On Thu, 21 May 1998 11:57:22 +0200 (CEST),
henning loeser <loeser@ma1304.physik.uni-marburg.de> wrote:
> Sorry it took me a while to get to this again. I had to write some papers 

No prob.

> To the configuration :
> Q700, 20MB Ram, 2MB Video (at 256 greys), MacCon Nubus Ethernetcard, HD.
> 
> After doing a make build again on the 1.3.1 sorces and waiting for about 
> 26 hours I exited from single user and the system seemd to come up fine.
> I could log in from console. Starting the xdm was when the problem 
> occured again. Ken's Xmac68k.OSFA server duped core, gdb gives:
> 
> #(gdb) core Xmac68k.core
> #Core was generated by `Xmac68k'.
> #Program terminated with signal 11, Segmentation fault.
> #Cannot access memory at address 0x40a0.

Hmm.... This very strange...  I have no idea what was going on.

> Reinstalling the base.tgz got everything working again.
> May I not use the -O2 -m68040 options on compilation time ?

I don't have full confidence in GCC's optimization.  There's at least
one program that gives very different results when compiled with -O0
(no optimization) and with -O2.  The binary compiled with -O0 works as
expected but the one with -O2 doesn't.

However, NetBSD is compiled with -O2 by default and we don't usually
have any problem with the system, so it is conceivable that -m68040 is
havinga a bad interaction with -O2...  Maybe we should wait until
mac68k switches to egcs?

Sorry for not being able to help much...

Ken