Re: Patch fixing 3.3 bug PR 9745 and PR 10021

>>>>> Mark Mitchell writes:

Mark> It's up to you and Geoff (as RS6000 maintainers) to decide whether the
Mark> RS6000 port is better off generating bad code or slow code.

Mark> We are going to be stuck in one state or the other until someone is
Mark> willing to invest some serious effort in this problem.

	You need to look at the Bugzilla report more carefully: PR 10021
affects m68k, not PowerPC, but it is a manifestation of the same problem.
The aliasing bug also affects S/390, although there is not a PR at the
moment.  Basically the bug affects all ports.  Once Apple's IMI patch or
something similar is merged into GCC, this bug will be much more prevalent
because it's effect is exacerbated by aggressive inlining within loops.

	If Jim's patch or something similar is applied, GCC will decrease
from being somewhat competitive on performance to completely
uncompetitive and uninteresting, which will drive users to proprietary
compilers that are compatible with GCC.

	Expanding on Jim's original analysis, we need to find a way for
the current GCC aliasing infrastructure to correctly deduce and return a
valid ADDRESS RTL instead of a naked hard register.

David