This is the mail archive of the
mailing list for the GCC project.
Re: [RFC] Early inlining pass
- From: Richard Guenther <richard dot guenther at gmail dot com>
- To: Jan Hubicka <jh at suse dot cz>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Wed, 1 Jun 2005 10:27:03 +0200
- Subject: Re: [RFC] Early inlining pass
- References: <20050531230850.GM11574@kam.mff.cuni.cz>
- Reply-to: Richard Guenther <richard dot guenther at gmail dot com>
On 6/1/05, Jan Hubicka <firstname.lastname@example.org> wrote:
> this patch adds early inlining pass that inline the "obvious" stuff (ie
> functions having body smaller than the call construct and always inline
> functions). This is followed by cleanup_cfg and true inlining pass.
> And finally I am not sure if the idea sounds like good way to go, but I am
> overall quite surprised that it is somewhat effective even for "common" code,
> not only tramp3d it was originally designed as a quick hack for, so I don't
> seem to come with good reasons against..
I think early inlining of wrappers & simple accessor methods in C++ is very
much important if profile-driven inlining is supposed to be useful. Also only
with early inlining pre-inline optimizations can do useful work (as you noted),
and we may close the gap a little between the size estimate heuristics useful
for C and C++ code.
If this patch goes in in some form, I promise to do an extensive analysis of
code size estimate vs. text size again, for C and C++ testcases.
Thank you very much!