[03/12] powerpc: Replace cpu_up/down with device_online/offline

The core device API performs extra housekeeping bits that are missing
from directly calling cpu_up/down.

See commit a6717c01ddc2 ("powerpc/rtas: use device model APIs and
serialization during LPM") for an example description of what might go
wrong.

This also prepares to make cpu_up/down a private interface for anything
but the cpu subsystem.

Signed-off-by: Qais Yousef <qais.yousef@arm.com>
CC: Benjamin Herrenschmidt <benh@kernel.crashing.org>
CC: Paul Mackerras <paulus@samba.org>
CC: Michael Ellerman <mpe@ellerman.id.au>
CC: Enrico Weigelt <info@metux.net>
CC: Ram Pai <linuxram@us.ibm.com>
CC: Nicholas Piggin <npiggin@gmail.com>
CC: Thiago Jung Bauermann <bauerman@linux.ibm.com>
CC: Christophe Leroy <christophe.leroy@c-s.fr>
CC: Thomas Gleixner <tglx@linutronix.de>
CC: linuxppc-dev@lists.ozlabs.org
CC: linux-kernel@vger.kernel.org
---
 arch/powerpc/kernel/machine_kexec_64.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

Qais Yousef <qais.yousef@arm.com> writes:
> The core device API performs extra housekeeping bits that are missing
> from directly calling cpu_up/down.
>
> See commit a6717c01ddc2 ("powerpc/rtas: use device model APIs and
> serialization during LPM") for an example description of what might go
> wrong.
>
> This also prepares to make cpu_up/down a private interface for anything
> but the cpu subsystem.
>
> Signed-off-by: Qais Yousef <qais.yousef@arm.com>
> CC: Benjamin Herrenschmidt <benh@kernel.crashing.org>
> CC: Paul Mackerras <paulus@samba.org>
> CC: Michael Ellerman <mpe@ellerman.id.au>
> CC: Enrico Weigelt <info@metux.net>
> CC: Ram Pai <linuxram@us.ibm.com>
> CC: Nicholas Piggin <npiggin@gmail.com>
> CC: Thiago Jung Bauermann <bauerman@linux.ibm.com>
> CC: Christophe Leroy <christophe.leroy@c-s.fr>
> CC: Thomas Gleixner <tglx@linutronix.de>
> CC: linuxppc-dev@lists.ozlabs.org
> CC: linux-kernel@vger.kernel.org
> ---
>  arch/powerpc/kernel/machine_kexec_64.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)

My initial though is "what about kdump", but that function is not called
during kdump, so there should be no issue with the extra locking leading
to deadlocks in a crash.

Acked-by: Michael Ellerman <mpe@ellerman.id.au>

I assume you haven't actually tested it?

cheers

> diff --git a/arch/powerpc/kernel/machine_kexec_64.c b/arch/powerpc/kernel/machine_kexec_64.c
> index 04a7cba58eff..ebf8cc7acc4d 100644
> --- a/arch/powerpc/kernel/machine_kexec_64.c
> +++ b/arch/powerpc/kernel/machine_kexec_64.c
> @@ -208,13 +208,15 @@ static void wake_offline_cpus(void)
>  {
>  	int cpu = 0;
>  
> +	lock_device_hotplug();
>  	for_each_present_cpu(cpu) {
>  		if (!cpu_online(cpu)) {
>  			printk(KERN_INFO "kexec: Waking offline cpu %d.\n",
>  			       cpu);
> -			WARN_ON(cpu_up(cpu));
> +			WARN_ON(device_online(get_cpu_device(cpu)));
>  		}
>  	}
> +	unlock_device_hotplug();
>  }
>  
>  static void kexec_prepare_cpus(void)
> -- 
> 2.17.1

On 11/19/19 12:19, Michael Ellerman wrote:
> Qais Yousef <qais.yousef@arm.com> writes:
> > The core device API performs extra housekeeping bits that are missing
> > from directly calling cpu_up/down.
> >
> > See commit a6717c01ddc2 ("powerpc/rtas: use device model APIs and
> > serialization during LPM") for an example description of what might go
> > wrong.
> >
> > This also prepares to make cpu_up/down a private interface for anything
> > but the cpu subsystem.
> >
> > Signed-off-by: Qais Yousef <qais.yousef@arm.com>
> > CC: Benjamin Herrenschmidt <benh@kernel.crashing.org>
> > CC: Paul Mackerras <paulus@samba.org>
> > CC: Michael Ellerman <mpe@ellerman.id.au>
> > CC: Enrico Weigelt <info@metux.net>
> > CC: Ram Pai <linuxram@us.ibm.com>
> > CC: Nicholas Piggin <npiggin@gmail.com>
> > CC: Thiago Jung Bauermann <bauerman@linux.ibm.com>
> > CC: Christophe Leroy <christophe.leroy@c-s.fr>
> > CC: Thomas Gleixner <tglx@linutronix.de>
> > CC: linuxppc-dev@lists.ozlabs.org
> > CC: linux-kernel@vger.kernel.org
> > ---
> >  arch/powerpc/kernel/machine_kexec_64.c | 4 +++-
> >  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> My initial though is "what about kdump", but that function is not called
> during kdump, so there should be no issue with the extra locking leading
> to deadlocks in a crash.
> 
> Acked-by: Michael Ellerman <mpe@ellerman.id.au>

Thanks.

> 
> I assume you haven't actually tested it?

Only compile tested it I'm afraid. Would appreciate if you can give it a spin.
Otherwise I'd be happy to try it out on qemu.

Cheers

--
Qais Yousef