Re: Random signals in {build,install}world recently?

On Mon, Oct 20, 2003 at 11:57:48PM -0700, Terry Lambert wrote:
> Barney Wolff wrote:
> > I don't think so.  I tried that on my A7M266D with no effect.  I believe
> > something in recent pmap code doesn't like this mobo, or maybe dual
> > athlons in general.  I can run RELENG_5_1 rock solid, and -current from
> > 9/24/03 rock solid, but -current from 10/3 or later gets random sigs
> > and eventually panics.  I have scsi disks so it's not ata.
> 
> I think you need to define "random"; do you mean "rare in frequency
> over time at unpredicatable intervals" or "you never know what
> program is going to get shot in the head, every 5 seconds, like
> clockwork"?

The latter, with the interval being more like a very few minutes rather
than seconds.  The system would sit fine when idle, but never make it
all the way through buildworld, or buildkernel, or building a port.
In one notorious instance, the -current kernel didn't even make it
through installworld without a sig.  grep, sed, awk, sh have all gotten
sigs, not just cc.

> My impression so far in this therad is that it's the former.  If
> it's the latter, then I need to think about the problem differently.
> 
> 
> Note that you can identify the patch that caused the problem, if
> there's an 8 day difference, in no more than 4 kernel recompiles
> (log2(8)+1), if you have a local CVS mirror.

The frequency of patches to pmap was quite high, and that was not the
only file patched.  But yes, I can do a binary search.  I haven't,
because the changes to pmap were architectural; we're not talking about
a typo here.  Of course it's just my assumption that it's pmap because
the eventual panic is there.

btw, I use sched_bsd so _ule is not a suspect.

-- 
Barney Wolff         http://www.databus.com/bwresume.pdf
I'm available by contract or FT, in the NYC metro area or via the 'Net.