[Bug target/114224] New: popcount RTL cost seems wrong with cssc

[Bug target/114224] New: popcount RTL cost seems wrong with cssc

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114224

            Bug ID: 114224
           Summary: popcount RTL cost seems wrong with cssc
           Product: gcc
           Version: 14.0
            Status: UNCONFIRMED
          Keywords: missed-optimization
          Severity: normal
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: pinskia at gcc dot gnu.org
  Target Milestone: ---
            Target: aarch64

I noticed this with:
```
#include <stdint.h>
bool is_power_of_2_1(uint64_t value)
{
  return __builtin_popcountg(value) == 1;
}
```

Which I had expected to produce:
```
        cnt     x0, x0
        cmp     x0, 1
        cset    w0, eq
```

But instead we get:
```
        sub     x1, x0, #1
        eor     x0, x0, x1
        cmp     x0, x1
        cset    w0, hi
        ret
```

[Bug target/114224] popcount RTL cost seems wrong with cssc

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114224

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

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

--- Comment #1 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
This is a matter of rtl costs.
.POPCOUNT expansion in expand_POPCOUNT expands the sequence both ways and asks
the backend what is cheaper:
  /* If .POPCOUNT call has 2 arguments, match_single_bit_test marked it
     because the result is only used in an equality comparison against 1.
     Use rtx costs in that case to determine if .POPCOUNT (arg) == 1
     or (arg ^ (arg - 1)) > arg - 1 is cheaper.
     If .POPCOUNT second argument is 0, we additionally know that arg
     is non-zero, so use arg & (arg - 1) == 0 instead.  */

[Bug target/114224] popcount RTL cost seems wrong with cssc

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114224

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |ASSIGNED
           Assignee|unassigned at gcc dot gnu.org      |pinskia at gcc dot gnu.org
   Last reconfirmed|                            |2024-03-04
     Ever confirmed|0                           |1

--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Interesting:
```
int h1(unsigned a)
{
        return  __builtin_popcountg(a) == 1;
}
```
works.


Anyways I will be adding POPCOUNT's rtl cost here.

We don't even handle POPCOUNT for vector modes either ...