LKML Archive on lore.kernel.org
help / color / mirror / Atom feed
From: Lorenz Bauer <lmb@cloudflare.com>
To: John Fastabend <john.fastabend@gmail.com>,
	Daniel Borkmann <daniel@iogearbox.net>,
	Jakub Sitnicki <jakub@cloudflare.com>,
	Lorenz Bauer <lmb@cloudflare.com>,
	Eric Dumazet <edumazet@google.com>,
	"David S. Miller" <davem@davemloft.net>,
	Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>,
	Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org>,
	Jakub Kicinski <kuba@kernel.org>,
	Alexei Starovoitov <ast@kernel.org>
Cc: kernel-team@cloudflare.com, netdev@vger.kernel.org,
	bpf@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: [PATCH bpf-next v4 02/12] skmsg: update saved hooks only once
Date: Mon,  9 Mar 2020 11:12:33 +0000	[thread overview]
Message-ID: <20200309111243.6982-3-lmb@cloudflare.com> (raw)
In-Reply-To: <20200309111243.6982-1-lmb@cloudflare.com>

Only update psock->saved_* if psock->sk_proto has not been initialized
yet. This allows us to get rid of tcp_bpf_reinit_sk_prot.

Signed-off-by: Lorenz Bauer <lmb@cloudflare.com>
Reviewed-by: Jakub Sitnicki <jakub@cloudflare.com>
Acked-by: John Fastabend <john.fastabend@gmail.com>
---
 include/linux/skmsg.h | 20 ++++++++++++++++----
 net/ipv4/tcp_bpf.c    | 16 +---------------
 2 files changed, 17 insertions(+), 19 deletions(-)

diff --git a/include/linux/skmsg.h b/include/linux/skmsg.h
index 4d3d75d63066..2be51b7a5800 100644
--- a/include/linux/skmsg.h
+++ b/include/linux/skmsg.h
@@ -347,11 +347,23 @@ static inline void sk_psock_update_proto(struct sock *sk,
 					 struct sk_psock *psock,
 					 struct proto *ops)
 {
-	psock->saved_unhash = sk->sk_prot->unhash;
-	psock->saved_close = sk->sk_prot->close;
-	psock->saved_write_space = sk->sk_write_space;
+	/* Initialize saved callbacks and original proto only once, since this
+	 * function may be called multiple times for a psock, e.g. when
+	 * psock->progs.msg_parser is updated.
+	 *
+	 * Since we've not installed the new proto, psock is not yet in use and
+	 * we can initialize it without synchronization.
+	 */
+	if (!psock->sk_proto) {
+		struct proto *orig = READ_ONCE(sk->sk_prot);
+
+		psock->saved_unhash = orig->unhash;
+		psock->saved_close = orig->close;
+		psock->saved_write_space = sk->sk_write_space;
+
+		psock->sk_proto = orig;
+	}
 
-	psock->sk_proto = sk->sk_prot;
 	/* Pairs with lockless read in sk_clone_lock() */
 	WRITE_ONCE(sk->sk_prot, ops);
 }
diff --git a/net/ipv4/tcp_bpf.c b/net/ipv4/tcp_bpf.c
index 7d6e1b75d4d4..3327afa05c3d 100644
--- a/net/ipv4/tcp_bpf.c
+++ b/net/ipv4/tcp_bpf.c
@@ -637,20 +637,6 @@ static void tcp_bpf_update_sk_prot(struct sock *sk, struct sk_psock *psock)
 	sk_psock_update_proto(sk, psock, &tcp_bpf_prots[family][config]);
 }
 
-static void tcp_bpf_reinit_sk_prot(struct sock *sk, struct sk_psock *psock)
-{
-	int family = sk->sk_family == AF_INET6 ? TCP_BPF_IPV6 : TCP_BPF_IPV4;
-	int config = psock->progs.msg_parser   ? TCP_BPF_TX   : TCP_BPF_BASE;
-
-	/* Reinit occurs when program types change e.g. TCP_BPF_TX is removed
-	 * or added requiring sk_prot hook updates. We keep original saved
-	 * hooks in this case.
-	 *
-	 * Pairs with lockless read in sk_clone_lock().
-	 */
-	WRITE_ONCE(sk->sk_prot, &tcp_bpf_prots[family][config]);
-}
-
 static int tcp_bpf_assert_proto_ops(struct proto *ops)
 {
 	/* In order to avoid retpoline, we make assumptions when we call
@@ -670,7 +656,7 @@ void tcp_bpf_reinit(struct sock *sk)
 
 	rcu_read_lock();
 	psock = sk_psock(sk);
-	tcp_bpf_reinit_sk_prot(sk, psock);
+	tcp_bpf_update_sk_prot(sk, psock);
 	rcu_read_unlock();
 }
 
-- 
2.20.1


  parent reply	other threads:[~2020-03-09 11:13 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
     [not found] <20200309111243.6982-1-lmb@cloudflare.com>
2020-03-09 11:12 ` [PATCH bpf-next v4 01/12] bpf: sockmap: only check ULP for TCP sockets Lorenz Bauer
2020-03-09 11:12 ` Lorenz Bauer [this message]
2020-03-09 11:12 ` [PATCH bpf-next v4 03/12] bpf: tcp: move assertions into tcp_bpf_get_proto Lorenz Bauer
2020-03-09 11:12 ` [PATCH bpf-next v4 04/12] bpf: tcp: guard declarations with CONFIG_NET_SOCK_MSG Lorenz Bauer
2020-03-09 11:12 ` [PATCH bpf-next v4 05/12] bpf: sockmap: move generic sockmap hooks from BPF TCP Lorenz Bauer
2020-03-09 11:12 ` [PATCH bpf-next v4 06/12] bpf: sockmap: simplify sock_map_init_proto Lorenz Bauer
2020-03-09 11:12 ` [PATCH bpf-next v4 07/12] bpf: add sockmap hooks for UDP sockets Lorenz Bauer
2020-03-09 11:12 ` [PATCH bpf-next v4 08/12] bpf: sockmap: add UDP support Lorenz Bauer
2020-03-09 11:12 ` [PATCH bpf-next v4 09/12] selftests: bpf: don't listen() on UDP sockets Lorenz Bauer
2020-03-09 11:12 ` [PATCH bpf-next v4 10/12] selftests: bpf: add tests for UDP sockets in sockmap Lorenz Bauer
2020-03-09 11:12 ` [PATCH bpf-next v4 11/12] selftests: bpf: enable UDP sockmap reuseport tests Lorenz Bauer
2020-03-09 11:12 ` [PATCH bpf-next v4 12/12] bpf, doc: update maintainers for L7 BPF 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=20200309111243.6982-3-lmb@cloudflare.com \
    --to=lmb@cloudflare.com \
    --cc=ast@kernel.org \
    --cc=bpf@vger.kernel.org \
    --cc=daniel@iogearbox.net \
    --cc=davem@davemloft.net \
    --cc=edumazet@google.com \
    --cc=jakub@cloudflare.com \
    --cc=john.fastabend@gmail.com \
    --cc=kernel-team@cloudflare.com \
    --cc=kuba@kernel.org \
    --cc=kuznet@ms2.inr.ac.ru \
    --cc=linux-kernel@vger.kernel.org \
    --cc=netdev@vger.kernel.org \
    --cc=yoshfuji@linux-ipv6.org \
    --subject='Re: [PATCH bpf-next v4 02/12] skmsg: update saved hooks only once' \
    /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

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).