[Bug tree-optimization/108757] We do not simplify (a - (N*M)) / N + M -> a / N

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

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108757

--- Comment #21 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
(In reply to Jiu Fu Guo from comment #20)
> Interesting thing:
> the VR is always VR_VARYING, even for the below simple case:
> 
> typedef unsigned long INT;
> INT __attribute__ ((noinline)) foo (INT x)
> {
>   if (x < 4)
>     return 0;
>   INT a = x + 18446744073709551612ULL;
>   INT b = a >> 2;
>   return b + 1;
> }

Yes that is because x does not have a "global" range.

You could try the following testcase:
```
typedef unsigned long INT;
INT __attribute__ ((noinline)) foo (INT x)
{
  if (x < 4)
    __builtin_unreachable();
  INT a = x + 18446744073709551612ULL;
  INT b = a >> 2;
  return b + 1;
}
```

Which gets a (global) range for x_1(D) during forwprop3:
  # RANGE [irange] INT [4, +INF]
  INTD.2750 x_1(D) = xD.2751;