Re: linux-kernel-digest V1 #180

root (
Tue, 10 Oct 1995 01:07:31 -0700 (PDT)

> > From: (root)
> > Has anyone had experience with the "Linux Sudden Death Syndrome?"
> >
> > Symptoms include:
> > [...]
> > Does anyone know causes and perhaps cures for this mysterious syndrome?
> Yes and I'm getting utterly fed up of repeating how to get diagonostic
> traces of this to people so they can find out why their specific system
> crashed.

Stop. Lie down. Relax. Take deep breath. Exhale. Repeat.
There. Now don't we feel better?

> Boot without syslogk( or klogd as its sometimes called). Kernel messages
> will go to screen - tough live with it for this one

Done. Long time ago. Learned the hard way from tracing NE2000 panics.

> Wait till the box dies - read any panic messages of screen (they may not
> get logged to disk if for example it was a disk panic). If nothing is
> going on hit right-alt scroll lock, right-shift scroll lock and
> right-ctrl scroll lock. Write down the info. The CPU dump (the one,
> starting EIP:) do about 20 times and write down the EIP value for each.
> Reboot
> Look the EIP values for each up in package and mail. That way
> a developer can see where there kernel is going at the time.

If this information were easily available, I wouldn't have asked. I knew
the key combo existed but it isn't documented anywhere I could find. And I
dug. And dug. And web surfed. For hours. And never found anything.

/usr/src/linux/Documentation/kernel-debug.txt anyone?

It's amazing how many people don't know you can scroll VC's, and even fewer
know the debugging key combos. Developers could save themselves lots of
trouble by .. gasp .. DOCUMENTING this stuff. 8)

Of course, if it's documented somewhere, it isn't obvious. I didn't find it
in any of the HOWTOs, etc..

Now that I have all my ducks in a row (so to speak), next time we have
LSDS, I'll be prepared enough to get debugging info out of the kernel.