[PATCH] powerpc/cell: fix dependency in cpufreq

[Michael Neuling]

> > > cbe_cpufreq.c uses cbe_cpufreq_has_pmi which is provided by
> > > cbe_cpufreq_pmi.c.  Hence CBE_CPUFREQ depends on CBE_CPUFREQ_PMI.
> > >=20
> > > Signed-off-by: Michael Neuling <mikey at neuling.org>
> > > ---
> > > I'm not 100% sure is this the right fix.  Should CBE_CPUFREQ really
> > > depend on CBE_CPUFREQ_PMI?
> > 
> > No I don't think it is. Look at platforms/cell/cbe_cpufreq.h, the
> > intention seems to be that if PMI isn't built in then that variable is
> > hardcoded to 0 and we never use the pmi routines.
> 
> ---
> #if defined(CONFIG_CBE_CPUFREQ_PMI) || defined(CONFIG_CBE_CPUFREQ_PMI_MODULE)
> extern bool cbe_cpufreq_has_pmi;
> #else
> #define cbe_cpufreq_has_pmi (0)
> #endif
> --
> 
> So, I think this is bust.  If cpufreq_pmi is in a module, how does
> cpufreq use cbe_cpufreq_has_pmi if the module is not inserted?  
> 
> Arnd?

FYI the config to trigger this is below...

Mikey

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: cbe_cpufreq_has_pmi-missing.config
URL: <http://lists.ozlabs.org/pipermail/linuxppc-dev/attachments/20090210/064eac2e/attachment.asc>