Re: [PATCH 3/3] alarmtimer: switch spin_{lock,unlock}_irqsave() to guard()

[Thomas Gleixner <tglx@linutronix.de>]

On Thu, Apr 24 2025 at 16:59, John Stultz wrote:
> On Thu, Apr 24, 2025 at 7:48 AM Su Hui <suhui@nfschina.com> wrote:
>> -       spin_unlock_irqrestore(&freezer_delta_lock, flags);
>> +       scoped_guard(spinlock_irqsave, &freezer_delta_lock) {
>> +               min = freezer_delta;
>> +               expires = freezer_expires;
>> +               type = freezer_alarmtype;
>> +               freezer_delta = 0;
>> +       }
>
> I'm not necessarily opposed, but I'm not sure we're gaining much here.

>> @@ -352,13 +347,13 @@ EXPORT_SYMBOL_GPL(alarm_init);
>>  void alarm_start(struct alarm *alarm, ktime_t start)
>>  {
>>         struct alarm_base *base = &alarm_bases[alarm->type];
>> -       unsigned long flags;
>>
>> -       spin_lock_irqsave(&base->lock, flags);
>> -       alarm->node.expires = start;
>> -       alarmtimer_enqueue(base, alarm);
>> -       hrtimer_start(&alarm->timer, alarm->node.expires, HRTIMER_MODE_ABS);
>> -       spin_unlock_irqrestore(&base->lock, flags);
>> +       scoped_guard(spinlock_irqsave, &base->lock) {
>> +               alarm->node.expires = start;
>> +               alarmtimer_enqueue(base, alarm);
>> +               hrtimer_start(&alarm->timer, alarm->node.expires,
>> +                             HRTIMER_MODE_ABS);
>> +       }
>
> Similarly, this just seems more like churn, than making the code
> particularly more clear.

I disagree. scoped_guard() is actually superior as it makes it
visually entirely clear what the actual scope of the spin lock protected
code is. That's the whole point.

Especially in alarm_suspend() this would end up with a mix of scoped
guards and open coded spinlock regions. That's obstructing the reading
flow.

I'll bring them back for consistency when applying the series.

Thanks,

        tglx