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 8CEAB3858431; Wed, 30 Aug 2023 08:16:05 +0000 (GMT)
DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org 8CEAB3858431
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gcc.gnu.org;
	s=default; t=1693383365;
	bh=Vp7ajhcP7SS509PK/+//5Q1z5a4l2HR2Juj5MDccGew=;
	h=From:To:Subject:Date:In-Reply-To:References:From;
	b=u1nMXdMs79QZMQ1MTkJ2z6rgWBJ6GQXe11Y70BT+r+cRtCy1vtqu7fsBpkbAQROY2
	 KotEE0n2z1Ve74DDwEmnKanWw6YDiw2vw8tPcPtd31kihimUMn0eKQvefbupOH6oxH
	 6XAzT/UzUXbB2tfDi1tNg5pbcXGbQJ3wsm1lIyvg=
From: "gjl at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug rtl-optimization/110093] [12/13/14 Regression][avr] Move frenzy
 leading to code bloat
Date: Wed, 30 Aug 2023 08:16:02 +0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: rtl-optimization
X-Bugzilla-Version: 14.0
X-Bugzilla-Keywords: missed-optimization, needs-bisection, ra
X-Bugzilla-Severity: normal
X-Bugzilla-Who: gjl at gcc dot gnu.org
X-Bugzilla-Status: NEW
X-Bugzilla-Resolution: 
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org
X-Bugzilla-Target-Milestone: 12.4
X-Bugzilla-Flags: 
X-Bugzilla-Changed-Fields: 
Message-ID: <bug-110093-4-4z3gpJqzMo@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-110093-4@http.gcc.gnu.org/bugzilla/>
References: <bug-110093-4@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=3D110093

--- Comment #4 from Georg-Johann Lay <gjl at gcc dot gnu.org> ---
(In reply to Vladimir Makarov from comment #3)
> I propose to avoid the above RTL code by switching off subreg3
> pass (or -fsplit-wide-types) for AVR by default as it was for gcc-8.

Thanks for looking into this.

With v8, I don't see a difference with -f[no-]split-wide-types, everything
works fine.

Since v10 r280033 the default is -fsplit-wide-types-early, but that option =
has
no effect on testcase + master, only -fno-split-wide-types seems to "fix" t=
he
problem, regardless of -f[no-]split-wide-types-early.

>From my experience, -fno-split-wide-types has no clear edge over
-fsplit-wide-types, which very much depends on the code.  This is the reason
why -fsplit-wide-types is still the default.

So are you saying that the bug is actually in lower-subreg.cc ?=