https://gcc.gnu.org/bugzilla/show_bug.cgi?id=127103
--- Comment #4 from Drea Pinski <pinskia at gcc dot gnu.org> ---
Note `x + uMax(x, 1) -> uMax(x+x, 1)` is a special case of `minmax (a, b) + c
-> minmax (a + c, b + c)`.
That is:
```
t_4 = a_3(D) * 2;
a_5 = a_3(D) + 1;
_6 = MAX_EXPR <t_4, a_5>;
```
Can be optimized to `MAX_EXPR <t_4, 1>` as t_4 min is 0 while a_5 min is 1.
```
unsigned t(unsigned a)
{
if (a > __INT_MAX__)
return 0;
unsigned t = a*2;
a+=1;
return t < a ? a : t;
}
```
LLVM handles the above but not the `x + uMax(x, 1)` for some reason. I didn't
look into why. Anyways the second pattern is enough for remove the `>=` in the
original testcase too. Though I don't understand why my reduced testcase fails
for LLVM; even though the original code (of push_back) works; not my worry.
