Re: [PATCH bpf-next] libbpf: Add support for dynamic program attach target

["Eelco Chaudron" <echaudro@redhat.com>]

On 12 Feb 2020, at 22:52, Toke Høiland-Jørgensen wrote:

> Andrii Nakryiko <andrii.nakryiko@gmail.com> writes:
>
>> On Wed, Feb 12, 2020 at 5:05 AM Toke Høiland-Jørgensen 
>> <toke@redhat.com> wrote:
>>>
>>> Eelco Chaudron <echaudro@redhat.com> writes:
>>>
>>>> Currently when you want to attach a trace program to a bpf program
>>>> the section name needs to match the tracepoint/function semantics.
>>>>
>>>> However the addition of the bpf_program__set_attach_target() API
>>>> allows you to specify the tracepoint/function dynamically.
>>>>
>>>> The call flow would look something like this:
>>>>
>>>>   xdp_fd = bpf_prog_get_fd_by_id(id);
>>>>   trace_obj = bpf_object__open_file("func.o", NULL);
>>>>   prog = bpf_object__find_program_by_title(trace_obj,
>>>>                                            "fentry/myfunc");
>>>>   bpf_program__set_attach_target(prog, xdp_fd,
>>>>                                  "fentry/xdpfilt_blk_all");
>>>
>>> I think it would be better to have the attach type as a separate arg
>>> instead of encoding it in the function name. I.e., rather:
>>>
>>>    bpf_program__set_attach_target(prog, xdp_fd,
>>>                                   "xdpfilt_blk_all", 
>>> BPF_TRACE_FENTRY);
>>
>> I agree about not specifying section name prefix (e.g., fentry/). But
>> disagree that expected attach type (BPF_TRACE_FENTRY) should be part
>> of this API. We already have bpf_program__set_expected_attach_type()
>> API, no need to duplicate it here.
>
> Ah yes, forgot about that; just keeping that and making this function
> name only is fine with me :)

Toke/Andrii,

Thanks for the feedback, will send out a v2 soon.

//Eelco