https://gcc.gnu.org/bugzilla/show_bug.cgi?id=82188
Andrew Pinski <pinskia at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Severity|normal |enhancement
--- Comment #3 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
int main (void)
{
volatile int x = 1;
long t1 = 1/(2U-x);
int t2 = 0!=(((int)(2%(2+t1)))/2);
if (t1 != 1) __builtin_abort();
if (t2 != 1) __builtin_abort();
return 0;
}
---- CUT ----
x.1_2 = (unsigned int) x.0_1;
_3 = 2 - x.1_2;
_4 = 1 / _3;
_12 = _3 <= 1;
_8 = (long int) _12;
_6 = _8 << 1;
if (_4 != 1)
...
if (_6 != 2)
We know the range of _8 is [0, 1], so we should change _6 != 2 to _8 != 1
Which then gets simplified to _3 > 1.
As for the _4 != 1:
1/ (2-x) != 1 is always false?
