[Bug tree-optimization/96671] New: Failure to optimize a 3 xor+and pattern to xor + andnot

[Bug tree-optimization/96671] New: Failure to optimize a 3 xor+and pattern to xor + andnot

[gabravier at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96671

            Bug ID: 96671
           Summary: Failure to optimize a 3 xor+and pattern to xor +
                    andnot
           Product: gcc
           Version: 11.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: gabravier at gmail dot com
  Target Milestone: ---

[Bug tree-optimization/96671] Failure to optimize a 3 xor+and pattern to xor+andnot

[gabravier at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96671

Gabriel Ravier <gabravier at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |missed-optimization
            Summary|Failure to optimize a 3     |Failure to optimize a 3
                   |xor+and pattern to xor +    |xor+and pattern to
                   |andnot                      |xor+andnot

--- Comment #1 from Gabriel Ravier <gabravier at gmail dot com> ---
Oops, I accidentally pressed enter and filed this with no details. Here are the
details :

int f(int a, int b, int c)
{
    return (a ^ b) & ((b ^ c) ^ a);
}

This can be optimized to `return (a ^ b) & ~c`. This transformation is done by
LLVM, but not by GCC.

[Bug tree-optimization/96671] Failure to optimize a 3 xor+and pattern to xor+andnot

[gabravier at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96671

--- Comment #2 from Gabriel Ravier <gabravier at gmail dot com> ---
There is also a very similar pattern with this function :

int f(int a, int b, int c)
{
    return (a ^ b) | ((b ^ c) ^ a);
}

Which can be optimized to `return (a ^ b) | c;`.

[Bug tree-optimization/96671] Failure to optimize a 3 xor+and pattern to xor+andnot

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96671

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|                            |2020-08-25
             Status|UNCONFIRMED                 |NEW
           Keywords|                            |easyhack
     Ever confirmed|0                           |1

--- Comment #3 from Richard Biener <rguenth at gcc dot gnu.org> ---
Confirmed.

[Bug tree-optimization/96671] Failure to optimize a 3 xor+and pattern to xor+andnot

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96671

--- Comment #4 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Jeff Law <law@gcc.gnu.org>:

https://gcc.gnu.org/g:f44e6091627372bd8fc4e72874a003643b021dca

commit r11-5146-gf44e6091627372bd8fc4e72874a003643b021dca
Author: Eugene Rozenfeld <Eugene.Rozenfeld@microsoft.com>
Date:   Wed Nov 18 14:29:49 2020 -0500

    Optimize two patterns with three xors

    gcc/
            PR tree-optimization/96671
            * match.pd (three xor patterns): New patterns.

[Bug tree-optimization/96671] Failure to optimize a 3 xor+and pattern to xor+andnot

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96671

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jakub at gcc dot gnu.org

--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Has any testcase been committed for this?
Also, for the 2 xors, I don't think we have any canonicalization, so can't that
(b ^ c) ^ a
part e.g. in #c1 come in also in
(b ^ a) ^ c
and
(a ^ c) ^ b
forms (I know it uses :c and that handles
(c ^ b) ^ a
a ^ (b ^ c)
a ^ (c ^ b)
for the first one.

[Bug tree-optimization/96671] Failure to optimize a 3 xor+and pattern to xor+andnot

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96671

--- Comment #6 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Created attachment 49973
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=49973&action=edit
gcc11-pr96671.patch

Seems we handle all those cases, so I think we just should add a testcase and
close.

[Bug tree-optimization/96671] Failure to optimize a 3 xor+and pattern to xor+andnot

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96671

--- Comment #7 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Jakub Jelinek <jakub@gcc.gnu.org>:

https://gcc.gnu.org/g:aaec739250b1b148f7db5aa86719fe76645efc06

commit r11-6731-gaaec739250b1b148f7db5aa86719fe76645efc06
Author: Jakub Jelinek <jakub@redhat.com>
Date:   Fri Jan 15 19:26:01 2021 +0100

    testsuite: Add testcase coverage for already fixed [PR96671]

    The fix for this PR didn't come with any test coverage, I've added
    tests that make sure we optimize it no matter what order of the x ^ y ^ z
    operands is used.

    2021-01-15  Jakub Jelinek  <jakub@redhat.com>

            PR tree-optimization/96671
            * gcc.dg/tree-ssa/pr96671-1.c: New test.
            * gcc.dg/tree-ssa/pr96671-2.c: New test.

[Bug tree-optimization/96671] Failure to optimize a 3 xor+and pattern to xor+andnot

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96671

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |FIXED
             Status|NEW                         |RESOLVED

--- Comment #8 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Fixed.