Advice needed on when to synthesize <sym>.high_bound in ld

[Greg McGary]

Nick Clifton <nickc@redhat.com> writes:

> Symbols do have a size.  In fact it is the size of common symbols that
> is problematic, as it can change of a bigger definition of the symbol
> comes along.  (Would your patch handle this BTW ?)

I assumed that the common symbols had already been merged by the time
I checked for high bound, but I'm not certain that's justified.  I'll
investigate.

> Since we are talking ELF here, in theory all you need to do is to get
> hold of the elf_link_hash_entry structure for the symbol and examine
> its 'size' field.  You might look, for example at the function
> bfd_elf_link_hash_newfunc() in bfd/elf.c

OK.  Since I already have people trying to use this with COFF, I'd
like to get that working as well.  If COFF keeps sizes for its
init-data symbols, then it's easy.  If it doesn't, then I can
make gcc generate the high-bound symbols for the COFF case.

Greg