Re: [PATCH v2 2/3] powerpc: Call chained reset handlers during reset

[Andrey Smirnov]

On Sun, Jul 31, 2016 at 8:47 PM, Nicholas Piggin <npiggin@xxxxxxxxx> wrote:
> On Thu, 28 Jul 2016 16:07:17 -0700
> Andrey Smirnov <andrew.smirnov@xxxxxxxxx> wrote:
>
>> Call out to all restart handlers that were added via
>> register_restart_handler() API when restarting the machine.
>>
>> Signed-off-by: Andrey Smirnov <andrew.smirnov@xxxxxxxxx>
>> ---
>>
>> No changes compared to v1
>>
>>  arch/powerpc/kernel/setup-common.c | 4 ++++
>>  1 file changed, 4 insertions(+)
>>
>> diff --git a/arch/powerpc/kernel/setup-common.c
>> b/arch/powerpc/kernel/setup-common.c index 5cd3283..205d073 100644
>> --- a/arch/powerpc/kernel/setup-common.c
>> +++ b/arch/powerpc/kernel/setup-common.c
>> @@ -145,6 +145,10 @@ void machine_restart(char *cmd)
>>               ppc_md.restart(cmd);
>>
>>       smp_send_stop();
>> +
>> +     do_kernel_restart(cmd);
>> +     mdelay(1000);
>> +
>>       machine_hang();
>>  }
>>
>
> Ah, I see why you don't move smp_send_stop(). 3 other architectures
> call do_kernel_restart(). arm and arm64 call it with
> local_irq_disabled().

I am not very familiar with low-level SPM code, so take all below with
a grain of salt.

>From my understanding of the code ARM's implementation of
smp_send_stop() is different from MIPS/PowerPC ones in that it just
raises an IPI with a special "stop" flag set, which can and probably
should be done with IRQs disabled. Both MIPS and PowerPC call
smp_call_fuction() in their smp_send_stop() implementation, which if I
read the documentation correctly should not be called with interrupts
disabled, so it looks like the call to local_irq_disabled() could only
be placed after the call to smp_send_stop() on those platforms.

> arm and mips insert the 1s delay. All call it
> after smp_send_stop(). I don't see the harm in the delay. Should we
> call it with local interrupts disabled?
>

With all above being said I don't see any harm in disabling interrupts.

Thanks,
Andrey