[daip] Wierd Solaris compiler problem
Michael Bietenholz
michael at polaris.phys.yorku.ca
Thu Sep 18 12:15:14 EDT 2003
(Not that you guys need more compiler problems...)
I just found a slightly worrying thing: I was running BPASS and if
failed mysteriously (ie. ZABORS). I tracked it down to CHNDAT and/or
CHNCOP.
BPASS1: ZABORS: signal 10 received
BPASS1: ABORT!
Poking around, I found the following:
It worked when I COMRPL'ed CHNCOP and CHNDAT with "NOOPT", and failed
when I COMRPL'ed as usual (ie. with whatever the default optimiation
is). The optimization used when COMLNK'ing BPASS didn't seem to
matter.
This is on:
SunOS polaris.phys.yorku.ca 5.7 Generic_106541-25 sun4u sparc SUNW,Ultra-5_10
using f77: WorkShop Compilers 5.0 98/12/15 FORTRAN 77 5.0
ie., this is what COMRPL NOOPT chose:
/opt/SUNWspro/bin/f77 -c -fns -fsimple=1 -xarch=v8plusa -xlibmopt -dalign
-ansi -u -ansi -u -O3
Now I'd be happy enough about my recompiled albeit marginally slower
version of CHNDAT and CHNCOP, but I'm a little worried about what else
might be going wrong that doesn't happen to cause an ABORT
michael bietenholz
More information about the Daip
mailing list