Re: [patch] Add support for new mips relocs

[Richard Sandiford <rdsandiford@googlemail.com>]

Catherine Moore <clm@codesourcery.com> writes:
> *************** mips_reloc_class (int reloc_type)
> *** 890,903 ****
>   {
>     switch (reloc_type)
>       {
>       case R_MIPS_TLS_DTPMOD32:
>       case R_MIPS_TLS_DTPREL32:
>       case R_MIPS_TLS_TPREL32:
>         return RTYPE_CLASS_TLS;
>       default:
> !       /* MIPS lazy resolution stubs are local to the containing object,
> ! 	 so SHN_UNDEF symbols never participate in symbol lookup.  */
> !       return RTYPE_CLASS_PLT;
>       }
>   }
>   
> --- 914,929 ----
>   {
>     switch (reloc_type)
>       {
> +     case R_MIPS_COPY:
> +       return RTYPE_CLASS_COPY;
> +     case R_MIPS_JUMP_SLOT:
> +       return RTYPE_CLASS_PLT;
>       case R_MIPS_TLS_DTPMOD32:
>       case R_MIPS_TLS_DTPREL32:
>       case R_MIPS_TLS_TPREL32:
>         return RTYPE_CLASS_TLS;
>       default:
> !       return RTYPE_CLASS_VALID;
>       }
>   }
>   

I'm probably missing something, sorry, but how does the patch handle
the property mentioned in the original comment?  Relocations must never
resolve to another object's (traditional) lazy-binding stub.

Richard