------- Additional Comments From ericw at evcohs dot com 2005-02-28 22:10 ------- Subject: Re: [4.0/4.1 Regression] libgcc2.h Improperly determines required built-in function size requirements.
schlie at comcast dot net wrote: >------- Additional Comments From schlie at comcast dot net 2005-02-28 22:02 >------- >(In reply to comment #21) > > >>Hi, >> >>since this bug has been fixed by a patch of Roger Sayles a couple of weeks >>ago, I suggest to mark it as "fixed". >> >> > >It's true that the original failure mode, which nessesitated removing 64-bit >long longs >has been fixed, but observe that none the less; libgcc2.h still improperly >determines >built-in function types, if long-long types are not declared as being >supported by the >target. (the catch 22 is that one can't show a regression against an exiting >target, >because exiting targets would not build without 64-bit type support, therefore >the >only means to demonstraite it is to intentially remove such type support from >an exsiting >target, which should otherwise build fine, but wont')? > > > We've already gone over this. If you want to modify the sources to not declare the long long type for the AVR, fine, but that is on your experimental sources. See the previous discussion about this on bug # <http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20143> Please stop rehashing the same stuff in bug reports. If the original failure is fixed, then close the bug report. If there are different failures, then open up new bug reports, one per failure. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18887
