Re: [PATCH bpf-next v5 1/9] bpf: Generalize data copying for percpu maps

[Leon Hwang <leon.hwang@linux.dev>]

On 9/9/25 01:35, Alexei Starovoitov wrote:
> On Mon, Sep 8, 2025 at 7:37 AM Leon Hwang <leon.hwang@linux.dev> wrote:
>>

[...]

>> @@ -313,11 +312,7 @@ int bpf_percpu_array_copy(struct bpf_map *map, void *key, void *value)
>>         size = array->elem_size;
>>         rcu_read_lock();
>>         pptr = array->pptrs[index & array->index_mask];
>> -       for_each_possible_cpu(cpu) {
>> -               copy_map_value_long(map, value + off, per_cpu_ptr(pptr, cpu));
>> -               check_and_init_map_value(map, value + off);
>> -               off += size;
>> -       }
>> +       bpf_percpu_copy_data(map, pptr, value, size);
> 
> Same issue as before. This is not equivalent.
> Stop this "refactoring".
> 

Got it.

I’ll drop this refactoring and keep the original per-CPU copy logic as
is, and focus only on adding the BPF_F_CPU / BPF_F_ALL_CPUS support.

Thanks,
Leon