RS6000 problem in egcs

[meissner@cygnus.com]

|   In message <199708281932.PAA23113@ahab.rutgers.edu>you write:
|   > With Ian's help I now have the real stacktrace for the RS6000 problem:
|   > 
|   > cc1 920413-1.i -quiet -dumpbase 920413-1.c -version -o 920413-1.s
| I just tried this with the 970901 snapshot on a ppc and it doesn't
| fail; can you verify? (I assume you're refering to compile/920413-1.c?)

As I recall, the bug was against an AIX 3.2 system.  Note there are some
gotchas noted about earlier AIX systems in README.RS6000, particularly when
using the host compiler.  Do any of those apply?  Also, if the system is not at
least 3.2.5, gcc could have suffered decay when the 3.2.5 -mcpu=common support
went in years ago.

(Jeff, Kalessin is a 3.2.5 system).