Re: glibc CVS (2.3) results

[Jeff Bailey]

On Mon, Sep 02, 2002 at 08:44:46PM +0200, Marcus Brinkmann wrote:

> > I'm going out soon, but I'll take a look later to see if I can
> > figure out why it's not resolving the name to number mappings
> > right.

> According to the objdump output you send me it's the same bug as I
> had.  The problem is that the __fxstat64 etc symbols don't get a PLT
> entry in the linker.  The only question that remains is why ;)

And since we're having the conversation on irc and just reporting the
interesting bits on email *g*, I looked at the Versions.v for
__fxstat64 and found it in two locations:

libc {
  GLIBC_2.1 {
    __xstat64; __fxstat64; __lxstat64; __poll;
    creat64;
    fstatfs64; fstatvfs; fstatvfs64; ftw64;
    lockf64; lseek64;
    nftw; nftw64;
    open64;
    sendfile;
    statfs64; statvfs; statvfs64;
  }
** stuff removed **
}

and later at:

ld {
  GLIBC_2.0 {
    __hurd_sigthread_stack_base; __hurd_sigthread_stack_end;
    __hurd_sigthread_variables;
    __hurd_threadvar_stack_mask; __hurd_threadvar_stack_offset;
    __close; __fxstat64; __getcwd; __getpid; __libc_read; __libc_write;
    __libc_lseek64; __mmap; __open; __xstat64;
    _exit; _hurd_intr_rpc_mach_msg;
    abort;
  }
  GLIBC_PRIVATE {
    _dl_init_first;
  }
}


-- 
At last you cry out in anguish: "Why me?"
God answers: "Why not?"
 - Sheldon Kopp