Re: [PATCH 1/2] [powerpc] export symbols for use by lparcfg

[Will Schmidt <will_schmidt@vnet.ibm.com>]

On Mon, 2007-12-03 at 15:24 -0500, Nathan Lynch wrote:
> Will Schmidt wrote:
> > 
> > Updates the Kconfig to allow lparcfg to be built as a module, and
> > add the necessary EXPORT_SYMBOLS needed for a successful build.
> 
> Well, almost exactly a year ago lparcfg was changed to bool.  Do the
> reasons for that change still stand?
Hi Nathan, 

Thanks for pointing this out to me.  I was not aware there was a
specific and deliberate change to bool from tristate.   I thought it was
due to a sloppy change that neglected to export required symbols.

> commit 82dfdcae0d57c842e02f037758687eef42fb7af6
> Author: Paul Mackerras <paulus@samba.org>
> Date:   Tue Mar 14 11:35:37 2006 +1100
> 
>     powerpc: Disallow lparcfg being a module
>     
>     The lparcfg code needs several things which are pretty arcane internal
>     details and which we don't want to export, which means that lparcfg
>     doesn't work when built as a module.  This makes it a bool instead of
>     a tristate in the Kconfig so that users can't try to build it as a
>     module.
> 
> > +EXPORT_SYMBOL_GPL(lppaca);
> 
> > +EXPORT_PER_CPU_SYMBOL_GPL(cpu_usage_array);
> 
> > +EXPORT_SYMBOL_GPL(vdso_data);
> 
> Hmm, I don't think lparcfg needs to access vdso_data at all.

It does via the reference here: 

	partition_potential_processors = vdso_data->processorCount


> 
> In pseries_lparcfg_data we have:
> 
>         lrdrp = get_property(rtas_node, "ibm,lrdr-capacity", NULL);
> 
>         if (lrdrp == NULL) {
>                 partition_potential_processors = vdso_data->processorCount;
>         } else {
>                 partition_potential_processors = *(lrdrp + 4);
>         }
> 
>         partition_active_processors = lparcfg_count_active_processors();
> 
> But if there's no ibm,lrdr-capacity property then the system doesn't
> support adding processors, so partition_potential_processors should be
> equal to partition_active_processors.