From: Jakub Sitnicki <jakub@cloudflare.com>
To: Martin KaFai Lau <kafai@fb.com>
Cc: Lorenz Bauer <lmb@cloudflare.com>,
Networking <netdev@vger.kernel.org>, bpf <bpf@vger.kernel.org>,
dccp@vger.kernel.org, kernel-team <kernel-team@cloudflare.com>,
Alexei Starovoitov <ast@kernel.org>,
"Daniel Borkmann" <daniel@iogearbox.net>,
"David S. Miller" <davem@davemloft.net>,
Eric Dumazet <edumazet@google.com>,
Gerrit Renker <gerrit@erg.abdn.ac.uk>,
Jakub Kicinski <kuba@kernel.org>,
Marek Majkowski <marek@cloudflare.com>
Subject: Re: [PATCH bpf-next 02/17] bpf: Introduce SK_LOOKUP program type with a dedicated attach point
Date: Fri, 08 May 2020 10:54:06 +0200 [thread overview]
Message-ID: <87blmyvmc1.fsf@cloudflare.com> (raw)
In-Reply-To: <20200507205524.pv2pnujxdrbktdc4@kafai-mbp>
On Thu, May 07, 2020 at 10:55 PM CEST, Martin KaFai Lau wrote:
> On Wed, May 06, 2020 at 03:53:35PM +0200, Jakub Sitnicki wrote:
>> On Wed, May 06, 2020 at 03:16 PM CEST, Lorenz Bauer wrote:
>> > On Wed, 6 May 2020 at 13:55, Jakub Sitnicki <jakub@cloudflare.com> wrote:
>>
>> [...]
>>
>> >> @@ -4012,4 +4051,18 @@ struct bpf_pidns_info {
>> >> __u32 pid;
>> >> __u32 tgid;
>> >> };
>> >> +
>> >> +/* User accessible data for SK_LOOKUP programs. Add new fields at the end. */
>> >> +struct bpf_sk_lookup {
>> >> + __u32 family; /* AF_INET, AF_INET6 */
>> >> + __u32 protocol; /* IPPROTO_TCP, IPPROTO_UDP */
>> >> + /* IP addresses allows 1, 2, and 4 bytes access */
>> >> + __u32 src_ip4;
>> >> + __u32 src_ip6[4];
>> >> + __u32 src_port; /* network byte order */
>> >> + __u32 dst_ip4;
>> >> + __u32 dst_ip6[4];
>> >> + __u32 dst_port; /* host byte order */
>> >
>> > Jakub and I have discussed this off-list, but we couldn't come to an
>> > agreement and decided to invite
>> > your opinion.
>> >
>> > I think that dst_port should be in network byte order, since it's one
>> > less exception to the
>> > rule to think about when writing BPF programs.
>> >
>> > Jakub's argument is that this follows __sk_buff->local_port precedent,
>> > which is also in host
>> > byte order.
>>
>> Yes, would be great to hear if there is a preference here.
>>
>> Small correction, proposed sk_lookup program doesn't have access to
>> __sk_buff, so perhaps that case matters less.
>>
>> bpf_sk_lookup->dst_port, the packet destination port, is in host byte
>> order so that it can be compared against bpf_sock->src_port, socket
>> local port, without conversion.
>>
>> But I also see how it can be a surprise for a BPF user that one field has
>> a different byte order.
> I would also prefer port and addr were all in the same byte order.
> However, it is not the cases for the other prog_type ctx.
> People has stomped on it from time to time. May be something
> can be done at the libbpf to hide this difference.
>
> I think uapi consistency with other existing ctx is more important here.
> (i.e. keep the "local" port in host order). Otherwise, the user will
> be slapped left and right when writting bpf_prog in different prog_type.
>
> Armed with the knowledge on skc_num, the "local" port is
> in host byte order in the current existing prog ctx. It is
> unfortunate that the "dst"_port in this patch is the "local" port.
> The "local" port in "struct bpf_sock" is actually the "src"_port. :/
> Would "local"/"remote" be clearer than "src"/dst" in this patch?
I went and compared the field naming and byte order in existing structs:
| struct | field | byte order |
|----------------+------------+------------|
| __sk_buff | local_port | host |
| sk_msg_md | local_port | host |
| bpf_sock_ops | local_port | host |
| bpf_sock | src_port | host |
| bpf_fib_lookup | dport | network |
| bpf_flow_keys | dport | network |
| bpf_sock_tuple | dport | network |
| bpf_sock_addr | user_port | network |
It does look like "local"/"remote" prefix is the sensible choice.
I got carried away trying to match the field names with bpf_sock, which
actually doesn't follow the naming convention.
Will rename fields to local_*, remote_* in v2.
next prev parent reply other threads:[~2020-05-08 8:54 UTC|newest]
Thread overview: 34+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-05-06 12:54 [PATCH bpf-next 00/17] Run a BPF program on socket lookup Jakub Sitnicki
2020-05-06 12:54 ` [PATCH bpf-next 01/17] flow_dissector: Extract attach/detach/query helpers Jakub Sitnicki
2020-05-06 12:54 ` [PATCH bpf-next 02/17] bpf: Introduce SK_LOOKUP program type with a dedicated attach point Jakub Sitnicki
2020-05-06 13:16 ` Lorenz Bauer
2020-05-06 13:53 ` Jakub Sitnicki
2020-05-07 20:55 ` Martin KaFai Lau
2020-05-08 8:54 ` Jakub Sitnicki [this message]
2020-05-08 7:06 ` Martin KaFai Lau
2020-05-08 10:45 ` Jakub Sitnicki
2020-05-08 18:39 ` Martin KaFai Lau
2020-05-11 9:08 ` Jakub Sitnicki
2020-05-11 18:59 ` Martin KaFai Lau
2020-05-11 19:26 ` Jakub Sitnicki
2020-05-11 20:54 ` Martin KaFai Lau
2020-05-12 14:16 ` Jakub Sitnicki
2020-05-06 12:54 ` [PATCH bpf-next 03/17] inet: Store layer 4 protocol in inet_hashinfo Jakub Sitnicki
2020-05-06 12:55 ` [PATCH bpf-next 04/17] inet: Extract helper for selecting socket from reuseport group Jakub Sitnicki
2020-05-06 12:55 ` [PATCH bpf-next 05/17] inet: Run SK_LOOKUP BPF program on socket lookup Jakub Sitnicki
2020-05-06 12:55 ` [PATCH bpf-next 06/17] inet6: Extract helper for selecting socket from reuseport group Jakub Sitnicki
2020-05-06 12:55 ` [PATCH bpf-next 07/17] inet6: Run SK_LOOKUP BPF program on socket lookup Jakub Sitnicki
2020-05-06 12:55 ` [PATCH bpf-next 08/17] udp: Store layer 4 protocol in udp_table Jakub Sitnicki
2020-05-06 12:55 ` [PATCH bpf-next 09/17] udp: Extract helper for selecting socket from reuseport group Jakub Sitnicki
2020-05-06 12:55 ` [PATCH bpf-next 10/17] udp: Run SK_LOOKUP BPF program on socket lookup Jakub Sitnicki
2020-05-06 12:55 ` [PATCH bpf-next 11/17] udp6: Extract helper for selecting socket from reuseport group Jakub Sitnicki
2020-05-06 12:55 ` [PATCH bpf-next 12/17] udp6: Run SK_LOOKUP BPF program on socket lookup Jakub Sitnicki
2020-05-06 12:55 ` [PATCH bpf-next 13/17] bpf: Sync linux/bpf.h to tools/ Jakub Sitnicki
2020-05-06 12:55 ` [PATCH bpf-next 14/17] libbpf: Add support for SK_LOOKUP program type Jakub Sitnicki
2020-05-08 17:41 ` Andrii Nakryiko
2020-05-08 17:52 ` Yonghong Song
2020-05-08 17:59 ` Andrii Nakryiko
2020-05-11 8:12 ` Jakub Sitnicki
2020-05-06 12:55 ` [PATCH bpf-next 15/17] selftests/bpf: Add verifier tests for bpf_sk_lookup context access Jakub Sitnicki
2020-05-06 12:55 ` [PATCH bpf-next 16/17] selftests/bpf: Rename test_sk_lookup_kern.c to test_ref_track_kern.c Jakub Sitnicki
2020-05-06 12:55 ` [PATCH bpf-next 17/17] selftests/bpf: Tests for BPF_SK_LOOKUP attach point Jakub Sitnicki
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=87blmyvmc1.fsf@cloudflare.com \
--to=jakub@cloudflare.com \
--cc=ast@kernel.org \
--cc=bpf@vger.kernel.org \
--cc=daniel@iogearbox.net \
--cc=davem@davemloft.net \
--cc=dccp@vger.kernel.org \
--cc=edumazet@google.com \
--cc=gerrit@erg.abdn.ac.uk \
--cc=kafai@fb.com \
--cc=kernel-team@cloudflare.com \
--cc=kuba@kernel.org \
--cc=lmb@cloudflare.com \
--cc=marek@cloudflare.com \
--cc=netdev@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).