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 B498B3858402; Wed, 31 Jan 2024 02:33:15 +0000 (GMT)
DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org B498B3858402
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gcc.gnu.org;
	s=default; t=1706668395;
	bh=NxTH0D9qpX4hcrRSx5jxHCnXvRfqvbykovvAZdjfuAM=;
	h=From:To:Subject:Date:From;
	b=LEmHSJdRfNRJMAQdjGT/InPAY0GJ+NLoI30sqhojexLAE5VeNiVXKVxXthnkQsryv
	 T1kbCbhaWgTZGq9MSe8z1pYrm0x8UhrhML2Q4UfaO3CiwDXoin6eJiBKrXSVzz/TUR
	 uV9ACNNDEH6h5kqywqz3YXNpMWSF+FrjiiqnYdHw=
From: "pinskia at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug tree-optimization/113678] New: SLP misses up vec_concat
Date: Wed, 31 Jan 2024 02:33:15 +0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: new
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: tree-optimization
X-Bugzilla-Version: 14.0
X-Bugzilla-Keywords: missed-optimization
X-Bugzilla-Severity: normal
X-Bugzilla-Who: pinskia at gcc dot gnu.org
X-Bugzilla-Status: UNCONFIRMED
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: bug_id short_desc product version bug_status
 keywords bug_severity priority component assigned_to reporter
 target_milestone cf_gcctarget
Message-ID: <bug-113678-4@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=3D113678

            Bug ID: 113678
           Summary: SLP misses up vec_concat
           Product: gcc
           Version: 14.0
            Status: UNCONFIRMED
          Keywords: missed-optimization
          Severity: normal
          Priority: P3
         Component: tree-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: pinskia at gcc dot gnu.org
  Target Milestone: ---
            Target: x86_64

Take:
```
void f(char *a, char *b)
{
        int b0 =3D b[0];
        int b1 =3D b[1];
        int b2 =3D b[2];
        int b3 =3D b[3];
        int b4 =3D 0;
        int b5 =3D 0;
        int b6 =3D 0;
        int b7 =3D 0;
        a[0] =3D b0;
        a[1] =3D b1;
        a[2] =3D b2;
        a[3] =3D b3;
#if 0
        asm("":::"memory");
#endif
        a[4] =3D b0;
        a[5] =3D b1;
        a[6] =3D b2;
        a[7] =3D b3;
}
```

On x86_64 we get some mess because SLP decides to do this:
```
  _1 =3D *b_6(D);
  _2 =3D MEM[(char *)b_6(D) + 1B];
  _3 =3D MEM[(char *)b_6(D) + 2B];
  _4 =3D MEM[(char *)b_6(D) + 3B];
  _16 =3D {_1, _2, _3, _4, _1, _2, _3, _4};
```

But this is could be done as 2 stores (if we change the `#if 0` to `#if 1` =
we
get the better code):
```
  vect__1.5_18 =3D MEM <vector(4) char> [(char *)b_6(D)];
  MEM <vector(4) char> [(char *)a_7(D)] =3D vect__1.5_18;
  MEM <vector(4) char> [(char *)a_7(D) + 4B] =3D vect__1.5_18;
```

Or we could get one store even like LLVM gets:
```
        movd    xmm0, dword ptr [rsi]           # xmm0 =3D mem[0],zero,zero=
,zero
        pshufd  xmm0, xmm0, 0                   # xmm0 =3D xmm0[0,0,0,0]
        movq    qword ptr [rdi], xmm0
        ret
```=