Re: [coLinux-users] Random signal 11 problems?

[Doctor B. <do...@gm...>]

Yes.  The general problem with the FAQ is it assumes the gcc and
kernel have been well tested under all possible configurations.  I
have seen sig 11 problems on many different hardware platforms in many
different cases.  In most cases I attributed it to an inadequately
tested configuration, and the fact most people do not report the sig
11 problem when it occurs.  Why?  Well if the location of the error
depends on the hardware configuration you are using, it is unlikely
someone else would be able to successfully debug the problem without
access to the same machine.  Sometimes it is repeatable on the same
file, regardless of current memory usage, sometimes not.  Most
commonly I've seen the problem under Solaris, where the bootstrapping
process is very particular about what set of bin utils you use, what
compiler you bootstrap from etc.  It seems after each major upgrade of
gcc, the bootstrapping order I need to follow to get a compiler that
does not sig 11 under Solaris is always slightly different.

Does that mean it isn't a hardware problem?  Not at all.  It could be
that the many different boxes I've seen the problem in the past all
share a similar design flaw, and finding a way of doing a majic build
of gcc that does not have the problem is just working around the
problem.  It is also possible that in the upgrade, someone added code
to workaround the hardware problem.  Possible, but not likely.

Under coLinux, I believe the most common culprit is corrupt or heavily
fragmented swap files.

                                  Bill


On Thu, 09 Sep 2004 09:51:36 -0700, Chuck Adams <ch...@da...> wrote:
> Sandy Harris wrote:
> 
> > It may be outdated and may not apply to your
> > problem, but it is probably worth looking at
> > the sig 11 FAQ:
> > http://www.bitwizard.nl/sig11/
> 
> 
> Actually it's almost never worth looking at that FAQ.  It's pure
> baloney.  I remember this FAQ getting thrown at me before ... Yes,
> memory glitches may cause segfaults and if random apps are failing at
> random times, it's worth running memtest.  But there are claims in that
> FAQ that are just absurd.  I always found it odd how it was *just* gcc
> that somehow "stressed the memory" and therefore just gcc that would
> segfault.  Occam's razor was right all along of course -- it was a bug
> in gcc.  Either that or the next upgrade magically fixed my hardware.
> 
> chuck
> 
> -------------------------------------------------------
> This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170
> Project Admins to receive an Apple iPod Mini FREE for your judgement on
> who ports your project to Linux PPC the best. Sponsored by IBM.
> Deadline: Sept. 13. Go here: http://sf.net/ppc_contest.php
> 
> 
> _______________________________________________
> coLinux-users mailing list
> coL...@li...
> https://lists.sourceforge.net/lists/listinfo/colinux-users
>