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: 1008 segfaults in genattr


  In message <m0xYF7V-0004ecC@ocean.lucon.org>you write:
  > I suspect something is wrong with .fini on UnixWare. It will be
  > nice to set break point at _fini and see how many times it is
  > called on UnixWare. It may just be called twice. If it does, I
  > will be wondering why. One possibility is UnixWare doesn't support
  > one .fini on each object loaded into memory, including shared
  > libraries.
Note solaris != unixware.  I was speaking about possibly having to disable
.init/.fini on x86-solaris

x86-solaris2.5 has a different problem that has nothing to do with
multiple calls into the various exit routines.  The linker is null padding
the .init/.fini sections, which causes numerous problems.

At least that's how the problem was described to me.

jeff



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