http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60174
--- Comment #3 from Bernd Edlinger <bernd.edlinger at hotmail dot de> ---
(In reply to Richard Biener from comment #1)
> Can you check from the -fdump-tree-all dumps where i_34(ab) and i_399(ab)
> start to have overlapping life-ranges?
OK, i used grep 'i_\(34\|399\)(ab)' cc3305a.adb.*
i_399(ab) is first mentioned in cc3305a.adb.078t.dom1:
cc3305a.adb.078t.dom1:i_399(ab) -> { i_36(ab) }
cc3305a.adb.078t.dom1: # i_34(ab) = PHI <i_168(D)(ab)(2), i_168(D)(ab)(3),
i_168(D)(ab)(4), i_398(ab)(8), i_398(ab)(9), i_398(ab)(11), i_398(ab)(12),
i_398(ab)(13), i_35(ab)(14), i_35(ab)(15), i_35(ab)(16), i_35(ab)(17),
i_208(ab)(20), i_208(ab)(21), i_208(ab)(22), i_399(ab)(25), i_399(ab)(26),
i_399(ab)(28), i_399(ab)(30), i_399(ab)(31), i_36(ab)(32), i_36(ab)(34),
i_36(ab)(36), i_36(ab)(38), i_400(ab)(39), i_37(ab)(41), i_37(ab)(42),
i_37(ab)(43), i_38(ab)(46), i_38(ab)(47), i_38(ab)(48), i_401(ab)(51),
i_401(ab)(52), i_401(ab)(54), i_401(ab)(56), i_401(ab)(57), i_39(ab)(58),
i_39(ab)(60), i_43(ab)(90), i_402(ab)(61), i_40(ab)(63), i_40(ab)(64),
i_40(ab)(65), i_41(ab)(68), i_41(ab)(69), i_41(ab)(70), i_403(ab)(73),
i_403(ab)(74), i_403(ab)(76), i_403(ab)(78), i_403(ab)(79), i_42(ab)(80),
i_42(ab)(82), i_42(ab)(84), i_42(ab)(86), i_404(ab)(87), i_43(ab)(89)>
cc3305a.adb.078t.dom1: # i_399(ab) = PHI <i_36(ab)(24), i_333(93)>
cc3305a.adb.078t.dom1: _188 = i_399(ab) != 0;
cc3305a.adb.078t.dom1: _189 = i_399(ab) != 4;
cc3305a.adb.078t.dom1: # i_400(ab) = PHI <i_399(ab)(29), i_36(ab)(37),
i_36(ab)(38), i_399(ab)(30)>
cc3305a.adb.078t.dom1: # i_305 = PHI <i_34(ab)(5)>
cc3305a.adb.078t.dom1: # i_333 = PHI <i_34(ab)(23)>
cc3305a.adb.078t.dom1: # i_350 = PHI <i_34(ab)(49)>
cc3305a.adb.078t.dom1: # i_365 = PHI <i_34(ab)(71)>
is this already overlapping?
