[Bug rtl-optimization/124793] RISC-V RTL optimization for "(X EQ 0) ? 0 : FN (X) => FN (X)"

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=124793

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
     Ever confirmed|0                           |1
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2026-04-06

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Confirmed.
>We have match.pd code that covers this case for gimple
There is also extra phiopt code to handle popcount see
cond_removal_in_builtin_zero_pattern (because of a cast is not handled by
phiopt match yet).

The problem in the second popcount (the first one is handled) is we have:
```
  if (b_7(D) != 0)
    goto <bb 3>; [89.00%]
  else
    goto <bb 4>; [11.00%]

  <bb 3> [local count: 105119324]:
  _27 = (unsigned long) b_7(D);
  _2 = .POPCOUNT (_27);

  <bb 4> [local count: 118111600]:
  # prephitmp_21 = PHI <_10(2), _2(3)>
```

Oh I see the issue now.
This is b ? popcount(b) : c
Maybe the code in cond_removal_in_builtin_zero_pattern could be convert this
into:
t = popcount(b)
t != 0 ? t : c

When PROP_last_full_fold is set?