[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <DDE6EAC2-EB86-4A21-9AF7-D74E1538E87D@fb.com>
Date: Mon, 30 Sep 2019 23:35:36 +0000
From: Song Liu <songliubraving@...com>
To: Andrii Nakryiko <andrii.nakryiko@...il.com>
CC: Song Liu <liu.song.a23@...il.com>,
Andrii Nakryiko <andriin@...com>, bpf <bpf@...r.kernel.org>,
Networking <netdev@...r.kernel.org>,
"Alexei Starovoitov" <ast@...com>,
Daniel Borkmann <daniel@...earbox.net>,
"Kernel Team" <Kernel-team@...com>
Subject: Re: [PATCH bpf-next 2/6] libbpf: move bpf_helpers.h, bpf_endian.h
into libbpf
> On Sep 30, 2019, at 4:27 PM, Andrii Nakryiko <andrii.nakryiko@...il.com> wrote:
>
> On Mon, Sep 30, 2019 at 4:23 PM Song Liu <songliubraving@...com> wrote:
>>
>>
>>
>>> On Sep 30, 2019, at 3:58 PM, Andrii Nakryiko <andrii.nakryiko@...il.com> wrote:
>>>
>>> On Mon, Sep 30, 2019 at 3:55 PM Song Liu <liu.song.a23@...il.com> wrote:
>>>>
>>>> On Mon, Sep 30, 2019 at 1:43 PM Andrii Nakryiko <andriin@...com> wrote:
>>>>>
>>>>> Make bpf_helpers.h and bpf_endian.h official part of libbpf. Ensure they
>>>>> are installed along the other libbpf headers.
>>>>>
>>>>> Signed-off-by: Andrii Nakryiko <andriin@...com>
>>>>
>>>> Can we merge/rearrange 2/6 and 3/6, so they is a git-rename instead of
>>>> many +++ and ---?
>>>
>>> I arranged them that way because of Github sync. We don't sync
>>> selftests/bpf changes to Github, and it causes more churn if commits
>>> have a mix of libbpf and selftests changes.
>>
>> Aha, I missed this point.
>>
>>> I didn't modify bpf_helpers.h/bpf_endian.h between those patches, so
>>> don't worry about reviewing contents ;)
>>
>> Well, we need to be careful here. As headers in a library should be
>> more stable than headers shipped with the code.
>>
>> Here, I am a little concerned with the fact that we added BPF_CORE_READ()
>> to libbpf, and then changed its syntax. This is within one release, so
>> it is mostly OK.
>
> Well, I could bundle bpf_helpers move and fixing up selftests in one
> commit, but I think it just makes commit unnecessarily big and
> convoluted. BPF_CORE_READ in previous form was ever only used by
> selftests, so it was never "released" per se, so it seems fine to do
> it this way, but let me know if you disagree.
A better approach is to modify BPF_CORE_READ in selftests before moving
it to libbpf. But I am ok with current approach as-is.
Song
Powered by blists - more mailing lists