[Bug middle-end/98709] New: gcc optimizes bitwise operations, but doesn't optimize logical ones

[Bug middle-end/98709] New: gcc optimizes bitwise operations, but doesn't optimize logical ones

[vanyacpp at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98709

            Bug ID: 98709
           Summary: gcc optimizes bitwise operations, but doesn't optimize
                    logical ones
           Product: gcc
           Version: 10.2.1
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: middle-end
          Assignee: unassigned at gcc dot gnu.org
          Reporter: vanyacpp at gmail dot com
  Target Milestone: ---

GCC 10.2 produces very good code for this function noticing that both sides of
conjuntion are the same:

unsigned foo_bitwise(unsigned a, unsigned b)
{
    return (~a ^ b) & ~(a ^ b);
}

foo_bitwise(unsigned int, unsigned int):
        xor     edi, esi
        mov     eax, edi
        not     eax
        ret

But when I write a similar function with logical operations it doesn't notice
that:

bool foo_logical(bool a, bool b)
{
    return (!a ^ b) & !(a ^ b);
}

foo_logical(bool, bool):
        mov     eax, esi
        xor     eax, edi
        xor     eax, 1
        cmp     dil, sil
        sete    dl
        and     eax, edx
        ret

I believe that in a similar manner it can be optimized to something like this:

foo_logical(bool, bool):
        xor     edi, esi
        mov     eax, edi
        xor     eax, 1
        ret

[Bug middle-end/98709] gcc optimizes bitwise operations, but doesn't optimize logical ones

[glisse at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98709

--- Comment #1 from Marc Glisse <glisse at gcc dot gnu.org> ---
At the end of gimple, we have
  _6 = a_3(D) ^ b_4(D);
  _1 = ~_6;
  _2 = a_3(D) == b_4(D);
  _7 = _1 & _2;
I guess we are missing a simplification of ~(a^b) to a==b for bool (similar to
~(a!=b) be we canonicalize != to ^).

[Bug tree-optimization/98709] gcc optimizes bitwise operations, but doesn't optimize logical ones

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98709

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
     Ever confirmed|0                           |1
         Depends on|                            |101706
   Last reconfirmed|                            |2021-08-01
          Component|middle-end                  |tree-optimization
             Status|UNCONFIRMED                 |NEW
           Severity|normal                      |enhancement

--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Confirmed.

I noticed the ~(a^b) issue the other day (not even knowing about this bug) and
filed PR 101706 for that :)


Referenced Bugs:

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101706
[Bug 101706] bool0^bool1^1 -> bool0 == bool1

[Bug tree-optimization/98709] gcc optimizes bitwise operations, but doesn't optimize logical ones

[vanyacpp at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98709

Ivan Sorokin <vanyacpp at gmail dot com> changed:

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

--- Comment #3 from Ivan Sorokin <vanyacpp at gmail dot com> ---
This issue was fixed in PR106379 by Richard Biener.

https://gcc.gnu.org/g:375668e0508fbe173af1ed519d8ae2b79f388d94

[Bug tree-optimization/98709] gcc optimizes bitwise operations, but doesn't optimize logical ones

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98709
Bug 98709 depends on bug 101706, which changed state.

Bug 101706 Summary: bool0^bool1^1 -> bool0 == bool1
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101706

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