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 1C9D838582AD; Thu,  4 Jan 2024 17:06:57 +0000 (GMT)
DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org 1C9D838582AD
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gcc.gnu.org;
	s=default; t=1704388018;
	bh=JkaCdUw7bIjzzoQ7LqEOj7W61JTqNsv/vlaB5zHNPmA=;
	h=From:To:Subject:Date:In-Reply-To:References:From;
	b=Zwo3zAmKTovasrmeU0Ew0w3CvZ1RZZkv6u3hCvIUmnEt7cHRtgWW9FVDlU/DiyvUz
	 mOkjqpxqI4XY027uaE4OJOLf7DaKt34wrjj9RLQH+9Yj0MEvHMSVAb/XMCqkQ+8LHN
	 0cCriSFhoPTgtBfZ9HxyU1gxrjULZsINgSo2f70o=
From: "hubicka at ucw dot cz" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug tree-optimization/110852] [14 Regression] ICE: in
 get_predictor_value, at predict.cc:2695 with -O -fno-tree-fre and
 __builtin_expect() since r14-2219-geab57b825bcc35
Date: Thu, 04 Jan 2024 17:06:57 +0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: tree-optimization
X-Bugzilla-Version: 14.0
X-Bugzilla-Keywords: ice-on-valid-code
X-Bugzilla-Severity: normal
X-Bugzilla-Who: hubicka at ucw dot cz
X-Bugzilla-Status: NEW
X-Bugzilla-Resolution: 
X-Bugzilla-Priority: P1
X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org
X-Bugzilla-Target-Milestone: 14.0
X-Bugzilla-Flags: 
X-Bugzilla-Changed-Fields: 
Message-ID: <bug-110852-4-oPPKWvylDW@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-110852-4@http.gcc.gnu.org/bugzilla/>
References: <bug-110852-4@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=3D110852

--- Comment #14 from Jan Hubicka <hubicka at ucw dot cz> ---
> I thought the goal was to handle what is in predict-18.c, i.e.
> b * __builtin_expect (c, 0)
> or similar.  If it is about
> __builtin_expect_with_probability (b, 42, 0.25) *
> __builtin_expect_with_probability (c, 0, 0.42)
> sure, my version will merge the probabilities, while you'll pick the
> probability from
> the 0 case.

Probability from 0 case is better estimate, so I think it makes sense to
handle it right.  I did not take that much stats on how often it
happens, but on my TODO list is to turn this into value range predictor
which may have better chance of success. We can also handle other
constants than INTEGER_CST.

I will see if I can clean up the code bit more or add a comment, since
it is indeed bit confusing as written now.  Will also look into more
testcases.

Thanks a lot!
Honza=