hardware reboot

[Alex S. Crain]

>Where are those register dumps documented anyway? I'd much rather
>interpet them myself than depend on some bozo in Support Land to
>tell me what happened. 

	I would doubt that they are documented at all. When my machine 
first did it, I called Support land and they said that I had a hardware 
problem, and gave me a new motherboard. I've had 3 with the new board,
so I don't think that it helped.

	There are only a few things that can go wrong with the kernal, and
usually the panic says what happend; ie:

	page fault in kernal - these are cused by kernal bugs
	parity errors - these are caused by hardware bugs.
	running out of things, inodes, clists, etc, these are when the
system overloads past its configuration. Use ktune, or fix the user program
that caused it.

	There are a few more, but thats the jist of it. But since we can't 
recompile the kernal, nad only a few of us are willing to adb our way through
it over an occation panic, the second and third types are the only ones that
matter.

	The problem is that the kernal has bugs, just like every other program
of its size and complexity, and no one will ever fix them, ever. If you knew
what they were, what good would it do?

	If it helps, I've had 3 panics, all of which seem to be associated
with branching from bogus function pointers in a user program. I can't 
duplicate them, they're like buss errors gone astray, and rerunning the 
user program doesn't cause a panic. The message is always "page fault in
kernal" and I don't get a core image. I suppose that I could use the PC
to calculate where the kernal barfed, but since I don"t have source, and
those with source don't care....

-- 
					:alex.

nerwin!alex at umbc3.umd.edu
alex at umbc3.umd.edu