Arnd Bergmann writes:
> --- linux-2.6.15-rc.orig/arch/ppc/kernel/ppc_ksyms.c
> +++ linux-2.6.15-rc/arch/ppc/kernel/ppc_ksyms.c
> @@ -311,7 +311,6 @@ EXPORT_SYMBOL(__res);
>
> EXPORT_SYMBOL(next_mmu_context);
> EXPORT_SYMBOL(set_context);
> -EXPORT_SYMBOL_GPL(__handle_mm_fault); /* For MOL */
Why? What have you got against MOL? :)
Paul.
Paul Mackerras <[email protected]> wrote:
>
> Arnd Bergmann writes:
>
> > --- linux-2.6.15-rc.orig/arch/ppc/kernel/ppc_ksyms.c
> > +++ linux-2.6.15-rc/arch/ppc/kernel/ppc_ksyms.c
> > @@ -311,7 +311,6 @@ EXPORT_SYMBOL(__res);
> >
> > EXPORT_SYMBOL(next_mmu_context);
> > EXPORT_SYMBOL(set_context);
> > -EXPORT_SYMBOL_GPL(__handle_mm_fault); /* For MOL */
>
> Why? What have you got against MOL? :)
>
The export was moved to mm/memory.c. No explanation why though...
On Middeweken 16 November 2005 06:26, Andrew Morton wrote:
> >
> > Why? ?What have you got against MOL? :)
> >
>
> The export was moved to mm/memory.c. ? No explanation why though...
>
Sorry about the lack of explanation. There was a short discussion
about this in August, see http://lkml.org/lkml/2005/8/8/205 :
On Mon, 8 Aug 2005 11:42:03 -0700 (PDT), Linus Torvalds wrote:
> I don't see any reason not to make it global if there are two
> architectures that need it. Especially as long as it's marked GPL-only so
> that people don't start misusing it.
The __handle_mm_fault symbol is used by spu_base.ko because
the DMA page fault handler calls handle_mm_fault.
Of course at the point where ppc_ksyms.c gets merged into arch/powerpc,
there would again only be one architecture needing it...
Arnd <><