https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90273
--- Comment #12 from Richard Biener <rguenth at gcc dot gnu.org> --- Is # DEBUG INLINE_ENTRY NULL useful at all? And since I see # DEBUG D#1157 => &aOther_2(D)->mOffsetD.377222 # DEBUG thisD.1370683 => D#1158 # DEBUG aOtherD.1370684 => D#1157 # DEBUG INLINE_ENTRY __ct D.183324 MEM[(struct MaybeD.183104 *)this_5(D) + 16B].mIsSomeD.183308 = 0; _7 = MEM[(const struct Maybe &)aOther_2(D) + 16].mIsSomeD.183308; if (_7 != 0) goto <bb 3>; [50.00%] else goto <bb 4>; [50.00%] <bb 3> [local count: 536870913]: # DEBUG thisD.1370685 => D#1157 # DEBUG INLINE_ENTRY operator*D.183425 # DEBUG thisD.1370677 => D#1157 # DEBUG INLINE_ENTRY NULL # DEBUG thisD.1370678 => D#1157 # DEBUG INLINE_ENTRY NULL my guess that debug temps are always in the same BB as uses is wrong.