This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
[Bug ipa/66122] Bad uninlining decisions
- From: "pinskia at gcc dot gnu.org" <gcc-bugzilla at gcc dot gnu dot org>
- To: gcc-bugs at gcc dot gnu dot org
- Date: Mon, 18 May 2015 18:46:40 +0000
- Subject: [Bug ipa/66122] Bad uninlining decisions
- Auto-submitted: auto-generated
- References: <bug-66122-4 at http dot gcc dot gnu dot org/bugzilla/>
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66122
--- Comment #12 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
(In reply to Markus Trippelsdorf from comment #11)
> The compiler isn't psychic, e.g. it doesn't parse asm statements at all (so
> it cannot know how many insn it contains).
Actually it does some parsing just to see how many statements. That is it uses
';', and newline as an estimate of how many statements there are. And then
uses that for an estimate in the cost (I know it does this because I added this
support). But there are some many different heuristics for the inliner and it
could decide if it is one instruction not to inline it anyways for other
reasons.