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 4FACB3858D28; Wed,  9 Nov 2022 18:35:58 +0000 (GMT)
DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org 4FACB3858D28
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gcc.gnu.org;
	s=default; t=1668018958;
	bh=PwelN+qlxecBErwo/mYIFlXcLISCfUFIIYHI25bZM+I=;
	h=From:To:Subject:Date:In-Reply-To:References:From;
	b=AwaehmBgjuhvkuiOecpccYYBVu4++cj2bVtihfLdiPXmMI348iuKnDSdrxo+cejB0
	 dEJWsQCLjdaF2getpaBOian4tmUezJ5hY1Mtg8rRnNF7ZYCZHTdie6xosvxEKjHdi2
	 awBdd1BYKo1S7qSfRgaz09lgQOtSKA3C4dHLMf8g=
From: "jakub at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug tree-optimization/107591] range-op{,-float}.cc for x * x
Date: Wed, 09 Nov 2022 18:35:58 +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: 13.0
X-Bugzilla-Keywords: missed-optimization
X-Bugzilla-Severity: normal
X-Bugzilla-Who: jakub 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: ---
X-Bugzilla-Flags: 
X-Bugzilla-Changed-Fields: 
Message-ID: <bug-107591-4-gwteMP2KB6@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-107591-4@http.gcc.gnu.org/bugzilla/>
References: <bug-107591-4@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=3D107591

--- Comment #12 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
(In reply to Andrew Macleod from comment #11)
> no, I meant in addition to the VREL_EQ.  so=20
>   if (rel =3D=3D VREL_EQ && op1_range !=3D op2_range)
>      then you know you have something like  if (x =3D=3D y) z=3Dx*y and m=
ay have
> to check for various signed zero cobinations in each range,=20
> whereas is op1_range =3D=3D op2_range in this case, it should be perfectl=
y safe..

I don't understand.  Let's modify the testcase to:
float
foo (float x, y)
{
  if (x < -13.f || x > 26.f)
    return -1.0f;
  if (y < -13.f || y > 26.f)
    return -1.0f;
  if (x =3D=3D y)
    return x * y;
  return 42.0f;
}
I'd expect that fold_range on the x * y should see trio.op1_op2 () =3D=3D V=
REL_EQ
because of the guarding x =3D=3D y condition.  And the ranges of both are [=
-13.f,
26.f] +-NAN too.  Still, x could be -0.0f and y 0.0f or vice versa, and so
x * y could be -0.0f, so we need [-0.f, 676.f] +-NAN.  While if it is x * x=
, we
know the result will have always sign bit of 0 (except if NAN), so [0.f, 67=
6.f]
+-NAN.=