Re: [PATCH bpf-next v3 2/4] bpf: Verifier support for sleepable raw tracepoint programs

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

On 12/3/26 07:07, Kumar Kartikeya Dwivedi wrote:
> On Wed, 11 Mar 2026 at 19:53, <bot+bpf-ci@kernel.org> wrote:
>>
>>> diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
>>> --- a/kernel/bpf/verifier.c
>>> +++ b/kernel/bpf/verifier.c
>>
>> [ ... ]
>>
>>> @@ -25435,6 +25441,7 @@ static bool can_be_sleepable(struct bpf_prog *prog)
>>>               case BPF_MODIFY_RETURN:
>>>               case BPF_TRACE_ITER:
>>>               case BPF_TRACE_FSESSION:
>>> +             case BPF_TRACE_RAW_TP:
>>>                       return true;
>>>               default:
>>>                       return false;
>>> @@ -25464,7 +25471,7 @@ struct bpf_attach_target_info tgt_info = {};
>>>       }
>>>
>>>       if (prog->sleepable && !can_be_sleepable(prog)) {
>>> -             verbose(env, "Only fentry/fexit/fmod_ret, lsm, iter, uprobe, and struct_ops programs can be sleepable\n");
>>> +             verbose(env, "Only fentry/fexit/fmod_ret, lsm, iter, uprobe, struct_ops, and raw_tp programs can be sleepable\n");
>>
>> The updated error message adds raw_tp but still omits fsession,
>> which is listed in can_be_sleepable() just above. Should this
>> read "fentry/fexit/fmod_ret/fsession, lsm, iter, uprobe,
>> struct_ops, and raw_tp"?
>>
> 
> We can ignore AI for now, Leon has a change for this.
> 
Just to confirm the change: is it
https://lore.kernel.org/bpf/20260303150639.85007-2-leon.hwang@linux.dev/?

If yes, it is to add "fsession" to the verbose log message in
check_get_func_ip(). Then, the AI review is correct: "fsession" was
missing here.

After searching for "fexit" in verifier.c, "FENTRY/FEXIT" can be updated
to "FENTRY/FEXIT/FSESSION" in bpf_check_attach_target(), and "fsession"
is missing in some comments. I'm not sure whether it is worth a patch to
do such cleanup work.

cc Menglong

Thanks,
Leon