hamm@austoto.sps.mot.com (Steve Hamm) (01/11/91)
I am in danger of losing what little sanity I have left, in dealing
with the 6.8 C compiler. To be exact:
C compiler 68K Rev 6.8(105) - cr1.0 beta
run from an SR10.3 BSD environment using /bin/cc.
I have a relatively large program that refuses to act the same on the
Apollo as it does on other platforms. I can't get the program to fail
consistently: Run from dde and it dies one way; run from the shell and
it dies differently. Doing things in a different order causes
different failures.
(For anyone who's interested, the program is a Motorola internal
modification of Berkeley SPICE 3b1. If anyone has compiled SPICE 3b
for Apollo and gotten around the more obvious Berkeley bugs, I can
send you some benchmark circuits...)
I have run using the malloc.dbg stuff to check that memory isn't
getting trashed; it isn't. I'm suspicious of trashing the stack,
though, but can't figure out how to check that. A colleague wasted a
couple of weeks of his life trying to track this down.
Just for reference, the program runs fine on all flavors of Suns and
the IBM RS/6000. Also for reference, it does the same thing, except
more often, under SR10.1. I'd prefer to not even speak of SR9.7.
I'd like suggestions on where (and perhaps how) to look for the
problem. I'd really like to kill this problem, preferably through
slow torture. Any help GREATLY appreciated.
--Steve
--
Steve Hamm ------- Motorola Inc. Semiconductor Systems Design Technology
3501 Ed Bluestein Blvd., MD-M2, Austin TX 78762
Ph: (512) 928-6612 Internet: hamm@austoto.sps.mot.com
Fax:(512) 928-7662 UUCP: ...cs.utexas.edu!oakhill!austoto!hamm