[PATCH] Fix PR rtl-optimization/21041 (yet another reload problem)

Ulrich Weigand uweigand@de.ibm.com
Sun Dec 18 16:11:00 GMT 2005


Roger Sayle wrote:

> This is OK for mainline, 4.1 and 4.0.  The patch looks reasonable,
> and I'll defer to your reload expertise that this is the correct
> fix.  Is there any chance we'll generate worse code with this patch
> or fail to find reloads on some !STRICT_ALIGNMENT targets?

Well, in those cases where we need to make the replacement (eliminable
registers), the result should be a properly aligned stack slot, so we
should be fine.  In other cases where we cannot prove proper alignment,
I guess the old behaviour would have been a latent bug anyway ...

> It might
> make sense to apply this patch to mainline, wait a few days and if
> nothing bad happens backport to 4.0 and 4.1 once we're confident
> its safe.

OK, I've applied the patch to mainline now, and will wait until the
end of the week before backporting.

Thanks,
Ulrich

-- 
  Dr. Ulrich Weigand
  Linux on zSeries Development
  Ulrich.Weigand@de.ibm.com



More information about the Gcc-patches mailing list