This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: Help with DG/UX Intel port




  In message <06f501be9336$b4d6ddd0$65c8c8c8@ehrpc.listworks.com>you write:
  > the operation of the compiler.  When I prove that it all works, I'll be
  > happy to post the patches.
Thanks.

  > So, the compiler now fails the stage 2 to stage 3 comparisons (I used "make
  > bootstrap").  How do I figure out why?
There's no easy answer.  It could be a bug in egcs, it could be a bug in the
compiler you used to build egcs, it could be an assembler issue, etc.

The first thing to do is find out what is different between two of the .o
files.  Sometimes assemblers put timestamps or temporary filenames into the
.o file which causes comparison failures.

So, first find out what part of the .o is different.  ie, is it a difference
in the contents of a section like .text or .data?  Or is the difference
somewhere in the header of the .o?


  > I went passed that and ran the test suite anyway.  The first test that
  > failed is gcc.c-torture/execute/980526-1.c, which core dumps when -O2 is
  > used, but works fine without it.  How do I even start to fix this?
Don't bother :-)  980526-1.c is known to fail with certain versions of egcs.
jeff


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]