On Tue, May 07, 2002 at 03:35:27PM +0200, Thomas Runge wrote:
> > On Tue, May 07, 2002 at 07:46:28PM +1000, Murray Armfield wrote:
> 
> > Although, can anybody explain why those kernels usually fail?
> 
> Smells a bit like the mystical Sig11 problem:
> 
> http://www.bitwizard.nl/sig11/

Useful, but don't let it lead you to assume hardware failure every time
you see a sig11 or similar problem.

The fact is, it's much easier to blame and fix hardware than it is to
find obscure kernel and compiler bugs.

So, yes, check your hardware; most people forget this.  

However, I too have occasional problems with install kernels, especially
FreeBSD, and on known good hardware.  I have always blamed this on kernel bugs
simply because an install kernel has a lot more code in it than my custom
kernels do.  I just assume it's an issue of exposing more bugs and dependency
problems otherwise known as bugs.

If you have trouble with an install but your system runs fine after
that--and you use all available memory and devices--then the problem is
not likely to be your hardware. I have had many systems run fine for
years, despite several problems during install, including the infamous
sig11 bug.

> The problem gets even worse now that we have UBC integrated. We are now
> using all available memory to cache data and faulty memory shows up quite
> early.

I've found this does catch more systems with faulty memory.

Of course, it has also caught more kernel bugs too.

-- 
UNIX/Perl/C/Pizza__________________________________shannon@widomaker.com