This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Reload problem
> > If we are seriously concerned about performance there, we can cache the
> > insn code or even the entire emitted instruction for use by gen_reload.
>
> I think that would be a good idea, if only to reduce stress on the
> memory manager.
Is it really possible to cache the entire insn? If we did so, the insn would
be frozen at the end of push_reload. Now it turns out that gen_reload calls
find_replacement on both operands of the PLUS, so if some replacement was
scheduled after the reload had been pushed, I think we would end up with
an incorrect insn.
--
Eric Botcazou