950503-1.c CSE bug on PowerPC / rs6000

[David Edelsohn]

	This testsuite case has started failing on PowerPC / rs6000.
Looking at recent testsuite result postings, I am a little confused why it
is not failing consistently, but maybe there is confusion over which
snapshot actually is being tested.  The failure has been shown in both AIX
and Linux testsuits reports.

	When any optimization is enabled, the executable solely calls
abort() without any computation.  The actual computation and comparison is
removed in CSE phase, so something seems to be going wrong there.  The
.jump dump looks okay, but .cse dump does little more than call abort().

David