https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118551
--- Comment #6 from Hongtao Liu <liuhongt at gcc dot gnu.org> --- (In reply to Jan Hubicka from comment #5) > as discussed in PR111551 the SPEC train run does not include hottest loop of > MorphologyApply, so MeanShiftImage may have same issue and auto-fdo may be > kind of correct here. I'm afraid issue for autofdo is different. For auto-fdo, I'm collecting perf.data for ref-run, and the function body is hot(total:17008941), it's just that the sampling profile doesn't get any count for the function entry, then ipa scaling all function body to count zero, compiler optimized it for size.
