[PATCH] Fix store motion

[Andreas Jaeger]

Daniel Berlin <dan@cgsoftware.com> writes:

[...]
> Store motion can't possibly work anyway.
> Because the test for set regs is reversed, it can only cause the
> exact wrong thing to happen.
> This appears to have happened because the code for store_ops_ok was
> originally copied from expr_killed_p.  Of course, as the names imply,
> store_ops_okay returns 1 for okay expressions, expr_killed_p returns 0
> for okay expressions.  Somebody forgot to reverse the test, to take
> this into account.
> Store motion *should* be disabled until it is fixed (reversing the test exposes
> many other bugs).
> I submitted a one line patch to do this, but it was never reviewed.

Which one?  I found only this one:
http://gcc.gnu.org/ml/gcc-patches/2001-08/msg00268.html

Can you send the patch again, and let's hope that somebody reviews it
or looks at those issues and fixes them.

Thanks,
Andreas
-- 
 Andreas Jaeger
  SuSE Labs aj@suse.de
   private aj@arthur.inka.de
    http://www.suse.de/~aj