Re: [PATCH v2 11/22] arm64: Populate cpuinfo after notify_cpu_starting

[Suzuki K. Poulose]

On 08/10/15 11:46, Suzuki K. Poulose wrote:
> On 08/10/15 11:15, Catalin Marinas wrote:
> > On Mon, Oct 05, 2015 at 06:02:00PM +0100, Suzuki K. Poulose wrote:
> > > diff --git a/arch/arm64/kernel/smp.c b/arch/arm64/kernel/smp.c
> > > index cb3e0d8..6987de4 100644
> > > --- a/arch/arm64/kernel/smp.c
> > > +++ b/arch/arm64/kernel/smp.c
> > > @@ -163,14 +163,14 @@ asmlinkage void secondary_start_kernel(void)
> > >           cpu_ops[cpu]->cpu_postboot();
> > >
> > >       /*
> > > -     * Log the CPU info before it is marked online and might get read.
> > > +     * Enable GIC and timers.
> > >        */
> > > -    cpuinfo_store_cpu();
> > > +    notify_cpu_starting(cpu);
> > >
> > >       /*
> > > -     * Enable GIC and timers.
> > > +     * Log the CPU info before it is marked online and might get read.
> > >        */
> > > -    notify_cpu_starting(cpu);
> > > +    cpuinfo_store_cpu();
> > >
> > >       smp_store_cpu_info(cpu);
> >
> > You can move the cpuinfo_store_cpu() call directly to
> > smp_store_cpu_info().
>
> That looks better, Thanks.

On a second look, smp_store_cpu_info() is also called from
smp_prepare_cpus() by the Boot CPU to update its topology
information, just after init_cpu_topology(). So moving the
cpuinfo_store_cpu() could be called for the boot CPU (which
we don't want, as it takes a different route to storing the
info).

One thing we could do is:

- Move init_cpu_topology() to setup_processor()
- Have the boot CPU update the topology from setup_processor()
  there, without going via the smp_store_cpu_info()
- Remove the smp_store_cpu_info() from smp_prepare_cpus().

Should we do that ? Or retain the code as above ?

Suzuki

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/