This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH] Use alias-oracle for memmove -> memcpy folding
On Wed, 2 Sep 2009, Eric Botcazou wrote:
> > I failed to test ada and the patch caused some regressions there
> > (basically using the oracle when we're not yet in gimple is fragile
> > due to the FEs not setting TREE_ADDRESSABLE properly and also due
> > to extra weird trees popping up). The proper fix is to move this
> > folding to fold_stmt only - I don't have time for this right now
> > so I have reverted the patch.
>
> Thanks. I presume you're referring to this couple of new failures:
>
> FAIL: c52102a
> FAIL: c52102c
Yes.
> If there is something we can do differently/better in gigi...
Naa - I was just being overly optimistic that the oracle can deal
with frontend trees just fine. Obviously overly optimistic ;)
Richard.