[Bug tree-optimization/112093] New: (X & Y) < X (unsigned) and (X & Y) != X should produce the same code

[Bug tree-optimization/112093] New: (X & Y) < X (unsigned) and (X & Y) != X should produce the same code

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=112093

            Bug ID: 112093
           Summary: (X & Y) < X (unsigned) and (X & Y) != X should produce
                    the same code
           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: ---

Take:
```
int f(unsigned a, unsigned b)
{
        return (a&b) < a;
}

int f1(unsigned a, unsigned b)
{
        return (a&b) != a;
}

int g(unsigned a, unsigned b)
{
        return (a&b) >= a;
}

int g1(unsigned a, unsigned b)
{
        return (a&b) == a;
}
```

f and f1 are equivalent but don't produce the same result.
Likewise for g and g1.

This is a Canonical issue really.

[Bug tree-optimization/112093] (X & Y) < X (unsigned) and (X & Y) != X should produce the same code

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=112093

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|normal                      |enhancement

[Bug tree-optimization/112093] (X & Y) < X (unsigned) and (X & Y) != X should produce the same code

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=112093

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           See Also|                            |https://gcc.gnu.org/bugzill
                   |                            |a/show_bug.cgi?id=101590

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
| has a similar issue:
```
int f_or(unsigned a, unsigned b)
{
        return (a|b) <= a;
}

int f1_or(unsigned a, unsigned b)
{
        return (a|b) == a;
}

int g_or(unsigned a, unsigned b)
{
        return (a|b) > a;
}

int g1_or(unsigned a, unsigned b)
{
        return (a|b) != a;
}
```

Note for other comparisons, see PR 101590 which I am about to submit a patch
for since they simplify down to a constant.

[Bug tree-optimization/112093] (X & Y) < X (unsigned) and (X & Y) != X should produce the same code

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=112093

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |ASSIGNED
   Last reconfirmed|                            |2023-11-04
     Ever confirmed|0                           |1
           Assignee|unassigned at gcc dot gnu.org      |pinskia at gcc dot gnu.org

--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
(for cmp  (lt      ge      le      gt     )
     bop  (bit_and bit_and bit_ior bit_ior)
     rcmp (ne      eq      eq      ne     )
 (simplify
  (cmp (bop@2 @0 @1) @0)
  (if (tree_expr_nonnegative_p (@0)
       && (bop == BIT_AND
           || tree_expr_nonnegative_p (@1)))
    (rcmp @2 @0))))

I have not looked into what is needed for the IOR case for
!tree_expr_nonnegative_p (@1) yet but I suspect it is similar to what I did for
PR 101590.

Anyways mine.