[Bug tree-optimization/127915] New: When `a <= b` and `b <= CST` then `b - a` is known to have a range of [0, CST] (for unsigned)

[pinskia at gcc dot gnu.org via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=127915

            Bug ID: 127915
           Summary: When `a <= b` and `b <= CST` then `b - a` is known to
                    have a range of [0, CST] (for unsigned)
           Product: gcc
           Version: 17.0
            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
            Blocks: 85316, 127913
  Target Milestone: ---

Take:
```
tyepdef unsigned U;

void f(U first, U last)
{
    if (first > last || last > 100)
        __builtin_unreachable();

  U t = last - first;
  if (t > 100)
    __builtin_trap ();
}
```

The range of t should be [0, 100] also.  But we don't currently get that range:

  # RANGE [irange] U [0, 100][4294967196, +INF]
  t_6 = last_5(D) - first_4(D);

Also the `[0, 100][4294967196, +INF]` range is only picked up by DOM2 and not
by evrp.

Oh and it seems like we don't get `[0, 100]` range for first either.


Referenced Bugs:

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=85316
[Bug 85316] [meta-bug] VRP range propagation missed cases
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=127913
[Bug 127913] Unnecessary 128-bit arithmetic in ranges::lower_bound over a
bounded iota_view