Re: [PATCH RFC bpf-next 2/2] selftests/bpf: Add LoongArch support for get_preempt_count()

On 4/20/26 5:47 PM, Tiezhu Yang wrote:
> On 2026/4/20 下午9:35, [email protected] wrote:
>> Sashiko AI review found 1 potential issue(s):
>> - [High] BPF programs including `bpf_experimental.h` will fail to compile on LoongArch due to a non-existent `thread_info` field in `struct task_struct`.
>> -- 
>>
>> commit 82453686fe5eacd9f219e2a050dc016b297cfc79
>> Author: Tiezhu Yang <[email protected]>
>>
>> selftests/bpf: Add LoongArch support for get_preempt_count()
>>
>> This commit adds LoongArch support for get_preempt_count() in the BPF
>> experimental helpers. It enables bpf_in_interrupt() and related macros to
>> work properly on LoongArch systems.
>>
>>> diff --git a/tools/testing/selftests/bpf/bpf_experimental.h b/tools/testing/selftests/bpf/bpf_experimental.h
>>> index e5c039bc44cb4..5847bbf12d240 100644
>>> --- a/tools/testing/selftests/bpf/bpf_experimental.h
>>> +++ b/tools/testing/selftests/bpf/bpf_experimental.h
>>> @@ -507,6 +507,8 @@ static inline int get_preempt_count(void)
>>>       return bpf_get_lowcore()->preempt_count;
>>>   #elif defined(bpf_target_riscv)
>>>       return bpf_get_current_task_btf()->thread_info.preempt_count;
>>> +#elif defined(bpf_target_loongarch)
>>> +    return bpf_get_current_task_btf()->thread_info.preempt_count;
>>
>> Will this compile successfully on LoongArch?
>>
>> Since LoongArch does not select CONFIG_THREAD_INFO_IN_TASK, struct
>> thread_info is allocated on the kernel stack rather than being embedded
>> within struct task_struct. Because of this, the generated BTF and vmlinux.h
>> for LoongArch will not contain a thread_info field in task_struct.
>>
>> Could this lead to a compiler error indicating that there is no member named
>> thread_info in struct task_struct when building BPF selftests?
> 
> As mentioned in the cover letter, it needs to apply the kernel patch
> "LoongArch: Implement CONFIG_THREAD_INFO_IN_TASK" [1] before testing.

Hi Tiezhu,

I think it makes sense to include the selftests changes in the
series implementing CONFIG_THREAD_INFO_IN_TASK for loongarch [1].

With respect to riscv part of this, I suggest you drop it unless you
have the means to run the tests. Landing without any testing is silly.

Also I think you may drop the RFC tag. AFAIU you're trying to bring
loongarch to parity with mainstream arch-es, and that shouldn't be
controversial.

Thanks.

[1] https://lore.kernel.org/bpf/[email protected]/

> 
> [1] https://lore.kernel.org/loongarch/[email protected]/
> 
> Thanks,
> Tiezhu
>