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 B4F273858D28; Thu, 26 Jan 2023 11:27:50 +0000 (GMT)
DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org B4F273858D28
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gcc.gnu.org;
	s=default; t=1674732470;
	bh=+9efky2+pqu1IH9zL8tNw4NxeJKrh2dZtihst92KJ1o=;
	h=From:To:Subject:Date:In-Reply-To:References:From;
	b=W2Um3Q/tVmVPZBOgoDE6MK+8HRsJJtIC3VQaEMMwsaQbRp1+J1v978Q1/g0NWJ+iK
	 UzySFQustzJZCNSWLmiqga3rpCnaUE22mJKOv+MAl5Qb5R4ZyZs1qMP1sh2MIcbmdH
	 WxLriD1E0kUZdJrx06M8j9DI0EHZcGTH0nMZcMrE=
From: "jakub at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug tree-optimization/108540] [13 Regression] Frange miscompilation
 of ruby since r13-3261
Date: Thu, 26 Jan 2023 11:27:49 +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: wrong-code
X-Bugzilla-Severity: normal
X-Bugzilla-Who: jakub at gcc dot gnu.org
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: 13.0
X-Bugzilla-Flags: 
X-Bugzilla-Changed-Fields: 
Message-ID: <bug-108540-4-OVVTtRYZ9m@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-108540-4@http.gcc.gnu.org/bugzilla/>
References: <bug-108540-4@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=3D108540

--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
This goes wrong during thread1 pass I think.
Before that we have in the IL early exits for x !=3D 42, d NAN or d +-INF, =
and
then:
  <bb 6> [local count: 96926676]:
  if (d_16(D) =3D=3D 0.0)
    goto <bb 7>; [50.00%]
  else
    goto <bb 8>; [50.00%]

  <bb 7> [local count: 48463338]:
  _2 =3D __builtin_signbit (d_16(D));
  if (_2 !=3D 0)
    goto <bb 9>; [50.00%]
  else
    goto <bb 8>; [50.00%]

  <bb 8> [local count: 72695007]:

  <bb 9> [local count: 164282501]:
  # RANGE [frange] double [-0.0 (-0x0.0p+0),
1.79769313486231570814527423731704356798070567525844996599e+308
(0x0.fffffffffffff8p+1024)]
  # d_7 =3D PHI <d_17(5), -0.0(7), d_16(D)(8)>
  # RANGE [irange] char [0, 0][45, 45] NONZERO 0x2d
  # sign_10 =3D PHI <45(5), 45(7), 0(8)>
  _3 =3D ch_18(D) =3D=3D 97;
  _4 =3D ch_18(D) =3D=3D 65;
  _5 =3D _3 | _4;
  if (_5 !=3D 0)
    goto <bb 10>; [50.00%]
  else
    goto <bb 13>; [50.00%]

  <bb 10> [local count: 82141250]:
  _13 =3D VIEW_CONVERT_EXPR<long long int>(d_7);
  _24 =3D _13 < 0;
  # RANGE [irange] int [0, 1] NONZERO 0x1
  _25 =3D (int) _24;
  if (d_7 =3D=3D 0.0)
    goto <bb 12>; [50.00%]
  else
    goto <bb 11>; [50.00%]

  <bb 11> [local count: 41070625]:

  <bb 12> [local count: 82141250]:
  # PT =3D
  # cp_8 =3D PHI <"WRONG"(11), "ZERO"(10)>
(#c1 testcase at -O2 on x86_64-linux in sra dump).  The path through the IL=
 on
arguments the function is called with is
d_16(D) =3D=3D 0.0, __builtin_signbit (d_16(D)) is non-zero and so d_7 is -=
0.0,
sign_10 45, then _5 is non-zero, d_7 =3D=3D 0.0 is
again true and so cp_8 is "ZERO".
But in thread1 this changes:
  <bb 6> [local count: 96926676]:
  if (d_16(D) =3D=3D 0.0)
    goto <bb 7>; [50.00%]
  else
    goto <bb 10>; [50.00%]

  <bb 7> [local count: 48463338]:
  _2 =3D __builtin_signbit (d_16(D));
  if (_2 !=3D 0)
    goto <bb 8>; [50.00%]
  else
    goto <bb 10>; [50.00%]

  <bb 8> [local count: 24231669]:
  # RANGE [frange] double [-0.0 (-0x0.0p+0),
1.79769313486231570814527423731704356798070567525844996599e+308
(0x0.fffffffffffff8p+1024)]
  # d_11 =3D PHI <-0.0(7)>
  # RANGE [irange] char [0, 0][45, 45] NONZERO 0x2d
  # sign_26 =3D PHI <45(7)>
  _27 =3D ch_18(D) =3D=3D 97;
  _28 =3D ch_18(D) =3D=3D 65;
  _29 =3D _27 | _28;
  if (_29 !=3D 0)
    goto <bb 9>; [50.00%]
  else
    goto <bb 14>; [50.00%]

  <bb 9> [local count: 12115835]:
  _30 =3D VIEW_CONVERT_EXPR<long long int>(d_11);
  _31 =3D _30 < 0;
  # RANGE [irange] int [0, 1] NONZERO 0x1
  _32 =3D (int) _31;
  goto <bb 12>; [100.00%]
...
  <bb 12> [local count: 41070625]:
  # RANGE [irange] char [0, 0][45, 45] NONZERO 0x2d
  # sign_33 =3D PHI <sign_10(11), sign_26(9)>
  # RANGE [irange] int [0, 1] NONZERO 0x1
  # _35 =3D PHI <_25(11), _32(9)>

  <bb 13> [local count: 82141250]:
  # PT =3D
  # cp_8 =3D PHI <"WRONG"(12), "ZERO"(11)>
Now the path through for the arguments is again d_16(D) =3D=3D 0.0, signbit
non-zero, d_11 is correctly -0.0 but this is in a threaded block, _29 is
non-zero, but we come into bb 13 from bb 12 and so cp_8 is "WRONG" rather t=
han
"ZERO".

Aldy, can you please have a look?=