Committed: fix undefweak hidden references for CRIS

[Nick Clifton]

Hi Guys,

>> The bug caught a bfd_assert, which *REALLY* isn't helpful, as
>> this is just a call to _bfd_error_handler and then linking
>> continues with *no* exit error.  I've changed bfd_assert locally
>> to do like bfd_abort and also call _exit (EXIT_FAILURE), and I'd
>> like to poll interest in doing that or the equivalent for the
>> FSF code too.  If nothing else, a helpful code comment why it
>> isn't done.  I guess letting it continue is helpful for some
>> situations and bfd users, but not the common case of the linker.
>> Perhaps do it optionally, default on for the linker?
> 
> I can't find the message now, but I've suggested this too.
> I think having the linker abort would be a big improvement.

No objection here.  Is there a patch in the offering.

Cheers
   Nick