This is the mail archive of the
binutils@sourceware.org
mailing list for the binutils project.
Re: Your gas changes caused weakref failures for cris-elf,supposedly others too
>>> Alexandre Oliva <aoliva@redhat.com> 27.10.05 16:08:09 >>>
>On Oct 27, 2005, "Jan Beulich" <JBeulich@novell.com> wrote:
>
>> Hmm, interesting. Which way to fix this actually depends on the
>> intentions with weak references - if they're supposed to be
>> re-definable, then assign_symbol() or s_weakref() would need to
take
>> care of this. If not, the test is broken. Alexandre, can you clarify
the
>> intentions?
>
>Either way is fine. I wouldn't mind if they were not redefinable,
but
>I wanted to play along with .set et al, that allowed redefinitions.
>As long as it's consistent (i.e., .weakref can be a redefinition
*and*
>can be redefined, or neither of these), I'm open to change. GCC, the
>prime user of this feature, will never issue redefinitions for
correct
>programs anyway.
Actually, looking a little closer at these tests I'm not certain I'm
ready to change them. For example, in the section where re-definitions
are tested, I don't really understand why nr7 doesn't appear in any of
the *.d files even though I would think it should. So maybe it'd be
better if you adjusted it...
Thanks, Jan