RS6000 vs. egcs

[meissner@cygnus.com]

| I now have a version of egcs which is working better.  It turns out
| that I was bitten by a bug in an alpha version of g77, which caused
| all kinds of problems.  I now have 970901 (the move-if-change patch
| works) and the output of "make check" is:

	...

Of the gcc tests, the following I know about and are long standing:

compile/961203-1.c	is a generic bug, fails for all platforms
execute/931004-11.c	is a bug in passing small structures
execute/931004-12.c	is a bug in passing small structures

I've seen execute/{ieee,fp-cmp-1,complex-5}.c fail for other architectures as
well as the rs600/powerpc.