Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!tut.cis.ohio-state.edu!STEER.CALSTATE.EDU!r-hcsc01 From: r-hcsc01@STEER.CALSTATE.EDU (Mike Stump) Newsgroups: gnu.gcc.bug Subject: (none) Message-ID: <8909271754.AA14522@steer.CalState.EDU> Date: 27 Sep 89 17:54:49 GMT Sender: daemon@tut.cis.ohio-state.edu Distribution: gnu Organization: GNUs Not Usenet Lines: 34 Subject: bug in gcc 1.35.9[6789], 1.36.0 I think the below still needs fixing... It show up on machines that like 64 bits and not 32 bits. When widening immediates that are VOIDmode to 64 bits for compares.... If there is a reason why it is wrong to reverse them, can someone clue me in? Please mail clues to me. Thanks.... Diffs for 1.35.99: *** optabs.c.~1~ Tue Sep 5 13:22:31 1989 --- optabs.c Sat Sep 23 14:56:58 1989 *************** *** 730,738 **** enum mode_class class; enum machine_mode wider_mode; - class = GET_MODE_CLASS (mode); - if (mode == VOIDmode) mode = GET_MODE (y); /* They could both be VOIDmode if both args are immediate constants, but we should fold that at an earlier stage. With no special code here, this will call abort, --- 730,738 ---- enum mode_class class; enum machine_mode wider_mode; if (mode == VOIDmode) mode = GET_MODE (y); + + class = GET_MODE_CLASS (mode); /* They could both be VOIDmode if both args are immediate constants, but we should fold that at an earlier stage. With no special code here, this will call abort,