Re: 2.6.8-rc3-mm2

[William Lee Irwin III <wli@holomorphy.com>]

* William Lee Irwin III <wli@holomorphy.com> wrote:
>> This is serial port IO; would that make the same kind of difference?

On Tue, Aug 10, 2004 at 03:01:22PM +0200, Ingo Molnar wrote:
> serial port IO is even more heavy, it also generates IRQ traffic. I'd
> suggest the following tests: modified kernel without serial console, and
> original kernel without serial console. This way you can find out 
> whether the hang itself is triggered by the serial console.
> another attack angle is to find out what state the system has during the
> hang. Is it debuggable?

I can get a very small amount of register state out of POD but most of
it is clobbered by trapping with the NMI/MCA I send from the L2 to get
into the POD (Prom Only Debugger), and it's not entirely clear what may
have been clobbered in this process. The state I can extract from this
is very difficult to interpret. It's unclear how well the consolidated
L2 console (a.k.a.  firmware console) and machine console interoperate
with e.g. kgdb and the like. I might as well try that, too.

It's unclear how much of the progress I can track without the serial
console as there are no other output devices besides the serial console.
I'll try it anyway, since the worst that can happen is a boot where it
crashes.


-- wli