Date:   Mon, 02 Dec 2019 22:15:35 +0100
From:   Toke Høiland-Jørgensen <toke@...hat.com>
To:     Andrii Nakryiko <andrii.nakryiko@...il.com>,
        Jiri Olsa <jolsa@...nel.org>
Cc:     Arnaldo Carvalho de Melo <acme@...nel.org>,
        lkml <linux-kernel@...r.kernel.org>,
        Networking <netdev@...r.kernel.org>, bpf <bpf@...r.kernel.org>,
        Ingo Molnar <mingo@...nel.org>,
        Namhyung Kim <namhyung@...nel.org>,
        Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
        Peter Zijlstra <a.p.zijlstra@...llo.nl>,
        Michael Petlan <mpetlan@...hat.com>,
        Jesper Dangaard Brouer <brouer@...hat.com>,
        Daniel Borkmann <daniel@...earbox.net>,
        Alexei Starovoitov <alexei.starovoitov@...il.com>,
        Martin KaFai Lau <kafai@...com>,
        Song Liu <songliubraving@...com>, Yonghong Song <yhs@...com>,
        Andrii Nakryiko <andriin@...com>,
        Quentin Monnet <quentin.monnet@...ronome.com>
Subject: Re: [PATCHv4 0/6] perf/bpftool: Allow to link libbpf dynamically

Andrii Nakryiko <andrii.nakryiko@...il.com> writes:

> On Mon, Dec 2, 2019 at 5:19 AM Jiri Olsa <jolsa@...nel.org> wrote:
>>
>> hi,
>> adding support to link bpftool with libbpf dynamically,
>> and config change for perf.
>>
>> It's now possible to use:
>>   $ make -C tools/bpf/bpftool/ LIBBPF_DYNAMIC=1
>>
>> which will detect libbpf devel package and if found, link it with bpftool.
>>
>> It's possible to use arbitrary installed libbpf:
>>   $ make -C tools/bpf/bpftool/ LIBBPF_DYNAMIC=1 LIBBPF_DIR=/tmp/libbpf/
>>
>> I based this change on top of Arnaldo's perf/core, because
>> it contains libbpf feature detection code as dependency.
>>
>> Also available in:
>>   git://git.kernel.org/pub/scm/linux/kernel/git/jolsa/perf.git
>>   libbpf/dyn
>>
>> v4 changes:
>>   - based on Toke's v3 post, there's no need for additional API exports:
>>
>>     Since bpftool uses bits of libbpf that are not exported as public API in
>>     the .so version, we also pass in libbpf.a to the linker, which allows it to
>>     pick up the private functions from the static library without having to
>>     expose them as ABI.
>
> Whoever understands how this is supposed to work, can you please
> explain? From reading this, I think what we **want** is:
>
> - all LIBBPF_API-exposed APIs should be dynamically linked against libbpf.so;
> - everything else used from libbpf (e.g., netlink APIs), should come
> from libbpf.a.
>
> Am I getting the idea right?
>
> If yes, are we sure it actually works like that in practice? I've
> compiled with LIBBPF_DYNAMIC=1, and what I see is that libelf, libc,
> zlib, etc functions do have relocations against them in ".rela.plt"
> section. None of libbpf exposed APIs, though, have any of such
> relocations. Which to me suggests that they are just statically linked
> against libbpf.a and libbpf.so is just recorded in ELF as a dynamic
> library dependency because of this extra -lbpf flag. Which kind of
> defeats the purpose of this whole endeavor, no?
>
> I'm no linker expert, though, so I apologize if I got it completely
> wrong, would really appreciate someone to detail this a bit more.
> Thanks!

Ah, that is my mistake: I was getting dynamic libbpf symbols with this
approach, but that was because I had the version of libbpf.so in my
$LIBDIR that had the patch to expose the netlink APIs as versioned
symbols; so it was just pulling in everything from the shared library.

So what I was going for was exactly what you described above; but it
seems that doesn't actually work. Too bad, and sorry for wasting your
time on this :/

-Toke

Powered by blists - more mailing lists