https://gcc.gnu.org/bugzilla/show_bug.cgi?id=24568
Tobias Schlüter <tobi at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|NEW |RESOLVED
Resolution|--- |FIXED
--- Comment #11 from Tobias Schlüter <tobi at gcc dot gnu.org> ---
Take milliDiff = INT_MIN (<0).
milliDiff_6 = -milliDiff_5(D); // milliDiff_6 = INT_MIN (still <0)
minutesDiff_13 = milliDiff_6 / 60000;// minutesDiff_13 = INT_MIN/60000 =
-35791;
minutesDiff_8 = -minutesDiff_13; // minutesDiff_8 = 35791 (>0!)
So negation and division don't commute over the two's complement integers, and
I don't think there is any bug left. I'm taking the liberty to close this.
Link to a demonstration on the Compiler Explorer:
https://godbolt.org/z/hnjcMdjnq
Why the person who wrote that original bit of code wanted to deal with INT_MIN
(LONG_MIN in the original, but the testcase would overflow if sizeof(int) <
sizeof(long)) in this way is another question.
