Re: Debugging the kernel when it doesn't crash

Martin von Loewis (martin@mira.isdn.cs.tu-berlin.de)
Sun, 29 Jun 1997 22:05:32 +0200


> I would like to give some info to the developers but I cannot find how to
> do it. Is there a way to force a crash dump, for instance? How to help
> debugging that version on the Bleeding Edge?

This sounds like a memory leak somewhere: Something eats more and more
memory, until only little memory is left to the applications. I don't
know whether there are already ideas about searching memory leaks in the
kernel; a good start would be watching /proc/slabinfo. If you see that some
pool continually grows, you know at least what kind of object consumes all
the memory. Finding out where it was allocated is left as an exercise :-)

Regards,
Martin