Re: one more h8300 ld -r problem left

[Joel Sherrill <joel dot sherrill at OARcorp dot com>]

Nick Clifton wrote:
> 
> Hi Joel,
> 
> : It appears the ld -r adds ctor/dtor symbols to files.
> :
> : Any ideas?

Thanks again

I am in the process of rebuilding with these changes:

2000-11-02     Nick Clifton <nickc@redhat.com>

       * reloc16.c: Add code to detect the bad relocs and issue a
       suitable error message, rather than core dumping.

2000-11-02     Alan Modra <alan@linuxcare.com.au>

       * coffcode.h (coff_slurp_reloc_table): Move asect->relocation
test
       from here..
       (coff_canonicalize_reloc): ..to here.  Adjust reloc addresses for
       relocatable link.

2000-11-02  Nick Clifton  <nickc@redhat.com>

        * scripttempl/h8300.sc (.tors): Only create symbols during a
        final link.  This prevents the section from being created
        during a partial link.
        * scripttempl/h8300h.sc (.tors): As above.
        * scripttempl/h8300s.sc (.tors): As above.

I will report on that as soon as I can.

A question though.  Don't these script templates have the same problem?

bash$ grep ctors_end * | grep -v RELOC 
h8500.sc:	___ctors_end = . ;
h8500b.sc:	___ctors_end = . ;
h8500c.sc:	___ctors_end = . ;
h8500m.sc:	___ctors_end = . ;
h8500s.sc:	___ctors_end = . ;
pj.sc:    ___ctors_end = . ;
sh.sc:    ___ctors_end = . ;
w65.sc:	___ctors_end = . ;

I know that the sh-coff RTEMS target links OK. Weird.

-- 
Joel Sherrill, Ph.D.             Director of Research & Development
joel@OARcorp.com                 On-Line Applications Research
Ask me about RTEMS: a free RTOS  Huntsville AL 35805
   Support Available             (256) 722-9985