From: Lorenz Bauer <lmb@cloudflare.com>
To: Cong Wang <xiyou.wangcong@gmail.com>
Cc: Networking <netdev@vger.kernel.org>, bpf <bpf@vger.kernel.org>,
duanxiongchun@bytedance.com, wangdongdong.6@bytedance.com,
jiang.wang@bytedance.com, Cong Wang <cong.wang@bytedance.com>,
John Fastabend <john.fastabend@gmail.com>,
Daniel Borkmann <daniel@iogearbox.net>,
Jakub Sitnicki <jakub@cloudflare.com>
Subject: Re: [Patch bpf-next v2 2/9] sock: introduce sk_prot->update_proto()
Date: Tue, 2 Mar 2021 16:22:17 +0000 [thread overview]
Message-ID: <CACAyw9-SjsNn4_J1KDXuFh1nd9Hr-Mo+=7S-kVtooJwdi1fodQ@mail.gmail.com> (raw)
In-Reply-To: <20210302023743.24123-3-xiyou.wangcong@gmail.com>
On Tue, 2 Mar 2021 at 02:37, Cong Wang <xiyou.wangcong@gmail.com> wrote:
...
> @@ -350,25 +351,12 @@ static inline void sk_psock_cork_free(struct sk_psock *psock)
> }
> }
>
> -static inline void sk_psock_update_proto(struct sock *sk,
> - struct sk_psock *psock,
> - struct proto *ops)
> -{
> - /* Pairs with lockless read in sk_clone_lock() */
> - WRITE_ONCE(sk->sk_prot, ops);
> -}
> -
> static inline void sk_psock_restore_proto(struct sock *sk,
> struct sk_psock *psock)
> {
> sk->sk_prot->unhash = psock->saved_unhash;
Not related to your patch set, but why do an extra restore of
sk_prot->unhash here? At this point sk->sk_prot is one of our tcp_bpf
/ udp_bpf protos, so overwriting that seems wrong?
> - if (inet_csk_has_ulp(sk)) {
> - tcp_update_ulp(sk, psock->sk_proto, psock->saved_write_space);
> - } else {
> - sk->sk_write_space = psock->saved_write_space;
> - /* Pairs with lockless read in sk_clone_lock() */
> - WRITE_ONCE(sk->sk_prot, psock->sk_proto);
> - }
> + if (psock->saved_update_proto)
> + psock->saved_update_proto(sk, true);
> }
>
> static inline void sk_psock_set_state(struct sk_psock *psock,
> diff --git a/include/net/sock.h b/include/net/sock.h
> index 636810ddcd9b..0e8577c917e8 100644
> --- a/include/net/sock.h
> +++ b/include/net/sock.h
> @@ -1184,6 +1184,9 @@ struct proto {
> void (*unhash)(struct sock *sk);
> void (*rehash)(struct sock *sk);
> int (*get_port)(struct sock *sk, unsigned short snum);
> +#ifdef CONFIG_BPF_SYSCALL
> + int (*update_proto)(struct sock *sk, bool restore);
Kind of a nit, but this name suggests that the callback is a lot more
generic than it really is. The only thing you can use it for is to
prep the socket to be sockmap ready since we hardwire sockmap_unhash,
etc. It's also not at all clear that this only works if sk has an
sk_psock associated with it. Calling it without one would crash the
kernel since the update_proto functions don't check for !sk_psock.
Might as well call it install_sockmap_hooks or something and have a
valid sk_psock be passed in to the callback. Additionally, I'd prefer
if the function returned a struct proto * like it does at the moment.
That way we keep sk->sk_prot manipulation confined to the sockmap code
and don't have to copy paste it into every proto.
> diff --git a/net/core/sock_map.c b/net/core/sock_map.c
> index 3bddd9dd2da2..13d2af5bb81c 100644
> --- a/net/core/sock_map.c
> +++ b/net/core/sock_map.c
> @@ -184,26 +184,10 @@ static void sock_map_unref(struct sock *sk, void *link_raw)
>
> static int sock_map_init_proto(struct sock *sk, struct sk_psock *psock)
> {
> - struct proto *prot;
> -
> - switch (sk->sk_type) {
> - case SOCK_STREAM:
> - prot = tcp_bpf_get_proto(sk, psock);
> - break;
> -
> - case SOCK_DGRAM:
> - prot = udp_bpf_get_proto(sk, psock);
> - break;
> -
> - default:
> + if (!sk->sk_prot->update_proto)
> return -EINVAL;
> - }
> -
> - if (IS_ERR(prot))
> - return PTR_ERR(prot);
> -
> - sk_psock_update_proto(sk, psock, prot);
> - return 0;
> + psock->saved_update_proto = sk->sk_prot->update_proto;
> + return sk->sk_prot->update_proto(sk, false);
I think reads / writes from sk_prot need READ_ONCE / WRITE_ONCE. We've
not been diligent about this so far, but I think it makes sense to be
careful in new code.
--
Lorenz Bauer | Systems Engineer
6th Floor, County Hall/The Riverside Building, SE1 7PB, UK
www.cloudflare.com
next prev parent reply other threads:[~2021-03-03 3:55 UTC|newest]
Thread overview: 30+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-03-02 2:37 [Patch bpf-next v2 0/9] sockmap: introduce BPF_SK_SKB_VERDICT and support UDP Cong Wang
2021-03-02 2:37 ` [Patch bpf-next v2 1/9] sock_map: introduce BPF_SK_SKB_VERDICT Cong Wang
2021-03-02 2:37 ` [Patch bpf-next v2 2/9] sock: introduce sk_prot->update_proto() Cong Wang
2021-03-02 16:22 ` Lorenz Bauer [this message]
2021-03-02 18:23 ` Cong Wang
2021-03-03 9:35 ` Lorenz Bauer
2021-03-03 18:20 ` Cong Wang
2021-03-04 9:30 ` Lorenz Bauer
2021-03-04 23:52 ` Cong Wang
2021-03-06 0:27 ` John Fastabend
2021-03-06 0:57 ` Cong Wang
2021-03-06 1:55 ` John Fastabend
2021-03-09 17:53 ` Cong Wang
2021-03-10 6:33 ` John Fastabend
2021-03-02 2:37 ` [Patch bpf-next v2 3/9] udp: implement ->sendmsg_locked() Cong Wang
2021-03-02 2:37 ` [Patch bpf-next v2 4/9] udp: implement ->read_sock() for sockmap Cong Wang
2021-03-03 6:26 ` Yonghong Song
2021-03-02 2:37 ` [Patch bpf-next v2 5/9] udp: add ->read_sock() and ->sendmsg_locked() to ipv6 Cong Wang
2021-03-02 16:23 ` Lorenz Bauer
2021-03-02 17:59 ` Cong Wang
2021-03-02 2:37 ` [Patch bpf-next v2 6/9] skmsg: extract __tcp_bpf_recvmsg() and tcp_bpf_wait_data() Cong Wang
2021-03-02 2:37 ` [Patch bpf-next v2 7/9] udp: implement udp_bpf_recvmsg() for sockmap Cong Wang
2021-03-02 2:37 ` [Patch bpf-next v2 8/9] sock_map: update sock type checks for UDP Cong Wang
2021-03-03 6:37 ` Yonghong Song
2021-03-03 18:02 ` Cong Wang
2021-03-03 18:50 ` Yonghong Song
2021-03-02 2:37 ` [Patch bpf-next v2 9/9] selftests/bpf: add a test case for udp sockmap Cong Wang
2021-03-02 16:31 ` Lorenz Bauer
2021-03-02 18:05 ` Cong Wang
2021-03-03 10:20 ` Lorenz Bauer
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='CACAyw9-SjsNn4_J1KDXuFh1nd9Hr-Mo+=7S-kVtooJwdi1fodQ@mail.gmail.com' \
--to=lmb@cloudflare.com \
--cc=bpf@vger.kernel.org \
--cc=cong.wang@bytedance.com \
--cc=daniel@iogearbox.net \
--cc=duanxiongchun@bytedance.com \
--cc=jakub@cloudflare.com \
--cc=jiang.wang@bytedance.com \
--cc=john.fastabend@gmail.com \
--cc=netdev@vger.kernel.org \
--cc=wangdongdong.6@bytedance.com \
--cc=xiyou.wangcong@gmail.com \
/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).