the PALcode, which was
my first guess.  I know that it probably isn't a simple uninitialized
stack pointer read starting the thread of destruction.  Past that, I
have little in the way of solid facts.

I have, however, collected several new interesting tools for sanity
checking the kernel, so I guess it hasn't been a complete waste of
time.  But it sure feels like it.

Jay just got back from vacation and mentioned that the PC164 has
broken interrupt hardware, in that it cannot mask interrupts.  So
if you don't get the device to shut up, you'll continually get
sent back to the interrupt handler.  A hypothesis that does fit 
all the facts to date.

I'll see what I can see about that tomorrow.  Anyway, it is the only
good lead I have at the moment.


r~