freebsd stable - Mar 2006 - Segfaults from gcc, awk and Zend; advice needed

For some reason I'm getting more or less random segfaults compiling  
kernels, world or PHP5 on BETA-3 (Python and perl went ok). So far I  
haven't succeeded building a fresh kernel or world. This system is an  
Athlon XP with 1MB RAM and 4GB swap, compiling is done in the usual  
places in /usr, which is a set of two gstriped partitions. Dmesg  
attached.

Previously this system has been running 5-STABLE w/o any such  
problems (uptime before installing 6 was about 80 days, started from  
an installworld). This machine doesn't regularly see load yet, except  
for the few build(world/kernel)/install(&)s.

I haven't seen any other messages of this kind, so I suspect a local  
problem. So far I think I need to look at:
- memory; run memtestx86 and cross fingers
- I recollect seeing mention of problems with large amounts of swap  
on this list in the past; turn one of the swap partitions off or  
something like that
- Optimization flags; I've tried setting CPUTYPE?=athlon-xp and  
CLFAGS=, CFLAGS=-O -pipe, CFLAGS=-O2 -pipe so far. All w/o avail...
- gcc 3.4.4, awk, PHP are all borken (all were built using gcc 3.4.4,  
I suppose)

I'm a bit low on spare time, so I'd like to tackle this problem  
efficiently. Anything I forgot or to help me find the culprit?

Regards,

-------------- next part --------------
A non-text attachment was scrubbed...
Name: dmesg.out
Type: application/octet-stream
Size: 5879 bytes
Desc: not available
Url :
http://lists.freebsd.org/pipermail/freebsd-stable/attachments/20060313/af46a02c/dmesg.obj
-------------- next part --------------

--
Alban Hertroys

		Sometimes you wake up and you think:
		"Galileo was right,
		 The world does rotate"

Alban wrote:
> For some reason I'm getting more or less random segfaults compiling
> kernels, world or PHP5 on BETA-3 (Python and perl went ok). So far I
> haven't succeeded building a fresh kernel or world. This system is an
> Athlon XP with 1MB RAM and 4GB swap, compiling is done in the usual
> places in /usr, which is a set of two gstriped partitions. Dmesg attached.
If the crashes aren't repeatable (ie, the compiler segfaults in different
places
if you re-do the make), that's an almost sure sign of hardware problems like
overheating.
> - Optimization flags; I've tried setting CPUTYPE?=athlon-xp and
CLFAGS=,
> CFLAGS=-O -pipe, CFLAGS=-O2 -pipe so far. All w/o avail...
If you're having problems, use the default compiler flags.  Trying to add
machine-specific optimizations into the mix is going to introduce spurious
issues and make it harder to figure out the real problem.
> - gcc 3.4.4, awk, PHP are all borken (all were built using gcc 3.4.4, I
> suppose)
> 
> I'm a bit low on spare time, so I'd like to tackle this problem
> efficiently. Anything I forgot or to help me find the culprit?
I suppose you could wait until 6.1 is released and do a binary install/upgrade?

-- 
-Chuck