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 9B5973858CD1; Mon, 11 Mar 2024 11:21:26 +0000 (GMT)
DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org 9B5973858CD1
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gcc.gnu.org;
	s=default; t=1710156086;
	bh=W8tx0/diwoT0NybsBmrvu1JSv8itudCxuH8zXDDU1RI=;
	h=From:To:Subject:Date:In-Reply-To:References:From;
	b=kq/NWUceRIGZHbZMgcHdFBUaNfKWudoCnFNoz1NRkt30iebI5a32tCV51N1iTJGnp
	 CRJ+3NfeVPgslVDQjEqFco3LxsADXFqaTx+AxgdxfeflJAat1byCVN1i8yljW0+NDr
	 t0EUfRTZmIlKFxMnBibvTCOLlUIfp6aiwbbhDaRU=
From: "jakub at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug ipa/113907] [11/12/13/14 regression] ICU miscompiled since on
 x86 since r14-5109-ga291237b628f41
Date: Mon, 11 Mar 2024 11:21:23 +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: 14.0
X-Bugzilla-Keywords: wrong-code
X-Bugzilla-Severity: normal
X-Bugzilla-Who: jakub at gcc dot gnu.org
X-Bugzilla-Status: ASSIGNED
X-Bugzilla-Resolution: 
X-Bugzilla-Priority: P1
X-Bugzilla-Assigned-To: hubicka at gcc dot gnu.org
X-Bugzilla-Target-Milestone: 14.0
X-Bugzilla-Flags: 
X-Bugzilla-Changed-Fields: 
Message-ID: <bug-113907-4-mt7j4Ulgmq@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-113907-4@http.gcc.gnu.org/bugzilla/>
References: <bug-113907-4@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=3D113907

--- Comment #54 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Slightly adjusted #c41 testcase, which indeed still fails with my patch at =
-O2:

int d[100], c;

static __attribute__((noinline))
int foo (int x, unsigned int y)
{
  if (y > 30)
    ++c;
  return x + y;
}

static int
bar (unsigned int x)
{
  if (x > 100)
    __builtin_unreachable ();
  if (__builtin_expect (d[x] !=3D 0, 1))
    return d[x];
  for (int j =3D 0; j < 100; j++)
    d[x] +=3D foo (d[j], x & 1 ? x + 17 : x + 16);
  return d[x];
}

static int
baz (unsigned int x)
{
  if (x > 10)
    __builtin_unreachable ();
  if (__builtin_expect (d[x] !=3D 0, 1))
    return d[x];
  for (int j =3D 0; j < 100; j++)
    d[x] +=3D foo (d[j], x & 1 ? x + 17 : x + 16);
  return d[x];
}

int
main ()
{
  int r =3D baz (1) + baz (2) + baz (3) + bar (4) + bar (30);
  if (c !=3D 100)
    __builtin_abort ();
  return r;
}

I still think we want both my patch and some fix for the jump functions, not
just the latter.
Anyway, can we in the spot my patch changed just walk all source->node->cal=
lees
cgraph_edges, for each of them find the corresponding cgraph_edge in the al=
ias
and for each walk all the jump_functions recorded and union their m_vr?
Or is that something that can't be done in LTO for some reason?=