[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <87r1vpuwzx.fsf@toke.dk>
Date: Tue, 12 May 2020 15:02:42 +0200
From: Toke Høiland-Jørgensen <toke@...hat.com>
To: Alan Maguire <alan.maguire@...cle.com>
Cc: Alexei Starovoitov <alexei.starovoitov@...il.com>,
Daniel Borkmann <daniel@...earbox.net>,
Andrii Nakryiko <andriin@...com>,
"David S. Miller" <davem@...emloft.net>,
Andrey Ignatov <rdna@...com>,
Networking <netdev@...r.kernel.org>, bpf <bpf@...r.kernel.org>
Subject: Re: bpf: ability to attach freplace to multiple parents
Alan Maguire <alan.maguire@...cle.com> writes:
> On Tue, 12 May 2020, Toke Høiland-Jørgensen wrote:
>
>> Alexei Starovoitov <alexei.starovoitov@...il.com> writes:
>>
>> >> > Currently fentry/fexit/freplace progs have single prog->aux->linked_prog pointer.
>> >> > It just needs to become a linked list.
>> >> > The api extension could be like this:
>> >> > bpf_raw_tp_open(prog_fd, attach_prog_fd, attach_btf_id);
>> >> > (currently it's just bpf_raw_tp_open(prog_fd))
>> >> > The same pair of (attach_prog_fd, attach_btf_id) is already passed into prog_load
>> >> > to hold the linked_prog and its corresponding btf_id.
>> >> > I'm proposing to extend raw_tp_open with this pair as well to
>> >> > attach existing fentry/fexit/freplace prog to another target.
>> >> > Internally the kernel verify that btf of current linked_prog
>> >> > exactly matches to btf of another requested linked_prog and
>> >> > if they match it will attach the same prog to two target programs (in case of freplace)
>> >> > or two kernel functions (in case of fentry/fexit).
>> >>
>> >> API-wise this was exactly what I had in mind as well.
>> >
>> > perfect!
>>
>
> Apologies in advance if I've missed a way to do this, but
> for fentry/fexit, if we could attach the same program to
> multiple kernel functions, it would be great it we could
> programmatically access the BTF func proto id for the
> attached function (prog->aux->attach_btf_id I think?).
Yes! I pushed for adding this to the GET_LINK_INFO operation, but it
wasn't included the first time around; I still think it ought to be added.
Actually in general, getting the btf_id of the currently running
function for any type of BPF program would be good; e.g., for xdpdump we
want to attach to a running XDP program, but we need the btf_id of the
main function. And because the name can be truncated to BPF_OBJ_NAME_LEN
when returned from the kernel, we have to walk the BTF info for the
program, and basically guess...
-Toke
Powered by blists - more mailing lists