https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81165
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Keywords| |missed-optimization
--- Comment #5 from Richard Biener <rguenth at gcc dot gnu.org> ---
Note that whatever clever optimization we did in GCC 7 it only happened during
RTL optimization. The root cause of the regression is a change in the GIMPLE
IL.
Andrew figured we miss some simplification patterns. I think that for
x1.1_4 ={v} x1;
_5 = (unsigned short) x1.1_4;
_6 = _5 ^ 8;
_7 = (short int) _6;
_29 = _7 & 1;
_8 = (int) _29;
_9 = _8 + -1;
_15 = (unsigned int) _9;
_30 = 1 >> _15;
we miss to optimize (int)((short int)((unsigned short) x1.1_4 ^ 8) & 1) to
(int)(x1.1_4 & 1) as well.
Not enough to eliminate the loop. Somebody has to investigate what RTL opt did
which transform to do that...
