This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: RFA: redirect_jump_1: fix REG_EQUAL notes
- From: amylaar at spamcop dot net
- To: Roger Sayle <roger at eyesopen dot com>
- Cc: gcc-patches at gcc dot gnu dot org, joern dot rennecke at st dot com
- Date: Tue, 1 Mar 2005 22:50:26 -0500
- Subject: Re: RFA: redirect_jump_1: fix REG_EQUAL notes
> Any reason not to move invert_br_probabilities into redirect_jump_2
> as well, controlled by the "invert" parameter?
- slightly more code to execute in redirect jump
- more code churn (i.e. larger patch)
Not really strong reasons, I'm happy to include that change.
> optimization (at least not in our lifetimes :). Instead, how about:
>
> XEXP (note, 0) = simplify_replace_rtx (XEXP (note, 0), olabel, nlabel);
- This won't work. LABEL_REF is a RTX_CONST_OBJ, so simplify_replace_rtx
won't (and shouldn't) change the inside. And since LABEL_REFs may be shared,
but need not be unique, no LABEL_REF that you pass to simplify_replace_rtx
or replace_rtx is 'right' .
- simplify_replace_rtx does simplifications, which I don't think make any
sense in this context. Its overhead is likely higher than that of
redirect_exp_1.
- redirect_exp_1 can also translate RETURN.