Hi,

the story goes on :-) - today, with the patch from Antti Katee, the
machine survived for about 4 hours, but then it stopped responding again.

The console was scrolling the following messages:

Alignment error: pid=4814.1 comm=milter-greylist dsfsr=00000000:00800003 dsfar=32e:63697489 isfsr=00000000:00000000 pc=4090bc28
Alignment error: pid=4814.1 comm=milter-greylist dsfsr=00000000:00800003 dsfar=32e:63697489 isfsr=00000000:00000000 pc=4090bc28
Alignment error: pid=4814.1 comm=milter-greylist dsfsr=00000000:00800003 dsfar=32e:63697489 isfsr=00000000:00000000 pc=4090bc28
Alignment error: pid=4814.1 comm=milter-greylist dsfsr=00000000:00800003 dsfar=32e:63697489 isfsr=00000000:00000000 pc=4090bc28
Alignment error: pid=4814.1 comm=milter-greylist dsfsr=00000000:00800003 dsfar=32e:63697489 isfsr=00000000:00000000 pc=4090bc28

- which doesn't actually surprise me, as I *know* that milter-greylist
is giving me lots of problems.

If I understand the code correctly, I see these messages now because I
didn't run a kernel with "option DEBUG" before - and the offending line
of code *should* be found with "list *0x4090bc28" from inside gdb?


Well - I can't find it, but that seems to be due to "it's somewhere
inside a shared library"...

I wouldn't mind if the kernel would actually just SIGSEGV/SIGBUS the 
offending process that produced the alignment error (and have it core 
dump, which is something I'm more familiar in debugging...)

Reading the kernel sources, this should be possible, but it depends on
"(l->l_proc->p_md.md_flags & MDP_FIXALIGN)", which is something I don't
know how to set/clear...  - so how to make the kernel kill processes
that show alignment misbehaviour?

thanks for your advice,

gert

-- 
USENET is *not* the non-clickable part of WWW!
                                                           //www.muc.de/~gert/
Gert Doering - Munich, Germany                             gert@greenie.muc.de
fax: +49-89-35655025                        gert@net.informatik.tu-muenchen.de