Simon Burge <simonb@netbsd.org> writes:

> [[ CC'd to port-mips too ]]
> 
> Matthias Drochner wrote:
> 
> > kleink@ira.uka.de said:
> > > Since pmaxen are 32-bit-only machines I'd say explicitly using ELF32
> > > symbols would be the appropriate solution.  (... and apologies for
> > > missing this one!) 
> > 
> > There seems to be one more: The <mips/elf.h> header duplicates a lot of
> > the new <sys/exec_elf.h>:
> > 
> > [[ current errors ]]
> >
> > (The SHN_MIPS_ACOMMON there looks bogus btw.)
> 
> I think I can get the kernel to stop using <mips/elf.h> fairly easily.
> Luckily the SHN_ sutff isn't used by the mips-specific ELF interpreter.
> Klaus - are the definitions of the SHN_MIPS_* macros "authorative"?

They are, but they're not tied to <mips/elf.h> but <elf.h>. (Like
Matthias pointed out, the SHN_MIPS_ACOMMON macro in <mips/elf.h> is
bogus.)

> Does anyone know if <mips/elf.h> is required by the MIPS API?

It's not; unless there are reasonable objections, I'd like to remove
it from the tree.


-klaus