This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
Re: g77 - mips-sgi-irix6.2 - Internal compiler error in function copy _to_mode_reg
- To: craig at jcb-sc dot com
- Subject: Re: g77 - mips-sgi-irix6.2 - Internal compiler error in function copy _to_mode_reg
- From: Jeffrey A Law <law at hurl dot cygnus dot com>
- Date: Sun, 14 Feb 1999 23:12:55 -0700
- cc: David dot Billinghurst at riotinto dot com dot au, egcs-bugs at cygnus dot com
- Reply-To: law at cygnus dot com
In message <19990215060712.29282.qmail@deer>you write:
> >I believe this is a bug in the fortran front end.
>
> I'm inclined to agree. I'll talk to ezmlm about getting the original
> test case, to put it on my list of things to look at. Even if the
> problem doesn't evidently reproduce on my Pentium II, the front end
> *does* probably make the same mistake on it, and the back end might
> just be "coping". I can sometimes find the bug anyway.
>
> However, if someone wants to help me, send me the testcase, and maybe
> try debugging `f771', feeding it the source file (see `g77 -v ...' to
> see what command-line args it gets passed, if you are concerned
> about reproducability), and get to where the bug happens.
>
> Then do `p lineno' and `p input_filename', maybe a backtrace, and
> send me the output. (egcs/gcc/README.g77 describes this a bit
> more at the bottom.)
I suspect the problem is related to sizes of basic data types. On a pure
32bit box you probably won't get the differing modes. Thus the bug does
not exhibit itself.
I suspect if you build a cross compiler for mips-sgi-irix6 that you'll
be able to trigger the problem using the "-mabi=64" switch.
jeff