From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-bugzilla@gcc.gnu.org>
Received: by sourceware.org (Postfix, from userid 48)
	id 25299384CB97; Wed, 15 May 2024 09:05:46 +0000 (GMT)
DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org 25299384CB97
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gcc.gnu.org;
	s=default; t=1715763946;
	bh=RRtWpqlmqsg0posJsxhFZsw2v1B9MoRSqPSMasXfzco=;
	h=From:To:Subject:Date:In-Reply-To:References:From;
	b=Nvjhsp/bAYeLhDM8NPJp1bzYhqfWZ7QvyMq1zpVN2CJUYSIOr3zYxTMJEOu/VbU64
	 mXDQkviPxrciSZ/Gf9GOLpdF3vytKNgCBQpaIElCW9iuHxYygb9AZER9u9V17VbWTL
	 UQIWYyQAhRm3Z0+2N5CDTRfok2IEt0akYgUbiixE=
From: "hubicka at ucw dot cz" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug ipa/115097] Strange suboptimal codegen specifically at -O2 when
 copying struct type
Date: Wed, 15 May 2024 09:05:43 +0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: ipa
X-Bugzilla-Version: 15.0
X-Bugzilla-Keywords: missed-optimization
X-Bugzilla-Severity: normal
X-Bugzilla-Who: hubicka at ucw dot cz
X-Bugzilla-Status: RESOLVED
X-Bugzilla-Resolution: DUPLICATE
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org
X-Bugzilla-Target-Milestone: ---
X-Bugzilla-Flags: 
X-Bugzilla-Changed-Fields: 
Message-ID: <bug-115097-4-uXDZuvusFf@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-115097-4@http.gcc.gnu.org/bugzilla/>
References: <bug-115097-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
List-Id: <gcc-bugs.sourceware.org>

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=3D115097

--- Comment #7 from Jan Hubicka <hubicka at ucw dot cz> ---
> and then we inline them back, introducing the extra copy.  Why do we use
> tail-calls here instead of aliases?  Why do we lack cost modeling here?
Because the function is exported and we must keep addresses different.
Cost modeling is somewhat hard here, since it is not clear what will
help inliner and whether inliner will inline back.

For example if we icf two function calling third function. the third one
may become called once and we get better code by inlining it and eliding
offline copy rather than keeping the caller duplicated.

So the idea is to get code into as canonical form as possible (with no
obvious duplicates) and then let the inliner heuristics to decide what
should and what should not be duplicated in the final code.=