Re: Re: [PATCH] bpf: Simplify code by using for_each_cpu_wrap()

[Punit Agrawal]

Hi Alexei,

Alexei Starovoitov <alexei.starovoitov@xxxxxxxxx> writes:

> On Wed, Aug 17, 2022 at 6:08 AM Punit Agrawal
> <punit.agrawal@xxxxxxxxxxxxx> wrote:
>>
>> No functional change intended.
>
> ?
>
>> -       orig_cpu = cpu = raw_smp_processor_id();
>> -       while (1) {
>> +       for_each_cpu_wrap(cpu, cpu_possible_mask, raw_smp_processor_id()) {
>>                 struct pcpu_freelist_head *head;
>>
>>                 head = per_cpu_ptr(s->freelist, cpu);
>> @@ -68,15 +67,10 @@ static inline void ___pcpu_freelist_push_nmi(struct pcpu_freelist *s,
>>                         raw_spin_unlock(&head->lock);
>>                         return;
>>                 }
>> -               cpu = cpumask_next(cpu, cpu_possible_mask);
>> -               if (cpu >= nr_cpu_ids)
>> -                       cpu = 0;
>> -
>> -               /* cannot lock any per cpu lock, try extralist */
>> -               if (cpu == orig_cpu &&
>> -                   pcpu_freelist_try_push_extra(s, node))
>> -                       return;
>>         }
>> +
>> +       /* cannot lock any per cpu lock, try extralist */
>> +       pcpu_freelist_try_push_extra(s, node);
>
> This is obviously not equivalent!

Thanks for taking a look. You're right - I missed the fact that it's an
infinite loop until the node gets pushed to one of the lists.

I'll send an update with that fixed up.